Green Energy and Technology

Nabil Derbel · Quanmin Zhu Editors

Modeling, Identification
and Control Methods
in Renewable Energy
Systems
Green Energy and Technology
More information about this series at http://www.springer.com/series/8059
Nabil Derbel • Quanmin Zhu
Editors

Modeling, Identification and

Control Methods in
Renewable Energy Systems

123
Editors
Nabil Derbel Quanmin Zhu
Sfax Engineering School University of the West of England
University of Sfax Bristol, UK
Sfax, Tunisia

ISSN 1865-3529 ISSN 1865-3537 (electronic)

Green Energy and Technology
ISBN 978-981-13-1944-0 ISBN 978-981-13-1945-7 (eBook)
https://doi.org/10.1007/978-981-13-1945-7

Library of Congress Control Number: 2018964938

© Springer Nature Singapore Pte Ltd. 2019

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, express or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore
Preface

This book entitled Modeling, Identification and Control Methods in Renewable

Energy Systems consists of a representation of theoretical results related to renew-
able energy systems. It is not our intention to cover all renewable energy systems
of recent control-related contributions, but the priority has been placed on the
application of wind energy and solar energy with works related to their appli-
cations, such as vibration energy, converters, FPGA-based control, and MPPT
techniques. Intelligent approaches such as fuzzy systems, neural networks, and
genetic algorithm constitute powerful tools applied to the best work of these energy
systems. This book is made of 17 invited chapters that have been written by leading
researchers covering recent theoretical developments and applications of renewable
energy systems.
The book presents theoretical explorations on several fundamental problems for
several kinds of renewable energy systems. By integrating fresh concepts and state-
of- the-art results to form systematic approaches for controlling these systems,
fundamental theoretical approaches as well as practical framework have been
established.
This book is intended for readers from a wide spectrum related to photovoltaic
systems and wind energy systems. It is expected that readers require background
knowledges to understand various concepts and results presented in this book.
Moreover, this book provides not only theoretical results but also experimental case
studies to highly motivate young students and researchers.
The book has been organized into the following chapters which are regrouped
into three parts.
• The first part contains four chapters detailing fundamentals of photovoltaic
systems. The first chapter of this part presents the modeling and performance
analysis of a solar PV power system connected to a three phase. The second
chapter is dedicated to the photovoltaic power prediction using recurrent neural
networks. A comparison between two behavior MPPT techniques applied to
photovoltaic systems has been given in Chap. 3. Chapter 4 is dedicated to
discussion on cascaded boost converters for photovoltaic chains.

v
vi Preface

• The second part is dedicated to wind energy systems. It contains seven chapters.
The fifth chapter applies the control of the power of a DFIG generator using
the MPPT technique. Chapter 6 studies the ripple correlation on the optimal
point determination in a medium power wind conversion system. Chapter 7
considers the control of wind turbine based on PMSG using pitch angle control.
Chapter 8 consists to apply a nonlinear control of a variable speed wind energy
conversion system-based PMSG. Chapter 9 is an application of the direct power
control of a DFIG using sliding mode control approach. Chapter 10 presents a
comparative study between PI and sliding mode control for the DFIG of a wind
turbine. Chapter 11 is an emulation of wind turbine for stand-alone wind energy
conversion systems.
• The third part is dedicated to tools for renewable energy systems. It is composed
by six chapters. Chapter 12 starts this part by the vibration control by piezoelec-
tric elements for renewable energy systems. Chapter 13 considers the feedback
T-S fuzzy controller in a finite frequency for wind turbines. Chapter 14 considers
an intelligent load frequency control in the presence of wind power generation.
Chapter 15 proposes a fault tolerant control of switch power converter in WECS
based on a DFIG. Finally, Chapters 16 and 17 consider FPGA-based control
applied to a DFIG-based wind power system and to a PMSG on variable wind
speed turbine.
We wish to express our sincere gratitude to colleagues who have contributed
to this book. First of all, we are particularly indebted to our colleagues who have
contributed their excellent research in order to bring valuable materials for graduate
students, researchers, and practitioners. We greatly appreciate the contributors
to this book for their patience and time taken to collaborate with us to finally
complete this book. Next, particular thanks go to students and colleagues who
helped engaging in the preparation and assisted us to improve this book. Finally,
we would like to sincerely express our deepest gratitude to the Springer editorial
staff for their continuous support, assistance, and significant improvement in the
manuscript. Without their help, the book would not be published as scheduled.

Sfax, Tunisia Nabil Derbel

Bristol, UK Quanmin Zhu
May 2018
Contents

1 Modeling and Performance Analysis of a Solar PV Power

System Connected to a Three Phase Load Under Irradiation
and Load Variations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Fatima-Ezzahra Tahiri, Khalid Chikh, and Mohamed Khafallah
2 Photovoltaic Power Prediction Using Recurrent Neural Networks . . . 25
Rim Ben Ammar and Abdelmajid Oualha
3 A Comprehensive Comparison of Two Behavior MPPT
Techniques, the Conventional (Incremental Conductance
(INC)) and Intelligent (Fuzzy Logic Controller (FLC))
for Photovoltaic Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Aouatif Ibnelouad, Abdeljalil El Kari, Hassan Ayad, and Mostafa
Mjahed
4 Modeling and Comparison of Boost Converter With Cascaded
Boost Converters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Nesrine Boujelben, Ferdaous Masmoudi, Mohamed Djemel,
and Nabil Derbel
5 Control of Power of a DFIG Generator with MPPT Technique
for Wind Turbines Variable Speed. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Yasmine Ihedrane, Chakib El Bekkali, Badre Bossoufi,
and Manale Bouderbala
6 The Ripple Correlation Optimal Point Determination in a
Medium Power Wind Conversion System and Performance
Evaluation with Respect to Conventional Algorithms . . . . . . . . . . . . . . . . . . 131
Hassan Abouobaida and Said EL Beid
7 Control of Wind Turbine Based on PMSG Using Pitch Angle
Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151
Aroua Slimen, Hatem Tlijani, Mehdi Dhaoui, and Rached B. Younes

vii
viii Contents

8 Nonlinear Control of a Variable Speed Wind Energy

Conversion System Based PMSG . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
Marwa Ayadi and Nabil Derbel
9 Direct Power Control of DFIG Using Sliding Mode Control
Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193
Aicha Daoud and Nabil Derbel
10 A Comparative Study Between PI and Sliding Mode Control
for the DFIG of a Wind Turbine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205
Basma Boujoudi, Elm’kaddem Kheddioui, Nadia Machkour,
and Mohammed Bezza
11 Emulation of Wind Turbine for Standalone Wind Energy
Conversion Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
L. Benaaouinate, M. Khafallah, A. Mesbahi, A. Martinez,
T. Bouragba, and D. Breuil
12 SSDI-Max Control and Its Applications in Renewable Energies. . . . . . 245
Aida Chérif, Djamila Zehar, Nabil Derbel, and Claude Richard
13 Feedback T-S Fuzzy Controller in Finite Frequency for Wind
Turbine. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
Youssef Berrada, Abderrahim El-Amrani, and Ismail Boumhidi
14 Intelligent Load Frequency Control in Presence of Wind
Power Generation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 281
Nour EL Yakine Kouba and Mohamed Boudour
15 Fault Tolerant Control of Switch Power Converter in WECS
Based on a DFIG. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
Amina Tamer, Azeddine Bendiabdellah, Bilal Djamal Eddine Cherif,
and Djillali Toumi
16 ZedBoard-FPGA Control of the DFIG Based Wind Power System . . 333
Hassane Mahmoudi, Marouane El Azzaoui, and Chafik Ed-Dahmani
17 An FPGA-Based Control of the PMSG on Variable Wind
Speed Turbine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357
Hassane Mahmoudi, Chafik Ed-Dahmani, and Marouane El Azzaoui
Chapter 1
Modeling and Performance Analysis of a
Solar PV Power System Connected to a
Three Phase Load Under Irradiation and
Load Variations

Fatima-Ezzahra Tahiri, Khalid Chikh, and Mohamed Khafallah

Abstract This chapter describes a stand-alone solar photovoltaic system with a

robust controllers which are Incremental Conductance and Perturb and Observe
used to enhance the maximum power point tracking and stable output power under
variations of solar irradiation, temperature and electrical loads. Indeed, to supply an
alternative load with a sinusoidal line without harmonic distortion under weather
conditions, algorithms techniques are used to control a DC-DC boost converter to
generate the Maximum Power Point of the photovoltaic generator to alternative load
via a PWM (Pulse Width Modulation) three phase inverter. The computer simulation
results, in Matlab/Simulink environment, presented in this chapter validate the full
PV system components and show that the photovoltaic simulation system can track
the maximum power point accurately using the two maximum power point tracking
techniques.

Keywords Photovoltaic system (PV) · Boost converter · Maximum power point

tracking (MPPT) · Maximum power point (MPP) · Three phase inverter · Load
variations · Weather conditions · Alternative load · Perturbation and observation
(P&O) · Incremental conductance (INC)

F.-E. Tahiri () · M. Khafallah

Energy and Electrical Systems Laboratory, Hassan II University ENSEM Casablanca,
Casablanca, Morocco
e-mail: [email protected]
K. Chikh
Engineering and Applied Technology Laboratory (LITA), High School of Technology, Sultan
Moulay Slimane University, Béni Mellal, Morocco

© Springer Nature Singapore Pte Ltd. 2019 1

N. Derbel, Q. Zhu (eds.), Modeling, Identification and Control Methods
in Renewable Energy Systems, Green Energy and Technology,
https://doi.org/10.1007/978-981-13-1945-7_1
2 F.-E. Tahiri et al.

Nomenclature

IP V : Solar cell current (A)

VP V : Solar cell output voltage (V)
I0 : Diode saturation current (A)
RS : Solar cell series resistance (Ω)
Eg : Energy band gap (eV)
n: Ideality factor (between 1 and 2)
q: Electron charge (1.6 × 10−19 C)
k: Boltzmann constant (1.38 × 10−23 J/K)
G: irradiance (W/m2 )
T: Cell temperature in Kelvin (K)
Tref : Reference cell temperature (298.15 K)
IL : Light generated current (A)
Icc : Short circuit current (A)
Voc : Open circuit voltage (V)
D: Duty cycle
VS : Output voltage of the boost converter (V)
VP V : Output voltage of the PV generator (V)
fC : Cut-off frequency (Hz)
Lf : Filter inductor (H)
Cf : Filter capacitor (F)

1.1 Introduction

The solar photovoltaic system is one of the important renewable energy sources.
It converts sunlight into electricity and offers many advantages such as the energy
produced is not polluting, requiring little maintenance, most promising and inex-
haustible (Jiang et al. 2005).
The basic device of a photovoltaic system is the photovoltaic cell. Cells may
be grouped to form panels or modules. Panels can be grouped to form large
photovoltaic arrays. The term array is usually employed to describe a photovoltaic
panel (with several cells connected in series and/or parallel) or a group of panels
(Bauer 2010).
However, the PV system still has relatively low conversion efficiency. Indeed,
the power delivered by the PV module depends on the irradiance, temperature, and
electrical loads and it has a maximum (MPP) at a certain working point. At the MPP
(Maximum Power Point), the PV operates at its highest efficiency (Freeman 2010;
Sera et al. 2006a,b; Suryakumari et al. 2013; Tahiri et al. 2016, 2017).
Therefore, to extract the maximum power under the different conditions stated
earlier, a maximum power point tracking (MPPT) technique is used to control the
fluctuating operating power point of the PV array via a DC-DC converter.
1 Solar PV Power System Connected to a Three Phase Grid 3

The MPPT controller can be realized based on different methods and algorithms.
The most popular methods are known as Perturb and Observe (P&O) and Incremen-
tal Conductance (INC).
This chapter treats the deepened modeling of a whole photovoltaic system
composed of a photovoltaic generator, power electronic components, filter and local
loads.
This work is organized as follows: solar cell modeling and the simulation
behavior of the PV module at different operating conditions will be presented and
discussed in Sects. 1.2 and 1.3, respectively. In Sect. 1.4, the MPPT techniques are
discussed and tested. Whereas in Sect. 1.5 the others components of the PV system
are presented and the simulation results in Matlab/Simulink environment of the full
PV power system are talked in order to evaluate its performance under different
conditions. Eventually, conclusion is presented in Sect. 1.6.

1.2 Modeling of the Solar Cell

Solar cells consist of a p-n junction fabricated in thin wafer or layer of semicon-
ductors (Kinjal et al. 2015), whose electrical characteristics differ very little from
a diode represented by the equation of Shockly (Alsadi and Alsayid 2012). Thus
the simplest equivalent circuit of a solar cell is a current source in parallel with a
diode. The output of the current source is directly proportional to the light falling
on the cell. For this research work, a model of moderate complexity was used. The
circuit diagram for the solar cell is shown in Fig. 1.1 (Villalva et al. 2009; Walker
and Geoffrey 2000).
Equations which describe the I − V characteristics of the cell are: (Makhlouf
et al. 2012)
   
q(VP V + Rs IP V )
IP V = IL − I0 exp −1 (1.1)
nkT
G
IL (G, T ) = Icc [1 + a(T − Tref )] (1.2)
1000

Fig. 1.1 Equivalent model of PV cell

4 F.-E. Tahiri et al.

 3   
T n qEg 1 1
I0 (T ) = I0 (Tref ) exp − (1.3)
Tref nk Tref T
Icc
I0 (Tref ) =   (1.4)
qV0C
exp −1
nkTref

dPP V  nkTref
Rs = − −   (1.5)
dIP V V0C qV0C
Icc q exp
nkTref

1.3 Photovoltaic Module

1.3.1 Photovoltaic Module Characteristics

BP SX 150 solar array PV module is chosen for a MATLAB simulation model.

The module is made of 72 multicrystalline silicon solar cells in series and provides
150W of nominal maximum power. Table 1.1 shows its electrical specification take
from datasheet (Datasheet, BP SX 150, 2002).

1.3.2 Photovoltaic Module Simulation

Figure 1.2 shows the modeling circuit of the PV module in environment Mat-
lab/Simulink. The modeling of the PV is done applying the equations seen before,
(1.1), (1.2), (1.3), (1.4) and (1.5). Note that the irradiance and temperature are the
inputs of the system.
The P − VP V and IP V − VP V characteristics at standard temperature condition
are given in Figs. 1.3 and 1.4.
The characteristic P − VP V has only one maximum power point called MPP.
Simulation results are identical to the values given by the datasheet.

Table 1.1 Electrical Maximum Power (Pmax) 150W

characteristics data of the BP
Voltage at Pmax (Vmp) 34.5V
SX 150 SOLAR AT 25 ◦ C,
5AM, 1000 W/m2 Current at Pmax (Imp) 4.35A
Open-circuit voltage (Voc) 43.5V
Short-circuit current (Isc) 4.75A
(KI) Temperature coefficient of Isc (0.065 ± 0.015)%/◦ C
(KV) Temperature coefficient of Voc −(160 ± 20)mV/ ◦ C
NOCT 47± 2 ◦ C
1 Solar PV Power System Connected to a Three Phase Grid 5

Fig. 1.2 PV Module Model in environment Matlab/Simulink

Fig. 1.3 Simulated P − VP V curve of BP SX 150S PV module (1 kW/m2 , 25 ◦ C)

Fig. 1.4 Simulated IP V − VP V curve of BP SX 150S PV module (1 kW/m2 , 25 ◦ C)

6 F.-E. Tahiri et al.

1.3.3 Effect of Changes in Parameters on the Characteristics

of PV Module

Figures 1.5 and 1.6 show the effect of irradiance variation (400, 500, 600, 700, 800,
900, 1000) W/m2 at constant temperature (T = 25 ◦ C) respectively on IP V − VP V
and P −VP V characteristics. As one can see in these figures, the module current and
the MPP are proportional to the irradiation, while the open-circuit voltage changes
slightly with irradiation (Ferchichi et al. 2016) (Yadav et al. 2016).

Fig. 1.5 Effect of the irradiation on IP V − VP V characteristic of BP SX 150S PV module at

T = 25 ◦ C

Fig. 1.6 Effect of the irradiation on P − VP V characteristic of BP SX 150S PV module at

T = 25 ◦ C
1 Solar PV Power System Connected to a Three Phase Grid 7

Fig. 1.7 Effect of the temperature on IP V − VP V characteristic of BP SX 150S PV module at

G = 1000 W/m2

Fig. 1.8 Effect of the irradiation on P −VP V characteristic of BP SX 150S PV module at G = 1000
W/m2

In Figs. 1.7 and 1.8 the simulation results are shown for different values of tem-
perature (T = 25 ◦ C, T = 50 ◦ C, T = 75 ◦ C) at constant irradiance (G = 1000 W/m2 )
respectively on IP V − VP V and P − VP V characteristics.
From Fig. 1.7 the generated power is gradually decreased and maximum power
available more at low temperature. From Fig. 1.8 the current is constant up to some
voltage and then it will decrease gradually.
Moreover, when a PV module is directly coupled to a load, the PV module’s
operating point depends strongly on the characteristics of the load (Fig. 1.9). In other
term, the impedance of load dictates the operating condition of the PV module and
only the optimal load, which passes through its characteristic MPP, allows extract
the maximum power (Nemsi et al. 2013).
From Figs. 1.5, 1.6, 1.7, 1.8, and 1.9, we conclude that the MPP of the PV
generator changes continuously under variations of solar irradiation, temperature
8 F.-E. Tahiri et al.

Fig. 1.9 IP V − VP V curves of BP SX 150S PV module and various resistive loads Simulated with
the MATLAB model (G = 1000 W/m2 , T = 25 ◦ C)

PV ARRAY

CHOPPER LOAD

IPV VPV
PWM
Generator

MPPT
Algorithm

Fig. 1.10 Block diagram of a PV array connected to the load

and electrical loads. Therefore a MPPT technique is required to exploit the

maximum power of the module and optimize the efficiency of PV cells.

1.4 Maximum Power Point Tracking (MPPT) Techniques

MPPT is an adaptive system used to control a static converter between the load and
the PV panel (Fig. 1.10).
1 Solar PV Power System Connected to a Three Phase Grid 9

This converter is designed to fit every time the apparent impedance of the load to
the impedance of PV field corresponding to the maximum power point. This method
is based on the use of a search algorithm of maximum power of the photovoltaic
panel curve (Mohssine et al. 2015). There are many different MPPT techniques
available in the literature; the most largely used algorithms are described in the
following sections.

1.4.1 Perturbation and Observation (P&O)

The most commonly used MPPT algorithm is the Perturb and Observe (P&O), due
to its ease of implementation in its basic form. In Fig. 1.11, if the operating voltage

Fig. 1.11 Flowchart of the P&O MPPT method

10 F.-E. Tahiri et al.

dP
of the PV array is perturbed in a given direction and > 0, it is known that the
dV
perturbation moved the array’s operating point toward the MPP. The P&O algorithm
dP
would then continue to perturb the PV array voltage in the same direction. If <
dV
0, then the change in operating point moved the PV array away from the MPP,
and the P&O algorithm reverses the direction of the perturbation (Hohm and Ropp
2000).
The advantage of the P&O method is that it is easy to implement. However, it
has some limitations, like oscillations around the MPP in steady state operation,
slow response speed, and even tracking in wrong way under rapidly changing
atmospheric conditions (Sera et al. 2006a).

1.4.2 Incremental Conductance (INC) Method

dI
The incremental conductance uses the PV array’s incremental conductance dV to
dP
compute the sign of . It does this using an expression derived from the condition
dV
dP
that, at the MPP, = 0. Beginning with this condition, it is possible to show that,
dV
dI I
at the MPP + = 0. Thus, incremental conductance can determine that the
dV V
MPPT has reached the MPP and stop perturbing the operating point. If this condition
is not met, the direction in which the MPPT operating point must be perturbed can
dI I
be calculated using the relationship between and − (Hohm and Ropp 2000)
dV V
The INC can track rapidly increasing and decreasing irradiance conditions
with higher accuracy than P&O. However, because of noise and errors due to
measurement and quantization, this method also can produce oscillations around
the MPP; and it also can be confused in rapidly changing atmospheric conditions.
Another disadvantage of this algorithm is the increased complexity when compared
to perturb and observe. This increases computational time, and slows down the
sampling frequency of the array voltage and current (Fig. 1.12) (Sera et al. 2006a).

1.4.3 Simulation Results

Figure 1.13 shows the modeling circuit of the PV system in Matlab/Simulink

environment. The PV generator connected to a resistive load through a Boost
converter controlled by two algorithms MPPT: P&O and INC.
To test the performance of the two algorithms (P&O and INC) in the same
simulation conditions, we vary the irradiation, temperature and load as shown below.
1 Solar PV Power System Connected to a Three Phase Grid 11

Fig. 1.12 Flowchart of the INC MPPT method

1.4.3.1 Variation of Irradiation

Figure 1.14 shows the convergence of the output power of the PV system to the
MPP using P&O and INC commands under variation of irradiation as indicated in
Fig. 1.15.

1.4.3.2 Temperature Variation

Figure 1.16 shows the convergence of the output power of the PV system to the
MPP using P&O and INC commands under variation of temperature as indicated in
Fig. 1.17.
12 F.-E. Tahiri et al.

Fig. 1.13 PV system in environment Matlab/Simulink

Fig. 1.14 Output power of the PV system verses time using P&O algorithm and INC algorithm at
temperature 25 ◦ C and different solar irradiation levels 1000, 800, 700, 600 and 400 W/m2

1.4.3.3 Load Variation

Figures 1.18, 1.19 and 1.20 show the output power of the PV system respectively
without MPPT algorithm, using INC command and using P&O command.
1 Solar PV Power System Connected to a Three Phase Grid 13

Fig. 1.15 Variation of solar irradiation

Fig. 1.16 Output power of the PV system verses time using P&O algorithm and INC algorithm at
solar irradiation 1000 W/m2 and different temperature levels 0, 10, 25, 30, 40 and 20 ◦ C

1.5 The Full PV Power System

Figure 1.21 shows the simplified block structure of the investigated system. It
contains a photovoltaic generator, power electronic components, filter and local
loads.
The DC/DC converter with an MPPT (Maximum Power Point Tracker) con-
nected to the solar array to optimize the PV output, a second DC/DC converter
is connected to the output of this converter to increase the voltage received from the
photovoltaic generator to the voltage level required by the voltage source inverter
(VSI) via a LC filer.
14 F.-E. Tahiri et al.

Fig. 1.17 Variation of temperature

Fig. 1.18 Output power of the PV system verses time without MPPT algorithm at different value
of resistive load 10Ω, 30Ω, 50Ω and 100Ω

1.5.1 DC Voltage Control

The continuous part of the PV system is presented in Fig. 1.22. The first DC/DC
converter is a boost chopper circuit controlled by an MPPT algorithm for rise the
low PV power to an optimal level under different weather conditions (Benkhelil and
Gherbi 2012).
The second DC/DC converter is a boost which raises the input voltage to the
voltage level demanded by the VSI and works with a fixed duty cycle (Bauer 2010).
1 Solar PV Power System Connected to a Three Phase Grid 15

Fig. 1.19 Output power of the PV system verses time using INC algorithm at different value of
resistive load 10, 30, 50 and 100 Ω

Fig. 1.20 Output power of the PV system verses time using P&O algorithm at different value of
resistive load 10, 30, 50 and 100 Ω

Fig. 1.21 Block diagram of the system

16 F.-E. Tahiri et al.

Fig. 1.22 DC voltage control

Fig. 1.23 Three-phase VSI

1.5.2 Voltage Source Inverter (VSI) Model

The VSI used in this work converts DC voltage to three-phase AC voltages.

Figure 1.23 shows the detail circuit of the inverter. The power part of the inverter is
composed of three arms consisting each one two switches. Each switch is composed
of transistor (IGBT, MOSFET. . . ) and of a diode coupled in parallel. The L−C filter
is connected to the inverter output, it’s designed to reduce high order harmonics
produced by the pulse-width modulation (PWM) inverters (Benadli et al. 2015).
1 Solar PV Power System Connected to a Three Phase Grid 17

Fig. 1.24 Reference voltages and triangular wave carrier

1.5.3 PWM Control

A pulse width modulation (PWM) signal is a method for generating an analog

signal using a digital source. A PWM signal consists of two main components that
define its behavior: a duty cycle and a frequency. The interest of the PWM is to
provide a variable DC voltage from a fixed DC voltage source. PWM provides a
way to decrease the Total Harmonic Distortion (THD) of load current. The THD
requirement can be met when the output of a PWM inverter is filtered since the
unfiltered output of a PWM based inverted will have a relatively high distortion.
There are different types of PWM schemes, the most used ones being sinusoidal
PWM (SPWM), hysteresis PWM (HPWM), space vector modulation (SVM) and
optimal PWM (Bratt 2011). The inverter switching, used in this studied system,
is controlled by the SPWM. The high-frequency triangular carrier wave VCR is
compared with a sinusoidal reference Vm of the desired frequency (shown in
Fig. 1.24) to realize SPWM and generate the gating signal for the transistors (shown
in Fig. 1.25).

1.6 LC Filter

The filter is an essential element after semiconductor converter. The filter reduces
the effects caused by switching semiconductor devices on other devices. L − C
filter is one of the main existing harmonic filter topologies for three-phase inverters.
It delivers a sinusoidal output current with THD less than 5% (Nordin et al. 2014).
Consequently, the low-pass L − C filter is designed so that the cut-off frequency,
fC is higher than the load current and voltage frequency and lower than the inverter
switching frequency, based following equation:
18 F.-E. Tahiri et al.

Fig. 1.25 Pulses generated to the first arm of inverter

1
fC =  (1.6)
2π Lf Cf

1.6.1 Simulation Results

The power system is simulated in Matlab/Simulink power system toolbox software.

In this work, the simulation model for the full PV power system components is
designed (shown in Fig. 1.26).
A Perturbation and observation MPPT controller has been implemented in order
to deliver the maximum power. The performance of the designed model is evaluated
under changing weather conditions through the following case studies.

1.6.1.1 Case 1: Irradiation and Load Fixes

Figure 1.27 shows the simulation results of the power system model in stable
weather conditions. In this case the resulting output line voltage is stable at 400 V.
1 Solar PV Power System Connected to a Three Phase Grid 19

Fig. 1.26 Model for the full PV power system in Matlab/Simulink environment

Fig. 1.27 PV system output voltage and current waveforms with a constant irradiation and load
(G = 1000 W/m2 and R = 3 kΩ)

1.6.1.2 Case 2: Irradiation Variation

Figure 1.29 presents the simulation results of the power system model using P&O
commands under irradiation variation as indicated in Fig. 1.28. The resulting output
line and phase voltage remain constant.
20 F.-E. Tahiri et al.

Fig. 1.28 Variation of irradiation

Fig. 1.29 PV system output voltage in case of irradiation variation

1.6.1.3 Case 3: Load Variation

Figure 1.30 presents the simulation presents of the power system model under load
variation. For this case, the output current varies depending on the electrical load
value and the output voltage remains constant in steady state after a fast dynamic
state.
1 Solar PV Power System Connected to a Three Phase Grid 21

Fig. 1.30 PV system output voltage and current waveforms in case of load variation (R = 3 kΩ
before t = 0.3 s and R = 5 kΩ after this time) and constant irradiation G = 1000 W/m2 .

1.7 Conclusion

In this chapter, modeling of a stand-alone solar PV power system has been evaluated
and validated under changing weather conditions and load variations through
different cases in Matlab/Simulink environment. According to the results of the
simulation, we conclude that:
• The PV generator performance deteriorates with increasing temperature, decreas-
ing of solar irradiation and variation of the electrical load.
• The two MPPT algorithms (P&O and INC) make the output power of the PV
system to be kept at the maximum when the irradiation and temperature varied
and adjusts the PV generator to the load.
• The three phase voltage waveform, in the output inverter, remains constant in
spite of irradiation or load variations.

References

Alsadi, S., & Alsayid, B. (2012). Maximum power point tracking simulation for photovoltaic
systems using perturb and observe algorithm. International Journal of Engineering and
Innovative Technology (IJEIT), 2(6), 83.
Bauer, J. (2010). Single phase voltage source inverter photovoltaic application. ActaPolytechnica,
50(4), 7–11.
22 F.-E. Tahiri et al.

Bellia, H., Youcef, R., & Fatima, M. (2014). A detailed modeling of photovoltaic module using
MATLAB. NRIAG Journal of Astronomy and Geophysics, 3, 53–61.
Benadli, R., Khiari, B., & Sellami, A. (2015). Three-phase grid-connected photovoltaic system
with maximum power point tracking technique based on voltage-oriented control and using
sliding mode controller. In 6th International Renewable Energy Congress, USA.
Benkhelil, E., & Gherbi, A. (2012). Modeling and simulation of grid-connected photovoltaic
generation system. In Revue des Energies Renouvelables SIENR’12 Ghardaïa.
Bratt, J. (2011). Grid connected PV inverters: Modeling and simulation. Thesis, Faculty of San
Diego State University.
Ferchichi, M., Zaidi, N., & Khedher, A. (2016). Comparative analysis for various control
strategies based MPPT technique of photovoltaic system using DC-DC boost converter. In 17th
International Conf. on Sciences and Techniques of Automatic control & computer engineering,
STA’2016, Sousse, Tunisia.
Freeman, D. (2010). Introduction to photovoltaic systems maximum power point tracking. Appli-
cation Report, SLVA 446. Texas Instruments.
Hohm, D. P., & Ropp, M. E. (2000). Comparative study of maximum power point tracking
algorithms using an experimental, programmable, maximum power point tracking test bed.
In Conference Record of the Twenty-Eighth IEEE Photovoltaic Specialists, Anchorage, Alaska
(pp. 1699–1702).
Jiang, J., Huang, T., Hsiao, Y., & Chen, C. (2005). Maximum power tracking for photovoltaic
power systems. Tamkang Journal of Science and Engineering, 8(2), 147–153.
Kinjal, P., Shah, K. B., & Patel, G. R. (2015). Comparative Analysis of P&O and INC MPPT
Algorithm for PV System. In International Conference on Electrical, Electronics, Signals,
Communication and Optimization (EESCO), Visakhapatnam, Andhrapradesh, India.
Makhlouf, M., Messai, F., & Benalla, H. (2012). Modeling and simulation of grid-connected
photovoltaic distributed generation system. Journal of Theoretical and Applied Information
Technology, 45(2), 378–386.
Mohssine, H., Kourchi, M., Bouhouch, H., & Debbagh, F. (2015). Perturb and observe (P&O) and
incremental conductance (INC) MPPT algorithms for PV panels. International Journal of Soft
Computing and Engineering, 5(2), 123–126.
Nemsi, S., Barazane, L., Diaf, S., & Malek, A. (2013). Comparative study between two maximum
power point tracking (MPPT) techniques for photovoltaic system. Revue des Energies Renou-
velables, 16(4), 773–782.
Nordin, A., Omar, A., & Zainuddin, H. (2014). Modeling and simulation of grid inverter in grid-
connected photovoltaic system. International Journal of Renewable Energy Research, 4(4),
949–957.
Sera, D., Kerekes, T., Teodorescu, R., & Blaabjerg, F. (2006a). Improved MPPT algorithms
for rapidly changing environmental conditions. In 12th International Conference on Power
Electronics and Motion Control EPE-PEMC, USA.
Sera, D., Kerekes, T., Teodorescu, R., & Blaabjerg, F. (2006b). Improved MPPT algorithms
for rapidly changing environmental conditions. In 12th International Conference on Power
Electronics and Motion Control EPE-PEMC, USA.
Suryakumari, J., Sahiti, G., & Sudhakar, G. (2013). Analysis and simulation of perturb and observe
MPPT technique. International Journal of Engineering Research and Applications (IJERA),
3(4), 1662–1668.
Tahiri, F. E., Chikh, K., Khafallah, M., & Saad, A. (2016). Comparative study between two
Maximum Power Point Tracking techniques for Photovoltaic System. In 2nd International
Conference on Electrical and Information Technologies ICEIT, Tangier, Morocco.
Tahiri, F. E., Chikh, K., Khafallah, M., Saad, A., & Breuil, D. (2017). Modeling and performance
analysis of a solar PV power system under irradiation and load variations. In 13th International
Multi-conference on Systems, Signals and Devices SSD, Marrakech, Morocco.
1 Solar PV Power System Connected to a Three Phase Grid 23

Villalva, M. G., Gazoli, J. R., & Ruppert, E. (2009). Modeling and circuit-based simulation of
photovoltaic arrays. Brazilian Journal of Power Electronics, 14(1), 35–45.
Walker, G. (2000). Evaluating MPPT converter topologies using a Matab PV model. In Innovation
for Secure Power (pp. 138–143). Brisbane: Queensland University of Technology.
Yadav, P., Bhargava, A., & Sharma, M. (2016). MPPT using soft computing technique with MLI in
photovoltaic system. In International Conference on Micro-Electronics and Telecommunication
Engineering, Ghaziabad, Uttarpradesh, India. IEEE.
Chapter 2
Photovoltaic Power Prediction Using
Recurrent Neural Networks

Rim Ben Ammar and Abdelmajid Oualha

Abstract The intermittent characteristic of the photovoltaic power, due to the

variability of the weather conditions, involves many problems in grid energy
management. Therefore, the PV power forecasting becomes crucial to ensure grid
stability and economic dispatch. Artificial neural network (ANN) techniques present
alternative approaches to solve nonlinear problems in various areas. They can be
trained and applied for prediction. A particular type of ANN namely the recurrent
neural network (RNN) has shown powerful capabilities for PV power forecasting.
The paper investigates and compares the efficiency of several RNN architectures
specifically the modified Elman, Jordan and the hybrid model combining the latest
topologies.
The useful data for prediction are acquired from the National Institute of
Meteorology. The performance of these topologies is validated by calculating the
Root Mean Squared Error, the Mean Absolute Error and the Correlation Factor.
The results show that forecasting with the modified Elman outperforms the Jordan
and the hybrid networks.

Keywords Artificial neural network · PV power · Forecasting · Prediction ·

Recurrent neural network · Elman · Jordan

2.1 Introduction

Over the recent years, photovoltaic systems have been progressively installed
worldwide as a means of energy production. The intermittent characteristic of
the photovoltaic power engenders many problems in grid management (Kou
et al. 2013). Thus, PV system power prediction becomes highly recommended to
guarantee the grid stability and facilitate the energy dispatching (Grazia De Giorgi

R. B. Ammar () · A. Oualha

Sfax Engineering School, University of Sfax, Sfax, Tunisia
e-mail: [email protected]

© Springer Nature Singapore Pte Ltd. 2019 25

N. Derbel, Q. Zhu (eds.), Modeling, Identification and Control Methods
in Renewable Energy Systems, Green Energy and Technology,
https://doi.org/10.1007/978-981-13-1945-7_2
26 R. B. Ammar and A. Oualha

et al. 2014). Various approaches are used to forecast the pv power production
capacity. Physical prediction methods consist in establishing an equation related
to the PV system characteristic and the weather data. However, it is limited to
a particular weather condition and to a specific PV system (Pelland et al. 2013;
Zhu et al. 2015). The artificial neural network presents another potential alternative
for developing predictive models as it can solve complex non-linear relationships
between input and output data.
There are two major categories of the neural network: the Feed-forward which
transmits the data forward from input to output and the recurrent which disposes of
a feedback loop where data can be fed back (Abdulkarim 2016).
In the present work, recurrent neural networks were used for short, medium and
long term PV power forecasting. A comparative study between them is carried out.
The remaining of this paper was organized as follows. Section 2.2 provided the
artificial neural network forecast models. Section 2.3 presented the recurrent neural
network topologies. Section 2.4 defined the recurrent neural network training. The
recurrent neural network evaluation was proven in Sect. 2.5. In Sect. 2.6, simulation
results of the proposed forecasting approaches and their performances evaluation
were discussed before drawing our major conclusion in Sect. 2.7.

2.2 Artificial Neural Network Forecast Models

An Artificial Neural Network (ANN) is an information processing paradigm that

models operations of biological neural systems.
It consists of an interconnected processing elements (neurons) working in
harmony to solve specific problems. The fundamental structure of an ANN includes
an input, hidden and output layers. Figure 2.1 shows that the neuron involves two
parts: the weights summation and the activation function. The weights summation
defines the network inputs combination inside the neurone. The output of the
network is related to the inputs through an activation function. The most commonly
used activation function is the sigmoid one (Mellit 2009).

Fig. 2.1 The basic neuron

2 Photovoltaic Power Prediction Using Recurrent Neural Networks 27

The ANN showed their notable capability in diverse applications like prediction.
The ANN is able to learn the complex relationships between the output and the input
after training. Different types of artificial neural networks have been used to solve
the forecasting problems. The Radial Basis Function Neural Network (RBFNN) and
the Back Propagation neural network (BPNN) are among the tools for the PV power
prediction (Dragomir and Dragomirb 2014). Their combination with the wavelet
transform leads to more accurate results (Mandal et al. 2012).
Two main types of artificial neural networks have been used for PV system
power forecasting. These are the general regression neural network (GRNN) and
the feed-forward back propagation (FFBP). A simulation test has shown that the
FFBP outperforms the GRNN model (Saberian et al. 2014).
In Zhu et al. (2015), the authors confirm that the hybrid model formed by the
ANN and the wavelet decomposition is recommended for PV power prediction.
A methodology based on ANN and the Analog Ensemble (AnEn) technique to
procreate forecasts of the photovoltaic power based on weather and astronomical
predictions have been presented in Cervone et al. (2017).
In Chow et al. (2012), the authors have used a classical ANN architecture
model: multilayer perceptron (MLP) for short-term prediction. The input data are
the solar azimuth angles, the solar elevation, the solar radiation data and the dry
bulb temperature.
A hybrid method which combines the clear sky PV curves and the ANN called
Physical Hybrid Artificial Neural Network (PHANN) have been used for PV system
power prediction (Dolara et al. 2015).
The recurrent neural network RNN is another category of ANN.
This methodology is more efficient for PV power production prediction as it
supplies feedback to the network. Actually, it provides a dynamic internal feedback
loops to store information for later use and improve the learning efficiency. The
recurrent neural network is able to deal with non linear prediction and guarantees a
faster convergence. Thus, it is the best solution for problems that include dynamic
changes over time and dependency on the past values time-series to predict those of
the future. This is the case of the PV power forecasting (Mellit and Kalogirou 2008;
Wang et al. 2016; Wysocki and Lawryn 2016). The major type of the feedback
network is the partially recurrent networks. Among these topologies we can cite the
Elman, Jordan and the combination between them.

2.3 Recurrent Neural Networks Topologies

2.3.1 The Modified Elman Neural Network

The modified Elman neural network was proposed by Pham and Liu based on its
origin structure presented by Pham. The difference between them lies in the context
neuron which has a self feed-back coefficient. The recurrent topology has not only
classical layers namely input, hidden, and output, but also, a context layer. The
28 R. B. Ammar and A. Oualha

Output Layer

Hidden Layer

α α

Context Layer Input Layer

Fig. 2.2 The modified Elman network architecture

context unit is used to store a copy from the output of the hidden layer and from
itself (Pham and Liu 1996).
The feedback path enables the modified ENN to perceive temporal and spatial
patterns. Figure 2.2 shows the architecture of the modified Elman neural network.
The hidden layer output is led for both, output and context layers.
The input layer and the context layer output are sent to the hidden layer input. The
weights of the recurrent connections between hidden and context neurons are fixed
to one. Hidden and context layers have the same number of neurons (RamaKrishna
et al. 2014).
The mathematical model of the modified ENN is depicted in these expressions:


n
vi (k) = x
ωij (k − 1)xjc (k) + ωiu (k − 1)u(k) (2.1)
j =1

xi (k) = f (vi ) (2.2)

xjc (k) = xi (k − 1) + αxjc (k − 1) (2.3)


n
y
y(k) = ωi (k − 1)xi (k) (2.4)
i=1

Where n is the number of the context neurons which is equal to the number of the
hidden neurons, u is the input of the modified Elman network, y is denoted as the
output, vi presents the total input of the ith hidden unit, xi and xjc are the output
of the ith hidden unit and the jth context unit, α is the self feed-back coefficient,
2 Photovoltaic Power Prediction Using Recurrent Neural Networks 29

x designs the weights between context and hidden layers, ωu defines the weights
ωij i
y
between input and hidden layers, ωi characterizes the weights between hidden and
output layers.
f is the sigmoid activation function for the hidden layer given by (2.5) (Wei et al.
2007).

1
f (x) = (2.5)
1 + exp(−x)

For the output layer, the activation function is often taken as a linear function.

2.3.2 The Jordan Neural Network

Figure 2.3 shows The Jordan Neural Network (JNN) which was proposed by De
Mulder et al. (2015). The JNN output, which is turned back, is stored in a context
unit. Each unit receives a copy from the output neurons and from itself. Similar
to the modified ENN, the context unit has a self feed-back coefficient which lies
between 0 and 1. The output of the context layer is directly connected to the input
of the hidden layer. Sigmoid and linear activation functions for the hidden and the
output layers, respectively. The Output and context layers have the same number of
neurons (Husaini et al. 2011).
The mathematical model of the JNN and the modified ENN differs only in the
equation of the context layer which is described by the following expression:

xjc (k) = y(k − 1) + αxjc (k − 1) (2.6)

Fig. 2.3 The Jordan neural 1

network architecture

Output Layer

Hidden Layer

Input Layer
α
Context Layer
30 R. B. Ammar and A. Oualha

Output Layer

1
1

Hidden Layer

α α α
Input Layer
Context Layer

Fig. 2.4 The modified Elman-Jordan neural network architecture

Where xjc denotes the output of the jth context unit, y presents the output of the
Jordan neural network and α is the self feed-back coefficient.

2.3.3 Hybrid Model: Modified Elman-Jordan Neural Network

The hybrid model is a combination of the modified ENN and the JNN. Figure 2.4
presents the architecture of this network. The output unit is led to the first context
layer and the outputs of the hidden nodes are transmitted to the second context layer.
The first and the second context units are presented by (2.6) and (2.3) respectively
(RamaKrishna et al. 2014).

2.4 Recurrent Neural Network Training

One of the most important characteristics of the artificial neural networks is their
ability of learning from patterns. ANN learning paradigms can be arranged as
supervised and unsupervised learning. The supervised learning model, as shown
in Fig. 2.5, acquires the existence of the desired output for a given set of inputs. It
teaches the network the correct response for each sample. The weights are adjusted
relying the error between the produced and the required outputs during the training.
However, the unsupervised learning model identifes the pattern class information
heuristically and does not require the knowledge of the desired outputs (Du and
Swamy 2014; Sathya and Abraham 2013).
2 Photovoltaic Power Prediction Using Recurrent Neural Networks 31

Target

Neural Network
including connections Compare
(weights) between
Input neurons Output

Adjust
Weight

Fig. 2.5 Supervised learning

One of the most popular supervised training algorithms is the standard back
propagation. It is used to compute the necessary weights corrections. The first step
consists on making a random choice of the weights values of the network. After
running, the network outputs are calculated and compared with the desired outputs.
Then, the error from the desired and actual outputs is considered. This error is used
to update the weights in a backward pass. The whole process is repeated until the
error becomes negligible (Pham and Liu 1996).
The back propagation algorithm can be defined by the following expressions:
y
Δωi (k) = η(yd (k) − y(k))xi (k) (2.7)
y
Δωiu (k) = η(yd (k) − y(k))ωi (k − 1)fvi u(k) (2.8)
y
x
Δωij (k) = η(yd (k) − y(k))ωi (k − 1)fvi xjc (k) (2.9)

y
Δωiu (k) = η(yd (k) − y(k))ωi (k − 1)fvi xjc (k) (2.10)

∂f
fvi = = f (vi )(1 − f (vi )) = xi (k)(1 − xi (k)) (2.11)
∂vi
y
where Δωi presents the weight variation between hidden and output layers, Δωiu
characterizes the weight variation between input and hidden layers, Δωij x defines

the weight variation between context and hidden layers, yd is the desired output, y
is the output of the network, xi is the output of the ith hidden unit and vi presents
the total input of the ith hidden unit.
η determines the learning rate, which presents the weight updating step size. This
learning rate should be large enough to fit the data well, but small enough to keep
away from over-fitting.
It is to be noted that for recurrent neural networks, the learning algorithm is
more complex, taking into account the historic of neurons. For this reason, we have
used the standard back propagation algorithm dedicated for non recurrent neural
32 R. B. Ammar and A. Oualha

networks. Sometimes, this algorithm gives an unstable network. In this case, the
procedure consists on applying the learning algorithm with other initial conditions
and tests numerically the stability of the network.

2.5 Recurrent Neural Networks Design

This section detailed all the steps of the neural network design. The first step
consisted in defining the number of the neurons in the different layers. In the second
step, the necessary data for training and testing the recurrent architectures was
presented. Finally, a test of the different topologies was performed using the neural
performances evaluation.

2.5.1 Recurrent Neural Networks Design

The number of hidden layers that are necessary to ensure a satisfactory network
performance has been a controversial topic. Due to the absence of a theoretical
answer, heuristics processes have been assigned and generated that one hidden
layer is enough to characterize the task because several hidden layers may produce
undesired complexity to the problem (Gabriel 2014). Thus, for the prediction of the
photovoltaic power, one hidden layer would be enough.
The modified Elman network has 2 neurons in the input, hidden and context
layers and 1 neuron in the output layer. The Jordan network has 2 neurons in the
input and hidden layers and 1 neuron in the context and output layers. The hybrid
model has 2 neurons in the input and hidden layers, 3 neurons in the context layer
and 1 neuron in the output layer.

2.5.2 Data Collection

Meteorological conditions, especially temperature and solar radiation, may affect

PV power system efficiency. Thus, we put these two parameters as input data for the
recurrent topologies.
The historical data, from 2012 to 2015, are collected from the National Institute
of Meteorology. These data are related to the region of Sfax, Tunisia.
The historical data are divided into two categories: the data pertaining to the
first 3 years were used for training, whereas, the observations of the last year were
reserved for testing the different networks.
The historical data do not have the same magnitude. Thus, they should be
normalized in the range of [0 1]. The following formula converts them to the same
scale (Carcano et al. 2008).
2 Photovoltaic Power Prediction Using Recurrent Neural Networks 33

x − xmin
xnorm = (2.12)
x − xmax

Where xnorm is the normalized value, x is the actual value, xmax and xmin are the
maximum and minimum values.

2.6 Recurrent Neural Networks Evaluation

The connection weights were initially chosen as random numbers.The weights were
updated during training using the standard back propagation process. When the error
was low enough, the new weights were kept. These values were used for testing the
networks.
There is a wide range of functions for neural network performance evaluation
such as the Root Mean Squared Error (RMSE), Mean Absolute Error (MAE) and
the Correlation Factor given by (2.12), (2.13), and (2.14) (Gabriel 2014; Mellit and
Pavan 2010; Zhou et al. 2007).

1
n
1
RMSE = 100 (y − yd )2 (2.13)
(ymax − ymin ) n
i=1

1
n
1
MAE = 100 |y − yd | (2.14)
(ymax − ymin ) n
i=1

⎛ 
n ⎞
(y − yd )2
⎜ ⎟
R 2 = 100 ⎜ ⎟
i=1
⎝1 − 
n ⎠ (2.15)
(y − Ȳ )2
i=1

where ymax , ymin and Ȳ present the maximum, minimum and mean value of the
actual data, respectively. y and yd are the measured and the predicted data.

2.7 Forecast Results and Discussion

The different topologies of the recurrent neural network used for forecasting are
created in an m-file using MATLAB. There are three time horizons in the PV power
prediction as shown in Fig. 2.6. Short term forecasting is aheived for one day to one
week, medium term prediction for more than one weak to one month and long term
forecasting for more than one month to one year (Yadav et al. 2015).
The photovoltaic panel used in simulation is SOLAREX-MSX60. Figure 2.7
presents the equivalent electrical circuit model of a PV cell: a current source parallel
34 R. B. Ammar and A. Oualha

Types of PV power
forecasting

Short term Medium term Long term

forecasting forecasting forecasting

Fig. 2.6 Types of photovoltaic power forecasting

Icell Rs

Id Ish

Iph Vcell
Rsh

Fig. 2.7 PV cell Equivalent electrical circuit model

with one diode. Rs and Rsh are the shunt and the series resistances, respectively
(Keles et al. 2013).
The current, the voltage and the power provided by the PV cell are formulated
by (2.15), (2.16), (2.17) and (2.18) (Keles et al. 2013).
The cell temperature can be calculated from the Nominal Operating Cell
Temperature (NOCT) and the air temperature using (2.19) (Babalola et al. 2014)
 
G G
Icell = Ir + α (Tc − Tcr ) + − 1 Isc (2.16)
Gr Gr

Vcell = −β(Tc − Tcr ) − Rs Δ(I ) + Vr (2.17)

 
G G
Δ(I ) = α (Tc − Tcr ) + − 1 Isc (2.18)
Gr Gr

P = Vcell Icell (2.19)

G
Tc = Ta + (N OCT − 20) (2.20)
800
where, Icell and Vcell are the cell current and voltage. Ir , Vr represent the reference
current and voltage. G and Gr are the radiation and the reference radiation.
Tc and Tcr are the cell temperature and the reference cell temperature respec-
tively. Isc is the short circuit current. Δ(I ) is the current variation. P is the
2 Photovoltaic Power Prediction Using Recurrent Neural Networks 35

Table 2.1 Typical electrical characteristics of SOLAREX-MSX60

Pmax Vr Ir α β Isc Rs NOCT
60 17.1 V 3.5 3 × 10−3 73 × 10−3 3.8 0.47 47 ± 2
W V A mA/◦ C mW/◦ C V Ω

70

60

50
Power (W)

40

30

20

10

0
0 200 400 600 800 1000
Day

Fig. 2.8 Photovoltaic power

photovoltaic power. Ta defines the ambient temperature. α, β are the short circuit
current and the open circuit voltage temperature coefficient.
The parameters of the photovoltaic panel, SOLAREX-MSX60, used in this study
are depicted in Table 2.1 for a solar radiation of 1000 W/m2 and the cell temperature
is around 25 ◦ C. These values correspond to Gr and Tcr , respectively.
The PV powers, during this period, are calculated using (2.1), (2.2), (2.3), (2.4)
and (2.5). Figure 2.8 displays the daily photovoltaic powers of the different years,
covering a period of 1097 days.

2.7.1 Short Term Photovoltaic Power Forecasting

2.7.1.1 Daily Forecasts

The curves of the measured and the forecasted power of the photovoltaic panel
using the modified ENN, JNN and the hybrid model, are shown in Figs. 2.9, 2.11,
and 2.13. From these graphical depictions, it is clear that the predictions are fairly
36 R. B. Ammar and A. Oualha

80
Measured Power
70 Predicted Power

60

50
Power (W)

40

30

20

10

0
0 100 200 300
Day

Fig. 2.9 Measured and predicted PV powers with the modified Elman network

close to the actual measurements. The average error of the modified ENN, illustrated
in Fig. 2.10, doesn’t exceed 2.5%. It is lower than the error of the JNN and the hybrid
model displayed in Figs. 2.12 and 2.14, which is in the range of 4% and 6%. Thus
the modified Elman network outperforms the other topologies.
To estimate the success of the different recurrent structures, three accuracy
evaluators are introduced in Table 2.2 namely RMSE, MAE and the correlation
factor, are calculated according to Eqs. (2.12), (2.13) and (2.14).
The lower values of the RMSE and MAE that do not exceed 2.6%, and the
higher percentages of the correlation factor which are around 99%, show that a
considerable accuracy level was reached by the feedback networks. Nevertheless,
the modified ENN demonstrates better forecasting results with a high value of
the correlation factor equal to 99.8% and low percentages of RMSE and MAE in
the range of 1.3% and 1%. Therefore, it is more efficient for a daily PV power
prediction.

2.7.1.2 Weekly Forecasts

In this part, the feedback networks were used for weekly forecasts. The curves of
predicted and measured power, showed in Figs. 2.15, 2.16 and 2.17, demonstrate a
good agreement between them.
2 Photovoltaic Power Prediction Using Recurrent Neural Networks 37

80
Measured Power
70 Predicted Power

60

50
Power (W)

40

30

20

10

0
0 100 200 300
Day

Fig. 2.10 Measured and predicted PV powers with the Jordan neural network

80
Measured Power
70 Predicted Power

60

50
Power (W)

40

30

20

10

0
100 200 300
Day

Fig. 2.11 Measured and predicted PV powers with the Hybrid neural network
38 R. B. Ammar and A. Oualha

2.5

1.5
Error (%)

0.5

0
50 100 150 200 250 300 350
Day

Fig. 2.12 Daily relative forecasting error with the modied Elman network

4
Error (%)

0
100 200 300
Day

Fig. 2.13 Daily relative forecasting error with the Jordan neural network
2 Photovoltaic Power Prediction Using Recurrent Neural Networks 39

Error (%) 4

0
100 200 300
Day

Fig. 2.14 Daily relative forecasting error with the Hybrid neural network

Table 2.2 Daily prediction accuracy evaluation

Accuracy-evaluators Modified Elman Jordan Modified Elman-Jordan
RMSE % 1.3088 2.6358 2.3750
MAE % 1.0007 2.0904 1.8184
R2 % 99.7962 99.1733 99.3288

In addition, the error rate presented in Fig. 2.18 was less than 9%, which
highlighted the accuracy of the predictors.
Furthermore, as it can be seen in Table 2.3, the percentage of the correlation
factor obtained from the modified Elman model is equal to 99.8%, the MAE value
is only 1.16% and the RMSE value is around 2.1. These percentages are lower
than those of the JNN, which are equal to 99.16%, 7.6% and 2.55%, respectively.
Also, these rates are lower than those given by the hybrid model, which are equal
to 99.43%, 16.41% and 2.3%. Thus, the modified ENN presents a high degree of
accuracy for the weekly forecasts.

2.7.2 Medium Term Photovoltaic Power Forecasting: Monthly

Forecasts

In the next stage of the study, medium term photovoltaic power forecasting is
discussed. The recurrent networks were developed to forecast the monthly PV
power. Just like the daily forecasts, the new weights generated after training were
40 R. B. Ammar and A. Oualha

500 Predicted Power

Measured Power
450

400

350
Power (W)

300

250

200

150

100
10 20 30 40 50
Week

Fig. 2.15 Measured and predicted PV powers with the modified Elman network

500 Predicted Power

Measured Power
450

400

350
Power (W)

300

250

200

150

100
10 20 30 40 50
Week

Fig. 2.16 Measured and predicted PV powers with the Jordan neural network
2 Photovoltaic Power Prediction Using Recurrent Neural Networks 41

500 Predicted Power

Measured Power
450

400

350
Power(W)

300

250

200

150

100
10 20 30 40 50
Week

Fig. 2.17 Measured and predicted PV powers with the Hybrid neural network

10
Modified Elman

JNN
8
Hybrid Model

6
Error (%)

0
0 10 20 30 40 50
Week

Fig. 2.18 Weekly relative forecasting error

42 R. B. Ammar and A. Oualha

Table 2.3 Weekly prediction accuracy evaluation

Accuracy-evaluators Modified Elman Jordan Modified Elman-Jordan
RMSE% phantom02.1341 phantom07.5934 16.4117
MAE% phantom01.1686 phantom02.5540 phantom02.2921
R2 % 99.8349 99.1631 99.3433

2200
Measured Power
2000 Predicted Power

1800

1600
Power (W)

1400

1200

1000

800

600

1 2 3 4 5 6 7 8 9 10 11 12
Month

Fig. 2.19 Measured and predicted PV powers with the modified Elman network

used for testing the feedback networks. Figures 2.19, 2.20 and 2.21 show the
comparison between the measured and the predicted data.
The curves presented in Fig. 2.22 indicate that the error rate does not exceed 7%.
Similar to the daily and weekly predictions, these monthly prognoses seem to be
closer to their measured values.
The Several statistical error measures are detailed in table IV. For the modified
Elman network, the determination coefficient R 2 is found to be 99.8%; it is higher
than 99.2% and 99.3% for the Jordan and the Hybrid networks, respectively. In
addition, the RMSE and MAE rates are equal to 1.09% and 1.16%. These values
are lower than 3.8% and 2.5% achieved by Jordan network. Also, they are much
lower than 8.3% and 2.4% of the hybrid model. All these results confirm that the
best predictions are obtained by the modified ENN.
The Monthly average error is smaller than the weekly one. For the modified
Elman network, the maximum of error in weekly forecasts is 8%. In the monthly
forecasts, the value becomes in the range of 6%. Also, for the Jordan network
the maximum of error is reduced from 4.95% to 4.7%. Similarly, the hybrid
model recorded a rate decrease from 6.6% to 4.2%. Thus, the forecasting accuracy
increases as the forecasting interval is reduced (Table. 2.4).
2 Photovoltaic Power Prediction Using Recurrent Neural Networks 43

2200
Measured Power
2000
Predicted Power

1800

1600
Power (W)

1400

1200

1000

800

600
1 2 3 4 5 6 7 8 9 10 11 12
Month

Fig. 2.20 Measured and predicted PV powers with the Jordan neural network

2200
Measured Power
2000 Predicted Power

1800

1600
Power (W)

1400

1200

1000

800

600

1 2 3 4 5 6 7 8 9 10 11 12
Month

Fig. 2.21 Measured and predicted PV powers with the Hybrid neural network
44 R. B. Ammar and A. Oualha

7
Modified ENN
6 Hybrid Model
JNN
5

4
Error (%)

0
1 2 3 4 5 6 7 8 9 10 11 12
Month

Fig. 2.22 Monthly relative forecasting error

Table 2.4 Monthly prediction accuracy evaluation

Accuracy-evaluators Modified Elman Jordan Modified Elman-Jordan
RMSE % 1.0901 3.8489 8.2965
MAE % 1.1620 2.5019 2.3950
R2 % 99.8526 99.2340 99.3732

Table 2.5 Yearly prediction Accuracy Modified Jordan Modified

accuracy evaluation
evaluators Elman Elman-Jordan
RMSE % 0.3115 1.0998 2.3708
MAE % 0.3115 1.0998 2.3708
R2 % 99.9990 99.9879 99.9438

2.7.3 Long Term Photovoltaic Power Forecasting: Yearly

Forecasts

Finally, long term photovoltaic power forecasting is introduced. The feedback neural
networks were also used to try to forecast the yearly photovoltaic power output.
The statistical test between measured and predicted powers is summarized in
Table 2.5. This table shows a good statistical behavior for our predictions with
a correlation factor of around 99%. For the Jordan network the RMSE and MAE
percentages do not exceed 1.1%. In the hybrid model, the average error of the MAE
2 Photovoltaic Power Prediction Using Recurrent Neural Networks 45

and RMSE is equal to 2.3%. It can be noticed for the modified Elman network,
the MAE is less than 0.4%. Therefore, this proposed network shows superior
forecasting performance over the Jordan and the hybrid models.

2.8 Conclusion

The effective analysis of the different topologies of recurrent neural network

proposed in this paper for short, medium and long term photovoltaic power
forecasting, proved its accuracy with a high correlation factor rate about 99%. The
validity of the proposed RNN architectures is confirmed by comparing the predicted
and the measured powers. The average of MAPE, RMSE, which do not exceed
2.2%, acquired by the modified Elman neural network is much lower than those
obtained with the JNN and the Hybrid models. Hence, the modified ENN shows a
higher level of performance to forecast PV system power output. Therefore, it is
more efficient when context units are fed from the hidden layer instead of the output
layer. Also, the combination between the Jordan network and the modified Elman
network induces more complexity on the network without yielding more accuracy
than the modified Elman topology. The feedback networks present more accurate
results as the forecasting interval is reduced. In fact, the recorded rates of the RMSE
and MAE for the monthly forecasts are of smaller values compared to the weekly
forecasts.
In conclusion, it can be deduced that the recurrent topologies can be easily
adopted for forecasting the PV power and could be widely used for an efficient
planning of the renewable systems utilization.

References

Abdulkarim, S. A. (2016). Time series prediction with simple recurrent neural networks. Bayero
Journal of Pure and Applied Sciences, 9, 19–24.
Babalola, O. S., Komolafe, O. A., Jegede, O. O., & Ayoola, M. A. (2014). Photovoltaic generating
system parameter sizing for building. Journal of Energy Technologies and Policy, 4, 65–73.
Casaca de Rocha Vaz, A. G. (2014). Photovoltaic forecasting with artificial neural networks. Ph.D.
Thesis, Faculty of Sciences, University of Lisboa.
Carcano, E. C., Bartolinia, P., Musellib, M., & Piroddic, L. (2008). Jordan recurrent neural
network versus IHACRES in modelling daily stream flows. Journal of Hydrology, 362, 291–
307. Elsevier.
Cervone, G., Harding, L. C., Alessandrini, S., & Monache, L. D. (2017). Short term photovoltaic
power forecasting using artificial neural networks and an analog ensemble. Renewable Energy,
108, 274–286.
Chow, S. K. H., Lee, E. W. M., & Li, D. H. W. (2012). Short-term prediction of photovoltaic energy
generation by intelligent approach. Energy Build, 55, 660–667.
De Giorgi, M. G., Maria, C. P., & Malvoni, M. (2014). Photovoltaic power forecasting using
statistical methods: Impact of weather data. IET Science Measurement and Technology, 8, 90–
97.
46 R. B. Ammar and A. Oualha

De Mulder, W., Bethard, S., & Moens, M. F. (2015). A survey on the application of recurrent neural
networks to statistical language modeling. Computer Speech & Language, 30, 61–98.
Dolara, A., Grimaccia, F., Leva, S., Mussetta, M., & Ogliari, E. (2015). A physical hybrid artificial
neural network for short term forecasting of PV plant power output. Energies, 8, 1138–1153.
Dragomir, F., & Dragomirb, O. E. (2014). Forecasting of photovoltaic power generation by RBF
neural networks. Advanced Materials Research, 918, 200–205.
Du, K. L., & Swamy, M. N. S. (2014). Neural networks and statistical learning. Lon-
don/Heidelberg/New York/Dordrecht: Springer.
Husaini, N. A., Ghazali, R., Nawi, N. M., & Ismail, K. H. (2011). Jordan Pi-Sigma neural network
for temperature prediction. Communications in Computer and Information Science, 151, 547–
558.
Keles, C., Alagoz, B. B., Akcin, M., Kaygusuz, A., & Karabiber, A. (2013). A photovoltaic system
model for matlab/simulink simulations. In 4th International Conference on Power Engineering,
Energy and Electrical Drives.
Kou, J., Liu, J., Li, Q., Fang, W., Chen, Z., Liu, L., & Guan, T. (2013). Photovoltaic power
forecasting based on artificial neural network and meteorological data. In TENCON 2013 –
2013 IEEE Region 10 Conference, 31194 (pp. 1–4).
Pelland, S., Remund, J., Kleiss, J., Oozeki, T., & Brabandere, K. D. (2013). Photovoltaic and solar
forecasting: State of the art. IEA Intenatioanl Energy Agency, 14, 1–36.
Mandal, P., Teja, S. S., Ui Haque, A., Meng, J., & Mineda, R. L. (2012). Forecasting power output
of solar photovoltaic system using wavelet transform and artificial intelligence techniques.
Progress in Energy and Combustion Science, 12, 332–337. Elsevier.
Mellit, A. (2009). Recurrent neural network-based forecasting of the daily electricity generation of
a photovoltaic power system. In EVER’09, Monaco.
Mellit, A., & Kalogirou, S. A. (2008). Artificial intelligence techniques for photovoltaic applica-
tions. Progress in Energy and Combustion Science, 34, 574–632. Elsevier.
Mellit, A., & Pavan, A. M. (2010). 24-h forecast of solar irradiance using artificial neural network,
application for performance prediction of a grid-connected PV plant at Trieste. Solar Energy,
84, 221–226. Elsevier.
Pham, D. T., & Liu, X. (1996). Training of Elman networks and dynamic system modelling.
International Journal of Systems Science, 27, 221–226.
RamaKrishna, K., Ramam, V. A., & Rao, R. S. (2014). Mathematical neural network, MaNN. Mod-
els part IV, recurrent neural networks, RecNN. In bio-/chemical-tasks. Journal of Applicable
Chemistry, 3(6), 2209–2311.
Saberian, A., Hizam, H., Radzi, M. A. M., Abkadir, M. Z. A., & Mirzaei, M. (2014). Modelling and
prediction of photovoltaic power output using artificial neural networks. International Journal
of Photoenergy, 2014, 1–10.
Sathya, R., & Abraham, A. (2013). Comparison of supervised and unsupervised learning algo-
rithms for pattern classification. International Journal of Advanced Research in Artificial
Intelligence, 2(2), 34–38.
Wang, J., Wang, J., Fang, W., & Niu, H. (2016). Financial time series prediction using Elman
recurrent random neural networks. Computational Intelligence and Neuroscience, 2016, 1–14.
Wei, G. H., Qian, F., Chun, L. Y., Du, W., & Wang, L. (2007). Identification and control of nonlinear
systems by a dissimilation particle swarm optimization-based Elman neural network. Nonlinear
Analysis Real Word Applications, 9, 1345–1360.
Wysocki, A., & Lawryn, C. M. (2016). Elman neural network for modeling and predictive control
of delayed dynamic systems. Archives of Control Sciences, 26, 117–142.
Yadav, H. K., Pal, Y., & Tripathi, M. M. (2015). Photovoltaic power forecasting methods in smart
power grid. In Annual IEEE India Conference (INDICON), New Delhi.
Zhou, W., Yang, H., & Fang, Z. (2007). A novel model for photovoltaic array performance
prediction. Applied Energy, 84, 1187–1198. Elsevier.
Zhu, H., Li, X.,Sun, Q., Nie, L., Yao, J., & Zhao, G. (2015). A power prediction method for
photovoltaic power plant based on wavelet decomposition and artificial neural networks.
Energies, 9, 11.
Chapter 3
A Comprehensive Comparison of Two
Behavior MPPT Techniques, the
Conventional (Incremental Conductance
(INC)) and Intelligent (Fuzzy Logic
Controller (FLC)) for Photovoltaic
Systems

Aouatif Ibnelouad, Abdeljalil El Kari, Hassan Ayad, and Mostafa Mjahed

Abstract This chapter presents a detailed procedure to study and discuss the
behavior of different maximum power point tracking (MPPT) techniques applied
to PV systems. In this work, we presented a review on the state-of-the-art of
photovoltaic System, DC/DC converter and power point tracking techniques such as
conventional one incremental conductance (INC) and soft computing method fuzzy
logic controller (FLC) are evaluated. The simulation results obtained are developed
under software MATLAB/Simulink. Both methods (INC) and (FLC) are used with a
boost DC/DC converter and a load. These results show that the fuzzy logic controller
is better and faster than the conventional incremental conductance (INC) technique
in both dynamic response and steady state in normal operation.

Keywords MPPT · PV modeling · Technique INC · Technique FLC · Boost

DC/DC converter

3.1 Introduction

In recent years, a demand for electric power has been increasing as well as the
constraints linked to its production, such as the effect of pollution and global
warming, conduct research into the development of renewable energy sources.

A. Ibnelouad () · A. El Kari · H. Ayad

Department of Applied Physics, Laboratory of Electrical Systems and Telecommunications,
Faculty of Sciences and Technologies, Cadi Ayyad University, Marrakech, Morocco
M. Mjahed
Department of Mathematics and Systems, Royal School of Aeronautics, Marrakech, Morocco

© Springer Nature Singapore Pte Ltd. 2019 47

N. Derbel, Q. Zhu (eds.), Modeling, Identification and Control Methods
in Renewable Energy Systems, Green Energy and Technology,
https://doi.org/10.1007/978-981-13-1945-7_3
48 A. Ibnelouad et al.

Fig. 3.1 Characteristic I = f (V ) of the photovoltaic panel

In this context, photovoltaic solar energy is one of the most important sources
of renewable energy, which is an outcome to our problems of energy production.
Moreover, this energy seems the most promising, non-polluting and inexhaustible.
Nevertheless, the production of this energy is nonlinear and it varies according
to the luminous intensity and the temperature. Therefore, the operating point of the
photovoltaic panel PV does not always coincide with the maximum power point
(Fig. 3.1).
We use a mechanism that allows the search and tracking of the maximum power
point tracking (MPPT) so that the maximum power is permanently generated (Azab
2008).
A MPPT is used for harvesting the maximum energy from the photovoltaic panel
PV and transporting that energy to the load, on condition that using an appropriate
duty cycle to configure the DC/DC converter. It is essential technique for optimal
operation of the photovoltaic system. The principle of this control is based on the
automatic variation of the duty cycle of the DC-DC converter, to the optimum
value to maximize the power delivered by the PV panel. It used for matching the
characteristics of the load with those of the solar panels. A DC/DC converter ensures
to transferring maximum energy from photovoltaic panel PV to load. A DC/DC
converter is the interface that regulates the adaptation between the photovoltaic
panel PV and the load to ensure our load closer to the MPP.
As a result, several studies have focused on photovoltaic systems. They have tried
to develop algorithms to extract the maximum energy converted by the panel and
then allow an optimal operation of the system photovoltaic system (Zainudin and
Mekhilef 2010). Since the 1970s, a significant number of MPPT control techniques
have been developed, beginning with simple techniques such as MPPT controllers
based on the feedback of voltage and current (Salas et al. 2006), to more efficient
controllers using algorithms to calculate MPP of the PVG photovoltaic generator
(photovoltaic panel PV), among the most used technique (Incremental conductance
(INC)).
In recent years, more robust control techniques have been associated with the
MPPT control such as fuzzy logic controller (FLC) in order to increase the efficiency
of solar panels.
This chapter is ordered as follows. Section 3.2 describes the PV system modeling,
and demonstrates basic operation principle equations for the PV system and differ-
3 A Comprehensive Comparison of MPPT Techniques for Photovoltaic Systems 49

ent connexion with load. A definition and details of calculations to demonstrate the
operating principle of the converter DC/DC in its different state (continuous and
discontinuous) are given in Sect. 3.3. A brief description of the considered MPPT
(INC & FLC) system’s architecture is provided in Sect. 3.4. Simulation results for
the PV modeling and DC/DC Converter are addressed in Sect. 3.5. Analysis and
discussions are illustrated in Sect. 3.6. Finally, conclusions are given in Sect. 3.7.

3.2 PV Systems

A photovoltaic (PV) system directly converts sunlight into electricity. The basic
device of a PV system is the PV cell. Cells may be grouped to form panels or arrays
(Villalva et al. 2009). These panels prove to be a source of electrical energy that is
safe, reliable, maintenance-free and non-polluting. The majority of solar modules
on the market today are backed by guarantees of more than 20 years, and they work
well beyond that period.
Millions of systems have been installed worldwide, with different power ranging
from a fraction of a watt to several megawatts. For many applications, electric solar
systems are not only cost effective, but they can also be the least expensive option.
In this chapter, we will begin with the notion and characteristics of the sun and
then the operating principle of the photovoltaic cells, thus, we present the different
types of photovoltaic cells and their modeling. Finally, we will do a simulation for
a direct connection of the photovoltaic generator.

3.2.1 How a PV Cell Works

A photovoltaic cell is a semiconductor diode whose P − N junction is exposed

to light (Moller 1993; Sedra et al. 2006). The PV cell produces an effect, which
enables the solar energy to be converted directly into electrical energy by means of
a semiconductor material carrying the electrical charges. It is broken down into two
parts, the first one has an excess of electrons and the other an excess of so-called
doped holes, respectively N -type and P-type, as illustrated in Fig. 3.2. When the
radiation passes through the PV cell, the photons tear off electrons, which creates
free electrons and holes. The electrons accumulate in the N layer, while the holes
in the P layer. This gives rise to a potential difference and consequently the flow of
a current between the two layers of the cell. The P -doped zone plays the role of the
positive pole and the doped zone N plays the role of the negative pole. Subsequently,
the growth of the intensity of light will cause the increase of the emission of
photoelectrons in the PV material. An anti-reflective material generally covers PV
cells so that they can absorb as much solar radiation as possible. In practice, the
light absorbed by the solar cell will be the combination of direct solar radiation in
addition to the scattered rays radiated from the surrounding surface (Kanji 2012).
50 A. Ibnelouad et al.

Fig. 3.2 Structure of a cell

Photovoltaic cells are made of several types of semiconductors using different

manufacturing processes: Polycrystalline, Monocrystalline and Amorphous thin-
film silicon.
1. Polycrystalline silicon: Polycrystalline silicon consists of several single crystals
juxtaposed in different orientations giving the cell a mosaic appearance. It is the
most widely used technology in the world market because of its good efficiency
(about 15%) for controlled manufacturing costs.
2. Monocrystalline silicon: Monocrystalline silicon consists of a single crystal
providing the cell with a perfect arrangement of atoms. It has a slightly higher
yield than polycrystalline silicon (about 19%).
3. Amorphous thin-film silicon: The silicon is deposited in a thin layer on a glass
plate or other flexible support. The irregular organization of its atoms gives it in
part a poor semi-conduction.

3.2.2 Solar Panels

The PV cell is the basic unit of a solar panel, but it is intended to produce a small
power of a few watts. The need for high power causes the necessity of gathering
identical cells to form a module or a solar panel, in order to increase the power
generated.
Figure 3.3 shows a typical PV cell assembly. The series connection of several
solar cells makes it possible to easily increase the voltage, while the parallel
connection leads to a summation of the currents while maintaining the voltage.
Therefore, a series-parallel combination is used to obtain a PS with the desired
characteristics, that is to say a desired voltage and current.
3 A Comprehensive Comparison of MPPT Techniques for Photovoltaic Systems 51

Fig. 3.3 Combination of the

PV cell

Ipv
Anti return diode
Ipv1 Ipv2
Gpv2
Gpv1

Vpv
Chain A Chain A

18 Cells in series

Chain B Chain B

Diodes by-pass

Fig. 3.4 Combination of two PV modules with their protective diodes

3.2.3 Protection of a PV Generator

In order to ensure a long service life for PV generators that tends towards 25 years,
it is necessary to add some protections in order to avoid destructive failures resulting
from the association of the cells in series and modules in parallel. Often we
encounter the By-pass diodes and the anti-return diodes.
1. The anti-return diodes: Fig. 3.4 shows modules associated in series with anti-
return diodes. Under the effect of the sun as long as the voltage produced by
52 A. Ibnelouad et al.

Fig. 3.5 Equivalent model of + V

a GPV
Rs
I
V
Vd
+ I
Id Ip
Cell
= Isc Rp
– I

– I

the PV generator is higher than that of the battery, the battery charges. However,
when the darkness occurs, no voltage is produced from the PV generator; the
battery voltage would cause a current that would flow in the opposite direction
through the panels, which can lead to the destruction of the PV generator. Hence,
the use of the anti-return diodes will be beneficial to block the current flow and
protect the PV generator.
2. By-pass diodes: These diodes are shown in Fig. 3.4. The Bypass diode is
connected in parallel but in reverse polarity to a PV cell. In normal operation,
each solar cell is directly biased and therefore the Bypass diode will be reverse
biased and behaves as an open circuit. However, a cell is reverse biased due
to a short circuit between the cells, the By-pass diode conducts while allowing
the current to pass from the good solar cells to the external circuits. Then, the
maximum inverse polarization across the low cell is reduced to about a voltage
drop of a single diode, thereby limiting current, and attenuating hot spots.

3.2.4 Modeling of PV Devices

To be able to study and analyze the behavior of the cells during their operation,
modulation of the PV cells is necessary. A PV cell is equivalent to a current
generator to which we have added a diode in parallel, Fig. 3.5. The output of the
current generator is directly proportional to the light falling on the cell. The current
I from the cell is then written (Zagrouba et al. 2010).

3.2.4.1 Ideal PV Cell

Figure 3.6 shows the equivalent circuit of the ideal PV cell. The basic equation
from the theory of semiconductors (Rauschenbach 1980; Villalva et al. 2009) that
mathematically describes the I − V characteristic of the ideal PV cell is:
3 A Comprehensive Comparison of MPPT Techniques for Photovoltaic Systems 53

Practical PV device
Ideal PV Cell I

Rs V
Id

Ipv Rp

Fig. 3.6 Single-diode model pf the theoretical PV cell and equivalent circuit of a practical PV
device including the series and parallel resistances (Villalva et al. 2009)

Ipv Id I

– =

v v v

Fig. 3.7 Characteristic I − V curve of the PV cell. The net cell current I is composed of the
light-generated current Ipv and the diode current Id (Villalva et al. 2009)

   
qV
I = Ipv,cell − I0,cell exp −1 (3.1)
aT

where:
• Ipv,cell is the current generated by the incident light (it is directly proportional to
the Sun irradiation).
• Id is the Shockley diode equation.
• I0,cell is the reverse saturation or leakage current of the diode.
• q is the electron charge (1.60217646 ×10−19 C).
• k is the Boltzmann constant (1.3806503 ×10−23 J/K).
• T (in Kelvin) is the temperature of the P − N junction.
• a is the diode ideality constant.
Figure 3.7 shows the I − V curve originated from (3.1).

3.2.4.2 Modeling the PV Array

The basic equation (3.1) of the elementary PV cell does not represent the I − V
characteristic of a practical PV array. Practical arrays are composed of several
connected PV cells and the observation of the characteristics at the terminals of
the PV array requires the inclusion of additional parameters to the basic equation
54 A. Ibnelouad et al.

(Rauschenbach 1980; Villalva et al. 2009).

   
V + Rs I V + Rs I
I = Ipv − I0 exp −1 − (3.2)
aVt Rp

where:
• Ipv and I0 are the photovoltaic (PV) and saturation currents of the array,
respectively.
Ns kT
• Vt = is the thermal voltage of the array with Ns cells connected in series.
q
Cells connected in parallel increase the current and cells connected in series
provide greater output voltages. If the array is composed of parallel connections
of cells the PV and saturation currents may be expressed as: Ipv = Np Ipv,cell ,
I0 = Np I0,cell .
In (3.2), Rs is the equivalent series resistance of the array and Rp is the equivalent
parallel resistance. This equation originates the I − V curve in Fig. 3.8, where three
remarkable points are highlighted: short circuit (0, Isc ), MPP (Vmp , Imp ), and open
circuit (V0c , 0).
Equation (3.2) describes the single-diode model presented in Fig. 3.6. Some
authors have proposed more sophisticated models that present better accuracy and
serve for different purposes. For example, in Gow et al. (1999) and Hyvarinen
et al. (2003) an extra diode is used to represent the effect of the recombination
of carriers. A three-diode model is proposed in Nishioka et al. (2007) to include the
influence of effects that are not considered by the previous models. For simplicity,
the single diode model of Fig. 3.6 is studied in this paper. This model offers a
good compromise between simplicity and accuracy (Carrero et al. 2007), and has
been used by several authors in previous works, sometimes with simplifications
but always with the basic structure composed of a current source and a parallel
diode (Koutroulis et al. 2008; Xiao et al. 2004; Yi-Bo et al. 2008). The simplicity
of the single-diode model with the method for adjusting the parameters and the
improvements proposed in this paper make this model perfect for power electronics
designers who are looking for an easy and effective model for the simulation of PV
devices with power converters (Villalva et al. 2009).
All PV array datasheets bring basically the following information: the nominal
open-circuit voltage (V0c,n ), the nominal short-circuit current (Isc,n ), the voltage at
the MPP (Vmp ), the current at the MPP (Imp ), the open-circuit voltage/temperature
coefficient (Kv ), the short circuit current/temperature coefficient (KI ), and the
maximum experimental peak output power (Pmax,e ). This information is always
provided with reference to the nominal condition or standard test conditions
(STCs) of temperature and solar irradiation. Some manufacturers provide I − V
curves for several irradiation and temperature conditions. These curves make easier
the adjustment and the validation of the desired mathematical I − V equation.
Fundamentally, this is all the information one can get from datasheets of PV arrays
(Villalva et al. 2009).
3 A Comprehensive Comparison of MPPT Techniques for Photovoltaic Systems 55

Short circuit Maximum Power

V=0 V=mp
I=Isc I=Imp

I-V Curve
3,500 A

3,000 A

2,550 A
Current (A)

2,000 A
Open Voltage
1,500 A
V=Voc
1,000 A
I=0
0,500 A

0,000 A
0 2,5 5 7,5 10 12,5 15 17,5 20 22,5 25
Voltage (v)

Fig. 3.8 Characteristic I −V curve of a practical PV device and the three remarkable points: short
circuit (0, Isc ), MPP (Vmp , Imp ), and open circuit (Voc , 0)

Fig. 3.9 Principle of a direct Non return Diode

connection between a PV and
a load
PV DC
Array Load

3.2.5 Connection (PV Generator-Charge)

3.2.5.1 Direct Connection

When connecting directly a load with a PV array without passing through another
electrical device, it is in the case of a direct connection. In this case, the operating
point of the PV array depends on the impedance of the load to which it is connected
(Amarouayache 2014).
Figure 3.9 shows the principle of a direct connection between a PV and a
load. In general, the power extracted from a PV generator directly connected to an
application is often very far from the maximum power (MPP) that the PV generator
can deliver, as shown in Fig. 3.9.
As shown in Fig. 3.10, a PV generator can be connected directly to three types of
DC loads:
56 A. Ibnelouad et al.

Isc Resistive Load Current (A)

Power (W)
Imp Source of voltage

Power (watt)
current (A)

C
MPP

PC
Source of current
A
PA
B
PB
Vmp Voc

Voltage (V)

Fig. 3.10 Operating points of a GPV in direct connection, depending on the load

• Direct current source type load.

• DC source load.
• Purely resistive load.
Three respective operating points A, B, C can be identified providing power PA ,
PB and PC . These scenarios show that the GPV is badly exploited and does not
provide the maximum potential power. The difference being lost as dissipated heat
in the generator.

3.3 DC/DC Converter

DC/DC converters convert DC power to another DC level (voltage/current to

voltage/current). Their outputs are controlled by pulse width modulation (PWM)
of the gate signals. Switching frequency of the gate signal, in most cases, is
high and constant. In practice, their outputs are usually regulated by closed loop
control. Output DC voltage level of DC/DC converters can be stepped up, stepped
down, both stepped up and down, and inversed. In this project, we are specifically
interested in study of buck converter according to the project specification.

3.3.1 Topologies of DC/DC Converters

DC/DC converters can be classified as non-isolated DC/DC converters and isolated

DC/DC converters. Six most popular topologies of non-isolated DC/DC converters
(also called classical type DC/DC converters) are as below. All of them are single-
transistor converters. They are:
3 A Comprehensive Comparison of MPPT Techniques for Photovoltaic Systems 57

• Buck converter (step-down voltage only).

• Boost converter (step-up voltage only).
• Buck/Boost converter (inverse voltage, and step-up/step-down voltage).
• Cuk converter (inverse voltage, and step-up/step-down voltage).
• Zeta converter (step-up/step-down voltage).
• Sepic converter (step-up/step-down voltage).
There are four most popular topologies of isolated DC/DC converters. All of
them are able to step-up or step-down voltage. They are:
• Fly back converter (single-transistor).
• Forward converter (single-transistor).
• Single-ended two-transistor forward converter.
• Phase shift full-bridge converter.

3.3.2 Applications of DC/DC Converters

The DC-DC converter products are used extensively for diverse applications in the
healthcare (bio life science, dental, imaging, laboratory, medical), communications,
computing, storage, business systems, test and measurement, instrumentation, and
industrial equipment industries. They are used in electric motor drives, in Switch
mode DC power supplies etc.

3.3.3 Comparison of Montages

Table 3.1 summarizes the voltage gains and the stresses on the switches of the
different assemblies. For these converters, if more than one arrangement can be
considered as an elevator, especially if the duty cycle is greater than 0.5, only the
Boost system is lifted over the entire range of the duty cycle. For a cyclic ratio of 0.5,
for example, the Boost has a double output voltage of the input voltage. Whereas

Table 3.1 Characteristics of continuous and discontinuous conduction systems

Voltage Vkmax ikmax Current
Converter gain |Vdmax | |iDmax | source
1 Ve ΔVs Is αVe
Boost + + Continuous
1−α 1−α 2 1−α 2Lf
α(1 − α)Ve
Buck α Ve Is + Discontinuous
2Lf
α Ve ΔVs Is αVe
Buck-Boost + + Discontinuous
1−α 1−α 2 1−α 2Lf  
α Ve ΔVs + ΔVc Is αVe 1
Sepic + + + 1
Continuous
1−α 1−α 2 1−α 2f L1 L2
mα Vs mIs αVe
Fly-Back Ve + + Continuous
1−α m 1−α 2L1 f
58 A. Ibnelouad et al.

for the other risers the output voltage for this value of the duty cycle is equal to the
input voltage. It is only when the duty cycle is close to 1 that the other risers tend to
resemble the Boost circuit (Naffouti 2012).

3.3.4 DC-DC Adaptation Circuit of Type BOOST

In order to have a higher DC voltage at the output necessary to supply the various
loads directly or indirectly (through an inverter for example), a value which will
be mentioned hereafter according to the study to be carried out later, which
depends ontThe characteristic of the coupled load. Taking into account the economic
constraints, this brings us to the use of a voltage-boosting DC/DC converter, this is
the Boost chopper assembly (also called boost chopper or parallel chopper). This
type of static converter makes it possible to convert a DC voltage into another DC
voltage of higher value (Naffouti 2012).
In the following, we will study the operating of this structure in order to
determining the relationships between the different electrical signals.

3.3.4.1 Operation of the Boost Chopper

Presentation of the Structure

The schematic diagram of a Boost chopper is given in the Fig. 3.11. This structure is
composed mainly of an inductor L and two switches and D. Depending on the state
of these two switches, two operating phases can be distinguished (Reza Reisi et al.
2013):
• The active phase when the switch is closed and the switch D is open. During this
sequence, the current flowing through the inductor L will increase linearly and
an energy is stored in L. The capacitor C supplies energy to the load R.

+ VL –

IL iD I out
iin
+ – + –
L Diode
+
+ +
+ Vout
– Vin Switch C
PS

– – –
iC

Fig. 3.11 Diagram of a boost converter DC/DC

3 A Comprehensive Comparison of MPPT Techniques for Photovoltaic Systems 59

Fig. 3.12 Control signal of

the switch Tr

0 αT T t

• The freewheel phase when the switch is open, the switch D is closed. During this
sequence, the energy stored in the inductance L is restored to the capacitor and
to the charge R. During this phase, the fact that the inductance L is in series with
the source of input voltage makes it possible to obtain a boosting circuit.
Figure 3.12 shows the pattern of the control signal applied to switch Tr . It is a
rectangular signal of frequency f whose the duration is in the high state (duration
of conduction of the switch Tr , denoted TON ) is adjusted by the parameter α. This
parameter, called the duty cycle, Is defined as the ratio between the duration of
TON
conduction of the switch Tr and the cutting period T of the latter: α = , we
T
have: T = TON + TOF F , whither TOF F Corresponds to the duration of blocking of
the switch Tr .
The duration of conduction, Ton is between 0 and T therefore, the duty cycle is
between 0 and 1. The duration of conduction and of blocking of the switch Tr can
be expressed as a function of α and T :
• Duration of conduction: TON = αT
• Blocking time: TOF F = (1 − α)T
In the following study, we will make the following assumptions:
• The supply voltage is continuous and constant.
• The value of the capacitor C is large enough so that the output voltage can be
considered as continuous.
• The components are ideal.
Two conduction regimes can be distinguished:
• Continuous conduction, which corresponds to the case where the current through
the inductance never vanishes.
• The discontinuous conduction, which corresponds to the case where the current
passing through the inductance cancels before the next active phase.

Continuous Conduction

In the following, we explain the behavior of the structure as a function of these two
conduction regimes. The main objective is to determine the relationships between
the quantities electrical input and output of the converter as well as formulas for
dimensioning the various components (Naffouti 2012).
60 A. Ibnelouad et al.

VL iD
vD
iin
+ – – –
L iTr Diode iC
is
+ + +
+ Vin vTr R
– C V out
– – –

Fig. 3.13 Equivalent diagram of the Boost chopper during the active phase

VL iD vD
iin
+ –

L iTr Diode iC
is
+ + +
+ vTr
– Vin C R V out
– – –

Fig. 3.14 Equivalent diagram of the Boost chopper during the freewheeling phase

Sequence 1: Active Phase: 0 < t < αT

At the instant t = 0, the switch is closed for a period αT . The voltage across the
diode D is equal to VD = VT r − Vout . Since the switch Tr is closed, we have
VT r = 0, which implies that VD = −Vout . The diode is thus blocked since Vout > 0.
Under these conditions, we obtain the equivalent diagram of Fig. 3.13.
The voltage at the terminals of the inductance is then:
di
VL = Vin = L (3.3)
dt
By solving this differential equation, we obtain the following expression, which
expresses the evolution of the current flowing through the inductance:
Vin
iL = t + ILmin (3.4)
L

Sequence 2: Freewheel Phase; αT < t < T

At the instant t = αT , the switch is opened for a duration (1 − α)T . To ensure
the continuity of the current, the diode D goes into conduction. We then obtain the
equivalent diagram of Fig. 3.14.
3 A Comprehensive Comparison of MPPT Techniques for Photovoltaic Systems 61

The voltage at the terminals of the inductance is then:

diL
VL = Vin − Vout = L >0 (3.5)
dt
By solving this differential equation, we obtain the following formula, which
expresses the evolution of the current passing through the inductance:

Vin − Vout
iL = (t − αT ) + ILmax (3.6)
L

Expressions of Vout and Is

By definition:
 T  αT  T
1 1 1
VL  = VL dt = Vin dt + (Vin − Vout )dt (3.7)
T 0 T 0 T αT

Since the mean voltage at the terminals of a steady state, inductance is zero, we can
write:

VL  = αT Vin + (1 − α)(Vin − Vout ) = 0 (3.8)

Finally, we obtain the following relation:

Vin
Vout = vout (t) = (3.9)
1−α

The duty cycle α is between 0 and 1, so the output voltage Vs is necessarily

higher than the input voltage Vbat (boosting circuit).
If we assume that the input current is perfectly continuous, we can write:
 T
1
Is = is (t) = Iin dt (3.10)
T αT

which leads to:

Is = Iin (1 − α) (3.11)

This expression clearly shows that the Boost chopper is a step-down device.
With regard to these various expressions, it can be noted that the duty cycle
α makes it possible to adjust the average output voltage (or mean output
current) for a given average input voltage (respectively an average input
current). It is therefore possible to adjust the average power transfer between
the input and the output of the structure from the duty cycle α (Naffouti 2012).
62 A. Ibnelouad et al.

The average power transfer is:

P = p = (1 − α)Vout Iin (3.12)

Expression of ΔIL :
The absolute ripple of the current iL is defined by ΔIL = ILmax − ILmin . From the
Vin
above relations, at t = αT , we can write ILmax = αT + ILmin . We deduce the
L
αVin
following expression of ΔIL = . This expression shows that the current ripple
Lf
decreases when the switching frequency f increases the value of the inductor L. As
Vin = (1 − α)Vout , we can write:

α(1 − α)Vout
ΔIL = (3.13)
Lf

dΔIL
By solving = 0, it is found that the current ripple ΔIL is maximum for
dt
α = 2 . The dimensioning of the inductance L, from a given current ripple is carried
1

out using the following inequality:

Vout
L≥ (3.14)
4f ΔILmax

Voltage Ripple
To determine the expression of the voltage ripple ΔVout , it is assumed that the
dVout
current Is is perfectly constant. We have the following relation: ic = C , and
dt
for 0 < t < αT : we have ic = −Is , then the resolution of this differential equation
gives us:

1
Vout = − Is t + Voutmax (3.15)
C
For t = αT , we have:

1
Vout (αT ) = Voutmin = − Is αT + Voutmax (3.16)
C
Then, we have:

1
ΔVout = Voutmax − Voutmin = Is αT (3.17)
C
3 A Comprehensive Comparison of MPPT Techniques for Photovoltaic Systems 63

Finally:

αVout
ΔVout = (3.18)
RCf

This expression shows that the voltage ripple decreases when the switching
frequency f or the value of the capacitor C increases.
The dimensioning of the capacitor C, starting from a given ripple voltage, is
carried out using the following inequality:

αmax Vout
C> (3.19)
Rf ΔVout

Waveforms of the Main Signals

The waveforms of the main signals are given in Fig. 3.15. From these waveforms,
the mean and effective values of the currents flowing through the diode D and the
switch Tr can be expressed. We can also deduce the maximum stresses in voltage
and current from the switches. These relationships will be used when dimensioning
the various components of the structure (Naffouti 2012).
• Average current flowing through diode D:

ID = iD  = Is (3.20)

• Efficient current flowing through diode D:

 2 
Is ΔIL2
IDeff = (1 − α) + (3.21)
1−α 12

• Average current flowing through the switch Tr :

Is
IT r = iT r  = = Ipv (3.22)
1−α

• Efficient current flowing through switch T :

 2 
Is ΔIL2
IDeff = α + (3.23)
1−α 12

The voltage and current stresses on the controlled switch Tr and the diode D are
the same.
64 A. Ibnelouad et al.

Fig. 3.15 The appearance of the voltages and currents in continuous conduction mode

• Maximum voltage constraints

Vin ΔVout
VT rmax = |vDmax | = voutmax = + (3.24)
1−α 2

• Maximum current constraints

Is ΔIL
IT rmax = iDmax = iLmax = + (3.25)
1−α 2

The dimensioning of the switching cell takes place in the worst case. To calculate
the worst-case voltage and current constraints, we need to replace in the above
expressions α by αmax .
3 A Comprehensive Comparison of MPPT Techniques for Photovoltaic Systems 65

iL
T

vpv vpv–vs
L L

0 t

iD

0 aT bT T t

Fig. 3.16 Current flowing through the inductor in continuous conduction

Discontinuous Conduction

In discontinuous conduction, a phase is added during which the diode D does not
conduct.
From the curve of iL in Fig. 3.16, Δt can be deduced from the cancellation time
of this current: for αT < t < βT , Δt = (β − α)T :

Vin − Vout
iL (t) = (t − αT ) + ILαT (3.26)
T
and iL (βT ) = 0.
Vin Vin
In addition, with iL (αT ) = αT , we have: Δt = αT . The average
T Vin − Vout
output current is:
 βT
1 1
Is = is  = iD  = iL (t)dt = iL (αT )(β − α) (3.27)
T αT 2

which leads to:

2T
β 2 Vin
Is = (3.28)
2L(Vout − Vin )

or else:
 
2 Vin
Vout = Vin 1 + α (3.29)
2Lf Is
66 A. Ibnelouad et al.

In this operating mode, the output voltage is therefore load-dependent for a fixed
duty cycle. This implies that, in discontinuous conduction, it is necessary to
implement a control loop.

Power Available or Exchanged

 T 2
1 α2T Vout Vin
P = Vout is (t)dt = × (3.30)
T 0 2L Vout − Vin

Critical Conduction
This phase defines the limit between continuous and discontinuous regimes. For
αlimit , we have: Δt = (β − α)T = (1 − α)T that is to say βT = T .

Vin
Δt = αlim T = (1 − αlim )T (3.31)
Vout − Vin

which ultimately gives us:

Vout − Vin
αlim = (3.32)
Vout

3.4 Maximum Power Point Tracking (MPPT) Techniques

As mentioned before, the temperature and the irradiation depend on the atmospheric
conditions, which are not constant during the year and not even during a single day;
they can vary rapidly due to fast changing conditions such as clouds. This causes the
MPP to move constantly, depending on the irradiation and temperature conditions
in Fig. 3.17. If the operating point is not close to the MPP, great power losses occur.
Hence, it is essential to track the MPP in any conditions to assure that the maximum
available power is obtained from the PV panel. In a modern solar power converter,
this task is entrusted to the MPPT algorithms.
Maximum power point tracking (MPPT) aims to ensure that at any irradiation
or temperature, maximum achievable power is extracted from the PV system. This
is done by adjusting the duty cycle of the DC-DC converter, i.e. the converter’s
duty cycle is adjusted in a way that the operating point matches maximum point of
P-V curve. MPPT is a very important problem in PV systems, since extraction of
maximum achievable power from PV systems is of utmost importance. A MPPT
system directs the operating point of PV system toward maximum power point
(Rezaee Jordehi 2016). An efficient MPPT strategy must feature the following
properties:
3 A Comprehensive Comparison of MPPT Techniques for Photovoltaic Systems 67

20

18

16 I
sc
14 Imp
MPP
Current (I) Amps

12

ts
at
W
10

0
0 10 20 30 40 50 60 Vmp 70 80 Voc 90
Voltage (V) Volts
IV Curve:Current vs.Voltage Power Curve:Power vs.Voltage

Fig. 3.17 Maximum power point (MPP)

• It should provide high accuracy and be able to find true global maximum power
point (MPP). An accurate MPPT system results in a PV system with higher
efficiency.
• It should have high tracking speed. Slow tracking speed results in reduction of
extracted power and low efficiency of PV systems.
• It should be able to perform effectively both in uniform insolation conditions and
partial shading conditions. In partial shading conditions, there are several local
optima in P − V curve, therefore finding the true global maximum power point
is a challenging task.
• It should be system-independent, i.e. it should perform effectively for different
PV systems.
• It should not be too complex. Simplicity is a merit.
• It should not oscillate around maximum power point.
• It should be able to effectively track maximum power point after sudden drastic
changes in environmental conditions. Figure 3.17 illustrates the maximum power
point (MPP) of the solar panel computed by the MPPT controller.
Several MPPT algorithms have been developed, it differ in many aspects such
as required sensors, complexity, cost, range of effectiveness, convergence speed,
correct tracking when irradiation and/or temperature change, hardware needed for
the implementation or popularity, among others. The following algorithms are the
most common:
In this chapter, two MPPT techniques have been selected for purpose of
comparison; Incremental conductance (INC) and fuzzy logic controller (FLC).
68 A. Ibnelouad et al.

3.4.1 Incremental Conductance (INC) Technique

The Incremental Conductance (INC) algorithm is based to know the variation of

conductance of the photovoltaic generator, and the consequences on the position
of the operating point by comparison at the MPP. Thus, the conductance of the
photovoltaic module is defined by the relationship between the current and the
voltage of the GPV as indicated below:

Ipv
G= (3.33)
Vpv

Therefore, an elementary variation (increment) of conductance can be defined by:

∂Ipv
dG = (3.34)
∂Vpv

On the other hand, the GPV power evolution by voltage gives the position of the
operation point by to the MPP. When the power derivative is zero, it means MPP
has been reached, if positive the operating point is to the left of the maximum, when
negative, is on the right. Figure 3.18 shows the following conditions:
∂Ppv
• Left of MPP: >0
∂Vpv
∂Ppv
• Right of MPP: <0
∂Vpv
∂Ppv
• At MPP: =0
∂Vpv

dPpv
Ppv =0
dVpv

MPP

Left Right
dPpv
-< 0
dVpv
dPpv
>-0
dVpv

0
Vpv

Fig. 3.18 The operation of incremental conductance method

3 A Comprehensive Comparison of MPPT Techniques for Photovoltaic Systems 69

The relationship between the conductance given by (3.33) and the derivative of
∂Ppv
the power can be described by the following equation:
∂Vpv

∂Ppv ∂(Ipv Vpv ) ∂Ipv ΔIpv

= = Ipv + Vpv  Ipv + Vpv (3.35)
∂Vpv ∂Vpv ∂Vpv ΔVpv

Hence, we can write new conditions on the variation of conductance.

ΔIpv Ipv
• Left of MPP: >−
ΔVpv Vpv
ΔIpv Ipv
• Right of MPP: <−
ΔVpv Vpv
ΔIpv Ipv
• At MPP: =−
ΔVpv Vpv
The Maximum Power Point (MPP) can thus be tracked by comparing the
Ipv ΔIpv
instantaneous conductance to the incremental conductance (Abdelhak
Vpv ΔVpv
and Boubaker 2014), as illustrated by the algorithm of Fig. 3.19 (Abdelhak and
Boubaker 2014; Verma et al. 2016).

3.4.2 MPPT Controller with Fuzzy Logic

The objective is to present a reminder of the fuzzy sets and a general overview
on the fuzzy logic as well as its application for the optimization of a photovoltaic
system. We first introduce some notions of fuzzy logic; we will limit ourselves
to essential properties and then apply this optimization technique to maximize PV
system performance.
The output power of the photovoltaic modules is influenced by the intensity of
the solar radiation, the temperature and surface of the cells, as well as the charge.
Therefore, to maximize the efficiency of the renewable energy system, it is necessary
to continue the maximum power point (MPP) of the input source. We propose an
MPPT control system based on the theory of fuzzy sets to improve the efficiency of
PV energy conversion. The fuzzy algorithm based on the linguistic rules is applied
to the control of the DC-DC converter (boost converter) for the MPPT.

3.4.2.1 Comparison Between Fuzzy Logic and Boolean Logic

Fuzzy logic is widespread instead of Booleen logic, which cannot explain most
human reactions because the response cannot take only 0 or 1. On the other hand,
fuzzy logic can take all values between 0 and 1, following Functions belonging.
70 A. Ibnelouad et al.

Fig. 3.19 State-flow chart of INC MPPT technique

1
0.8
0.6
0.4
0.2

0 140 150 160 170 180 190 200

Price

Fig. 3.20 Square membership functions

Example

Figure 3.20 shows the price of a product: expensive or not expensive in classical
logic, we introduce a value “Threshold”: below this value, the product will not be
expensive and beyond it will be part of expensive. In fuzzy logic, we introduce
3 A Comprehensive Comparison of MPPT Techniques for Photovoltaic Systems 71

1 Degree of membership

0.8

0.6 Pricey state

0.4 Low pricey state

0.2

0
140 150 160 170 180 190 200

Fig. 3.21 Membership functions of price

Fuzzification Defuzzification
Interface Engine

Decision
Inputs E(k) Output
Making

Δ E(k)

Rule Base

Fig. 3.22 FLC block diagram

membership functions, which define, according to the parameter (price), the degrees
of belonging to each state. In this example, a product of 170 DH price will be
50% expensive and 50% cheap and another of 180 DH will be cheap to 20% and
expensive to 80%. Figure 3.21 shows membership functions of price.

3.4.2.2 Fuzzy Logic Controller (FLC) Technique for PV MPPT

Recently, fuzzy logic control (FLC) has been presented in many researches as in
(Eltamaly 2010; Eltamaly et al. 2010) to make PV work closer to MPP (Rezk and
Eltamaly 2015). In one hand, FLC offers the advantage of being robust and relatively
simple to develop and it does not require the exact knowledge of the regulate.
In the other hand, the FLC has the advantages of working with imprecise inputs,
not needing an accurate mathematical model, and handling nonlinearity. A FLC
generally consists of three stages: fuzzification, aggregation, and defuzzification
Fig. 3.22 (Seyedmahmoudian et al. 2016).
72 A. Ibnelouad et al.

NB NM NS ZE PS PM PB

Fig. 3.23 Membership function of FLC

Fuzzification

During fuzzification, numerical input variables are converted into a membership

function. The input values of this module are Ipv and Vpv . These values are used
to calculate the power from PV array. Then the error signal E(k) can be calculated
depending on (3.36). The value of ΔE(k) is calculated as shown in (3.37).

Ppv (k) − Ppv (k − 1)

E(k) = (3.36)
Vpv (k) − Vpv (k − 1)
ΔE(k) = E(k) − E(k − 1) (3.37)

Thus, these variables are expressed of linguistic variables such as P B (Positive Big),
P M (Positive Medium), P S (Positive Small), ZE (Zero), N S (Negative Small),
NM (Negative Medium), NB (Negative Big) using basic fuzzy subset. Each of
these acronyms is defined by mathematical membership functions, MF as shown in
Fig. 3.23 (Rezk and Eltamaly 2015).

Rule Inference

This stage is designed to control the output variables according to a calculation

and conversion to the linguistic variables based on MF between E and ΔE. The
change of duty cycle, ΔD of the power converter, which is the FLC output, can be
searched in a rule base given in Table 3.2 (Rezk and Eltamaly 2015). The strategy
of these rules is based on the “if-then” concept and necessitates information about
the systems. In this algorithm, the Mamdani Inference method uses the approach of
the max-min operation (Seyedmahmoudian et al. 2016).

Defuzzification

We must carry out the inverse operation of the fuzzification, here; we have to
calculate a numerical value comprehensible by the external environment from a
Fuzzy definition.
3 A Comprehensive Comparison of MPPT Techniques for Photovoltaic Systems 73

Table 3.2 Rules for a fuzzy e

system
Δe NB NS NM EZ PS PM PB
NB NB NB NB NB NM NS EZ
NM NB NB NB NM NS EZ PS
NS NB NB NM NS EZ PS PM
EZ NB NM NS EZ PS PM PB
PS NM NS EZ PS PM PB PB
PM NS EZ PS PM PB PB PB
PB EZ PS PM PB PB PB PB

3.5 Simulation of PV Modeling and DC/DC Converter

3.5.1 Simulation of PV Modeling

The photovoltaic system consists of four blocks as shown in Fig. 3.24. The first
block represents the energy source (photovoltaic panel), the second block is a static
DC-DC converter, the third block represents the load and the fourth block represents
the control system (MPPT). Radiation (R) is incident on the photovoltaic panel. It
generates a voltage (V ) and current (I ). The temperature of the PV solar is measured
at T (Seyedmahmoudian et al. 2016). The main role of the static converter is to
ensure impedance matching so that the photovoltaic panel PV delivers maximum
energy. For commanding the DC/DC converter, we have been carried out using
MPPT based on two techniques; INC and FLC (Abbes et al. 2014).

3.5.1.1 PV Solar Module

The PV solar module used in this study consists of 66 polycrystalline silicon solar
cells electrically configured as five series strings. Its main electrical specifications
are shown in Table 3.3.

Mathematical Models of the PV Panel

Mathematical models of the PV panel are defined below. Figure 3.25 shows the
equivalent circuit of a solar panel. A solar panel is composed of several photovoltaic
cells employing series or parallel or series-parallel external connections.
The following equation (3.38) describes the I − V characteristic of a solar panel
(Villalva et al. 2009).
   
V + Rs I V + Rs I
I = Ipv − I0 exp −1 − (3.38)
aVt Rp
74 A. Ibnelouad et al.

Fig. 3.24 Block diagram of a

general photovoltaic system
3 A Comprehensive Comparison of MPPT Techniques for Photovoltaic Systems 75

Table 3.3 Technical data of Maximum power (W) 305

the model manufacturer
Open circuit voltage Voc (V) 64.2
SUNPOWER SPR-305E
(T = 25 ◦ , G = 1000 W/m2 ) Short-circuit current Isc (A) 5.96
Current at maximum power point Imp(A) 5.58
Voltage at maximum power point Vmp (V) 54.7

I
+ –
ID Rs

+ +

Ipv Diode Rp V
–

Fig. 3.25 Equivalent circuit of solar module

where:
Ipv is the PV current.
I0 is the saturated reverse current.
a is a constant known as the diode ideality factor.
Rs and Rp are the series and parallel equivalent resistances of the solar panel
respectively;
Ns KT
Vt = is the thermal voltage associated with the cells.
q
Ns is the number of cells connected in series.
q is the charge of the electron.
K is the Boltzmann constant.
T is the absolute temperature of the p − n junction.
Ipv has a linear relationship with light intensity and varies with temperature
variations. I0 is dependent on temperature variations. Values of Ipv and I0 are
calculated from the following equations:
G
Ipv = (Ipvn + Ki ΔT ) (3.39)
Gn
Iscn + Ki ΔT
I0 =   (3.40)
V0cn + Kv ΔT
exp −1
aVt
where: Ipvn , V0cn and Iscn are the PV current, open circuit voltage and short circuit
current respectively under standard conditions (Tn = 25 ◦ C and Gn = 1000 W/m2 ).
Kv is the ratio of the open circuit voltage to temperature;
76 A. Ibnelouad et al.

Characteristic I=f (V) of the photovoltaic for different values of radiance

10
8
Current (A)

1 kW/m2
6
0.8 kW/m2
4 0.6 kW/m2
0.4 kW/m2
2 0.2 kW/m2

0
0 10 20 30 40 50 60 70
Voltage (V)

Fig. 3.26 Characteristic I = f (V ) of the photovoltaic system for different values of radiation

Characteristic P=f (V) of the photovoltaic for

× 104 differents values of radiation
16
14
12
1 kW/m2
Power (W)

10
0.8 kW/m2
8
0.6 kW/m2
6
0.4 kW/m2
4
0.25 kW/m2
2
0
0 50 100 150 200 250 300 350
Voltage (V)

Fig. 3.27 Characteristic P = f (V ) of the photovoltaic system for different values of radiation

Ki is the coefficient of short-circuit current variation with temperature;

ΔT = T − Tn is the deviation from standard temperature, G the light intensity
(Reza Reisi et al. 2013).

Influence of Irradiation on PV

(T = 25 ◦ C, G = 200, 400, 600, 800, 1000 W/m2 ) Figures 3.26 and 3.27 show
the characteristics of the PV solar module for different values of radiation. When
the radiation rate increases, Isc increases and Vco decreases slightly, therefore, the
maximum power undergoes a significant increase.

Influence of Temperature

Figures 3.28 and 3.29 illustrate the variations of the characteristic curves when there
is a change in temperature.
3 A Comprehensive Comparison of MPPT Techniques for Photovoltaic Systems 77

Fig. 3.28 Characteristic I (V ) of a module for different temperatures (at 1000 W/m2 )

Fig. 3.29 Characteristic P (V ) of a module for different temperatures (at 1000 W/m2 )

The above figures show the influence of temperature on the PV cell. We note that
when the ambient temperature increases, increases slightly and decreases according
to the temperature coefficient of the cell, consequently there is an increase of the
MPP in a negligible way.

3.5.2 Simulation of the BOOST Converter DC/DC

In the simulation, we examine the adaptation of a resistive load, which is fed by a

field of the panels, through a boost converter. The structure of the boosting circuit
in Matlab-Simulink environment is given in Fig. 3.30.

3.5.2.1 Simulation Result of Boost Converter

Figure 3.31 shows the simulation of the boosting circuit:

78 A. Ibnelouad et al.

Fig. 3.30 Simulation diagram of the boost converter

600
500
Voltage (V)

400
300
200 Ve
100 Vs
0
1 2 3 4 5 6
Time

Fig. 3.31 Simulation result of Boost converter

The Input voltage is Ve = 273 V and the output voltage is Vs = 500 V. After
simulation, this circuit shows us that the output voltage is higher than the input
voltage. So, this type give us the result that we looking for it.

3.6 Analysis and Discussion of the Simulation Results

3.6.1 The Simulation Results

3.6.1.1 Results with Temperature = 25 ◦ C and Irradiation = 1000 W/m2

Simulation results to compare two algorithms (INC and FLC) previously studied
are given in Figs. 3.32, 3.33 and 3.34. These figures show the output values of the
power, the voltage and the current for a pair of the temperature and the irradiation
equal to (25 ◦ C and 1000 W/m2 ).
3 A Comprehensive Comparison of MPPT Techniques for Photovoltaic Systems 79

Optimum PV system Power (w)

of two MPPT Techniques
120

100

80
Power

60

40

20 Incremental Conductance
Fuzzy Logic

0
0 2 4 6 8 10 12
Time × 104

Fig. 3.32 Optimum PV system power (W) of two MPPT techniques

Optimum PV system voltage (V)

of two MPPT Techniques
350

300

250
Voltage

200

150

100

50 Incremental Conductance
Fuzzy logic

0
0 2 4 6 8 10 12 14
Time × 105

Fig. 3.33 Optimum PV system voltage (V) of two MPPT techniques

Figures 3.32, 3.33 and 3.34 that are successively the responses of the power,
the voltage and the current of two algorithms for the photovoltaic system show
a convergence towards the optimal values with oscillations for the INC algorithm
while the FLC algorithm converges rapidly and without oscillation.
80 A. Ibnelouad et al.

Optimum PV system Current (A)

of two MPPT Techniques
400

300

200

100
Current

-100

-200 Incremental Conductance

Fuzzy Logic

-300
0 2 4 6 8 10 12 14
Time × 105

Fig. 3.34 Optimum PV system Current (A) of two MPPT techniques

Irradiation variation of Power

120

100

80

60
Power

40

20

Incremental
0 Conductance
Fuzzy Logic

-20
0 2 4 6 8 10 12 14 16
Time × 104

Fig. 3.35 Irradiation variation of the PV system optimal power of two MPPT techniques

3.6.1.2 Results with Temperature = 25 ◦ C and Different Values of

Irradiation

The simulation results of the algorithms previously studied are given in

Figs. 3.35, 3.36 and 3.37. These figures show the output values of the power,
the voltage and the current for a pair of the temperature equal to 25 ◦ C and different
values of irradiations.
3 A Comprehensive Comparison of MPPT Techniques for Photovoltaic Systems 81

Irradiation variation of the PV system optimal Voltage

of two MPPT Techniques
350

300

250

200
Voltage

150

100

50
Incremental
0 Conductance
Fuzzy Logic

-50
0 0.5 1 1.5 2 2.5
Time × 106

Fig. 3.36 Irradiation variation of the PV system optimal voltage of two MPPT techniques

Irradiation variation of PV System optimal

Current of two MPPT Techniques
400

300

200

100
Current

-100
Incremental
-200 Conductance
Fuzzy Logic

-300
0 0.5 1 1.5 2 2.5
Time × 106

Fig. 3.37 Irradiation variation of the PV system optimal current of two MPPT techniques

After the variation of radiations, the two algorithms undergo a great modification
that show in their responses of PV system, however, the algorithm INC remains still
a great oscillation at the maximum point while FLC no oscillations.
82 A. Ibnelouad et al.

3.6.2 Analysis and Discussion

T = 25 ◦ C and Ir = 1000 W/m2

The algorithm based on Incremental conductance (INC), gives fast response in

transient state but it works at the optimal point with large oscillations in steady
state.
The algorithm based on fuzzy logic is a robust and efficient algorithm. Indeed,
this algorithm works at the optimal point without oscillations. In addition, it is
characterized by good behavior in a transient state. However, the implementation
of this type of algorithm is more complex than the classical algorithms. Moreover,
the efficiency of this algorithm depends very much on the inference table.

T = 25 ◦ C and different values of irradiation

The same things when our Pv system has a variation irradiation, even if, each
algorithm in the beginning, shows a large movement due to different values of
irradiation, but always the FLC has more stable and better then the INC technique.
Since this gives more oscillation, that make our PV system far than the MPP,
therefore no energy has extract.

3.7 Conclusion

The work presented in this chapter focuses on the tracking of the maximum power
point MPP of a photovoltaic system. In this chapter, we have shown the importance
of photovoltaic systems, their properties, their operating, their modeling, their PV
characteristics and the variations of these characteristics due to the variation of
temperature and radiation, especially at point of maximum power.
In this work, we have studied the modeling of a photovoltaic chain, which
consists of a photovoltaic generator (with a single diode), an adaptation stage
(DC/DC converter Boost), a resistive load and a MPPT control stage to drive the
boost converter.
Different simulations, under the Matlab/Simulink environment, are carried out
to show the influence of the variations of the climatic conditions (illumination and
temperature) on the MPP where the operating power of the system is maximum.
The conventional or intelligent MPPT technique allows the best connection
between the photovoltaic generator (GPV) and the load and forces it to always
operate at the maximum power point, however, the comparative study of the
simulation results has found that Intelligent MPPT command FLC has better
performance (reduced response time and steady state error).
3 A Comprehensive Comparison of MPPT Techniques for Photovoltaic Systems 83

Therefore, the results of the simulations show in the both test that, FLC technique
has a good behavior in transient state and it presents a lack oscillation in steady state.
However, the INC presents a large oscillation in steady state. The algorithm, “INC”
is widely used, but fuzzy logic control FLC shows the best behavior and it is capable
to harvest the maximum energy from the PV system due to his membership’s
functions in comparison with INC technique.

References

Abbes, H., Abid, H., Loukil, K., Toumi, A., & Abid, M. (2014). Etude comparative de cinq
algorithmes de commande MPPT pour un système Photovoltaique. Revue des Energies
Renouvelables, 17(3), 435–445.
Abdelhak, B., & Boubaker, A. (2014). Contribution à l’optimisation d’une chaine de conversion
d’énergie Photovoltaique. Ph.D. thesis, Universite de Constantine.
Amarouayache, M. (2014). Contribution à l’optimisation d’une chaine de conversion d’énergie
Photovoltaique. Ph.D. thesis, Constantine 1 University.
Azab, M. (2008). A new maximum power point tracking for photovoltaic systems. World Academy
of Science, Engineering and Technology. International Journal of Electrical, Computer,
Energetic, Electronic and Communication Engineering, 2(8).
Carrero, C., Amador, J., & Arnaltes, S. (2007). A single procedure for helping PV designers to
select silicon PV module and evaluate the loss resistances. Renewable Energy, 32(15), 2579–
2589.
Eltamaly, A. M. (2010). Modeling of fuzzy logic controller for photovoltaic maximum power point
tracker. In Solar Future 2010 Conference Proceedings, Istanbul (pp. 4–9).
Eltamaly, A. M., Alolah, A. I., & Abdulghany, M. Y. (2010). Digital implementation of general
purpose fuzzy logic controller for photovoltaic maximum power point tracker. In Power
Electronics Electrical Drives Automation and Motion (SPEEDAM), International Symposium
on Digital Object Identifier (pp. 622–627).
Gow, J. A., & Manning, C. D. (1999). Development of a photovoltaic array model for use in power-
electronics simulation studies. IEE Proceedings Electric Power Applications, 146(2), 193–200.
Hyvarinen, J., & Karila, J. (2003). New analysis method for crystalline silicon cells. In Proceedings
of 3rd World Conference on Photovoltaic Energy Conversion (Vol. 2, pp. 1521–1524).
Kanji, B., (2012). Intelligent techniques for the tracking of the maximum power point of a
supervised photovoltaic system. MA thesis, Lebanese University.
Koutroulis, E., Kalaitzakis, K., & Tzitzilonis, V. (2008). Development of a FPGA-based system
for real-time simulation of photovoltaic modules. Microelectronics Journal, 40(7), 1094–1102.
Moller, H. J. (1993). Semiconductors for solar cells. Norwood: Artech House.
Naffouti, S. E. (2012). Dimensionnement et commande d’un hacheur parallèle alimenté par une
source Photovoltaique. M.S. thesis, University of Burgundy, France.
Nishioka, K., Sakitani, N., Uraoka, Y., & Fuyuki, T. (2007). Analysis of multicrystalline silicon
solar cells by modified 3-diode equivalent circuit model taking leakage current through
periphery into consideration. Solar Energy Materials and Solar Cells, 91(13), 1222–1227.
Rauschenbach, H. S. (1980). Solar cell array design handbook. New York: Van Nostrand Reinhold.
Rezaee Jordehi, A. (2016). Maximum power point tracking in photovoltaic (PV) systems: A review
of different approaches. Renewable and Sustainable Energy Reviews, 65, 1127–1138.
Rezk, H., & Eltamaly, A. M. (2015). A comprehensive comparison of different MPPT techniques
for photovoltaic systems. Solar Energy, 112, 1–11.
Reza Reisi, A., Moradi, M.H., & Jamasb, S. (2013). Classification and comparison of maximum
power point tracking techniques for photovoltaic system: A review. Renewable and Sustainable
Energy Reviews 19, 433–443.
84 A. Ibnelouad et al.

Salas, V., Olias, E., Barrado, A., & Lazaro, A. (2006). Review of the maximum power point
tracking algorithms for stand-alone photovoltaic systems. Solar Energy Materials and Solar
Cells, 90(11), 1555–1578.
Sedra, A. S., & Smith, K. C. (2006). Microelectronic circuits. London: Oxford University Press.
Seyedmahmoudian, M., Horan, B., Kok Soon, T., Rahmani, R., Thango, A. M., Mekhilef S.,
& Stojcevskiet, A. (2016). State of the art artificial intelligence-based MPPT techniques for
mitigating partial shading effects on PV systems-a review. Renewable and Sustainable Energy
Reviews, 64, 435–455.
Verma, D., Nema, S., Shandilya, A. M., & Dash, S. K. (2016). Maximum power point tracking
(MPPT) techniques: Recapitulation in solar photovoltaic systems. Renewable and Sustainable
Energy Reviews, 54, 1018–1034.
Villalva, M. G., Gazoli, J. R., & Ruppert Filho, E. (2009). Comprehensive approach to modeling
and simulation of photovoltaic arrays. IEEE Transactions on Power Electronics, 24(5), 1198–
1208.
Xiao, W., Dunford, W. G., & Capel, A. (2004). A novel modeling method for photovoltaic cells.
In Proceedings of IEEE 35th Annual Power Electronics Specialists Conference (PESC) (Vol. 3,
pp. 1950–1956).
Yi-Bo, W., Chun-Sheng, W., Hua, L., & Hong-Hua, X. (2008). Steady-state model and power flow
analysis of grid-connected photovoltaic power system. In IEEE International Conference on
Industrial Technology, ICIT 2008 (pp. 1–6).
Zainudin, H. N., & Mekhilef, S. (Dec 2010). Comparison study of maximum power point tracker
techniques for PV systems. In 14th Middle East Power Systems Conference, Mepcon’10, Cairo
University.
Zagroubaa, M., Bouaïchaa, M., Sellamia, A., Ksouric, M., et al. (2010). Optimisation par les
Algorithmes Genetiques et modélisation par la méthode LPV d’un systéme photovoltaique.
Vème Congrès Int. sur les Energies Renouvelables et l’Environnement, 4–6 Novembre, Sousse.
Chapter 4
Modeling and Comparison of Boost
Converter With Cascaded Boost
Converters

Nesrine Boujelben, Ferdaous Masmoudi, Mohamed Djemel, and Nabil Derbel

Abstract Photovoltaic (PV) energy is a very important renewable energy source.

The output voltage in renewable energy sources a is improved using DC-DC
converters, which are the key part in a photovoltaic chain. Among the classic DC-
DC topologies, the boost converter is the most used because of its simplicity and
high efficiency. The aim with this converter that the switching frequency is limited
so the output voltage is reduced. A possible solution to this problem is to use
other topologies; the quadratic boost converter results from the generalized cascaded
boost topologies with a single switch and the double cascade boost results from the
association of two identic elementary boost converters connected in tandem. In this
chapter a comparison of the efficiency of the classic boost converter with the two
cascaded boost converters is discussed.

Keywords Boost converter · Quadratic boost converter · Double cascade boost

converter · Efficiency

4.1 Introduction

For many years, fossil fuel cells have been the primary source of energy. The wide
use of these sources has leaded to the emission of green houses gases. Climate
change, caused by these greenhouse gases, has seriously damaged the environment.
Due to the limited supply of these sources on the earth and their environmental
effects, the world trend nowadays is to find a non-depletable and clean source of
energy.

N. Boujelben () · F. Masmoudi · M. Djemel · N. Derbel

Control and Energy Management Laboratory (CEMLab), Sfax Enginering School,
University of Sfax, Sfax, Tunisia
e-mail: [email protected]; [email protected];
[email protected]; [email protected]

© Springer Nature Singapore Pte Ltd. 2019 85

N. Derbel, Q. Zhu (eds.), Modeling, Identification and Control Methods
in Renewable Energy Systems, Green Energy and Technology,
https://doi.org/10.1007/978-981-13-1945-7_4
86 N. Boujelben et al.

Therefore, renewable energy sources such as Fuel cells, photovoltaic (PV) and
Wind etc, are experiencing a fast development. Unlike conventional energy sources,
such systems are environmentally friendly and produce clean electricity. Solar
energy is considered one of the most promising energy sources due to its infinite
power (Shen zt al. 2012; Suntio et al. 2010).
PV systems should be designed to operate at their maximum output power levels
for any temperature and solar irradiation level at all times.
The demand of PV generation systems pretends to be increased for both
standalone and grid-connected modes of PV systems.
The difference between the PV peak power voltage and the peak amplitude of
the grid voltage waveform requires high voltage gain. Both transformer-based and
transformer-less topologies can be proposed. The increase of the size and the cost of
the transformer make the transformer-less solutions a more attractive choice (Lopez-
Santos and Martinez-Salamero 2015; Lopez-Santos et al. 2013a).
Statics converters have the capability to step-up a low voltage to a high-voltage
level. Hence, different topologies of boost are used for the boost up action (Priya
et al. 2016; Veera Raghava and Ranga Rao 2014).
The development of new technologies is requiring wider conversion ratios. The
possibility to overcome this problem has been introduced over 40 years ago by
Matsuo el al., with their work on the cascaded connection of the conventional
converter topologies (Maksimovic and Cuk 1991).
The presented work deals with the comparison of the conventional boost
converter with two topologies of cascaded boost converters. The first one is the
quadratic boost converter with a single switch. The second topology named the
double cascade boost converter.
The remainder of this chapter is organized as follows. In Sect. 4.2, design and
modeling of the converters are given. Simulation results are developed in Sect. 4.3.
A discussion about the three converters is presented in Sect. 4.4. This chapter
concludes with some final remarks given in Sect. 4.5.

4.2 Converter Modeling

4.2.1 Boost Converter

A boost converter is used to convert a DC voltage from lower level to higher level.
The input to this converter can be from any DC source like solar panel, batteries etc
(Erickson and Maksimovic 1999; Park and Choi 2010; Selva Kumar et al. 2016).
The circuit diagram for the boost converter is shown in Fig. 4.1 where Vin is the
input voltage, S an active switch, D a passive switch, L and C are the corresponding
eliminate of inductance and capacitance of the converter.
As it is shown in Fig. 4.2, the boost converter has two possible structures in
continuous mode. In on-state, switch S is turned on. The inductor gets charged via
the supply voltage from the capacitor and stores the energy. During this state, the
inductor current increases gradually. In off-state, switch S is turned off. The energy
stored in the inductor changes its polarity to charge the capacitor through the diode.
4 Modeling and Comparison of Cascaded Boost Converter 87

L D
Vin S C R Vout

Fig. 4.1 Circuit diagram for the boost converter

D
L
Vin S C R Vout

L D
Vin S C R Vout

Fig. 4.2 Circuit configurations of the boost converter. (a) On-state; (b) Off-state

4.2.1.1 Dimensioning of the Circuit

The duty cycle is given by:

Vin
D =1− (4.1)
Vout

where Vin is the input voltage and Vout is the output voltage
The load R is calculated as follows:
2
Vout
R= (4.2)
Ppv

where Ppv is the input power

The output and the input currents are given by:

Vout
Iout = (4.3)
R
Iout
Iin = (4.4)
1−D
88 N. Boujelben et al.

The inductor value is designed using:

DVin
L= (4.5)
f ΔiL

where f is the switching frequency and ΔiL is the ripple current

The capacitor value is designed using:

DVout
C= (4.6)
f RΔvc

Where Δvc is the ripple voltage

The input power is equal to 50 w. The input voltage given to the circuit is 18 v
and it works on 40 kHz.

4.2.2 Quadratic Boost Converter

A quadratic boost converter results by assembling the components of two boost

converters by using single switch. This converter allows covering gain ranges which
are not available with the single boost converter (Lopez Santos et al. 2013b; Silveira
et al. 2014).
This converter is a fourth-order structure having two inductors and two capaci-
tors. The circuit diagram of a quadratic boost converter is shown in Fig. 4.3. This
converter comprises of an input voltage Vin , an active switch S, three passive
switches D1 , D2 and D3 , two capacitors C1 and C2 and two inductors L1 and L2 .
Thus, diodes D1 and D2 are replaced by voltage sources, and diode D3 and the
transistor switch by the corresponding current source (Lopez-Santos 2015; Selva
Kumar et al. 2016).
As it is shown in Fig. 4.4, the quadratic boost converter has two possible
structures in continuous mode. In on-state, switch S and diode D3 are turned on,
and diodes D2 and D1 are turned off. In off-state, switch S and diode D3 are turned
off, and diodes D2 and D1 are turned on.

D3

L1 D2 L2 D1
Vin C1 S C2 R Vout

Fig. 4.3 Circuit diagram for the quadratic boost converter

4 Modeling and Comparison of Cascaded Boost Converter 89

a
D3

L1 D2 L2 D1
Vin C1 S C2 R Vout

b
D3

L1 D2 L2 D1
Vin C1 S C2 R Vout

Fig. 4.4 Circuit configurations of the quadratic boost converter. (a) On-state; (b) Off-state

4.2.2.1 Dimensioning of the Circuit

The duty cycle is given by (Hauke 2014):

√
Vin
D =1− √ (4.7)
Vout

The output and the inductor currents are calculated as follows:

Vout
Iout = (4.8)
R
Iout
IL1 = (4.9)
(1 − D)2
Iout
IL2 = (4.10)
(1 − D)

The inductors values are given by:

DVin
L1 = (4.11)
f ΔiL1
DVin
L2 = (4.12)
f ΔiL2

The capacitors values are designed using:

DVin
C1,2 = (4.13)
(1 − D)Δvc Rf
90 N. Boujelben et al.

4.2.3 Double Cascade Boost Converter

The double cascade boost converter shown in Fig. 4.5 results from the association of
two identic elementary boosts connected in tandem. It consists of an input voltage
source Vin , two independently controlled switches S1 and S2 , two freewheeling
diodes D1 and D2 , two capacitors C1 and C2 and two inductors L1 and L2 (Chen
et al. 2014; Sira-Ramirez and Silva-Origoza 2006).
• Step 1: This step begins when the switch S1 is turned on. During this step, S2 will
remain off. The inductor L1 gets loaded through the supply voltage and stores the
energy. This step finishes when S1 is turned off.
• Step 2: During this period of operation, S1 is turned off and S2 is turned on. The
output of the first stage is now the input of the second stage. The inductor L2 gets
loaded through the supply voltage and stores the energy. This step finish when S2
Is turned off.
Figure 4.6 shows different structures of the double cascade boost converter in
continuous mode.

L1 D1 L2 D2
Vin S1 C1 V1 S2 C2 R Vout

Fig. 4.5 Circuit diagram for the double cascade boost converter

L1 D1 L2 D2
Vin S1 C1 V1 S2 C2 R Vout

L1 D1 L2 D2
Vin S1 C1 V1 S2 C2 R Vout

Fig. 4.6 Circuit configurations of the double cascade boost converter. (a) Step 1; (b) Step 2
4 Modeling and Comparison of Cascaded Boost Converter 91

4.2.3.1 Dimensioning of the Circuit

The voltage ratio is calculated as follows (Aamir and Shinwari 2010; Meddah et al.
2012):
Vin 1 1
= (4.14)
Vout (1 − D1 ) (1 − D2 )
where D1 is the duty cycle for the first boost and D2 is the duty cycle for the second
boost.
The output and the inductor currents are given by:
Vout
Iout = (4.15)
R
Iout
IL2 = (4.16)
(1 − D2 )
IL2
IL1 = (4.17)
(1 − D1 )
where IL2 is the output current of the first boost.
The inductors values are calculated as follows:
D1 Vin
L1 = (4.18)
f ΔiL1
D2 Vin
L2 = (4.19)
f ΔiL2
The capacitors values are designed using:
D1 V1
C1 = (4.20)
ΔVc1 f R
D2 Vout
C2 = (4.21)
ΔVc2 f R
where V1 is the output voltage for the first boost.

4.3 Simiulation Results

The three topologies are simulated using the SABER simulator.

4.3.1 Boost Converter

The simulation circuit diagram for the boost converter is shown in Fig. 4.7.
 92

tf rat pol 1
− Voltage to in2
Control b 1 s + b0
+ Interface a1 s + a0
Current to in1
v2var
Control
Interface
i2var
Current to
Control
1m mur1560 in1 tf rat pol 1
Interface
irf450 d i2var
2.35 mult b1 s + b0
60m a1 s + a0
Vdc =18V s 3200 in2
+ + v pulse + Voltage to
− − Control
Initial: 0 − Interface
Pulse: 10 v2var

Fig. 4.7 Simulation circuit diagram for boost converter

N. Boujelben et al.
4 Modeling and Comparison of Cascaded Boost Converter 93

Fig. 4.8 Waveforms of electrical signals. (a) The input; (b) The output of the boost converter

Simulation results are shown in Fig. 4.8.

As it shown in Fig. 4.8a, the input voltage is constant at the value of 18 V and
the inductor current is maintained constant at the value of 3 A after the transitional
regime. Figure 4.8b illustrates the waveform of the output voltage which achieve to
400 V and the waveform of the output current which is stabilized at 0.1 A.
94 N. Boujelben et al.

4.3.2 Quadratic Boost Converter

The simulation circuit diagram for the quadratic boost converter is shown in Fig. 4.9.
Simulation results are shown in Figs. 4.10 and 4.11.
The input voltage is kept constant at the value of 18 V as well as the input current
at the value of 3 A as it shown in Fig. 4.10a. Ripple in the input current is very less.
Figure 4.10.b illustrates the output voltage waveform as a constant at 400 V and
the output current at 0.1 A The waveform of the current IL2 is shown in Fig. 4.11.
Ripple in the inductor current is also very less.

4.3.3 Double Cascade Boost Converter

The simulation circuit diagram for the double cascade boost converter is shown in
Fig. 4.12.
Simulation results are shown in Fig. 4.13.
Figure 4.13a illustrates the waveform of the input voltage and the output
voltage where the values are 18 V and 3 A respectively. Figure 4.13b represents the
waveforms of electrical signals at the output of the converter, the voltage reaches
400 V and the current is kept constant at the value of 0.1 A. Figure 4.13c shows the
waveforms of the electrical signals at the output of the first stage.when the steady
state is achieved, the voltage is maintained constant at 100 V and the current IL2
attain 0.5 A. Ripple in the input current and in the inductor current IL2 are very less.
The value of the output voltage of the first boost is determined by evaluating the
efficiency of the converter with different value of the voltage V1 .
The result is shown in Fig. 4.14.
At the value 100 V, the converter gives the high efficiency 0.945.

4.4 Discussion

For the boost converter, Table 4.1 represents different values of duty ratio for
different values of input voltage. To satisfy the operating conditions, the duty ratio
must be very high for low values of input voltage. For example, with an input voltage
of 9 V, the structure is unable to assure 400 V at the output.
In the Table 4.2, different values of duty ratio are presented for different values
of input voltage for the quadratic boost converter. For low values of input voltage,
the structure is unable to satisfy the condition of operation.
For the double cascade boost converter, different values of duty ratio are
presented in Table 4.3 and this for different values of input Voltage. For low
values of input voltage, the structure is able to satisfy the operating conditions. For
 − Voltage to mult
in2 b 1 s + b0
Control
a1 s + a0
+ Interface
v2var in1 tf rat pol 1
Current to
640m Control
Interface Current to
i2var Control
Interface
i2var
stta2006p 3m
Current to
Control
stta2006p in1 tf rat pol 1
Interface
+ Voltage to stta2006p i2var b1 s + b0
Control mult
a1 s + a0
− Interface
in2
4 Modeling and Comparison of Cascaded Boost Converter

v2var irf450 + Voltage to

d Control
2.35 − Interface
10m 10m
Vdc =18V v2var
+ + s 3200
v pulse
− −
Initial: 0
Pulse: 10

Fig. 4.9 Simulation circuit diagram for the quadratic boost converter
95
96 N. Boujelben et al.

Fig. 4.10 Waveforms of electrical signals. (a) The input; (b) The output of the quadratic boost
converter

example, with a 9 V of input voltage, we get 400 V at the output with an efficiency
of the order of 0.86.
The performance study of each converter has been evaluated in two steps:
• Step 1: Evaluation of the efficiency by varying the input voltage with a fixed
output voltage
4 Modeling and Comparison of Cascaded Boost Converter 97

0.8

0.6

0.4
(A)

0.2

(A) : t(s)
0.0
IL2
–0.2
89.8m 89.85m 89.9m 89.95m 90.0m 90.05m 90.1m
t(s)

Fig. 4.11 Waveform of current at the inductor L2

• Step 2: Evaluation of the efficiency by varying the output voltage with a fixed
input voltage
The efficiency is calculated as follows:

Pout
eff = (4.22)
Ppv

where Pout is the output power and Ppv is the input power
Results are shown in Fig. 4.15.
Table 4.4 shows values of the efficiency of each converter at the nominal point
(18 V, 400 V).
 98

tf rat pol 1
in1
Current to
− Voltage to in2 b 1 s + b0
Control
Control Interface
a1 s + a0
+ Interface v2var
i2var + Voltage to Current to
v2var
Control Control
− Interface Interface
i2var
Current to
Control
0.67m mur1510 18m mur1560 in1 tf rat pol 1
Interface
irf150 d irf450 d i2var
4.7 2.35 b1 s + b 0
Vdc =18V 12m mult
a1 s + a0
46m
+ + s + s + Voltage to in2
v pulse v pulse 3200
− − − Control
Initial: 0 Initial: 0 − Interface
Pulse: 10 Pulse: 10 v2var

Fig. 4.12 Simulation circuit diagram for the double cascade boost converter
N. Boujelben et al.
4 Modeling and Comparison of Cascaded Boost Converter 99

From Table 4.4 and Fig. 4.15, we remark that the double cascade boost converter
has the best yield matched to the conventional boost converter and the quadratic
boost converter. To have an idea about the losses of each converter, a study of the
losses is evaluated by varying the input power. Figure 4.16 represents the result of
this study.
It’s clear from Fig. 4.16 that for different values of the input power, the double
cascade boost converter has the minimum losses compared to the other converters.

4.5 Conclusion

Renewable energy sources are experiencing a fast development. Solar energy seems
to be most attractive in present days. PV-systems offer a wide range of possibilities
and configurations for the use of power electronic converters. The need to have
converters with wider conversion rates has brought us to work with DC-DC cascade
converters. Various converter topologies have been proposed in the literature, to
improve performance and adapt to requirements.
In this chapter a comparison of three topologies of DC/DC converters has been
presented. A complete modeling and a dimensioning of each topology have been
evaluated. Simulation results for these converters have been discussed. An idea
about the waveforms of electrical signals for the three topologies has been delivered.
The dynamic behavior analysis shows that the double cascade boost has the best

Fig. 4.13 Waveforms of electrical signals. (a) The input; (b) The output of the double cascade
boost; (c) The output of the first boost
100 N. Boujelben et al.

Fig. 4.13 (continued)

4 Modeling and Comparison of Cascaded Boost Converter 101

95

94,5
efficiency (%)

94

93,5

93
40 60 80 100 120 140
voltage V1

Fig. 4.14 Efficiency of the double cascade boost converter according to the voltage V1

Table 4.1 Duty ratio according to input voltage for the boost converter
Input voltage (V) Output voltage (V) Duty ratio Efficiency %
9 X X X
18 400 0.955 85.32
48 420 0.88 96.98

Table 4.2 Duty ratio according to input voltage for the quadratic boost converter
Input voltage (V) Output voltage (V) Duty ratio Efficiency %
9 X X X
18 400 0.787 85.37
48 406 0.653 93.2

Table 4.3 Duty ratio according to input voltage for the double cascade boost converter
Input Intermediary Output Duty Duty Efficiency
voltage (V) voltage V1 (V) voltage (V) Ratio D1 Ratio D2 %
9 100 400 0.91 0.75 85.7
18 100 400 0.82 0.75 94.54
48 100 420 0.52 0.75 96.92

response as the result of its ability to boost up the voltage even with a low input
values. The efficiency analysis reveals that the best topology is the double cascade
boost converter. The obtained results of the losses study show that the double
cascade boost converter has the minimum losses. However, the operation with
a single active switch such as the quadratic boost converter is a very attractive
although its efficiency can be slightly lower.
102 N. Boujelben et al.

a b
100
96
95 94
90 boost 92 boost

90
efficiency (%)

85

efficiency (%)
88
80
quadratic 86 quadratic
75 boost boost
84
70
82
65 double double
cascade 80
cascade
60 boost 78 boost
6 12 18 24 30 36 42 48 275 325 375 425
input voltage (v) output voltage (v)

Fig. 4.15 Efficiency for the three converters according to. (a) The input voltage; (b) The output
voltage

Table 4.4 Efficiency of different converters at the nominal point

Type of converter Boost Quadratic boost Double cascade boost
Efficiency % 85.32 85.37 94.54

15
boost

13 quadratic boost
double cascade boost

11
losses (watt)

1
20 30 40 50 60 70 80 90 100 110 120 130 140 150
input power (watt)

Fig. 4.16 losses of the three converters according to the input power
4 Modeling and Comparison of Cascaded Boost Converter 103

References

Aamir, M., & Shinwari, M. Y. (2010). Design, implementation and experimental analysis of
two-stage boost converter for grid connected photovoltaic system. In 3rd IEEE International
Conference on Computer Science & Information Technology, 5, 194–199.
Chen, Z., Yong, W., & Gao, W. (2014). PI and Sliding Mode Control of a Multi-Input-Multi-Output
Boost-Boost Converter. WSEAS Transactions on Power Systems, 9, 87–102.
Erickson, R. W., & Maksimovic, D. (1999). Fundamentals of power electronics. Boston: Kluwer
Academic.
Hauke, B. (2014). Basic calculation of a boost converter’s power stage. Application Report
SLVA372C, Texas Instruments.
Lopez-Santos, O. (2015). Contribution to the DC-AC conversion in photovoltaic systems: Module
oriented converters (pp. 99). Ph.D. thesis. LAAS-Toulouse, France.
Lopez-Santos, O., & Martinez-Salamero, L. (2015). Robust sliding-mode control design for a
voltage regulated quadratic boost converter. IEEE Transaction on Power Electronics, 30(4),
2313–2327.
Lopez-Santos, O., Martinez-Salamero, L., Garcia, G., Valderrama-Blavi, H., & Sierra-Polanco, T.
(2013a). Comparison of quadratic boost topologies operating under sliding-mode control. In
2013 Brazilian Power Electronics Conference, Gramado, Brazil.
Lopez Santos, O., Martinez-Salamero, L., Garcia, G., Valderrama-Blavi, H., & Mercuri, D. O.
(2013b). Efficiency analysis of a sliding-mode controlled quadratic boost converter. IET Power
Electronics, 6, 364–373.
Maksimovic, D., & Cuk, S. (1991). Switching converters with wide DC conversion range. IEEE
Transaction on Power Electronics, 6(1), 151–157.
Meddah, M., Bourahla, M., & Bouchetata, N. (2011). Synthèse des convertisseurs statiques DC/AC
pour les systèmes photovoltaïques. Revue des Energies Renouvelables ICESD’11 Adrar (pp.
101–112).
Park, S., & Choi, S. (2010). Soft-switched CCM boost converters with high voltage gain for high-
power applications. IEEE Transaction on Power Electronics, 25(5), 1211–1217.
Priya, P., Shabbeer Basha, G., Sujith Niranjan, S. V., & Seyezha, R. Dr. (2016). Investigation of sic
mosfet based quadratic boost converter for photovoltaic applications. International Journal of
Precious Engineering Research & Applications, 1(3), 26–29.
Selva Kumar, R., Gayathri Deivanayaki, V. P., Vignesh, C. J., & Naveena, P. (2016). Design
and comparison of quadratic boost converter with boost converter. International Journal of
Engineering Research & Technology, 5(1), 878–881.
Shen, J.-M., Jou, H.-L., & Wu, J.-C. (2012). Novel transformerless grid-connected power converter
with negative grounding for photovoltaic generation system. IEEE Transaction on Power
Electronics, 27(4), 1818–1829.
Silveira, G. C., Lessa Tofoli, F., Santos Bezerra, L. D., & Torrico-Bascope, R. P. (2014). A
nonisolated DC-DC boost converter with high voltage gain and balanced output voltage. IEEE
Transaction on Industrial Electronics, 61(12), 6739–6746.
Sira-Ramirez H., & Silva-Origoza, R. (2006). Control design techniques in power electronics
devices. New Mexico City: Springer.
Suntio, T., Leppaaho, J., Huusari, J., & Nousiainen, L. (2010). Issues on solar-generator interfacing
with current-fed MPP-tracking converters. IEEE Transaction on Power Electronics, 25(9),
2409–2419.
Veera Raghava, J. H. V., & Ranga Rao, C. (2014). A new converter topology for grid connected
PV application. International Journal of Engineering & Science Research, 4(6), 256–269.
Chapter 5
Control of Power of a DFIG Generator
with MPPT Technique for Wind Turbines
Variable Speed

Yasmine Ihedrane, Chakib El Bekkali, Badre Bossoufi,

and Manale Bouderbala

Abstract This chapter presents the study and the modeling of a variable speed wind
system based on Doubly Fed Induction Generator controlled by a linear control type
PI, in order to independently control the active and reactive stator powers generated
by the wind turbine system.
The proposed control is applied to a Doubly Fed Induction Generator whose
stator is directly connected to the grid in contrast to the rotor which is connected via
“Back-to-Back” converters.
The objective of the modeling is to apply the direct and indirect control of the
active and reactive power generated by the wind turbine based on the Doubly Fed
Induction Generator via the Maximum Power Point Tracking (MPPT) strategy, to
extract the maximum power from the wind.
Simulation results are tested and compared in order to evaluate the performance
of the proposed system.

Keywords Doubly fed induction generator (DFIG) · Wind turbine · Direct field
oriented control (DFOC) · Indirect field oriented control (IFOC) · Maximum
power point tracking (MPPT)

5.1 Introduction

Today, energy demands has increased enormously. Fossil fuel sources are declining
and the preoccupations concern the levels of pollution in the environment are the
main drivers for electricity generation from renewable energy sources.

Y. Ihedrane () · C. E. Bekkali

Laboratory of STIC, Faculty of Sciences Dhar El Mahraz, Sidi Mohamed Ben Abdellah
University, Fez, Morocco
B. Bossoufi · M. Bouderbala
Laboratory of Electrical Engineering and Maintenance, Higher School of Technology,
EST-Oujda, University of Mohammed I, Oujda, Morocco

© Springer Nature Singapore Pte Ltd. 2019 105

N. Derbel, Q. Zhu (eds.), Modeling, Identification and Control Methods
in Renewable Energy Systems, Green Energy and Technology,
https://doi.org/10.1007/978-981-13-1945-7_5
106 Y. Ihedrane et al.

Fig. 5.1 Wind conversion

chain

Renewable energies, such as solar energy and wind energy, are clean, inex-
haustible energies, because of all these factors there has been a great deal of interest
in wind power generation in recent years (Kazemi and Kojabadi 2010).
The induction machine in variable speed drives poses some difficulties which
make its control more complex compared to a DC machine. Indeed, this type of
machine is characterized by a multivariable, nonlinear mathematical model, with
a strong coupling between the two control variables, which are the magnetic Field
and the electromagnetic torque (de Wit 2000). In other words, it is not possible
to control independently the input variables (voltage or current). Moreover, some
state variables are not accessible to measurements, and its parameters are subject to
variations over time. For this reason, several methods of controlling the DFIG have
emerged, among them, the field oriented control technique.
The principle of this technique was developed by BLASCHKE in the early
1970s; it consists in orienting the field along one of the axes of the referential d,
q in order to make this machine’s behavior similar to that of a Separately Excited
DC Machine. This control is based on PI controllers.
In this chapter, we begin with the modeling of the turbine. Next, a tracking
technique operating point at maximum power point tracking (MPPT) will be
presented (Bekakra and Attous 2011; Ghoudelbourk et al. 2016). Then, we will
present a model of the DFIG in the referential (d, q). After that, we will tackle the
principle of field oriented control. Finally with the use of MATLAB/SIMULINK,
we will present and analyze the simulation results to validate our theoretical
study.

5.2 Modeling of the Wind Conversion Chain

The turbine studied in this chapter comprises three blades of length R. They are
fixed on a drive shaft rotating at a turbine speed Ωt connected to a gain multiplier G
which drives an electric generator as shown in Fig. 5.1 (Ihedrane et al. 2017a).
5 DFIG Generator with MPPT Technique for Wind Turbines 107

5.2.1 Model of the Turbine

5.2.1.1 Turbine Power

The kinetic power of the wind is defined by the following equation (Ihedrane et al.
2017a,b)

ρ ×S ×V3
Pv = (5.1)
2
where:
ρ: Density of the air
S: Surface swept by the turbine
v: Wind Speed
The turbine can convert only a percentage of the wind power presented by the
coefficient Cp , hence the expression of the aerodynamic power can be described as
follows:

ρ ×S ×V3
Paero = Cp (λ, β) × (5.2)
2

5.2.1.2 Power Coefficient

The power coefficient Cp (λ, β) represents the aerodynamic efficiency of the wind
turbine. The latter depends on the specific speed λ and the angle of the orientation
of the blades β. It is approximated by the following function (Bossoufi et al. 2014;
Ihedrane et al. 2017a):
   
C2 C5
Cp (λ, β) = C1 × − C3 × β − C4 × exp − + C6 × λ (5.3)
A A

1 1 0.035
where: = −
A (λ + 0.08 × β) β 3 + 1

5.2.1.3 Specific Speed

Specific speed λ is defined as the ratio of the linear turbine speed and the wind speed
(Ihedrane et al. 2017a; Youcef 2014), it is given by (5.4):

Ωt
λ=R× (5.4)
V
108 Y. Ihedrane et al.

0.6
Cp(l, b)
0.4
b = 2°
0.2
4
0
6
–0.2
8
–0.4
b = 10°
–0.6

–0.8
b = 0°
–1

–1.2
λ
–1.4
0 2 4 6 8 10 12 14 16 18 20

Fig. 5.2 Power coefficient CP in function of speed ratio λfor different values of β

Figure 5.2 shows the variation of power coefficient CP in function of speed ratio
λ for different values of β (Bossoufi et al. 2014; Ihedrane et al. 2017a). From this
figure we can note that the power coefficient Cp reaches its maximum 0.5506 for a
speed ratio λopt = 8 and β = 0◦ .

5.2.1.4 Mechanical Torque

From the expression of the power produced by the turbine and knowing the speed
turbine, the mechanical torque can be expressed by 5.5 (Ihedrane et al. 2017a):

Paero ρ ×S ×V3
Caero = = Cp (λ, β) × (5.5)
Ωt 2 × Ωt

5.2.2 Model of the Multiplier

The multiplier transforms the turbine speed into the generator speed and the
aerodynamic torque into the generator torque according to the following system
(Ihedrane et al. 2017a; Youcef 2014):
5 DFIG Generator with MPPT Technique for Wind Turbines 109

⎧
⎪ Ω
⎨Ωt = mec
G (5.6)
⎪ C
⎩Cg = aero
G
where:
Ωmec : rotation speed of the generator.
Cg : The generator torque.

5.2.3 Dynamic Equation of the Transmission Shaft

From the mechanical torque exerted on the rotor shaft of the wind turbine and the
electromagnetic torque Cem , the fundamental equation of the dynamics makes it
possible to determine the evolution of the mechanical speed (Guda 2005):

dΩt
Cmec = J = Cg − Cem − Cf (5.7)
dt
with:
J : the total inertia, it is constituted by the turbine inertia Jt , and the generator
inertia Jg (Aimani 2004):

Jt
J = + Jg (5.8)
G
The block diagram corresponding to the model of the turbine is given in Fig. 5.3.

Fig. 5.3 Block diagram of the wind converter model

110 Y. Ihedrane et al.

5.2.4 Control Strategy for Wind Turbine

5.2.4.1 Maximum Power Point Search Method

The characteristic of the optimal power of a wind turbine is strongly non-linear and
“bell-shaped” (Johansen et al. 2001).
The wind system must find the maximum power, for each wind speed, which is
equivalent to finding the optimum rotational speed.
Figure 5.4 illustrates the characteristic curves of the wind turbine in the plane
(power,rotational turbine speed). Each curve corresponds to a wind speed Vv . The
vertices of these characteristics present the optimal points sought which define a
curve called the optimal power curve given by:

ρ × π × R2 × V 3
Popt = Cp opt(λopt ) × (5.9)
2
The wind system requires a perfect tracking of the optimum power curve in order to
have an ideal operation, to do that, a specific command called Maximum Power
Point Tracking (MPPT) must be used. The strategy of this control consists in
controlling the electromagnetic torque in order to adjust the mechanical speed in
a way to maximize the electrical power generated.
There are two approaches:
• The first approach considers that the characteristic Cp = f (λ) is not known
(Johansen et al. 2001).
• The second one considers that the characteristic Cp = f (λ) is known. Simply
follow the optimum power curve for the wind turbine to be in optimum
conditions.

P(W)
Popt =f(Ωopt)

u5

u4

u3

u2
W (rad/s)
u1

Fig. 5.4 Characteristics of the wind turbine in the plane (power,rotational turbine speed)
5 DFIG Generator with MPPT Technique for Wind Turbines 111

P(w)

Zone 1 Zone 2 Zone 3 Zone 4

Wind Speed
umin unom umax

Fig. 5.5 Characteristics (Power, wind speed)

This approach Simplifies the algorithm of the maximum power and allows to use
more basic and less expensive converters.

5.2.4.2 Operation of a Variable Speed Wind Turbine

Figure 5.5 illustrates the different operating phases of a variable speed wind turbine
(Mourad 2016).
• Wind speeds are very low and insufficient to drive the wind turbine and produce
power.
• the wedge angle is kept constant, and the control of the electromagnetic torque
will be implemented in order to capture the maximum power for each wind
speed (MPPT principle). In this zone, the generator power curve keeps a rapid
progression (Mourad 2016).
• Zone 3: The generator speed is kept constant at its maximum as opposed to
a suitable torque. The increase in the wind speed leads to a decrease in the
coefficient Cp and a slower increase in the recovered power. When the maximum
of the power generator is reached, the angle of the blades (pitch) is modified
(Passage from β1 to β2 ) in order to degrade the coefficient Cp .
• Zone 4: In this zone, when the wind speed becomes too high VM , an emergency
device is used to stop the wind turbine (No electricity production) and to prevent
damage.

5.2.4.3 Maximum Power Extraction Technique

In order to extract the maximum power from the wind turbine, we need an algorithm
acting on the set point variables, to have a good efficiency of the device.
112 Y. Ihedrane et al.

In the literature we have found two types of control structures for the maximiza-
tion of power extracted:
• The MPPT control without mechanical speed control.
• The MPPT control with mechanical speed control
In this chapter we are interested in the MPPT control without mechanical speed
control.
In practice, an exact measurement of wind speed is difficult to achieve. Due to
the following reasons (Aimani 2004; Mourad 2016).
• The anemometer is located behind the rotor of the turbine, which gives an
erroneous reading of the wind speed.
• Since the diameter of the surface swept by the blades is large (typically 70 m
for a 1.5 m wind turbine), a significant variation in wind appears depending on
the height of the anemometer. So, the use of a single anemometer leads to use
only one local measurement of the wind speed which is therefore not sufficiently
representative of its average value appearing on all the blades.
An incorrect measurement of the speed therefore necessarily leads to a degradation
of the power captured according to the power extraction technique. This is why most
wind turbines are controlled without speed control. This control structure is based
on the assumption that the wind speed varies very steadily (Gaillard 2010):

dΩmec
J = Cg − Cem − f × Ωmec = 0 (5.10)
dt
According to this equation, and by neglecting the mechanical torque Cmec and the
effect of the couple of viscous friction f × Ωmec , we obtain the following equation:

Cem = Cg (5.11)

The reference electromagnetic torque is determined from an estimation of the

aerodynamic torque:

Caeroref
Cemref = (5.12)
G
We know that:
⎧
⎪
⎪ ρ × π × R2 × V 3
⎪
⎪ C aero = C p (λ, β)
⎪
⎨ 2 × Ωt
Ωmec
⎪ Ω t = (5.13)
⎪
⎪ G
⎪
⎪ Ωtest
⎩Vest = R
2 × λtest
5 DFIG Generator with MPPT Technique for Wind Turbines 113

Fig. 5.6 Block diagram of the MPPT strategy without speed control

After calculation and a series of combinations, we get the following equation:

Cp (λ, β) ρ × π × R3 Ω3
Cemref = × × mec (5.14)
λ3opt 2 G3

The block diagram of the MPPT strategy without mechanical speed measurement is
shown in Fig. 5.6.

5.3 Modeling of the Doubly Fed Induction Generator DFIG

5.3.1 Structure of the Electrical Machine

The wind turbine conversion system studied in this chapter is shown in Fig. 5.7. It
is divided into two main parts which will be modeled separately: the Doubly Fed
Induction Generator whose stator is directly connected to the grid, in contrast to
the rotor which is connected to the grid via powers converters “Back to Back”,
114 Y. Ihedrane et al.

Fig. 5.7 General structure of a wind turbine conversion chain

these converters being able to operate as a rectifier or an inverter according to the

direction of the energy transit.
This structure has been retained in this work, in order to establish a control
strategies with the objective of controlling the power exchange between the
generator and the electrical grid.
The stator magnitudes have the same frequency as that of the grid, that is why it
is useless to use power converters to transfer the stator power. On the other hand,
the rotor magnitudes don’t have the same frequency as the alternating magnitudes of
the grid, which depend on the speed of the rotor and consequently the wind speed,
which justifies the use of a voltage converter feeding the rotor.
In the context of the production of electrical energy from wind power, this
device can operate at fixed or variable speed, operate in hypo-synchronous or hyper-
synchronous mode.
The powers converter side generator allows the control of the active and reactive
power produced by the DFIG, and the powers converter side grid allows the control
of the DC bus voltage and the power factor at the connection of the wind turbine
with the power grid (Datta 2002).
The use of a doubly Fed Induction Generator presents an excellent compromise
between the range of speed variation and the dimensioning of the converters
compared to the nominal power of the machine.
5 DFIG Generator with MPPT Technique for Wind Turbines 115

5.3.2 Operation of the Doubly Fed Induction Machine in

Generator Mode

The Doubly Fed Induction Machine, unlike other electrical machines, is reversible
and can operate in generator mode and motor mode.
Figure 5.8 shows the form of the mechanical torque/speed characteristic of the
induction machine (Mourad 2016).
From Fig. 5.8 it can be seen that the induction machine operates in motor mode
for g > 0 and in generator mode in the opposite case.
For operation with Doubly fed Induction Machine, the stator is powered by a
first balanced three-phase voltage source of frequency f , the rotor is connected to a
second alternating frequency source fr .
The currents flowing in the stator windings create rotating field at the speed Ωs =
Ωs Ωr
, similarly, the rotor currents produce a rotor field rotating at a speed Ωr = .
P P
When the rotating field of the ror rotates in the same direction of rotation of the
stator field, the slip ‘g’ is positive and the rotor rotates slower than the stator field
Ω < Ωs , so, the machine operates in Hypo-Synchronous mode.
In the opposite case, the rotating field created by the windings of the rotor rotates
in the opposite direction than that of the stator, the slip ‘g’ becomes negative and
the rotor rotates faster than the rotating field created by the stator Ω > Ωs , so, the
machine operates in hyper-synchronous mode.

g>0 g<0
Motor mode Generator mode

80

60
Electromagnetic Torque (N.m)

40

20

–20

–40

–60

–80
0 ns 2.ns
Rotational speed (tr/min)

Fig. 5.8 Torque/speed characteristic of the induction machine

116 Y. Ihedrane et al.

For the conventional induction machine, a positive sign of slip ‘g’ corresponds to
a motor operation of the machine, whereas a negative sign translates to a generator
operation, which may lead to confusion in the understanding of the DFIM operating
principle, in which the sign of the slip reflects hypo or hyper-synchronous operation
and not the operating mode of the machine (motor or generator) (Bennani 2011).
The possibility of operating in hyper-synchronism or in hypo-synchronism, both
in motor mode and in generator mode, is obtained by controlling the amplitude and
phase of the rotor voltages, these making it possible to control the magnetic field at
inside the machine.

5.3.3 Modeling of the Doubly Fed Induction Generator

5.3.3.1 Simplifying Assumptions

The basic structure of the three-phase induction machine is given by Fig. 5.9.
Despite its simplicity of construction, the mathematical model of the Doubly Fed
Induction Gnerator DFIG is complex. In the three-phase referential fixed and linked
to the stator, the model of the induction machine present the disadvantage of arriving
at differential equations with variable coefficients as a function of the position of the
rotor, and therefore of time.
To remediate to this problem, several approaches have been used, among them
the approach based on the theory of the two axes RH Park (Park 2012) which

Fig. 5.9 Structure of the Bs

three-phase induction isb
machine vsb

br ar

irb
vrb
ira
vra q

O As
isa
vsa

vrc

vsc irc

isc

Cs cr
5 DFIG Generator with MPPT Technique for Wind Turbines 117

consists in transforming the three-phase system into an equivalent two-phase

system, moving from the fixed frame of reference (A, B, C) to a second mobile
reference frame (d, q): ‘d’ presents the direct axis and ‘q’ the quadrature axis,
perpendicular and mutually coupled, the angle between the two referential is called
‘the Park angle’.

5.3.3.2 Application of Park Transformation to DFIG

Figure 5.10 presents the DFIM model in the PARK referential.

From Fig. 5.10 we obtain:

θs = θr + θ (5.15)

So its derivative can be writing as follows:

dθs dθr dθ
= + (5.16)
dt dt dt
The modeling of the DFIG is identical to that of the induction machine cage. The
only difference lies in the fact that the rotor voltages are not null because these
windings are not ‘short-circuited’.
The general equations obtained after the application of the Park transformation
are given by the following equations.

Fig. 5.10 Graphical Direct axis ’d’

representation of DFIG in the
park referential
B

bR
qr
aR
qs
Quadrature axis ’q’ q
As
O

C
cR
118 Y. Ihedrane et al.

• Stator Voltages:
⎧
⎪
⎨Vsd = Rs × Isd +
dΦsd
− ωs × Φsq
dt (5.17)
⎪
⎩Vsq = Rs × Isq +
dΦ sq
+ ωs × Φsd
dt

• Rotor Voltages:
⎧
⎪
⎨Vrd = Rr × Ird +
dΦrd
− ωr Φrq
dt (5.18)
⎪
⎩Vrq = Rr × Irq +
dΦ rq
+ ωr × Φrd
dt

• Stator Field:

Φsd = Ls × Isd + M × Ird
(5.19)
Φsq = Ls × Isq + M × Irq

• Rotor Field:

Φrd = Lr × Ird + M × Isd
(5.20)
Φrq = Lr × Irq + M × Isq

• Electromagnetic Torque:

Cem = p × (Φsd × Isq − Φsq × Isd ) (5.21)

• The active and reactive powers of the stator:


Ps = Vsd × Isd + Vsq × Isq
(5.22)
Qs = Vsq × Isd − Vsd × Isq

• The active and reactive powers of the rotor:


Pr = Vrd × Ird + Vrq × Irq
(5.23)
Qr = Vrq × Ird − Vrd × Irq

5.4 Modeling of Power Converters

The converter powered by a source of perfect voltage imposes an AC voltage at its

output, which is formed by a succession of two levels rectangular square waves, this
5 DFIG Generator with MPPT Technique for Wind Turbines 119

D1 Q1 D3 Q3 D5 Q5
Ucc/2

iA
Ucc A
O uAB iB
B
uBC uCA iC
C

D4 Q4 D6 Q6 D2 Q2 vA vB vC
Ucc/2

Neutral

Fig. 5.11 Diagram of a three-phase voltage converter

being due to the opening and closing of the switches. The operating frequency is set
by the control of the switches.
The three-phase system obtained at the output of the converter is a balanced
voltage system containing odd harmonics that are different from rank three and their
multiples (Mourad 2016).
The structural diagram of a three-phase voltage converter is given by Fig. 5.11.
The compound voltages are given by the following relations:
⎧
⎪
⎨UAB = vAO − vBO
⎪
UBC = vBO − vCO (5.24)
⎪
⎪
⎩U = v − v
CA CO AO

At the level of the charge, the simple voltages are giving as follows:
⎧
⎪
⎨UAB = vA − vB
⎪
U = vB − vC (5.25)
⎪ BC
⎪
⎩U = v − v
CA C A

We have:
⎧
⎨ UAB − UCA = 2vA − (vB + vC )
U − UAB = 2vB − (vC + vA ) (5.26)
⎩ BC
UCA − UBC = 2vC − (vA + vB )
120 Y. Ihedrane et al.

The voltages (vA , vB , vC ) form a balanced three-phase system, so we can write:

⎧
⎪
⎨UAB − UCA = 3vA
⎪
UBC − UAB = 3vB (5.27)
⎪
⎪
⎩U − U = 3v
CA BC C

The expression of the simple voltages is presented by the following system (5.28)
⎧
⎪ 1
⎪
⎪vA = (UAB − UCA )
⎪
⎨ 3
1
vB = (UBC − UAB ) (5.28)
⎪
⎪ 3
⎪
⎪
⎩vC = 1 (UCA − UBC )
3

Hence, the matrix form of simple voltages is giving as follows:

⎡ ⎤ ⎡ ⎤⎡ ⎤
vA 2 −1 −1 vAO
1
⎣vB ⎦ = ⎣−1 2 −1⎦ ⎣vBO ⎦ (5.29)
3
vC −1 −1 2 vCO

Each arm of the converter consists of two switches which are supposed perfect and
operate in a complementary manner, it is therefore possible to associate a binary
control (Si ) value with each arm with (i = A; B; C) and such that:
⎡ ⎤ ⎡ ⎤
vAO SA
U
⎣vBO ⎦ = CC ⎣SB ⎦ (5.30)
3
vCO SC

By replacing the voltages with their values, we obtain:

⎡ ⎤ ⎡ ⎤ ⎡ ⎤
vA 2 −1 −1 SA
1
⎣vB ⎦ = ⎣−1 2 −1⎦ U CC ⎣SB ⎦ (5.31)
3 3
vC −1 −1 2 SC

The simple voltages delivered by the converter will be obtained directly from the
states of the control quantities SA , SB and SC which represent the control signals of
the switches.
The states of these quantities will also be determined by the envisaged control
strategy.
5 DFIG Generator with MPPT Technique for Wind Turbines 121

Fig. 5.12 Stator Bs

Br Axis d
Field-orientation along the d Axis q
axis
fs
qs
Vs qr
Stator Axis As
q

5.5 Field Oriented Control Strategy Applied to DFIG

5.5.1 Principle of Control by Field Orientation Control

Strategy

The principle of control by field oriented control consists in orienting field along
one of the axes in order to make the behavior of the induction machine similar to
that of machine DC separately excited (Aimani 2004; Ihedrane et al. 2017a).
We have chosen to orient the stator field along the axis ‘d’ as shown in Fig. 5.12
According to this orientation we obtain:

Φsd = Φs
(5.32)
Φsq = 0

In the referential (ABC), the voltage at the terminals of a phase ‘i’ of the stator is
expressed as follows:

dΦsi
vsi = Rs × Isi + (5.33)
dt
with: i = 1, 2, 3
Assuming the resistance of the stator winding ‘Rs is negligible for the medium
and high power machines generally employed in the wind energy conversion
(Ihedrane et al. 2017a; Mourad 2016), we obtain:

dΦsi
vsi = (5.34)
dt
We supposed that the voltages of the electrical grid are stable; the stator flux will
also be constant. So, we can write:

vsd = 0
(5.35)
vsq = vs = ωs × Φs
122 Y. Ihedrane et al.

We have chosen to orient the stator flux along the axis ‘d’, we obtain:
⎧
⎪ 1
⎪
⎪ Isd = × (Φs − M × Ird )
⎪
⎨ Ls
(5.36)
⎪
⎪
⎪
⎪ M
⎩Isq = − × Irq
Ls

So, the previous relations become as follows:

• Relation between the electromagnetic torque and the rotor currents:
The expression of the torque is written as follows:

M
Cem = p × Isq .Φs = −p × × Φs × Irq (5.37)
Ls

• Relation between stator powers and rotor currents:

We replaced Isq and Isd by their expressions in the expression of the stator
power (Ps and Qs ) and knowing that vsd = 0,the expression of the active and
reactive stator powers become as follows:
⎧
⎪ M
⎪
⎪Ps = Vsq × Isq = −Vs × × Irq
⎪
⎨ L s
(5.38)
⎪
⎪
⎪
⎪ Vs2 M
⎩Qs = Vs × Isd = − Vs × × Ird
ωs × Ls Ls

• Relation between rotor voltage and rotor current:

By replacing Isq and Isd with their expressions in Φrd and Φrq we get:
⎧
⎪ M2 M × vs
⎪
⎪ Φ = (L − ) × Ird +
⎪
⎨
rd r
Ls Ls × ωs
(5.39)
⎪
⎪
⎪
⎪
2
⎩Φrq = (Lr − M ) × Irq
Ls

By injecting these equations into the expressions of the rotor voltages, we obtain the
following expression:
⎧     
⎪
⎪ M2 M2
⎪
⎪V = R + S L − I − ω g L − Irq
⎪
⎨
rd r r
Ls
rd s r
Ls
⎪
⎪     
⎪
⎪ M2 M2 g M Vs
⎪
⎩Vrq = Rr + S Lr − Irq − ωs g Lr − Ird +
Ls Ls Ls
(5.40)
5 DFIG Generator with MPPT Technique for Wind Turbines 123

Fig. 5.13 Simplified model of the DFIG

with:
g: the slip of the induction machine.
ωr = g × ωs : the rotor pulsation
ωs : The stator pulsation
These equations make it possible to establish a block diagram of the electrical
system that can be regulated (Fig. 5.13):
According to this block diagram, we can note that the rotor voltages and the
stator powers are linked by transfer functions with the first order. This allows to set
up a vector control with the influence of the couplings, each axis can be controlled
independently with its own controller.
For these controllers, the reference values will be the active power for the rotor
axis ‘q’ and the reactive power for the rotor axis ‘d’ (Allam et al. 2014; Poitiers
2003).
The reactive power reference will be kept zero in order to ensure a unit power
factor on the stator side so as to optimize the quality of the energy sent to the grid.
The active power reference should allow keeping the wind power factor optimal.
To carry out the power control of this machine, there are two structures:
• Structure of the direct control.
• Structure of the indirect control.
124 Y. Ihedrane et al.

Fig. 5.14 Direct control of the doubly fed induction generator

Fig. 5.15 Indirect control of the doubly fed induction generator

5.5.1.1 Structure of Direct Control

The block diagram of the direct control of the Doubly Fed Induction Generator is
shown in Fig. 5.14.
This method consists in neglecting the coupling terms and setting up a regulator
on each axis in order to control independently the active and reactive powers (Sejir
2006; Tarfaya et al. 2015).

5.5.1.2 Structure of Indirect Control

The indirect control of the Doubly Fed Induction Generator is given by Fig. 5.15.
5 DFIG Generator with MPPT Technique for Wind Turbines 125

11
Wind speed profile (m/s)

10

Time (s)
4
0 1 2 3 4 5 6 7 8 9 10

Fig. 5.16 Wind profile

The second method consists in considering the coupling terms and compensates
them by performing a system with two loops in order to control the powers and the
rotor currents. This technique wich called indirect method, allows to controlling the
rotor currents in order to ensure protection of the generator DFIG by limiting the
currents and giving more flexibility to the machine (Allam et al. 2014; Ihedrane et
al. 2017a).

5.6 Simulation and Interpretations

In order to examine the robustness of the Field Oriented Control Technique with the
turbine control by the MPPT strategy, we applied a random wind profile given by
Fig. 5.16.
The rotational speed of the DFIG (Fig. 5.17) has less oscillation and high
reliability with respect to wind speed profile (Fig. 5.16).

5.6.1 Simulation Result of the Direct Field Oriented Control

DFOC

Figure 5.18 shows the simulation results of Direct Field Oriented Control DFOC:
126 Y. Ihedrane et al.

250
Wmec (rad/s): Mechanical speed according the MPPT

200

150

100

50

Time (s)
0
0 1 2 3 4 5 6 7 8 9 10

Fig. 5.17 Rotational speed of the DFIG ωmec

Fig. 5.18 Simulation results of direct field oriented control

5 DFIG Generator with MPPT Technique for Wind Turbines 127

• The active power (a) follows its reference values (which is the mechanical power
obtained according to the MPPT control strategy) for all the variations of the
wind, which shows the robustness of the field oriented control of the DFIG during
the variation of the wind speed.
• The stator reactive power (b) is maintained at zero in order to ensure a unit power
factor on the stator side.
• The output currents Isabc (c) in the wind system is almost sinusoidal with a fixed
frequency equivalent to that of the grid (50Hz ), which implies a good injection
into the grid.
• The power coefficient Cp (d) reach its maximum values Cp = 0.5505, which
implies that the system provide an optimal power from the wind.

5.6.2 Simulation Result of the Indirect Field Oriented Control

IFOC

Figure 5.19 presents the different performances of the Indirect Field Oriented
Control, which shows the robustness of the wind system and the good tracking.

Fig. 5.19 Simulation results of indirect field oriented control IFOC

128 Y. Ihedrane et al.

• The stator active and reactive power (a, b) show a perfect decoupling and a good
tracking of their references values with a reduction of the undulations, compared
to the direct control. Consequently, the harmonics are minimized.
• The stator currents (c) are also sinusoidal in shape, with an improvement in
quality compared to those obtained by the direct control, which implies a clean
power without harmonics echanged between the DFIG and the grid.
• From (d) we can note that the power coefficient kept around its optimum
Cpmax = 0.5505.
So, the results obtained from the IFOC show the reliability and robustness of the
proposed control and provide better control for injecting power into the grid.

5.7 Conclusion

This work has been devoted to the modeling and control of the wind turbine
conversion system based on the doubly-fed induction generator DFIG in order to
regulate the active and reactive stator power, with the MPPT strategy to extract the
maximum of the power generated by DFIG.
Thus, we presented two methods of control of the DFIG which were implemented
under MATLAB/SIMULINK in order to visualize the results of simulation.
Direct methods in which the active and reactive powers are measured and
compared with their references. The indirect method has the role to improve the
tracking of the instructions and therefore, ameliorate the performance of the wind
system.
We have seen that the direct control has good results with low implementation
complexity, On the other hand, the indirect method allows to obtain an efficient and
robust system. It is true that it is more complex but it makes possible to achieve
an optimal operation of the electrical generation system by minimizing the worries
linked to the parametric variations of the DFIG.
As perspective, this work can be continued and completed by the implementation
of this command in an FPGA.

References

Aimani, S. E. (2004). Modélisation de Différentes Technologies d’Eoliennes Intégrées dans un

Réseau de Moyenne Tension. Ph.D. thesis, Université des Sciences et Technologies de Lille.
Allam, M., Dehiba, B., Abid, M., Djeriri, Y., & Adjoudj, R. (2014). Etude comparative entre
la commande vectorielle directe et indirecte de la machine asynchrone à double alimenta-
tion(mada) dédiée à une application éolienne. Journal of Advanced Research in Science and
Technology, 1(2), 88–100.
Bossoufi, B., Karim, M., Lagrioui, A., Taoussi, M., & Hafyani, M. (2014). Backstepping control
of DFIG generators for wide-range variable-speed wind turbines. IJAAC International Journal
of Automation and Control, 8(2), 122–140.
5 DFIG Generator with MPPT Technique for Wind Turbines 129

Bekakra, Y., & Attous, D. B. (2011). Sliding mode controls of active and reactive power of a
DFIG with MPPT for variable speed wind energy conversion. Australian Journal of Basic and
Applied Sciences, 5, 2274–2286.
Bennani, H. (2011). Machine asynchrone à double alimentation: Les lois de commande en régime
permanent. Ph.D. thesis, Université de Laval, Québec.
Datta, R., & Ranganathan, V. R. (2002). Variable-speed wind power generation using doubly fed
wound rotor induction machine a comparison with alternative schemes. IEEE Transaction on
Energy Conversion, 17, 414–421.
de Wit, C. C. (2000). Commande des moteurs asynchrone modélisation contrôle vectoriel et DTC.
Lavoisier, 1, 10–22.
Gaillard, A. (2010). Systéme éolien basé sur une MADA: Contribution l’étude de la qualité de
l’énergie électrique et de la continuité de service. Ph.D. thesis, Université Henri Poincaré,
Nancy-I.
Ghoudelbourk, S., Dib, D., Omeiri, A., & Azar, A. T. (2016). MPPT control in wind energy
conversion systems and the application of fractional control (PIα ) in pitch windturbine.
International Journal of Modelling, Identification and Control, 26, 140–151.
Guda, S. R. (2005). Modeling and power management of a hybrid wind-microturbine power
generation. Master’s thesis, Université de Bozeman, Monata.
Ihedrane, Y., Chakib, E., & Bossoufi, B. (2017a). Power control of DFIG-generators for wind tur-
bines variable-speed. International Journal of Power Electronics and Drive Systems (IJPEDS),
8(1), 444–453.
Ihedrane, Y., Chakib, E., & Bossoufi, B. (2017b). Direct and indirect field oriented control
of DFIG-generators for wind turbines variable-speed. In IEEE 14th International Multi-
Conference on Systems, Signals & Devices, Marrakech, Morocco (pp. 27–32).
Johansen, P. R., Patterson, D., O’Keefe, C., & Swenson, J. (2001, January–March). The use of an
axial flux permanent magnet in wheel direct drive in an electric bicycle. Renewable Energy,
22(1–3), 151–157.
Kazemi, M. V., Yazdankhah, A. S., & Kojabadi, H. M. (2010). Direct power control of DFIG based
on discrete space vector modulation. Renewable Energy, 35, 1033–1042.
Mourad, L. (2016). Synthèse de lois de commande non-linéaires pour le contrôle d’une machine
asynchrone à double alimentation dédiée à un système aérogénérateur. Ph.D. thesis, Université
Aboubakr Belkaid – Tlemcen – Faculté de Technologie.
Park, R. H. (2012). Two-reaction theory of synchronous machines: Generalized method of analysis
– Part I. In Transaction of the AIEE, 48, 716–730.
Poitiers, F. (2003). Etude et commande de génératrice asynchrones pour l’utilisation de l’énergie
éolienne. Ph.D. thesis, Ecole polytechnique de l’Université de Nantes.
Sejir, K. (2006). Commande Vectorielle d’une Machine Asynchrone Doublement Alimentée
(MADA). Ph.D. thesis, Institut National Polytechnique de Toulouse, France.
Tarfaya, A., Dib, D., & Ouada M. (2015). Variable-speed wind power generation using doubly fed
wound rotor induction machine a comparison with alternative schemes. International Confer-
ence on Mechanical And Industrial Engineering ICMAIE’2015, Kuala Lumpur, Malaysia.
Youcef, B. (2014). Contribution à l’étude et à la Commande Robuste d’un Aérogénérateur
Asynchrone à Double Alimentation. Ph.D. thesis, Université Mohamed Khider – Biskra.
Chapter 6
The Ripple Correlation Optimal Point
Determination in a Medium Power Wind
Conversion System and Performance
Evaluation with Respect to Conventional
Algorithms

Hassan Abouobaida and Said EL Beid

Abstract This paper deals with the application of the method called ripple corre-
lation speed (RCS) in a power wind conversion system (PWCS) based a permanent
magnet synchronous generator (PMSG). This method is widely used in photovoltaic
systems for optimal maximum power point determination. The contribution of this
work is the use of two electrical sensors instead of mechanical sensors in order to
find the maximum power point (MPP). This method is justified by a ratio of pro-
portionality between the mechanical speed of the (PMSG) generator and the output
voltage of the three-phase rectifier. The reading of the power curve as a function of
the output voltage of the three-phase rectifier showed that the optimum point has
a maximum according to the wind speed. The desired point is characterized by a
determined voltage. This observation allows to apply the ripple correlation between
power and voltage at the output of the three-phase rectifier. The (PMSG) generator
associated with the three-phase diode rectifier is considered as a continuous source
whose voltage is dependent on the wind speed and hence the mechanical speed of
the generator. This paper uses a power conversion structure based on a synchronous
generator, a three-phase rectifier and a boost converter. The choice of this structure
is justified by the reliability, simplicity of implementation and control. The boost
converter ensures maximum power point tracking. The control duty cycle of the
boost converter is determined by the ripple correlation of the input voltage and
power of the Boost converter. This paper discusses the performance evaluation
of a MPPT method called RCS (ripple correlation speed) in comparison with the
conventional methods. To achieve this objective, four criteria are chosen: speed
convergence, need of mechanical sensors, accuracy and dependence of the previous
position. The main features presented in this paper are as follows: (a) modeling the

H. Abouobaida () · S. EL Beid

Laboratory of Engineering Sciences for Energy, National School of Applied Sciences,
Chouaib-Doukkali University, El Jadida, Morocco
e-mail: [email protected]

© Springer Nature Singapore Pte Ltd. 2019 131

N. Derbel, Q. Zhu (eds.), Modeling, Identification and Control Methods
in Renewable Energy Systems, Green Energy and Technology,
https://doi.org/10.1007/978-981-13-1945-7_6
132 H. Abouobaida and S. EL Beid

power wind conversion system, (b) presenting and detailing the conventional MPPT
method (c) detailing the proposed (MPPT), (d) presenting simulation results, (e)
conducting a discussion and evaluation performance of the proposed method and
comparison with conventional methods.

Keywords MPPT · TSR · RCC · PMSG · HCF · PSF · PWCS

6.1 Introduction

The production of electric energy from the wind power source is today one of the
most promising electrical energy sources and one of the most widely discussed
themes in the scientific research. For small wind generation (typically less than
100 kW) and for the reasons of simplicity of control, reliability and low cost, the
power conversion structure based a (PMSG) generator, an uncontrolled three-phase
rectifier with diodes and boost converter is chosen (Urtasun et al. 2013).
The intermittent nature of the weather conditions and to obtain a better efficiency
of the power wind conversion structure, the wind system must be equipped with
an MPPT algorithm that allows to permanently search for a point corresponding to
maximum power operation. The desired point is characterized by a optimal speed of
the (PMSG) generator as a function of the available wind speed (Laraa et al. 2015).
Several research papers in the literature have addressed the methods called
(MPPT: maximum power point tracking) in power wind conversion systems (Blaab-
jerg et al. 2011). Among the most popular methods are TSR (Tip Speed Ratio)
control, power signal feedback (PSF) and hill-climb search (HCS). The TSR method
is based on controlling the speed of the wind turbine in order to maintain the TSR
at its optimum value, which allows to operation at a maximum power. The PSF
method is based on the knowledge of the power curve of the wind generator. A
practical reading of the power curve as a function of the generator speed implies
prior knowledge of the optimum speed. The (HCS) is similar (MPPT) method to
the Perturb and Observ (P&O method Zhang et al. 2012). The optimum point
determination is based on the continuous calculation of the power variation and
the speed variation. The (TSR) and (PSF) MPPT method are based on the use of
mechanical sensors (speed and wind sensors). These sensors lack precision, which
implies an inaccuracy in the search for optimal point tracking. This problem is
explained in Dalala et al. (2013) and in Kot et al. (2013). The accuracy of these
methods is degraded following a change in ambient temperature, which presents a
second problem (Pena et al. 2011).
The disadvantage of the (HCS) MPPT method is the large ripple around optimal
point. A small variation in wind speed causes an optimal point shift which involves
adjusting the control to find the new point (MPP) and oscillating again around this
point. For sudden changes in wind speed, the (MPPT) method lose the direction of
the new optimal point displacement and therefore finds a difficulty in the optimal
point detection.
6 Ripple Correlation Optimal Point for a Wind Conversion System 133

Several works in the literature have addressed the study of different (MPPT)
and an evaluation of their performances. The paper (Kot et al. 2013) details the
comparison of the (TSR), (PSF) and (INC-COND) methods according to criteria
based on the efficiency, the flexibility, the speed, the cost and the complexity of the
implementation.
The paper (Pena et al. 2011) limits the study of the comparison of the (P&O) and
(ODC) method according to the efficiency of the power wind conversion system.
The evaluation according to the efficiency and the dynamics of searching maximum
power point is approached in Shirazi et al. (2009). The objective of this work is the
evaluation performance of three (MPPT) method. The first (MPPT) is called IHC
(intelligent Hill-Climb), the second is the optimum control (OC) and the third is
based on successive measurements of power.
In this paper, a new high-precision method named RCS (ripple correlation
speed) is proposed. The discussed method uses two electrical sensors instead of the
mechanical sensors used in the other method. The determination of maximum power
point is based a correlation between the power and voltage at the input of the DC/DC
converter. Limitation of accuracy of the mechanical sensors is avoided by the better
accuracy and the low cost of the electrical sensors. The rest of the paper is organized
as follows. While Sect. 6.2 deals mainly with the modeling of the power wind
conversion system (PWCS), a presentation of each (MPPT) method is explained and
detailed in Sect. 6.3. Presentation and simulation results of the proposed method in
Sect. 6.4. Section 6.5 presents a conclusions drawn from this work.

6.2 Wind Conversion System Modeling

The power wind conversion system (PWCS) is composed of blades that capture
wind speed and electrical generator that converts the wind power into electrical
energy. The electrical generator based (PMSG) machine is coupled directly to
the blades. The (PMSG) generator is cascaded to AC/DC rectifier and a DC/DC
Boost converter. The function of the adaptation circuit is to operate the (PWCS)
at maximum power (Lahfaoui et al. 2014; Lahfaoui et al. 2015a,b). Figure 6.1
illustrates the power wind conversion structure.

Fig. 6.1 Power wind conversion system (PWCS)

134 H. Abouobaida and S. EL Beid

6.2.1 The Turbine Model

The aerodynamic power Pt captured by the wind turbine is expressed by the

following relation:

1
P = π ρ R 2 Cp (λ) v 3 (6.1)
2
where the tip speed ratio λ is given by:

RΩm
λ= (6.2)
v
and where v is the wind speed, ρ is the air density, R is the rotor radius and Cp is
the power coefficient.
In a wind turbine, the power coefficient Cp is expressed as function of tip speed
ratio λ in place of pitch angle β as:
   
116 21
Cp = 0.576 − 0.4β − 5 exp + 0.0068λ (6.3)
λi λi

with:
1 1 0.035
= − (6.4)
λi λ + 0.08β 1 + β3

The aerodynamic power is also defined by:

P = Tm ωm (6.5)

where Tm is the aerodynamic torque and ωm is the rotor speed. The power coefficient
in term of rotor speed and the aerodynamic power in term of tip speed ratio are
respectively illustrated in Figs. 6.2 and 6.3.
In this paper, the control of the power wind conversion system is carried out
keeping the pitch angle equal to zero (β = 0). The following mechanical model
gives the dynamic of the turbine:

dωm
J = Tm − Tem − f ωm (6.6)
dt
where Tem is the electromagnetic torque of the synchronous generator, J is the
turbine total inertia and f is the turbine total external damping.
6 Ripple Correlation Optimal Point for a Wind Conversion System 135

x 104
2.5
2.25 V=12m/S
2
1.75
1.5
Power(W)

V=10m/S
1.25
1
0.75 V=8m/S
0.5
V=6m/S
0.25
0
0 5 10 15 20 25 30 35 40 45 50
Rotor Speed (rpm)

Fig. 6.2 Power curve a function of the rotor speed

0.5
Pach angle = 0

0.4

0.3
Cp

0.2

0.1

0
0 5 10 15 20
Tip speed ratio

Fig. 6.3 Power coefficient as function of tip speed ratio

6.2.2 The PMSG Model

The dynamic equations of a three-phase permanent magnetic synchronous generator

(PMSG) can be written in a synchronously rotating dq reference frame as Ben Ali
et al. (2017):

disd
Vsd = −Rs isd − Ld + Lq ωr isq (6.7)
dt
136 H. Abouobaida and S. EL Beid

Fig. 6.4 DC/DC boost converter

disq
Vsq = −Rs isq − Lq − Ld ωr isd + φst ωr (6.8)
dt
where Vsq and Vsd are the q-axis and d-axis stator voltages. isd and isq are the q-
axis and d-axis stator currents, Rs is the stator winding resistance, ωr = pωm is the
electrical angular velocity of the rotor and p is the number of pole pairs and φst is
the rotor flux linkage (Djagarov et al. 2016).
The output electrical power is expressed as the following equation:

3
P = (Vsd isd + Vsq isq ) (6.9)
2

6.2.3 The Boost Converter Model

The boost converter is illustrated in Fig. 6.4. The dynamic model of the boost
converter is expressed as follows:

C2 U̇s = (1 − u)IL − Is (6.10)

LI˙L = Udc − (1 − u)Us (6.11)

where L and IL represents the storage inductance and the current across it, Us is the
output voltage and u is the switched control signal that can only take the discrete
values 0 (switch open) and 1 (switch closed).

6.3 Maximum Power Point Tracking Methods (MPPT)

A wind system is characterized by the presence of a maximum power point in

the power curve (see Fig. 6.2) as a function of the generator speed (Almozayen
et al. 2016; Sahin et al. 2016). Depending on the wind speed, the maximum power
point can move up or down, which implies a change in the rotor speed at which
the synchronous generator must rotate (Asri et al. 2017; Biweta and Mamo 2017).
6 Ripple Correlation Optimal Point for a Wind Conversion System 137

Fig. 6.5 MPPT method based TSR control

Several methods of maximum power point tracking are discussed in the literature,
among the most popular methods are TSR, HCS, PSF. When the maximum power
point tracking (MPPT) method finds the optimum speed, a control loop is required
to adjust the rotational speed to its reference (Fantino et al. 2015; Koch et al. 2015).
In the literature, the (MPPT) methods are classified into three categories: Tip
Speed Ratio Control (TSR), Hill Climbing Searching Control (HCS) and Power
Signal Feedback Control (PSF). In the next paragraph, the operating principle of the
(MPPT) methods, the simulation results are presented, discussed and commented.

6.3.1 Tip Speed Ratio Control

The principle of the TSR method is based on a continuous measurement of the

wind speed. A prior knowledge of the tip speed ratio (TSR) allows to determine
the optimum speed which corresponds to maximum power operation. Figure 6.5
illustrates the principle of the (TSR) method.
The TSR method is based on the continuous measurement of wind speed. An
inaccuracy in the measurement by an anemometer cause an operating of the power
wind conversion system a far from its maximum power. The knowledge of the tip
speed ratio is necessary to determine the optimal speed.
The model of the power wind conversion system based (PMSG) generator,
AC/DC rectifier and the boost converter is built using Matlab/Simulink software.
The synchronous generator provides a power of 22 kW. The switching frequency
is set to 20 kHz. The simulations of the (TSR), (PSF), (HCS) and the proposed
method are carried out using the same parameters and the same power conversion
structure.The simulation parameters are given in Tables 6.1 and 6.2
The validation of the (TSR) method and the simulations results of the (TSR)
MPPT are given in Fig. 6.6. Figure 6.6a shows the wind profile applied to the wind
turbine blades. The profile used is characterized by two rising steps (8 to 10 m/s and
10 to 12 m/s).
138 H. Abouobaida and S. EL Beid

Table 6.1 Parameters of the Parameters Values Units

boost converter and RCS
MPPT Boost L 3 mH
Converter C2 470 μF
Us 800 V
f 20 kHz
RCS MPPT k 10−4 –

Table 6.2 Parameters of the Parameters Values Units

power wind conversion
system Turbine Vtn 12 m/s
R 4 m
PMSG U 500 V
Pn 22 kW
f 50 Hz
Rs 50 mΩ
Ld = Lq 0.6 mH
Jm 0.011 kgm2
P 10 –

Figure 6.6.b shows the power extracted as a function of time according to the
wind speed changes. According to the power curve given in Fig. 6.3, the mechanical
power is well regulated to its maximum value. In Fig. 6.6.b, the wind system reaches
the optimal point very quickly in transient mode and in the event of a sudden change
in wind speed. In addition, the oscillations around the optimal point are very weak.
This is justified by the ease of optimal point determination based on the formula
λopt
ωopt = v.
R
Figure 6.6c illustrates the mechanical speed of the synchronous generator.
According to Fig. 6.3, the rotor speed is well regulated to its optimal value.
Figure 6.6d illustrate the voltage at the input of the boost converter. According
to this figure, the voltage at the input of the boost converter is well adjusted to its
reference. Figure 6.6e shows the power coefficient as a function of time. According
to Fig. 6.3, the power coefficient is well maintained at its optimum value of 0.43 in
the presence of the sudden changes of wind speed.

6.3.2 Hill Climbing Searching Control

The (HCS) MPPT method determines the optimal point from the variation of the
mechanical power and the variation of the generator speed. The calculation of the
power variation and the rotor speed variation is carried out continuously in order
to detect any change in the wind speed. The detection of the speed change implies
6 Ripple Correlation Optimal Point for a Wind Conversion System 139

Fig. 6.6 TSR MPPT Results. (a) Wind speed (m/s). (b) Generator power (W). (c) Rotor Speed
ωm (rad/s). (d) Udc Voltage (V). (e) Power coefficient
140 H. Abouobaida and S. EL Beid

Fig. 6.7 Principle operation of the (HCS) Method

a positioning of the wind system in a new optimal point. The principle of (HCS)
method is summarized by the following two points:
• If the variation of the power and the variation of the speed have the same sign,
the optimal point is to the right of the actual point.
• If the variation of power and the variation of the voltage have a different sign,
the optimal point is to the left of the actual point. Figure 6.7 shows the operation
principle of the (HCS) method.
A very rapid change in the wind speed cause a possible loss of optimal point
because the new optimal point is determined from the previous point, which
constitutes a disadvantage of the (HCS) method.
The validation of the (HCS) method and the simulations results of this MPPT
are given in Fig. 6.8. Figure 6.8a shows the wind profile applied to the wind turbine
blades. Figure 6.8b illustrates the power as function of time. According to this figure,
the system finds the optimal point in transient mode after 200 milliseconds. During
abrupt changes in wind speed, the (HCS) method finds the new optimum point after
100 milliseconds. Figure 6.8c illustrates the mechanical speed of the synchronous
generator. According to Fig. 6.3, the generator speed is well regulated to the optimal
value. Figure 6.8d illustrate the voltage at the input of the boost converter. The
input voltage is well regulated to its optimal value. Figure 6.8e shows the power
coefficient as a function of time. According to Fig. 6.3, the power coefficient is well
maintained at its optimum value of 0.43 in the presence of the sudden changes of
the wind speed. According to Fig. 6.8e, the power coefficient deviates at t = 1S and
at time t = 2S from its optimal value of 0.43 due to sudden changes in wind speed.
The (HCS) MPPT was able to restore the power coefficient to its optimal value.
The simulations results show that the oscillations of the power coefficients and the
oscillation of the voltage around the optimal point using the (HCS) MPPT method
are greater than the oscillations observed using the (TSR) method.
6 Ripple Correlation Optimal Point for a Wind Conversion System 141

Fig. 6.8 HCS MPPT Results. (a) Wind speed (m/s). (b) Generator power (W). (c) Rotor Speed
ωm (rad/s). (d) Udc Voltage (V). (e) Power coefficient
142 H. Abouobaida and S. EL Beid

Fig. 6.9 Principle operation of the (PSF) method

6.3.3 Power Signal Feedback Control

The (PSF) method is based on a prior knowledge of the power characteristic as

a function of the generator speed. The first step consists in learning of the power
characteristic as a function of the speed of the generator. Depending on the wind
speed, the (PSF) MPPT method provides the optimum speed which corresponds to
a maximum power operation of the wind system. Figure 6.9 shows the principle
operation of the (PSF) method.
The disadvantage of this method lies in the change of the power curve due to
a degradation of the performance of the wind system which affects the optimal
point determination. A second disadvantage is the stopping of production of the
wind system at the time of learning of the power characteristic. The validation of
the (PSF) method and the simulations results of this MPPT are given in Fig. 6.10.
Figure 6.10a shows the wind profile applied to the wind turbine blades. The wind
profile applied to the (PSF) method is identical to the wind profile applied to
the (HCS) and (TSR) method. Figure 6.10b illustrates the power extracted as
a function of time. According to Fig. 6.10b, the power wind conversion system
reaches the optimal value after a few milliseconds (t < 10 mS). Figure 6.10c
illustrates the mechanical speed of the synchronous generator. The generator speed
is well regulated to the optimal value. Figure 6.10d illustrate the voltage at the
input of the boost converter. The voltage at the input of the boost converter is
adjusted to its reference. According to Fig. 6.10d, the input voltage is adjusted to
its reference. Figure 6.10e shows the power coefficient as a function of time. The
power coefficient is well maintained at its optimum value of 0.43.

6.4 Proposed MPPT Method

The proposed method is named (RCS: ripple correlation speed), uses the correlation
between the power and the generator speed (Abouobaida et al. 2014). The (RCS)
method is based on the following hypothesis: the output voltage of the boost
converter is assumed to be constant. If the wind system is cascaded with a
6 Ripple Correlation Optimal Point for a Wind Conversion System 143

Fig. 6.10 PSF MPPT Results. (a) Wind speed (m/s). (b) Generator power (W). (c) Rotor Speed
ωm (rad/s). (d) Udc Voltage (V). (e) Power coefficient.
144 H. Abouobaida and S. EL Beid

three-phase inverter, the output voltage of the boost converter is regulated using
another control loop (part not studied in this chapter).
The proposed (MPPT) method is inspired by the following remarks:
• If the variation of the generator speed and the change of power have the same
signs (positive or negative), the optimal point is to the right of the operating
point,
• If the variation of the generator speed and change of the power have opposite
signs, the optimal point is to the right.
By combining these observations, it can be concluded that the product of the
change of the speed and the change of the power Ṗ ω̇ is positive to the left of optimal
point and negative to the right of this point.
The choice of sensors is justified by the simplicity of implementation and the
limitation of the cost and precision. The justifications are explained as follows:
• Instead of measuring the generator speed with a tachometer speed sensor, a
higher accuracy voltage sensor is used at the input of the DC/DC converter since
this voltage is proportional to the generator speed.
• A second current sensor is used at the input of the DC/DC converter, which
allows to determine the power extracted by multiplying the voltage and the
current at the input of the DC/DC converter.
• A reading of the power curve as a function of the voltage at the input of the
DC/DC converter allows to identify the new curve of the wind system at the
input of the DC/DC converter. The advantage of this method is that it is easy to
determine the optimal point by two electrical sensors. The new observed power
curves are characterized by maximum power points according to the wind speed.
To show the linear relation between the generator speed and the input voltage
of the DC/DC converter, a reading is made for four wind speeds. According to
Fig. 6.11, the factor of proportionality between the speed and the voltage is constant
and independent of the wind speed.
The proportionality relation between the generator speed and the rectified voltage
Udc is justified by the following relation:
√
3 6
Udc = φeff P ωm (6.12)
π
where Udc is the rectified voltage, φeff is the effective value of the internal flux of
the (PMSG) generator, P is the number of poles and ωm the rotation speed of the
asynchronous generator (Marmouh et al. 2016; Saihi and Boutera 2016; Sarkar and
Khule 2016).
Figure 6.12 illustrate the principle of the proposed (MPPT) method.
The electrical power at the input of the boost converter is expressed by the
following relation:

Pdc = Udc Idc (6.13)

6 Ripple Correlation Optimal Point for a Wind Conversion System 145

Fig. 6.11 Udc voltage as a function of generator speed. (a) v = 12 m/s, (b) v = 10 m/s, (c)
v = 8 m/s, (d) v = 6 m/s.

Fig. 6.12 Principle of the proposed (MPPT) method

146 H. Abouobaida and S. EL Beid

Fig. 6.13 Power as function DC voltage Pdc = f (Udc ) at the input of the boost converter

where Udc and Idc are respectively the voltage and the current at the input of the
boost converter (Ayadi et al. 2017; Zou and He 2016).
According to Fig. 6.12, the duty cycle of controlling of the boost converter is
expressed by the following relation:

α=k= Ṗdc U̇dc dt (6.14)

where k is a positive parameter, Ṗdc and U̇dc are the time derivative of the power
and the time derivative of the voltage, respectively.
Figure 6.13 illustrates the power curve as a function of the voltage at the input of
the boost converter.
The validation of the (RCS) method and the simulations results are given in
Fig. 6.14. Figure 6.14a shows the wind profile applied to the wind turbine blades.
The profile used is characterized by a descending step (10 to 8 m/s) and two rising
steps (8 to 10 m/s and 10 to 12 m/s). The applied steps are abrupt which is the
extreme case of a wind scale and also allows to show the capacity of the proposed
(MPPT) of a tracking as quickly as possible to the new optimal point. Figure 6.14b
shows the extracted power at the input of the boost converter as a function of time.
According to the power curve given in Fig. 6.3, it is noted that the extracted power
is well regulated to maximum power. The difference between the extracted power
and the power given in Fig. 6.3 represents the power losses in the wind system
which is logical since the curve in Fig. 6.3 represents the mechanical power not
the available power at the input of the boost converter. A comparison between the
6 Ripple Correlation Optimal Point for a Wind Conversion System 147

Fig. 6.14 RCS MPPT Results. (a) Wind speed (m/s). (b) Generator power (W). (c) Rotor Speed
ωm (rad/s). (d) Udc Voltage (V). (e) Power coefficient
148 H. Abouobaida and S. EL Beid

power extracted and the power curve given in Fig. 6.9 shows a similarity between
these two values. Figure 6.14c illustrates the mechanical speed of the synchronous
generator. According to Fig. 6.3, the generator speed is well regulated to the optimal
value that correspond to a maximum power operation of the wind system.
Figure 6.14d illustrates the voltage at the input of the boost converter. Fig-
ure 6.14d shows that Udc voltage is well regulated to the optimal value. According
to Fig. 6.9, Udc voltage must be regulated at 400 V when the wind speed is 8 m/s
and adjusted to 520 V when the wind speed is equal to 10 m/s and regulated to
620 V when the wind speed is equal to 12 m/S. Figure 6.14d shows that Udc voltage
follows the desired value.
Figure 6.14e illustrates the power coefficient. According to Fig. 6.2, the optimal
value of power coefficient is equal to 0.43. According to Fig. 6.14, the power
coefficient is well regulated at its optimal value in the presence of sudden changes
in wind speed.
In summary, Fig. 6.14 justify the validity of the proposed approach to extract the
maximum of the available wind energy and thus a good exploitation of the power
wind conversion system in a situation characterized by sudden, fast and successive
changes in wind speed.
Table 6.3 summarizes a general evaluation and comparison of the proposed
method and the conventional methods. The criteria of comparisons are: speed of
convergence, need for mechanical sensors, precision and dependence of the previous
position.
According to Table 6.3, the proposed (MPPT) method has many advantages:
• a response time and therefore quick searching and optimal point tracking,
• the use of two electrical sensors with better accuracy, low cost and simplicity of
implementation instead of two mechanical sensors,
• a better accuracy than the other method that reflects a very small distance from
the optimum point and an operation at the maximum power point.

Table 6.3 Comparison between MPPT algorithms

RCS
MPPT TSR HCS PSF (Proposed)
Speed FAST Medium(1) Fast Good(2)
convergence (depends on step size) (depends on parameter k)
Mechanical Yes Yes Yes No(3)
sensor (2 electrical sensor)
Dependence of No Yes No Yes
previous position
Accuracy Good Medium Good Very Good
6 Ripple Correlation Optimal Point for a Wind Conversion System 149

6.5 Conclusion

This paper has addressed a new method of tracking maximum power point in a
power wind conversion system (PWCS). The proposed method called (RCS: ripple
correlation speed) is based on a correlation calculation between the power and the
voltage at the input of the DC/DC converter.
A summary table at the end of the chapter illustrate the comparison of the
suggested method and the conventional methods. The evaluation performance is
based on four criteria: convergence speed, sensors need, dependence on the previous
position and accuracy. According to the simulation results, the proposed method
showed a better performance in terms of optimal point detection speed, need
of the electrical sensors and better accuracy. Another advantage, the proposed
method seems simpler, efficient and promising due to the elimination of the wind
speed sensor and the generator speed sensor which implies a low cost and a
flexibility. The major advantage of this approach is that the control method does
not need to the parameters of the turbines which constitutes an inconvenience of
the other approaches because of the uncertainties in the measurement of the turbine
parameters. In conclusion, the simulation results proved the validity of the proposed
approach.

References

Abouobaida, H., EL Khayat, M., & Cherkaoui, M. (2014). Combination of RCC MPPT and
Backstepping controler to design a standard continuous source (12V–24V) supplied by a PV
panels. Journal of Electrical Engineering, 14(3), 364–371.
Almozayen, M. A., El-Nemr, M. K., Rashad, E. M., & Shobair., A. I. (2016). Finite-element-
based design of axial field PMSG for wind energy conversion with MPPT control. In 18th
International Middle East Power Systems Conference (MEPCON) (pp. 610–616).
Asri, A., Mihoub, Y., Hassaine, S., & Allaoui, T. (2017). Intelligent maximum power tracking
control of PMSG wind energy conversion system. In 5th International Conference on Electrical
Engineering – Boumerdes (ICEE-B), Algeria (pp. 1–6).
Ayadi, M., Naifar, O., & Derbel, N. (2017). Sensorless control with an adaptive sliding mode
observer for wind PMSG systems. In 14th International Multi-conference on Systems, Signals
& Devices (SSD), Morocco (pp. 33–37).
Ben Ali, R., Schulte, H., & Mami, A. (2017). Modeling and simulation of a small wind turbine
system based on PMSG generator. In Evolving and Adaptive Intelligent Systems (EAIS),
Ljubljana, Slovenia (pp. 1–6).
Biweta, M., & Mamo, M. (2017). Closed loop control strategy of back to back PWM converter
fed by PMSG using PLECS toolbox on Matlab/Simulink for wind energy application. In IEEE
AFRICON, Cape Town, South Africa (pp. 1313–1318).
Blaabjerg, F., Liserre, M., & Ma, K. (2011). Power electronics converters for wind turbine systems.
Energy Conversion Congress Exposit (ECCE), Arizona, USA (pp. 281–90).
Dalala, Z. M., Zahid, Z. U., Yu, W., Cho, Y., & Lai, J. S. (2013). Design and analysis of an
MPPT technique for small-scale wind energy conversion systems. IEEE Transaction on Energy
Conversion, 28, 756–767.
Djagarov, N., Djagarova, J., Grozdev, Z., Bonev, M., Pazderin, A., & Kokin, S. (2016). PI vector
control study for wind PMSG. In IEEE 16th International Conference on Environment and
Electrical Engineering (EEEIC), Italy (pp. 1–6).
150 H. Abouobaida and S. EL Beid

Fantino, R., Solsona, J., & Busada, C. (2015). Mechanical sensorless controller for a small size
PMSG-based wind energy conversion system. In XVI Workshop on Information Processing
and Control (RPIC), Cordoba, Argentina (pp. 1–6).
Koch, G., Gabbi, T., Henz, G., Vieira, R. P., & Pinheiro, H. (2015). Sensorless technique applied
to PMSG of WECS using sliding mode observer. In IEEE 13th Brazilian Power Electronics
Conference and 1st Southern Power Electronics Conference (COBEP/SPEC) (pp. 1–6).
Kot, R., Rolak, M., & Malinowski, M. (2013). Comparison of maximum peak power tracking
algorithms for a small wind turbine. Mathematics and Computers in Simulation, 91, 29–40.
Lahfaoui, B., Zouggar, S., Elhafyani, M. L., Seddik, M., Kadda, F. Z. (2014). Experimental
modeling and control of a small wind PMSG turbine. In International Renewable and
Sustainable Energy Conference (IRSEC), Ouarzazate (pp. 802–807).
Lahfaoui, B., Zouggar, S., Elhafyani, M. L., & Seddik, M. (2015a). Experimental study of P&O
MPPT control for wind PMSG turbine. 3rd International Renewable and Sustainable Energy
Conference (IRSEC), Marrakech-Ouarzazate, Morocco (pp. 1–6).
Lahfaoui, B., Zouggar, S., Elhafyani, M. L., Seddik, M., & Benslimane, A. (2015b). Modeling
validation and MPPT technique of small wind PMSG turbines using DSPACE hardware. In 3rd
International Renewable and Sustainable Energy Conference (IRSEC), Marrakech-Ouarzazate,
Morocco (pp. 1–6).
Laraa, D., Merinoa, G., & Salazarb, L. (2015). Power converter with maximum power point
tracking, MPPT for small wind-electric pumping systems. In Energy Convers Manage (pp.
53–62).
Marmouh, S., & Boutoubat, M., Mokrani, L. (2016). MPPT fuzzy logic controller of a wind
energy conversion system based on a PMSG. In 8th International Conference on Modelling,
Identification and Control (ICMIC), Algeria (pp. 296–302).
Pena, J. C. U., de Brito, M. A. G., e Melo, G. de A., & Canesin, C. A. (2011). A comparative
study of MPPT strategies and a novel single phase integrated buck-boost inverter for small
wind energy conversion systems. In XI Brazilian Power Electronics Conference (pp. 458–465).
Sahin, P., Resmi, R., & Vanitha, V. (2016). PMSG based standalone wind electric conversion
system with MPPT. In International Conference on Emerging Technological Trends (ICETT),
Kollam, India (pp. 1–5).
Saihi, L., & Boutera, A. (2016). Robust control of a variable-speed wind turbine with fixed pitch
angle and strategy MPPT control associated on a PMSG. In 8th International Conference on
Modelling, Identification and Control (ICMIC), Algeria (pp. 326–331).
Sarkar, J., & Khule, S. S. (2016). A study of MPPT schemes in PMSG based wind turbine
system. In International Conference on Electrical, Electronics, and Optimization Techniques
(ICEEOT), Chennai, Tamilnadu, India (pp. 100–105).
Shirazi, M., Hooshmand Viki, A., & Babayi, O. (2009). A comparative study of maximum Power
extraction strategies in PMSG wind turbine system. In IEEE Electrical Power & Energy
Conference (EPEC), Montreal, Canada (pp. 1–6).
Urtasun, A., Sanchis, P., & Marroyo, L. (2013). Small wind turbines sensorless MPPT: Robustness
analysis and lossless approach. In IEEE Energy Conversion Congress and Exposition, Denver,
Colorado, USA (pp. 2420–2427).
Zhang, X., Li, Q., Yin, M., Ye, X., & Zou Y. (2012). An improved hill-climbing searching method
based on halt mechanism. In Zhongguo Dianji Gongcheng Xuebao/Proceed Chinese Society of
Electrical Engineering, China (pp. 128–134).
Zou, Y., & He, J. (2016). Maximum power point tracking (MPPT) of sensorless PMSG wind power
system. In IEEE Energy Conversion Congress and Exposition (ECCE), Milwaukee (pp. 1–6).
Chapter 7
Control of Wind Turbine Based on
PMSG Using Pitch Angle Control

Aroua Slimen, Hatem Tlijani, Mehdi Dhaoui, and Rached B. Younes

Abstract this chapter focuses on wind power. At present, wind turbines are
installed with variable wind speed. Using intelligent systems, we can harness this
energy that stores the wind and transforms it into electrical energy relying on such
machines. In order to produce the maximum amount of electricity, the obstacles due
to synergistic effects must be taken into consideration. The chapter also presents a
solution enabling to convert the final energy demanded. A design of an intelligent
wind turbine (3.8 kW) pitch angle control with a permanent magnet synchronous
generator (PMSG) (3.5 kW), with the DC bus voltage controller, output from the
converter for a 150 V pumping load, where a mechanism of fuzzy inference is used
to estimate the upper limit of uncertainties. In addition, the fuzzy inference makes it
possible to adjust the output voltage, obtain greater efficiency in the use of wind
energy and protect the blades of the turbine. In the first part, we presented the
modeling system with a pitch angle control. In the second part, a presentation of
the results of the simulation system is analyzed and validated by the simulations in
the MatLab/simulink environment.

Keywords Turbine · PMSG · Wind · Pitch control · Fuzzy logic

A. Slimen ()
Unity of Physical Research, Computer Science and Mathematics, University of Gafsa, Gafsa,
Tunisia
National Engineering School of Gabes, University of Gabes, Gabes, Tunisia
H. Tlijani · R. B. Younes
Unity of Physical Research, Computer Science and Mathematics, University of Gafsa, Gafsa,
Tunisia
Faculty of Science of Gafsa, University of Gafsa, Gafsa, Tunisia
e-mail: [email protected]
M. Dhaoui
Unity of Photovoltaic Research, Wind and Geothermal Energy, University of Gabes, Gabes,
Tunisia
National Engineering School of Gabes, University of Gabes, Gabes, Tunisia

© Springer Nature Singapore Pte Ltd. 2019 151

N. Derbel, Q. Zhu (eds.), Modeling, Identification and Control Methods
in Renewable Energy Systems, Green Energy and Technology,
https://doi.org/10.1007/978-981-13-1945-7_7
152 A. Slimen et al.

Turbine
DC
AC Bus
PMSG Load
DC

βref
Pitch Angle
Control
Rotor
speed

System Vdc
Wind speed
management Vdc ref

Fig. 7.1 General diagram of the control system

7.1 Introduction

Generally, a wind turbine system is constructed with a pitch angle controller, pitch
control mechanisms, backup power and a feedback module monitoring the output
power of the generator. In addition, the pitch control system makes it possible to
adjust the output voltage and provide protection for the turbine blades. This chapter
focuses on dynamic modeling of the wind pump system, modeling of the pitch
control turbine based on a synchronous machine with permanent magnets controlled
by a hysteresis rectifier for pumping 150 V load. Pitch variable-speed wind turbines
is become the dominant type over the past years. For variable speed wind turbines,
there are generally two control strategies.
In low wind speed lower than a nominal value, the cruise control can always
adjust the rotor speed to maintain the speed at a level that gives the maximum power
coefficient, and otherwise, when the wind speed is higher in the event of saturation,
the controller stops the rotation of the blades to protect them, and then the efficiency
of the turbine is increased, the control system is illustrated in Fig. 7.1 and detailed
in the following (Akhmatov 2001, 2002; Akhmatov et al. 2003).

7.2 Mechanical Model of the Turbine

The mechanical power obtained by the conversion of kinetic energy (wind energy)
into mechanical energy available at the output of the wind turbine:

1
Pm = ρ Cp (λ, β) S v 3 (7.1)
2
7 Control of Wind Turbine Based on PMSG Using Pitch Angle Control 153

where, Cp (λ, β) is the power coefficient, λ is the speed ratio, β is the pitch angle,
ρ is the air density, V is the wind speed and S = π R 2 is the blades swept of the
turbine. The mechanical torque of wind is obtained as:

1 R3 v2
Tm = π ρ Cp (λ, β) S (7.2)
2 λ
The power coefficient Cp (λ, β) depending on the speed ratio λ and the pitch angle
β given by:

π(λ − 3)
Cp (λ, β) = (0.44 − 0.167β) sin − 0.00184(λ − 3)β (7.3)
15 − 0.3β
The speed ratio λ is obtained depended by the wind speed and the mechanical speed:
ROmegam
λ= (7.4)
V
where: Omegam is the angular velocity and R is the radius of the turbine blades.
The developed turbine torque is:
Pm
λ= (7.5)
Omegam
The maximum power is given when the power coefficient is at maximum of the
turbine and the speed ratio is at optimum, we can express by the following relation:

1 R 5 Ωm3
Pmax = π ρ Cpmax S (7.6)
2 λo pt 3

The power coefficient describes the production of the turbine, Fig. 7.2 shows the
shape of the power coefficient Cp according to the different values of the speed ratio

Fig. 7.2 Allure of Cp (λ, β)

154 A. Slimen et al.

and the pitch angle, the production is maximum at value Cpmax = 0.48 for a value
of speed ratio λopt = 7.1 and height angle βopt = 0◦ .

7.3 Pitch Angle Control

The wind goes through wind turbines at variable speed, to control the wind system,
and to optimize the extracted power, the only parameter that controls the extracted
power is the parameter Cp (Fig. 7.2). This parameter depends on the type of turbine
and the type of design of the manufacturer. It is also a function of the pitch angle
control, used mainly for optimizing the output power of wind turbines and thus
the DC bus voltage according to our needs, for strong winds (wind speed above
Nominal value) the mechanical power does not exceed the nominal power and
protection against overloading the structure of the turbine and the risks of damage to
its physical structure (Macêdo and Mota 2012). This procedure makes it possible to
control the power supply of the wind farm and to avoid a possible higher generation
in situations of strong winds and low consumption. This can be achieved by using
the blade angle control to stabilize the wind farm when contingencies occur in
the network (to reduce imbalance between generation and loads). This principle
of control must therefore be taken into account in the transient stability studies
and represented in the dynamic models of wind turbines. The purpose of the pitch
control system proposed in this chapter is to maintain the speed of the wind turbine
rotor at nominal value in strong winds. When the rotor speed is higher than the
nominal speed, each blade will be rotated about its axis to reduce the angle (Hafsi
et al. 2016).
The position of the angle β is controlled by the value Y , which can be
characterized in various ways such as:
1. An electric value: for example, electric power Pe as described in Akhmatov
(2001);
2. A single mechanical value: for example, the mechanical power Pm , or the
generator speed ωg , as described in Hansen et al. (2004), Lopes and Almeida
(2004), Slootweg (2003), Akhmatov (2002), Akhmatov et al. (2003), Mota
(2006);
3. A combination of mechanical and electrical power values, according to Miller
et al. (2003), Akhmatov et al. (2003).
Ideally, the control parameters should be chosen depending on wind speed, but
this is not an appropriate procedure, because it is not possible to precisely measure
this variable. The wind speed can be changed by the turbine nacelle. Assuming that
the turbine system is well controlled, the generated power can be used as a control
parameter (Macêdo and Mota 2012).
In this chapter, power supply error criteria are used. The error between the
reference and the measured power at the output of the PMSG is sent to a controller
to generate the reference value βref of the pitch angle (Anjun et al. 2011), two
7 Control of Wind Turbine Based on PMSG Using Pitch Angle Control 155

Xref βref β
error 1 1
PI/FL
+ + Tservo S
– – β

Fig. 7.3 Strategy of the pitch angle controller

Pref βref 1 1 β
error
PI
+ + Tservo S
– – β
Pg

Fig. 7.4 Power regulation pitch angle controller by PI

controllers are applied, a control system by the conventional PI regulator and a

regulator by fuzzy logic is illustrated is in following.

7.3.1 PI Control for the Pitch Angle

The generic blade angle control model (Anjun et al. 2011), using the PI control
block is shown in Fig. 7.3.
Figure 7.4 shows the comparison between the generated power Pg with its value
of reference obtained a error, by applying the PI controller to get the βref , Tservo =
0.01 is a time constant, it is the servomechanism used to correct the rate of change
of the adjustment angle.

7.3.2 Fuzzy Logic Control for the Pitch Angle

Fuzzy logic control considered one of the most control Fuzzy logic, proposed by
Lotfi Zadeh (1965), essentially generates output values without the need for precise
inputs.
Provides a method for translating vague, imprecise and qualitative verbal
expressions common to human communication into numerical values (Simães and
Shaw 2007). Its main advantage is not to require an analytical description of the
156 A. Slimen et al.

Δe
d/dt

Pref e βref 1 1 β
+ + Tservo S
– – β
Pg

Fig. 7.5 Power regulation pitch angle controller by fuzzy logic

Table 7.1 Inference rules for e

the pitch angle
Δe NB NS EZ PS PB
NB NB NB NS NS EZ
NS NB NS NS EZ PS
EZ NB NS EZ PS PS
PS NS EZ PS PS PB
PB EZ PS PS PB PB

controlled system (Macêdo and Mota 2012). The fuzzy logic control is considered
one of the most effective control methods. The fuzzy logic algorithm is suitable
for controlling wind turbines with nonlinear models. The fuzzy controller uses
observation to obtain the optimum pitch angle without knowing the characteristic
of the wind turbine (Akhmatov et al. 2003).
The structure of the fuzzy logic controller basic on three stages: Fuzzification,
decision making Unit and defuzzification unit. The inputs of the fuzzification are
selected as error and rate of change of error and the output is the pitch angle
reference. The input variables are calculated (Nguen and Naidu 2011).

e(k) − e(k − 1)
e(k) = Pref (k) − Pg (k) , Δe(k) = (7.7)
T
where, Pref (k) and Pg (k) are respectively the reference and generated power at step
k (Akhmatov et al. 2003). In Fuzzification stage the crisp variables e(k) and Δe(k)
are converted into fuzzy variables which can be identified by membership function
(Fig. 7.5). In Fuzzification stage the crisp variables e(k) and Δe(k) are converted
into fuzzy variables which can be identified by membership function. The proposed
controller uses following linguistic labels: NB (Negative Big), NS (Negative Small),
EZ (Equal Zero), PS (Positive Small), PB (Positive Big) (Table 7.1).
Triangular membership functions are used to define both inputs (error e, variation
error Δe) and output (pitch angle βref ) as shown as on the different figure: Figs. 7.6,
7.7, 7.8.
7 Control of Wind Turbine Based on PMSG Using Pitch Angle Control 157

Negative Negative Positive Positive

Zero
Big Small Small Big

error ‘e’
–1.5 –1 –0.5 0 0.5 1 1.5

Fig. 7.6 Fuzzy inference for the input error e

Negative Negative Positive Positive

Zero
Big Small Small Big

rate of change
of error ‘Δe’
–1.5 –1 –0.5 0 0.5 1 1.5

Fig. 7.7 Fuzzy inference for the input rate of change of error Δe

Negative Negative Positive Positive

Zero
Big Small Small Big

pitch angle
βref
–1.5 –1 –0.5 0 0.5 1 1.5

Fig. 7.8 Fuzzy inference for the output pitch angle βref

7.3.3 Simulation Results

For evaluating the robustness of the fuzzy logic controller compared with the
conventional PI controller at variable wind speed, the simulations have been carried
out using MATLAB/Simulink software.
Figure 7.9 shows the variation of the wind model, the wind varies around
its nominal value 10 m/s and reaches a maximum value 25 m/s to evaluate the
robustness of the control of the pitch angle control.
Figure 7.10 shows the results of the control strategy of the pitch angle by the
conventional regulator PI.
158 A. Slimen et al.

30
Wind speed(m/s)

25

20
Wind Speed(m/s)

15

10

0
0 1 2 3 4 5 6 7 8 9 10
Time(s)

Fig. 7.9 Wind speed (m/s)

40

30
βPI
β(°)

20

10

0
0 2 4 6 8 10
Time(s)

Fig. 7.10 Pitch angle control with PI

Figure 7.11 shows the results of the control strategy of the pitch angle by the
regulator fuzzy logic.
Figure 7.12 shows the results of the power mechanical with the different regulator
fuzzy logic and PI. The Comparisons of different control strategies of the pitch angle
β with the conventional PI regulator and the fuzzy controller show that the control
strategies of the fuzzy logic the control of the pitch angle.
7 Control of Wind Turbine Based on PMSG Using Pitch Angle Control 159

50

40
βFL

30
β(°)

20

10

0
0 2 4 6 8 10
Time(s)

Fig. 7.11 Pitch angle with fuzzy logic

Fig. 7.12 Comparisons 10000

power with pitch angle
control strategies
8000
PmPI
6000 PmLF
Power(W)

4000

2000

-2000
0 2 4 6 8 10
Time(s)

7.4 Model of the PMSG

The wind turbine drives a permanent magnet synchronous generator PMSG. By

considering only the fundamental harmonic of the flux distribution in the air-gap of
the machine and by neglecting the homopolar component, the theory of the space
vector gives the dynamic equations of the stator currents as follows:
160 A. Slimen et al.

⎧
⎪ isd
⎪
⎨ Vsd = −Rs isd − Ls dt − pΩLs isq
⎪
isq (7.8)
⎪
⎪ Vsq = −Rs isq − Ls + pΩLs isd + pωφ
⎪
⎩ dt
Tem = pφisq

where Rs is the phase resistance of the stator winding, Ls is the stator cyclic
inductance, φ is the flux of the permanent magnetic, Vsd and Vsq are the d-q
components of the stator voltages respectively, isd and isq are the d-q components of
the stator currents respectively, p is the number of pairs of pole, Ω is the mechanical
speed rotor and Tem is the electromagnetic torque.

7.5 DC Bus Control

The generator-side inverter is controlled to catch maximum power from available

wind power. According to (7.8), in order to control the electromagnetic torque Tem ,
this study just controls the q-axis current isq with the assumption that the d-axis
current isd is equal to zero. Furthermore, Veermani and Mohan (2011), Rolan et al.
(2009) show that, in order to catch maximum power, the optimum value of the
rotation speed is adjusted. The tip speed ratio λ is taken into account due to the
equation being addressed as follows:

λopt V
Ωref = (7.9)
R
where Ωref is the blade angular velocity reference, λopt is the tip speed ratio
optimum and V is the wind speed. We apply two regulator, a conventional regulator
PI and à fuzzy logic regulator, between the reference voltages (Vdcref ) and the
voltage (Vdc ) to get the error and sent it to the controllers to adjust the voltage
to a desired quantity which is detailed in the following.

7.6 Fuzzy Logic Controller for DC Bus

The fuzzy logic controller for the DC bus has three parameters. The first input is the
error of DC voltage, the second is the rate of change of error and the output is the
power controlling current component. The inputs variables are calculated at every
sampling time k (Yin et al. 2007).

e(k) − e(k − 1)
e(k) = Vdcref (k) − Vdc (k) , Δe(k) = (7.10)
T
7 Control of Wind Turbine Based on PMSG Using Pitch Angle Control 161

Table 7.2 Rules table for e

DC bus
Δe NB NS NM EZ PS PM PB
NB NB NB NB NB NM NS EZ
NM NB NB NB NM NS EZ PS
NS NB NB NM NS EZ PS PM
EZ NB NM NS EZ PS PM PB
PS NM NS EZ PS PM PB PB
PM NS EZ PS PM PB PB PB
PB EZ PS PM PB PB PB PB

Δe
Pref
d/dt
isqref
×
1/ke
Vdcref ÷
e
+
–

Vdc Ωref

Fig. 7.13 Strategy fuzzy DC bus controller

where Vdcref (k) and Vdc (k) are the reference and measured voltage at time
k, respectively. In Fuzzification stage the crisp variables e(k) and Δe(k) are
converted into fuzzy variables which can be identified by membership function. The
fuzzification maps the error and change in error to linguistic labels of fuzzy sets.
The proposed controller uses following linguistic labels: NB (Negative Big),
NM (Negative Medium), NS (Negative Small), EZ (Zero), PS (Positive Small),
PM(Positive Medium), PB(Positive Big) (Table 7.2).
The activation function of both the output layer are linear and unbiased while
that of the hidden layer is sigmoid and biased (Hafsi et al. 2016).

7.6.1 PI Controller DC Bus

For the regulation of the DC bus controller, we apply the regulator PI between
the reference voltage Vdcref and the voltage measured Vdc as shown in Fig. 7.14
(Slimen et al. 2016).
For the control of the DC bus, it is assumed that the transfer function on the
rectifier side is equal to zero G(s) = 1 detailed on Fig. 7.15 (Slimen et al. 2016).

kpv kiv
Vdc s+
= C C (7.11)
Vdcref k pv kiv
s2 + s+
C C
162 A. Slimen et al.

1 +
kp.s+ki
Vdcref 1
2 – s
Pref
Vdc
Add
Transfer Fcn

Fig. 7.14 DC bus PI controller

1 Vdc
Vdcref kiv
kpv + G(s)=1
s Cs

Fig. 7.15 Transfer function DC bus

The characteristic equation function is a second-order dynamics is of the form:

kpv kiv
s2 + s+ =0 (7.12)
C C
The parameters of the PI regulator can be expressed by the canonical form by
identifying the denominator with:

kpv = 2Cxωp , kiv Cωp2 (7.13)

using the value of the capacitor C = 380 μF and the clean pulse ωp = 60 rad/s
(Slimen et al. 2016).

7.6.2 Simulation Results and Discussion

The simulation show that the combination of PI and Fuzzy Logic DC bus controller.
In order to evaluate the performance of the control system for the variable DC bus
reference concept with PMSG, we set reference to 100 V in the first, next 130 V and
finally 150 V.
Figure 7.16 shows the behavior of the control response of the DC bus voltage by
the fuzzy logic regulator with the desired reference voltage.
Figure 7.17 shows the behavior of the control response of the DC bus voltage by
the PI regulator with the desired reference voltage.
Figures 7.16, 7.17 and 7.18 show the responses of the system when using
respectively IP and fuzzy logic controllers. From Fig. 7.18, showing the comparison
of the two responses; we note that the use of fuzzy provides a response with a smaller
overshoot and faster compared to the conventional PI controller.
7 Control of Wind Turbine Based on PMSG Using Pitch Angle Control 163

200

150
Vdc(V)

100 VdcFL(V)
Vdcref(V)

50

0
0 5 10 15 20 25 30
Time(s)

Fig. 7.16 DC bus control with fuzzy Logic regulator

200

150

100 VdcPI(V)
Vdc(V)

Vdcref(V)
50

-50
0 5 10 15 20 25 30
Time(s)

Fig. 7.17 DC bus control with the regulator PI

The fuzzy control is efficient in terms of reducing the oscillations of monitored

variables compared to the conventional PI control.

7.7 Conclusion

This chapter presents the dynamic model of PSMG wind generation system in
Matlab/Simulink. The model of generator side can capture the maximum power
from variable wind. In the first, we begin with a control design scheme proposed
164 A. Slimen et al.

Fig. 7.18 Comparisons DC 200

bus control with the
regulators PI and fuzzy logic
150

VdcPI(V)
100

Vdc(V)
data2
VdcFL(V)
50
Vdcref(V)

-50
0 5 10 15 20 25 30
Time(s)

for the control of the blade pitch angle in variable speed wind turbine using two
regulator PI and fuzzy logic (Akhmatov 2002). Simulation results demonstrated
good performance for both proposed schemes. In the second part, we control the
DC bus side generator (Yang and Zhang 2013).
By combining the two controllers, pitch angle control and the DC bus control,
using the two tested regulators, PI regulator and fuzzy logic regulator, good
performance to track the power maximum to can feed our pump load.

References

Akhmatov, V. (2001). Note concerning the mutual effects of grid and wind turbine voltage stability
control. Wind Engineering, 25, 367–371.
Akhmatov, V. (2002). Speed wind turbines with doubly-fed induction generators – Part I:
Modelling in dynamic simulation tools. Wind Engineering, 26, 85–108.
Akhmatov, V., Knudsen, H., Nielsen, A. H., & Poulsen, J. (2003). Modelling and transient stability
of large wind farms. International Journal Electrical Power and Energy Systems, 25(2), 123–
144
Anjun, X., Hao, X., Shuju, H., & Honghua, X. (2011). Pitch control of large scale wind turbine
based on expert PID control. In International Conference on Electronics, Communications and
Control (ICECC), Ningbo, China, September 9–11, 2011 (pp. 3836–3839).
Hafsi, S., Dhaoui, M., & Sbita, L. (2016). Advanced control of a PMSG wind turbine. International
Journal of Modern Nonlinear Theory and Application, 5, 1–10.
Hansen, A. D., Sorensen, P., Loy, F., & Blaabjer, F. (2004). Speed wind turbines with doubly-fed
induction. Engineering, 28, 247–263.
Lopes, J. A. P., & Almeida, R. G. (2004). Descrição de Modelos Matemáticos de Máquinas
de Indução Convencional e Duplamente Alimentada e Estratégias de Controlo para Estudo
Dinâmico e de Estabilidade Transitória, Trabalho de Consultoria parao ONS Brasil, INESC
PORTO.
7 Control of Wind Turbine Based on PMSG Using Pitch Angle Control 165

Macêdo, A. V. A., & Mota, W. S. (2012). Wind turbine pitch angle control using fuzzy logic. In
IEEE/PES Transmission and Distribution: Latin America Conference and Exposition (T&D-
LA), Montevideo, Uruguay, September 3–5, 2012 (pp. 1–6).
Miller, N. W., Price, W., & Sanchez-Gasca, J. J. (2003). Dynamic modelling of GE 1.5 and 3.6
wind turbine generators (p. 31). Technical Report, Power Systems Energy Consulting, General
Electric Int., Shenectady.
Mota, W. S. (2006). Simulação de Transitórios Eletromec â Potência (p. 368). Campina Grande,
Epgraf.
Nguen, H. M., & Naidu, S. (2011). Advanced control strategies for wind energy systems. In IEEE
PES, Power Systems Conference & Exposition (PSCE), Phoenix, AZ, USA, March 20–23,
2011.
Rolan, A., Luna, A., Vazquez, G., Aguilar, D., & Azevedo, G. (2009). Modeling of a variable
speed wind turbine with a permanent magnet synchronous generator. In IEEE International
Symposium on Industrial Electronics (ISlE 2009), Seoul, South Korea, July 5–8, 2009 (pp.
734–739).
Simães, M. G., & Shaw, I. S. (2007). Controle modelagem fuzzy (2nd ed., p. 186). SãoPaulo:
Blucher.
Slimen, A., Tlijani, H., Dhaoui, M., & Ben Younes, R. (2016). Study of wind turbine system based
on PMSG with fuzzy-PI controller. In International Journal of Control Theory and Applications
(pp. 111–123).
Slootweg, J. G. (2003). Modelling and Impact on Power System Dynamics. Ph.D. Thesis,
University of Technology, Delft.
Veermani, C., & Mohan, G. (2011). A fuzzy based pitch angle control for variable speed wind
turbines. In IEEE, Student Conference on Research and Development (SCOReD) (pp. 36–39).
Yang, S., & Zhang, L. (2013). Modeling and control of the PMSG wind generation system with a
novel controller. In 3rd International Conference on Intelligent System Design and Engineering
Applications, Hong Kong, China, January 16–18, 2013 (pp. 946–949).
Yin, M., Li, G., & Zhou, M. (2007). Modeling of the wind turbine with a permanent magnet
synchronous generator for integration. In IEEE, Power Engineering Society General Meeting
(pp. 1–6).
Zadeh, L. A. (1965). Fuzzy sets. Information and Control, 8, 338–353.
Chapter 8
Nonlinear Control of a Variable Speed
Wind Energy Conversion System Based
PMSG

Marwa Ayadi and Nabil Derbel

Abstract In this chapter, we present a control system for a variable speed wind
turbine based direct-drive permanent magnet synchronous generator. In order to
capture the optimal power from the wind and to ensure a maximum efficiency
for this system, nonlinear control laws namely backstepping controller and sliding
mode controller have been synthesized. Moreover, a blade pitch angle controller has
been introduced above rated wind speed to keep the generated power at the designed
limit. Finally, to avoid the power fluctuation, the transition between the two control
regions has been investigated.

Keywords Wind turbine · PMSG · Blade pitch control · Robust control ·

Stability

8.1 Introduction

Nowadays, Renewable energy systems receive a great interest in the research

community thanks to economic, social and environmental benefits.
In fact, because of the earth greenhouse effects, the oil price increases, the nuclear
world security and the rapid growth of the global economy, the renewable energy,
which includes geothermal energy, wind energy and photovoltaic energy, etc., has
been widely investigated and rapidly developed (Bull 2001).
Wind energy is considered to be the most competitive and the fastest growing
renewable energy as it is a clean energy source with an exhaustible supply (Burton
et al. 2011).
Several types of wind turbine exist nowadays with variation in the number of
blades, the generator type and the control surface. The most prevalent turbine

M. Ayadi () · N. Derbel

National School of Engineering of Sfax (ENIS), University of Sfax, Sfax, Tunisia
e-mail: [email protected]; [email protected]

© Springer Nature Singapore Pte Ltd. 2019 167

N. Derbel, Q. Zhu (eds.), Modeling, Identification and Control Methods
in Renewable Energy Systems, Green Energy and Technology,
https://doi.org/10.1007/978-981-13-1945-7_8
168 M. Ayadi and N. Derbel

existing in the control literature is the 3-blade, horizontal axis wind turbine
(HAWT).
For several decades, the wind turbine has been operated at a fixed speed. Despite
it has a simple structure and a lower construction and maintenance cost, this type
presents some problems as the poor power quality and the low efficiency. Thus, it
has been replaced by a variable speed wind turbine (VSWT) to enhance efficiency,
reduce audible noise and mechanical stresses (Abulanwar et al. 2016; Burton et al.
2011; Yang et al. 2014b).
For VSWT, the direct− driven permanent magnet synchronous generators
(PMSG) are widely used in small and medium-size wind turbine systems. There
are several advantages of using PMGs for electrical power generations.
First, they can provide high-reliability and high-efficiency, since there are no
copper losses in the rotor circuit and no need for external excitation (Cai et al. 2000).
Second, PMGs with high−power−density are small in size, which reduces the
weight and the cost of the wind turbine system (Michalke et al. 2007).
In addition, the wind turbine with a direct-drive PMSG offers the ability to
remove the gearbox. According to the statistical data reported in Ribrant and
Bertling (2007), failures of gearboxes cause about 19.4% downtime of WTGs.
Without gearboxes, the system needs less maintenance and have a higher reliability
and a reduced downtime.
The overall configuration of a direct-drive PMSG wind turbine shown in Fig. 8.1.
This structure includes a generator control for maintaining the desired rotor speed,
a pitch angle control for limiting the output power, a constant DC link control and a
grid side inverter control for active and reactive power control.
The main challenge in wind energy conversion system (WECS) control is to
extract the optimal accessible power and cope with the nonlinearity of the system.
Therefore, in this chapter, we will focus only on analyzing modern nonlinear control
techniques for the generator-side to achieve maximum performances.
For the generator side, there are typically two control strategies according to the
wind speed velocity.

Wind Turbine Grid

idc imv id
AC ic DC
V
PMSG Vdc

DC AC
is

Wm Vdc
Generator Side Grid Side
Control Control
Pm
Pitch Control
Pref

Fig. 8.1 Process structure

8 Control of Variable Speed Wind Energy System Based PMSG 169

In partial load region, the speed controller can continually adjust the speed of the
rotor to ensure as much as possible energy yield. Thus, the efficiency of the system
will be increased.
For full load zone, the pitch angle regulation is required to keep the generated
power constant.
In wind turbine system, control strategies hold a crucial role to enhance the
efficiency. Without a good control system, a considerable portion of the produced
power is wasted.
Several works of the literature are addressing the issue of controlling the wind
turbine system. The simplest techniques are based on PI controllers and the Perturb
and Observe algorithm (Kesraoui et al. 2011; Mansour et al. 2011). However,
these classical methods show limited performances. Even though neural networks
may guarantee a fast response, the performance of control fails along the system
variation.
Moreover, fuzzy logic approaches have been widely applied and have presented
better effectiveness (Bahraminejad et al. 2014; Jerbi et al. 2009; Simoes et al. 1997),
but, the major problems are the strict theoretical analysis and the lack of stability.
A backstepping approach has been studied and developed to the control of the
PMSG (Karabacak and Eskikurt 2012; Wang et al. 2013; Yang et al. 2014a). This
technique provides a good performance result. Sliding mode control emerges also as
a suitable option to deal with such complex system and it has been applied in several
works. Lee et al. (2010) propose the sliding mode control scheme to maximize the
generated power below the rated wind speed, Results demonstrated by research (Vu
et al. 2013; Zhang et al. 2013) involving the use of SMC with PMSG continue to
prove the strength of SMC as a viable strategy for effective control of synchronous
machines
Although the positive attributes of SMC, the phenomenon known as chattering
gives an understandable level of criticism. Because of the nonidealities in switching
devices, the response of the system oscillates with high frequencies about the desired
reference, known as the sliding surface which leads to low control accuracy and even
damages of mechanical parts.
In order to face this undesirable phenomenon, the literature proposes different
improvement techniques Such as the approach presented in Kachroo and Tomizuka
(1996) which places a boundary layer around the switching surface such that the
relay control is replaced by a saturation function. Another class of techniques
is based on introducing high order sliding mode algorithms for speed control
(Feng et al. 2012; Valenciaga and Puleston 2008). In Evangelista et al. (2013), the
speed tracking was achieved using a super-twisting sliding mode control based on
Lyapunov techniques.
From this point of view, The main objective of this paper is to study the control
algorithms for a variable speed wind turbine based permanent magnet synchronous
generator.
In order to achieve a maximum power capture efficiency for the system in safety
conditions, a nonlinear control strategy based on backstepping controller and sliding
170 M. Ayadi and N. Derbel

mode controller has been proposed. Furthermore, a blade pitch angle controller is
proposed to enhance system performances in high wind speed region.
The rest of the paper is organized as follow. Section 8.2 presents the system
modeling. The control technique applied to the wind system is introduced in
Sect. 8.3. The Control below rated wind speed: MPPT zone is presented in Sect. 8.4.
Section 8.5 describes the control above rated wind speed (the pitch control). The
transition between the two control regions is detailed in Sect. 8.6. Finally, some
conclusion remarks are drawn in the last Section.

8.2 Modeling of the Wind Turbine System

8.2.1 Wind Turbine Modeling

Consider a horizontal axis wind turbine (HAWT). The potentially available power
by the wind is defined as:

1
Pw = πρR 2 V 3 (8.1)
2
where:
• R denotes the blade radius of the WT(m)
• ρ is the air density (1.25 kg/m in normal atmosphere)
• V is the wind speed (m/s)
However, this expression can only stand if the wind passes through the swept area
of the wind turbine.
In 1919, A German physicist Albert Betz state that no more than 59% of the
kinetic energy, contained in a steam tube that shares the same cross-section as the
disc, can be converted to useful works.
Figure 8.2 presents a schematic of fluid flow which helps to contextualize the
statement.

Fig. 8.2 Schematic of fluid

flow through a wind turbine
8 Control of Variable Speed Wind Energy System Based PMSG 171

Thus, a power coefficient term is introduced and can be understood as:

mechanical power captured by the wind turbine

Cp = (8.2)
maximum power available in the wind

Cp represents the performance of the wind turbine. It depends on the tip-speed ratio
(TSR) λ and the pitch angle β which is the angle between the orientation of the
blade and the wind velocity vector:
  −12.5
116
Cp (λ, β) = 0.22 − 0.4β − 5 e λi
λi
1 1 0.035
= − 3 (8.3)
λi λ + 0.08β β +1

with
Ωt R
λ= (8.4)
V
where Ωt is the turbine speed
The nonlinear relation between Cp , λ and β is presented in Fig. 8.3.

0.45
β=0°
0.4
β=2
0.35 β=4
β=6
0.3 β=8

0.25 β=10
Cp

0.2

0.15

0.1

0.05

0
0 1 2 3 4 5 6 7
λ

Fig. 8.3 Aerodynamic power coefficient variation Cp against the speed ratio λ and the pitch
angle β
172 M. Ayadi and N. Derbel

It is obvious that Cp is maximised when β =0 and it decreases with the increase

of β. For a constant pitch angle, there is a unique peak of the curve corresponding
to Cpmax .
The aerodynamic torque is given by:

Pm 1 R2 V 3
Tm = = π ρ Cp (λ, β) (8.5)
Ωt 2 Ωt

8.2.2 PMSG Modeling

The mathematical model of the PMSG in the d-q reference is Kim et al. (2010):
⎧
⎪ di
⎨ Ud = Rs Id + Ld d − Ωe Φq
dt (8.6)
⎪ di
⎩ Uq = Rs Iq + Lq q + Ωe Φd
dt

where Rs is the stator resistance. Φd and Φq are the d and q axis flux linkage,
respectively:

Φd = Ld Id + Φm (8.7)
Φq = Lq Iq

Furthermore, consider Lq = Ld = Ls denotes the inductance of the stator (a PMSG

with a nonsalient pole is considered in this study).

8.3 Control Technique

Wind Turbine systems present intense nonlinearities, and are subjected to distur-
bances and parameter uncertainties. Such factors transform their control to a hard
task.
Therefore, a high level control is often required in order to ensure a certain
convergence of the system to the requested state condition and to operate at
maximum power coefficient over a wide range of wind speeds.
In fact, wind turbines should be driven differently according to the speed of the
wind.
Therefore, the WT generator needs to be controlled to operate in three different
modes as shown in Fig. 8.4.
• Parking Mode (I): The wind speed is lower than the cut-in speed which is (2 m/s)
in this system, the wind turbine stays in a parking status and does not rotate.
8 Control of Variable Speed Wind Energy System Based PMSG 173

Parking
Mode (I) MPPT Mode (II) Constant power Mode (III)
Power (W)

cut in wind speed rated wind speed

t (s)

Fig. 8.4 Power versus wind speed characteristic

• MPPT (Maximum Power Point Tracking) mode (II): The system starts to work
and generates electrical power. The MPPT is applied until arrived to the rated
wind speed, (12.7 m/s), for this case-study system. The blade angle is usually
controlled at zero degree to maximize the wind energy capture.
• Constant power Mode (III): It is necessary to control the extracted power above
rated wind value by adjusting properly the blade pitch angle of the wind turbine
to keep the power at constant power region, as its name implies.

8.4 Control Below Rated Wind Speed: MPPT Zone

In this region, the wind speed is below the rated value. The objective of the controller
is to maximize the wind energy capture and to achieve the Maximum Point Power
Tracking (MPPT). For that, The tip speed ratio λ is set to its optimal value λopt
which contributes to the power coefficient peak Cpmax . Therefore, the optimum
rotor speed of the Wind turbine can be written as follows:

λopt v
Ωopt = (8.8)
R
174 M. Ayadi and N. Derbel

Thus, the maximum power output of the wind turbine can be expressed as:

Pmax = kopt Ωopt

3
(8.9)
 3
1 R
kopt = Cpmax ρπ R 2
2 λopt

In addition, to achieve the MPPT operation mode, the pitch angle is kept at zero
degree to ensure a high efficiency of energy capture.
The optimum rotor speed given in Eq. (8.8) is considered as the speed reference
of the PMSG. Consequently, by using a robust control technique the generator
speed will reach its reference value in the static state, and then the MPPT control is
achieved.
In this work, the control system proposed to command the wind turbine generator
is the Field Oriented Control (FOC), which controls the torque indirectly by
adjusting the stator currents. In this method, the torque angle is maintained at
90◦ . The d-axis current is set to zero and then the stator current has only the q-
axis component. Thus, the torque depends only on the q-axis stator current. This
technique removes current cross coupling between the d and q axes components.
Based on the FOC technique, different controller design have been investigated
in this work to the PMSG to optimize the wind turbine power output.
• Proportional Integral (PI) control
• backstepping control
• Sliding mode control
Remark 1 For the control design, the following assumptions are required:
Assumption 1 Variables iq , id and Ω are available.
Assumption 2 The reference speed Ωref is determined referring to the MPPT
algorithm.
Assumption 3 The d-axis reference current should be equal to zero id = 0.

8.4.1 Vector Control Based PI Controlller

8.4.1.1 Controller Design

This technique is the most commonly applied algorithm for the PMSG control
(Ayadi and Derbel 2017). It involves two control loops:
• two inner current loops are used to adjust the generator currents id and iq
allowing the control of the flux and the torque, respectively.
• an external mechanical loop in order to ensure the speed regulation.
8 Control of Variable Speed Wind Energy System Based PMSG 175

Then, to improve the dynamic response, compensation voltage, ΔU q and ΔU d are

added into the control law.

ΔU d = Ωe iq Ls
ΔU q = Ωe iq Ls + Ωe Φm (8.10)

• d-axis current controller design

The control of this component aims to keep the stator flux constant. Its reference
value is set to zero.
1
id Rs
= (8.11)
Ud 1 + s LRds

The PI transfer’s function is:

 
1 + τd s
G(s) = kd (8.12)
τd s

Based on the compensation pole method, the synthesis of this PI can be determined
R2
τd = LRds , Kd = Lds , and the closed loop transfer function between the actual and
the desired direct current is:
1
Fd (s) = Ls
(8.13)
1+sR s

• q-axis current controller design

The control of the quadrature current iq affects directly the electromagnetic torque.
The reference value of iq is determined from the speed control loop. The control law
Ls
is implemented in the same way as the d-axis current controller yielding: τq = R s
Rs2
and Kq = Ls .
• Speed control loop
The reference value of this loop is the optimal speed Ωopt determined based on the
MPPT strategy Eq. (8.8). The transfer function between Ωr and iq is given by:

3
iq pΦm
= 2 (8.14)
Ωr f + sJ
176 M. Ayadi and N. Derbel

Table 8.1 Turbine and Wind turbine parameters PMSG parameters

PMSG parameters
3 blades Pnom = 25 KW
Variable speed Rs = 0.82 m
R = 4.5 m Lsd = Lsq = 15.1 mH
Cpmax = 0.42 φ = 0.5 Wb
Jtot = 99 e-4 f = 1e-3
ρ = 1.225 kg/m p=4

The PI transfer’s function is:

 
1 + τΩ s
GΩ (s) = kΩ (8.15)
τΩ s

Thus, the closed-loop transfer function is:

1 + τΩ s
FΩ (s) = f
(8.16)
1 + (τΩ + 3 )s + 3
J
s2
2 pφm KΩ 2 pφm KΩ

The output of the PI represents the quadrature reference current īq .

8.4.1.2 Simulation Results

Results have been presented, using numerical simulations carried under the Matlab-
SIMULINK tool to test the control strategy and to evaluate the performance of the
system.
Characteristics of the turbine and the Permeant Magnetic Synchronous Machine
are presented in Table. 8.1.
System simulation has been performed considering an aleatory wind profile to
evaluate the behavior of the system similar to real conditions.
As shown in Fig. 8.4, for different regions of the wind speed, there is a specific
control strategy. Therefore in this part, The wind speed will be limited to (11.5 m/s)
as we are interested in the control below rated wind speed region(II).
• Figure 8.5a presents the wind speed profile.
• Figure 8.5b shows that the power coefficient Cp is maintained close to its
maximum value Cpmax = 0.41. Thus, the Maximum Power Point Tracking
MPPT has been achieved.
• The system response under PI is shown in Fig. 8.6.
Figure 8.6a shows the rotational speed response. It converges to its reference
track which is supplied from the Maximum Power Point Tracking (MPPT)
controller. Figure 8.6b presents the rotational speed error which exceeds the
0.2(rad/s). The direct and the quadrature currents are shown in Fig. 8.6c, e respec-
tively. It is obvious from Fig. 8.6d, f that current errors present a considerable
distorted values due to the nonlinear characteristics of the system
8 Control of Variable Speed Wind Energy System Based PMSG 177

a b
12 0.5

0.4
11

0.3
v(m/s)

Cp
10
0.2

9
0.1

8 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
t(s) t(s)

Fig. 8.5 wind speed profile

This control laws gives good dynamic response of the speed tracking and remarkable
rejection of disturbances.
However, it should be noted that it is better to minimize as possible the tracking
error, hence we propose two non-linear regulators namely the backstepping control
and the sliding mode control in the sequel.

8.4.2 Backstepping Controller

8.4.2.1 Controller Design

This control approach is designed in such a way it keeps the same general structure
of the vector control while ensuring at the same time the regulation and the
limitation of the currents. The objective of the nonlinear backstepping controller
is to track the speed of PMSG with the choice of appropriate regulated variables
(Ayadi and Derbel 2017). Based on the PMSG model given in Eq. (8.6), The control
design can be determined by following three steps as shown in Fig. 8.7.
• Step 1: Control design based on rotational speed dynamics:
The speed error is:

eΩ = Ω̄r − Ωr (8.17)

The error dynamics derived can be obtained as:

Tm 3p B
ėΩ = Ω̄˙ r − + Φm iq + Ωr (8.18)
J 2J J
Consider the following Lyapunov function candidate:
1 2
VΩ = e (8.19)
2 Ω
178 M. Ayadi and N. Derbel

a b
20 1

15 0.5
Ωr(rad/s)

ew
10 0

Ωref
5 Ωr −0.5

0 −1
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
t(s) t(s)

c d
0.5 0.2
id
0.4 0.1
idref
0.3 0

0.2 −0.1
ed
id

0.1 −0.2

0 −0.3

−0.1 −0.4

−0.2 −0.5
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
t(s) t(s)

e f
700 20

600
10
500

400
eq

0
iq

300

200
−10
iqref
100
iq
0 −20
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
t(s) t(s)

Fig. 8.6 Simulation results of the FOC based PI controller. (a): rotational speed, (b): Speed error,
(c):direct current, (d): direct current error, (e):quadrature current, (f):quadrature current error

Its derivative is:

 
Tm 3p B
V˙Ω = eΩ ėΩ = eΩ Ω̄˙ r − + Φm iq + Ωr (8.20)
J 2J J
8 Control of Variable Speed Wind Energy System Based PMSG 179

Fig. 8.7 Internal structure of the backstepping controller block

The direct and quadrature current id and iq are chosen as virtual control elements:
⎧
⎨ īd = 0
2J  ˙ Tm
 (8.21)
⎩ īq = −Ω̄r + J − B
J Ωr − KΩ eΩ
3pΦm

where KΩ > 0
Thus:

V˙Ω = −KΩ eΩ
2
(8.22)

• Step 2:Current components control design

The current error can be determined as:

ed = īd − id
eq = īq − iq (8.23)
180 M. Ayadi and N. Derbel

The error dynamics derived from Eqs. (8.18), (8.21) and (8.23) are:

Rs 1
ėd = id − piq Ωr − Ud
Ls Ls
   
2J ¨ B Rs Φm 1
ėq = −Ω̄r − Ω̇r − KΩ eΩ + iq + p id + Ωr − Uq
3pΦm J Ls Ls Ls
3pΦm
ėΩ = − eq − KΩ eΩ
2J

• Step 3: Control laws design

The Lyapunov function is determined as:

1 2
V2 = (e + ed2 + eq2 ) (8.24)
2 Ω
Its time derivative is:
3pΦm
V̇2 = (eΩ ėΩ + ed ėd + eq ėq ) − KΩr eΩ
2
r
− Kd ed2 − Kq2 eq − eΩ eq
2J
  
2J B Rs Φm
+eq −Ω̄¨ r − Ω̇r − KΩ eΩ + iq + p(id + )Ωr
3pΦm J Ls Ls
  
1 Rs 1
− Uq + Kq eq +ed id − piq Ωr − Ud + Kd ed (8.25)
Ls Ls Ls

The control inputs which stabilize The current tracking errors dynamics are:

v̄d = Rs id − Ls piq Ωr + Ls Kd ed (8.26)

 
2J Ls B
v̄q = −Ω̄¨ r − Ω̇r − KΩ eΩ + Rs iq + p(id Ls + Φm )Ωr + Kd Ls ed
3pΦm J

This choice yields:

V̇2 = −KΩ eΩ
2
r
− Kd ed2 − Kq eq2 ≤ 0 (8.27)

where Kd and Kq are positive gains.

8.4.2.2 Simulation Results

The same wind speed profile is applied in this part with varying degrees of wind
speed between (8 m/s) and (11.5 m/s).
8 Control of Variable Speed Wind Energy System Based PMSG 181

a b
20 0.15
PI
Backstepping
15 0.1

speed error(%)
Ωr (rad/s)

10 0.05

5 0
PI
Ωref
Backstepping
0 −0.05
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
t(s) t(s)

c d
3 0.01
PI
idref
2 Backstepping 0
ed(%)
id (A)

1 −0.01

0 −0.02
PI
Backstepping
−1 −0.03
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
t(s) t(s)
e f
800 5
PI
4 Backstepping
600
3
eq(%)
iq (A)

400 2

1
200 PI
iqref 0
Backstepping
0 −1
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
t(s) t(s)

Fig. 8.8 Simulation results

• The system response under PI controller and nonlinear backstepping control has
been presented in Fig. 8.8.
Figure 8.8a presents the rotational speed response which converges perfectly
to its reference track. Figure 8.8b shows the speed error. Its is neglected for the
backstepping control compared to PI controller. Figures 8.8c, e illustrate the
direct and the quadrature currents, respectively. The evolution of the currents
errors (Fig. 8.8d, f) shows that the backstepping control has a good dynamic
response.
182 M. Ayadi and N. Derbel

8.4.3 Sliding Mode Control

8.4.3.1 Controller Design

The sliding mode control has proved to be an efficient technique able to cope with
aforementioned complex characteristics. In this context, this technique is applied
in this section to design a robust controller for nonlinear multi-input multi-output
permanent magnet synchronous generator based WECS. The controller design is
ensured by the following steps.
1. Sliding surfaces:
Based on the system model, the sliding surfaces are to be defined as:

Sd (t) = ed (t) + λ1 ed (t)dt (8.28)

Sq (t) = eq (t) + λ2 eq (t)dt (8.29)

SΩr (t) = eΩr (t) + λ3 eΩr (t)dt (8.30)

where ed (t) = id (t) − idref (t), eq (t) = iq (t) − iqref (t) and eΩr (t) = Ωr (t) −
Ωrref (t).
The integral action has been included to overcome the state error and to
improve the sliding surface where λ1 , λ2 , λ3 are the integral gains.
2. Reachability:
The reachability condition for Eqs. (8.28), (8.29) and (8.30) are defined
respectively as:

Sd (t)Ṡd (t) < 0 (8.31)

Sq (t)Ṡq (t) < 0 (8.32)
SΩr (t)ṠΩr (t) < 0 (8.33)

3. d-axis current control design

The control of id component current aims to keep the stator flux constant. Its
reference value is set to zero.
From Eq. (8.21), the inequality Eq. (8.31) can be rewritten as:
  
Rs 1
Sd (t) λ1 − id + piq Ωr + Ud < 0 (8.34)
Ls Ls

The d-axis control law is given by:

Ud (t) = Udeq (t) + ΔUd (8.35)

8 Control of Variable Speed Wind Energy System Based PMSG 183

with the equivalent controlUdeq is:

Udeq (t) = (Rs − λ1 Ls )id − piq Ωr Ls (8.36)

The switching control ΔUd is defined as:

ΔUd = Kd sign[Sd (t)] (8.37)

where Kd is a positive constant

4. q-axis current control design
Based on the system model and following the similar method, Eq. (8.32)
becomes:
    
Rs Φm 1
Sq (t) λ2 + i q − p id + Ωr + Uq + λ2 iqref < 0
Ls Ls Ls

Denoting the control law as:

Uq (t) = Uqeq (t) + ΔUq (8.38)

The equivalent control can be given as:

diqref
Uqeq = (Rs + Ls λ2 )iq + p(Ls id + φm )Ωr − λ2 Ls iqref − Ls (8.39)
dt
where iqref is a dynamic value that will be determined as a resulting output of
the rotational speed control.
The switching portion of the Uq (t) can be expressed by:

ΔU q = Kq sign[Sq (t)] (8.40)

where Kq is a positive constant

5. Rotational speed dynamics control
This controller should ensure that the rotational speed. It is driven to its
reference determined by the MPPT algorithm. Considering the rotational speed
motion equation, the inequality Eq. (8.33) becomes:
   
Tm 3 B
SΩr (t) − pΦm iq + λ3 − Ωr − λ3 Ωref < 0 (8.41)
J 2J J

The control variable for the rotational speed controller is the quadrature axis
reference current which is designed as:

iqref (t) = iqeq,ref (t) + Δiqref (8.42)

184 M. Ayadi and N. Derbel

The equivalent control law is given as:

2 ! "
iqref = Tm + (λ3 J − B)Ωr − λ3 Ωref (8.43)
3pφm

It then follows that:

Δiqref = Kiq sign[SΩr (t)] (8.44)

where Kiq is a positive constant

In order to reduce the chattering effect, a smooth control discontinuity is
introduced around the switching surface. The sign [S(t)] is replaced by sat[S(t),Δ]
which can be expressed as:
 S(t)
 if |S(t)| ≤ Δ
sat[S(t)] =  Δ (8.45)
sign(S(t)) otherwise

where Δ is the boundary layer thickness.

8.4.3.2 Simulation Results

The system response under the Sliding Mode concept has been presented in Fig. 8.9.
The rotational speed Ωr (t) is shown in Fig. 8.9a. The second result shown is
that of id (t). As it can be seen by Fig. 8.9c, the d-axis current accurately tracks the
reference value idref = 0. Figure 8.9e. shows the actual value of iq (t), accurately
tracking the defined reference.

8.5 Control Above Rated Wind Speed: Pitch Control

8.5.1 Introduction

The key part of a variable-speed wind turbine is the pitch system. Its main purpose
is to keep the generated power constant above the rated wind speed.
In fact, when the wind speed becomes larger than the rated value and the MPPT
algorithm is still applied, the power generated by the system will exceed the optimal
power if the blade angle β stays fixed at zero degrees.
This will make the power devices and the generator work in a higher zone than
optimum output, which is harmful to the mechanical system if sustained for a period
of time.
Based on this concern, a control system for the regulation of the pitch angle
according to different wind conditions is required for the wind turbine.
8 Control of Variable Speed Wind Energy System Based PMSG 185

a b
20 20

15
15

10

ew
Ωr

10
5

5
0

0 −5
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
t(s) t(s)

c d
0.6 0.2

0.4 0
ed

0.2 −0.2
id

0 −0.4

−0.2 −0.6
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
t(s) t(s)
e f
700 800

600
600
500

400 400
eq
iq

300 200
200
0
100

0 −200
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1

Fig. 8.9 Simulation results of the Sliding mode strategy

Small changes in the blade angle can have dramatic effects on the output power.
The advantages of the pitch control system might be summarized as follows:
1. Optimize the output power of the wind turbine. Indeed, to give the maximum
power in partial load region, the pitch setting should be at its optimum value
2. Prevent input mechanical power to exceed the design limits. This controller
provides a good regulation of the aerodynamic power and the loads produced
by the rotor.
186 M. Ayadi and N. Derbel

3. Minimize the fatigue loads of the turbine mechanical components. The design of
the controller must take into account the load.
Thus, the blade pitch angle control should have a good dynamic quality and a strong
robustness.
Several works on pitch control have been done to improve the quality of the
generated power. For example, a gain scheduling controller, proposed in Lescher
et al. (2005), changes the controller gain with the variation of the wind speed.
However, the practical implementation of the gain scheduling is very difficult
because the wind speed is usually measured on the tower and does not represent
the wind speed at the turbine plant. Disturbance accommodating control (DAC) has
been developed in Stol et al. (2000, 2003), but it presents a large estimation error
when the system operates far away from the rated wind speed. H∞ controllers for
blade pitch control and electromagnetic torque control have been used in Rocha et al.
(2003). However, this controller has not been experienced on a nonlinear turbine
model.
Based on the analysis of the above studies, we propose two types of controllers
to solve the variable pitch control problem of a direct-drive PMSG wind turbine
(Ayadi et al. 2017):
• Pitch Control based on a Proportional Integrator controller
• Pitch Control based on a Sliding Mode controller

8.5.2 System Description

The blade pitch control system is the key part of a variable speed wind turbine. It
acts directly on the system responses for different wind speed zones (Fig. 8.10).
To regulate the blade angle, the power deviation from its reference value is used.
An electromechanical system called actuator is then used to put the blade angle
into the desired position. It can be modeled as a first-order system as Hwas and
Katebi (2012):

Fig. 8.10 Schematic

illustration of the Kinematics
of the WT
8 Control of Variable Speed Wind Energy System Based PMSG 187

k0
G(s) = (8.46)
1 + τs

k0 is the proportional gain and τ is the time constant.

According to industrial recommendations, it is necessary to include a pitch rate
adjusted to 10◦ /s

8.6 Transition Between the Two Control Regions

8.6.1 Switched Wind Turbine System

Variable speed wind turbines operate within a large range of wind speeds. During
the power generation and depending on the prevailing wind speed, the wind turbine
switches between two operation modes as shown in Fig. 8.4: the low wind speed
region (also called Region II) where the control objective is to capture the maximum
power from the wind and the high wind speed region (also called Region III) where
the control aim is to limit the output power to the required power.
The control of a variable speed wind turbine is usually achieved based on two
distinct control laws for Regions (II) and (III), respectively. The switching or
called also transition between these two regions cause undesirable effects such as
vibrations in drive train, tower and blades (Pao and Johnson 2009) which cause
system fatigue and premature failure.
Therefore, it is necessary to develop a control algorithm to smooth out the
undesirable effects of the switching actions and also to stabilize the switching
operation.
The synthesis of the blade angle controller is divided typically into two zones
according to the wind speed value:
• Partial load zone: v < vref −→ β = βref = 0
• Full load zone: v ≥ vref −→ β changes with the wind variation to keep P =
Popt
There are usually two controllers for the variable-speed wind turbines which are
cross-coupled each other, shown as in Fig. 8.11.
In order to smooth the wind power fluctuations in the intermediate zone, a fuzzy
logic controller has been investigated.

8.6.2 Simulation Results

To validate the stability of the system when it switches between Region (II) and
Region (III) modes, simulations have been performed on a system with the following
conditions:
1. Maximum Power Coefficient (Cpmax ) = 0.4109
188 M. Ayadi and N. Derbel

Power
wind Wind Turbine Power
and generator Converter

Rotor speed
β
Power Set point

Cross Coupling
Pitch Speed
Controller Controller

Fig. 8.11 Structure of a model of variable-speed wind turbine

a b
x 104
25 3

2.5
20
Generated Power (W)

2
Wind Speed

15 Region
1.5 II
Region III
10
1

5 Rated wind speed =11.7 0.5

0 0
0 10 20 30 40 50 0 10 20 30 40 50
Time(s) Time(s)
c d
x 104
30
2.8
’zoom in ’ of power curve

20 2.7
Beta

2.6
10

2.5
0
2.4

−10 2.3
0 10 20 30 40 50 20 30 40
Time(s) Time(s)

Fig. 8.12 Simulation results in full range of operation

2. Rated wind speed Vm = 11.7

3. Nominal Demand Power (Pmax ) = 2.5 kW
Figure 8.12a presents the wind speed profile which involves the two operational
areas, i.e., the low wind speed area (V < 11.7 m/s) and the high wind speed area (V
≥ 11.7 m/s).
8 Control of Variable Speed Wind Energy System Based PMSG 189

a b
30 30

25 25
P(kW)

P(kW)
20 20

15 15
10 20 30 40 50 60 10 20 30 40 50 60
t(s) t(s)

Fig. 8.13 Simulation in the transition region

The power characteristic is shown in Fig. 8.13b. The turbine starts operating
when the wind speed exceeds cut-in wind speed 3 m/s. In this region (II), The MPPT
algorithm and the torque control come into action with a constant pitch angle.
When the wind speed exceeds its nominal value (region III), the control objective
shifts from maximizing the power capture to regulate the power to the turbine rated
output.
Figure 8.12c shows the pitch angle variation. Below the rated wind speed the
blade pitch angle is set at 0. It is the optimal value that allows the turbine to extract
the maximum energy from the incident wind. Otherwise, the pitch control is active:
the angle changes with the variations of the wind to avoid over rated power.
Figure 8.12d presents a zoomed in section of Fig. 8.13b. It is obvious that
Between the high and the low wind speed region, the curve of the output power
presents a noticeable fluctuation caused by the transition between the two regions
which causes system fatigue and premature failure.
As a solution, a fuzzy logic controller has been applied with the aim of smoothing
the wind power fluctuations in the intermediate zone. Figure 8.13a presents the wind
speed profile with varying degrees. Figure 8.13b presents the power coefficient Cp .
It can be observed that the Cp curve is maintained close to its maximum value
Cpmax = 0.41 which proves that the MPPT has been achieved. In region (III) as
the power coefficient depends on the blade pitch angle, it varies with the blade
variations.
Figure 8.13c presents the generated power for the 2 control zones: It is remark-
able that the fluctuation between the intermediate zone has been eliminated. The
wind speed moves from zone to another fluently. The variation of the pitch angle
is illustrated in Fig. 8.13d. The pitch angle controller shows a perfect capability
of giving to the actuator the right angle demand in different wind speed zones.
The rotational speed is presented in Fig. 8.13e, (in region II) the speed is varying
to optimal power extraction, overrated wind speed, the speed control input is
considered as its maximum value.
190 M. Ayadi and N. Derbel

8.7 Conclusion

This chapter discusses the control systems used for a variable speed wind turbine
equipped with PMSG. In the first part, a control algorithm has been synthesized
to extract the maximum of the power based on two nonlinear control techniques
namely the backstepping controller and the sliding mode controller. Comparative
studies between these algorithms have been presented. Then, above rated wind
speed, a pitch angle regulation has been investigated to keep the generated power at
the designed limit.
Finally, the transition between the two control regions has been discussed to
avoid the power fluctuation.

References

Abulanwar, S., Hu, W., Chen, Z., & Iov, F. (2016). Adaptive voltage control strategy for variable-
speed wind turbine connected to a weak network. IET Renewable Power Generation, 10(2),
238–249.
Ayadi, M., & Derbel, N. (2017). Nonlinear adaptive backstepping control for variable-speed wind
energy conversion system-based permanent magnet synchronous generator. The International
Journal of Advanced Manufacturing Technology, 92(1–4), 39–46.
Ayadi, M., Ben Salem, F., & Derbel, N. (2017). Power control of a variable speed wind turbine
based on direct torque control of a permanent magnet synchronous generator. International
Journal of Digital Signals and Smart Systems, 1(3), 204–223.
Bahraminejad, B., Iranpour, M. R., & Esfandiari, E. (2014). Pitch control of wind turbines using
IT2FL controller versus T1FL controller. International Journal of Renewable Energy Research
(IJRER), 4(4), 1065–1077.
Bull, S. R. (2001). Renewable energy today and tomorrow. Proceedings of the IEEE, 89(8), 1216–
1226.
Burton, T., Jenkins, N., Sharpe, D., & Bossanyi, E. (2011). Wind energy handbook (2nd ed.). John
Wiley & Sons.
Cai, W., Fulton, D., & Reichert, K. (2000). Design of permanent magnet motors with low torque
ripples: A review. International Conference on Electrical Machines (pp. 1384–1388).
Evangelista, C., Puleston, P., Valenciaga, F., & Fridman, L. M. (2013). Lyapunov-designed super-
twisting sliding mode control for wind energy conversion optimization. IEEE Transactions on
Industrial Electronics, 60(2), 538–545.
Feng, Y., Chen, B., Yu, X., & Yang, Y. (2012). Terminal sliding mode control of induction
generator for wind energy conversion systems. In IECON 2012: 38th Annual Conference on
IEEE Industrial Electronics Society (pp. 4741–4746). IEEE.
Hwas, A., & Katebi, R. (2012). Wind turbine control using PI pitch angle controller. IFAC
Proceedings Volumes, 45(3), 241–246.
Jerbi, L., Krichen, L., & Ouali, A. (2009). A fuzzy logic supervisor for active and reactive power
control of a variable speed wind energy conversion system associated to a flywheel storage
system. Electric Power Systems Research, 79(6), 919–925.
Kachroo, P., & Tomizuka, M. (1996). Chattering reduction and error convergence in the sliding-
mode control of a class of nonlinear systems. IEEE Transactions on Automatic Control, 41(7),
1063–1068.
8 Control of Variable Speed Wind Energy System Based PMSG 191

Karabacak, M., & Eskikurt, H. I. (2012). Design, modelling and simulation of a new nonlinear
and full adaptive backstepping speed tracking controller for uncertain PMSM. Applied
Mathematical Modelling, 36(11), 5199–5213.
Kesraoui, M., Korichi, N., & Belkadi, A. (2011). Maximum power point tracker of wind energy
conversion system. Renewable Energy, 36(10), 2655–2662.
Kim, H.-W., Kim, S.-S., & Ko, H.-S. (2010). Modeling and control of PMSG-based variable-speed
wind turbine. Electric Power Systems Research, 80(1), 46–52.
Lee, S.-H., Joo, Y.-J., Back, J., & Seo, J. H. (2010). Sliding mode controller for torque and pitch
control of wind power system based on pmsg. In IEEE International Conference on Control
Automation and Systems (ICCAS) (pp. 1079–1084).
Lescher, F., Zhao, J. Y., & Borne, P. (2005). Robust gain scheduling controller for pitch regulated
variable speed wind turbine. Studies in Informatics and Control, 14(4), 299.
Mansour, M., Mansouri, M., & Mmimouni, M. (2011). Study and control of a variable-speed wind-
energy system connected to the grid. International Journal of Renewable Energy Research,
1(2), 96–104.
Michalke, G., Hansen, A. D., & Hartkopf, T. (2007). Control strategy of a variable speed wind
turbine with multipole permanent magnet synchronous generator. In European Wind Energy
Conference and Exhibition.
Pao, L. Y. & Johnson, K. E. (2009). A tutorial on the dynamics and control of wind turbines and
wind farms. In IEEE American Control Conference, ACC’09 (pp. 2076–2089).
Ribrant, J., & Bertling, L. (2007). Survey of failures in wind power systems with focus on swedish
wind power plants during 1997–2005. In Power Engineering Society General Meeting. IEEE,
(pp. 1–8). IEEE.
Rocha, R., Filho, L. S. M. (2003). A multivariable H∞ control for wind energy conversion system.
In IEEE Conference on Control Applications, CCA (pp. 206–211).
Simoes, M. G., Bose, B. K., Spiegel, R. J. (1997). Fuzzy logic based intelligent control of a variable
speed cage machine wind generation system. IEEE Transactions on Power Electronics, 12(1),
87–95.
Stol, K., Rigney, B., & Balas, M. (2000). Disturbance accommodating control of a variable-speed
turbine using a symbolic dynamics structural model. In 2000 ASME Wind Energy Symposium
(p. 29).
Stol, K. A., Balas, M. J., et al. (2003). Periodic disturbance accommodating control for blade load
mitigation in wind turbines. Transactions-American Society of Mechanical Engineers Journal
of Solar Energy Engineering, 125(4), 379–385.
Valenciaga, F., & Puleston, P. (2008). High-order sliding control for a wind energy conversion
system based on a permanent magnet synchronous generator. IEEE Transactions on Energy
Conversion, 23(3), 860–867.
Vu, N. T.-T., Yu, D.-Y., Choi, H. H., & Jung, J.-W. (2013). T–S fuzzy-model-based sliding-mode
control for surface-mounted permanent-magnet synchronous motors considering uncertainties.
IEEE Transactions on Industrial Electronics, 60(10), 4281–4291.
Wang, G.-D., Wai, R.-J., & Liao, Y. (2013). Design of backstepping power control for grid-side
converter of voltage source converter-based high-voltage DC wind power generation system.
IET Renewable Power Generation, 7(2), 118–133.
Yang, F., Li, S.-S., Wang, L., Zuo, S., & Song, Q.-W. (2014a). Adaptive backstepping control
based on floating offshore high temperature superconductor generator for wind turbines. In
Abstract and applied analysis. Hindawi Publishing Corporation.
Yang, W., Tavner, P. J., Crabtree, C. J., Feng, Y., & Qiu, Y. (2014b). Wind turbine condition
monitoring: Technical and commercial challenges. Wind Energy, 17(5), 673–693.
Zhang, X., Sun, L., Zhao, K., & Sun, L. (2013). Nonlinear speed control for PMSM system using
sliding-mode control and disturbance compensation techniques. IEEE Transactions on Power
Electronics, 28(3), 1358–1365.
Chapter 9
Direct Power Control of DFIG Using
Sliding Mode Control Approach

Aicha Daoud and Nabil Derbel

Abstract This paper investigates direct active and reactive power control for a
variable-speed wind energy conversion system with doubly fed induction generator
(DFIG). The stator windings of DFIG are directly connected to the grid whereas
rotor windings are connected to an inverter. A nonlinear sliding mode control (SMC)
strategy is employed to control the instantaneous active and reactive powers of
DFIG. Simulation results show the effectiveness of the proposed control law.

Keywords Wind energy conversion system · Doubly fed induction generator ·

Direct power control(DPC) · Sliding mode control (SMC)

9.1 Introduction

Over the last decade, wind energy conversion systems have rapidly evolved,
wherefore, efficient and reliable exploitation tools are necessary to make these
installations more profitable (Cheng 2014). Several types of wind energy conversion
system subsist. The most popular is based on the control of the doubly fed induction
generator (DFIG). The benefits of DFIG consist in low converter cost, reduced
power losses, limited operating speed range (−30 to +20%).
The development of power electronics related to the onset of rapid switches
and the progress of digital technologies have permitted the control of powerful of
high efficient controllers. New control strategies have been suggested such as direct
power control (DPC) (Giglia et al. 2007).
Several DPC strategies with a constant switching frequency have also been
proposed for the DFIG (Hu et al. 2010). The switching states were initially selected
based on conventional LUT in Zhang et al. (2012), despite the fact that their

A. Daoud () · N. Derbel

Control and Energy Management Laboratory (CEMLab), Sfax Engineering School,
University of Sfax, Sfax, Tunisia
e-mail: [email protected]; [email protected]

© Springer Nature Singapore Pte Ltd. 2019 193

N. Derbel, Q. Zhu (eds.), Modeling, Identification and Control Methods
in Renewable Energy Systems, Green Energy and Technology,
https://doi.org/10.1007/978-981-13-1945-7_9
194 A. Daoud and N. Derbel

durations were calculated based on the objectives of reduced active and reactive
power oscillations (Zhang et al. 2011).
In this work a new strategy is proposed to control the power of doubly fed
induction generator connected to grid, using sliding mode control (SMC) (Bouaziz
and Bacha 2013). This method is a nonlinear robust control that alters the dynamics
of a nonlinear system by designing the discontinuous control signal, which slides
the system states towards the normal behavior (Lajimi and Shahabi 2011).
Moreover, the paper is organized as follows. Section 9.2 presents the mathemat-
ical model of a wind turbine conversion system. A nonlinear Sliding Mode Control
(SMC) strategy is designed in Sect. 9.3. Section 9.4 presents simulation results to
demonstrate performances of the proposed DPC strategy. Finally, some conclusion
remarks are made in the last section.

9.2 Wind Turbine Conversion System

9.2.1 Wind Turbine Model

The mechanical power available on the rotor is given by:

1
Pm = π ρ R 2 V 3 Cp (λ, β) (9.1)
2
where ρ designates the air density, R is the blade radius, V is the wind speed and
Cp represents the power coefficient. It depends on the tip speed ratio λ and the blade
pitch angle β:

   12.5 
116 −
Cp (λ, β) = 0.22 − 0.4 β − 5 e λi (9.2)
λi

where λi is given by:

1 1 0.035
= − 3 (9.3)
λi λ + 0.08 β β +1

with
Ωt R
λ= (9.4)
V
Referring to Equation (9.1), there is a unique optimal power which shows a cubic
function of the generator speed for each wind speed (Cardenas et al. 2005) (Fig. 9.1).
Their relationship can be given by:

1 # $
Pm−opt = Cpmax λopt ρ S V 3 (9.5)
2
9 Direct Power Control of DFIG Using Sliding Mode Control Approach 195

16
V=12m/s
14

12
MPPT
10
Pm(KW)

V=10m/s
8

6
V=8m/s
4

2 V=6m/s

0
0 50 100 150 200 250 300 350
Ωm(rad/s)

Fig. 9.1 Characteristic of the generated power based on the mechanical speed and wind speed

Then, the value of the reference power can be expressed by:

Psref = Kopt Ωm
3
(9.6)
 3
R
where Kopt = 12 π R 2 Cpmax ρ
λopt

9.2.2 Doubly Fed Induction Generator Model

A DFIG can be modelled in the (d-q) reference frame adopting active sign conven-
tion for stator and passive sign convention by the following equations (Belmokhtar
et al. 2014; Shukla and Tripathi 2014).
The stator and rotor voltages can be written as:
⎧
⎪ d
⎨ vsd = −Rs isd − dt φsd − ωs φsq
(9.7)
⎪
⎩ d
vsq = −Rs isq − dt φsq + ωs φsd
⎧
⎪ d
⎨ vrd = Rr ird + dt φrd − ωr φrq
(9.8)
⎪
⎩ d φ +ω φ
vrq = Rr irq + dt rq r rd
196 A. Daoud and N. Derbel

and the stator and rotor fluxes are expressed as:

⎧
⎨ φsd = Ls isd + Mird
(9.9)
⎩
φsq = Ls isq + Miqr

⎧
⎨ φrd = Lr ird + Misd
(9.10)
⎩
φrq = Lr irq + Misq

The electromagnetic torque is given as:

3
Tem = − Np (φsd isq − φsq isd ) (9.11)
2
The mechanical equation of the doubly fed generator is:

dΩm
Tm − Tem = J (9.12)
dt
The instantaneous stator active and reactive power can be expressed in terms of the
stator voltages and currents as follows:
⎧
⎨ Ps = 2 (Vsd isd + Vsq isq )
3

(9.13)
⎩
Qs = 3
2 (Vsq isd − Vsd isq )

According to (9.9) and (9.10), the rotor flux vector can be expressed as:

Lr
φrdq = φsdq + M σ isdq (9.14)
M
where ωs is the electric angular speed for stator windings, ωr is the electric angular
speed for rotor windings and ωm the electric angular speed for rotor windings. Let’s
Lr Ls
the following notations: σ = 1 − and α = Lr Rs + Ls Rr .
M2
The time differentiation of the stator current can be expressed as follows:

disd 1 α
= [αisd −(ωs −ωr )Lr φsq −Rr φsd +Lr Vsd ]+ Vrd +ωr isq (9.15)
dt σ M2 σM
disq 1 α
= 2
[αisq +(ωs −ωr )Lr φsd −Rr φsq +Lr Vsq ]+ Vrq −ωr isd (9.16)
dt σM σM
9 Direct Power Control of DFIG Using Sliding Mode Control Approach 197

By (9.13), time derivative of the stator active and reactive power can be deter-
mined as:
⎡ ⎤ ⎡ ⎤
dPs α     
⎢ dt ⎥ ⎣ σ M 2 − ω r Ps 3(ωs − ωr )Lr Vsq −Vsd φsd
⎣ dQs ⎦ = α ⎦ +
ωr Qs 2σ M 2 −Vsd −Vsq φsq
σM 2
dt
     
3Rr Vsd Vsq φsd 3 Vsd Vsq Vrd
− +
2σ M 2 Vsq −Vsd φsq 2σ M Vsq −Vsd Vrq
 
3Lr 2 +V2
Vsd
+ sq (9.17)
2σ M 2 0

9.3 Conception of Sliding Mode Controller

A non linear sliding mode strategy is based on the conception of the discontinuous
control signal that drives the system states toward special manifolds in the state
space (Rajendran and Jena 2014; Utkin 1993). These manifolds are chosen in such
a way that the control system will have the desired behavior as the states converge
to them (Barambones et al. 2014; Kairous and Belmadani 2015). The strategy of
the DPC uses a nonlinear sliding-mode control scheme to directly generate the rotor
voltage reference components from the stator active and reactive power errors. The
conception of SMC can be defined by the following steps:
1. Sliding Surface:
' (T
S = SP SQ (9.18)

Two integral switching functions has been included to overcome the state errors
and to improve the sliding surfaces. The sliding surfaces of the active and reactive
powers are chosen as:
 t
SP = eP + KP eP dt (9.19)
0
 t
SQ = eQ + KQ eQ dt (9.20)
0

where eP = Ps − Ps∗ , eQ = Qs − Q∗s are errors between the actual values and
the references of instantaneous stator active and reactive powers, respectively.
KP and KQ are the integral gains.
198 A. Daoud and N. Derbel

2. Define the condition of the convergence by the Lyapunov theory as follows:

SP Q ṠP Q < 0 (9.21)

3. Law of SMC: The control signal Vrdq is determined as follows:

Vrdq = Vreq + Vr (9.22)

with Vreq is the equivalent control and Vr is the switching control. The
equivalent command is obtained on the basis of the system behaviour by:

S˙P = P˙s − P˙s∗ + KP (Ps − Ps∗ ) = 0

S˙Q = Q̇s − Q̇∗s + KQ (Qs − Q∗s ) = 0 (9.23)

Substituting (9.25) into (9.23) leads to

Ṡ = F (x) + G(x)Vrdq (9.24)

with F = (FP FQ )T and Vrdq = (Vrd Vrq )T

⎡ α ⎤
  − ω     
FP r Ps 3(ωs − ωr )Lr Vsq −Vsd φsd
= ⎣ σM ⎦
2
α +
FQ ωr Qs 2σ M 2 −Vsd −Vsq φsq
σ M2
    
3Rr 3Lr 2 +V2
Vsd
Vsd Vsq φsd
− + sq
2σ M 2 Vsq −Vsd φsq 2σ M 2 0
 
KP (Ps − Ps∗ )
+ (9.25)
KQ (Qs − Q∗s )

and:
 
3 1 Vsd Vsq
G=
2 σM Vsq −Vsd

Let us consider the following Lyapunov function:

1 T
V = S S (9.26)
2
The time derivative of the quadratic Lyapunov function is:

V̇ = S T Ṡ = S T [−H sign(S) − δ S] (9.27)

H and δ are diagonal matrices as described below.

9 Direct Power Control of DFIG Using Sliding Mode Control Approach 199

Grid
Vsabc
dq
abc
Isabc

Ps computing Qs computing
DFIG
Ps∗ Q∗s

Siding Mode
Control
Vrd Vrq
Udc
SVM

Fig. 9.2 Basic DPC bloc diagram using SMC approach

In view of the following expression of the switching control:

 
HP sign(SP ) + δP SP
Vr = − (G(x))−1 (9.28)
HQ sign(SQ ) + δQ SQ

where HP , HQ , δP and δQ are positive control gains, sign(SP ) and sign(SQ ) are
switch functions for active and reactive powers.
4. Synthesis of SMC:
Figure 9.2 shows the bloc diagram of the basic DPC strategy, using the SMC
approach. Actual values of stator active and reactive powers are compared to their
corresponding references. Errors are achieved to stator active and reactive powers
SMC controller. Considering SMC approach, rotor voltage reference values are
determined.
Then, the Space Vector Modulation (SVM) technique is used to transform the
switching patterns of the RSC with a constant switching frequency.

9.4 Simulation Results

Simulation works have been carried out considering a sampling period T0 = 10 μs

and a bus voltage Vdc = 620 V. The control design exposed in Fig. 9.2 is simu-
lated in the MATLAB/Simulink environment with parameters given in Tables. 9.1
and 9.2.
200 A. Daoud and N. Derbel

Table 9.1 Parameters of the Denotation Numerical value of the parameter

turbine
Radius of the wind 3.24 m
G 2.5
Air density (ρ) 1.22 kg/m3

Table 9.2 DFIG parameters

Denotation Numerical value of the parameter
Power output 20 kW
Voltage 220 V
Speed (ωs , fs ) 1500 tr/mn; 50 Hz
Number of pole pairs (Np ) 4
Stator inductance (Ls ) 0.05 H
Rotor inductance (Lr ) 0.05 H
Mutual inductance (M) 0.0473 H
Stator resistance (Rs ) 0.29 Ω
Rotor resistance (Rr ) 0.38 Ω
Moment of inertia (J) 1 kg.m2

The simulation is effected supposing which the chain of wind energy conversion
expands, according to the MPPT, the stator active power reference (Ps ref =
s ref = 0.
Kopt Ωm 3 ), and imposing a stator reactive power Q

Figure 9.3 demonstrates that the mechanical speed pursues the wind but with a
delay that is due to slow dynamics of the generator against the very rapid variation
of the wind. Moreover, stator currents have honest evolutions regarding to their
amplitudes and their frequency.
Foremost, investigate the expression of the ias current, around a steady state
operating point (Bekakra and Attous 2011), as follows:
∞
 # $
ias (t) = In cos nωs t + ϕn (9.29)
n=1

|I1 | is the amplitude of the fundamental, and |In | is the amplitude of the harmonic n.
The index criterion is the average total harmonic distortion (THD) of the stator
current which is determined as follows:
)

∞
|In |2
n=2
THD = (9.30)
|I1 |

In Figs. 9.4, 9.5 and 9.6, the frequency spectrum of stator current ias has been
visualized by different wind values. They consist to observe amplitudes of all
 15
Wind (m/s) a
10

t (s)
5
0 5 10 15 20 25 30
85
Ωm (rad/s) b
80

t (s)
75
0 5 10 15 20 25 30
50
isabc (A) c
0

t (s)
−50
0 5 10 15 20 25 30

Fig. 9.3 Simulation results. (a): The trial wind profile, (b): Mechanical speed of DFIG Ωm , (c):
currents of stator isabc

20
a
10
0
−10
−20
22 22.002 22.004 22.006 22.008 22.01 22.012 22.014 22.016 22.018 22.02

1 b

0.5

0
0 5 10 15 20 25 30 35 40 45 50

0.06
c
0.04

0.02

0
0 5 10 15 20 25 30 35 40 45 50

Fig. 9.4 Simulated basic signal waveforms and stator current harmonic spectrum under purely
sinusoidal stator voltage for V = 10 m/s. (a) one period of ias current, (b) reduced spectrum of ias
with respect to its fundamental, and (c) harmonics of ias (fundamental = 1)
 40
a
20
0
−20
−40
12 12.002 12.004 12.006 12.008 12.01 12.012 12.014 12.016 12.018 12.02

1 b

0.5

0
0 5 10 15 20 25 30 35 40 45 50

0.06
c
0.04

0.02

0
0 5 10 15 20 25 30 35 40 45 50

Fig. 9.5 Simulated basic signal waveforms and stator current harmonic spectrum under purely
sinusoidal stator voltage for V = 13 m/s. (a) one period of ias current, (b) reduced spectrum of ias
with respect to its fundamental and (c) harmonics of ias (fundamental = 1)

10
a
5
0
−5
−10
5 5.002 5.004 5.006 5.008 5.01 5.012 5.014 5.016 5.018 5.02

1 b

0.5

0
0 5 10 15 20 25 30 35 40 45 50

0.06
c
0.04

0.02

0
0 5 10 15 20 25 30 35 40 45 50

Fig. 9.6 Simulated basic signal waveforms and stator current harmonic spectrum under purely
sinusoidal stator voltage for V = 8 m/s. (a) one period of ias current, (b) reduced spectrum of ias
with respect to its fundamental, and (c) harmonics of ias (fundamental = 1)
9 Direct Power Control of DFIG Using Sliding Mode Control Approach 203

Table 9.3 Evaluation of Wind values (m/s) 8 10 13

THD
THD (%) 3.303 1.22 0.96

Cem (Nm) a
200

−200
t (s)
0 5 10 15 20 25 30

20
Ps (KW) b
10
0
−10
t (s)
−20
0 5 10 15 20 25 30

20
Qs (KVar) c
10
0
−10
t (s)
−20
0 5 10 15 20 25 30

Fig. 9.7 Simulation results. (a): Electromagnetic torque of the DFIG, (b): Stator active power
transmitted to the grid, (c): Reactive power of the stator

harmonics in terms of harmonic frequencies. Subscribing the fast Fourier transform

(FFT), the performance index T H D is evaluated.
Table 9.3 shows the evaluation of performance index T H D using a sliding mode
control. By increasing wind values, it is clear that results donated from the SMC
controllers have lower values of the performance index.
Figure 9.7 presents the electromagnetic torque and stator active and reactive
powers. It is found that, the responses of the SMC regulator follow the wind profile
with a very fast time response.

9.5 Conclusion

In this work, a nonlinear sliding mode control approach of a DFIG based on

wind turbine has been investigated to improve the efficiency of the wind energy
conversion system. Using the Lyapunov stability theorem, the synthesis of this
204 A. Daoud and N. Derbel

controller is developed. As demonstrated by the obtained results, it is observed that

this tuning is useful to improve the damping of the oscillatory modes and also to
enhance the fault ride-through capability of the DFIG system.

References

Barambones, O., Cortajarena, J. A., Alkorta, P., & de Durana, J. M. G. (2014). A real-time sliding
mode control for awind energy system based on a doubly fed induction generator. Energies, 7,
6412–6433.
Bekakra, Y., & Attous, D. B. (2011). Sliding mode controls of active and reactive power of a DFIG
with mppt for variable speed wind energy conversion. Australian Journal of Basic and Applied
Sciences, 5, 2274–2286.
Belmokhtar, K., Doumbia, M., & Agbossou, K. (2014). Novel fuzzy logic based sensorless
maximum power point tracking strategy for wind turbine systems driven DFIG (doubly-fed
induction generator). Energy, 76, 679–693.
Bouaziz, B., & Bacha, F. (2013). Direct power control of grid-connected converters using sliding
mode controller. Electrical Engineering and Software Applications, 5, 801–806.
Cardenas, R., Pena, R., Perez, M., Clare, J., Asher, G., & Wheeler, P. (2005). Control of a
switched reluctance generator for variable-speed wind energy applications. IEEE Transactions
on Energy Conversion, 20, 781–791.
Cheng, M. Y. Z. (2014). The state of the art of wind energy conversion systems and technologies:
A review. Energy Conversion and Management, 88, 332–347.
Giglia, G. M., Pucci, C. S., & Vitale, G. (2007). Comparison of control techniques for three-
phase distributed generation based on VOC and DPC, I.S.S.I.A.-C.N.R. (Institute on intelligent
systems for the automation). IEEE Transactions on Energy Conversion, 1, 1–12.
Hu, J., Nian, H., Hu, B., He, Y., & Zhu, Z. Q. (2010). Direct active and reactive power regulation
of DFIG using sliding-mode control approach. IEEE Transactions on Energy Conversion, 25,
1028–1039.
Kairous, D., & Belmadani, B. (2015). Robust fuzzy-second order sliding mode based direct power
control for voltage source converter. International Journal of Advanced Computer Science and
Applications, 6, 167–175.
Lajimi, A. B., & Shahabi, M. (2011). Modeling and control of a DFIG-based wind turbine during a
grid voltage drop. ETASR Engineering Technology and Applied Science Research, 1, 121–125.
Rajendran, S., & Jena, D. (2014). Control of variable speed variable pitch wind turbine at above
and below rated wind speed. Electrical Engineering and Software Applications, 1, 1–14.
Shukla, R. D., & Tripathi, R. K. (2014). A novel voltage and frequency controller for standalone
DFIG based wind energy conversion system. Renewable and Sustainable Energy Reviews, 37,
69–89.
Utkin, V. (1993). Sliding mode control design principles and applications to electric drives. IEEE
Transactions on Industrial Electronics, 40, 23–36.
Zhang, Y., Zhu, J., & Guo, Y. (2011). A simple method to reduce torque ripple in direct torque-
controlled permanentmagnet synchronous motor by using vectors with variable amplitude and
angle. IEEE Transactions on Energy Conversion, 58, 2848–2859.
Zhang, Y., Li, Z., & Xu, W. (2012). A novel three-vectors-based predictive direct power control of
doubly fed induction generator for wind energy applications. In Energy Conversion Congress
and Exposition (ECCE) (pp. 793–800).
Chapter 10
A Comparative Study Between PI and
Sliding Mode Control for the DFIG of a
Wind Turbine

Basma Boujoudi, Elm’kaddem Kheddioui, Nadia Machkour,

and Mohammed Bezza

Abstract This chapter presents a comparative study between the indirect field-
oriented control (IFOC) with Proportional-Integral controller (PI) and the sliding
mode control of a doubly fed induction machine (DFIG), dedicated to the production
of the electrical energy in a wind system. Both types of control are presented in
this paper to compare their performances in terms of follow-up of deposits and
robustness towards parametric variations of the DFIG.

Keywords DFIG · Indirect field-oriented control · IFOC · PI · Sliding mode ·

Wind turbine

10.1 Introduction

Nowadays, due to multiple benefits of wind turbines, their integration into the field
of alternative and renewable energy sources has been growing increasingly. Several
research studies are currently focusing on the management and control of wind
conversion systems, and more particularly the control of the electrical machine
which forms the core of the process of energy conversion. The aim of my research
work is to study the connection of the wind turbine system to an electrical network
affected by voltage dip, in earlier work (Boujoudi et al. 2016) a classification of
existing voltage dips and their impacts are presented.
In this paper we compare the performances of the two control modes to choose
the appropriate command for the wind turbine system, but in future work we are

B. Boujoudi () · E. Kheddioui · N. Machkour

Laboratory of Physics of the atmosphere, materials and modeling, Faculty of Sciences and
Techniques of Mohammedia, Hassan II University of Casablanca, Mohammedia, Morocco
M. Bezza
Laboratory of Electronics, Energy, Automation and Information Processing, Faculty of Sciences
and Techniques of Mohammedia, Hassan II University of Casablanca, Mohammedia, Morocco

© Springer Nature Singapore Pte Ltd. 2019 205

N. Derbel, Q. Zhu (eds.), Modeling, Identification and Control Methods
in Renewable Energy Systems, Green Energy and Technology,
https://doi.org/10.1007/978-981-13-1945-7_10
206 B. Boujoudi et al.

Fig. 10.1 Doubly-fed induction generator in wind power generation

going to study the command performances of the system connected to an electrical

network affected by voltage dip.
The used machine in this work is the doubly fed asynchronous machine (DFIG).
Its stator is connected directly to the electrical network, while its rotor is connected
to the network by means of two converters (Fig. 10.1). The converter C1 maintains
the unit power factor while the C2 acts on the regulation of the active and reactive
powers injected into the network. In this chapter we present two control strategies
of the converter C2 : one concerns the control by a PI regulator and the other is
articulated around the nonlinear control based on the sliding mode. A modeling of
the wind turbine and DFIG assembly was carried out, in order to synthesize the
indirect field-oriented control with the PI and the sliding mode.
Simulation results of both commands are presented in Matlab-Simulink environ-
ment. The overall system is illustrated in Fig. 10.1.
Firstly, the modeling of the “wind turbine – DFIG” is exposed, then, we present
the strategy of controlling the wind turbine with the MPPT. At last, we synthesize
the indirect vector control by the two methods. A comparative study of both methods
is established to present the performances of these two control techniques.

10.2 System Modeling (Turbine & DFIG)

10.2.1 Wind Turbine Modeling

A wind Vv is applied to the blades of radius R of the wind turbine to create its
rotation (Delenclos 2016). This rotation generates a mechanical power which is
transmitted to the mechanical shaft of the turbine. It is expressed as follows (Allam
10 Control for the DFIG of a Wind Turbine 207

et al. 2014; Bedoud et al. 2013; Sabzevari et al. 2017; Tamaarat et al. 2013):
1
Pm = ρ Cp (λ, β) S Vv3 (10.1)
2
The mechanical torque can be represented by:
Pt 1
Tt = = π ρ Cp (λ, β) R 3 Vv2 (10.2)
Ωt 2λ
where Cp (λ, β) is the power coefficient of the wind turbine which represents the
aerodynamic efficiency of the wind turbine and also depends on the characteristic
of the turbine. Several forms of this coefficient can be found in the literature. In this
chapter we opt for the following expression (Bedoud et al. 2013):
   
151 184
Cp (λ, β) = 0.46 − 0.58β − 0.002β 214 − 1.32 exp − (10.3)
λi λi
1 1 0.003
= − (10.4)
λi λ + 0.2β 1 + β3
where λ represents the tip-speed ratio which is the ratio of the blade-tip linear speed
to the wind speed, defined as:

RΩt
λ= (10.5)
Vv

The mechanical part of the turbine comprises three steerable blades of length R.
They are fixed on a drive shaft rotating at a speed of rotation Ωt , connected to a
gain multiplier Gm which drives the electric generator. The multiplier, considered
as perfect, guarantee the adaptation between the speed of rotation of the turbine and
that of the generator according to the following relation:

Tt Ωmec
Gm = = (10.6)
Tm Ωt

The wind turbine system mechanical equation can be represented by:

 
Jt dΩm
Tm − Tem = Jm + + fv Ωm (10.7)
dt dt

where Tem is the torque produced by the electrical generator.

The block diagram of the wind turbine is illustrated in Fig. 10.2.
In order to extract the maximum power from the incident wind energy, it is
necessary to permanently regulate the speed of rotation of the turbine so that it
adapts to the wind speed. This permanent search technique is called “Maximum
Power Point Trackin”.
208 B. Boujoudi et al.

Fig. 10.2 Modeling of the mechanical part of the wind turbine

Fig. 10.3 MPPT control with rotational speed control

The search for the maximum is made constantly and the wind turbine adapts to
each variation of the wind to be in a configuration of maximum power extraction
(El Aimani 2004; Mirecki 2005).
There are several techniques to pursue MPPT. These techniques differ according
to the type of information they need to deliver the speed reference, the technique
used in this paper is that of power maximization with speed control (Fig. 10.3).

10.2.2 DFIG Modeling

In general, a squirrel cage induction machine can operate in two modes: motor
or generator. Indeed, the machine operates as a motor when it rotates below its
synchronous speed and it functions as a generator when it exceeds this speed (El
Aimani 2004).
10 Control for the DFIG of a Wind Turbine 209

In the case of a DFIG, it is no longer the speed of rotation which imposes

the mode of operation. It is the control of the rotor voltages which control the
magnetic field inside the machine, thus offering the possibility of operating in
hyper-synchronism or in hypo-synchronism in both the motor and generator modes.
The following equation system describes the global modeling of DFIG in Park’s
repository (Bedoud et al. 2013; El Aimani 2004; Uma and Manikandan 2013).
• Electrical equations
– Stator voltage
⎧ ϕ
⎨ Vsd = Rs isd + sd − θ̇s ϕsq
dt (10.8)
⎩ Vsq = Rs isq + ϕsq + θ̇s ϕsd
dt
– Rotor voltage
⎧ ϕ
⎨ Vrd = Rr ird + rd − θ̇r ϕrq
dt (10.9)
⎩ Vrq = Rr irq + ϕrq + θ̇r ϕrd
dt

• Magnetic equations
M
Tem = − (ird ϕsq − irq ϕsd ) (10.10)
Ls

Thus, the equations of the machine are established, we notice that the mechanical
equation requires a decoupling to be able to control independently the active and
reactive powers.
In order to facilitate this control, one chooses the synchronously rotating
reference frame so the stator flux vector ϕs is placed on the axis d (Bühler 1986)
and this results in (Fig. 10.4):
*
ϕsd = ϕs
(10.11)
ϕsq = 0

Fig. 10.4 Reference frame

related to the stator rotating
field
210 B. Boujoudi et al.

Assuming that the stator flux is constant and neglecting the stator resistance Rs ,
the stator voltages in the plane dq are presented below (Betin 2003).
*
Vsd = 0√
(10.12)
Vsq = 3Vs = Us = ωs ϕsd

Thus, equations of the stator flux become:

*
ϕsd = ϕs = Ls isd + Mird
(10.13)
ϕsq = 0 = Ls isq + Mirq

From the above equations and taking into account the simplifying hypotheses we
obtain the following equations:
• Stator Current
⎧
⎪ 1 M
⎨ isd = ϕs − ird
Ls Ls (10.14)
⎪ M
⎩ isq = − irq
Ls

• Active power

M
Ps = −Vsq isq (10.15)
Ls

• Electromagnetic torque

M
Tem = p ϕsd irq (10.16)
Ls

• Reactive power

Vsq ϕsd Vsq M

Qs = − ird (10.17)
Ls Ls

• Rotor voltage
⎧
⎪ di
⎨ Vrd = Rr ird − σ Lr rd − σ Lr ωr irq
dt (10.18)
⎪ di M
⎩ Vrq = Rr irq + σ Lr rq + σ Lr ωr ird + ωr ϕsd
dt Ls

Where σ is the dispersion coefficient of the DFIG, and ωr is the rotor pulsation.
From these equations, Fig. 10.5 presents the block diagram of the system to be
regulated (El Aimani 2004; Uma and Manikandan 2013).
10 Control for the DFIG of a Wind Turbine 211

Fig. 10.5 Block diagram of the DFIG

The expressions of the active and reactive powers show that, assuming that the
stator flux along the axis dq is kept constant, the choice of the reference dq makes
the electromagnetic torque and consequently the stator active power proportional to
the rotor current of axis q, as regards the reactive power is not proportional to the
rotor current irq due to a constant imposed by the network. This allows the control
of the stator powers independently of one another, by regulating the rotor currents
in the axis dq.

10.3 Indirect Field-Oriented Control of the DFIG

The wind power generation system has two converters, the first on the grid side and
the second on the rotor side of the DFIG. In this chapter, we are interested only in the
converter rotor side, responsible for the regulation of active and reactive power. The
rotor-side converter is controlled by the PI control at first, then by the sliding mode
control, a setpoint and robustness test is performed to compare the performance of
these two control strategies. The reference current irq is derived from the MPPT
control via the reference electromagnetic torque or via the reference stator power,
the expression of the current is presented by:

Ls
irq−ref = − Tem−ref (10.19)
pMϕsd−est

which we can written as:

Ls
irq−ref = − Ps−ref (10.20)
MVsq
212 B. Boujoudi et al.

The reference current ird is derived from the control of the stator reactive power.

ϕsd−est Ls
ird−ref = − Qs−ref (10.21)
M MUs

10.3.1 Indirect PI Field-Oriented Control of the DFIG

Expressions above show that the stator flux is necessary for the generation of the
reference rotor currents (Allam et al. 2014; Bedoud et al. 2013; El Aimani 2004;
Uma and Manikandan 2013). The stator flux of axis d can be estimated in open
loop from the measurements of the stator and rotor currents with axis d (El Aimani
2004):

ϕsd−est = Ls isd + Mird (10.22)

The control diagram of the converter C2 is illustrated in Fig. 10.6.

PI regulators have been used. Their parameters depend on the system parameters.
The open loop transfer function of the rotor current regulators can be written as
follows:
⎛ ⎞ ⎛ ⎞
Kpr 1
⎜ 1 + Kir s ⎟ ⎜ Rs ⎟
OLT F = Kir ⎜ ⎝
⎟ ⎜
⎠ ⎝
⎟ (10.23)
s σ L r ⎠
1+ s
Rr

Fig. 10.6 Machine-side converter control with PI

10 Control for the DFIG of a Wind Turbine 213

Using the pole compensation method for determining the parameters of the
regulator, we obtain:
3Rr
Kir = (10.24)
tr5%
3σ Lr
Kpr = (10.25)
tr5%

10.3.2 Indirect Field-Oriented Control of the DFIG with the

Slidding Mode

The principle of this command is to choose a surface or a region of slip, and to attract
the states of the system in this region, then to elaborate a law of control which allows
the maintenance of the system in this region.
The choice of the non-linear sliding surface is based on the system state variables.
The variables to be regulated are expressed by the model previously established
(Betin 2003).
The design of the sliding mode algorithm takes into account the problems of
stability and good performance in a systematic way in its approach, which is carried
out mainly in three complementary steps (Bühler 1986; Bekakra and Ben Attous
2009; Bekakra and Ben Attous 2010; Bekakra and Ben Attous 2011; Ben attous et
al. 1998)
• Choice of sliding surfaces;
• Definition of the conditions of existence and convergence of the rolling regime;
• Determination of the control law.
Based on the model established for control with the PI controller, the slip surfaces
chosen in this chapter are the direct and quadrature rotor current, since the rotor
currents are the image of the stator powers.

10.3.2.1 Active Power Control (Ps )

In the following, an indirect control of the active power of the DFIG is performed
by acting on the current irq. The error of the quadrature rotor current irq is defined
by Ben attous et al. (1998), Namuduri and Sen (1986), and Young (1989)

e = irq−ref − irq (10.26)

For n = 1, the surface of the quadrature rotor current can be defined from the
following equation:

Sq (irq ) = eq = irq−ref − irq (10.27)

214 B. Boujoudi et al.

dirq
By taking the time derivative of (10.27) and by replacing the expression with
dt
its expression according to the term of the command, one obtains:
   
d 1 1 M2 1
Ṡq = ėq = irq−ref − − + irq − gωs ird + Vrq
dt σ Tr Lr Ls Ts σ Lr
(10.28)
The expression of Vrq is:

Vrq = Vrq−eq + Vrq−n (10.29)

During the sliding mode:

Sq = Ṡq = 0 (10.30)

This gives the so called equivalent control:

   
d 1 1 M2
Vrq−eq = σ Lr irq−ref + + irq + gωs ird (10.31)
dt σ Tr Lr Ls Ts

Consequently:

Vrq−n = Kq sat (Sq ) (10.32)

where Kq is a positive constant.

In this chapter we use the function “sat” instead of “sign” in order to eliminate
the chattering of the control.

10.3.2.2 Reactive Power Control

The reactive power generated by the DFIG is controlled by controlling the direct
rotor current ird . The error of the direct rotor current is defined by Ben attous et al.
(1998), Hashimoto et al. (1986), Namuduri and Sen (1986), and Young (1989):

ed = ird−ref − ird (10.33)

For n = 1, the surface of the direct rotor current can be defined from the
following equation:

Sd (ird ) = ed = ird−ref − ird (10.34)

Its derivative:
 
d 1 1
Ṡd = ėd = ird−ref − ird + gωs irq + Vrd (10.35)
dt σ Tr σ Lr
10 Control for the DFIG of a Wind Turbine 215

The expression of Vrd is then:

Vrd = Vrd−eq + Vrd−n (10.36)

During the sliding mode:

Sd = Ṡd = 0 (10.37)

This gives the so called equivalent control:

 
d 1
Vrd−eq = σ Lr ird−ref − ird − gωs irq (10.38)
dt σ Tr

Consequently:

Vrq−n = Kd sat (Sq ) (10.39)

where Kd is a positive constant.

Figures 10.7 and 10.8 summarize the principle of the sliding mode.

Fig. 10.7 Principle of control with sliding mode with addition of equivalent control
216 B. Boujoudi et al.

Fig. 10.8 Control of the machine side converter with sliding mode

Table 10.1 Parameters of the DFIG and the wind turbine

Parameter Pn Rs Rr Ls Lr M F J R Gm
Value 20 0.455 0.19 70 21.3 34 0.0024 0.53 3 5.4
Unit W Ω Ω mH mH mH Nm/s kg m2 m –

10.4 Simulation Result, Performance Analysis

This part deals more specifically with the comparison and analysis of the perfor-
mances of the regulators synthesized and detailed in the previous section. To do
this, a series of tests were carried out with different conditions in order to establish
a model closest to the experimental device. Parameter of the DFIG and the wind
turbine are given in Table 10.1.

10.4.1 Follow-Up of the Reference

Test conditions are chosen as:

• The wind speed is assumed to be constant 7 m/s, which means that the generator
will run at a speed of 900 rpm.
• At t = 1 s active power step (Pref switches from 0 to −10 kW).
• At t = 2 s step of reactive power (Qref varies from 0 to 6 kVAR).
Figures 10.9, 10.10, 10.11, 10.12, 10.13, and 10.14 show the simulation results
obtained with MATLAB / SIMULINK in the case of the indirect control of the DFIG
with the sliding mode and the PI controller.
Fig. 10.9 Active and reactive Ps_Qs (PI)
stator power with PI
controller 6000

4000 Psref
Ps
Qsref
2000 Qs

Ps (W) Qs (VAR)
0

–2000

–4000

–6000

–8000

–10000

0.5 1 1.5 2 2.5 3

Time (s)

Ps_Qs (Sliding mode)

6000 Psref
Ps
Qsref
Qs
4000

2000
Ps (W) Qs (VAR)

–2000

–4000

–6000

–8000

–10000

0.5 1 1.5 2 2.5 3

Time (s)

Fig. 10.10 Active and reactive stator power with sliding mode controller
Fig. 10.11 Reactive power at t = s with the PI

Fig. 10.12 Reactive stator power at t = 1 s with the sliding mode controller
Fig. 10.13 Active stator power at t = 2 s with PI controller

Fig. 10.14 Active stator power at t = 2 s with sliding controller

220 B. Boujoudi et al.

The previous figures show that the active and reactive powers generated by the
DFIG follow their references perfectly. Obviously the results assert that the active
stator power depends on the quadrature rotor current and that the reactive stator
power depends on the direct rotor current, hence the effect of decoupling between
the d-axis and q. The almost zero static error is approximately the same for both
control modes. The system has a satisfactory fast dynamic and no overshoot. The
coupling between the two powers is virtually absent in the case of the control by
the sliding mode, contrary to the indirect vector control with the PI, it is observed
that the stator powers undergo a transition peak. To conclude, the control of active
and reactive stator power by the sliding mode brings a remarkable improvement
compared to the PI regulator.

10.4.2 Robustness Testing

The calculation of the regulators is based on transfer functions whose parameters

are assumed to be fixed. However, in reality, these parameters may vary because of
the various physical phenomena, in particular the heating of the resistors. In order
to get closer to the actual behavior of the system, we will perform a robustness tests
which consist in varying the parameters of the machine.
• A 50% increase in resistances Rs and Rr .
• A 20% decrease in mutual inductance M.
• The wind speed is set at 7 m/s, which causes the machine to run at 900 rpm.
• At t = 1 s the active power goes from 0 to −10 kW.
• At t = 1.5 s the wind speed increases from 7 to 10 m/s, so the machine speed
increases from 900 to 1290 rpm.
• At t = 2 s the reactive power increases from 0 to 6 kVAR.
Figures 10.15, 10.16, 10.17, 10.18, 10.19, 10.20, 10.21, 10.22, and 10.23, show
the result obtained for the two types of control in the case of a parametric variation.
It can be seen that the PI regulator loses its robustness, contrary to the sliding mode
control. From the figures, it can be seen that the reactive stator power controlled
by the PI is slightly short of its reference value, thus causing a static error of
5%, contrary to that regulated by the sliding mode. In the end, a very satisfactory
performance is obtained by using the sliding mode in terms of follow-up of the
reference when a variation of the parameters of the machine.

10.5 Conclusion

This chapter presents a comparative study between the indirect field-oriented control
with the PI controller and the sliding mode control of the active and reactive stator
powers generated by the DFIG in a wind turbine. In a first step we have established
Fig. 10.15 Mechanical Mechanical rotation speed
rotation speed 150

140

130

120

Omega (rad/s)
110

100

90

80

70
0.5 1 1.5 2 2.5 3
Time (seconds)

Fig. 10.16 Active and Ps_Qs (PI)

reactive stator power with PI
controllers 8000 Psref
Ps
Qsref
Qs
6000

4000

2000
Ps (W) Qs (VAR)

–2000

–4000

–6000

–8000

–10000

0.5 1 1.5 2 2.5 3

Time (s)
Fig. 10.17 Active and Ps_Qs (Sliding mode)
reactive stator power with 8000
sliding mode controller
6000
Psref
4000 Ps
Qsref
Qs
2000

Qs (VAR)
–2000

–4000

–6000

–8000

–10000

–12000
0.5 1 1.5 2 2.5 3
Time (s)

Fig. 10.18 Reactive stator power at t = 1 s with PI controller

10 Control for the DFIG of a Wind Turbine 223

Fig. 10.19 Reactive stator power at t = 1 s with sliding mode controller

a model of the set consisting of the wind turbine and the DFIG. Secondly, the
synthesis of the two types of regulators was carried out. Then we effectuated two
tests in order to compare the performance of each control technique from the point
of view of the follow up of the reference and the robustness. The results obtained
show that the control by the sliding mode presents better performances than that with
the PI controller, whether for the reference tracking or the robustness thus showing
a good insensitivity to the disturbances and the parametric uncertainties. Given the
satisfactory results of the sliding mode, the next step is to inject the different types
of voltage dips into the grid in order to study the behavior of the system controlled
by the sliding mode when it’s affected by voltage dips.
Fig. 10.20 Active stator x104
Ps (PI)
power at t = 2 s with the PI –0.92
controller Psref
Ps
–0.94

–0.96

–0.98

–1

Ps(W) –1.02

–1.04

–1.06

–1.08

–1.1
1.5 2 2.5 3
Time (s)

Fig. 10.21 Active stator power at t = 2 s with sliding controller

Fig. 10.22 Stator currents Isdq (PI)
(axis dq) with PI controller 30
Isd
Isq

25

20

15
Isdq (A)

10

–5
0.5 1 1.5 2 2.5 3
Time (s)

Fig. 10.23 Stator currents Isdq (Sliding mode)

(axis dq) with sliding mode 30
controller Isd
Isq
25

20

15
Isdq (A)

10

10

–5
0.5 1 1.5 2 2.5 3
Time (seconds)
226 B. Boujoudi et al.

References

Allam, M., Allam, M., & Djeriri, Y. (2014). Etude comparative entre la commande vectorielle
directe et indirecte de la Machine Asynchrone à Double Alimentation (MADA) dédiée à une
application éolienne. Journal of Advanced Research in Science and Technology, 1(2), 88–100.
Bedoud, K., Lakel, R., Ali-Rachedi, M., Lekhchine, S., & Bahi, T. (2013). Modelisation d’une
chaine de conversion d’energie eolienne basee sur une machine asynchrone a double alimenta-
tion. 21 Congres Francais de Mecanique, Bordeaux.
Bekakra, Y., & Ben Attous, D. (2009). A sliding mode speed and flux control of a doubly fed
induction machine. In Electrical and Electronics Engineering, IEEE Conference, (I) (pp. 174–
178).
Bekakra, Y., & Ben Attous, D. (2010). Speed and flux control for DFOC of doubly fed induction
machine using sliding mode controller. Acta Electrotechnica and Informatica, 10(4), 75–81.
Bekakra, Y., & Ben Attous, D. (2011). Sliding mode controls of active and reactive power of a
DFIG with MPPT for variable speed wind energy conversion. Australian Journal of Basic and
Applied Sciences, 5(12), 2274–2286.
Ben attous, D., Golea, A., & Abdessemed, R. (1998). Commande à structure variable par mode
glissant pour la commande vectorielle d’un moteur asynchrone. In ICEL98, USTOran.
Betin, F. (2003). Commande d’actionneurs électriques soumis à une charge mécanique à
fortes variations paramétriques. Habilitation à diriger des recherches, Centre de Robotique
d’Electrotechnique et d’Automatique, Université de Picardie Jules Verne, 02880 Cuffies.
Boujoudi, B., Machkour, N., & Kheddioui, Elm. (2016). New method for detection and character-
ization of voltage dips. Molecular Crystals and Liquid Crystals, 641(1), 86–94.
Bühler, H. (1986). Réglage par Mode de Glissement. Presses Polytechniques Romandes, EPFL,
Ecublens, Lausanne.
Delenclos, S. (2016). L’énergie éolienne. http://gte.univ-littoral.fr
El Aimani, S. (2004). Modélisation de différentes technologies d’éoliennes intégrées dans un
réseau de moyenne tension. Thesis, Ecole Centrale de Lille, France.
Hashimoto, H., Yamamoto, H., Yanagisawa, D., & Harachima, F. (1986). Brushless servomotor
control using VSS approach. In IEEE TAS Annuel Meeting (pp. 72–79).
Mirecki, A. (2005). Etude comparative de chaînes de conversion d’énergie dédiées à une éolienne
de petite puissance. Thesis, Institut National Polytechnique, Toulouse.
Namuduri, C., & Sen, P. C. (1986). A servo control system using a self-controlled synchronous
motor (SCSM) with sliding mode controller. In IEEE IAS Annuel Meeting (pp. 56–65).
Sabzevari, S., Karimpor, A., & Monfared, M. (2017). MPPT control of wind turbines with
wind speed estimation using direct adaptive fuzzy-pi controller. Journal of Renewable and
Sustainable Energy, 9, 013302. https://doi.org/10.1063/1.4973447
Tamaarat, A., Benakcha, A., & Menacer, A., (2013). Commande des puissances active et réactive
dans d’une Chaine de Conversion basée sur une Machine Asynchrone à Double Alimentation.
http://hdl.handle.net/123456789/2513
Uma, S. P., & Manikandan, S. (2013). Control technique for variable speed wind turbine
using PI controller. In IEEE International Conference on Emerging Trends in Computing,
Communication and Nanotechnology (pp. 640–643).
Young, M. (1989). The technical writer’s handbook. Mill Valley: University Science.
Chapter 11
Emulation of Wind Turbine for
Standalone Wind Energy Conversion
Systems

L. Benaaouinate, M. Khafallah, A. Mesbahi, A. Martinez, T. Bouragba,

and D. Breuil

Abstract In this chapter, a Wind Turbine Emulator (WTE) based on permanent

magnet DC motor controlled by Pulse-Width Modulated (PWM) is proposed to
satisfy the further study of the wind energy conversion system (WECS) in the
laboratory. The main objective of the emulation system is to reproduce the dynamic
behavior of the wind turbine torque on the generator shaft. The electrical, mechani-
cal, aerodynamic models of horizontal axis wind turbines (HAWTs) are analyzed in
detail. The control of a permanent magnet DC motor and the mathematical models
of the wind turbine are elaborated using Matlab/Simulink environment. Simulation
and experiment results are also presented to demonstrate the functionality and
effectiveness of the system and to confirm that a WTE can provide all necessary
parameters of the wind turbine system such as wind speed, output torque, power
coefficient and tip speed ratio.

Keywords Wind turbine emulator · DC motor drive · Real time control ·

Modeling · Electrical generator · Wind turbine

11.1 Introduction

Recently, with the growing energy demands and environmental crisis such as
global warming, there had led to the development of renewable sources for the

L. Benaaouinate () · M. Khafallah · M. Mesbahi

High National School of Electricity and Mechanics, Hassan II University, Casablanca, Morocco
e-mail: [email protected]
A. Martinez · T. Bouragba · D. Breuil
Industrial systems engineering school, La Rochelle University, La Rochelle, France
e-mail: [email protected]; [email protected]; [email protected]

© Springer Nature Singapore Pte Ltd. 2019 227

N. Derbel, Q. Zhu (eds.), Modeling, Identification and Control Methods
in Renewable Energy Systems, Green Energy and Technology,
https://doi.org/10.1007/978-981-13-1945-7_11
228 L. Benaaouinate et al.

production of electricity. One of the leading prospects in this search for clean and
safe renewable power sources is wind generation system, as it is environmentally
friendly, socially beneficial and economically competitive.
The wind energy has been exploited for thousands of years. The oldest applica-
tions of wind power were providing mechanical power to pump water from wells
and making flour out of grain. During the last decade of the twentieth century,
worldwide wind capacity doubled approximately every three years. By the end of
2016, an increase of 12.5% compared to the previous year of the worldwide total
cumulative installed electricity generation capacity from wind power amounted to
486.790 MW. Currently, five countries have achieved relatively high levels of wind
power penetration, such as 39% of stationary electricity production in Denmark,
18% in Portugal, 16% in Spain, 14% in Ireland and 9% in Germany (Council 2011).
In order to increase the profits of wind power and to improve the performance
and reliability of next generation wind technologies, a great deal of research has
been focused on the better way to exploit and generate this kind of energy while
lowering the cost of wind energy. The research has helped to increase the power
plant productivity from 22% for wind turbines installed before 1998 to an average
of 33% today (Council 2011).
A wind power conversion system involves wind turbine, induction or syn-
chronous generator, power electronics, and control. The analysis of the functionality
and efficiency of wind turbines is an important step to improve the design of
wind energy conversion systems and to develop efficient effective generator control
strategies. Wind speed is not predictable or reliable all time, therefore it is difficult
to use a real wind turbine in the laboratory environment. So there is the need to build
a WTE to emulate different aerodynamic phenomena present in the rotor of a wind
turbine and to provide an easier testing environment for generator control strategies.
WTE can drive an electrical generator in similar ways as a real wind turbine for
given wind velocities in a controlled and safe environment. Moreover, it is an
excellent tool for academic purposes such as control, operation, and characterization
of a wind turbine.
The goal of this project was to develop such a test platform using as much real-
world data as possible. The WTE is composed of a 0.8 KW DC motor coupled to
the shaft of a permanent magnet synchronous generator (PMSG). The DC motor
has an electrical drive system controlled in real time by a software developed in
Matlab/Simulink environment. The interface between software and drive system is
performed by a Data Acquisition Card NI6009 (DAC).

11.2 Literature Survey

Many studies describe wind turbine emulators either based on DC motor (Arifuj-
jaman et al. 2008; Kouadria et al. 2013; Li et al. 2007), induction motor (Kenneth
et al. 2016; Martinello et al. 2016; Soltoski et al. 2016; Voltolini et al. 2012) or
permanent magnet synchronous machine (PMSM) (Yan et al. 2016). The most
11 Emulation of Wind Turbine for Standalone Wind Energy Conversion Systems 229

common type of motor used in the construction of WTE is DC motor because it

is easy to implement and because the direct relationship between armature current
and the torque produced by the machine.
There are several approaches for emulators used in different situations depending
on the desired requirements such as the power range and the type of wind
turbine. The input to the emulator is wind speed profile, which is instantiated in
a mathematical wind turbine model to determine the reference torque output which
in turn is used to control the torque of the motor shaft. This mathematical model is
one of the key issues for the efficiency of WTE. The fundamental aspect of turbine
modeling is the aerodynamics effects, that is, a model which interprets the effects of
the wind on the turbine system. The fundamentals of this idealized model are widely
available in the literature (Han and Chen 2009; Lu et al. 2012; Pao and Johnson
2009) where the aerodynamic power output of the wind turbine is a function of
parameters such as wind speed, turbine rotor speed, air density and turbine rotor
area.
Another key issue in the aerodynamic model of the wind turbine is the blade pitch
control. The wind turbine operates mainly in four regions delimited by the critical
operating points. The first critical point encountered by the wind turbine is the cut-
in speed. Below this speed the wind turbine does not operate at all; once this speed
reached, the turbine begins to produce power. As the wind speed increases the next
critical point that is reached is the rated speed. At this point, the blades have been
pitched to capture as much energy from the wind as it can and the generator produces
as much power as possible. Once the wind speed exceeds the rated wind speed,
the turbine blades begin to adjust their pitch to minimize the capturing of energy
by shedding some of the energy in the wind. The turbine continues to produce its
maximum power even though the wind speed is larger than the turbines rated value.
As the wind speed increases, the turbine continues to adjust the pitch of its blades
until they reach their mechanical limit. At this point, the turbine must shut-down
to prevent damages. This point is well known the cut-out speed (Clark et al. 2010;
Hardy and Jewell 2011).
These regions of operation are shown in Fig. 11.1 where θ represents the blade
pitch angle. The best way to exploit the maximum capacity of the wind in wind
energy conversion systems is through the variable speed wind turbines with pitch
angle control. However previous studies into the emulation of wind turbine system
do not pay attention to the control of pitch angle system (Van et al. 2015).
In order to make the WTE system more efficient and able to provide a complete
substitution of the real wind turbine, we presented the pitch control using a Fuzzy
Logic Controller for their usefulness when the system is nonlinear, such as the wind.
230 L. Benaaouinate et al.

Fig. 11.1 Regions of wind turbine operation

Aerodynamics Mechanical Electrical

Stage Stage Stage

Gearbox
Wind

Generator
Transformer
ωT TT
Grid

ωG TG

Capacitive
bank

Fig. 11.2 Wind power system

11.3 Dynamic Models for Wind Power System

A wind power system can be described using three mathematical models: aerody-
namic, mechanical, and electrical model. Figure 11.2 shows the basic structure of
the wind power system.
11 Emulation of Wind Turbine for Standalone Wind Energy Conversion Systems 231

11.3.1 Aerodynamic Model

11.3.1.1 Aerodynamic Power

The mechanical power extracted by the wind turbine is expressed as:

1
PT = πρR 2 V 3 Cp(λ, β) (11.1)
2
where, ρ is the air density (kg/m3), R is the turbine radius (m), V is the wind speed
(m/s) and Cp is the power coefficient which is a function of both blade pitch angle
β and tip speed ratio (TSR) λ. The tip speed ratio λ is expressed as:

ωR
λ= (11.2)
V
where, ω is the angular speed of the turbine rotor (rad/s), and R is the radius of
turbine blades and V is the wind speed (m/s). The curve power that relates Cp to
the different value of pitch angle β and TSR λ is different for each wind turbine
and it is obtained from direct measurements of the turbine in operation. The generic
equation used to model the Cp is obtained from the curve power for different points
of operation of the turbine. A numerical approximation that is commonly used for
the determination of Cp is:

c2 −c5
Cp (λ, β) = c1 ( − c3 β − c4 ) exp + c6 λ (11.3)
λi λi

1 1 c8
= − 3 (11.4)
λi λ + c7 β β +1

Table 11.1 shows the parameters of the performance coefficient.

The torque developed in the shaft of a wind turbine is expressed as:

1 3 2
2 πρR V Cp(λ, β)
Ta = (11.5)
λ
According to Eq. (11.5), the mechanical torque of the wind turbine is relative to
the wind speed. Therefore it is expected to produce a torque with a stochastic profile.

Table 11.1 Parameters of c1 c2 c3 c4 c5 c6 c7 c8

the performance coefficient
0.5109 116 0.4 5 21 0.0068 0.08 0.0035
232 L. Benaaouinate et al.

R
L

Extrados
Rotor Plane rΩ (1+a’) D

β dos
φ Intra
α C
V (1–a)

w V

Fig. 11.3 Profile of a turbine blade

11.3.1.2 Blade Aerodynamics

Blade Element Momentum (BEM) method can be used to examine how a wind
turbine works and to predict the blade forces produced due to the interaction of the
blade and the wind. When the wind passes through the rotor plane and interacts with
the moving rotor, a resultant relative velocity at the blade is introduced, depicted as
W in Fig. 11.3, which is the difference between the undisturbed wind velocity and
the blade tip velocity.

W = V 2 (1 − a 2 ) + Ω 2 r 2 (1 + a  )2 (11.6)

Where V is the wind velocity perpendicular to the rotor plane, r is the distance of
blade element from the rotor axis, a and a’ are the flow factors and Ω is the angular
velocity of blades.
The aerodynamics forces on the blade are the lift force and the drag. The lift
force is:
1
L= ρcW 2 CL (11.7)
2
The drag force is:

1
D= ρcW 2 CD (11.8)
2
where CL and CD are respectively the lift and drag coefficient, c is the chord length
of the blade and ρ is the air density.
The resultant relative velocity at the blade W creates an angle with the blade
chord C, depicted as α in Fig. 11.3, the angle of attack. It is controlled by rotating
the blade about its axis. The pitch angle β is the angle formed by the blade chord C
and the rotor plane. The power of the wind turbines depends on the wind velocity,
11 Emulation of Wind Turbine for Standalone Wind Energy Conversion Systems 233

Gearbox

wt
Tls wls
kls Jgen
Ta
Ths

Br whs Tgen
Bls

Bg

Fig. 11.4 Mechanical model of wind turbine

it increases as the wind increases. Therefore the wind turbine must be protected
against the risk of damages from the strong wind. This can be achieved by rotating
the blade about its axis in a position where a part of incoming wind will pass by the
wind turbine. The angle of attack α decreases when the pitch angle β increases. The
lift force L decreases as well and this reduces the wind turbine’s mechanical power.

11.3.2 Coupling System and Transmission

The mechanical model of a wind turbine is currently modeled by two rotating

masses. Such model includes the rotor of the wind turbine and the generator shaft.
A gearbox or direct coupling connects them. Figure 11.4 shows the structure of the
mechanical model of the wind turbine.
The parameters presented in Fig. 11.4 are described below:

Ta : Torque of the turbine applied on the rotor;

Tls : Torque applied on low-speed shaft;
Ths : Torque applied on high-speed shaft;
Tgen : electromagnetic torque of generator;
ωt : rotor angular speed;
ωls : angular speed of the low-speed shaft;
ωhs : angular speed of the high-speed shaft;
Br : rotor damping effect;
Bls : low speed damping effect;
Bg : high speed damping effect;
kls : Stiffness of low speed shaft;
Jg : moment of inertia of the generator;
ng : speed multiplication ratio.

Thus, the equations for the mechanical model of the wind turbine are:
234 L. Benaaouinate et al.

Ta − Tl s = Jr ẇt + Br wt (11.9)

Tl s = Bl s(wt − wl s) + kl s(θt − θl s) (11.10)

Th s − Tg en = Jg w˙h s + Bg wh s (11.11)

Tl s
ng = (11.12)
Th s

11.3.3 Electrical Model

The electrical dynamic equations of permanent magnet synchronous generator

PMSG are expressed as:

did
Vd = Rs id − ωe Lq iq + Ld (11.13)
dt

diq
Vq = Rs iq + ωe (Ld id + Ψf ) + Lq (11.14)
dt

did 1
= (−Rs id + ωe (Lq s + Ll s)iq + Vd ) (11.15)
dt Ld s + Ll s

diq 1
= (−Rs id − ωe [(Ld s + Ll s)id + Ψf ] + Vq ) (11.16)
dt Lq s + Ll s

Where ωe is the angular speed of the reference frame (ωe = pωg ), p is the number
of pole pairs, Ld and Lq are the inductance of the stator winding in the synchronous
reference frame, Łl s is the leakage inductance, Rs is the stator resistance, and ψf is
the flux constant of the generator rotor.
Figure 11.5 shows the equivalent circuit of the PMSG in d-q synchronous rotating
reference frame.

a b
Lds + Lls Rs Lqs + Lls Rs

id iq
We(Lqs+Lls).iq We((Lds+Lls).id+ψf)
vq

Fig. 11.5 Equivalent circuit of the PMSG: (a) d-axis equivalent circuit, (b) q-axis equivalent
circuit
11 Emulation of Wind Turbine for Standalone Wind Energy Conversion Systems 235

Fig. 11.6 Structure of wind turbine emulator

The electrical torque Tg of the generator is expressed as:

3
Tg = (ψp miq + (Ld − Lq )id iq ) (11.17)
2

11.4 Wind Turbine Emulator Design

The structure of the WTE using a permanent magnet DC motor is represented in

Fig. 11.6.
In order to emulate the real behavior of the wind turbine with a DC machine and
its control, it is necessary to determine the electromagnetic reference torque. The
inputs to the wind turbine’s mathematical model are angular speed, pitch angle, and
wind speed. The reference torque derived from the model of the turbine is multiplied
by Gear Ratio and motor constant to obtain the reference current, which is compared
with the actual DC motor armature current. The difference between the two currents
is the input of a proportional-integral (PI) controller. It is then used to generate
PWM gate pulses to drive the transistor of the DC-DC buck converter that regulates
the DC motor armature voltage. Then the variation in armature voltage generates
a variation in the speed of DC motor, which allows obtaining a mechanical torque
which depends on the wind velocity and varies over the time in a similar way as the
torque of the wind turbine.
236 L. Benaaouinate et al.

11.5 Pitch Angle Control

11.5.1 Dynamic Actuator

The main objectives for blade angle control are:

• Optimization of the wind turbine output power, by producing as much power as
possible with the available wind.
• Prevention for mechanical power does not exceed the rated power in high winds.
• Protection against overloading of the turbine structure and risks of damage to
their physical structure at the strong wind.
The pitch angle actuator consists of a mechanical and hydraulic system, it is used
to turn the blades along their longitudinal axis. The blade angle control can be done
in two stages (Macedo and Mota 2012)
1. Pitch control: the mechanical power is reduced when the pitch angle grows. This
principle is applied, in most cases, in variable wind speed turbine. Fixed speed
turbines can also use this control;
2. Active stall control: the mechanical power is reduced when the pitch angle
decreases. This type of control is commonly applied in fixed wind speed turbines.

11.5.2 Pitch Angle Control Using Fuzzy Control

As the aerodynamic power depends on the wind speed, it is nonlinear. Therefore the
control of variable speed wind turbine could not perform correctly with conventional
methods (Sahoo et al. 2016). The use of fuzzy logic controller FLC can overcome
these problems. It does not require a well-known system, ease of its modeling and
its implementation. It is presented as an alternative for Proportional-Integral (PI)
controllers currently used in wind turbine Pitch Angle Control (Yin et al. 2015). The
FLC uses the rules determined with linguistic expressions derived from a human
expert who has some experience and knowledge about the system. Figure 11.7
shows the diagram of general FLC. It consists of three stages: an input stage, a
processing stage, and output stage. The Fuzzification stage translates the input data
to the fuzzy representation incorporating the vagueness and imprecision in a natural
language. The processing stage or the inference consists in evaluating the rules and
generates results. The De-Fuzzification stage converts the results into a specific
control output value.
Figure 11.8 shows the pitch angle control block diagram using fuzzy control.
The present work developed the fuzzy control using the Fuzzy Logic ToolboxTM
of Matlab/Simulink. Figure 11.9 represents Membership function plots of input and
output of the FLC using Mamdani’s fuzzy inference method to define how each
point in the universe of discourse is mapped to a degree of membership between 0
and 1.
11 Emulation of Wind Turbine for Standalone Wind Energy Conversion Systems 237

Fig. 11.7 Diagram of general FLC

Fig. 11.8 Pitch angle block diagram using Fuzzy Logic controller

a b

Fig. 11.9 The Membership function plots of the FLC: (a) Membership function input, (b)
Membership function output

Table 11.2 Rules base for ΔP NL NM NS ZE PS PM PL

FLC
βr NL NM NS ZE PS PM PL

Table 11.2 gives the rule base modeled in this work for the FLC.
The abbreviations used in the presented tables are defined as follows:
• NM is negative medium;
• NS is negative small;
• ZE is Zero;
• PL is positive large;
• PM is positive medium;
• PL is positive large.
238 L. Benaaouinate et al.

Fig. 11.10 Block diagram of the WTE

11.6 Simulation Results

Based on the different mathematical models of wind power system presented

previously, a software application was developed in Matlab/Simulink to compare
them with the waveforms of the experimental results. Figure 11.10 presents the
block diagram of the implemented and simulated WTE system. The wind profile
exploited in this simulation is presented in Fig. 11.14 and it is modeled in the
deterministic form of a sum of several harmonics:

V (t) = 9 + 0.2 sin 0.0021t + 2 sin 0.0053t + sin 0.0259t + 0.2 sin r0.0733t
(11.18)
As it can be analyzed from Fig. 11.11, there is a high level of correspondence
between wind speed profiles applied to wind turbine model and the characteristics of
different responses of WTE such as the torque, the speed and the armature current of
DC motor. This demonstrates that the PI controller has a good performance tracking.
Figure 11.11 shows the different values of the pitch angle β for different wind
velocities. Adjusting the pitch angle of the blades provides an effective way to
regulate or limit the turbine performance in strong wind speeds. The pitch angle
β, increases and the angle of attack, decreases when the wind speed increases. The
lift force decreases as well and the mechanical power of WTE is then reduced.
The simulation waveforms obtained in this section will be used to verify the
emulator dynamics in the next section.

11.7 Experimental Results

In order to duplicate the similar waveform of the aerodynamic torque of the wind
turbine on a small scale, an adaptation of the torque’s magnitude is required. This
adaptation is simply achieved through a gain which is the ratio between the rated
power of the machine and that of the turbine (Fig. 11.12):
 a

Speed (rd/s)
200

100

0
0 500 1000 1500 2000
Time (s)
b 5
Armature Current (A)

1
0 500 1000 1500 2000
Time (s)
c 14
Wind Speed (m/s)

12

10

6
0 500 1000 1500 2000
Time (s)
d 5.2

5
Torque (Nm)

4.8

4.6

4.4

0 500 1000 1500 2000

Time (s)
e 4
Pitch angle β(°)

3.5

2.5

2
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Time (s)

Fig. 11.11 Simulation results of WTE: (a) Variation of rotational speed of DC motor, (b) Variation
of armature current of DC motor, (c) Wind speed profile, (d) Variation of torque of DC motor, (e)
Variation of pitch angle β
240 L. Benaaouinate et al.

Fig. 11.12 Adaptation of the aerodynamic torque

Table 11.3 Parameters of Equipment Parameters

the DC motor and PMSG
DC motor Type MSS-8
Permanent magnet
Rated power: 0.8 kW
Rated armature voltage: 90 V
Rated armature current: 10.7 A
Rated speed: 3000 rpm
PMSG Type MA-6
Stall torque: 3.6 N
Rated armature voltage: 127 V
Stall current: 4.2 A
Max mechanical speed: 6000 rpm

Pd cm
G= (11.19)
PT

The test setup consists of the following:

• DC motor with rated values shown in Table 11.3;
• Permanent Magnet Synchronous Generator (PMSG) with rated values shown in
Table 11.3;
• DC-DC Buck Converter;
• Electronic card with voltage sensors and current sensors;
• Calculator;
• Data acquisition card NI6009.
Table 11.3 shows the parameters of DC motor and PMSG.
Figure 11.13 shows the hardware structure of the wind energy system and the
photograph of the experimental setup.
A real time calculator monitors the control parameters whilst in operation. The
Data Acquisition Card NI6009 (DAC) allows communication between the setup and
the user. Various turbine parameters like the wind velocity and pitch angle can be
set by the user. To calculate the current reference the program reads wind velocity
and pitch angle value from the input data file and measured armature current from
current sensor and speed from the tachometer. The calculated reference current is
11 Emulation of Wind Turbine for Standalone Wind Energy Conversion Systems 241

a b

Fig. 11.13 Experiment setup: (a) Hardware structures of wind energy systems, (b) Photograph of
experiment setup

given to the PI controller. The output of PI controller is passed to hardware interface

(PWM Control Circuits) through analog output channel of DAC and finally, DC-
DC buck converter drives the DC motor where the firing pulses are generated by
changing the armature voltage. The output voltage profile of the generator (PMSG)
directly depends on the shaft speed.
Most of the effort during the experimentation was focused on adjusting the
weighting for the PI controller for DC motor.
Figure 11.14 shows the results of rotational speed, armature current and torque
characteristics of WTE for different wind velocities. As it can be concluded from
the figures, there is a high level of correspondence between the wind speed profiles
applied to wind turbine model and characteristics of different responses of WTE
such as the torque, the speed, and the power coefficient. It means that the PI
controller has a good performance tracking.
As it can be depicted from obtained figures, there is a good deal of match between
simulated and experimental results.
The results verified that the WTE can efficiently reproduce the steady-state and
dynamic behavior of a wind turbine for a given wind velocity.

11.8 Conclusion

This chapter described the main components of a WTE based a on permanent

magnet DC motor controlled by PWM implemented in the laboratory.
In order to check the emulator’s dynamics, an experimental evaluation was pre-
sented and compared with simulated ones. Considering simulation and experimental
waveforms for different wind conditions and their effects on several aerodynamic
and mechanical variables, The WTE can supply all required parameters of the wind
turbine system such power coefficient, tip speed ratio, output torque and output
power.
242 L. Benaaouinate et al.

The control of pitch angle is one of the main steps in the emulation of the wind
turbine for providing more realistic conditions for the emulator and a complete
substitution of the real wind turbine. However, most of the previous studies into
the WTE systems do not take into account the control of pitch angle despite the best
way to extract the maximum power of the wind is through the variable speed wind
turbines with pitch angle control.
Fuzzy Logic Controller FLC is used in the pitch angle control system for their
high accuracy, fast dynamic response, stability, solving nonlinear systems such as
the wind which has the nonlinearity as its main characteristic.

a
400

350

300
Speed (rd/s)

250

200

150

100

50

0
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Time (s)

b
8

6
Armature Current(A)

0
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Time (s)

Fig. 11.14 Experimental results of WTE: (a) Variation of rotational speed of DC motor, (b)
Variation of armature current of DC motor, (c) Wind speed profile, (d) Variation of torque of
DC motor
11 Emulation of Wind Turbine for Standalone Wind Energy Conversion Systems 243

c
14

12
Wind (m/s)

10

4
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Time (s)

d
20

15
Torque (Nm)

10

0
0 200 400 600 800 1000 1200 1400 1600 1800 2000
Time (s)

Fig. 11.14 (continued)

A key limitation of this research is the high cost of the DC motor in the structure
of WTE systems compared to other electrical machines in the same power range
and also it demands higher maintenance requirements due to its commutators and
brushes. The next stage of our research will be an emulation system of a wind
turbine based on an induction motor (IM) to overcome this limitation.

References

Arifujjaman, M., Iqbal, M., & Quaicoe, J. E. (2008). Emulation of a small wind turbine system
with a separately-excited DC machine. IU-Journal of Electrical & Electronics Engineering,
8(1), 569–579.
Clark, K., Miller, N. W., & Sanchez-Gasca, J. J. (2010). Modeling of GE wind turbine-generators
for grid studies. GE Energy, 4, 4.13–4.17.
244 L. Benaaouinate et al.

Council, G. W. E. (2011). Annual market update 2011. Global Wind Report, Online: http://www.
gwec.net
Han, K., & Chen, G.-Z. (2009). A novel control strategy of wind turbine MPPT implementation
for direct-drive PMSG wind generation imitation platform. In Power Electronics and Motion
Control Conference, 2009. IPEMC’09. IEEE 6th International, Wuhan, China (pp. 2255–
2259). IEEE.
Hardy, T., & Jewell, W. (2011). Emulation of a 1.5 MW wind turbine with a DC motor. In Power
and Energy Society General Meeting, 2011 IEEE, Detroit, MI, USA (pp. 1–8). IEEE.
Kenneth, A., et al. (2016). Emulator for fixed pitch wind turbine studies. Renewable energy.
London: Taylor & Francis Ltd.
Kouadria, S., Belfedhal, S., Meslem, Y., & Berkouk, E. M. (2013). Development of real time wind
turbine emulator based on DC motor controlled by hysteresis regulator. In 2013 International
Renewable and Sustainable Energy Conference (IRSEC), Ouarzazate, Morocco (pp. 246–250).
IEEE.
Li, W., Xu, D., Zhang, W., & Ma, H. (2007). Research on wind turbine emulation based on
DC motor. In 2nd IEEE Conference on Industrial Electronics and Applications, ICIEA 2007,
Harbin, China (pp. 2589–2593). IEEE.
Lu, L., Xie, Z., Zhang, X., Yang, S., & Cao, R. (2012). A dynamic wind turbine simulator of the
wind turbine generator system. In 2012 Second International Conference on Intelligent System
Design and Engineering Application (ISDEA), Sanya, Hainan, China (pp. 967–970). IEEE.
Macedo, A., & Mota, W. (2012). Wind turbine pitch angle control using fuzzy logic. In
Transmission and Distribution: Latin America Conference and Exposition (T&D-LA), 2012
Sixth IEEE/PES Montevideo, Uruguay (pp. 1–6). IEEE.
Martinello, D., Carati, E. G., da Costa, J. P., Cardoso, R., & Stein, C. M. (2016). Emulation of wind
turbines. In A. G. Aissaoui & A. Tahour (Eds.), Wind turbines-design, control and applications.
Rijeka: InTech.
Pao, L. Y., & Johnson, K. E. (2009). A tutorial on the dynamics and control of wind turbines and
wind farms. In American Control Conference, ACC’09, St. Louis, MO, USA (pp. 2076–2089).
IEEE.
Sahoo, S., Subudhi, B., & Panda, G. (2016). Pitch angle control for variable speed wind turbine
using fuzzy logic. In 2016 International Conference on Information Technology (ICIT),
Bhubaneswar, India (pp. 28–32). IEEE.
Soltoski, J. C. F., dos Santos, P. T. P., & Font, C. H. I. (2016). Development of a small scale wind
turbine emulator work bench. In 12th IEEE International Conference on Industry Applications
(INDUSCON), 2016, Curitiba, Brazil (pp. 1–8). IEEE.
Van, T. L., Nguyen, T. H., & Lee, D.-C. (2015). Advanced pitch angle control based on fuzzy
logic for variable-speed wind turbine systems. IEEE Transactions on Energy Conversion, 30(2),
578–587.
Voltolini, H., Granza, M. H., Ivanqui, J., & Carlson, R. (2012). Modeling and simulation of
the wind turbine emulator using induction motor driven by torque control inverter. In 10th
IEEE/IAS International Conference on Industry Applications (INDUSCON), 2012, Fortaleza,
Brazil (pp. 1–6). IEEE.
Yan, J., Feng, Y., & Dong, J. (2016). Study on dynamic characteristic of wind turbine emulator
based on PMSM. Renewable Energy, 97, 731–736.
Yin, X.-x., Lin, Y.-g., Li, W., Gu, Y.-j., Wang, X.-j., & Lei, P.-f. (2015). Design, modeling and
implementation of a novel pitch angle control system for wind turbine. Renewable Energy, 81,
599–608.
Chapter 12
SSDI-Max Control and Its Applications
in Renewable Energies

Aida Chérif, Djamila Zehar, Nabil Derbel, and Claude Richard

Abstract The vibration control using piezoelectric elements is an area interesting

several industrial sectors. Modal synchronized switch damping on inductor control
is a vibration damping technique that combines advantages of passive and active
control techniques based on a modal strategy. Within this framework, we propose
an improved control technique, which is called modal SSDI-Max. The particularity
of this new approach is to maximize the self-generated voltage amplitude by
a proper definition of the switch instants (voltage inversion) according to the
chosen targeted mode. Following the basic modal synchronized switch damping on
inductor technique, the switch is synchronized with the chosen modal coordinate
extremum. In the investigated approach, the voltage is increased by waiting for
the next voltage extremum following immediately any targeted modal coordinate
extremum in a given time window. This chapter presents simulations performed on
a model representative of a clamped plate. Damping results are given in the case
of multimodal, pulse or noise excitations. The chapter analyses the control time
window influence on the damping performance of the system.

Keywords Piezoelectric · Smart structures · Modal control · Semi active

control · SSDI-Max · Time window

A. Chérif ()
LAS Laboratory, University of Bordj Bou Arréridj, El Anasser, Algeria
e-mail: [email protected]
D. Zehar
LSI Laboratory, University of Bordj Bou Arréridj, El Anasser, Algeria
e-mail: [email protected]
N. Derbel
CEMLab Laboratory, University of Sfax, Sfax, Tunisia
e-mail: [email protected]
C. Richard
LGEF Laboratory, INSA, University of Lyon, Lyon, France
e-mail: [email protected]

© Springer Nature Singapore Pte Ltd. 2019 245

N. Derbel, Q. Zhu (eds.), Modeling, Identification and Control Methods
in Renewable Energy Systems, Green Energy and Technology,
https://doi.org/10.1007/978-981-13-1945-7_12
246 A. Chérif et al.

12.1 Introduction

All mechanical systems are subjected to various conditions that may result in
vibration motion. These vibrations often lead to material fatigue, structural damage
and failure, deterioration of system performances and increase of noise levels.
These effects are usually prominent around natural frequencies of the system. Such
vibrations cannot be tolerated and, therefore, developing a strategy for reducing
these vibrations has been a major focus of research. This subject has considerable
industrial consequences. Vibrations and acoustic control are used in many domains
such as automobiles, aeronautics, aerospace, machine tools, etc.
In vibration control the aim is the amplitude limitation of the vibrating structure
in order to improve the reliability of their functions. Vibration control of a structure
also allows increasing its resistance to fatigue then increasing the duration of its live.
One of recent strategies of vibration and acoustic control is the control by intelligent
materials. In general, three types of vibration control using piezoelectric inserts are
pointed out in the literature:
• Passive techniques (Eielsen and Fleming 2010; Forward 1979; Hagood and
Von Flotow 1991) consist of connecting a passive electrical network (usually
an inductive network) to the piezoelectric insert(s), these techniques can be
difficult to implement: attainable trade-off between performances and bandwidth
or feasibility of the required inductor are the usual drawbacks (Wu and Bicos
1997).
• Active techniques (Fuller and Jones 1987; Meyer and Collet 2011; Preumont
et al. 2002; Snyder and Hansen 1991) consist of a control system with actuators
supplied by a controlled electric field usually generated by an amplifier driven by
a feedback loop using sensors and according to a given control theory approach.
It is a very powerful and global technique allowing a diversity of strategies. In
these techniques, there is an external source supplying both the control energy
for the control loop circuitry and the operative energy delivered by the power
amplifier.
• Semi active techniques (Clark 2000; Cunefare and Holdhusen 2004; Nieder-
berger et al. 2004; Petit et al. 2004; Richard et al. 2000) are techniques using
no external operative energy. They rely on switches allowing management of the
piezoelement electrostatic energy and/or electric boundary conditions.
Among these vibration damping techniques are synchronized switch damping
(SSD) techniques originally developed by (Guyomar and Badel 2006). The overall
principle is to introduce, using a proper switching sequence, strong nonlinearities in
the voltage generated by the piezoelectric inserts, which have the effect of modifying
significantly the voltage amplitude and phase, resulting in increased dissipation and
mode damping. Among these methods, synchronized switch damping on inductor
(SSDI) consists in inverting the piezo voltage each time it reaches an extremum.
Another way to address the multimodal problem is the so-called modal SSDI
control developed by Harari et al. (2009). The principle is to combine advantages
12 SSDI-Max Control and Its Applications in Renewable Energies 247

of both active control techniques and the semi active SSD damping. Relying on
a modal model of the structure, this method allows controlling a given targeted
mode. The principle is to shape the piezoelement voltage nearly out of phase with
the targeted modal speed. To do so, the technique relies on a modal filter allowing
the computation of the modal coordinates of the vibrating structure and on an SSDI
switch device synchronized with the extremum of the targeted modal coordinate. An
improvement of the modal SSDI control was proposed by Richardet et al. (2009);
the purpose of this latter technique called “SSDI-Max” is to use part of the energy
of the non targeted mode to improve the piezoelement voltage. The inversion is
scheduled at the maxima of the target modal coordinate, unless the voltage varies
with the corresponding slope at that moment.
This chapter presents an analysis of the performance of the SSDI-Max damping
technique with a plate-structure. It relies on simulations, made with the Matlab-
Simulink environment, using a realistic model of a plate structure previously
identified. For this structure the piezoelement coupling coefficient are not optimal
and very representative of a real target application structure. Moreover, compared to
the beam case, frequency modes are rather close to each other. The proposed method
aims at maximizing the amplitude of the piezoelectric actuator by the definition
of an optimal switching time according to the chosen targeted mode. The smart
structure modeling along with the plate structure investigated is described. The
SSDI-Max strategy is then exposed. The performances of the SSDI Max method for
the control of single modes of the structure are described in the case of sinusoidal,
pulse and white noise excitation. Finally, the influence of the delay time window
used is described again for various type of excitation.

12.2 Vibration Controls by Piezoelectric Elements

In many industrial and defense applications, vibrations present a major scientific

and technological issue. The objectives sought by vibration damping have different
aspects, such as improving comfort in acoustics and mechanics, the stability of
increasingly light structures or increasing the life of materials. . . In the field of
embedded systems, the aim is to reduce vibrations as efficiently as possible while
limiting the mass and the volume of the control system. In order to achieve this
goal, it is necessary to use efficient methods while reducing or even eliminating the
energy required to control these vibrations.
In this part we present the state of the art on the vibration control, the different
types of control that excite and finally we briefly give a summary on nonlinear
methods developed in the laboratory of electrical and ferroelectric engineering
LGEF for the control.
248 A. Chérif et al.

12.2.1 Passive Control

Passive control has the advantage of requiring neither energy nor information to
function. It consists of connecting the piezoelectric element to a circuit consisting
of passive components, for example a resistor. The purpose of this circuit is to
dissipate the electrical energy provided by the piezoelectric element Joule effect.
The advantages of passive techniques are:
• Easy adjustment of the electrical system,
• A weight and a size of elements to be added to the structure insignificant,
• An unconditionally stable system.
While passive damping can greatly improve the damping of the system, beyond
their simplicity, passive techniques possess certain number of limitations. First, for
low structural frequencies, the value of inductance to the resonant shunt circuit
can be quite large. A passive inductor of this size could easily preclude its use
for lightweight applications (air/space, tennis rackets, etc.). Secondly, the resonant
shunt technique can only be compatible to one structural mode (Carriere 2010;
Franklin et al. 1990; Wu and Bicos 1997).

12.2.2 Active Control

This type of control is usually more performing than passive technique. But the
fundamental difference is on the need of the external source of energy for driving
control actuators. Nevertheless, they are complex for practical implementation and
in general rely on a complete chain comprising sensors, calculation unit, power
amplifiers and actuators. The raw principle of the active control uses an external
source to generate forces that interfere with the wave or the system vibration in
order to minimise it. This method needs a network of sensors along with a good
model of the structure to get a situation of the vibration or acoustic and actuators
to act on the vibration or the acoustic field. These diverse elements are controlled
by an advanced system of management (microcontroller, DSP. . . ) that calculate the
optimal control permitting the minimisation of waves or vibrations of the system
(Davis and Lesieutre 2000).

12.2.3 Semi-active Control

In order to improve damping performances without using external operative energy

for the actuator drive, several semi-active vibration control techniques have been
proposed and developed. The advantages of semi-active vibration control present a
much lower energy requirement, a rather overall low cost, and a lower complexity
12 SSDI-Max Control and Its Applications in Renewable Energies 249

compared to active control systems. Better damping performances can be obtained

compared to passive control systems. The Synchronized Switch Damping (SSD)
techniques have been developed by Richard et al. (2000). SSD techniques reach a
good trade-off between the simplicity, external energy requirements and vibration
control performances. This nonlinear technique consists of connecting intermit-
tently the piezoelectric element to a passive shunt (small resistance or inductor)
each time the voltage (or strain) reaches an extremum. It is implemented with a
simple switch driven during short periods synchronously with the structure motion.
No external operative energy is needed. It is extracted from the structure itself as a
result of the switch process.

12.2.3.1 SSDI Control

The SSDI technique consists of a switching device in parallel with the piezoelectric
elements shown on Fig. 12.1a. The capacitance of piezoelectric element C0 , induc-
tance L and resistance R could constitute an LCR oscillator when the switch is
turned on. The switch is triggered-on each time the piezoelectric voltage reaches
an extremum. When the switch closes, an oscillating discharge is initiated. The on-
time ti is set to one half of the electric period of oscillator network as defined in
Eq. (12.1). The switch is therefore kept closed until the piezoelectric voltage has
been reversed. The voltage inversion results from the oscillating discharge of the
piezoelectric element blocked capacitance in the inductance during the switching
time. Because of a small inductor L value, this switching time ti is much smaller
than a mechanical oscillation period.
If the structure is excited by a sinusoidal force, when this switching sequence
is repeated, the displacement of the structure and the piezoelectric voltage result in
the waveforms shown on Fig. 12.1b. It is seen that the switch is nearly always open
excepted for each voltage extremum. Due to the resistance R, the absolute value of
piezoelectric voltage after the inversion is smaller than the initial value. This loss
is modeled by the electrical quality factor Qi of the oscillation network, which is
expressed in Eq. (12.2). It is also quantified by the inversion factor γ defined by
Eq. (12.3):

ti  π LC0 (12.1)
)
1 L
Qi = (12.2)
R C0
 
Vaf ter π
γ = = exp − (12.3)
Vbef ore 2Qi

where, Vbef ore and Vaf ter are the absolute value of the piezoelectric element voltage
before and after the switching, respectively.
250 A. Chérif et al.

Fig. 12.1 Classic SSDI method. (a) SSDI circuit. (b) SSDI waveforms

Fig. 12.2 Sum of the displacement of three modes

12.2.3.2 Modal-SSDI Control

The classical low-powered SSD techniques perform well in mono-modal excitation.

However, it hardly deals with cases of the multimodal and complex vibrations. The
principal difficulty is the definition of the proper switching sequence. It is illustrated
on Fig. 12.2 with a composite displacement u(t) built as the superposition of these
modal components. Many extremum are visible making the choice of the optimal
switching time.
To overcome this limitation and work on a multimodal problem with a wide
bandwidth excitation, a modal approach has been implemented. It is assumed that
the motion of the vibrating structure can be decomposed on a given modal basis.
This decomposition is a good approximation in the case of structures with low
viscous losses for which the resonance modes are clearly distinct and uncoupled.
Based on the principle of the modal approach, Modal-SSDI method was developed
by Harari et al. (2009).
12 SSDI-Max Control and Its Applications in Renewable Energies 251

Fig. 12.3 Piezoelectric

voltage V waveform when
inversion is focused on the
modal displacement of the
first mode

The modal SSDI control is developed and based on basic SSDI method. It
consists of synchronizing the switch sequence on a given modal coordinate instead
of the voltage. The voltage across the actuator is therefore inversed when the
displacement of the targeted mode is extremum as shown on Fig. 12.3. The principle
and strategy of semi-active Modal-SSD for multimodal control is illustrated in
Fig. 12.4.
The targeted modal displacement qi can not be directly monitored by any kind
of direct sensor. Several methods can be used to derive this displacement by a
set of appropriate sensors. The technique implemented by Chérif et al. (2012), is
based on the use of an observer which is designed using a modal model of the
structure. Figure 12.4 is a schematic of the global control loop for modal SSDI
control. On the upper side, the real system comprises the structure itself, which
can be represented by a state-space equation. It is excited by an external force
and controlled via the actuator voltage Va . The motion of the structure can be
represented by the state space vector (q, q̇) consisting of modal displacements and
modal speed coordinates. Moreover, many actuation or sensing piezoelectric patches
are installed on the structure. The open-circuit voltages of the sensors Vs can be used
for state monitoring of the structure. The voltage Va developed on the piezoelectric
element used as control actuator is processed according to the usual SSDI technique.
On the lower side, there is the control observation loop. The purpose of this loop is
to derive from the measurement voltages Vs , the proper modal coordinate qi used
to trigger the switch device. This observation loop is based on an observer, which
estimates both the space-state vector (q̂, q̇)ˆ and the sensor voltages measurement
V̂s . This observer consists of a feedback loop gain L and of the space-state model
of the structure. Modal state x is estimated and corrected due to the convergence,
more or less rapidly as a function of gain L, between the measured and estimated
outputs Vs and V̂s . The gain matrix L must be chosen so that the error on the state
is globally stable and vanishes exponentially with a dynamic much quicker than the
structure itself. To calculate this gain of the observer, two methods can be used such
as pole placement or LQG techniques (Chérif et al. 2012).
252 A. Chérif et al.

Fig. 12.4 Modal SSDI control architecture (Chérif et al. 2012)

12.3 Smart Structure Definition

12.3.1 Smart-Structure Definition

The structure coupled with one piezoelectric element is in the simplest approach
described by the single of degree of freedom electromechanical model as shown
in Fig. 12.5. The structure mass M is driven by an external force F . u is the
displacement of the vibration structure and V is the voltage of the piezoelectric
element. Based on usual hypothesis, the mechanical and electric equations govern-
ing the motion of the single degree of freedom structure are: The electromechanical
behavior equations of a smart structure using usual assumption are (Harari et al.
2009):

mδ̈ + cδ̇ + kE δ = −αV + βF (12.4)

I = α δ̇ − C0 V̇ (12.5)
12 SSDI-Max Control and Its Applications in Renewable Energies 253

Fig. 12.5 Structure of the F

electromechanical model for u
a single degree of freedom
structure corresponding to
mono-modal with one M
piezoelectric element
Piezoelectric
element (C0)
C I
KE
V

with δ the nodal displacement vector, m, c and kE are the mass, the damping and the
stiffness matrices, respectively, when piezoelectric patches are in short circuit, α is
the electromechanical coupling matrix, V is the voltage vector of the i piezoelectric
patches, I is the electric current vector, and C0 is the diagonal capacitance matrix.
F is the force applied to the system.
By using the following variable change where φ is the mode shape matrix limited
to n modes and q the modal displacement vector:

δ=φq (12.6)

Equations (12.4) and (12.5) can be well represented by the projection in the modal
basis by:

M q̈ + C q̇ + KE q = −θ V + βF (12.7)
I = θ T q̇ − C0 V̇ (12.8)

θ is the modal electromechanical coupling matrix with a [n, i] matrix size. θ is

defined as follows:

θ = φT α (12.9)

M, C, KE are respectively the mass, damping and stiffness modal matrices. The
structure is assumed to be lightly damped, with proportional damping and with
decoupled modes. Equation (12.7) is normalized in order to have the modal mass
matrix equal to identity. Modal matrices can therefore be written as a function of ξ
the modal damping vector, ωE the short circuit frequency vector and ωD the open
circuit frequency vector as follows:

M = Id , C = 2(ξ )diag(ωD ) , KE = (ωE ) (12.10)

254 A. Chérif et al.

By separating the actuators and sensors voltages, named Va and Vs , respectively,

Eq. (12.7) becomes:

M q̈ + C q̇ + KE q = −θa Va − θs Vs + βF (12.11)

In an open circuit or when the sensor voltage is monitored with a voltage amplifier,
sensor intensity is null, therefore:

θsT q − C0s Vs = 0 (12.12)

By reintroducing Eq. (12.12) in Eq. (12.11):

# −1 T $
M q̈ + C q̇ + KE θs C0s θs q = −θa Va + βF (12.13)

Linear systems (12.13) and (12.12) can be written in modal state under form:
⎡ ⎤
*     q
ẋ = Ax + Bu q F
, x= , u= , y = ⎣ q̇ ⎦ (12.14)
y= Cx q̇ Va
Vs

with x being the state vector, u is the control vector, y is the output vector, A, B, C
are the state matrices:
⎡ ⎤
    Id 0
0 Id 0
A= , B= , C=⎣ 0 Id ⎦ (12.15)
A0 −M −1 C M −1 (β − θa ) −1 T
C0s θs 0
# −1 T $
with: A0 = −M −1 KE θs C0s θs and:

−1 T
Va = C0s θa q (12.16)

and C0a is the capacity of actuator and the matrix C0s is capacity sensors.

12.3.2 Smart-Structure Definition

The structure that will be used in the following simulations and analyses is a steel
clamped plate equipped with four P188 piezoelectric inserts. Its dimensions and
physical properties are given in Tables 12.1 and 12.2. This plate is clamped on all
four sides. Figure 12.6 illustrates this plate which is representative of a protective
panel. This structure has been identified according to the previously described
model. The measurement process and parameter identification is described in
Richard (2007).
12 SSDI-Max Control and Its Applications in Renewable Energies 255

Table 12.1 Characteristics of the plate (Richard 2007)

Length Width Thickness Young modulus Poisson ratio Density
0.6 m 0.4 m 1 mm 210 GPa 0.345 7500

Table 12.2 Characteristics Properties Symbol Value

of P ZT P189 piezoelectric
inserts (Richard 2007) Density ρ 7650 kg m3
sE11 10.66 × 10−12 Pa−1
Compliances sE12 −3.34 × 10−12 Pa−1
Short-circuit sE13 −4.52 × 10−12 Pa−1
sE33 13.25 × 10−12 Pa−1
Permittivity T
ε33 10.17 nF m−1
Piezoelectric coefficient d13 −108 pC N−1

Fig. 12.6 Protective panel

structure used in simulations.
The plate is clamped on the
four sides. Each four
piezoelectric elements are
12 × 4 cm and 600 μm thick

Table 12.3 Frequencies of the four simulated modes of the plate embedded (Richard 2007)
Modes Mode 1 Mode 2 Mode 3 Mode 4
Frequency 96 Hz 175 Hz 204 Hz 228 Hz

Table 12.3 summarizes the frequencies of the four modes considered in the
model. Due to the position of various piezoelectric patches, and according to
symmetry conditions, only symmetric modes are efficiently electromechanically
coupled. As a result, on the lowest part of the frequency spectrum, only a few modes
are significantly coupled, which means that they can be targeted effectively by the
proposed control if the patches are used as actuators, it also means that only these
modes can produce a significant response if the patches are used as sensors. These
modes are taken into account by the model used in simulations.
The resonance frequencies of the lower modes are given in Table 12.3.
256 A. Chérif et al.

12.4 SSDI Max Modal Control

The main disadvantage of Modal-SSDI control technique is the need for numerical
computation. The technique can not be self-powered. However, once implemented,
the numerical control loops offer different opportunities such as hybridization with
active techniques or alternative switching sequences improved in terms of voltage
inversion.
The technical implementation SSDI-Max is an alternate modal-SSDI control
which increases the growth of the piezoelectric actuator voltage Va . The damping
performance is strongly dependent on this voltage amplitude. The SSDI allows a
natural growth of this voltage using a cumulative effect (Richardet et al. 2009).
However, if this cumulative effect is affected by local maximum, the voltage
magnification is not optimal. It is the objective of the modal-SSDI-Max technique
to correct this point.
Modal-SSDI-Max strategy in comparison with Modal-SSDI consists in delaying
the switch instant to the next voltage extremum immediately following the targeted
modal coordinate extremum. The switch still allows the brief Va voltage inversion.
According to this strategy, the inversion is triggered from a higher voltage,
resulting in an amplification of the cumulative effect. However, it could imply
an important added phase shift resulting with a voltage only approximately syn-
chronized with the targeted modal displacement. This could result in a significant
damping depreciation. In order to limit this desynchronization, the delay is limited
to a given time window. This process is illustrated on Fig. 12.7 which schematically
illustrates the gain attainable in the voltage amplitude when modifying the switch
instant. The dashed line represents the targeted modal coordinate. Usual modal-
SSDI technique would trigger the switch at each extremum. The thick line represents
the open-circuited actuator voltage Va−oc . It is clear that due to the modal content
of the vibration, maximum of the voltage and maximum of the targeted modal
coordinate do not necessarily match. Note that in Fig. 12.7, Va−oc is represented
for sake of simplicity, but it is clear that it is the switched voltage Va that have
normally to be taken into account The grey area starting after each modal coordinate
extremum represents the imposed time window limitation. Finally, arrows figure the
effective switching time defined following the proposed strategy. The (A) switching
example illustrates the fact that immediate switching would result in an actuator
voltage with the wrong phase because the voltage is negative when the modal
coordinate is maximum positive. Waiting for the next positive maximum is therefore
necessary to ensure both the correct phase and the optimal amplitude. The (B)
switching example illustrates that even if the voltage sign is correct, waiting for the
next extremum allows to increase the voltage amplitude. If the voltage extremum
occurs immediately before the modal coordinate extremum as illustrated by (C),
it is appropriate to switch immediately. Example (D) illustrates the fact that if no
voltage extremum with the correct sign is located in the maximum time allowed, the
switch must occur at the end of the time window to limit the resulting voltage phase
shift. Finally, this procedure can be seen as a way to use the modal energy of the
other modes to increase the voltage devoted to the control of the targeted mode.
12 SSDI-Max Control and Its Applications in Renewable Energies 257

Fig. 12.7 Illustration of the SSDI-Max strategy: definition of the switch instant according to the
targeted modal coordinate, the piezoelectric actuator open-circuit voltage and the maximum delay
authorized

This methodology can be summarized by the following algorithm. When a

maximum of the modal displacement q̂i occurs the limitation time window starts.
Then the signs of the voltage Va and its derivative are considered during the
window:
1. If the voltage is positive and the derivative is negative, the switch trigger is
immediate.
2. If the voltage is positive and the derivative is positive, the system waits for the
next maximum voltage. This delay is nevertheless limited by the time window.
3. If the voltage is negative, the system waits for one of the above conditions.
4. If no switching occurred and the end of the time window is reached, the switching
is triggered.
This algorithm is anti-symmetric if a minimum modal displacement is reached.

12.5 Simulations

Simulations are performed using the Matlab/SimulinkTM software environment.

The inversion coefficient γ is set to 0.7 which is a realistic value (Richard et al.
2000). The first step in SSDI-Max validation is the control of a single mode of the
structure. This single mode control is considered with different type of multimodal
excitation: multi-sinusoidal excitation or broadband excitation with a pulse or white
noise. The second stage of validation protocol consists in the control of the four
modes simultaneously excited with a pulse or with a white noise. The effect of the
time window on SSDI-Max performance is established in the later section.
258 A. Chérif et al.

× 10-8
5
No control
0

-5
0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9
Modal coordinate of mode 1 (m)

Time (s)

× 10-8
5
SSDI

-5
0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9
Time (s)

× 10-8
5
SSDI-Max
0

-5
0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9
Time (s)

Fig. 12.8 Simulation of the mode 1 modal coordinate with a pulse excitation in the case of mode
1 control

12.5.1 Pulse Excitation

The excitation is a wide frequency square force pulse 50 μs long, and with a
normalized amplitude. This signal is applied on the structure space-state block
Force input shown Fig. 12.4. Figure 12.8 illustrates simulation results in the time
domain in the case of mode 1 targeting. A clear strong improvement of damping
using SSDI max is clearly visible. The voltage build-up improvement is illustrated
in Fig. 12.9 showing that small shifts in the switch instants result in amplification
of the self-generated control voltage. This result is remarkable since the reference
which corresponds to the structure controlled using modal SSDI is already well
damped.

12.5.2 White Noise Excitation

In this case the plate model is excited with a white noise of normalized amplitude
applied on the structure space-state block Force input shown in Fig. 12.3. Like a
pulse this type of broadband excitation results in vibration energy shared on all the
four considered modes of the model.
12 SSDI-Max Control and Its Applications in Renewable Energies 259

0.06
SSDI
SSDI-Max

Actuators voltage (V) 0.04

0.02

-0.02

-0.04

-0.06
0.48 0.5 0.52 0.54 0.56 0.58 0.6 0.62 0.64
Time (s)

Fig. 12.9 Simulation of the actuator voltage with a pulse excitation in the case of mode 1 control

500
Actuator voltage (V)

SSDI
SSDI-Max
-500
0.26 0.27 0.28 0.29 0.3 0.31 0.32 0.33 0.34 0.35
Time (s)
× 10 -4
2
Modal displacement
of mode 1 (m)

-1

-2
0.26 0.27 0.28 0.29 0.3 0.31 0.32 0.33 0.34 0.35
Time (s)

Fig. 12.10 Modal displacement and the voltage across the actuator when the mode 1 is targeted
by the control

Figure 12.10 shows a comparison of the modal SSDI and modal SSDI-Max
damping performances. It shows a partial time window representative of the 3 s
long simulation time. The lower plot presents the first modal coordinate which is
targeted by the control. The upper illustrates the control voltage actuator. Again it
is clearly visible that a slight shift in the switching time definition leads to a much
improved actuator voltage amplitude, thus resulting in more efficient damping.
260 A. Chérif et al.

a No control
400
SSDI
Actuator voltage (V)

SSDI-Max
200

-200

-400
2.02 2.03 2.04 2.05 2.06 2.07 2.08
Time (s)
b × 10-4
1
Modal displacement (m)

0.5

-0.5

-1
2.02 2.03 2.04 2.05 2.06 2.07 2.08
Time (s)

Fig. 12.11 Modal displacement (b) and the piezoelectric actuator voltage (a) for single mode
control (mode 1) and in the case of multi-sine excitation. The ratio of the controlled mode energy
to the structure vibrational energy is E1/E2 = −3.8 dB

12.5.3 Sinusoidal Excitation

In this case the excitation is composed of the sum of four sine signals whose
frequencies are those of the four modes of the structure. The interest of this type
of excitation relies in the fact that it is possible to adjust the energy associated with
the various modal contents of the vibration.
Figure 12.11a illustrates the increase in self-generated control power: A slight
variation of the moment of inversion (time window is 1/10 of the first mode period)
can lead to a significant change in the control voltage responsible for mitigation.
It should be noted that this change led to a much greater reduction of the targeted
mode (Fig. 12.11b).
12 SSDI-Max Control and Its Applications in Renewable Energies 261

12.5.4 Influence of the Time Window on the Performance of

SSDI Max

According to the SSDI-Max strategy the time window, which is used to limit the
possible time shift prior to switching, is a very important and critic parameter. If it
is too small, the voltage will not have the possibility to increase and no significant
enhancement will be observed. If it is too long, there is a risk of desynchronization of
the actuator voltage with the targeted modal speed, thus resulting altered damping.
In order to define an optimal time window, simulations were made while varying
the value of this window from zero (pure modal SSDI) to 2/5th of the targeted mode
period. This analyze was conducted for single mode control (mode 1). Figure 12.12
illustrates the variation of the targeted mode damping as a function of the time
window for pulse excitation.
As shown on Fig. 12.12, a time window lying between 1/10th and 1/5th of the
period is nearly optimal for SSDI-Max control, ensuring an increase of the actuator
voltage while not inducing a too important phase shift between the voltage and the
targeted modal speed, thus resulting in additional damping.

12.6 Conclusions

It has been shown that performances of the modal SSDI control can be significantly
improved using an algorithm that exploits the complexity of the shape of the
piezoelectric voltage in the case of multimodal vibration. Instead of inversing the
voltage on the extremum of the modal displacement, the proposed technique, named
modal SSDI-Max, consists of waiting for the next voltage extremum immediately
after an extremum of the targeted modal coordinate to inverse the voltage across the
piezoelectric element used as an actuator. Thus, the inversion is done from a higher
voltage, resulting in improved voltage build-up. The resulting quasi square control
voltage is still approximately synchronized with the targeted modal displacement
mode and nearly in phase with the targeted modal speed which implies a significant
damping improvement.
Validation of this concept has been done using numeric simulation in the case
study of a clamped steel plate. The entire control scheme has been simulated and
evaluated. Modal SSDI-Max Damping simulation results showed nearly improved
damping performances compared to modal SSDI for the control of a single mode of
the structure, in the case of sinusoidal excitation on the four modes of the plate, pulse
or white noise excitation. Remarkable gains in attenuation lying between 6 and 11 db
were obtained with actuators presenting a rather low coupling coefficient (lower than
9%). Finally, the influence of the maximum time delay between the targeted modal
coordinate extremum and the switch instant has been evaluated. Simulations show
that a maximum delay lying between 1/10th and 1/5th of the targeted mode period
is nearly optimal for considered cases.
 a
-7
1/2 of 1/1.5 of
SSDI 1/2.5 of
-8 period period
period

-9 1/30 of period
Mode 1 damping (dB)

-10 1/20 of period

-11
1/3 of
period
-12 1/10 of
period
-13 1/5 of period

-14

-15
0 1 2 3 4 5 6 7
The time windows (s) × 10-3

b × 10-8
2
SSDI
1/10
1.5 1/5
Modal displacement of mode 1 (m)

0.5

-0.5

-1

-1.5
0.54 0.55 0.56 0.57 0.58 0.59 0.6
Time (s)

Fig. 12.12 Influence of the time window for pulse excitation. (a) Influence of the time window on
mode 1 damping. (b) Corresponding time domain results
12 SSDI-Max Control and Its Applications in Renewable Energies 263

References

Carriere, M. S. (2010). Crossed synthesis of controllers and observers for robust control of
synchronous machine. Ph.D. Thesis, University of Toulouse, France.
Chérif, A., Richard, C., Guyomar, D., Belkhiat, S., & Meddad, M. (2012). Simulation of
multimodal vibration damping of a plate structure using a modal SSDI-Max technique. Journal
of Intelligent Material Systems and Structures, 23(6), 675–689.
Clark, W. W. (2000). Vibration control with state-switching piezoelectric materials. Journal of
Intelligent Material and Structures, 11, 263–271.
Cunefare, K. A., & Holdhusen, M. (2004). Optimization of a state-switched absorber applied to
a continuous vibrating system. In SPIE SSM Conference Passive Damping and Isolation, San
Diego.
Davis, C. L., & Lesieutre, G. A. (2000). Actively tuned solid-state vibration absorber using
capacitive shunting of piezoelectric stiffness. Journal of Sound and Vibration, 232, 601–617.
Eielsen, A. A., & Fleming, A. J. (2010). Passive shunt damping of a piezoelectric stack
nanopositioner. In American Control Conference, Baltimore (pp. 4963–4968).
Forward, R. L. (1979). Electronic damping of vibrations in optical structures. Applied Optics,
18(5), 690–697.
Franklin, G. F., Powell, J. D., & Workman, M. L. (1990). Digital control of dynamic systems (2nd
ed.). Reading: Addison-Wesley.
Fuller, C. R., & Jones, J. D. (1987). Experiments on reduction of propeller induced interior noise
by active control of cylinder vibration. Journal of Sound and Vibration, 112(2), 398–395.
Guyomar, D., & Badel, A. (2006). Nonlinear semi-passive multimodal vibration damping: An
efficient probabilistic approach. Journal of Sound and Vibration 294(1–2), 249–268.
Hagood, N. W., & Von Flotow, A. (1991). Damping of structural vibrations with piezoelectric
materials and passive electrical networks. Journal of Sound and Vibration, 146(2), 243–268.
Harari, S., Richard, C., & Gaudiller, L. (2009). New semi-active multi-modal vibration control
using piezoceramic components. Journal of Intelligent Material Systems and Structures, 20,
1603–1613.
Meyer, Y., & Collet, M. (2011). Active vibration isolation of electronic components by piezocom-
posite clamped-clamped beam. Mechanical Systems and Signal Processing, 25, 1687–1701.
Niederberger, D., Morari, M., & Pietrzko, S. (2004). A new control approach for switching shunt
damping. In SPIE SSM Conference Passive Damping and Isolation, San Diego.
Petit, L., Lefeuvre, E., Richard, C., & Guyomar, D. (2004). A broadband semi-passive piezoelectric
technique for structural damping. In SPIE SSM Conference, Damping and Isolation, San Diego.
Preumont, A., François, A., Bossens, A., & Abu-Hanieh, A. (2002). Force feedback versus
acceleration feedback in active vibration isolation. Journal of Sound and Vibration, 257(4),
605–613.
Richard, T. (2007). Diminution du coefficient de transmission acoustique d’une paroi à l’aide
d’amortisseurs piézoélectriques semi-passifs. Thèse de doctorat, Institut National des Sciences
Appliquées de Lyon, no 2007ISAL0112.
Richard, C., Guyomar, D., Audigier, D., & Bassaler, H. (2000). Enhanced semi passive damping
using continuous switching of a piezoelectric device on an inductor. In SPIE Smart Structures
and Materials Conference, Passive Damping and Isolation, Newport Beach (Vol. 3689,
pp. 288–299).
Richard, C., Harari, S., & Gaudiller, L. (2009). Enhanced piezoelectric voltage build-up for semi-
active control of smart structures. In SPIE Smart Structures and Materials, San Diego.
Snyder, S. D., & Hansen, C. H. (1991). Mechanisms of active noise control by vibration sources.
Journal of Sound and Vibration, 147(3), 519–525.
Wu, S., & Bicos, A. S. (1997). Structural vibration damping experiments using improved
piezoelectric shunts. Proceedings of the SPIE, 3045, 40–50.
Chapter 13
Feedback T-S Fuzzy Controller in Finite
Frequency for Wind Turbine

Youssef Berrada, Abderrahim El-Amrani, and Ismail Boumhidi

Abstract This chapter investigates a feedback finite frequency Takagi-Sugeno (T-

S) fuzzy controller synthesis for a variable speed wind turbine. The proposed
control design is based on both the T-S fuzzy modeling and the finite frequency
approach. The T-S fuzzy model is used to deal with a nonlinear behavior of
wind turbine system, and the finite frequency approach allows the command in a
specific domain of frequency. The control constraints are given in terms of a set
of LMIs which can be efficiently solved using existing numerical tools. In order to
illustrate the performance of the proposed control algorithm, numerical simulations
are performed using Matlab software.

Keywords Finite frequency · Takagi-Sugeno fuzzy · Variable wind speed ·

Wind turbine

Notations

• The superscript T stands for matrix transposition.

• I denotes an identity matrix with appropriate dimension.
• “diag” stands for block diagonal matrix.
• “tr(A)” denotes the trace of matrix A.
• A > 0 (resp. A < 0) mean that matrix A is positive definite (resp. negative
definite).

Y. Berrada () · A. El-Amrani · I. Boumhidi

LESSI Laboratory, Sciences Faculty, Department of Physics, University of Sidi Mohammed Ben
Abdellah, Fez, Morocco
e-mail: [email protected]; [email protected]; ismail.
[email protected]

© Springer Nature Singapore Pte Ltd. 2019 265

N. Derbel, Q. Zhu (eds.), Modeling, Identification and Control Methods
in Renewable Energy Systems, Green Energy and Technology,
https://doi.org/10.1007/978-981-13-1945-7_13
266 Y. Berrada et al.

13.1 Introduction

Global warming effect and fossil fuel pollution has caused great damage in the
earth’s environment, and has become a serious problem for people. As renewable
energies sources are environment-friendly and sustainable, wind and solar energy
have attracted increased attention during the last years. In light of this, improving
the efficiency of wind turbine becomes an important research topic. Furthermore,
control techniques have a major effect on wind energy conversion systems, and
remain a key factor in maximizing the extracted energy from the wind and reducing
the stresses caused by aerodynamic loads. To achieve satisfactory wind turbine
performances, Takagi-Sugeno (T-S) fuzzy has received increasing attention during
the past decades from many researchers. There are a lot of research results on
the T-S fuzzy in the literature, such that, in Lasheen et al. (2016) the authors
proposed a new algorithm of fuzzy predictive for collective pitch control of large
wind turbines. A Neuro-fuzzy inertia controller has been addressed in Hafiz et al.
(2016) to control parameter selection which ensures the optimal use of available
Kinetic Energy reserve. Neural network is also coupled in Medjber et al. (2016)
with fuzzy logic controller to monitor maximum power for wind energy conversion
system. A new kind of T-S fuzzy control technique used for capturing maximum
wind energy under multi-operating condition has been discussed in An et al. (2015).
A data driven design methodology able to generate a Takagi-Sugeno fuzzy model for
maximum energy extraction from variable speed wind turbines has been examined
in Galdi et al. (2008). A sensorless wind energy conversion system maximum
wind power point tracking using Takagi-Sugeno fuzzy cerebellar model articulation
control to achieve maximum power transfer under various wind speeds without
actual measurement of the wind velocity has been proofed in Liu et al. (2015).
Note that all those control techniques and other given in this area can improve the
wind turbine robustness against the random wind speed and maximize the extracted
wind power. However, when the external disturbance belong to a certain frequency
range which is known beforehand, it is not favorable to control the system in the full
frequency domain, because this may introduce some conservatism and poor system
performance. Recently, the control synthesis in a finite frequency domain has been
addressed, and there have appeared many results in this domain (Berrada et al. 2017;
Chen et al. 2010; El-Amrani et al. 2016; Li et al. 2015; Zhang et al. 2014).
In light of the above, we propose feedback Finite Frequency Takagi-Sugeno
Fuzzy (FFTSF) to control wind turbine under various wind conditions. We first
represent the wind turbine, which is a two mass model, as a highly nonlinear
dynamical model. To carry out the FFTSF design, we then rewrite the wind
turbine model as a T-S fuzzy representation. Next the proposed feedback control is
established by the finite frequency approach to command the wind turbine system in
a specific band of frequency. Based on the generalized Kalman-Yakubovich-Popov
(GKYP) lemma (Iwasaki et al. 2005), the controller constraints are given in terms
of linear matrix inequalities (LMIs) which can be efficiently solved numerically.
The control technique acts on generator in order to apply the electromagnetic
13 Feedback T-S Fuzzy Controller in Finite Frequency for Wind Turbine 267

torque reference and on the pitch actuator in order to control the pitch angle of
the blades according to wind speed value, calculated from the measurements of the
rotational speed of the shaft at the generator side, and of the speed of the wind by
an accelerometer located at the top of the tower.
The paper is organized as follows. In Sect. 13.2, we will describe the model of the
wind turbine system. In the next section, we will rewrite the obtained model as a T-
S Fuzzy representation, and the controller will be designed by the Finite frequency
technique. In Sect. 13.4, the performances of the proposed control strategy will be
shown carried out by simulation results. Finally, some conclusions are given in
Sect. 13.5.

13.2 Wind Turbine Model

The wind turbine is established by combining a model of a mechanical structure

represent the drive trains and nonlinear model representing the blades aerodynamic
properties. The mathematics model of the wind turbine is clearly described in
Bououden et al. (2012), which is represented by the following state representation:
⎡ ⎤ ⎡ ⎤
0 1 −1 0 0
⎢ Ks Bs Bs ⎥ ⎢ 1 ⎥ 
⎢− − ⎥ ⎢ 0 ⎥
⎢
ẋ0 = ⎢ Jr ⎥ ⎢
Jr Jr ⎥ x0 + ⎢ Jr ⎥ Ta (13.1)
⎥
⎣ Ks Bs Bs ⎦ ⎣ 1 ⎦ Tg
− − 0
Jg Jg Jg Jg

where x0 = [θs , ωr , ωg ]T , θs , ωr , ωg , Bs and Ks are the torsion angle, the rotor

speed, the generator speed, the damping of the transmission and the stiffness of the
transmission, respectively. Jr and Jg are the inertia of the rotor and the generator,
respectively. Tg is the generator torque which is a nonlinear function depends to the
generator speed and the zero-torque speed ωz .

Tg = Bs (ωg − ωz ) (13.2)

Ta is the aerodynamic torque which is expressed as:

1
Ta = π ρ R 3 Cq (λ, β) v 2 (13.3)
2
1
with Cq (λ, β) = Cp (λ, β) is the torque coefficient. Cp (λ, β) is the power
λ
coefficient that is a nonlinear function of the pitch angle β and the reduced speed λ.
So, Ta is a nonlinear function depends to wind speed, rotor speed and pitch angle,
can be linearized by the following expression:

Ta = Tav v + Taβ β + Taλ λ (13.4)

268 Y. Berrada et al.

where Tav , Taβ and Taλ are individual partial derivatives of the aerodynamic torque
Tav for rotor speed, wind speed and pitch angle at the operating point, respectively.
The actuator describes the dynamic behavior between the desired pitch βd and
the actuation of this desired pitch β is modeled as:

1
β̇ = (βd − β) (13.5)
τ
where τ is a time constant.
Finally, according to the Eqs. (13.1), (13.2), (13.3), (13.4), and (13.5), and
replacing Ta and Tg with their approximated expressions, the dynamic model of
the wind turbine can be represented as:

ẋ = Ax + Bu + Ev
y = Cx (13.6)

where x = [θs , ωr , ωg , β]T , u = [βd , ωz ]T , y = ωg and matrices A, B and E are

given by:
⎡ ⎤ ⎡ ⎤
0 1 −1 0 0 0 ⎡ ⎤
⎢ Ks Bs Bs Taβ ⎥ ⎢0 0 ⎥ 0
⎢− − ⎥ ⎢ ⎥ ⎢ Tav ⎥
⎢ Jr Jr Jr Jr ⎥ ⎢ ⎥ ⎢ ⎥
⎢ ⎥ Bg ⎥
A = ⎢ Ks Bs + Bg Bs ⎥, B = ⎢
⎢0
⎢
⎥ , E = ⎢ Jr
⎥
⎥
⎢− − 0 ⎥ ⎢ J ⎥ ⎣ 0 ⎦
⎢ Jg Jg Jg ⎥ ⎣1
g
⎦
⎣ 1 ⎦
0 0
0 0 0 − τ
τ
(13.7)
Let ξ = x − xref so x = xref + ξ and then ẋ = ẋref + ξ̇ .
The system dynamic model (13.6) becomes:

ξ̇ = Aξ + Bu + Dw
χ = Cξ (13.8)

where w = [xref , ẋref , v]T and D = [A, −I, E] with xref is the reference signal.

13.3 Finite Frequency T-S Fuzzy Control

13.3.1 T-S Fuzzy Representation

The best possible performance from the highly nonlinear system (13.6) can be
obtained using T-S fuzzy model. The wind turbine variables are assumed varying in
the operating range v1 ≤ v ≤ v2 , β1 ≤ β ≤ β2 . Consequently, The T-S fuzzy model
13 Feedback T-S Fuzzy Controller in Finite Frequency for Wind Turbine 269

of system (13.6) is established by the following rules IF-THEN for i = 1, . . . , 4 and

l, k = 1, 2:
*
ξ̇ = Ai ξ + Bi u + Di w
If(βisMl )And(visNk )Then (13.9)
χ = Ci ξ

Membership functions are given by:

h1 = M1 (β)N1 (v), h2 = M1 (β)N2 (v), h3 = M2 (β)N1 (v), h4 = M2 (β)N2 (v)

with:
β − β1 β2 − β v − v1 v2 − v
M1 (β) = , M2 (β) = , N1 (v) = , N2 (v) =
β2 − β1 β2 − β1 v2 − v1 v2 − v1


4
The fuzzy basis functions satisfies: hi ≥ 0 and hi = 1.
i=1
The fuzzy system can be written as following form:

ξ̇ = A(h)ξ + B(h)u + D(h)w

χ = C(h)ξ (13.10)

where:


4 
4 
4 
4
A(h) = hi Ai , B(h) = hi Bi , D(h) = hi Di , C(h) = h i Ci
i=1 i=1 i=1 i=1

The fuzzy state feedback controller can be designed as:


4
u= hi Ki ξ = K(h)ξ (13.11)
i=1

Combining (13.10) and (13.11) together, we can get the following closed-loop fuzzy
system:

ξ̇ = Ac (h)ξ + D(h)v
χ = C(h)ξ (13.12)

where: Ac (h) = A(h) + B(h)K(h).

The transfer function from the input ξ to the output χ is given by:

G(j ω) = c(h)[j ωI − A(h)]−1 D(h) (13.13)

270 Y. Berrada et al.

13.3.1.1 Problem Description

The objective is to design a state feedback controller in (13.11) for system (13.12)
such that two conditions are satisfied:
1. Closed-loop system (13.12) is asymptotically stable.
2. The following finite frequency index holds:
 
χ (ω)T χ (ω)dω ≤ w(ω)T w(ω)dω (13.14)
ω1 ≤ω≤ω2 ω1 ≤ω≤ω2

where ω1 and ω2 are known scalars.

13.3.2 Finite Frequency

We start this section by introducing some basic lemmas, which we will be used in
the proof of our results.
Lemma 1 Let be a given scalar. For the system (13.12) is asymptotically stable, and
the FF H∞ (13.14) is satisfied if there exists Hermitian matrices, such P = P T that:
 T   
Ac (h) D(h) −Q(h) P (h) + j ωc Q(h) P (h) − j ωc Q(h) −ω1 ω2 Q(h)
I 0 I 0 I 0
 
C(h)T C(h) 0
+ < 0 (13.15)
0 −γ 2 I

where ωc = 12 (ω1 + ω2 ).
Proof First, suppose (13.15) holds. Post multiplying by [ξ T w T ] from the left and
by its conjugate transpose from the right, we have:

2ξ̇ T P ξ − ξ̇ T P ξ̇ + j ωc ξ̇ Qξ − j ωc ξ T Qξ̇ − ω1 ω2 ξ T Qξ + χ T χ − γ 2 w T w ≤ 0
(13.16)
Note that for any vectors φ and ϕ, the equality φ T Qϕ = tr(ϕφ T Q) holds.
Then (13.16) can be rewritten as

d T ' # $# $T (
(ξ P ξ ) + χ T χ − γ 2 w T w ≤ tr He ω1 ξ + j ξ̇ ω2 ξ + j ξ̇ Q (13.17)
dt
Taking the integrating from t = 0 to ∞ using the stability property, we have
 ∞  ∞
ξ(∞) P ξ(∞) +
T
χ χ dt − γ
T 2
w T wdt ≤ tr[He (S)Q] (13.18)
0 0
13 Feedback T-S Fuzzy Controller in Finite Frequency for Wind Turbine 271

where:
 ∞# $# $T
S= ω1 ξ + j ξ̇ ω2 ξ + j ξ̇ dt (13.19)
0

Note that ξ(∞)T P ξ(∞) ≥ 0 for P > 0, then we have:

 ∞  ∞
χ χ dt ≤ γ
T 2
w T wdt + tr[He (S)Q] (13.20)
0 0

By the Parseval’s theorem (Goodwin et al. 2001; Skenton et al. 1998), we have:
 ∞
1
S= (ω1 − ω)(ω2 − ω)X(ω)X(ω)T dt (13.21)
2π −∞
 ∞  ∞
1
χ χ dt =
T
Y (ω)T Y (ω)dw (13.22)
0 2π −∞
 ∞  ∞
1
w T wdt = W (ω)T W (ω)dw (13.23)
0 2π −∞

And hence S is Hermitian and the bound on the right-hand side of (13.20) becomes,
and hence (13.20) is equivalent to:
 ∞  ∞
χ T χ dω − γ 2 w T wdω ≤ 2π tr(SQ) (13.24)
0 0

Note that XT QX ≥ 0 for Q > 0, and (ω1 − ω)(ω2 − ω) ≤ 0 for ω1 ≤ ω ≤ ω2 .

Then we have

2π tr(SQ) = (ω1 − ω)(ω2 − ω) tr(XXT Q)dω
ω1 ≤ω≤ω2

= (ω1 − ω)(ω2 − ω)XT QXdω ≤ 0 (13.25)
ω1 ≤ω≤ω2

From (13.24) and (13.25), we have statement (13.14), and hence the finite
frequency performance is satisfied.
Remark 1 If all the matrices in Lemma 1 are independent on h, then the fuzzy
system becomes a linear system, and Lemma 1 is reduced to the generalized KYP
lemma (Iwasaki et al. 2005), which proved to be an effective tool to deal with the
Finite Frequency problem of linear time-invariant systems.
Lemma 2 (Projection Lemma Apkarian et al. 2001) Given a symmetric matrix
Φ and two matrices Γ , Π of column dimension m, there exists a symmetric matrix
F such that the following LMI holds

Φ + Γ F Π + ΠT F T Γ T < 0 (13.26)
272 Y. Berrada et al.

If and only if the following projection inequalities with respect to F are satisfied:

Γ ⊥ ΦΓ ⊥T < 0 (13.27)
⊥ ⊥T
Π ΦΠ <0 (13.28)

Now, an important theorem which can guarantee the asymptotical stability and the
FF H∞ performance of the system in (13.12) is going to be proposed.
Theorem 1 For a given constant γ > 0, consider the closed-loop system (13.12),
if there exist symmetric matrices Q(h) > 0, X > 0, Z(h) and general matrix Y (h)
such that the following linear matrices inequality holds
⎡ ⎤
−Q(h) −X + Z(h) + j ωc Q(h) 0 0
⎢  Υ D(h) XC(h)T ⎥
Ψ =⎢
⎣ 
⎥ < 0 (13.29)
⎦
 −γ 2 I 0
   −I
S = A(h)X + XA(h)T + B(h)Y (h) + Y (h)T B(h)T < 0 (13.30)

with: Υ = A(h)X + XA(h)T − B(h)Y (h) − Y (h)T B(h)T − ω1 ω2 Q(h).

The controller gains Ki are given by:

K(h) = Y (h)X−1 (13.31)

Proof Using the Lemma 1, and according to the close loop system (13.12). The
inequality (13.15) can be rewritten as:
⎡ ⎤T ⎡ ⎤
A(h) D(h) A(h) D(h)
⎣ I 0 ⎦ Φ ⎣ I 0 ⎦<0 (13.32)
0 I 0 I

where:
⎡ ⎤
−Q(h) P (h) + j ωc Q(h) 0
Φ = ⎣ P (h) − j ωc Q(h) ω1 ω2 Q(h) + C(h)T C(h) 0 ⎦ (13.33)
0 −γ 2 I

On the other hand, we can obtain

⎡ ⎤T ⎡ ⎤
I 0 I 0  
⎣ 0 0 ⎦ Φ ⎣ 0 0 ⎦ = −Q(h) 0 <0 (13.34)
0 −γ 2 I
0I 0I
13 Feedback T-S Fuzzy Controller in Finite Frequency for Wind Turbine 273

According to the projection Lemma 2, with:

 
A(h)T I 0
Γ⊥ = (13.35)
D(h)T 0 I
 
I 00
Π⊥ = <0 (13.36)
00I

The following inequality is a sufficient condition for (13.15).

⎡ ⎤ ⎡ ⎤T
−I ! " ! " −I
Φ + ⎣ A(h)T ⎦ F 0 I 0 + 0 I 0 F ⎣ A(h)T ⎦ < 0
T
(13.37)
D(h)T D(h)T

By substituting Ac (h) = A(h) + B(h)X, we obtain:

⎡ ⎤
−Q(h) −F + P (h) + j ωc Q(h) 0
⎣  Λ F D(h) ⎦ < 0 (13.38)
  −γ 2 I

with: Λ = A(h)T F + F A(h) + F B(h)K(h) + K(h)T B(h)T F − ω1 ω2 Q(h) +

C(h)T C(h).
Multiplying both sides of (13.36) by the full rank matrix diag(F −1 F −1 I ) and its
transpose from the left and right, and defining new variables Q(h) = F −1 Q(h)F −1 ,
Z(h) = F −1 P (h)F −1 , Y (h) = K(h)F −1 then (13.36) is rewritten as follows:
⎡ ⎤
−Q(h) −X + Z(h) + j ωc Q(h) 0
⎣  Λ D(h) ⎦ < 0 (13.39)
  −γ 2 I

with: Λ = A(h)X+CA(h)T +B(h)Y (h)+Y (h)T B(h)T F −ω1 ω2 Q(h)+XC(h)T +

C(h)X.
Applying the Schur complement to inequality (13.39), we obtain exactly the
inequality (13.29).
For the second sufficient condition of Theorem 1, let us construct a Lyapunov
function inequality, A(h) is stable if and only if there exists F = F T such that:

Ac (h)T F + F Ac (h) < 0 (13.40)

By substituting, Ac (h) = A(h) + B(h)K(h), we get

A(h)T F + F A(h) + F B(h)K(h) + K(h)T B(h)T F < 0 (13.41)

274 Y. Berrada et al.

Multiplying both sides of (13.41) by the matrix F −1 and its transpose from the left
and right, then we can get

F −1 A(h)T + A(h)F −1 + B(h)K(h)F −1 + F −1 K(h)T B(h)T < 0 (13.42)

Let X = F −1 and Y (h) = K(h)F −1 , then (13.40) became exactly the inequal-
ity (13.28).
Theorem 2 Given a positive scalar γ > 0, there exists a T-S fuzzy control
law (13.11) which makes the H∞ norm of the T-S fuzzy system (13.12) less than
γ in the frequency domain, if there exist matrices Qi > 0, X > 0, Zi and general
matrix Yi , such that the following LMIs hold for all i < j = 1, . . . , 4.

Ψii < 0 (13.43)

Ψij + Ψj i < 0 (13.44)
Sii < 0 (13.45)
Sij + Sj i < 0 (13.46)

where:
⎡ ⎤
−Qi −X + Zi + j ωc Qi 0 0
⎢  A X + XAT − B Y − Y T B T − ω ω Q Di XCiT ⎥
⎢ i i j 1 2 i ⎥
Ψij = ⎢ i j i
⎥(13.47)
⎣   −γ I
2 0 ⎦
   −I

Proof Taking the following summations:


4 
4 
4 
4
hi hj Ψij ; hi hj Sij (13.48)
i=1 j =1 i=1 j =1

Using Theorem 1, the proof is completed.

13.4 Simulation Results

The wind turbine model (13.6) with the numerical values listed in Table 13.1 is
considered under the variable wind speed 12 ≤ v ≤ 35 m/s. The Rotor speed
is maintained around the nominal speed value in the high speed region, and the
operated range of pitch angle is −2 ≤ β ≤ 24◦ . Matrices of the T-S fuzzy model are:
13 Feedback T-S Fuzzy Controller in Finite Frequency for Wind Turbine 275

Table 13.1 Wind turbine parameters

Parameter Ks Bs Bg Jg Jr ρ R τ
Value 1.566×106 3029.5 15.993 5.9 83×104 1.225 30 500
Unit N/m Nms/rad Nms/rad kg m2 kg m2 kg/m3 m μs

⎡ ⎤
0 1 −1 0
⎢ Ks Bs Bs Taβ1 ⎥
⎢− − ⎥
⎢ Jr Jr Jr Jr ⎥
⎢ Bs + Bg ⎥
A1 = A2 = ⎢ Ks Bs ⎥ (13.49)
⎢− − 0 ⎥
⎢ Jg Jg Jg ⎥
⎣ 1 ⎦
0 0 0 −
τ
⎡ ⎤
0 1 −1 0
⎢ Ks Bs Bs Taβ3 ⎥
⎢− − ⎥
⎢ Jr Jr Jr Jr ⎥
⎢ Bs + Bg ⎥
A3 = A4 = ⎢ Ks Bs ⎥ (13.50)
⎢− − 0 ⎥
⎢ Jg Jg Jg ⎥
⎣ 1 ⎦
0 0 0 −
τ

⎡ ⎤
0 0
⎢0 0 ⎥
⎢ ⎥
⎢ Bg ⎥
⎢
B1 = B 2 = B 3 = B 4 = ⎢ 0 ⎥ (13.51)
⎥
⎢ Jg ⎥
⎣1 ⎦
0
τ

⎡ ⎤ ⎡ ⎤
0 0
⎢ Tav1 ⎥ ⎢ Tav2 ⎥
⎢ ⎥ ⎢ ⎥
D1 = ⎢
⎢ Jr
⎥ , D1 = ⎢ Jr
⎥ ⎢
⎥
⎥ (13.52)
⎣ 0 ⎦ ⎣ 0 ⎦
0 0

where Taβ1 = Taβ (β = β1 ), Taβ2 = Taβ (β = β2 ), Tav1 = Tav (v = v1 ) and

Tav1 = Tav (v = v2 ).
The simulation is carried out under the following operating conditions:
• Wind speed profile 17 ≤ v ≤ 31 m/s (see Fig. 13.1)
• The operating frequency domain represents the rotation frequency maximum
and minimum of the aerodynamics rotor (3–42 rpm) which is equivalent to
276 Y. Berrada et al.

Fig. 13.1 Wind speed input profile

[0.05, 0.7] Hz, and for the generator (1500–1550 rpm) which is equivalent to
[25, 25.83] Hz.
By solving the linear matrix inequalities (13.43), (13.44), (13.45), and (13.46) for
i < j = 1, . . . , 4 with the optimized variable γ > 0 and the ranges of frequency:
• [ω1 = 0.05 Hz, ω2 = 0.7 Hz],
• [ω1 = 25 Hz, ω2 = 25.83 Hz],
for the regulators of pitch angle and zero-torque speed, respectively, we can obtain
the control gains:

K1 = 103 [1.049 3.484 − 0.953 − 0.052 − 97.814 0.160 − 4.743 − 0.005]

K2 = 103 [1.064 3.531 − 1.042 − 0.053 − 97.874 − 0.040 − 2.363 − 0.002]

K3 = 103 [0.585 2.029 − 0.731 − 0.030 − 97.877 − 0.050 − 2.089 − 0.002]

K4 = 103 [0.621 2.139 − 0.595 − 0.032 − 98.022 − 0.527 − 2.721 0.005]

The performance of the proposed Feedback T-S fuzzy controller in finite

frequency (FFTSF) is illustrated using a comparison with the predictive controller
(MPC) strategy [16] carried out by Bououden et al. (2012).
From the simulation results, Fig. 13.2 represents the time response of the
generator speed equipped with controller FFTSF and controller MPC in dashed and
13 Feedback T-S Fuzzy Controller in Finite Frequency for Wind Turbine 277

Fig. 13.2 Response time of the generator speed

dotted lines respectively, the solid line represents the optimal signal of the generator
speed. The time response of controller laws: Pitch angle βd and Zero-torque speed
Ω are shown in Figs. 13.3 and 13.4, respectively. Figure 13.5 represents the variation
of the membership functions.
According to these results, the generator speed with the proposed control strategy
converges faster without any oscillatory behavior to its optimal value (Fig. 13.2)
compared to this of the MPC controller which has some oscillators at the beginning
of convergence. We can also observe that the proposed control laws, the pitch
angle βd (Fig. 13.3) and the Zero-torque Ω (Fig. 13.4), yields the best performance
in terms of stability and convergence. Moreover, the applied effort at the regime
transient is too small despite the presence of strong variations in wind speed.

13.5 Conclusions

This chapter proposes a Feedback T-S fuzzy controller in finite frequency synthesis
for a variable speed wind turbine, which is designed using GKYP lemma extended
to T-S fuzzy model. The T-S fuzzy is used to deal with the highly nonlinear term
of aerodynamics torque, and the finite frequency in order to design the feedback
control. The main objective of the developed control strategy is to improve the
robustness of the system in a certain finite frequency range where the system
operates or the external disturbance has existed. The effectiveness of the control
Fig. 13.3 Desired pitch angle control

Fig. 13.4 Zero-torque speed control

13 Feedback T-S Fuzzy Controller in Finite Frequency for Wind Turbine 279

Fig. 13.5 Membership functions

strategy compared with a predictive controller has been realized, and simulation
results have shown an interesting performance of the proposed control strategy in
terms of stability and convergence speed, allowing a better regulation of the power
generated by the wind turbine.

References

An, A., Yang, G., Zhang, H., & Du, P. (2015). PMSG maximum wind power tracking control
based on TS fuzzy method. In IEEE 27th Chinese Control and Decision Conference (CCDC),
Qingdao (pp. 2844–2849).
Apkarian, P., Tuan, H. D., & Bernussou, J. (2001). Continuous-time analysis, eigenstructure
assignment, and H/sub 2/synthesis with enhanced linear matrix inequalities (LMI) characteri-
zations. IEEE Transaction on Automatic Control, 46(12), 1941–1946.
Berrada, Y., El Amrani, A., & Boumhidi, I. (2017). Finite frequency T-S fuzzy control for a
variable speed wind turbine. In 14th IEEE International Multi-conference on Systems, Signals
& Devices (SSD), Marrakech (pp. 505–510).
Bououden, S., Chadli, M., Filali, S., & El Hajjaji, A. (2012). Fuzzy model based multivariable
predictive control of a variable speed wind turbine: LMI approach. Renewable Energy, 37(1),
434–439.
Chen, Y., Zhang, W., & Gao, H. (2010). Finite frequency H∞ control for building under earthquake
excitation. Mechatronics, 20(1), 128–142.
El-Amrani, A., Hmamed, A., Boukili, B., & El Hajjaji, A. (2016). H∞ filtering of TS fuzzy systems
in finite frequency domain. In 5th IEEE International Conference on Systems and Control
(ICSC), Marrakech (pp. 306–312).
280 Y. Berrada et al.

Galdi, V., Piccolo, A., & Siano, P. (2008). Designing an adaptive fuzzy controller for maximum
wind energy extraction. IEEE Transaction on Energy Conversion, 23(2), 559–569.
Goodwin, G. C., Graebe, S. F., & Salgado, M. E. (2001). Control system design. Upper Saddle
River: Prentice Hall.
Hafiz, F., & Abdennour, A. (2016). An adaptive neuro-fuzzy inertia controller for variable-speed
wind turbines. Renewable Energy, 92, 136–146.
Iwasaki, T., & Hara, S. (2005). Generalized KYP lemma: Unified frequency domain inequalities
with design applications. IEEE Transaction on Automatic Control, 50(1), 41–59.
Lasheen, A., & Elshafe, I. A. L. (2016). Wind-turbine collective-pitch control via a fuzzy predictive
algorithm. Renewable Energy, 87, 298–306.
Li, X. J., & Yang, G. H. (2015). Adaptive H∞ control in finite frequency domain for uncertain
linear systems. Information Sciences, 314, 14–27.
Liu, P., Yang, W. T., Yang, C. E., & Hsu, C. L. (2015). Sensorless wind energy conversion system
maximum power point tracking using Takagi-Sugeno fuzzy cerebellar model articulation
control. Applied Soft Computing, 29, 450–460.
Medjber, A., Guessoum, A., Belmili, H., & Mellit, A. (2016). New neural network and fuzzy
logic controllers to monitor maximum power for wind energy conversion system. Energy, 106,
137–146.
Skelton, R. E., & Iwasaki, T. (1998). A unified algebraic approach to linear control design.
London/Bristol: Taylor and Francis.
Zhang, H., Wang, R., Wang, J., & Shi, Y. (2014). Robust finite frequency H∞ static-output-
feedback control with application to vibration active control of structural systems. Mechatron-
ics, 24(4), 354–366.
Chapter 14
Intelligent Load Frequency Control in
Presence of Wind Power Generation

Nour EL Yakine Kouba and Mohamed Boudour

Abstract With the advent of large-scale interconnected power systems, many

new problems have emerged, which include frequency fluctuations problem. In
many parts of the world, installed capacity and energy production levels for
electric generation from non-conventional renewable resources such as wind power
generation are growing rapidly. However, the fluctuations of these generators affect
the system frequency. The purpose of this work is to design an intelligent load
frequency control (LFC) strategy based on Fuzzy Logic-PID controller to suppress
all the fluctuations of the total power output of the wind generation and maintain
the constancy of the system frequency. To show the effectiveness of the proposed
control strategy, a two-area multi-sources power system was investigated for the
simulation. The observed simulation results of the proposed Fuzzy Logic-PID
controller are compared with the results obtained by the classical Ziegler-Nichols
method and the meta-heuristic Particle Swarm Optimization (PSO) technique. The
transient responses showing the integration impact of the wind farm are depicted and
the results are tabulated as a comparative performance in view of peak overshoot and
settling time. The results are compared and the ability of the proposed approach to
evaluate load frequency control over large wind farm integration is confirmed.

Keywords Load frequency control (LFC) · PID controller · Fuzzy logic control
(FLC) · Wind power generation · Wind farm

N. EL. Y. Kouba () · M. Boudour

Laboratory of Electrical and Industrial Systems, Faculty of Electrical Engineering and
Computing, University of Sciences and Technology Houari Boumediene, Bab Ezzouar, Algiers,
Algeria
e-mail: [email protected]; [email protected]

© Springer Nature Singapore Pte Ltd. 2019 281

N. Derbel, Q. Zhu (eds.), Modeling, Identification and Control Methods
in Renewable Energy Systems, Green Energy and Technology,
https://doi.org/10.1007/978-981-13-1945-7_14
282 N. EL. Y. Kouba and M. Boudour

14.1 Introduction

The endeavor for a more sustainable power generation has led to a fast increase
in green power generation from renewable energy sources (RES), such as wind,
biomass, hydro and solar power. These sources are of a fluctuating nature; conse-
quently, many new challenges must be introduced in power system stability and
control. Amongst them, a wind farm type of power generation is considered the
most viable alternative (Kouba et al. 2016c,d). The generation of wind energy
is clean, exhaustless and more economic. However, this energy is not stable and
cannot ensure the constant power generation because it depends on the wind as a
natural source (Bevrani and Daneshmand 2012). In addition, a large wind farm can
include hundreds of wind turbines and the output of each one of them depends on
wind direction and velocity, which result into frequency oscillations (Eduardo et al.
2011; Kouba et al. 2016a). The control of system frequency is a vital aspect for
a secure and stable electrical network with good power quality. Frequency control
can be called upon for a variety of conditions ranging from a gradual change in load
levels over time to a sudden loss of generation or step increase/decrease in load. A
continuous balance between active power generated and active power consumed by
the load and losses is required to maintain frequency constant at nominal system
frequency value within an acceptable tolerance (Patnaik and Dash 2015; Sahu et al.
2013). The frequency and load have an inverse relationship; therefore any imbalance
in active power will result in a frequency deviation. Maintaining the frequency
at its nominal value requires that both active power produced and consumed be
controlled to keep the load and supply in equilibrium. The increasing size of the
interconnected power systems has been accompanied with the appearance of the
frequency fluctuations problems, which may results in disconnection actions, loss
of several lines, zone isolation and black-out. Therefore, the frequency control on
an interconnected electrical network is particularly challenging function and can be
considered to be one of the most crucial aspects of ancillary services (Bihui et al.
2011; Kouba et al. 2016b). In large-scale interconnected power system, balancing
power production and consumption is usually guaranteed by Automatic Generation
control (AGC) scheme adjusting production of some power plants to meet the
current demand (Ramakrishna and Bhatti 2008). If there is a power mismatch,
system frequency will change as the rotating mass in generators will be either
accelerated, thus increasing frequency, if too much electric power is produced, or
decelerated, thus decreasing frequency, if the demand is bigger than the production
(Kiaee et al. 2013; Prakash and Sinha 2014). In the case of any contingencies
such as the change in load, failure of a plant or the outage of a line, if no
action is taken and the power mismatch remains, system frequency will diverge
until a critical point is reached resulting in a black-out, which is an undesirable
case. Generally, in the Transmission System Operators (TSOs) the AGC scheme is
mapped to three control levels which are: primary, secondary and tertiary control.
The primary control is implemented through the governor control system and is
used to stabilize the system frequency. The secondary control named also load
14 Intelligent Load Frequency Control for Wind Power Generation 283

frequency control (LFC), reacts slower than primary control and is used to relieve
the primary control and restore the system frequency to its pre-disturbance nominal
value. Finally, Tertiary control refers to the economic dispatching control (EDC)
of each unit and is used to relieve the secondary control loop (LFC) (Kouba et al.
2014b; Tofighi et al. 2015). As major functions of automatic generation control
(AGC), power system frequency regulation named load frequency control (LFC)
become one of the most important research topics in power system operation. The
main goals of LFC are to suppress the fluctuations of the system frequency and
maintain the frequency and the power interchanges with neighboring control areas
at the scheduled values. To satisfy these objectives, a control error signal called the
Area Control Error (ACE) is measured, which represents the real power imbalance
between generation and load, where this signal combines both frequency and net
interchange flow deviations (Kiaee et al. 2013; Ramakrishna and Bhatti 2008). A
PID controller is used to support the frequency regulation LFC loop in each control
area, while the measured ACE signal is the input of the PID controller in each
control area (Kouba et al. 2014b). In fact, small variations in system frequency
will not result in a reduction of system reliability and security. In contrast, large
frequency fluctuations in particularly with the increasing integration of renewable
energy sources such as wind power generation can have a serious impact on
power system equipments and power quality (Chung et al. 2011; Hooshmand et al.
2012). The large increase integration of wind farm in the grid causes extensive
changes in power systems, and the generation does not equal scheduled generation
at all times, while this mismatch adds to the usual imbalances between supply
and demand. However, if a large amount of wind power generation is installed
in the grid; it becomes difficult to remove frequency fluctuations. Therefore, it’s
necessary to study the effects on the frequency control and assess the impact of
wind integration. In the aim to balance the deviation between power production
and power demand in presence of a large penetration of wind power generation, a
robust LFC controller scheme is needed to satisfy these objectives. Various studies
consider new control methods dealing with the design of the secondary frequency
controller (LFC) have been proposed and discussed in last decades. In 1942, Ziegler
and Nichols proposed two heuristic approaches based on their experience and some
simulations to quickly adjust the controller parameters: P, PI, and PID (Kouba
et al. 2014b). Many others techniques were used for tuning the PID controller
parameters such as the stochastic particle swarm optimization (PSO) (RamaSudha
et al. 2010), bacterial foraging optimization algorithm (BFOA) (Nanda et al. 2009),
hybrid algorithm between bacterial foraging and particle swarm optimization (BF-
PSO) (Kouba et al. 2014a), genetic algorithm (GA) (Demirören et al. 2002; Panda
and Yegireddy 2013), differential evolution algorithm (DEA) (Pandey et al. 2013),
firefly algorithm (FA) (Saikia and Sahu 2013), and many others strategies have been
proposed for the LFC study such as the artificial neural network (ANN) (Kouba et al.
2014c; Mahabuba and Khan 2009; Saikia et al. 2011) and H-infinity techniques
(Singh et al. 2013). In this work, a dynamic study of the load frequency control
(LFC) with a large penetration of wind power generation is presented. The LFC
problem has been analyzed using a new algorithm based on the implicit integration
284 N. EL. Y. Kouba and M. Boudour

Trapezoidal rule with variable time step and iterative Newton-Raphson method. The
wind farm is modeled using a high-number of wind turbines and its impact on the
system frequency and the tie-line power flow is examined. A dynamic model of the
IEE Japan East 107-bus 30-generator power system is investigated for the purpose
of studying long-term frequency stability and control on an interconnected two-
area power system concerning high penetration of wind farm. The obtained results
are compared to the classical LFC based Ziegler-Nichols method and, thereby to
the optimal LFC based particle swarm optimization (PSO) technique. This work
is organized as follows. In Sect. 14.2, the interconnected power system model is
presented. In Sect. 14.3, the Load Frequency Control (LFC) model is described, and
the algorithm used for the frequency stability analysis in this work is explained in
detail. Section 14.4 illustrates the proposed Fuzzy Logic-PID controller technique.
Section 14.5 is devoted to the modeling of the wind farm. Results of the wind farm
integration analysis are given and discussed in Sect. 14.6. Finally, Sect. 14.7 includes
the conclusion of this work.

14.2 Interconnected Power System Model

To investigate the performance of the proposed control strategy, the multi-machines

power generation (hydro, thermal, and nuclear) IEE Japan East is considered
as the test system. This electrical system is widely used in the literature as a
standard system for testing of new power system analysis and control strategies.
Figure 14.1 shows a single-line diagram of the test system (Arita et al. 2006). This
electrical network consists of 107 buses, 30 generators, 31 loads, 191 branches,
136 transmission lines and 55 transformers. The power system is divided to two
interconnected control areas. All power generating units in the power system are
equipped with generator, turbine and speed governor. The interconnection between
the control areas is made using a tie-line power flow. It is noted that, outputs of
hydro, thermal and nuclear power plants are controlled automatically with the LFC
signal.
In this work, all proposed thermal and nuclear units are modeled with an
equivalent thermal unit and all proposed hydro power plans are modeled with
an equivalent hydro unit. For the load frequency control study, a classical model
of synchronous machine is needed, which includes the mechanicals equations of
the generator (swing equation) and the speed regulation system (governor-turbine)
model. In our problem, the differential equations representing dynamics of rotating
machines are linearized by a numerical integration method, such as the implicit
integration Trapezoidal rule, and solved using the iterative Newton-Raphson method
for each time step Δt. The control loop model of a synchronous machine which is
used in this work is shown in Fig. 14.2.
14 Intelligent Load Frequency Control for Wind Power Generation 285

Fig. 14.1 IEE Japan East 107-bus 30-machine power system model

Fig. 14.2 Load frequency control loop of a synchronous machine

14.2.1 Generator Model

The relationship between the mechanical power ΔPm and the electrical power ΔPe
is given by Arita et al. (2006):
286 N. EL. Y. Kouba and M. Boudour

dΔω
M = ΔPm − ΔPe (14.1)
dt
The load in a power system is the sum of various power demands and varies
randomly over time. Some loads depend on the change of frequency and others
don’t. In general, the expression of the electrical power which depends on the
frequency change can be expressed by Kouba et al. (2014b):

ΔPe = ΔPL + DΔω (14.2)

In multi-machines system, if all generators are assumed to turn with the same speed
of synchronism, the equations of the equivalent generator can be expressed with:
• The equivalent inertia constant:


i=n
Mi
i=1
Meq = (14.3)
n
• The equivalent load-damping constant:


i=n
Di
i=1
Deq = (14.4)
n
The block diagram representation of equivalent generator used in this work is shown
in Fig. 14.3.
The equivalent generator can be expressed by the following equation:

dΔω 1 ! "
= ΔPmth + ΔPmh − (αΔPtie ) − ΔPL − (Deq Δω) (14.5)
dt Meq

Fig. 14.3 Block diagram representation of equivalent generator

14 Intelligent Load Frequency Control for Wind Power Generation 287

ΔPg SCV

GFmax
SCVmax CVopen
CV
1 – + 1 1
ΔW KG
1 + S *TSR TMS S
+ –
GFmin
U SCVmin CVclose
Speed Governor

Servomotor
ΔPHP ΔPLP

+ ΔPmth
1 1
CV FLP
1 + S *TSC 1 + S *TCO
+

FHP
Turbine

Fig. 14.4 Block diagram of governor-turbine model for thermal/nuclear unit

14.2.2 Governor-Turbine System Model

The main role of a speed governor control system is to adjust the turbine valve to
stabilize the system frequency. The schematics of such governors control systems
used in this work are shown in Fig. 14.4 for the thermal and nuclear unit and in
Fig. 14.5 for the hydro unit (Kouba et al. 2014b).
The equations representing the thermal/nuclear power plant are given by:
⎧
⎪ dΔPg 1
⎪
⎨ = (KG Δω − ΔPg )
dt TSR
(14.6)
⎪
⎪
⎩
W ith : GFmin < ΔPg < GFmax
⎧
⎪ 1
⎪
⎨ SCV = (U − ΔPg − CV )
TSM
(14.7)
⎪
⎪
⎩
W ith : SCVmin < SCV < SCVmax
⎧
⎪ dCV 1
⎪
⎨ = (U − ΔPg − CV )
dt TSM
(14.8)
⎪
⎪
⎩
W ith : CVclose < CV < CVopen
288 N. EL. Y. Kouba and M. Boudour

ΔPgh DCV

GFmax DCVmax CVopen

– +
Δw 1 1 1 ΔPmh
KGh
1 + S∗TSRh TSMh S
+
–
GFmin DCVmin CVclose
U

Fig. 14.5 Block diagram of governor-turbine model for hydro unit

⎧ dΔPH P 1
⎪
⎪ = (CV − ΔPH P )
⎪
⎨ dt TSC
(14.9)
⎪
⎪
⎪
⎩ dΔPLP 1
=( )(ΔPH P − ΔPLP )
dt TCO

The mechanical power for the thermal unit is calculated by this expression:

ΔPmth = (FH P ΔPH P ) + (FLP ΔPLP ) (14.10)

The equations representing the hydro power plant are given by:
⎧
⎨ dΔPgh 1
=( )((ΔωKGh ) − ΔPgh )
dt TSRH (14.11)
⎩
W ith : GFmin < ΔPgh < GFmax
⎧
⎨ DCV = (
1
)(U − ΔPgh − ΔPmh )
TSMh (14.12)
⎩
W ith : DCVmin < DCV < DCVmax
⎧
⎨ dΔPmh
=(
1
)(U − ΔPgh − ΔPmh )
dt TSMh (14.13)
⎩
W ith : CVclose < CV < CVopen

14.2.3 Tie-Line Model

Using DC load flow method and assuming that the tie-line is from area-1 to area-2,
the deviation ΔP12 from the nominal flow can be expressed (Kouba et al. 2014a) by:
14 Intelligent Load Frequency Control for Wind Power Generation 289

Fig. 14.6 Block diagram of Δw1

the tie-line power flow model

+
ΔPT12 1
T
S
–

Δw2

dΔPtie
= T12 (Δω1 − Δω2 ) (14.14)
dt
The block diagram representation for the tie-line is shown in Fig. 14.6.

14.3 Load Frequency Control (LFC) Model

The frequency has an inverse relationship with the load that is changing continually.
Hence, the change in real power affects the system frequency. To maintain the
system frequency constant, the power supply must follow the momentary power
load change. Corresponding to the demand fluctuation, three control strategies as
mentioned in the introduction are used in the electrical network to solve frequency
fluctuation problem. The primary control: is a local control through the speed
governor control system, is used to stabilize the frequency and is effective for the
load change within ten seconds of the disturbance. The secondary control: named
the load frequency control (LFC) is used for the disturbance with the period of
several minutes to about 30 min. The objective of the LFC loop is to adjust operating
point reference of governor in the control area and maintain the system frequency at
the nominal value. The tertiary control: refers to the economic dispatching control
(EDC) of units, and presents a part of the regular market clearing mechanism.
Tertiary control acts on minute-to-hours time scale (30 min–h) (Kassem et al. 2013;
Liu et al. 2015; Pan and Das 2015; Pandey et al. 2013; Rahmani and Sadati 2013).
In large and interconnected power systems, the load frequency control is considered
as the most important control strategy to remove the fluctuations with a long-period.
In the centralized LFC model, to evaluate the area requirement (AR), the actual
frequency and net interchange power flow are measured by the independent system
operator (ISO). The LFC output power signal is sent to each generator when the
frequency deviation is detected. Then, each speed governor output of the selected
power plant is adjusted by the LFC signal to change the power plant output. To better
understanding the function of LFC, Fig. 14.7 depicts the general LFC algorithm.
To keep the system frequency and power balance at the scheduled values, each
generator is equiped with PID controller. The PID parameters are tuned using the
290 N. EL. Y. Kouba and M. Boudour

AGC Algorithm

1 Gen.units output error Ramp Up /Ramp 2

Frequency Tie-Line Flow
Measurement Measurement

Generating Units

Power injection

Power Network

Indicates Power Lines Tie Line Flow

Frequency Sensor
Indicates communication Lines Sensor

Fig. 14.7 AGC/LFC algorithm

Kp
+
Δf + ACE Ki + U
β
S
+ +
α Kd.s
ΔPtieij

Fig. 14.8 Block diagram of LFC model

traditional Ziegler-Nichols method, the meta-heuristic particle swarm optimization

(PSO) algorithm, and the Fuzzy Logic (FL) strategy. The diagram of the LFC model
used in this work is shown in Fig. 14.8 (Kouba et al. 2014a). The input signal of each
PID controller is the area control error (ACE), which is given by:

ACEi = αΔPtie + βf i Δωi (14.15)

The control equation U in each control area is given by:

Ki
U = ACE(Kp + + Kd S) (14.16)
S
In order to analyze the LFC problem in the two-area interconnected power
system, the differential equations system in Eq. (14.17) is algebraized using the
implicit Trapezoidal integration rule. After that, the resulting algebraic equations
are solved using the iterative Newton-Raphson method at each time step based on
14 Intelligent Load Frequency Control for Wind Power Generation 291

the work developed in Kouba et al. (2014b). By application of implicit integration

Trapezoidal rule with a variable integration time step (h = Δt), the following
equations can be obtained:
   
Δt Δt
F = Xn+1 − f (Xn+1 , tn+1 ) − Xn + f (Xn , tn ) (14.17)
2 2

The differential equations system for the LFC analysis is given by:
⎧
⎪
⎪ dΔPg 1
⎪
⎪ = (KG Δω − ΔPg )
⎪
⎪ dt TSR
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪
dCV
=
1
(U − ΔPg − CV )
⎪
⎪
⎪
⎪ dt TSM
⎪
⎪
⎪
⎪
⎪
⎪ dΔPH P 1
⎪
⎪ = (CV − ΔPH P )
⎪
⎪
⎪
⎪ dt TSC
⎪
⎪
⎪
⎪
⎪
⎪ dΔPLP 1
⎪
⎪ = (ΔPH P − ΔPLP )
⎪
⎪
⎪
⎪ dt TCO
⎪
⎪
⎪
⎨
dΔPgh 1
= (KGh Δω − ΔPgh ) (14.18)
⎪
⎪ dt TSRH
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪ dΔPmh 1
⎪
⎪ = (U − ΔPgh − ΔPmh )
⎪
⎪ dt TSMh
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪ dΔω 1
⎪
⎪ = (ΔPmth + ΔPmh − αΔPtie − ΔPL − Deq Δω)
⎪
⎪ dt Meq
⎪
⎪
⎪
⎪
⎪
⎪
⎪
⎪ dUI
⎪
⎪ = KI (αΔPtie − βf Δω)
⎪
⎪ dt
⎪
⎪
⎪
⎪
⎪
⎪ dΔPtie
⎪
⎩ = T12 (Δω1 − Δω2 )
dt

By applying Eq. (14.17) to the deferential equations system (14.18), the obtained
algebraic equations system is given in (14.20), where:
⎡ ⎤
F11(j,i)
   
⎢ F12(j,i) ⎥ ! "
F1(j,i) = ⎢ ⎥ ; F2(j,i) = F21(j,i) ; F3(i) = F31(i) ; F4(i) = F4(i)
⎣ F13(j,i) ⎦ F22(j,i) F32(i)
F14(j,i)

F = [F1 , F2 , F3 , F4 ]t (14.19)
292 N. EL. Y. Kouba and M. Boudour

At each time step h, the equation [F ] = 0 is solved by Eq. (14.21).

⎧  
⎪
⎪ Δt
⎪
⎪ F11 (xn + 1, xn ) = ΔPgn+1 − (KG Δωn+1 − ΔPgn+1 )
⎪
⎪  2TSR 
⎪
⎪
⎪
⎪ Δt
⎪
⎪ − ΔPgn + [((Δωn KG ) − ΔPgn )]
⎪
⎪  2TSR 
⎪
⎪
⎪
⎪ Δt
⎪
⎪ F12 (xn + 1, xn ) = CVn+1 − (Un+1 − ΔPgn+1 − CVn+1 )
⎪
⎪  2TSM 
⎪
⎪
⎪
⎪ Δt
⎪
⎪ − CVn + (Un − ΔPgn − CVn )
⎪
⎪  2TSM 
⎪
⎪
⎪
⎪ Δt
⎪
⎪ F13 (xn + 1, xn ) = ΔPH Pn+1 − (CVn+1 − ΔPH Pn+1 )
⎪
⎪  2TSC 
⎪
⎪
⎪
⎪ Δt
⎪
⎪ − ΔPH Pn + (CVn − ΔPH Pn )
⎪
⎪  2TSC 
⎪
⎪
⎪
⎪ Δt
⎪
⎪ F14 (xn + 1, xn ) = ΔPLPn+1 − (ΔPH Pn+1 − ΔPLPn+1 )
⎪
⎪  2TCO 
⎪
⎪
⎪
⎪ Δt
⎪
⎪ − ΔPLPn + (ΔPH Pn − ΔPLPn )
⎪
⎪  2TCO 
⎪
⎪
⎪
⎨ F21 (xn + 1, xn ) = Δt
ΔPghn+1 − (KGh Δωn+1 − ΔPghn+1 )
 2TSRH 
⎪
⎪ Δt
⎪
⎪ − ΔP + (K Δω − ΔP )
⎪
⎪ gh n Gh n gh n
⎪
⎪  2TSRH 
⎪
⎪ Δt
⎪
⎪ F (x + 1, x ) = ΔP − (U − ΔP − ΔP )
⎪
⎪ 22 n n mh n+1 gh n+1 mh n+1
⎪
⎪  2TSMh 
⎪
⎪ Δt
⎪
⎪ − ΔP + (U − ΔP − ΔP )
⎪
⎪ mh n gh n mh n
⎪
⎪ 2TSMh
⎪
⎪ Δt
⎪
⎪ F31 (xn + 1, xn ) = (Δωn+1 − Δωn ) − ×
⎪
⎪ 2Meq
⎪
⎪ '
⎪
⎪
⎪
⎪ (ΔPmthn+1+ΔPmhn+1−αΔPtien+1−ΔPL−Deq Δωn+1 )
⎪
⎪ (
⎪
⎪
⎪
⎪ −(ΔP mth + ΔP mh − αΔP tie − ΔP L − D eq Δω n )
⎪
⎪
n n n
⎪
⎪ Δt
⎪
⎪ F32 (xn + 1, xn ) = (UIn+1 − UI n ) − ×
⎪
⎪ ' 2 (
⎪
⎪
⎪
⎪ KI α(ΔPtien+1 − ΔPtien ) − βf (Δωn+1 − Δωn )
⎪
⎪
⎪
⎪
⎪
⎪ F4 (xn + 1, xn ) = (ΔPtien+1 − ΔPtien ) −
Δt
×
⎪
⎪
⎪
⎪ ' 2 (
⎪
⎩ T12 (Δω1n+1 − Δω1n ) − (Δω2n+1 − Δω2n )
(14.20)
In our problem of LFC, the Newton-Raphson iterates are:

(J )k [Δxn+1 ]k = −[F ]k (14.21)

14 Intelligent Load Frequency Control for Wind Power Generation 293

⎛ ⎞k ⎛ ⎞k ⎛ ⎞k
J1 J2 J3 J4 J5 J6 ΔX1 (1) F1 (1)
⎜J J12 ⎟ ⎜ ⎟ ⎜ F (1) ⎟
⎜ 7 J8 J9 J10 J11 ⎟ ⎜ ΔX2 (1) ⎟ ⎜ 2 ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎜ J13 J14 J15 J16 J17 J18 ⎟ ⎜ ΔX3 (1) ⎟ ⎜ F3 (1) ⎟
⎜ ⎟ ⎜ ⎟ =⎜ ⎟ (14.22)
⎜ J19 J20 J21 J22 J23 J24 ⎟ ⎜ ΔX4 ⎟ ⎜ F4 ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎝ J25 J26 J27 J28 J29 J30 ⎠ ⎝ ΔX1 (2) ⎠ ⎝ F1 (2) ⎠
J31 J32 J33 J34 J35 J36 ΔX3 (2) F3 (2)

The state vector is given by:

[X] = [ΔPg1 CV1 ΔPH P1 ΔPLP1 ΔPgh1 ΔPmh1 Δω1 UI1 ΔPtie

ΔPg2 CV2 ΔPH P2 ΔPLP2 Δω2 UI2 ]t

(14.23)
The Jacobian matrix J is given by:
⎡ dF (1) dF1 (1) dF1 (1) dF1 (1) dF1 (1) dF1 (1) ⎤
1
⎢ dX1 (1) dX2 (1) dX3 (1) dX4 dX1 (2) dX3 (2) ⎥
⎢ dF (1) dF2 (1) ⎥
⎢ 2 dF2 (1) dF2 (1) dF2 (1) dF2 (1) ⎥
⎢ ⎥
⎢ dX1 (1) dX2 (1) dX3 (1) dX4 dX1 (2) dX3 (2) ⎥
⎢ dF3 (1) dF3 (1) dF3 (1) dF3 (1) dF3 (1) dF3 (1) ⎥
⎢ ⎥
⎢ dX (1) dX3 (2) ⎥
J =⎢ 1
⎢ dF4
dX2 (1)
dF4
dX3 (1)
dF4
dX4
dF4
dX1 (2)
dF4 dF4 ⎥
⎥ (14.24)
⎢ ⎥
⎢ dX1 (1) dX2 (1) dX3 (1) dX4 dX1 (2) dX3 (2) ⎥
⎢ ⎥
⎢ dF1 (2) dF1 (2) dF1 (2) dF1 (2) dF1 (2) dF1 (2) ⎥
⎢ ⎥
⎢ dX1 (1) dX2 (1) dX3 (1) dX4 dX1 (2) dX3 (2) ⎥
⎣ dF (2) dF2 (2) dF2 (2) dF2 (2) dF2 (2) dF (2) ⎦
2 2
dX1 (1) dX2 (1) dX3 (1) dX4 dX1 (2) dX3 (2)

The new solution at the iteration k + 1 is calculated using the flowing equation:

x k+1 = Δx k + x k (14.25)

1. The initial condition values x 0 are fixed.

2. The convergence of iterative Newton-Raphson is based on the max absolute error
max |F (x k+1 )| and the specified tolerance  = 10−6 .
3. If max|F (x k+1 )| < , iterative Newton-Raphson have converged.
With: i = 1 : ng number of area, and j = 1 : m number of machines; (In our case
ng = 2 and m = 30).
294 N. EL. Y. Kouba and M. Boudour

14.4 Fuzzy Logic Control

Nowadays, one of promising control methods in industrial automation and process

control is the Fuzzy Logic Control (FLC). Because of simplicity, robustness, and
reliability fuzzy logic is used in almost all research areas, including solving a wide
range of control problems in power system control and operation. Fuzzy logic is
an intelligent technique developed by Professor Lotfi Zadeh. He first introduced
the theory of fuzzy sets and fuzzy logic in 1965 when he wrote his first paper
entitled Fuzzy Sets (Pothiya and Ngamroo 2008; Shayeghi et al. 2008). This logic
is the mathematical representation of the formation of human concepts and of
reasoning concerning human concepts. Several published papers have shown and
proven that fuzzy systems are strong and efficacy solution schemes. Moreover, the
FLC is used in many commercial, domestic, automotive control applications, and
has been successfully applied to many control problems (e.g. washing machines,
televisions and photocopiers) because no mathematical modeling is involved. This
work addresses the problem of tuning the optimal PID controller parameters using
fuzzy logic strategy. Our aim in this work is to design a robust load frequency
controller based Fuzzy Logic-PID controller to regulate the system frequency
concerning large wind power penetration. A general scheme for Fuzzy Logic-
PID controller based LFC system is designed as depicted in Fig. 14.9. As shown,
the parallel combination between Fuzzy Logic and PID controller is adopted in
this system. The Fuzzy Logic controller is implemented in three phases given as
follow:
1. Fuzzification module (Fuzzifer).
2. Rule base and Inference engine.
3. Defuzzification module (Defuzzifier).
For the purpose of load frequency control analysis, the inputs of the fuzzy logic
controller are the variables error (ACE) and change of error (dACE), and the
outputs of the fuzzy logic are the PID parameters. The KP , KI , KD values are
calculated out according to offline rules in fuzzy controller.
A label set corresponding to linguistic variables of the input control signals,
ACE(Z) and dACE(Z), with a sampling time of 0.01 s is as follows:

Li(ACE, dACE) = (NB, NS, ZE, P S, P B) (14.26)

A label set corresponding to linguistic variables of the output control signals is as

follows:

Lo(KP , KI , KD ) = (ZE, P S, P M, P B) (14.27)

The membership function for the control input variables are shown in Fig. 14.10,
and the membership function for the control output variables are given as shown in
Fig. 14.11.
14 Intelligent Load Frequency Control for Wind Power Generation 295

Fig. 14.9 Structure of fuzzy logic-PID controller

Fig. 14.10 Membership function for the control input variables

The control rules are built from the statement: if input a and input b then output
Z, while Table 14.1 resumes the control rules used in this work. In this work the
Triangular membership functions is used. The two input signals (ACE, dACE)
are converted to fuzzy numbers first in fuzzifier using five membership functions
(NB, NS, ZE, P S, P B). Then they are used in the rule table shown in Table 14.1
to determine the fuzzy number of the compensated output signals. The proposed
FLC model is shown in Fig. 14.12. The Fuzzy Logic Controller(FLC) is used to
reach the optimal PID controller parameters. The proposed approach is compared
to the heuristic Particle Swarm Optimization (PSO) technique, and to the classical
Ziegler-Nichols method.
296 N. EL. Y. Kouba and M. Boudour

Fig. 14.11 Membership function for the control output variables

Table 14.1 Control rules ACE

dACE NB NS ZE PS PB
NB PB PB PB PB PS
NS PB PM PM PB PB
ZE PB PM ZE PB PB
PS PB PM PM PM PB
PB PS PB PM PM PB

Fig. 14.12 Proposed fuzzy logic-PID controller structure

14.5 Wind Power Generation System

The increasing share of fluctuating renewable energy sources (RES) in the electrical
networks poses new challenges for power systems operation and control. On the
one hand, renewable energy sources can cover part of the increasing demand and
provide electricity production with low marginal costs and reduce CO2 emissions.
In the other hand, large RES integration could influence the power quality and
disturb the system stability. Presently, the grid integration of variable distributed
generations (DG), such as wind power plants presents one of the most important
issues in power system stability and control. Among all renewable energy sources,
14 Intelligent Load Frequency Control for Wind Power Generation 297

Fig. 14.13 Doubly fed induction generator model

the wind power is the most used in power system. Wind energy is a renewable
electricity production from converting kinetic energy of moving air masses into
electricity. The wind power is the most renewable energy source utilized in power
systems, while this energy depends on the fluctuating nature of the wind direction
and its velocity. In the last decades, wind turbines are effectively displacing classical
generators and their rotating machinery (e.g. thermal, gas, or nuclear). Nowadays
wind power plays an important role in the generation of electrical power as a single
wind power plant with operational capabilities similar to a conventional power
plant (Jafarian and Ranjbar 2013; Mandal et al. 2014; Wang and McCalley 2013).
Therefore, the large deployment of wind turbine led to significant generation shares
of wind farms in electrical power networks. Currently, wind power technology is
becoming very important, while in the wind farms construction, the variable-speed
wind turbine type is the most used technology. Today the doubly fed induction
generator (DFIG) shown in Fig. 14.13 is the most used wind turbine, because of
his high power control capability since a partial-scale power converter and variable
speed operation. The impact analysis of wind farms on power system stability and
control requires the development of suitable models. Many dynamic models of wind
farms have been developed by researchers and network operators with different
detail levels depending on the scope of the study (Ge et al. 2013; Michigami and
Oishi 2001). However, high wind turbines penetration causes many implications in
frequency dynamics and making frequency stability and control more challenging.
When a large penetration of wind power generation is integrated in a small control
area, it influences the area frequency control and the tie-line power flow. For that,
it is necessary to study the impact of large-scale integration of wind power into
interconnected power systems. Hence, maintaining the frequency stability in the
electrical network depending to the active power balance is a necessary requirement
for a good power quality. This makes the importance of impact study of wind power
298 N. EL. Y. Kouba and M. Boudour

integration into the grid a major issue especially during a contingency, which in the
scope of this work. The purpose of this work is to analyze the impact of large wind
farm integration in interconnected power system using the wind turbine model. Two
hundred wind turbines (of 1.5 MW each) are used to represent a 300 MW wind farm,
while the mathematical model of such wind turbine is given below.

14.5.1 Wind Turbine Model

Wind turbines produce electricity by using the power of the wind to drive an
electrical generator. The power in the airflow is given by Hang et al. (2016),
Schlechtingen et al. (2013), Zhang et al. (2013), Sarrias-Mena et al. (2015),
and Attya and Hartkopf (2012):

1
Pair = (ρ.S.V 3 ) (14.28)
2
The extracted power from the wind can be expressed as follows:

1
Pmech = (ρ.π.R 2 .V 3 .Cp(λ, β)) (14.29)
2
The tip-speed ratio λ is defined by:

ωtr .R
λ= (14.30)
V
The power coefficient Cp is given by:

1 π.(λ + 1)
Cp(λ, β) = ( − 0, 167).β. sin( ) − 0, 00184.(λ − 3).β (14.31)
2 18, 5 − 0, 3.β

Figure 14.14 depicts the 1.5 MW wind turbine simulation model in MATLAB.

14.5.2 Wind Farm Model

Increased penetration of wind farm in the interconnected network poses new

challenges to conventional power system operation and control. The impacts of
increasing a large wind power penetration on system frequency regulation and
active power control are of significant interest in the industry. As a result, there is
a rising interest in the supplementary services such as the frequency regulator (i.e.
load frequency control) in presence of wind farm. In a traditional interconnected
power system (without RES), the purpose of the secondary control LFC is to take
14 Intelligent Load Frequency Control for Wind Power Generation 299

Fig. 14.14 1.5 MW wind turbine simulation model in MATLAB

back the system frequency to nominal value and maintain the balancing between
the interconnected control areas. In contrast, due to the fluctuating nature of wind
power, equilibrium between production and demand in presence of wind turbines is
not an easy operation (Attya and Hartkopf 2012; Vrakopoulou 2013). Wind turbines
are non-dispatchable and non-controllable such as the conventional power plants.
Therefore, employing the stochastic nature of wind power may result in an adverse
effect on the network. Furthermore, the increasing integration of wind farms to
satisfy consumption it may has an opposite influence on both frequency control and
power quality. It should thus be apparent that it is necessary to design a robust load
frequency controller for optimal solution in presence of wind power generation.
In a normal state of power system, automatic load frequency control takes place
to keep the system within the safety margins, following generation and demand
fluctuations. This LFC loop is also depending on the ability of the interconnected
network to withstand disturbances. In presence of wind farm, the supply-demand
active power mismatch occurs and the frequency will deviate from its nominal value.
However, to analyze the additional fluctuation caused by wind farm, optimal LFC
controller needs to be employed. The objective is to bring frequency fluctuation
back to zero and maintain the power flow on the tie-lines that connect it with
the other control areas at its scheduled value. Therefore, each power plant could
be controlled using the supplementary frequency control LFC loop (Vrakopoulou
2013). In the case of a high number of wind turbines are installed into a balancing
control area, fluctuations in wind power may reduce the balancing areas ability and
makes the frequency control more difficult as shown in Fig. 14.15. Towards a stable
300 N. EL. Y. Kouba and M. Boudour

Fig. 14.15 Impact of integration wind farm on LFC

Fig. 14.16 Layout of a typical wind farm

power system operation, the study of possibility of integration a large wind farm
in interconnected electrical networks and analysis its impact, an issue which serves
as motivation for this work. In this work, a typical wind farm composed of 200
wind turbines of 1.5 MW each is used for the simulation as shown in Fig. 14.16. It
is assumed that a large control area is interconnected to a small control area with
a large penetration of wind turbines. The wind farm is mapped into eight groups,
where each group produces a power of 37.5 MW. This wind farm is based on the
wind turbine model, and the wind profile is presented using the dynamic load model
proposed by Michigami and Oishi (2001) as shown in Fig. 14.17.
An equivalent wind speed calculated in Eq. (14.32) is applied to the wind farm.
We propose that the same wind profile shown in Fig. 14.18 is applied to all wind
14 Intelligent Load Frequency Control for Wind Power Generation 301

1
Band-Limited 3000s*1
White Noise2 LPF1
Product1 + –
Scope1
1 + K– +
+ – 1
Constant2 Gain2
+ Wind
Subract Subract1
300s Subract2
Band-Limited 300s*1 4
White Noise1 Product2 LPF Constant1
0.15
Constant

Fig. 14.17 Wind speed model

14

12
wind profil (m/s)

10

2
0 100 200 300 400 500 600
time (s)

Fig. 14.18 Wind speed output

turbines of the same group: (nr is number of groups):


i=n
Vi
i=1
Veq = (14.32)
nr

The dynamic load model proposed by T. Michigami and T. Ishii is used to represent
the wind speed (Michigami and Oishi 2001). The wind speed profile is generated by
a block diagram shown in Fig. 14.17. A load consists of base and fringe components
obtained from the white noise generators, since the component with period shorter
than 5 min or longer than 30 min are out of LFC control, the component is eliminated
by applying filters HPF and LPF. The energy penetration of wind farm is defined as:

Ew
p (%) = × 100 (%) (14.33)
E
302 N. EL. Y. Kouba and M. Boudour

On the whole and in a simplified manner, the wind farm model is present by:


n 
n 
n 
n
Seq = Si ; P meq = P mi ; Qeq = Qi ; Cpeq = Cpi (14.34)
1 1 1 1

where: the subscript i represents the single turbine.

14.6 Simulation Results and Discussion

To satisfy the load frequency control objectives concerning the integration of

wind farm, a dynamic time domain simulation has been carried out using the
interconnected IEE Japan East 107-bus, 30-machine. It consists of two areas
interconnected by a tie-line power flow. As mentioned, the main goal is to design
an effective load frequency controller model with a desirable performance in the
presence of large-scale integration of wind turbines. An additional 300 MW wind
farm generation is assumed to be installed in area-2. As presented in Sect. 14.5.2,
Fig. 14.16 shows the wind farm structure considered in this work. It presents
200 × 1.5 MW DFIG groups, while the wind farm model operates with the same
variable wind profile. All wind turbines are assumed connected to the same PCC.
The algorithm developed to analyze the load frequency control LFC is shown in
Figs. 14.19 and 14.20. In order to verify the effectiveness of the proposed algorithm,
the new LFC model proposed in this work is verified by comparing their responses
with those of the conventional LFC based Ziegler-Nichols and the optimal LFC
based PSO. The numerical parameters used in the simulation are included in the
“Appendix” (Arita et al. 2006) (Tables 14.5, 14.6, 14.7, 14.8, and 14.9).
Using the proposed wind farm model, the impact of wind power fluctuations
on the system frequency and tie-line power flow in the two-area interconnected
network is examined. Figures 14.21 and 14.22; depict the output power of the first
wind turbine and the power coefficient Cp (λ, β) characteristics respectively. The
total wind farm power generation is shown in Fig. 14.23 generation is analyzed.
For comparison purpose, four cases are carried out as presented in Table 14.2,
whilst the PID controller parameters are given in Table 14.3. The deviations in
system frequency and tie-line power flow are shown in Figs. 14.24, 14.25, and 14.26,
respectively. It can be seen that the fluctuations of system frequency and tie-line
power flow are very important when the penetration of wind power generation is
large. However, the system frequency and the tie-line power flow are suppressed
most effectively if both areas adopt LFC based optimal PID controller. The effect
of the PID controller on the system fluctuations caused by the wind farm is clear
that in the difference between the blue corves (without control) and the other
corves (using LFC with PID controller). It is clear that in the case of using the
conventional LFC design shows more oscillations. In contrast, using the PSO
method, global and local solutions could be found simultaneously for a better
agreement of the PID controller parameters. Also, these figures show the superior
Fig. 14.19 Proposed flowchart for LFC analysis
Fig. 14.20 Proposed flowchart for variable time step

x 105
16

14

12
first turbine power(MW)

10

0
0 100 200 300 400 500 600
time (s)

Fig. 14.21 Output power of first wind turbine

 Power Coefficient Cp
0.6

0.5

0.4
Cp

0.3

0.2

0.1

0
0 2 4 6 8 10 12 14
Tip−Speed Ratio

Fig. 14.22 Power coefficient CP (λ, β) characteristic

x 108
3.5

3
wind farm power(MW)

2.5

1.5

0.5

0
0 100 200 300 400 500 600
time (s)

Fig. 14.23 Output power of the wind farm

Table 14.2 Simulation case Case 1 Case 2 Case 3 Case 4

study
Without LFC Ziegler-Nichols PSO Fuzzy-PID
306 N. EL. Y. Kouba and M. Boudour

Table 14.3 PID controller Parameters

parameters
Methods KP KI KD
Ziegler-Nichols 0.06168 0.0173 0.055
PSO 13.9506 0.2524 1.8796
Fuzzy logic 3.33 1.65 3.33

0.3
Without LFC
Frequency deviation in Area−1(HZ)

0.25 With LFC−PID (Ziegler−Nichols)

With LFC−PID (PSO)
With LFC−Fuzzy−PID
0.2

0.15

0.1

0.05

−0.05
0 100 200 300 400 500 600
time (s)

Fig. 14.24 Frequency fluctuation in area-1

0.3
Without LFC
Frequency deviation in Area−2(HZ)

0.25 With LFC−PID (Ziegler−Nichols)

With LFC−PID (PSO)
With LFC−Fuzzy−PID
0.2

0.15

0.1

0.05

−0.05

−0.1
0 100 200 300 400 500 600
time (s)

Fig. 14.25 Frequency fluctuation in area-2

14 Intelligent Load Frequency Control for Wind Power Generation 307

50
Tie−line power flow deviation(MW)
0

−50 Without LFC

With LFC−PID (Ziegler−Nichols)
With LFC−PID (PSO)
−100 With LFC−Fuzzy−PID

−150

−200
0 100 200 300 400 500 600
time (s)

Fig. 14.26 Tie-line power flow fluctuation

Table 14.4 Results and comparison

Tuning PID controller Without Ziegler- Fuzzy
techniques control Nichols PSO logic
Max frequency deviation 0.2633 0.09184 0.02176 0.00474
in Area-1 [Hz]
Max frequency deviation 0.2827 0.1074 0.05776 0.0148
in Area-2 [Hz]
Max tie-line power flow 194.1 16.06 3.735 1.044
deviation [MW]

Table 14.5 Data of the Area

two-area power system
Parameters Area-1 Area-2
Reference frequency [Hz] 50 50
Inertia constant [s] 8.85 9.02
Load-damping constant [pu] 2 2

performance of the proposed Fuzzy Logic-PID controller based LFC scheme to the
other conventional and optimal LFC models. Using the proposed Fuzzy Logic-PID
controller, the fluctuations of the system frequency and the tie-line power flow are
better suppressed compared to the results given by the classical Ziegler-Nichols
method and the PSO technique. In other hand, the proposed algorithm based on
the implicit integration Trapezoidal rule with variable time step and the iterative
Newton-Raphson method is proven to be very efficient for the frequency stability
study. The results are compared in view of peak overshoot and settling time; where,
the proposed Fuzzy logic-PID controller based LFC scheme is proven to be better
as shown in Table 14.4.
308 N. EL. Y. Kouba and M. Boudour

Table 14.6 Thermal and nuclear unit parameters

KG TSR TSM TSC TCO FH P FLP CVclose CVopen SCVmin SCVmax
20 0.2 0.2 0.25 0.9 0.3 0.7 0 1.05 −1000 0.2

Table 14.7 Hydro unit parameters

KGh TSRh TSMh CVclose CVopen DCVmin DCVmax
20 10 0.3 0 1.02 −1000 0.1

Table 14.8 Generation data

Area Rated capacity Initial output Inertia constant Total load
[MW] [MW] [Sec] [MW]
Area-1 53509 30247 8.85 33090
Area-2 11560 10600 9.02 7090

Table 14.9 Wind turbine Blades Rotor Gearbox Moment Coefficient

parameters number diameter ratio inertia of friction
3 36.5 1 : 104 50 0.0071

14.7 Conclusion

This work investigates the impact of integration a large wind farm on frequency
stability and control, which is an important issue in power system control and
operation. For this purpose, a Fuzzy Logic structure was proposed to design a
new robust load frequency control (LFC) scheme in two-area interconnected power
system with diverse sources of power generation concerning high penetration of
wind turbines. As the wind power fluctuations influence power system frequency,
this study investigated the impact of large wind farm integration on the system
frequency and the tie-line power flow in large-scale interconnected electrical
network. A typical wind farm equipped with 200 DFIG wind turbines of 1.5 MW
each was used for dynamic study, from the frequency stability point of view and
control of the electrical power system. The fuzzy logic strategy was used to achieve
the optimal values of the PID controller parameters. A new methodology to solve
the load frequency control based on the implicit integration Trapezoidal rule with
variable time step and iterative Newton-Raphson method was used in this work.
The main objective of the proposed algorithm and the suggested LFC scheme
is to analyze the frequency stability in an interconnected electrical network in
presence of wind farm and solve the frequency fluctuations problem by keeping
system frequency and tie-line power interchange between interconnected areas
within an acceptable range close to the scheduled values. The proposed strategy
was examined on the IEE Japan East 107-bus, 30-machine power system including
wind farm, while this system is divided into two control areas: a big control area
is interconnected into a small control area. The wind farm is installed in the small
control area, where the deviations of the system frequency and the tie-line power
14 Intelligent Load Frequency Control for Wind Power Generation 309

flow in this network are examined to improve the LFC capacity of the conventional
power units. The results obtained using the proposed Fuzzy Logic-PID controller
based LFC scheme was compared with those of the conventional LFC based Ziegler-
Nichols method and the optimal LFC based PSO technique. The results show that
the main advantage of using Fuzzy Logic structure is to reduce the fluctuations,
while achieving a good performance of the whole response of system in presence of
wind farm. Further, the robustness of the proposed control strategy is confirmed and
the LFC scheme provides desirable performance against wind power fluctuations.

Appendix

List of Abbreviations and Symbols

Abbreviations

ACE: Area Control Error.

AGC: Automatic Generation Control.
ANN : Artificial Neural Network.
AR: Area Requirement.
BF OA: Bacterial Foraging Optimization.
BF − P SO: Hybrid Bacterial Foraging and Particle Swarm Optimization.
DEA: Differential Evolution Algorithm.
DF I G: Doubly Fed Induction Generator.
DG: Distributed Generations.
EDC: Economic Dispatching Control.
F A: Firefly Algorithm.
F LC: Fuzzy Logic Control.
GA: Genetic Algorithm.
HPF: High-Pass Filter.
I SO: Independent System Operator.
LF C: Load Frequency Control.
LP F : Low-Pass Filter.
P CC: Point of Common Coupling.
P I D: Proportional-Integral-Derivation Controller.
P SO: Particle Swarm Optimization.
RES: Renewable Energy Sources.
T SOs: Transmission System Operators.
310 N. EL. Y. Kouba and M. Boudour

Symbols

M: Inertia Constant.
Meq : Equivalent Inertia Constant.
D: Load-Damping.
Deq : Equivalent Load-Damping.
Δω: Frequency Deviation.
ΔPm : Mechanical Power Variation.
ΔPe : Electrical Power Variation.
ΔPL : Non-Frequency-Sensitive Load Change.
ΔPtie : Tie-Lien Power Flow Deviation.
ΔPmh : Hydro Unit Mechanical Power Variation.
ΔPmth : Thermal Unit Mechanical Power Varia-
tion.
ΔPg : Thermal Unit Governor Power Variation.
ΔPgh : Hydro Unit Governor Power Variation.
CV : Control Valve.
SCV : Speed Control.
DCV : Distributor Valve and Gate Servomotor.
CVopen , CVclose : Valve or Gate Position Limits.
SCVmin , DCVmin , SCVmax , DCVmax : Valve/ Gate Servomotor Rate Limits.
GFmin , GFmax : Speed Governor Limits.
TCO : Time Constants for the Cross Over.
TSC : Time Constants for the Steam Chest.
FH P : High-Pressure Turbine Power Fraction.
FLP : Low-Pressure Turbine Power Fraction.
KG : Thermal Unit Speed Governor Regula-
tion Gain.
KGh : Hydro Unit Speed Governor Regulation
Gain.
TSR : Thermal Unit Time Constant of the
Speed Relay.
TSM : Thermal Unit Time Constant of the Ser-
vomotor.
TSRh : Hydro Unit Time Constant of the Speed
Relay.
TSMh : Hydro Unit Time Constant of the Servo-
motor.
U: Control Signal.
UI : Integral Control Signal.
βf : Frequency Bias.
a: Constant (a = [1, −1]).
T12 : Tie-Line Rigidity Factor.
XT 12 : Tie-Line Reactance.
14 Intelligent Load Frequency Control for Wind Power Generation 311

Ts : Simulation Time.
tp : Perturbation time.
Pmech : Power from Wind.
ρ: Air Density.
R: Blade Radius.
S: The Swept Area of the Rotor.
V: Wind Speed.
Veq : Equivalent Wind Speed.
Cp: Power Coefficient.
λ: Tip-Speed Ratio.
β: Blade Pitch Angle.
wtr : Turbine Rotation Speed.
Ew : The Energy Supplied by Wind Farm.
E: The Energy Supplied by all Generation.
Seq : Apparent Power of the Whole Wind
Farm.
P meq : Equivalent Active Power of the Whole
Wind Farm.
Qeq : Equivalent Reactive Power of Whole
Wind Farm.
Cpeq : Equivalent Power Coefficient of Whole
Wind Farm.
Δtorh: Integration Time Step.
hOld , hN ew : Old and New Integration Time Step.
: Tolerance.
c: Constant between the interval [0.6, 0.9].
q: Order of the Method, for the Trapezoidal
Method q = 2.
|e|: Is a Weighted Root Square Mean Norm.
k: Iteration Number.

References

Arita, M., Yokoyama, A., & Tada, Y. (2006). Evaluation of battery system for frequency control
in interconnected power system with a large penetration of wind power generation. In IEEE
International Conference on Power System Technology, Chongqing (pp. 1–7).
Attya, A., & Hartkopf, T. (2012). Penetration impact of wind farms equipped with frequency
variations ride through algorithm on power system frequency response. Electrical Power and
Energy Systems, 40, 94–103.
Bevrani, H., & Daneshmand, P. (2012). Fuzzy logic-based load-frequency control concerning high
penetration of wind turbines. IEEE Systems Journal, 6, 173–180.
Bihui, L., Hong, S., Yong, T., Hongyun, Z., Feng, S., & DongFu, L. (2011). Study on the frequency
control method and AGC model of wind power integration based on the full dynamic process
simulation program. In International Conference on Advanced Power System Automation and
Protection, Beijing (pp. 246–251).
312 N. EL. Y. Kouba and M. Boudour

Chung, I.-Y., Liu, W., Cartes, D., & Moon, S.-I. (2011). Control parameter optimization for
multiple distributed generators in a microgrid using particle swarm optimization. European
Transactions on Electrical Power, 21, 1200–1216.
Demirören, A., Kent, S., & Günel, T. (2002). A genetic approach to the optimization of automatic
generation control parameters for power systems. European Transactions on Electrical Power,
12, 275–281.
Eduardo, V.-N., Andreas, S., Oriol, G.-B., Adriã, J.-F., & Marcela, M.-R. (2011). Pitch control
system design to improve frequency response capability of fixed-speed wind turbine systems.
European Transactions on Electrical Power, 21, 1984–2006.
Ge, B., Wang, W., Bi, D., Rogers, C. B., Peng, F. Z., de Almeida, A. T., & Abu-Rub, H. (2013).
Energy storage system-based power control for grid-connected wind power farm. Electrical
Power and Energy Systems, 44, 115–122.
Hang, J., Zhang, J., & Cheng, M. (2016). Application of multi-class fuzzy support vector machine
classifier for fault diagnosis of wind turbine. Fuzzy Sets and Systems, 297, 128–140.
Hooshmand, R., Ataei, M., & Zargari, A. (2012). A new fuzzy sliding mode controller for load
frequency control of large hydropower plant using particle swarm optimization algorithm and
kalman estimator. European Transaction on Electrical Power, 22, 812–830.
Jafarian, M., & Ranjbar, A. (2013). The impact of wind farms with doubly fed induction generators
on power system electromechanical oscillations. Renewable Energy, 50, 780–785.
Kassem, A. M., Hasaneen, K. M., & Yousef, A. M. (2013). Dynamic modeling and robust power
control of DFIG driven by wind turbine at infinite grid. Electrical Power and Energy Systems,
44, 375–382.
Kiaee, M., Cruden, A., Infield, D., & Chladek, P. (2013). Improvement of power system frequency
stability using alkaline electrolysis plants. Proceedings of the Institution of Mechanical
Engineers, Part A: Journal of Power and Energy, 227, 115–123.
Kouba, N. E. Y., Menaa, M., Hasni, M., Boussahoua, B., & Boudour, M. (2014a). Optimal load
frequency control based on hybrid bacterial foraging and particle swarm optimization. In IEEE,
11th International Multi-conference on Systems, Signals and Devices (SSD-PES), Barcelona
(pp. 1–6).
Kouba, N. E. Y., Menaa, M., Hasni, M., Boussahoua, B., & Boudour, M. (2014b). Optimal load
frequency control in interconnected power system using PID controller based on particle swarm
optimization. In IEEE, International Conference on Electrical Sciences and Technology in
Maghreb (CISTEM), Tunis (pp. 1–8).
Kouba, N. E. Y., Menaa, M., Hasni, M., & Boudour, M. (2014c). Application of artificial neural
networks to load frequency control in multi-area power system. In Proceedings of the 3rd
International Conference on Information Processing and Electrical Engineering (ICIPEE),
Tebessa (pp. 357–362).
Kouba, N. E. Y., Menaa, M., Hasni, M., & Boudour, M. (2016a). Design of intelligent load
frequency control strategy using optimal fuzzy-PID controller. International Journal of Process
Systems Engineering, 4, 41–64.
Kouba, N. E. Y., Menaa, M., Hasni, M., & Boudour, M. (2016b). Frequency stability enhancement
in two-area deregulated power system based competitive electricity markets with redox flow
batteries and power flow controllers. In IEEE 8th International Conference on Modelling,
Identification and Control (ICMIC), Algiers (pp. 1029–1036).
Kouba, N. E. Y., Menaa, M., Hasni, M., & Boudour, M. (2016c). LFC enhancement concerning
large wind power integration using new optimised PID controller and RFBS. IET Generation,
Transmission and Distribution, 10, 4065–4077.
Kouba, N. E. Y., Menaa, M., Hasni, M., & Boudour, M. (2016d). A novel optimal frequency control
strategy for an isolated wind-diesel hybrid system with energy storage devices. SAGE, Wind
Engineering, 40, 497–517.
Liu, P., Yang, W.-T., Yang, C.-E., & Hsu, C.-L. (2015). Sensorless wind energy conversion system
maximum power point tracking using Takagi-Sugeno fuzzy cerebellar model articulation
control. Applied Soft Computing, 29, 450–460.
14 Intelligent Load Frequency Control for Wind Power Generation 313

Mahabuba, A., & Khan, M. (2009). Small signal stability enhancement of a multi-machine
power system using robust and adaptive fuzzy neural network-based power system stabilizer.
European Transactions on Electrical Power, 19, 978–1001.
Mandal, P., Zareipour, H., & Rosehart, W. D. (2014). Forecasting aggregated wind power
production of multiple wind farms using hybrid wavelet-PSO-NNS. International Journal of
Energy Research, 28, 1654–1666.
Michigami, T., & Oishi, T. (2001). Construction of dynamic fluctuation load model and simulation
with AFC control of BTB interconnection. Electrical Engineering in Japan, 136, 15–25.
Nanda, J., Mishra, S., & Saikia, L. (2009). Maiden application of bacterial foraging-based
optimization technique in multiarea automatic generation control. IEEE Transactions on Power
Systems, 24, 602–609.
Pan, I., & Das, S. (2015). Fractional-order load-frequency control of interconnected power systems
using chaotic multi-objective optimization. Applied Soft Computing, 29, 328–344.
Panda, S., & Yegireddy, N. (2013). Automatic generation control of multi-area power system
using multi-objective non-dominated sorting genetic algorithm-II. Electrical Power and Energy
Systems, 53, 54–63.
Pandey, S., Mohanty, S., & Kishor, N. (2013). A literature survey on load frequency control for
conventional and distribution generation power systems. Renewable and Sustainable Energy
Reviews, 25, 318–334.
Patnaik, R., & Dash, P. (2015). Impact of wind farms on disturbance detection and classification in
distributed generation using modified adaline network and an adaptive Neuro-fuzzy informa-
tion system. Applied Soft Computing, 30, 549–566.
Pothiya, S., & Ngamroo, I. (2008). Optimal fuzzy logic-based PID controller for load-frequency
control including superconducting magnetic energy storage units. Energy Conversion and
Management, 49, 2833–2838.
Prakash, S., & Sinha, S. (2014). Simulation based Neuro-fuzzy hybrid intelligent PI control
approach in four-area load frequency control of interconnected power system. Applied Soft
Computing, 23, 152–164.
Rahmani, M., & Sadati, N. (2013). Two-level optimal load-frequency control for multi-area power
systems. Electrical Power and Energy Systems, 53, 540–547.
Ramakrishna, K. S. S., & Bhatti, T. S. (2008). Automatic generation control of single area power
system with multi-source power generation. Proceedings of the Institution of Mechanical
Engineers, Part A: Journal of Power and Energy, 222, 1–11.
RamaSudha, K., Vakula, V., & Shanthi, R. V. (2010). PSO based design of robust controller for two
area load frequency control with nonlinearities. International Journal of Engineering Science
and Technology, 2, 1311–1324.
Sahu, R. K., Panda, S., & Padhan, S. (2013). A novel hybrid gravitational search and pattern search
algorithm for load frequency control of nonlinear power system. Applied Soft Computing, 13,
4718–4730.
Saikia, L., & Sahu, S. (2013). Automatic generation control of a combined cycle gas turbine plant
with classical controllers using firefly algorithm. Electrical Power and Energy Systems, 53,
27–33.
Saikia, L., Mishra, S., Sinha, N., & Nanda, J. (2011). Automatic generation control of a multi area
hydrothermal system using reinforced learning neural network controller. Electrical Power and
Energy Systems, 33, 1101–1108.
Sarrias-Mena, W., Fernãndez-Ramírez, L., García-Vázquez, C., & Jurado, F. (2015). Dynamic
evaluation of two configurations for a hybrid DFIG-based wind turbine integrating battery
energy storage system. Wind Energy, 18, 1561–1577.
Schlechtingen, M., Santos, I. F., & Achiche, S. (2013). Wind turbine condition monitoring based on
scada data using normal behavior models. Part 1: System description. Applied Soft Computing,
13, 259–270.
Shayeghi, H., Jalili, A., & Shayanfar, H. (2008). Multi-stage fuzzy load frequency control using
PSO. Energy Conversion and Management, 49, 2570–2580.
314 N. EL. Y. Kouba and M. Boudour

Singh, V., Mohanty, S., Kishor, N., & Ray, P. (2013). Robust H-infinity load frequency control in
hybrid distributed generation system. Electrical Power and Energy Systems, 46, 294–305.
Tofighi, M., Alizadeh, M., Ganjefar, S., & Alizadeh, M. (2015). Direct adaptive power system
stabilizer design using fuzzy wavelet neural network with self-recurrent consequent part.
Applied Soft Computing, 28, 514–526.
Vrakopoulou, M. (2013). Optimal decision making for secure and economic operation of power
systems under uncertainty. Ph.D. thesis, Eidgenössische Technische Hochschule ETH, Zürich.
Wang, C., & McCalley, J. (2013). Impact of wind power on control performance standards.
Electrical Power and Energy Systems, 47, 225–234.
Zhang, W., Wang, J., Wang, J., Zhao, Z., & Tian, M. (2013). Short-term wind speed forecasting
based on a hybrid model. Applied Soft Computing, 13, 3225–3233.
Chapter 15
Fault Tolerant Control of Switch Power
Converter in WECS Based on a DFIG

Amina Tamer, Azeddine Bendiabdellah, Bilal Djamal Eddine Cherif,

and Djillali Toumi

Abstract In the field of Wind Energy Conversion Systems(WECS), the prediction

and early detection of wind power converter failures is one of the most promising
ways to control and optimize the operational costs. This chapter presents the
detection of converter open switch faults and the fault tolerant of power converters
fed Doubly Fed Induction Generator (DFIG) in Wind Energy Conversion Systems
(WECS). The proposed control approach is based on using the grid side converter to
regulate the DC link voltage constant. The mean of the rotor side converter is to track
the maximum power point for the wind turbine and to maintain unity power factor
at stator terminals. The description of the proposed system is presented with the
detailed dynamic modeling equations. The mean value of the rotor current diagnosis
technique is adopted for fault detection. The fault tolerant topology used for service
continuity is that with a redundant leg. The simulations are performed using the
Matlab/Simulink environment and Sim power. The fault operation is analyzed and
the detection method and fault tolerant topology used are tested with encouraging
results.

Keywords DFIG · WECS · Open switch · Detection · Fault tolerant ·

Redundancy

15.1 Introduction

The large industrialization during the last decade has led to considerable needs in
the electrical energy. To meet this high demand, countries are increasingly turning

A. Tamer () · A. Bendiabdellah · B. D. E. Cherif

Laboratory of Electrical Drives Development (LDEE), University of Science and Technology of
Oran, Oran, Algeria
D. Toumi
Laboratory of Electrical and Computer Engineering (L2GEGI), Ibn Khaldun University, Tiaret,
Algeria

© Springer Nature Singapore Pte Ltd. 2019 315

N. Derbel, Q. Zhu (eds.), Modeling, Identification and Control Methods
in Renewable Energy Systems, Green Energy and Technology,
https://doi.org/10.1007/978-981-13-1945-7_15
316 A. Tamer et al.

to the use of clean and renewable energy sources. Indeed, in the medium term, these
countries have committed themselves to increase the share of renewable energy in
their electricity production.
Among the various renewable energy sources, wind energy is considered as the
one with the greatest potential. The power from wind turbines and wind farms
installed in the world is increasing every year. The (WECS) although designed
initially as an active power generator linked to the grid system, it is now required
to provide system services such as reactive power compensation and electric power
quality improvement.
Nowadays, most of the wind turbines installed are equipped with the Doubly Fed
Induction Generator (DFIG), which has become very popular. The DFIG benefits
from certain advantages over all other types with variable speed. It allows operating
over a wide range of wind speed, and drawing the maximum possible power for
each of its speeds. The use of the (DFIG) associated with the (WECS) as a wind
turbine has grown dramatically in recent years. Indeed, the energy converter used
as rectifier-inverter for the alternating currents of the rotor has a nominal power
fractional to that of the generator, which reduces its cost compared to the competing
topologies (Boukhriss et al. 2014; Touati et al. 2014).
The presence of power convertors as key elements in the wind turbine structure
impacts negatively the reliability and may jeopardize the whole system. The power
semiconductor switches are the most fragile components in power converters. If the
(WECS) stops because of a fault in a convertor switch, it may take longer before
it can be repaired. As a consequence, the concepts of reliability and availability
have become increasingly an issue of paramount importance for the (WECS). In
order to guarantee the needed reliability and availability many researchers have
focused their investigation to designing (WECS) to include fault detection and fault
tolerant systems. The fault tolerance has the ability to tolerate faults by detecting
failures and isolating the faulty converter switches so that the rest of the system can
operate correctly and the continuity of service is achieved (Bensouda and Haddi
2014; Polinder et al. 2009; Sarati and Kyeong-Hwa 2014).
Several overview papers have covered the topics on power convertor switch fault
diagnosis methods (Cherif et al. 2016; Lu and Sharma 2009; Mohamed et al. 2013).
In reference (Cherif et al. 2015) the mean value stator currents and measurement
of semiconductor switch currents techniques were introduced and applied to a
three phase voltage inverter fed induction motor, in which the faults are detected
(ElKachani et al. 2014).
In this present chapter, the open-circuit fault in the power converter switches of
the rotor side is investigated. The diagnosis method adopted for the detection and
localization of this fault is based on the mean value of the rotor currents technique.
The fault tolerant power converter topology used for continuity of service is that
with a redundant leg reconfiguration (Guediri and Attous 2015; Riouch and Bachtiri
2014).
The rest of the chapter is organized as follows. Section 15.2 describes the
wind Energy Conversion System (WECS) with (DFIG). Section 15.3 presents a
detailed mathematical modelling of the various parts of the wind system with a
15 Fault Tolerant Control of Switch Power Converter Based on a DFIG 317

Fig. 15.1 Structure diagram of the WECS-DFIG

DFIG. Section 15.4 explains the detection and localization techniques of power
convertor open switch faults. Section 15.6 presents the topology of the fault-tolerant
redundancy reconfiguration. Finally, some conclusion remarks are drawn in the last
Section.

15.2 WECS-DFIG Description and Modelling

The structure of a wind turbine-DFIG set used in this paper is shown in Fig. 15.1.
The structure of the (WECS) is constituted of two main parts; the mechanical
part of the turbine; which is mechanically coupled to the DFIG; is composed of
steerable blades, a gearbox (speed reducer), a coupling shaft system. The electrical
part is composed by a doubly fed induction generator (which is fed from stator and
rotor sides), an electronic power converters chain with its continuous bus. The stator
is directly connected to the grid, whereas the rotor is also connected to the grid, but
through a back-to-back converter via a Pulse Width Modulation (PWM) control
(Tamer et al. 2014). The converter on the rotor side will be referred as the rotor
side converter (RSC), while the converter on the grid side as the grid side converter
(GSC) (Cordeiro et al. 2014). The rotor side converter (RSC) usually provides active
and reactive power control of the machine while the grid-side converter (GSC)
keeps the voltage of the DC-link constant. In this way, the active and reactive stator
power is controlled indirectly by means of the inner rotor current control loop.
The maximum power point tracking (MPPT) control is integrated to ensure the
optimum active power output. The control technique is based on the field oriented
vector, whereas the rotor current is controlled in the stator flux-oriented referential
(Elazzaoui 2015).
318 A. Tamer et al.

15.2.1 Wind Turbine Modelling

A wind turbine converts the wind power to mechanical power. The mechanical
power Pm is given by the following equation:

1 Cp (λ, β) 3 2
Pm = πρ R V1 (15.1)
2 λ
where ρ the air density (kg/m3), R the wind turbine rotor radius (m), V1 the
equivalent wind speed (m/s), β the pitch angle of rotor (deg), Cp the power
coefficient with its maximum value at 0.48 (Betz limit) and λ is the tip speed ratio.
The ratio λ can be expressed by the following equation:

R Ωtur
λ= (15.2)
V1

The power coefficient is intrinsic to the formation of wind turbine and depends
on the profiles of the blades. The power coefficient can be modeled with a single
equation that depends on the speed ratio λ and the orientation angle β of the blades
as follows equation:
   
1 −C5
Cp (λ, β) = C1 C2 − C3 β − C4 exp + C6 λ (15.3)
λi λi

In this formula, the parameter also depends on λi and β.

1 1 0.035
= − (15.4)
λi λ + 0.08 β 1 + β3

Figure 15.2 shows curves of the power coefficient as a function of λ for various β
values. This gives a maximum power coefficient of 0.48 for a speed ratio λ which is
8.1 maintaining β at 0◦ . By setting β and λ respectively to their optimal values, the
wind system will provide optimum electrical power.
This particular value of λ is defined as the optimal value λopt where the maximum
power is captured from wind by the wind turbine. So it is necessary to keep the rotor
speed at an optimum value of the tip speed ratio λopt . If the wind speed varies, the
rotor speed should be adjusted to follow the change. The aerodynamic torque on the
slow axis(speed reducer) is expressed by:

Paero ρ S V13 1
Taero = = Cp (15.5)
Ωtur 2 Ωtur

Mechanical speed is related to the speed of rotation of the turbine by the coefficient
of the gearbox. The torque on the slow axis is connected to the torque on the fast
axis (generator side) by the coefficient of the gearbox (Fig. 15.3).
15 Fault Tolerant Control of Switch Power Converter Based on a DFIG 319

Fig. 15.2 Power coefficient as a function of λ for various β values

Fig. 15.3 Wind turbine modeling

The total inertia J is made up of the inertia of the turbine plotted on the generator
rotor and inertia of the generator:
Jt
J = + Jg (15.6)
G2
Jt : inertia of the turbine.
Jg : inertia of the generator.
The evolution of the mechanical speed from the total mechanical torque Tm is
determined by the fundamental equation of dynamics.
dΩmec
J = Tmec = Tf a − Tem − f Ωmec (15.7)
dt
320 A. Tamer et al.

V1 lcpmaxy1 Ω∗mr 1 Ω∗m Ki T*em

+ Kp +
R G – P
Ωm

Fig. 15.4 MPPT control

Ωm : mechanical speed of DFIG.

Tf a : aerodynamic torque on the fast axis of the turbine.
Tem : Electromagnetic torque.
f : coefficient of friction.

15.2.2 Extraction of Maximum Power MPPT

In order to capture the maximum power of the incident wind, the rotational speed
of the turbine must permanently be adjusted to that of the wind. The optimum
mechanical speed of the turbine corresponds 0◦ . The speed of DFIG is used as a
reference value for a proportional-integral controller type. The latter determines the
set of the command that is the electromagnetic torque that we will apply to the
machine to run the generator at its optimum speed. The torque thus determined by
the regulator is used as reference variable torque model of the turbine (Fig. 15.4).

15.2.3 DFIG Modelling

The doubly fed induction generator is typically modeled in the Park plane, leading
to the following equations (Singh et al. 2010; Tapia et al. 2003):
⎧
⎪
⎪
dφsd
⎪ νsd = Rs isd +
⎪ − ωs φsq
⎪
⎪ dt
⎪
⎪ dφ
⎨ νsq = Rs isq + sq + ωs φsd
dt (15.8)
⎪
⎪ = +
dφrd
−
⎪
⎪ νrd R r i rd ω r φ rq
⎪
⎪ dt
⎪
⎪ dφ
⎩ νrq = Rr irq + rq + ωr φrd
dt
With:
⎧
⎪
⎪ φsd = Ls isd + Msr ird
⎨
φsq = Ls isq + Msr irq
(15.9)
⎪ φrd
⎪ = Lr ird + Msr isd
⎩
φrq = Lr irq + Msr isq
15 Fault Tolerant Control of Switch Power Converter Based on a DFIG 321

The expression of the electromagnetic torque of the DFIG based on the flux and
stator currents is written as follows:
# $
Tem = p φsd isq − φsq isd (15.10)

The active and reactive power of the stator are written as follows
*
Ps = Vds Ids + Vqs Iqs
(15.11)
Qs = Vqs Ids − Vds Iqs

15.3 WECS-DFIG Control Strategy

The control strategy is based on:

• Control of RSC by controlling active and reactive stator power of DFIG.
• Control of GSC by controlling the voltage of the DC bus

15.3.1 RSC Control

The control principle of the rotor side converter (RSC) allows the control of active
and reactive power as well as the extraction of maximum wind power. We will adopt
conventional controllers (PI) necessary to achieve control of reactive and active
power of the DFIG. The control of rotor currents is done by Proportional Integral
(PI) controllers, as shown in Fig. 15.5.

Fig. 15.5 Block diagram of RSC control

322 A. Tamer et al.

Fig. 15.6 Block diagram of GSC control

15.3.2 GSC Control

In the grid side control, mainly voltage of the DC bus is compared with a reference
DC bus voltage and an error is fed to a PI controller to maintain constant DC bus
voltage Fig. 15.6.

15.4 Fault Detection Technique

15.4.1 Open Switch Faults

The converter is the most sensitive subsystem of the (WECS-DFIG) from a

reliability point of view and hence it is the element that the fault detection is the most
required and profitable. Within the converters used in the (WECS-DFIG) structure,
the most common and most critical faults are related in most cases to the control
of the power semiconductor switch. A failure of the power switch may also be due
to the degradation caused by excessive electrical, thermal or mechanical stress. Its
origin can be internal and related to the function itself, or externally and linked to
the environment or to an out of specification use (Ahmed and Zobaa 2016). A 2011
survey (Yang et al. 2011) indicates that the power switches are often the weakest
link in converters, presenting 30% of total failures.
The two main types of switch failures in converters are short-circuit faults and
open circuit switch faults. While short-circuit faults are usually very destructive and
15 Fault Tolerant Control of Switch Power Converter Based on a DFIG 323

require immediate actions to shut down the system, open switch faults do not cause
a stop of the system and can stay undetected for some time. Unfortunately, such
faults may generate secondary problems in the system that can cause failure in other
parts of the system and can eventually lead to high repair costs (Yanhui et al. 2013).
For this reason, open switch faults should be detected as soon as the fault occurs and
properly localized.
In this chapter, only the open switch fault of the converter is considered. Several
papers have been published for the detection and localization of faults. Most
techniques are based on the analysis of output voltages or currents within the system.
In Cherif et al. (2016), the authors suggested the park vector method to detect and
localize the open switch fault. This technique is based on the mean value of the
vector currents of the Park in the α-β plane and the determination of the phase
angle and it is fully described in the following subsection.

15.4.2 Mean Value of the Rotor Current Detection Technique

This technique uses the mean phase current value for fault detection. It is to note
that, a fault in a converter semiconductor switch can produce offsets in the currents
of the electrical machine phases. The principle of this detection technique is to
determine the mean value of the tree mean value phase rotor currents from which the
fault can be detected and localized. A current threshold is usually defined in order to
distinguish between the converter open switch faults. After the introduction of a fault
in the converter power switch, a change can be clearly noticed in the rotor current
waveforms shape and their corresponding means values. This technique enables us
therefore to easily calculate the mean value of the rotor current phases and hence
detect and localize the open switch faults in a converter. Table 15.1 summarizes the
open switch faults characteristics of a power converter rotor side (Tamer et al. 2017).

Table 15.1 Open switch fault characteristics of a power converter rotor side
State Leg A Leg B Leg C
Healthy 0 0 0
K1 open −Ithresh < I amean < 0 I bmean > 0 I cmean > 0
K2 open I amean > 0 −Ithresh < I bmean < 0 I cmean > 0
K3 open I amean > 0 I bmean > 0 −Ithresh < I cmean < 0
K1 open 0 < I amean < Ithresh I bmean < 0 I cmean < 0
K2 open I amean < 0 0 < I bmean < Ithresh I cmean < 0
K3 open I amean < 0 I bmean < 0 0 < I cmean < Ithresh
324 A. Tamer et al.

15.5 Fault Tolerant Reconfiguration in Case

of Open Switch Fault

Several researchers have been engaged in investigating the behaviour of the

static converters during the occurrence of a power semiconductor switch fault as
well as the associated fault tolerance topologies and therefore suggesting various
propositions in order to overcome failures and thus to reduce such problems.
The fault tolerant topology Fig. 15.7 includes a redundant leg connected using
bidirectional voltage and current switches (triacs) to the converter. This redundant
leg is constituted of two power switches K and K  and replaces one of the other
converter legs when a fault occurs on a switch of that leg (Gaillard et al. 2007,
2013; Yanhui et al. 2013).
To ensure the continuity of service of the wind systems in the case of open switch
fault, the following steps should be applied:
• Calculate the mean value of the rotor currents
• Compare it with the threshold current
• Isolate the faulty leg

Fig. 15.7 Fault tolerant converter topology for WECS-DFIG

15 Fault Tolerant Control of Switch Power Converter Based on a DFIG 325

Fig. 15.8 Detection and reconfiguration algorithm of open switch fault

• Reconfigure by closing one of the bidirectional switches (triac) corresponding to

the faulty phase
• At the same time, apply of the order command switches of the redundant leg
• Stop function “Fault detection”.
Figure 15.8 below depicts in a synoptic diagram manner the required steps for
converter open switch fault detection and localization.

15.6 Results Presentation and Discussion

This section presents simulation results of the (WECS-DFIG) control in the

MATLAB-Simulink environment. Firstly, the simulation results of the vector con-
trolled (DFIG) efficiency for normal operation condition, are presented in Fig. 15.9,
it can be seen that the dynamics of the speed (Fig. 15.9b) is fast and it follows
its reference under MPPT control. The active and reactive powers (Fig. 15.9c, d)
control shows a good correlation compared to their references. Similar results for
direct and quadrature rotor currents (Fig. 15.9e, f) are obtained. The DC link voltage
(Fig. 15.9g) is well regulated.
326 A. Tamer et al.

a b
8 120
Wr ref Wr
Wind Speed (m/s)

7.5 110

Speed(rad/s)
7 100

6.5 90

6 80
0 1 2 3 4 0.5 1 1.5 2 2.5 3 3.5 4
time(s) time(s)

c d
5000 5000
Ps Ps ref Qs Qs ref

Reactive power(Var)
Active power(W)

−10000
0.5 1 1.5 2 2.5 3 3.5 4 0.5 1 1.5 2 2.5 3 3.5 4
time(s) time(s)

e f
20 20
Irq Irq ref Ird Ird ref
Quadrature rotor current

10 10
Direct rotor current(A)

0 0

−10 −10

−20 −20

−30 −30

−40 −40
0.5 1 1.5 2 2.5 3 3.5 4 0.5 1 1.5 2 2.5 3 3.5 4
time(s) time(s)

g
570.5
Vdc Vdc ref
Vdc voltage(V)

570

569.5
0.5 1 1.5 2 2.5 3 3.5 4
time(s)

Fig. 15.9 DFIG speed, active and reactive powers, direct and quadrature rotor currents, and VDC
voltage
15 Fault Tolerant Control of Switch Power Converter Based on a DFIG 327

x 104
8
Ps Qs
Active and reactive power
4

−4

−8
0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
time(s)

Fig. 15.10 Active and reactive powers with open switch fault K3

50
ira irb irc

25
Rotor curents(A)

−25

−50
0.95 1 1.05 1.1 1.15
time(s)

Fig. 15.11 Rotor phase current with open switch fault K3

The obtained results without fault detection and with compensation are shown
in the case of an open-circuit fault on switch (K3 ). The active and reactive power
in Fig. 15.10, presents oscillations and fluctuations at the moment of application the
fault (at time = 1 s). It is to note that the current in the phase concerned is zero during
half of the modulation period as seen in Fig. 15.11.
The mean value of the faulty phase rotor current is negative and the other values
are positive as depicted in Fig. 15.12. The command signal sent to the converter
switch K3 is as shown in Fig. 15.13. In the case of an open-circuit fault in the switch
 15 ira mean irb mean irc mean

Mean of rotor curents

5

−5

−15

0.95 1 1.05 1.1 1.15 1.2

time(s)

Fig. 15.12 Mean value of rotor current with open switch fault K3

1
command sent to the switch K3

0
0.95 1 1.05
time(s)

Fig. 15.13 Order command signal to switch K3

50
ira irb irc
Rotor curents(A)

−50
0.95 1 1.05 1.1 1.15
time(s)

Fig. 15.14 Rotor phase current with open switch fault K3
15 Fault Tolerant Control of Switch Power Converter Based on a DFIG 329

15
ira mean irb mean irc mean

Mean of rotor curents

−15
0.95 1 1.05 1.1 1.15
time(s)

Fig. 15.15 Mean value of rotor current with pen switch fault K3

x 104
Qs Ps
Active and reactive power

0.9 1 1.1 1.2 1.3 1.4 1.5

time(s)

Fig. 15.16 Open switch fault compensation in active and reactive powers

of the lower half-leg K3 (that is in the complementary converter switch) as shown
in both Figs. 15.14 and 15.15, it is to notice that at the instant of the occurrence of
the fault and in case of a fault in the switch of the upper half arm see Fig. 15.14, the
mean value of the current of the faulty phase is then positive and the other values
are negative as seen in Fig. 15.15. It can be also noted that with a short detection
time, the converter with redundant leg reconfiguration cancels the existing fault
and ensure the continuity of service of the wind systems as depicted in Figs. 15.16
and 15.17.
330 A. Tamer et al.

40
ira irb irc

courant rotorique (A) 20

−20

−40
0.95 1 1.05 1.1
time(s)

Fig. 15.17 Open switch fault compensation in rotor phase current

15.7 Conclusion

This chapter is addressing the issue of fault detection and fault tolerance reconfig-
uration for an open switch fault in power converter rotor side of a (WECS- DFIG).
First a detection of the open switch fault is performed using the method of the mean
value of the rotor currents technique. The use of this technique can detect the faulty
leg as well as the accurate location of the faulty switch in the converter leg. For
the continuity of service of the (WECS- DFIG), the fault tolerant power converter
reconfiguration is then carried out based on a leg redundancy approach. With the use
of the redundant leg fault tolerant topology, it is shown that the system can continue
to operate without any problems after detection and isolation of the fault with the
same performance as that before the fault occurrence. The performed simulation
results illustrate well the feasibility of the proposed chapter contributions. Due
the importance of the reliability in systems such as the (WECS- DFIG) which is
always associated to power convertors, rapid and efficient detection methods are
to be investigated to accurately localize the faulty converter switch. To ensure the
continuity of (WECS- DFIG) service, more research work for robust fault tolerant
control topologies needs to be investigated.

References

Ahmed, I., & Zobaa, A. (2016). Comparative power quality study of variable speed wind turbines.
International Journal on Energy Conversion, 4(4), 97–104.
Bensouda, F., & Haddi, A. (2014). Introduction on advances in diagnosis for power electronics
devices in wind energy, based on a doubly-fed asynchronous generator. In International
Conference on Green Energy and Environmental Engineering (GEEE), Tunisia.
15 Fault Tolerant Control of Switch Power Converter Based on a DFIG 331

Boukhriss, A., Nasser, T., Essadki, A., & Boualouch, A. (2014). Improved control for DFIG used in
wind energy conversion systems. International Review of Automatic Control (IREACO), 7(4),
403–411.
Cherif, B., Bendjebbar, M., & Bendiabdellah, A. (2015). Diagnostic of open-circuit fault in three
phase voltage inverter fed induction motor. In IEEE Conference on Electrical Engineering,
(ICEE 2015), Boumerdes, Algeria.
Cherif, B., Bendiabdellah, A., & Khelif, M. (2016). Detection of open-circuit fault in a three-phase
voltage inverter fed induction motor. International Review of Automatic Control, 9(6), 374–382.
Cordeiro, A., Palma, J., Maia, J., & Resende, M. (2014). Detection and diagnosis solutions for
fault-tolerant VSI. Journal of Power Electronics, 14(6), 1272–1280.
Elazzaoui, M. (2015). Modeling and control of a wind system based doubly fed induction
generator: Optimization of the power produced. Electrical & Electronic Systems Journal,
4(141), 1–8.
ElKachani, A., Chakir, E., Laachir, A., Jarou, A., Niaaniaa, T., & Sedra, J. A. (2014). AC grid
connected DFIG-based wind turbine with shunt active power filter. International Review on
Modelling and Simulations, 8, 354–361.
Gaillard, A., Karimi, S., Saadate, S., Poure, P., & Gholipour, E. (2007). A fault tolerant converter
topology for wind energy conversion system with doubly fed induction generator. In European
Conference on Power Electronics and Applications, Aalborg, Denmark (pp. 1–6).
Gaillard, A., Poure, P., & Saadate, S. (2013). FPGA-based reconfigurable control for switch fault
tolerant operation of WECS with DFIG without redundancy. Renewable Energy Journal, 55,
35–48.
Guediri, A., & Attous, D. B. (2015). Modeling and fuzzy control of a wind energy system based on
double-fed asynchronous machine for supply of power to the electrical network. International
Journal of System Assurance Engineering and Management, 8(1), 363–360.
Lu, B., & Sharma, S. K. (2009). A literature review of IGBT fault diagnostic and protection
methods for power inverters. IEEE Transactions on Industry Applications, 45, 1770–1777.
Mohamed, S., Haitham, Z., & Ashraf, S. (2013). Open-circuit fault diagnosis of three-phase
induction motor drive systems. Journal of Electrical Engineering, 13, 60–68.
Polinder, H., Lendenmann, H., Chin, R., & Arshad, W. (2009). Fault tolerant generator systems
for wind turbines. In International IEEE Conference on Electric Machines and Drives
(pp. 675–681).
Riouch, T., & Bachtiri, R. (2014). Advanced control strategy of doubly fed induction generator
based wind-turbine during symmetrical grid fault. International Review of Electrical Engineer-
ing, 9(4), 829–834.
Sarati, D. P., & Kyeong-Hwa, H. (2014). Voltage-based on-line fault detection and faulty
switch identification under multiple open-switches in grid-connected wind power converter.
International Journal of Control and Automation, 7(11), 419–434.
Singh, B., Aggarwal, S., & Chandra, T. (2010). Performance of wind energy conversion system
using a doubly fed induc-tion generator for maximum power point tracking. IEEE Transactions,
978, 4244–6395.
Tamer, A., Toumi, D., Bendiabdellah, A., & Boucherit, M. (2014). Inter-turn stator and rotor
fault in doubly-fed induction generator based wind power system. In Second International
Conference on Electrical Engineering and Control Applications ICEECA Constantine, Algeria.
Tamer, A., Toumi, D., Bendiabdellah, A., & Cherif, B. (2017). Open switch fault detection and fault
tolerant of power converter fed DFIG in WECS. International Review of Automatic Control,
10(3), 274.
Tapia, A., Tapia, G., & Ostolaza, J. (2003). Modeling and control of a wind turbine driven doubly
fed induction generator. IEEE, 18(2), 194–204.
Touati, A., Abouloifa, A., Elmounim, E. A., Aboulfatah, M., Majdoul, R., & Moutabir, A. (2014).
Modeling and control of a three phase DC-AC converter connected to the network for a wind
power system equipped with a DFIG. International Review on Modelling and Simulations
(IREMOS), 7(2), 231–237.
332 A. Tamer et al.

Yang, S., Bryant, A., Mawby, P., Xiang, D., Ran, L., & Tavner, P. (2011). An industry-based survey
of reliability in power electronic converters. IEEE Transactions on Industry Applications, 47,
1441–1451.
Yanhui, F., Junwei, Z., Yingning, Q., & Kai, F. (2013). Fault tolerance for wind turbine power
converter. In Renewable Power Generation Conference (RPGC), Beijing, China (pp. 1–4).
Chapter 16
ZedBoard-FPGA Control of the DFIG
Based Wind Power System

Hassane Mahmoudi, Marouane El Azzaoui, and Chafik Ed-Dahmani

Abstract This work deals with the field oriented control of the doubly fed induction
generator (DFIG) based wind energy conversion system using ZedBoard-FPGA.
Most of previous works use very expensive embedded processor to control the
doubly fed like Dspace. Using ZedBoard-FPGA reduces the cost of the complete
system. This is a quick way to create hardware/software for different algorithm
using Simulink. Simulation results show the performance of the field oriented
control of DFIG, and experiment results present an overview of this control using
ZedBoard-FPGA compared to simulation.

Keywords DFIG · ZedBoard · FPGA · FOC · Turbine emulator · HDL

16.1 Introduction

The developed research in this chapter focuses on modeling and the numerical
implementation of control algorithms applied to the doubly fed induction generator
(DFIG).
DFIG is a type of wind turbine that has been the subject of numerous studies in
recent years. It operates at variable wind speed. Its main feature is that it consists
of an asynchronous wound rotor machine and can thus provide active power to the
grid by the stator and rotor. It allows to obtain a good performance over a wide range
of wind speed (Meng et al. 2016). The DFIG is sized so that the power converters
on the rotor are only traversed by 30% of the nominal power. Therefore it use less
expensive converters (Yaramasu and Wu 2017).

H. Mahmoudi () · M. El Azzaoui · C. Ed-Dahmani

Power Electronics and Control Team, Department of Electrical Engineering Mohammadia School
of Engineers, Mohammed V University, Rabat, Morocco
e-mail: [email protected]; [email protected]; chafik.
[email protected]

© Springer Nature Singapore Pte Ltd. 2019 333

N. Derbel, Q. Zhu (eds.), Modeling, Identification and Control Methods
in Renewable Energy Systems, Green Energy and Technology,
https://doi.org/10.1007/978-981-13-1945-7_16
334 H. Mahmoudi et al.

Despite these advantages, it remains a problem: its connection to the network.

The power grid is not always stable. It regularly undergoes voltage deterioration
(overvoltage voltage type). These damages can lead to fault currents in the machine
and thus damage or even destroy the power converters (Amin and Uddin 2017).
The first implementations of control algorithms for electrical machines were
realized with analog solutions. These solutions ensure the execution of controls
having a high bandwidth and a high resolution. However, they lacked reliability
due to their sensitivity to perturbations and variations in control parameters related
to the thermal stresses of analog circuits. To remedy these drawbacks, digital
solutions naturally imposed themselves. The first numerical implementations of
control algorithms for electrical machines were performed using microcontrollers,
microprocessors and DSPs (Digital Signal Processor), which also solve problems
related to the use of analog controls. On the other hand, in spite of the advantages
offered by digital solutions, some advantages offered by analogue installations are
lost. This is mainly due to the fact that the discretization and the quantification
of the control algorithms to be implemented, as well as the calculation time
delays, deteriorate the control performance in terms of correction speed and control
resolution.
With technological advances in the field of microelectronics, new hardware
design solutions, such as Field Programmable Gate Array (FPGA) or Application
Specific Integrated Circuit (ASIC) such as ZedBoard, are available and can be
used as digital targets for the implementation of control algorithms (Azzaoui et al.
2016c).
ZedBoard is a low-cost development board for the Xilinx Zynq-7000 all pro-
grammable SoC (AP SoC). This board contains everything necessary to create
several application. Additionally, several expansion connectors expose the process-
ing system and programmable logic I/Os for easy user access. Take advantage
of the Zynq-7000 AP SoCs tightly coupled ARM processing system and 7-series
programmable logic to create unique and powerful designs with the ZedBoard.
Target applications include motor control.
In the following lines a field oriented control strategy of the doubly fed induction
generator is studied and implemented using ZedBoard FPGA.

16.2 Mathematical Model of Wind Energy Conversion

System (WECS)

Wind energy system is composed from different mechanical and electrical com-
ponents such as blades, gearbox, generator, power converter and more. . . In the
following, all these parts are described with their mathematical model, and in the
end of this section, a different controls strategies of the system are depicted.
16 ZedBoard-FPGA Control of the DFIG Based Wind Power System 335

Wind Blade

Gearbox
Grid
isa

isb
DFIG isc

Pitch
ira RSC GSC
irb
irc

Fig. 16.1 DFIG based wind energy conversion system

Figure 16.1 illustrates a classical structure of the DFIG on variable wind turbine.
The DFIG’s stator is connected directly to the grid, and its rotor is connected to the
grid via a power electronics system. The DFIG is powered by the grid voltage and
the voltage in the out of the power electronics system which is having a variable
frequency linked to the rotor of the generator. The power electronic system must be
bidirectional to feed the generator or to generate the power to the grid according
to the mode of operation (hypo or hyper synchrone). This is achieved by using a
back-to-back converter which is composed from : rotor side converter (RSC), grid
side converter (GSC), and DC bus voltage. GSC is connected to the grid through RL
filter to decrease the harmonics generated using the converters.
The Mechanical part is composed from orientable blades (variable pitch),
gearbox to increase the velocity, and the shaft allowing the transmission of the
power to the generator. In the next, we will develop the mathematical model of
the important parts.

16.2.1 Blade Model

Each turbine is characterized by its energy efficiency. This efficiency is the ratio
between the power extracted and the power contained in the wind. This ratio is
called power coefficient and it’s limited to 0.5926 according to Betz. Further, the
power extracted is expressed as following (Azzaoui et al. 2016b):

1
Pt = Cp (λ, β)ρRπ 2 v 3 (16.1)
2
336 H. Mahmoudi et al.

There is another expression of this power that can be writing as function of

mechanical velocity and turbine torque:

Pt = Ωt Tt (16.2)

From this two last equation the expression of the torque is deduced:

1
Tt = Cp (λ, β)ρRπ 2 v 3 (16.3)
2Ωt

λ ratio is defined as the fraction between the linear speed of the turbine Ωt and the
speed of the wind v, its expression is given by Abulanwar et al. (2013):

Ωt R
λ= (16.4)
v
with:
Ωt : Mechanical velocity of the turbine (rad/s).
ρ: Air density ρ = 1.225 kg/m3 .
R: Raduis of the turbine (m).
v: Wind speed (m/s).
β: Pitch angle (degree).
In this study, we consider a power coefficient expressed by this equation:
 
π(λ + 0.1)
Cp (λ, β) = (0.4 − 0.0167β) sin − 0.00184β(λ − 3) (16.5)
14.34 − 0.2β

Figure 16.2 shows the curves of power coefficient as function of λ and β. We get a
maximal Cp = 0.4 at λ = 7.

16.2.2 Gearbox Model

The Gearbox that is located between the turbine and the generator has the purpose
to adopt the slow speed of the turbine to the fast of the generator. It’s described by
the following two equations:

1
Ωt = Ωm (16.6)
G
1
Tm = Tt (16.7)
G
16 ZedBoard-FPGA Control of the DFIG Based Wind Power System 337

0.4 b = 0°
b = 2°
0.35 b = 4°
Power Coefficient Cp
0.3 b = 6°
b = 8°
0.25
b = 10°
0.2
0.15
0.1
0.05
0
0 2 4 6 8 10 12 14
λ ratio

Fig. 16.2 Curves of power coefficient

where:
Ωm : Mechanical velocity of the generator.
Tm : Torque applied to the generator.
G: Gain of gearbox.

16.2.3 Shaft Model

By applying the fundamental equation of dynamics, the dynamic of mechanical

velocity of the generator is given by (Fig. 16.3):

dΩm
J = Tm − Tem − f Ωm (16.8)
dt
With:
Tem : Electromagnetic torque of the DFIG.
J : Moment of inertia total.
f : Coefficient of viscosity total.
By grouping the blade, gearbox, and shaft models, we can construct the block
diagram of the turbine is presented in Fig. 16.3.

16.2.4 DFIG Model

Modeling of the DFIG in stationary frame reference using park transformation is

presented by Shehata (2015):
338 H. Mahmoudi et al.

Blade Gearbox Shaft

λ R Ωt Ωt 1
λ=
v G
Cp

v 1 .C (λ, β).ρ.π.R2.υ3 Tt 1 Tm 1 Ωm
p
2Ωt G Js + f

Tem

Fig. 16.3 Block diagram of turbine

• Voltage equations:

dϕsd
Vsd = Rs Isd + − ωs ϕsq (16.9)
dt
dϕsq
Vsq = Rs Isq + + ωs ϕsd (16.10)
dt
dϕrd
Vrd = Rr Ird + − ωr ϕrq (16.11)
dt
dϕrq
Vrq = Rr Irq + + ωr ϕrd (16.12)
dt
• Flux equations:

ϕsd = Ls Isd + Lm Ird (16.13)

ϕsq = Ls Isq + Lm Irq (16.14)

ϕrd = Lr Ird + Lm Isd (16.15)

ϕrq = Lr Irq + Lm Isq (16.16)

• Frequencies equations:

ωr = ωs − pΩ (16.17)

• Powers equations:

Ps = Vsd Isd + Vsq Isq (16.18)

16 ZedBoard-FPGA Control of the DFIG Based Wind Power System 339

Qs = Vsq Isd − Vsd Isq (16.19)

Pr = Vrd Ird + Vrq Irq (16.20)

Qr = Vrq Ird − Vrd Irq (16.21)

16.3 WECS Control Strategies

16.3.1 MPPT Control

This control structure is based on the assumption that the wind speed varies very
little in steady state. In this case, from the dynamic equation of the turbine, we
obtain the static equation describing the standing turbine speed (El Azzaoui 2015):

dΩm
J = 0 = Tmec = Tf a − Tem − f Ωm (16.22)
dt
The electromagnetic torque becomes:

∗ 1
Tem = Cpmax ρSv 3 − f Ωm (16.23)
2GΩt

Assuming that the orientation angle β of the blades is constant, the wind speed
can be written as follows:
RΩt
v= (16.24)
λopt

The electromagnetic torque reference will be expressed as:

∗ 1
Tem = Cpmax ρπ R 5 Ωm
2
− f Ωm (16.25)
2G3 λ3opt

Figure 16.4 present the block diagram of MPPT without control of speed.

16.3.2 Pitch Control

Pitch control regulates the power generated by the turbine by the nominal electrical
output of the DFIG. This control allows the automatic changing of the blades angle
over the maximal generator speed, therefore protects it before overloading at high
wind speeds. The optimal angle for the wind speed below the nominal value is
340 H. Mahmoudi et al.

λ R Ωt Ωt 1
λ=
v G
Cp

v 1 .ρ.π.R2.C (λ, β).υ3 Tt 1 Tg 1 Ω

p
2Ωt G Js + f
∗
Tem

1
Cpmax ρπR 5 Ω2 − f Ω
2G3 λ 3opt

Fig. 16.4 MPPT block diagram without control of the speed

Fig. 16.5 Model of pitch control in Simulink

approximately zero and then it increases with the wind speed growing. It have
considerable impact on the performance coefficient and on the value of the turbine
torque. During the modelling of this controller, it is very important to take into
account that changes of the blades angle must be fluent with a finite rate, maximal
10◦ per second, depending on the size of the wind turbine. The structure of the
aerodynamic controller in the Simulink, which is shown in Fig. 16.5, is relatively
simple.
The basic parameters set is formed by the nominal generator speed, the rate of
pitch angle and by the maximal and minimal value of this angle. The input value is
the speed of the generator, which is during the simulation compared with its nominal
value and the output is the value of the pitch angle of the blades

16.3.3 DFIG Control

To control active and reactive power independently, field oriented is applied by

aligning the stator flux with d-axis, this make quadratic flux equal to zero. In this
stage, assuming that the grid is stable, and neglecting stator resistance, we get the
simplified expression of stator voltages (Amrane et al. 2017).
16 ZedBoard-FPGA Control of the DFIG Based Wind Power System 341

ϕsd = ϕs ; ϕsq = 0 (16.26)

Vsd = 0 ; Vsq = Vs = ωs ϕs (16.27)

Consider the above assumption, rotor voltages are written as:

dIrd
Vrd = Rr Ird + σ Lr − gσ Lr ωs Irq (16.28)
dt
dIrq Vs Lm
Vrq = Rr Irq + σ Lr + gσ Lr ωs Ird + g (16.29)
dt Ls

Expressions of stator powers become:

Vs Lm
Ps = − Irq (16.30)
Ls

pLm
Tem = (ϕsq Ird − ϕsd Irq ) (16.31)
Ls

Vs Lm V2
Qs = − Ird + s (16.32)
Ls ωs Ls

16.3.3.1 Rotor Side Converter Control

The control of the rotor side converter (RSC) enables us to control the stator active
and reactive powers independently. From Eqs. (16.30) and (16.32), it’s clear that the
active and reactive powers are based on the q and d axis rotor currents respectively.
Therefore, the powers are checked by controlling the rotor currents. These currents
are controlled by a PI controllers. Then, we must add terms of compensation and
decoupling. The voltages obtained are transformed to abc frame using dq to abc
transformation which the angle is the difference between the stator angle obtained
using a PLL and the rotor angle (Azzaoui et al. 2016d). Finally, the abc results
voltages are converted to PWM signals by comparing these voltages with a carrier
signal in order to control the gates of IGBTs used in the rotor side converter (RSC)
as set forth in Fig. 16.6.

16.3.3.2 Grid Side Converter Control

The aim of controlling the grid side converter (GSC) is to produce a unitary power
factor in the connection node of the grid, and to regulate the DC bus voltage of the
back-to-back converter to a constant value. As presented in Fig. 16.7, a PI controller
is used to control the d-axis current which its reference is kept null to have a unitary
342 H. Mahmoudi et al.

LmVsq
r ( Lr I rd )
s Ls
ref
Tref
em
I rq V rqref
MPPT Eq. (8) PI dq Vra
I rq
Lr I
r rq
ref ref V rdref
Vrb
Q s I rd
Eq. (9) PI
Vrc
I rd
r
abc

carrier signal
dq ira
irb
irc
abc r

r DFIG
p

s
vsa
PLL vsb
vsc

Grid

Fig. 16.6 Rotor side converter control using PI

power factor. DC bus voltage is regulated using two control loops, an inner loop
that control the q-axis current using a PI controller, and an outer loop that regulate
the DC bus voltage using a second PI controller. Reference results voltages are
transformed in abc frame using the dq to abc transformation using stator angle.
Finally, the abc results voltages are converted to PWM signals by comparing these
voltages with a carrier signal in order to control the gates of IGBTs used in the grid
side converter (GSC).

16.3.4 Simulation Results

The entire wind energy conversion system is simulated in Matlab/Simulink environ-

ment as shown in Fig. 16.8 where the parameters of DFIG in DC motor are given
in “Appendix”. The reference DC bus voltage is set at 600 V. By cons, we will
vary the stator reactive power by adjusting its reference value in the RSC control.
8 kHz pulses width modulation (PWM) signals are generated with Space vector
modulation (SVM).
Figure 16.9a shows the wind speed used in simulation, while Fig. 16.9b presents
the mechanical speed derived from the turbine. Figure 16.10a, b illustrate that
both electromagnetic torque and stator reactive power follow their references.
16 ZedBoard-FPGA Control of the DFIG Based Wind Power System 343

Vdc

∗ ∗
∗ Ifq Vfq
Vdc
PI PI dq

∗ ∗ ∗
Qf Current Ifd
Vfd
calculation PI

qs
abc

carrier signal

dq ifa
ifb
ifc
abc qs

Grid

Fig. 16.7 Grid side converter control using PI

Figure 16.11a, b show the power generated by the turbine and the stator power pro-
duced by the DFIG. Figure 16.12a shows that the power coefficient is equal to it opti-
mal values, thus guaranteeing the operating in MPPT mode, and Fig. 16.12b presents
that the DC bus voltage is regulated perfectly to 600 V. Finally, Fig. 16.13a, b show
the curves of pith angle and speed ratio respectively. While the power exceeds its
nominal value (3 kW), the pitch increase to adjust this power to its nominal value.

16.4 Laboratory Experiment

This section covers the final laboratory setup, based upon the components covered
in the previous section.

16.4.1 Laboratory Setup

Figure 16.14 shows the block diagram of laboratory setup. The DFIG is driven
by a DC motor that emulates the mechanical part of the system (turbine). DC
motor excitation is supplied by a DC voltage source while the armature is power
via a dc chopper (in our case the seven arm in SomiKron Inverter) from the
rectifier. Stator of DFIG is linked directly to the grid and its rotor is linked to the
grid through the inverter and the rectifier cause in our study the DFIG is driven
344 H. Mahmoudi et al.

Fig. 16.8 WECS based DFIG in Simulink

in hypo-synchronous mode. Therefore in this case there will be no need for an

independent voltage source to power the armature of DC motor. thus the inverter
is replaced by the rectifier to simplify the control of the entire system. ZedBoard
FPGA control the system via two board; the first one is the acquisition which
acquire currents and voltages sensors and converts this information into digital
format using 8 channels MCP3208 from microchip. The second one allows adapting
PWM signal outgoing from 3.3 to 15 V needs to control the inverter and chopper. In
Final step, the ZedBoard communicate to Host computer through Ethernet protocol
to monitor data using external mode of simulink. Figure 16.15 presents Matalb
hardware/software workflow for ZedBoard. In this workflow, we create model in
Simulink that represent a complete dynamic system, and rapidly design hardware-
software implementations for ZedBoard directly from the this model. This approach
is structured as follow:
• developers used simulation in Simulink to create models for entire to evaluate
design concepts, make high-level tradeoffs, and partition algorithms into soft-
ware and hardware elements.
16 ZedBoard-FPGA Control of the DFIG Based Wind Power System 345

a b
14 1500

12

Mechanical speed (rpm)

10
Wind speed (m/s)

1000
8

6
500
4

0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Time (s) Time (s)

Fig. 16.9 Evolution of (a) wind speed, (b) mechanical velocity

a b
20 4000
15 Reference Reference
Electromaghnitic torque (N.m)

3000
Stator reactive power (Var)

Actual Actual
10
2000
5
0 1000

-5 0
-10
-1000
-15
-2000
-20
-25 -3000
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Time (s) Time (s)

Fig. 16.10 Evolution of (a) electromagnitic torque, (b) stator reactive power

• HDL code generation from Simulink enabled the creation of IP cores and high-
speed I/O processing on the fabric of the ZedBoard.
• C/C++ code generation from Simulink enabled the programming of the ARM
Cortex-A9 cores in ZedBord, supporting rapid embedded software iteration.
• Automatic generation of the AXI4 interfaces linking the ARM processing system
and programmable logic with support for Zynq in ZedBoard.
• Integration with downstream tasks, such as C/C++ compilation and building the
executable for the ARM, bitstream generation using Xilinx implementation tools
and downloading to ZedBoard development boards allows for a rapid prototyping
workflow.
This IP core is designed to be connected to an embedded processor with an AXI4
interface. The processor acts as master, and the IP core acts as slave. By accessing
the generated registers via the AXI4 interface, the processor can control the IP core,
and read and write data from and to the IP core (Fig. 16.16).
Figure 16.17 provides an overview of the laboratory setup.
Laboratory setup is composed from the following parts:
346 H. Mahmoudi et al.

a b
3500 3000

3000 2000

Stator active power (W)

2500
Turbine power (W)

1000
2000
0
1500
-1000
1000
-2000
500

0 -3000

-500 -4000
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Time (s) Time (s)

Fig. 16.11 Evolution of (a) turbine power, (b) stator active power

a b
0.5 800
0.4 700
0.3
Power coefficient Cp

600
DC bus voltage (V)

0.2
500
0.1
400
0
300
-0.1
-0.2 200

-0.3 100

-0.4 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Time (s) Time (s)

Fig. 16.12 Evolution of (a) power coefficient, (b) DC bus voltage

1. 3 kW Doubly fed induction generator (DFIIG) DELORENZO brand.

2. 3 kW DC motor DELORENZO brand.
3. Host computer.
4. 4 kW SomiKron inverter/rectifier which contains seven arms IGBT, the first six
are used as inverter, and the last one is used as chopper in our study, and six
diodes which are used as non-controlled rectifier
5. ZedBoard FPGA which is a complete development kit for designers interested
in exploring designs using the Xilinx Zynq-7000 All Programmable SoC. The
board contains all the necessary interfaces and supporting functions to enable a
wide range of applications. The expandability features of the board make it ideal
for rapid prototyping and proof-of-concept development.
6. Power board that generates three levels of voltages +15, 0, and −15 V, which are
used in adaptation and acquisition board.
7. Adaptation board convert the level of voltage from 3.3 V in the out of ZedBoard
FPGA to 15 V used by the inverter.
16 ZedBoard-FPGA Control of the DFIG Based Wind Power System 347

a b
3.5 10

3
8
Pitch Angle (degree)

2.5

Speed ratio λ
2 6

1.5 4
1
2
0.5

0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Time (s) Time speed (s)

Fig. 16.13 Evolution of (a) power coefficient, (b) DC bus voltage

8. Acquisition Board that contains currents and voltages sensors, and ADC to
convert analog data to digital, in our case, we use 8 channels MCP3208 from
Microchip.
9. Incremental encoder Omron E2b2 CWZ1x brand provides 2000 pulses per
revolution (PPR).

16.4.2 Turbine Emulator

In our laboratory there is no turbine, so, we decide to emulate it with a DC

motor, cause it’s simple to control it using a PI controller. Once the DC machine
control sets, must be added the turbine previously studied, this collection emulates
the mechanical part of the wind turbine, the synopsis of the control is shown in
Fig. 16.18.
The turbine has in output the mechanical speed which is considered as rotational
speed reference whose DC machine is to be driven. The purpose of DC motor
control is the regulation of its speed to follow the speed of the turbine. A simple
PI controller is used to control the DC motor as presented in Fig. 16.19. There are
many methods to synthesize the parameters of PI as poles placement presented in
this work (Azzaoui et al. 2016a).
Figure 16.20 presents the emulator program designed in Simulink and imple-
mented in Zedboard-FPGA. The main part of the program is the PI controller that
regulates mechanical speed of DC motor to mechanical speed in the out of the
turbine to emulates this last. MPPT block allow to follow the maximal operating
point and therefore to extract the maximal of power from kinetic energy of wind.
Pitch control is used here to for orienting angle β of the blades, so as, not to exceed
the maximum wind speed limit. There is also a encoder demodulation block that
allow to acquire the rotational speed. Finally, the out of PI block is modulated with
8kHz triangular wave to generate the pulse width modulation that used to control
the input chopper of the SomiKron inverter.
348 H. Mahmoudi et al.

Fig. 16.14 Block diagram of test bench

The previous program is flashed into the ZedBoard, and all connection with the
DC motor and chopper are made. Figure 16.21 shows that the mechanical speed of
the DC motor follows exactly the mechanical speed imposed by the turbine. Hence,
this test of control presents a good performance in term of tracking.

16.4.3 Experiment Results

After setup and implementing the laboratory setup. We did the parameters iden-
tifications of DFIG and DC motor. And we add the field oriented control studied
to previous turbine emulator. The model that generate the IP cores is presented in
Fig. 16.22. We did our test in hypo-synchronous mode as noted previously, and we
16 ZedBoard-FPGA Control of the DFIG Based Wind Power System 349

Fig. 16.15 Workflow Matlab and ZedBoard

Fig. 16.16 Generation of IP core

notice that a start-up approach is necessary before controlling the DFIG in normal
mode. The start-up process consists of two steps that are: calibration of the encoder
and grid voltage synchronization as described in these works (Abad et al. 2011;
Abo-Khalil 2012; Tapia et al. 2009). ZedBoard generate 8 kHz PWM to control both
inverted and chopper. Using external mode interface in simulink the parameters of
PI are tuning.
The next plots shows the experiment evolution of different measures.
350 H. Mahmoudi et al.

Fig. 16.17 Overview of test bench

Cp Ωt .R
v λ=
v
Cp Tt 1 Ωm
π
Tt = Cp (λ, β) ρR3 v 2 Js + f
2Gλ

Tem

Ωt
Duty
Ωm Ua Cycle
1
G M Buck Converter Speed control

Feedback speed

Fig. 16.18 Synopsis of the control of DC motor

Fig. 16.19 Block diagram of

DC machine control Ωm Ua Ω
PI DC motor

Speed regulator

Figure 16.23a, b present the profile of wind speed and the mechanical speed
of the DFIG. the waveform of the mechanical speed brings together that of wind
speed. Electromagnetic torque and stator reactive power follows their references
as shown in Fig. 16.24a, b. We observe that the power produced has the same wave
with tubine power as depicted in Fig. 16.25a, b. Figure 16.26a shows that the system
operates in MPPT mode. Finally, Fig. 16.26b shows that the pitch is triggered so as
not to exceed the nominal value of the power. Generally, the experimental results are
16 ZedBoard-FPGA Control of the DFIG Based Wind Power System 351

Fig. 16.20 Simulink model of turbine emulator

160
Reference
Measured
140
Mechanical speed (rad/s)

120

100

80

60

40

20

0
0 10 20 30 40 50 60
Time (s)

Fig. 16.21 Block diagram for the control of DC machine

somewhat identical to the results of the simulation except some noise due to errors
of measurements.

16.5 Conclusion

The aim of this work was to improve the performance of the doubly fed induction
generator and to analyze and highlight the contribution of ZedBoard-FPGAs in the
control of electrical machines. To this end, the work presented in this report was
mainly concerned with the modeling and implementation on the ZedBoard-FPGA
target of control algorithms of doubly fed induction generator. In order to reduce
352 H. Mahmoudi et al.

Fig. 16.22 Block diagram for the control of DC machine

a b
14 1500

12
Mechanical speed (rpm)

10
Wind speed (m/s)

1000
8

6
500
4

0 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Time (s) Time (s)

Fig. 16.23 Evolution of (a) wind speed, (b) mechanical velocity

the complexity of using FPGA solutions, a methodical and less intuitive approach
was used. This approach is based on the appropriate development methodology
that allows to answer the different constraints of the architecture of the control
algorithms. In addition, this approach enabled the implementation of a specific
library of reusable hardware architectures dedicated to the control of electrical
machines.
a b
40 4000
Reference Reference
Electromagnitic torque (N.m)

30

Stator reactive power (Var)

Measured 2000 Measured
20

10 0

0 -2000
-10
-4000
-20

-30 -6000
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Time (s) Time (s)

Fig. 16.24 Evolution of (a) electromagnitic torque, (b) stator reactive power

a b
3500 4000

3000 3000
Stator active power (W)

2500 2000
Turbine power (W)

2000 1000

1500 0

1000 -1000

500 -2000

0 -3000

-500 -4000
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Time (s) Time (s)

Fig. 16.25 Evolution of (a) turbine power, (b) stator active power

a b
0.5 3.5
0.4
3
0.3
Power coefficient Cp

Pitch angle (degree)

2.5
0.2
0.1 2

0 1.5
-0.1
1
-0.2
0.5
-0.3
-0.4 0
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Time (s) Time (s)

Fig. 16.26 Evolution of (a) power coefficient, (b) DC bus voltage

354 H. Mahmoudi et al.

Acknowledgements The authors would like to thank all researchers and professors of the
electrical department of Mohammadia School of Engineers.

Appendix

Table 16.1 DFIG nominal Designation Value

parameters
Rated power 3 kW
Rated stator voltage 220/380 V
Nominal frequency 50 Hz
Number of pole pairs p=2
Rotor resistance Rr = 0.44 Ω
Stator resistance Rs = 1.77 Ω
Rotor inductance Lr = 0.138 H
Stator inductance Ls = 0.309 H
Mutual inductance Lm = 0.103 H
Moment of inertia J = 0,017 kg.m2
Damping coefficient f = 0.0073 N.m.s.rd−1

Table 16.2 DC motor Designation Value

nominal parameters
Rated power 3 kW
Armature resistance Ra = 1.5 Ω
Armature inductance La = 0.9 H
Coefficient ke ke = 0.3
Moment of inertia Jdc = 0.021 kg.m2
Moment of inertia fdc = 0.004 N.m.s.rd−1

References

Abad, G., Lopez, J., Rodriguez, M. A., Marroyo, L., & Iwanski, G. (2011). Doubly fed induction
machine modeling and control for wind energy generation. Canada: Wiley.
Abo-Khalil, A. G. (2012). Synchronization of DFIG output voltage to utility grid in wind power
system. Renewable Energy, 44, 193–198.
Abulanwar, S., Chen, Z., & Bak-Jensen, B. (2013). Study of DFIG wind turbine fault ride-through
according to the Danish grid code. In 2013 IEEE Power Energy Society General Meeting,
Vancouver, BC, Canada (pp. 1–5).
Amin, I. K., & Uddin, M. N. (2017). MPPT based efficiently controlled DFIG for wind energy
conversion system. In 2017 IEEE International Electric Machines and Drives Conference
(IEMDC), Miami, FL, USA (pp. 1–6).
16 ZedBoard-FPGA Control of the DFIG Based Wind Power System 355

Amrane, F., Chaiba, A., Francois, B., & Babes, B. (2017). Experimental design of stand-alone
field oriented control for WECS in variable speed DFIG-based on hysteresis current controller.
In 2017 15th International Conference on Electrical Machines, Drives and Power Systems
(ELMA), Sofia, Bulgaria (pp. 304–308).
Azzaoui, M. E., Mahmoudi, H., Bossoufi, B., & Ghamrasni, M. E. (2016a). Emulation of wind
conversion chain equipped with a doubly fed induction generator. In 2016 International
Symposium on Fundamentals of Electrical Engineering (ISFEE), Bucharest, Romania (pp. 1–
6).
Azzaoui, M. E., Mahmoudi, H., & Ed-dahmani, C. (2016b). Backstepping control of a doubly
fed induction generator integrated to wind power system. In 2016 International Conference on
Electrical and Information Technologies (ICEIT), Tangier, Morocco (pp. 306–311).
Azzaoui, M. E., Mahmoudi, H., & Ed-Dahmani, C. (2016c). Backstepping control of the doubly
fed induction generator using Xilinx system generator for implementation on FPGA. In 2016
5th International Conference on Multimedia Computing and Systems (ICMCS), Marrakech,
Morocco (pp. 599–604).
Azzaoui, M. E., Mahmoudi, H., Ed-Dahmani, C., & Boudaraia, K. (2016d). Comparing perfor-
mance of PI and sliding mode in control of grid connected doubly fed induction generator.
In 2016 International Renewable and Sustainable Energy Conference (IRSEC), Marrakech,
Morocco (pp. 769–774).
El Azzaoui, M., & Mahmoudi, H. (2015). Modeling and control of a doubly fed induction generator
base wind turbine system optimizition of the power. Journal of Theoretical and Applied
Information Technology, 80(2), 304–314.
Meng, W., Yang, Q., & Sun, Y. (2016). Guaranteed performance control of DFIG variable-speed
wind turbines. IEEE Transactions on Control Systems Technology, 24(6), 2215–2223.
Shehata, E. (2015). Sliding mode direct power control of RSC for DFIGS driven by variable speed
wind turbines. Alexandria Engineering Journal 54, 1067–1075. Elsevier.
Tapia, G., Santamaría, G., Telleria, M., & Susperregui, A. (2009). Methodology for smooth
connection of doubly fed induction generators to the grid. IEEE Transactions on Energy
Conversion, 24(4), 959–971.
Yaramasu, V., & Wu, B. (2017). Control of DFIG WECS with voltage source converters (pp. 512–).
Hoboken: Wiley/IEEE Press.
Chapter 17
An FPGA-Based Control of the PMSG
on Variable Wind Speed Turbine

Hassane Mahmoudi, Chafik Ed-Dahmani, and Marouane El Azzaoui

Abstract This chapter presents a study of variable wind speed turbine based on
Permanent Magnet Synchronous Generator (PMSG) and turbine emulation system
by a DC motor with experimental validation. The proposed controls of maximum
power point tracking and pitch angle controls enables to extract the maximum
power from wind. A field oriented control (FOC) technique was implemented for
generator side converter. A new concept of controlling based on Matlab-Simulink
for developing compatible controllers with HDL coder in order to generate a VHDL
code, will be implemented on System on Chip (SoC) FPGA, the Xilinx Zynq-7000.
Simulation and experimental results concerning proposed controls of the Wind
Energy Conversion System (WECS) are presented and compared.

Keywords FPGA · PMSG · Wind energy conversion system · Vector control ·

VHDL · Turbine emulation · Matlab-Simulink · DC motor · HDL coder

17.1 Introduction

In recent years, the new interest related to wind energy results from two concerns,
firstly, the environmental view that is a clean energy. Secondly, the development
of wind turbines and power electronic technologies makes the conversion of the
wind kinetic energy more efficient and therefore its use becomes economically
competitive with other conventional sources. The wind turbines can be classified
into fixed-speed and variable turbine (Blaabjerg and Chen 2006; Blaabjerg and Ma
2013).

H. Mahmoudi · C. Ed-Dahmani () · M. El Azzaoui

Power Electronics and Control Team, Department of Electrical Engineering Mohammadia School
of Engineers, Mohammed V University, Rabat, Morocco
e-mail: [email protected]; [email protected]; marouane.
[email protected]

© Springer Nature Singapore Pte Ltd. 2019 357

N. Derbel, Q. Zhu (eds.), Modeling, Identification and Control Methods
in Renewable Energy Systems, Green Energy and Technology,
https://doi.org/10.1007/978-981-13-1945-7_17
358 H. Mahmoudi et al.

Comparing to fixed speed, the variable-speed turbine presents a several advan-

tages including increased wind energy output, improved quality, and reduced
mechanical stress (Ackermann and Söder 2005). But, it presents also drawbacks
such as a high cost of manufacturing and prower losses due to use of power
converters, and a complexity of the system. However, the drawbacks can be
compensated by the additional energy production.
The converter system enables the control of the generator speed that is mechani-
cally coupled with the wind turbine shaft through the gearbox system if needed (Wu
et al. 2011a).
Different generators have been used in wind energy systems over the past years
(Boldea 2005; Polinder et al. 2006). The Permanent Magnet Synchronous Generator
PMSG and Doubly Fed Induction Generator DFIG are widely used for the new
wind turbines. PMSG is normally controlled by a full-capacity power converter
which presents drawbacks comparing the DFIG (around 30% of total power), but
it is well suited for direct drive wind turbines, and is used for the high power rating
up to 7.5 MW. PMSG is known for their high efficiency and power density. Also,
the gearbox system can be eliminated for multi-poles PMSG architecture, which
can minimize the installation and maintenance costs, and mechanical disturbances.
Therefore, the PMSG is considered as a promising candidate for new designs in
Wind Energy Conversion Systems WECS (Pena et al. 2013).
The presented topology is a subject to a number of control structures such as
the Maximum Power point Tracking control for maintaining the desired tip speed
ratio λ, the pitch angle control for the safety of WECS in strong wind and to limit
the power at the rated value, and the generator control to maintain the desired rotor
velocity (Busca et al. 2010). The Generator Side Converter GSC, a rectifier, will
be controlled by the Field Orientation Control strategy (FOC) using space vector
modulation (SVM) technique, this method allows a precise speed and torque control
with minimal losses while reducing the torque ripple (Rogers et al. 2016).
The control algorithm GSC, needs to be implemented, the motor control
technology was dominated by the Digital Signal Processor (DSPs), but several
research has been using the Field Programmable Gate Arrays FPGA for driver
controls (Azzaoui et al. 2016; Kocur et al. 2014; Kung et al. 2016; Mohammadi
et al. 2016; Monmasson et al. 2011). Hardware solutions such as FPGA have
advantages in price, execution speed, large processing capability and flexibility but
are restricted for their poor calculation ability. The FPGA used in this chapter,was
a System on Chip (SoC) device, that include a powerful general purpose processors
and FPGA in the same chip, which is suitable for Real-Time (RT) simulation of
Electromechanical system (Tormo et al. 2016). The zynq-7000 allows to build in
one single device a complete system mixing soft and hardware functionalities which
increases integration in the test bench.
This chapter is organized as follows, in Sect. 17.2 a mathematical model of
WECS will be developed with determining the optimal conditions of work. Sec-
tion 17.3 the MPPT, pitch angle, and PMSG control strategies will be presented
with a diagram control schemes. In Sect. 17.4, a test bench and Turbine emulation
system will be given, with a comparison of simulation and experimental results.
17 FPGA-Based Control of the PMSG on Variable Wind Speed Turbine 359

Grid
PMSG
RL filter
Gearbox
system Power converters

Fig. 17.1 Wind energy conversion system based on permanent magnet synchronous generator
connected to the grid

17.2 Mathematical Model of Wind Energy Conversion

System (WECS)

The wind energy conversion system is composed of several components such as

wind turbine, PM synchronous generator, power converters, etc. In this section, the
different components are presented with their mathematical models.

17.2.1 Wind Turbine Model

The Turbine converts the kinetics energy of wind into mechanical energy (Bakouri
et al. 2015), the mechanical power extracted by the turbine is given by (17.1).

1
Pt = ρπ Cp R 2 v 3 (17.1)
2
Where, Pt is the mechanical power extracted by turbine, ρ the air density, Cp the
power coefficient, R the turbine blade radius, and v is the wind speed.
Due to various losses in the wind energy conversion system, the power extracted
by the turbine is less than the aerodynamic power of wind. Therefore, the power
coefficient comes as a ratio between the aerodynamic and turbine power. This
coefficient depends mainly on the blades inclination angle (pitch angle: β) and the
ratio of wind speed v and turbine speed wt (tip speed ratio: λ) as shown in (17.2)
and (17.3).
   
116 −21
Cp = 0.5 − 0.4β − 5 exp + 0.0068λ (17.2)
λi λi
360 H. Mahmoudi et al.

a b
2500 0.5
11.5m/s =0° =2°
2000 0.4 =4°

Power coefficient cp
Power turbine (W)

10.5m/s =6°
MPPT
1500 0.3
=8°
10m/s
1000 0.2 =10°
9m/s

8m/s
500 0.1
7m/s

0 0
0 20 40 60 80 100 0 2 4 6 8 10 12
Turbine speed (rad.s-1) Tip speed ratio (λ)

Fig. 17.2 Power turbine Pt and coefficient Cp curves under different values of wind speed and
pitch angle. (a): Power turbine curves. (b): Power coefficient curves

1 1 0.035
= − (17.3)
λi λ + 0.08β 1 + β3

R.wt
λ= (17.4)
v
By implementing previous equations, the nominal turbine power is given for wt =
63.43 rad.s−1 with the maximal power coefficient Cp−max = 0.48 which is achieved
for a specific value of tip speed ratio called optimal λopt = 8.1. Accordingly, for
a given value of wind speed, the power coefficient will depend on rotor turbine
velocity wt . In order to provide the power extraction from the wind, the turbine
must be operated with the optimal values of tip speed ratio and power coefficient
regardless of wind velocity, in that case, the power curve is optimal as shown in
Fig. 17.2.
For a variable wind speed, the turbine should be controlled in such away to
maintain the different variables (Cp , λ, wt ) at their optimal values and to guarantee
the safety work of the whole system during the operation, this achieved by deploying
the Maximum power point tracking (MPPT) control and controlling the pitch angle,
those controls will be discussed in Sect. 17.3.

17.2.2 PMSG Model

Permanent magnet synchronous generator is the main component that converts the
mechanical power of turbine into electrical power. It should be noted that, two
different types of PMSG exist depending on PM position in rotor which are Surface
mounted PM and Inner PM (SPM and IPM respectively). Generally, most turbine
with PMSG work with the SPM architecture with a high number of pole pairs,
17 FPGA-Based Control of the PMSG on Variable Wind Speed Turbine 361

which guarantee a high efficiency and it can eliminate the gearbox system (Wu et al.
2011c). In this chapter, the PMSG is considered with IPM architecture and with a
gearbox system.
The dynamical behavior of PMSG is given in a rotative synchronous dq-frame,
where the d-axis is aligned with rotor PM-flux (Yan et al. 2013). Then, the
mathematical model of the synchronous generator becomes more simplified as
shown for the electrical equations (17.5) and (17.6), and the electromechanical
equation (17.7), with the mechanical equation of the equivalent system on (17.8).

did
Vsd = −Rs .isd − Ld + we Lq .isq (17.5)
dt
diq
Vsq = −Rs .isq − Lq − we Ld .isd + we φf (17.6)
dt
3P # $
Te = φf .isq + (Lq − Ld )isd .isq (17.7)
2
dwm Tt
Jeq = − Te − feq wm (17.8)
dt G
Where, Rs is the stator resistance, Ldq are the stator-inductances in dq-frame, φf is
the PM-flux, P is the pole pairs number, we and wm are electrical and mechanical
speed, Vsdq , isdq are the stator voltages and currents components in dq-reference,
Te and Tt are respectively the electromagnetic and turbine torques, Jeq and feq
are respectively the equivalent system inertia and viscous damping, and G is the
gearbox gain.

17.3 WECS Control Strategies

The new wind turbines operate with variable wind speed, which involve to
implement a control strategy for each component of system in order to maximize the
extracted power. As mentioned in 17.2, the wind turbine should be controlled in such
away to maintain the parameters characteristics at their optimal values regardless of
wind speed value as shown in Fig. 17.3.
In this section a control strategy of turbine and PMSG will be presented.

17.3.1 MPPT Control

The purpose of Maximum Power Point Tracking control (MPPT) is to automatically

extract maximum power, and it is activated only if the wind speed between the cut-
on and rated value. There are three types of MPPT controller:
362 H. Mahmoudi et al.

Turbine power

Pn no function MPPT control Pitch angle control

1 2 3

Pmin
Wind speed
Cut-on rated Cut-off
speed wind speed speed

Fig. 17.3 Power turbine evolution with variable wind speed and controls

• MPPT with turbine power profile.

• MPPT with optimal speed ratio.
• MPPT with optimal torque control.
In this chapter, the MPPT with optimal speed ratio is chosen in Fig. 17.4, where
the maximum power operation. This is achieved by keeping the tip-speed ratio at
their optimal value λopt . The principle work of this method, is by measuring the
wind speed v to produce the reference generator velocity wm−ref according to λopt .
The mechanical generator velocity is controlled to tracks wm−ref in steady-state,
at which the MPPT is achieved. When the proposed control is activated, the new
expression of power, torque, and turbine speed are expressed as follow:
⎧  
⎪
⎪ 1 wt−opt 3
⎪ t−opt
⎪ P = π ρ C p−max R 5
⎪
⎪ 2 λopt
⎨ v λopt
wt−opt = (17.9)
⎪
⎪ R
⎪
⎪ 2
wt−opt
⎪
⎪ 1
⎩Tt−opt = π ρ Cp−max R 5 3
2 λopt

Depending on the wind speed profile, the MPPT regulates the generator velocity,
maintaining the turbine at optimal point, and maximize the power capture as shown
in the optimal curve of Fig. 17.2.
The MPPT with λopt was chosen because it guarantees a stale power coefficient
and tip speed ratio comparing to others controllers, but it requires a precise
measurement of wind speed.
17 FPGA-Based Control of the PMSG on Variable Wind Speed Turbine 363

Tt 1 Tm 1 wm Pn βref 1
v Turbine G +_ +_ PI +_ β
Jeq .s+feq τβ.s
Te-ref
PI
Pn Nominale power if : v ≤ vn
λopt wt-opt wm-opt
G +_
R Pg Generator power if : v>vn

Fig. 17.4 Diagram schemes of the MPPT and pitch angle controllers

17.3.2 Pitch Angle Control

To guarantee a safety work of WECS, the new turbines are equipped with a pitch
angle controller to limit the power capture. The Pitch angle control is enabled
against the strong wind (the wind speed is between the nominal and cut-off speed).
If the wind exceeds the nominal value, and as shown in Fig. 17.2, the pitch angle
will vary in such away the power coefficient Cp and tip-speed ratio λ decreased
to new optimal values depending on the wind speed which implies that the power
turbine Pt will be at their rated value, also the rotor velocity will be limited at the
optimal velocity wt−opt . The Pitch controller diagram is shown in Fig. 17.4.

17.3.3 PMSG Control

There are a several methods to control the PM synchronous generator to achieve

different objectives (Bose 2009; Wu et al. 2011b). In this chapter, a Field Oriented
Control (FOC) is used to maintain the PMSG working in the optimal region.
Where, the reference speed will be given by the MPPT controller and the stator
current d-axis component, isd , is controlled to be zero isd−r = 0 (Zero Direct-axis
Control: ZDC). The stator current is will be equal to its q-axis component (17.10).
The electromagnetic torque of the generator derived in (17.7), can be simplified
to (17.11).
+
is = 2 + i2 = i
isd sq sq (17.10)

3P
Te = φf .isq = Kt .is (17.11)
2
With a constant PM flux of rotor, the torque presents a linear relationship with stator
current, which implies that the behavior will be similar to a DC motor, where the
electromagnetic torque is proportional to the armature current.
The current controllers in dq-frame are the PI type, the output of the decoupled
controller can be expressed as:
364 H. Mahmoudi et al.

wind

Te-ref 1 isq-r we.Φf Park

MPPT
v controller Kt Transformer
_+ _+ vsq-r
isq PI _
abc va-r
we.Lq
vb-r
vc-r PMSG
we.Ld
dq
isd _ PI _+ vsd-r SVM Gearbox
+ Rectifier system
isd-r
wm.t
ZDC 0

Fig. 17.5 Diagram scheme of PMSG control strategy

 
Kid
vsd = − kpd + (isd−r − isd ) + we .Lq .isq (17.12)
s
 
Kiq
vsq = − kpq + (isq−r − isq ) − we .Ld .isd + we .φf (17.13)
s
 
With Kp + Ks i is the transfer function of PI controller.
As shown in Fig. 17.5, the reference torque Te−ref is given by the MPPT
controller. Then, the active power control is realized by dq-current closed-loops.
The measured three-phase stator currents isabc are transformed into dq-frame
isdq . The measured dq-currents are then compared to their references isd−r , and
isq−r respectively. The comparison errors are sent to PI-current controllers, which
generate the dq-axis reference voltage vsd−r , vsq−r for the generator side converter
(GSC). An important requirement is that the current controllers must be faster than
the speed controller.
To generate the signal gates for the GSC, the reference voltage dq-frame should
be transformed into the stationary abc-frame vsabc−r , and sent it to the Space Vector
Modulation (SVM) block. Then the generator power will be controlled by adjusting
the input GSC voltage vabc to their references.

17.4 Experimental Results and Discussion

17.4.1 Turbine Emulation System

In experimental studies, the turbine will be emulated by a DC motor, it will be

controlled in such a way it gives the same mechanical behavior of speed and torque
in the output of the gearbox system, then the DC motor will be directly coupled with
PMSG (Hardy and Jewell 2011; Ovando et al. 2007).
17 FPGA-Based Control of the PMSG on Variable Wind Speed Turbine 365

DC
chopper

wm-opt Tm-ref 1 idc-r vα-r α

+_ PI K
+_ PI PWM vdc

wm idc + vα _

DC
motor

Fig. 17.6 Turbine emulation system by a DC motor controlled by DC chopper

Pitch Pt Hsa 5
β Control wm-opt
Hsa
Theta theta
Lsa 6
Lsa
Pt Wm
β Hsb 7
FOC Hsb
Clk
PMSG Vd d
Cp 2 Cs Lsb 8
wt Cp 0 id-ref Lsb
Dout
3 ia zdc
λ Hsc 9
lambda 1 Din
ib
Din Hsc
Vq q
Tt Tt Tm 4
va 11 iab
Wind Lsc 10
Gearbox Tm vb Va
12
wm wt Lsc
actual_speed
Turbine System FOC PMSG SVM Rectifier block
GearBox ADC Converter controller block

Wind MPPT Wm-opt ref_speed_rpm

MPPT controller 2 Kp
Kp cmd cmd gates 1
3 Ki
wind gates_DC
Ki PWM generator
actual_speed
wind profile
DC motor emulation PI controller

Fig. 17.7 Diagram scheme control model of WECS in simulink

The DC motor was chosen as an emulator drive, due to its simplicity and high
torque comparing to AC drives. Also, it will be controlled by a DC-chopper with
PWM signal gates from the control block as shown in Fig. 17.6. The DC motor will
be controlled by two closed-loops, where the outer-loop will regulate the mechanical
speed to its reference given by the MPPT controller, and the inner-loop will regulates
the torque to be equal to the mechanical torque in the output of gearbox system
by controlling the DC-current idc . Then, by using the PWM block, the current
controller output that represents the control signal will be transformed into a duty
cycle signal α of the DC-chopper supplied by a DC supply, which generates a
variable DC-voltage vα to DC motor (Fig. 17.6).
366 H. Mahmoudi et al.

Control signals Signal gates

I/O (0/3.3V) (0/15V)
IP core generation FPGA Driver circuits DC chopper DC Speed
project Interface board 0/3.3V 0/15V Rectifier motor tachometer

Digital
signals PMSG Voltage/current

AD converter
Voltage /current
Adaptive sensors
Shifting
LPF

Fig. 17.8 Proposed design flow of the test bench

Fig. 17.9 Experimental test bench of WECS based on PMSG with turbine emulator system by a
DC motor

17.4.2 Test Bench Description

The experimental test will start by implementing the model control of WECS using
the DC motor emulator in FPGA board and generating a project interface. Then,
the control signals (3.3 V) generated by the FPGA will be transmitted to the IGBT
of DC-chopper and rectifier through a driver circuit from 3.3→15 V. The voltages,
currents, and speed informations from the DC motor and PMSG will be taken by
the voltage/current sensors and speed tachometer which is directly coupled with DC
motor, all informations should be adapted and shifted to signal range 0→3.3 V and
convert them by an analog to digital converter (ADC with 12 bits resolution and 8
channels) which is implemented in the data acquisition board and connected to the
zedboard by the pmod ports as shown in Figs. 17.8 and 17.9.
17 FPGA-Based Control of the PMSG on Variable Wind Speed Turbine 367

17.4.3 Experimental Results

The controllers are implemented using discrete-time domain function, due to their
compatibility with HDL Coder toolbox in Simulink for FPGA. Conventional PI
controllers with a specific gains are chosen. Also, to account for the inertia effects
of the turbine system, a numerical low pass filter was placed on the output speed
measurement block to limit the rate at which the controller could request the system
response. In order to guarantee the real time function, the sampling time of the
interface was fixed in Ts = 0.0001 s.

1 Rectifier 4 Acquisition board 7 DC motor

2 Adaptive board 5 Zedboard FPGA 8 Speed tachometer
3 Driver board 6 PMSG 9 Host interface

To validate the experimental results, a simulation was established as shown in

Fig. 17.12.
Figure 17.13 shows the experimental results of the PMSG and DC motor control,
which are based on wind profile as shown in Fig. 17.10 with rated speed vnom =
10 m/s and the parameters of the system are given in Tables 17.1, 17.2 and 17.3.
The switching frequency of the DC chopper for motor emulator system control was
fixed at 8 kHz, the PMSG was tested with 10 kHz for the rectifier.
To approximate the real dynamic of turbine, the cut-in speed is 3 m/s and cut-off
speed is 20 m/s.
The turbine model with MPPT and pitch angle controller was implemented in the
zedboard in order to give in real time the turbine behavior to the system emulator
by the DC motor.
The turbine controls, was evaluated as shown in Fig. 17.11. The performance of
MPPT can be evaluated by Fig. 17.11a, b, where Cp and λ are in their optimal
values, and are decreased when the wind speed exceeds the rated speed where
the pitch angle is enabled and gives pitch responses as shown in Fig. 17.11c. As
discussed previously, the Pitch angle controller limits the generated power at their
nominal value as verified in Fig. 17.11d. The Fig. 17.11e, f shows the speed and
torque of turbine under the mentioned wind speed profile, it is found that the rated
turbine speed is wt = 63.43 rad.s−1 and rated turbine torque Tt = 23.65 N.m.
To verify the performance of the PMSG control, a simulation test has been
carried out. In this simulation the PMSG control block received the reference
torque from the MPPT controller (Fig. 17.4), then the reference q-axis current
is derived from (17.11) and d-axis is chosen to be zero to simplify the control
algorithm. As shown in Figs. 17.12a, b shows a good tracking performance of the
optimal speed and reference electromagnetic torque given by the MPPT controller
block as mentioned before. Figure 17.12c verify the zero direct axis control by
maintaining the d-axis current component at zero.The magnitude of the three phase
stator currents depends on the q-axis current derived from (17.10), this is shown in
Fig. 17.12d with a nominal frequency equal to 50 Hz when the PMSG velocity is
maximal (157 rad.s−1 ).
368 H. Mahmoudi et al.

12

10
v (m.s-1)

4 Wind speed profile

Rated speed
2
0 5 10 15 20 25 30 35 40 45 50
Time (s)

Fig. 17.10 Wind speed v profile

a b
0.5 10

0.4 8

0.3 6
λ
Cp

0.2 4

0.1 2

0 0
0 10 20 30 40 50 0 10 20 30 40 50
Time (s) Time (s)
c d
5 1.5

4
1
Pt (kW)

3
β

2
0.5
1

0 0
0 10 20 30 40 50 0 10 20 30 40 50
Time (s) Time (s)
e f
70 25
60
20
50
wt (rad.s-1)

Tt (N.m)

40 15

30 10
20
5
10
0 0
0 10 20 30 40 50 0 10 20 30 40 50
Time (s) Time (s)

Fig. 17.11 Turbine dynamic performance with MPPT and pitch angle controllers by FPGA. (a):
Power coefficient Cp . (b): Tip speed ratio λ. (c): Pitch angle β. (d): Turbine power Pt . (e): Turbine
speed wt . (f): Turbine torque Tt
17 FPGA-Based Control of the PMSG on Variable Wind Speed Turbine 369

A prototype system is constructed as shown previously for experimental ver-

ification. The prime motor is a DC machine controlled in speed rotating to
emulate the turbine and gearbox system, where the reference speed given by the
MPPT controller block as discussed before. Figure 17.13a shows a high tracking

a b
12
150 10
wm-opt vs. wm (rad.s-1)

Te vs. Te-ref (N.m)

8
100
6

4
50
Ref mechanical speed 2 EM torque
Real mechanical speed Ref EM torque
0 0
0 10 20 30 40 50 0 10 20 30 40 50
Time (s) Time (s)
c d
6 6
Id
5 Iq 4
4
2
iabc (A)

3
Idq (A)

0
2 5
-2
1 0
ia
0 -4 -5
ib
ic
38 38.1
-1 -6
0 10 20 30 40 50 0 10 20 30 40 50
Time (s) Time (s)

Fig. 17.12 Simulation results: Dynamic performance with the field orientation control. (a):
Mechanical speed wm of PMSG. (b): Electromagnetic torque Te . (c): Stator currents in dq-frame
Idq . (d): Three phase stator currents iabc

a b
12
150
wm-opt vs. wm (rad.s-1)

10
8
Te (N.m)

100
6

50 4
Optimal speed wm-opt
Measured speed wm
2
0 0
0 10 20 30 40 50 0 10 20 30 40 50
Time (s) Time (s)
c d
10
Power generator Pg (kW)

1.5
5

1
ia (A)

0
10

0.5 0
-5
-10
34 34.1
0 -10
0 10 20 30 40 50 0 10 20 30 40 50
Time (s) Time (s)

Fig. 17.13 Experimental results: Dynamic performance with the Field Orientation control. (a):
Measured speed wm of DC motor. (b): Measured electromagnetic torque Te . (c): Measured power
from generator Pg . (d): Measured stator currents a-component ia
370 H. Mahmoudi et al.

performance by DC motor for the reference speed with a minimal initial offset due
to the mechanical time response of the system.
Figure 17.13b presents the produced electromagnetic torque which is derived
from the q-axis current isq , the nominal torque is 9.8 N.m.
The effectiveness of the pitch angle and MPPT controllers is shown in
Fig. 17.13c, where the generated power by PMSG reached the nominal power but is
suffers from a high harmonics which comes from the stator current measurement.
Figure 17.13d shows the output current of the PMSG with a high harmonics, in
nominal region the magnitude of stator current is stable and equal to 5 A which is
consistent with the calculation.

17.5 Conclusion

This chapter provided a comprehensive analysis of various components and control

schemes for wind energy conversion systems based on PM synchronous generator.
Three controls for the system are analyzed, including maximum power point
tracking (MPPT) control, pitch angle control, and generator side converter control
for the PMSG. Also, a new system of turbine emulation based on DC motor was
presented.
A test bench was constructed based on zedboard zynq-7000 system on chip
FPGA to verify and analyze the dynamic performance of the proposed controls and
to compare them with simulation results. Experimental results give a high tracking
performance comparing to simulation results but it suffers from the harmonics due
to current measurement.

Appendix

Table 17.1 Turbine data Parameter Value

Rated power Pn 1.5 kW
Rotor radius R 1.277 m
Maximum power coefficient Cp−max 0.48
Optimal tip speed ratio λopt 8.1
Rated wind speed vnom 10 m/s
Speed range [cut-in ; cut-off] [3 ; 20]
17 FPGA-Based Control of the PMSG on Variable Wind Speed Turbine 371

Table 17.2 PMSG data Parameter Value

Nominal power Pn 1.5 kW
Stator resistance Rs 2.6
d-inductance Ld 63.77 mH
q-inductance Lq 94.32 mH
PM flux φf 0.4 Wb
Equivalent inertia Jeq 0.0931 N.m.rad−1 .s2
Equivalent viscous damping 0.0153 N.m.rad−1 .s
Number of poles pairs P 2
Rated speed wnom 157 rad.s−1
Rated frequency f 50 Hz

Table 17.3 DC motor data Parameter Value

Nominal power 3.6 kW
Rated voltage Vdc 120 V
Rated current Idc 30 A
Resistance R 0.72 Ω
Motor constant K 0.5

References

Ackermann, T., & Söder, L. (2005). Wind power in power systems: An introduction (pp. 25–51).
Hoboken: Wiley.
Azzaoui, M. E., Mahmoudi, H., & Ed-dahmani, C. (2016). Backstepping control of the doubly
fed induction generator using Xilinx system generator for implementation on FPGA. In 2016
5th International Conference on Multimedia Computing and Systems (ICMCS), Marrakech,
Morocco (pp. 599–604).
Bakouri, A., Mahmoudi, H., Barara, M., Abbou, A., Bennassar, A., & Moutchou, M. (2015). A
complete control strategy of DFIG connected to the grid for wind energy conversion systems.
In 2015 3rd International Renewable and Sustainable Energy Conference (IRSEC), Marrakech,
Morocco (pp. 1–6).
Blaabjerg, F., & Chen, Z. (2006). Power electronics for modern wind turbines. San Rafael: Morgan
and Claypool.
Blaabjerg, F., & Ma, K. (2013). Future on power electronics for wind turbine systems. IEEE
Journal of Emerging and Selected Topics in Power Electronics, 1(3), 139–152.
Boldea, I. (2005). Variable speed generators. Boca Raton: Taylor & Francis.
Bose, B. K. (2009). Power electronics and motor drives recent progress and perspective. IEEE
Transactions on Industrial Electronics, 56(2), 581–588.
Busca, C., Stan, A. I., Stanciu, T., & Stroe, D. I. (2010). Control of permanent magnet synchronous
generator for large wind turbines. In 2010 IEEE International Symposium on Industrial
Electronics, Bari, Italy (pp. 3871–3876).
Hardy, T., & Jewell, W. (2011). Emulation of a 1.5 MW wind turbine with a DC motor. In 2011
IEEE Power and Energy Society General Meeting, Detroit, MI, USA (pp. 1–8).
Kocur, M., Kozak, S., & Dvorscak, B. (2014). Design and implementation of FPGA – Digital based
PID controller. In Proceedings of the 2014 15th International Carpathian Control Conference
(ICCC), Velke Karlovice, Czech Republic (pp. 233–236).
372 H. Mahmoudi et al.

Kung, Y. S., Risfendra, R., Lin, Y. D., & Huang, L. C. (2016). FPGA-based sensorless controller
for PMSM drives using sliding mode observer and phase locked loop. In 2016 International
Conference on Applied System Innovation (ICASI), Okinawa, Japan (pp. 1–4).
Mohammadi, D., Daoud, L., Rafla, N., & Ahmed-Zaid, S. (2016). ZYNQ-based SoC implemen-
tation of an induction machine control algorithm. In 2016 IEEE 59th International Midwest
Symposium on Circuits and Systems (MWSCAS), Abu Dhabi, United Arab Emirates (pp. 1–4).
Monmasson, E., Idkhajine, L., Cirstea, M. N., Bahri, I., Tisan, A., & Naouar, M. W. (2011). FPGAs
in industrial control applications. IEEE Transactions on Industrial Informatics, 7(2), 224–243.
Ovando, R. I., Aguayo, J., & Cotorogea, M. (2007). Emulation of a low power wind turbine
with a DC motor in matlab/simulink. In 2007 IEEE Power Electronics Specialists Conference,
Orlando, FL, USA (pp. 859–864).
Pena, R., Círdenas, R., & Asher, G. (2013). Overview of control systems for the operation of
DFIGS in wind energy applications. In IECON 2013 – 39th Annual Conference of the IEEE
Industrial Electronics Society, Vienna, Austria (pp. 88–95).
Polinder, H., Pijl, F. V. D., Vilder, G. D., & Tavner, P. (2006). Comparison of direct-drive and
geared generator concepts for wind turbines. IEEE Transactions on Energy Conversion, 21,
725–733.
Rogers, P., Kavasseri, R., & Smith, S. C. (2016). An FPGA-based design for joint control and
monitoring of permanent magnet synchronous motors. In 2016 International Conference on
ReConFigurable Computing and FPGAs (ReConFig), Cancun, Mexico (pp. 1–6).
Tormo, D., Idkhajine, L., Monmasson, E., & Blasco-Gimenez, R. (2016). Evaluation of SoC-based
embedded real-time simulators for electromechanical systems. In IECON 2016 – 42nd Annual
Conference of the IEEE Industrial Electronics Society, Florence, Italy (pp. 4772–4777).
Wu, B., Lang, Y., Zargari, N., & Kouro, S. (2011a). An introduction to wind energy conversion
systems. In B. Wu, Y. Lang, N. Zargari, & S. Kouro (Eds.), Power conversion and control of
wind energy systems (pp. 1–23). Oxford/Piscataway: Wiley-IEEE Press.
Wu, B., Lang, Y., Zargari, N., & Kouro, S. (2011b). Variable-speed wind energy systems with
synchronous generators. In B. Wu, Y. Lang, N. Zargari, & S. Kouro (Eds.), Power conversion
and control of wind energy systems (pp. 275–316). Oxford/Piscataway: Wiley/IEEE Press.
Wu, B., Lang, Y., Zargari, N., & Kouro, S. (2011c). Wind generators and modeling. In B. Wu, Y.
Lang, N. Zargari, & S. Kouro (Eds.), Power conversion and control of wind energy systems
(pp. 49–85). Oxford/Piscataway: Wiley-IEEE Press.
Yan, J., Lin, H., Feng, Y., Guo, X., Huang, Y., & Zhu, Z. Q. (2013). Improved sliding mode model
reference adaptive system speed observer for fuzzy control of direct-drive permanent magnet
synchronous generator wind power generation system. IET Renewable Power Generation, 7(1),
28–35.