Electrical Power and Energy Systems 156 (2024) 109738

Contents lists available at ScienceDirect

International Journal of Electrical Power and Energy Systems

journal homepage: www.elsevier.com/locate/ijepes

An optimized proportional resonant current controller based genetic

algorithm for enhancing shunt active power filter performance
Behnam Amini a, Hasan Rastegar a, *, Mohammad Pichan b
a
Department of Electrical Engineering, Amirkabir University of Technology (Tehran Polytechnic), Tehran, Iran
b
Department of Electrical Engineering, Arak University of Technology, Arak, Markazi, Iran

A R T I C L E I N F O A B S T R A C T

Keywords: Current harmonics produced by nonlinear loads have destructive effects that affect the power quality of the
Shunt Active Power Filter (SAPF) power system and cause serious problems throughout the network. Active Power Filter (APF) is one of the best
Optimal control and most common solutions to reduce or eliminate harmonic disturbances; But due to the complexity and
Power quality
sensitivity of these filters, the accurate and fast performance of the control technologies used in it is the
Current harmonics compensation
Proportional-Resonant Controller (PR)
important and basic requirement. Therefore, by performing accurate and optimal adjustments for APF control,
system performance and power quality level can be improved significantly. In this article, the design principles of
Shunt Active Power Filter (SAPF) are introduced and the optimal control method of this system is explained by
providing appropriate modeling. Also, as a suitable structure for the current control of the SAPF system, a
Proportional-Resonant (PR) controller optimized by Genetic Algorithm (GA) is proposed. The PR controller with
the capability of accurate tracking of harmonic currents and fast dynamic response is highly compatible with the
SAPF system. In general, the goal is to achieve better and more appropriate performance to reduce harmonic
disturbances and create a robust, faster, more accurate, flexible, stable, and optimal performance of these sys­
tems in detecting and generating compensation signals. The simulations have been done in MATLAB/SIMULINK,
and also the experimental tests of this device have been done and validated in a real DSP-based SAPF. Finally,
according to the results obtained under steady-state conditions with a favorable dynamic response, the Total
Harmonic Distortion (THD) of the grid current has been significantly reduced from 20.2 % to 4.2 % for the R-L
load and from 56.1 % to 4.5 % for the R-C load with a power factor of almost unity. These results confirm the
optimal and standard performance of the proposed method.

harmonics cause improper operation of many electrical devices and

vulnerable equipment throughout the network and reduce the efficiency
1. Introduction
of electrical energy by causing numerous problems. Large-scale
nonlinear load propagation is problematic for all utilities connected to
During the last decade, the advancement of technology along with
distribution lines and, in extreme cases, for the entire network, and
the growing trend of nonlinear loads has made them to be a major part of
many cases are affected by their destructive effects [3–5]. It also leads to
today’s power systems consumption. The presence of nonlinear loads in
the deterioration of electricity quality in distribution and transmission
the network causes harmonic components, nonlinear and distorted
systems in all industrial and civil sectors [6]; Therefore, these problems
characteristics. While one of the main challenges in the design and
make nonlinear loads and harmonics imposed due to them unsuitable for
operation of today’s power networks is to maintain or improve the
any network or power source and necessitates appropriate compensa­
power quality in the power system. Among the important and influ­
tion. Hence, thanks to the level of advanced technologies obtained in
encing parameters on power quality is the presence of harmonics in the
recent years by power electronics and controllers, by using compensa­
power grid, which has received great attention and research. The current
tion systems and carrying out a suitable design, adjustment, and control
harmonics is one of the most important type of harmonic distortion that
for them, the existing adverse effects can be effectively compensated [7].
exists in a power system [1]. Sometimes these harmonic currents with
Eliminating and reducing harmonics has always been an important
the same frequency caused by different non-ideal sources are vectorially
issue, so since the early 1950s, the need to use a tool to eliminate
added to each other and cause double destructive effects [2]. In general,

* Corresponding author.
E-mail addresses: [email protected] (B. Amini), [email protected] (H. Rastegar), [email protected] (M. Pichan).

https://doi.org/10.1016/j.ijepes.2023.109738
Received 10 February 2023; Received in revised form 3 December 2023; Accepted 21 December 2023
Available online 28 December 2023
0142-0615/© 2023 Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
B. Amini et al. International Journal of Electrical Power and Energy Systems 156 (2024) 109738

Nomenclature VGrid Grid voltage

iGrid Grid current
KP Proportional gain VLoad Load voltage
KR Resonant gain iLoad Load current
KI Integral gain VSAPF Shunt Active Power Filter voltage
KD Derivative gain iSAPF Shunt Active Power Filter current
dq0 Park transformation (direct-quadrature-zero) iSAPF,ref Shunt Active Power Filter reference current
SPWM Sinusoidal Pulse Width Modulation VDC DC voltage
LPF Low Pass Filter VDC,ref Reference DC voltage
VSI Voltage Source Inverter Lf Filter output series inductance
S Gate signals CDC-Link DC-link Capacitor
f Sampling Sampling frequency iC Compensation current
f Switching Switching frequency Vs Source voltage
θ Phase angle is Source current
Ts Sampling time IDC DC current estimated
PCC Point of Common Coupling iDC DC current required

harmonics was felt. Always one of the best and most common solutions fourth “current controller”. For this part in articles and studies, there are
available to minimize and deal with the problem of harmonics is to use different types of current control methods. Methods such as: Hysteresis
power filters. In the past, due to the lack of advanced technologies for with a simple structure and easy implementation, but with the problem
semiconductors, the use of passive filters became common as a solution of variable switching frequency [29] - Dead-Beat has a relatively accu­
to remove low-order harmonics and improve power quality at the same rate, robust, and fast performance, but dependence on the system model
time; But due to their many weaknesses and disadvantages such as causes delays and tracking errors due to poor performance [30] - Direct
resonance problems, limited capability, and low flexibility, this solution Power Control (DPC) with the fast dynamic response but complex
was not fully responsive in large scales [8]. Since 1961, the use of structure, variable switching frequency, and model dependence [9,31] -
innovative and efficient harmonic reduction tools known as active filters Sliding Mode Control (SMC) with robust performance to disturbances
began to replace them. With the emergence of new power electronics and uncertainties of system, but, has problems in terms of stability
technologies and the advancement of semiconductors, switching devices [9,32] - Fuzzy Logic Control (FLC) works with imprecise and ambiguous
with controllable power semiconductors such as IGBTs, availability of inputs but depends on human knowledge and expertise and requires
powerful controllers and processors, APFs were developed. Today, these regular updating of the rules [33] - Repetitive control has a very low
filters are popular as a common tool for compensating harmonics caused dynamic response [34] - Lyapunov function-based controller that has
by nonlinear loads and improving power factors and quality [9]. In high tracking accuracy but suffers from variable switching frequency,
general, there are various types, definitions, and structures of power high sampling frequency and high calculation [35,36].
filters, and based on different views, different divisions can be made for Also currently, there are many methods for setting and designing the
them. But usually, three-phase SAPF is used more. Because it is the most parameters of the controllers. The most popular of them is the Ziegler-
widely used and suitable choice for compensating current harmonics Nichols method or various analytical methods such as dominant poles
and reactive power compensation [10]. and numerical methods, which are usually based on the numerical
Of course, the most important point is that the effectiveness and evaluation of the objective function. In general, the methods of tuning
performance of APFs largely depend on the speed and accuracy of their and designing the parameters of the controllers can be done according to
control algorithms. Therefore, the control of these systems is very several strategies: Classical tuning methods (including analytical and
important and sensitive, and a suitable design should be considered for numerical methods) - Self-tuning or automatic adjustment methods
them [11]. In general, the control unit of the APF system includes four (based on Model Reference Adaptive Control (MRAC)) - Smart tuning
main and important control algorithms that each of which is responsible methods (including modern and optimizing methods). In classical
for performing a task in this system. The first case is harmonic detection methods, the tuning of the parameters of the controllers is done ac­
and extraction, and the types of this algorithm are classified under time cording to the analysis of the location of the roots and the exact place­
domain methods and frequency domain methods [12]. However, there ment of poles-zeros according to the specified requirements. For
are other Combined or modified versions of these approaches [13–15]. example, these requirements are: damping ratio, steady-state error of
Time domain methods such as: Synchronous Reference Frame (SRF), the system, settling or rising time, overshoot and undershoot, dynamic
which is also known as d-q theory or d-q frame that by providing characteristics, etc. Also, self-tuning methods based on adaptive control
excellent dynamics is the best choice for current-based applications are the reference model whose principles are well-defined and have
[11,12,16] - instantaneous power theory or known as PQ theory, which become a standard part of textbooks. But sometimes, because these
is mostly for power applications [12,14,17,18]. Also, frequency domain systems are designed and tuned according to old methods, they are
methods include: windowing-wavelet-based technique [19], Discrete almost obsolete and can no longer fully meet the needs of today’s
Fourier Transform (DFT) [20] and Recursive Discrete Fourier Transform standards and technologies. In addition to these, due to the advance­
(RDFT) [21,22], and methods based on Fast Fourier Transform (FFT) ments in the topic of optimization and smart methods, it is possible to
[12]. The second case is the DC-link capacitor voltage adjustment con­ achieve optimal performance by using them. For this purpose, smart
trol algorithm, whose algorithms are generally classified as methods tuning methods can be considered, because precise and optimal tuning
based on direct voltage error and self-charging methods [23,28]. The has a significant effect on the performance of such systems.
third case is the synchronizer algorithm, which is the Phase Lock Loop The current control algorithm is the most important and effective
(PLL) based on (SRF) [24,25], Zero Crossing Detectors (ZCD), [26] and part of APF performance, and the main discussion and innovations of
synchronizers based on Artificial Neural Network (ANN) [27] among its this research have been done in this part. It’s clear that if the control
widely used methods. methods have defects or errors, even have a strong and accurate system
But the last and most important control part of the APF system is the in terms of technical specifications, the result not be satisfactory and

2
B. Amini et al. International Journal of Electrical Power and Energy Systems 156 (2024) 109738

lead to the malfunction of the entire APF system. On the other hand, due none of them have been able to pass the desired standard value. Because,
to the complexity and sensitivity of filters and power quality compen­ for the proposed system by the IEEE-519, IEC62040, and IEEE1547
sators, analysis, design, and most importantly optimal and appropriate standards, the approach is to reduce the THD of the grid to less than 5 %,
control should be considered for them. Considering these issues and the which has been achieved in the studies conducted. Because the old
existence of problems in most other controllers for this application, the studies with their settings and methods were ineffective. - Considering
PR controller is preferred as the most efficient control method in har­ the specific settings in GA problem-solving parameters, the random
monic and nonlinear load conditions. In addition, this controller over­ search of the presented algorithm has been done purposefully and from
comes two disadvantages of conventional controllers such as instability different paths so that all possible answers can be checked to select
in tracking a sinusoidal reference with zero steady-state error and low suitable random answers without any restrictions in the search path. In
disturbance compensation capacity [37]. According to the many capa­ addition, it is not easily interrupted by small changes in the input values
bilities and advantages of the PR controller in harmonic applications, by or by significant amounts of noise in the system. It also has a high
applying ideas and expanding its efficiency, it can be used as a convergence speed in finding final solutions.
compatible and efficient controller in the field of APF. On the other According to the studies carried out so far, it can be understood that
hand, there are still many problems in the design and adjustment of the the design and regulation of the systems presented in most of these
conventional type of this controller, which allows further studies. studies are based on old methods and more than the conventional
Very limited similar studies have been done in this field, which are methods are used which are almost obsolete or can no longer meet the
often different from the work of this article; For example, in reference needs of today’s sensitivities and standards. It means that no specific,
[38] a resonant proportional current controller is presented for reactive standard, and optimal design has been done so far and they have been
power compensation and selective harmonic elimination in medium used in a classical way. This issue is very important because if the control
voltage industrial networks using a hybrid APF. The optimized design of part of these systems is not properly adjusted and designed, even leads to
modified proportional-resonant controller for the current control of instability. The importance of this issue doubles in critical situations
SAPF is reviewed in reference [39], which only presents simulation re­ such as the condition of the occurrence error and non-ideal source. Also,
sults without any experimental validation. Also, in the references these studies are mostly limited to computer simulation, while their
[40,41], various other types of similar studies in other fields of PR practical implementation and experimental tests have been less inves­
controller application and other types of APFs based on different control tigated. In addition, there is no standard quality measure and specific
methods are given, which mostly have flaws, and a comprehensive and evaluation function for them, so a complete analysis can be imagined for
optimal study has not been done in them. them. Because the SAPF and its performance indexes are different from
The specific problem related to PR controllers is related to the regular grid-connected converters. On the other hand, according to the
consideration of its resonators according to the correct analysis of sys­ limitations and problems of other studies conducted along with the
tem harmonics and how to accurately adjust the controller parameters. progress in the field of control and optimization, it is necessary to pro­
That is, according to the different types of nonlinear load and their vide newer and better controllers to use or to revise their design in order
different harmonics. First of all, the intended system should be fully to achieve a better performance of compensation. It is clear that precise
evaluated in terms of its performance, characteristics, and dominant design and optimal tuning have a significant effect on the performance
harmonic orders. So that the design of PR controller resonators can be of such systems, and with this perspective, the performance of the APF
designed exactly according to the needs and specific to the same system. system and the level of power quality can be significantly improved
It should be done not normally and without evaluation. In addition, due according to valid international standards [42].
to the high number of coefficients of PR controllers, it is not possible to In this paper, a SAPF based on a PR controller is proposed as an
set proportional coefficient, resonant coefficient, and harmonic resonant efficient method, and its application as a compatible harmonic
coefficients at the same time according to the old and usual methods. compensator with a fast dynamic response is discussed. The proposed PR
The system has at least five coefficients and achieving the best state for controller is implemented in the internal current control system of SAPF.
all of them is usually very complicated and difficult. This is while up to Considering the capabilities, many advantages, and the high ability of
now more than the usual methods and adjustments of individual co­ the PR controller in compensating harmonics, as well as its very good
efficients have been done and it can be seen that in this case the best performance in various other applications, by using this controller in the
adjustment was never achieved. Also, standard quality criteria should be SAPF system and performing an optimal and appropriate design for it,
determined based on the controller’s role and performance in the sys­ the performance of the system can be greatly improved. Also, in order to
tem, objective functions, and system outputs. In addition, the specific reach the most optimal adjustment mode for controller parameters, a
problem related to optimization techniques, especially GA, related to the new method based on Genetic Algorithm is proposed to determine the
consideration of the system’s specific objective functions and appro­ optimal controller coefficients. In addition, the controller settings and
priate index performance, how to perform general and specific algo­ the objective function provided for GA are defined and designed in a
rithm settings, the type of inputs, the appropriately defined limitation way that shows the optimization role and desirable performance of the
for the problem variables, the correct use of operators GA and creating a current controller. The results of this study confirm that by using the
suitable space for possible answers in feasible area. proposed method, improved and more appropriate performance can be
On the other hand, the difference between the GA in this work and achieved to reduce harmonic disturbances grid current and create a
other GAs that are used to adjust the proposed controller can be resistant, fast, accurate, and optimized performance of these systems in
mentioned in several things: - Considering a wider sample space, detecting and generating the compensation signal.
including more number of iterations and a larger population, which The rest of the paper is organized as follows. In Section 2, a more
better guarantees a complete, and comprehensive search of the algo­ detailed system description of the proposed scheme and the studied
rithm. - For GA, an objective function is usually determined, but in this system components including control units has been presented. Section
study, in addition to the main objective function, which is the THD% 3 introduces and describes the details of the proposed control system
grid current, other things such as system and controller errors are and optimization procedure. Simulation and experimental results while
considered based on IAE and error criteria. Hence, instead of a single- verifying the effectiveness of the proposed method are completely
sided objective function, a complete and multi-sided objective func­ explained and discussed in Section 4. Finally, this paper is concluded in
tion is pursued by GA. - So far, GA has been rarely used to adjust the section 5.
parameters of the controllers in the present form, and their use has only
been provided for a limited number of parameters, not all available
parameters. In addition, according to valid international standards,

3
B. Amini et al. International Journal of Electrical Power and Energy Systems 156 (2024) 109738

Fig. 1. System configuration for the three-phase shunt active power filter with control units.

2. System description based application, the synchronous reference frame (SRF) or dq-frame
method to detect and extract harmonic components, a classic PI
The scheme of system configuration for the three-phase SAPF with controller to regulate the DC-link capacitor voltage, and the PLL-based
the connection to the main supply, nonlinear load and control algo­ synchronizing algorithm as the best methods have been selected for
rithms is shown in Fig. 1. The SAPF is connected between the source this application that easily meets the needs of this study.
(Grid) and close to the harmonic generating load (Nonlinear load). In The extraction or detection of harmonic currents is the prerequisite
total, this system consists of five main components. The capacitor on the of the SAPF control unit. This algorithm is the first stage of SAPF control,
DC side of the inverter as an energy storage element - the inductor as a which is also known as the reference signal generation block. Its main
transmission element to the grid and connects the inverter phase to the task is to detect and separate the harmonic components and produce a
grid - Voltage Source Inverter (VSI) as a power converter - control units - suitable reference signal accordingly. The fast and accurate operation of
and other components such as sensors, fuses, heat sinks, etc. this section leads to the generation of the appropriate reference current
The non-linear load draws fundamental and harmonic currents (iL) that further controls the SAPF so that the compensator can regenerate
for its operation and the grid needs to supply both fundamental and the desired injection current to reduce harmonics according to this
harmonic currents to the non-linear load. Therefore the SAPF must correct reference current signal. These algorithms use a signal process­
generate the same harmonic currents supplied by the grid, except in ing function known as the distortion identifier, which takes the distorted
opposite polarity. The injection current as suitable compensation cur­ nonlinear load current signal to generate the corresponding reference
rent (iC) by the SAPF can compensate and neutralise the harmonic current signal by separating the distortion components from the original
components in the grid current and the shape of the grid current (iS) component. Extraction of harmonics based on SRF, which is also known
resumes to sinusoidal since the current has a fundamental component as theory (dq) or d-q frame, is done through mathematical calculations
only. Hence the amount of THD of the grid side current is up to reduce as of three-phase load current, which is done in the dq0 frame. Park’s
much as possible. In addition, the proposed structure for SAPF has the transformation matrix is applied to convert the three-phase load current
capability of reactive power compensation, and if reactive power in the abc frame (iLoad_abc) to the three-phase load current in the dq0
compensation is needed, this filter can provide it. This feature prevents frame (iLoad_dq0). Only d-q frame are involved in the extraction process
the imposition of additional reactive power sent or received to the and the reference phase angle (θ) is provided by the PLL synchronizer for
network, thereby improving the quality of the transmitted electrical (sin_cos) calculations. In d-q frame, the fundamental component (iL-dq-
energy. If the SAPF is supposed to provide reactive power, it is enough to (fund)) appears as fixed waveforms and the harmonic components (iL-dq-
add a part of the converted reactive power to the(q)component so that a (har)) appear as oscillating waveforms. Usually, a second-order low-pass
part of the reactive power required by the load is compensated by the filter (LPF) is used to separate the fundamental and harmonic compo­
SAPF and puts less pressure on the grid. nents. Finally, the desired reference current (iSAPF,ref) is created using
In recent years, different algorithms have been developed for each of the harmonic components extracted through the inverse Park transform.
these four main control parts of SAPF. Each of these methods has DC-link capacitor voltage regulator is a control algorithm that takes
strengths and weaknesses, and there are advantages to implementing the DC-link capacitor voltage (VDC) and compares it with the desired
some techniques over other techniques. Choosing the right method is reference value (VDC,ref), and estimates the appropriate amount of DC
actually a compromise between speed, accuracy, cost, ease of imple­ current for DC-link regulation. Estimated (IDC) is the value of (iDC)
mentation in terms of engineering, and desired performance. Therefore, required to maintain (VDC) at the desired reference value to adjust
choosing the appropriate control methods is an important and sensitive switching losses. The DC-link capacitor is an energy storage element to
process and has a great impact on the performance of the SAPF to ach­ provide the real power difference between the power requested by the
ieve better compensation. In this research, according to the current- load and the power provided by the source in transient states. It also

4
B. Amini et al. International Journal of Electrical Power and Energy Systems 156 (2024) 109738

Table 2
GA parameters implementation.
Description Name Value

Population size nPop 25

Maximum number of iterations MaxIt 50
Global NFE kP, kR , kR,5, kR,7, kR,11 -
Number of decision variables nVar 5
Size of decision variables matrix VarSize [1 nVar]
GA parameter ranges (Lower) VarMin [0, 0, 0, 0, 0]
GA parameter ranges (Upper) VarMax [50, 90, 90, 90, 90]
GA generation gap - 0.9
GA crossover percentage - 0.7
Fig. 2. PR controller with harmonic compensating resonant terms. GA selection rate depends on the rank of each individual
GA mutation percentage - 0.2
The objective function Fitness function IAE, THD, error

Table 1
Performance index estimation of pr controller.
Name Of Criterion Formulation
∫∞
IAE 0 |e(t) |dt
∫∞
ISE 2
0 e(t) dt
∫∞ ∫∞
ITSE te(t)2
dt 0 t|e(t) |dt
0

ITAE

Fig. 4. Best cost of GA and Convergence of the objective function.

Table 3
The optimal parameters of proposed optimal PR controller.
KP KR KR,5 KR,7 KR,11

2.8793 0.0657 5.9e-05 6.1e-05 6e-05

provides a voltage with a small ripple. In permanent states, the active

power provided by the source must be equal to the active power
required by the load plus the power caused by the losses in the SAPF,
which keeps the DC-link voltage constant at the reference value.
Whenever the load conditions change, the active power balance be­
tween the source and the load is disturbed, and the power difference
must be compensated by the DC-link capacitor. This causes a change in
the DC-link capacitor voltage and deviates from the reference value. For
the proper functioning of the capacitor, some active current must be
exchanged with the source. This active current can lead to the charging
or discharging of the capacitor and if the capacitor voltage of the DC link
is kept constant at the reference value, the actual consumed power will
be equal to the source power. Therefore, the amount of this active cur­
rent received from the source is controlled by a control circuit that keeps
the DC-link capacitor voltage constant. A capacitor voltage smaller than
the reference value means that the active power provided by the source
to provide the load requirement is low, as a result more active current
must be received from the source. Also, if the capacitor voltage is greater
than the reference value, the control block tries to reduce the active
Fig. 3. Flowchart of GA for optimal tuning PR controller parameters.
current received from the source. After fixing the DC voltage, the DC-
link capacitor is only responsible for providing the DC voltage, so the
reactive power that is supposed to be compensated for the load must be
compensated by this capacitor. The chosen method for this section is the

5
B. Amini et al. International Journal of Electrical Power and Energy Systems 156 (2024) 109738

Fig. 5. The Performance simulation results of SAPF with proposed PR control method: a) Grid Voltage, b) Load Current, c) Grid Current, d) SAPF Current, e) Load
reactive power, f) Grid reactive power, g) SAPF reactive power, h) DC-link Voltage, (i) FFT analysis of the grid current with and without SAPF operation.

use of a classic Proportional-Integral (PI) controller. measured frequency (Freq), in hertz. Angle (rad) varying between 0 and
The synchronizing algorithm is needed to ensure the SAPF phase 2π, synchronized on zero crossings of the fundamental (positive-
operation with the operating power system. In addition, it is necessary to sequence) of phase A (wt). The three-phase input signal is converted to a
provide the synchronization phase angle (θ) for transformations and dq0 rotating frame (Park transform) using the angular speed of an in­
calculations. The PLL (3ph) block models closed-loop control system, ternal oscillator. The quadrature axis of the signal, proportional to the
which tracks the frequency and phase of a sinusoidal three-phase signal phase difference between the abc signal and the internal oscillator
by using an internal frequency oscillator. In this method inputs are The rotating frame, is filtered with a Mean (Variable Frequency) block. A
normalized three-phase signals of grid voltage (VGrid_abc), in pu. The Proportional-Integral-Derivative (PID) controller, with an optional

6
B. Amini et al. International Journal of Electrical Power and Energy Systems 156 (2024) 109738

Table 4 Automatic Gain Control (AGC), keeps the phase difference to zero by
The SAPF simulation system configuration parameters. acting on a controlled oscillator. The PID output, corresponding to the
Description Name Value angular velocity, is filtered and converted to the frequency, in hertz,
which is used by the mean value.
Grid voltage (rms) VGrid,rms 70 V
Grid frequency f Grid 50 Hz Current control as a most important part is that the focus will be on
Sampling frequency fSampling 10 kHz designing a control approach for fast and accurate tracking of references
Switching frequency fSwitching 10 kHz and maintaining stable and robust conditions for.
Fundamental sample time Ts 1e-6 s

Fig. 6. Performance of SAPF with proposed PR control method at SAPF startup operation in t = 0.4 s: a) Load current, b) Grid current, c) SAPF current.

Fig. 7. Performance of SAPF with proposed PR under load change and start at t = 0.6 s: a) Load current, b) Grid current, c) SAPF current, d) DC-link voltage.

7
B. Amini et al. International Journal of Electrical Power and Energy Systems 156 (2024) 109738

Fig. 8. Simulation results under one phase voltage drop and unbalanced condition: a) Grid voltage, b) Load currents, c) Grid currents, d) Grid currents with SAPF
startup operation in t = 0.4 s, e) SAPF currents, f) DC-Link voltage.

Fig. 9. Bode plots of transfer function for proposed optimal PR and conventional PR.

8
B. Amini et al. International Journal of Electrical Power and Energy Systems 156 (2024) 109738

non-sinusoidal current and dealing with harmonic disturbances in order

to set the basic frequency of the grid is challenging. Also, in this case,
conventional controllers are not responsive. For example, PI controllers,
having one pole (with infinite gain) at zero frequency, are unable to
completely eliminate steady-state error unless adapted to the d-q frame.
Hence, the PR controller can be used as a suitable alternative to ineffi­
cient conventional methods [37].
Eliminating the harmonics in the inner loop of the control by reso­
nant terms has made the PR controller a suitable control system for
various applications. The internal loops are the resonators that are
added to the controller transformation function for the harmonic orders
in the system to create a more accurate expression for the controller
transformation function and current control function. They are equiva­
lent to a PI controller that moves both clockwise and counterclockwise.
The PR controller is chosen as the preferred method that can deal well
with harmonic, nonlinear and three-phase asymmetric systems. Elimi­
nating steady-state sinusoidal signal error in nonlinear load conditions,
Fig. 10. Experimental setup of the SAPF device: a) back view, b) front view. dealing with any disturbance even disturbance in frequencies other than
the main and base, easy and accessible implementation, harmonic
compensation without affecting system dynamics and overall band­
Table 5
width, good dynamics, and accuracy performing in calculations are
The SAPF experimental system configuration parameters.
another advantage of PR controller. Also, this controller has unlimited
Description Name Value/Type attention to the network frequency, which almost guarantees zero
Converter Full bridge VSI 3-phase / 3-legs steady-state error in both steady-state and transient states, and can track
Switches type IGBT SKM150GB12T4 the sinusoidal signal well. As well as, The other advantage of this
Processor DSP Control TMS-320F-28335 TI
method is the compensation of certain harmonics for the components of
Driver IGBTs Driver Prototype (IC-HCPL316j)
Transformer Step - down 220 V / 70 V both positive and negative sequences; Therefore, when the frequency of
Current sensors LEM sensors LA100P the reference signal is the same as the frequency of the controller, this
Nonlinear load 1 Diode bridge+R 6 pulses+10 Ω controller will have an appropriate response [43,44]. Due to the many
Nonlinear load 2 Diode bridge+RC 6 pulses+10 Ω+1100 μF capabilities and advantages of the PR controller, it can be used as an
DC-link capacitor CDC-Link 2200 μF
Series inductance Lf 1 mH
efficient and compatible controller for the current control of the SAPF
Series resistance Rf 10 mΩ system to effectively deal with the issue of harmonics.
In order to check the equation mechanism of the PR controller, the
control diagram of this method for internal current control is shown in
3. Proposed current control system Fig. 2:
The relation of the controller transformation function is a combina­
In this section, the main tasks of this study and all the necessary tion of the proportional term and the resonant terms, which in the ideal
aspects of the design of the proposed method are reviewed. First, the case with unlimited gain in ω is as described in the following relation:
design principles of the PR controller are introduced and the details of its
use are described in the SAPF current control section. Then the opti­ CPR (s) = Kp + KR
S
(1)
mization procedure, which is specifically the optimal tuning of PR S2 + ω2
controller parameters using GA, is examined. The current control in­
where ω is the resonant frequency (grid frequency 50 Hz), KP and KR
cludes an internal control algorithm that adjusts the converter perfor­
respectively are the proportional coefficient and resonant coefficient.
mance by generating appropriate switching pulses (S) so that the SAPF
Such a control has a high gain around the resonant frequency, so it is
acts as an effective compensator. This algorithm usually includes a Si­
able to remove the steady-state error or inject the sinusoidal signal ac­
nusoidal Pulse Width Modulation (SPWM) to generate the desired
cording to the rule of the internal model during tracking. The resonant
switching pulses and an internal current control loop that ensures that
frequency must be kept close to the system frequency to maintain good
the injected compensation current (IC) is generated according to the
controller performance. The ideal PR controller acts like a network with
reference current (ISAPF,ref). The basis of operation of these algorithms is
an unlimited quality factor, which is difficult to implement in reality;
based on the comparison of the actual current and the reference current,
Because firstly, the controller provides an unlimited gain that cannot be
and according to the difference between them or the error detected by
achieved in an analog or digital system. Secondly, to eliminate the
the controller, a suitable compensation current is produced. By injecting
harmonic effect created by the grid voltage, the gain of the controller is
this compensation current into the system, the waveform of the grid
further reduced in other frequencies that is not enough; Therefore, a
current is modified and thus the SAPF compensation process is
relatively ideal controller with a high-gain low-pass filter is used to solve
completed and the harmonic value is reduced based on the THD level.
the above problems. The transformation function of the controller in this
The proposed control method of this research is considered for this part,
case is as follows:
which is based on the optimized PR controller and innovations have
been made in it. 2ωc S
CPR (s) = Kp + KR (2)
S2 + 2ωc S + ω2
3.1. Proportional-resonant controller design In this case, the PR controller leads to a limited gain to minimize the
error and an expanded bandwidth by properly setting the cutoff fre­
The PR controller has recently attracted attention in many applica­ quency (ωc). This feature is useful when the network frequency changes
tions due to its good performance and numerous advantages. So far, this and reduces the sensitivity of the system.
controller has been mostly used for applications related to microgrids, The PR controllers are widely used in the control of converters in
distributed generations, and inverters connected to the grid to adjust the various other applications. But for the application of SAPF, whose pur
operation of the injected current. Usually, controlling or regulating the

9
B. Amini et al. International Journal of Electrical Power and Energy Systems 156 (2024) 109738

Fig. 11. The experimental results of SAPF based on the proposed PR controller: a) Grid voltage, b) FFT harmonic order analysis of grid voltage, c) RMS load currents,
d) THD of load currents, e) RMS grid currents, f) THD of grid currents, g) The power factor (PF) before the compensation, h) The power factor (PF) after the
compensation, i) FFT harmonic order analysis of grid currents.

pose is to improve performance in the face of harmonics, the harmonic ∑

KR S KRh S
compensator is designed as the following relationship: GPR (s) = KP + + (4)
S2 + ω2 h=5,7,11,⋯S2 + (ωh)2
∑ S
CHC (s) = KRh (3) While the discretization of this transformation function in the Z
h=5,7,11,⋯ S2 + (ωh)2
domain provided by the “Tustin” operator has been used.
where h is the harmonic orders and KRh is the harmonic orders resonant There are many sources of poor control performance in processes,
coefficients that can be easily added to the PR controller. Usually, in the and it is estimated that a high percentage of control systems have per­
three-phase power systems, odd harmonic orders such as 5th, 7th, 11th, formance problems. This issue is due to some reasons such as improper
13th, 17th, and 19th are the main and effective ones. Harmonic reso­ adjustment and design of the used controller, lack of attention to
nators are selected and designed according to the harmonic orders in the compensation for errors and damages, use of improper control structure,
network that are to be compensated, to be added to the conversion etc. In general, the most important point to achieve optimal perfor­
function of the PR controller. mance is the correct tuning of the gain of the controllers. Especially, the
For the SAPF current control system, the transfer functions of the PR gain of resonant sentences should be designed to be high enough and
controller can be written as: achieve zero or quasi-zero steady-state error for proper compensation.

10
B. Amini et al. International Journal of Electrical Power and Energy Systems 156 (2024) 109738

Fig. 11. (continued).

This requires a precise design and adjustment according to reliable that differentiates it from other methods, it can be comprehensive, more
methods, which methods based on intelligent optimization are the best complete and faster search among possible answers, parallel and
choice for tuning the proportional and resonant parameters of the pro­ simultaneous processing on a number of potential answers, work on the
posed PR controller. coded parameters aim to increase the calculation speed and high accu­
racy of this method [45–47]. Therefore, GA is the preferred method of
this study to optimally adjust the coefficients of the PR controller in the
3.2. Optimization procedure SAPF system.
The GA performs three steps to generate a new population from the
In this section, the optimal tuning PR controller parameters is done previous population:
using Genetic Algorithm (GA) as the preferred method. In general,
optimization is used in various sciences and has many applications and it 1. Starting with an initial random population and calculating the fitness
means achieving the best result in the given conditions. Also, depending score of each individual based on the fitness function.
on the type of problem, goal and limitations, there are different in­ 2. Prioritizing individuals based on merit and choosing a number of the
terpretations of the best. Intelligent optimization algorithms are based best according to their score.
on iteration and are often population oriented. It starts from an initial 3. Selected individuals mix with each other using genetic operators.
guess and continues toward the optimal solution by generating and
repeating an improving sequence. GA is the most well-known intelligent Next, by recalculating the fitness function for the new generation,
optimization method and evolutionary algorithm that has many appli­ this process is repeated until the best solution is found. Eventually, ac­
cations in various scientific and engineering fields and can be a very cording to the termination condition, the fitness value of the final gen­
suitable substitute for other conventional methods. GA is based on eration individuals converges to the overall optimal value according to
repetition, direct search and random sampling, inspired by natural the number of iterations, the best individual, error calculation and time
events and based on the science of genetics and natural selection (how to limit are applied. Finally, after the completion of the algorithm process,
inherit and transfer genes) based on the law of survival of the fittest. The the optimal responses are stored which are the optimal gains of the PR
purpose of its search is to optimize a function called the fitness function controller.
as a criterion for identifying optimal answers defined by the user. In In general, controllers try to minimize the error, which is the dif­
order to carry out this process, GA gathers a population of suitable ference between the desired response of the system and its actual
possible answers for this matter. Each of these candidates is considered response. Performance measures are mainly based on system errors,
an individual, and the set of characteristics of these individuals are which have different functions. According to Table 1, there are various
transferred to the next generation by their chromosomes. In each birth performance criteria for calculating the error, which for this study In­
period, which is called a generation, a new population is created, which tegral Absolute Error (IAE) is selected. In addition, minimizing the
generally includes people who are more compatible with the conditions amount of THD of the grid current and reducing the steady-state error
of the fitness function. Among other features of the genetic algorithm

11
B. Amini et al. International Journal of Electrical Power and Energy Systems 156 (2024) 109738

Fig. 12. The experimental results of SAPF based on the proposed PR controller under R-C load: a) THD of load currents, b) RMS load currents, c) THD of grid
currents, d) RMS grid currents.

are also considered. calculates the initial parameters of the proposed controller based on the
In the IAE formula, the integral of the absolute value of the error is current control implementation of the SAPF system simulated in Matlab-
calculated where t1 is the starting time and t2 is the total simulation time Simulink by generating a random initial population. Here, the fitness
in each simulation. score of each person is evaluated by the defined fitness function (based
∫ t2 on the minimum error mode, IAE, and THD). Then a population of
IAE = |i* (t) − i(t)| dt (5) possible and initial answers is transferred to the next generation based
t1 on the merit of the best ones. In each iteration, the optimization criteria
Also, the following equation is used for the voltage ripple formula: are checked to terminate, and as long as the evaluation criteria are not in
the best state compared to the previous iteration, this process is repeated
Vmax − Vmin
Voltage Ripple Percentage (rip%) = × 100 (6) until finding the best solutions in the same way by creating a new
Vaverage
generation and recalculating the fitness function. Also, to improve the
Due to the complexity and the high number of PR controller pa­ performance of the algorithm, the selected options are combined using
rameters, it is difficult to optimize these parameters. But GA is a suitable genetic operators (Selection-Crossover-Mutation-Replacement) in each
choice for finding optimal solutions, which works well even for complex iteration to obtain the best global solutions. Finally, after several itera­
or high-dimensional problems. Even if there are changes around the tions by satisfying the stopping conditions, the final parameters of the
operating point, the SAPF will still perform well because PR is a flexible proposed PR controller are calculated and the algorithm is finished by
controller. The parameters of the proposed PR controller will be saving and reporting the best and most optimal solutions. In Fig. 4, the
adjusted by GA. Because by using this algorithm and specifying the graph of the objective function is shown based on the number of itera­
criteria for the objective function, the most optimal adjusted state of the tions, and the best cost of GA does not change above iteration 700,
proposed controller for the SAPF system can be achieved. In general, the which indicates that GA has achieved the most optimal solutions.
goal is to achieve an improved and more suitable performance in order Finally, the results of the optimal parameters of the proposed optimal PR
to reduce harmonic disturbances and create a resistant, fast, accurate controller including KP, KR, KR,5, KR,7, KR,11 are given in Table 3.
and optimized performance of this active filter system in detecting and
generating a compensation signal. The approach of this research is in 4. Results and discussion
accordance with the IEEE-519 standard for the system, improving the
power quality and reducing grid THD caused by the nonlinear load In this section, the performance of the SAPF based on the proposed
current to less than 5 %. PR controller optimized by the GA is investigated and its performance is
The procedure for optimizing the parameters of the proposed PR validated by the practical implementation of this system. To confirm and
controller by GA is shown as a flowchart in Fig. 3. The working process analyze the performance of the proposed method, the results of both
of this algorithm according to its simple process is as follows: first, the simulation and experimental results are presented.
definition of the problem, the range of variables, and the initialization of
GA parameters are done according to Table 2 in MATLAB. Then GA

12
B. Amini et al. International Journal of Electrical Power and Energy Systems 156 (2024) 109738

Fig. 13. The comparison experimental results of SAPF based on the proposed PR controller under steady state condition: a) ch1:Grid current - ch4: Load current, b)
ch1: Grid current - ch4: SAPF current, c) ch1: Load current - ch4: SAPF current, d) ch1: Load current - ch4: Grid current, e) Synchronized grid voltage and current at
grid side (220 V), f) Synchronized voltage and current at grid side (70 V).

4.1. Simulation results on the grid side (Fig. 5-c). Usually, the maximum allowable compen­
sation current is assumed to be about 30 % of the grid’s peak current, in
Simulation tests in transient-state and steady-state using MATLAB/ order to meet the high dynamic response, the current compensation in
SIMULINK simulations are given to investigate the performance load changes with low current ripple. The SAPF injection harmonic
response and validate the proposed system. In these simulations, control current (Fig. 5-d) for the harmonic compensation. In addition, the
delay, sampling effect, dead time, and quantization effect as imple­ reactive power results for load, grid, and SAPF shown in (Fig. 5-e),
mentation limitations are considered. Fig. 5 shows the performance (Fig. 5-f), and (Fig. 5-g) respectively. Also, the complete compensation
simulation results of SAPF based on the proposed PR control method of reactive power with a unity power factor is guaranteed by the syn­
including grid voltage and current, load current, SAPF current, DC-link chronization of grid voltage and current. According to (Fig. 5-h) the DC-
voltage and grid current harmonics analysis. Also, the system parame­ link controller, with a very low ripple, has set and maintained the DC
ters are given in Table 4. voltage at the desired value. Also, the Fast Fourier Transform (FFT)
Based on this, it can be seen that the grid voltage (Fig. 5-a) is analysis of the grid current in both modes with and without SAPF
balanced and sinusoidal, but the nonlinear load current (Fig. 5-b) has compensation is shown in (Fig. 5-i). Based on this figure, the %THD of
very high harmonic values, which SAPF compensates for significant the grid current has decreased from 21.82 % to 3.86 %, where the
amounts of them effectively and has maintained the quality of current dominant harmonic components 5th, 7th, 11th, 13th, 17th, and 19th are

13
B. Amini et al. International Journal of Electrical Power and Energy Systems 156 (2024) 109738

Fig. 14. The performance under transient response with 50 % load changes: a) ch1: Grid current - ch4: Load current at start time, b) ch1: Load current, ch4: Grid
current during the system operation.

Fig. 15. The performance under step changes of DC-link voltage: a) DC-link voltage, b) ch1: DC-link voltage - ch4: network current.

greatly weakened. These resualts confirm the proper performance of the by providing sinusoidal current. Just as a problem, there are some low
SAPF system based on the proposed PR controller. fluctuations in the DC-Link voltage.
On the other hand, to evaluate the dynamic response of the proposed The stability analysis and frequency response analysis of the system
method, the sudden activation of the SAPF harmonic and reactive based on the results of the Bode diagram of the optimal PR and PR
compensator has been done at the moment of t = 0.4 s and the results are controllers are shown in Fig. 9. It can be seen that the Bode plot of both
shown in Fig. 6. The proposed control method with high dynamic controllers are stable with positive Gain Margin (G.M) and Phase Margin
response compensates for the changes immediately with a very low (P.M). But the proposed controller has experienced better conditions in
delay (within a few milliseconds) and confirms the proper performance terms of stability than the conventional type.
of the SAPF system. For better display, the outputs are given for the load,
grids and SAPF currents only include one phase and the other phases
4.2. Experimental results
have the same shape.
The results shown in Fig. 7 are to further investigate the dynamic
A prototype three-phase SAPF is built to validate the simulation re­
response of the SAPF under a load value change to 50 % of the nominal
sults and verify the effectiveness of the proposed control method, which
amplitude value at the moment of starting the system from zero and
is shown in Fig. 10 and its information is listed in Table 5. The SAPF is
applied at the moment of t = 0.6 s. Also, according to (Fig. 7-a), the load
placed between the source and the nonlinear load, and the control sys­
current changes from 0 to 100 %. Meanwhile, the SAPF controllers have
tem is implemented in the Digital Signal Processing (DSP) board. The
a good dynamic and transient response under load changes and the
power converter is a 3-phase Full Bridge VSI whose power module
compensation is very fast (under a few milliseconds). Only during the
consists of three IGBT legs. The prototype driver is used as the main
applying of the load change for a short moment, fluctuations in the DC-
control board of the power switchs for apply the DSP commands to the
link voltage of about 2 % overshoot and 2 % undershoot are observed.
IGBTs. The DSP controller needs a few different signals, three voltage
Also, sometimes the grid voltage may have non-ideal conditions such
signals (line voltage for the PLL) that go into the DSP board through
as unbalance condition or voltage drop. Since the SAPF is connected to
step-down transformers which are connected to the Analog-to-Digital
the grid in the PCC, this issue affects the performance of the SAPF.
Converter (ADC) and power supply IC (TPS767D301) of DSP board.
Therefore, to check this mode, the simulation is performed under un­
Six current signals (three filter output currents and three load currents).
balance condition and voltage drop, the results of which are shown in
On the other hand, LEM sensors are used to measure line, load and SAPF
Fig. 8. In this case, the proposed control method compensates the har­
currents. DC bus and DC capacitors are also available. In this part, MKP
monic components and the controller maintains its normal performance
capacitors can be used to reduce the spikes of IGBTs. Voltage

14
B. Amini et al. International Journal of Electrical Power and Energy Systems 156 (2024) 109738

Fig. 16. The experimental results of SAPF based on the proposed PR controller under unbalanced voltage condition: a) Unbalance grid voltage, b) Unbalance grid
voltage diagram, c) load currents, d) THD of grid currents.

Table 6
Comparison results.
Control methods Conventional PR Proposed optimal PR

THD% (grid current) 8.3 % for R-L load 4.2 % for R-L load
12.7 % for R-C load 4.5 % for R-C load
IAE 0.814 0.627
P.F > 0.96 > 0.99
rip% 1.65 1.34
Accuracy High Very High
Dynamic speed High Very High
Robustness High Very High
Flexibility High High
Stability analysis Good Excellent
Transient response Good Excellent
Complexity Medium Medium
Controller error Low Very low
Chattering effect NO NO
Convergence time Fast Fast
Standard Not passed Passed
Tuning method Ziegler-Nichols Optimization (GA)

Fig. 17. Robustness evaluation of proposed PR controller under model

parameter mismatches (ΔLf). is normal for real networks. However, the harmonic analysis of its orders
is given in (Fig. 11-b). Also, the three-phase currents of the nonlinear,
THD of load currents, the three-phase grid currents, and THD of grid
transformers are also used to measure voltages. SAPF requires three
currents after compensation are respectively shown in (Fig. 11-c),
series inductors to connect to the grid. In addition, the nonlinear load is
(Fig. 11-d), (Fig. 11-e), and (Fig. 11-f). As well as, the power factor (PF)
modeled by a three-phase diode bridge rectifier with a resistive load.
before and after the compensation shown in (Fig. 11-g), and (Fig. 11-h).
Likewise, other tools and components such as fuses, switches, heat sinks,
also the harmonic analysis of three-phase grid current is given in
multimeters and measurements are also connected to the system.
(Fig. 11-i), which according to the data provided by the FFT analyzer, it
Finally, the necessary outputs are measured by the power analyzer.
can be seen that the %THD of the grid current using the proposed
The experimental results of SAPF based on the proposed PR
method is reduced from 20.2 % to 4.2 % that has passed the desired
controller are shown in Fig. 11. The three-phase voltage of the balanced
standard.
grid is given in (Fig. 11-a), which has a small amount of harmonic, which

15
B. Amini et al. International Journal of Electrical Power and Energy Systems 156 (2024) 109738

So far, most of the studied systems have used diode-bridge rectifiers Also, the robustness of the control method, which is one of the most
with R-L load as nonlinear load simulators. Therefore, for a more ac­ important challenges, depends on the change of inductor and capacitor
curate investigation, the proposed controller has been tested by diode- values over time. The inductor value of the filter (Lf) is the main
bridge rectifier with R-C load. According to the results of Fig. 12 seen parameter of the model in the final control signals, and the robustness of
that under the R-C load, the amount of harmonic distortions and THD the control method varies accordingly, which should be evaluated by
level has increased greatly. But in this condition, the proposed method changing Lf. Hence, ΔLf changed from − 20 % to + 20 % to check the
has a very good performance and again the THD of the grid side current behavior of the proposed system under these conditions. The results of
has significantly decreased from 56.1 % to 4.5 % that again has passed this investigation are shown in Fig. 17 and it is clear that the proposed
the desired standard. method is very robust against the model mismatch. Because as the main
The comparison of the experimental results relative to each other in performance index, the %THD of grid current changed slightly and
the condition of steady-state operation for one phase is shown in Fig. 13, Power Factor (P.F) on the grid side remained fixed at one.
and the other phases are similar. These diagrams include load, grid, and
SAPF currents that are relative to each other, which are shown in 4.3. The performance comparison study
(Fig. 13-a), (Fig. 13-b), (Fig. 13-c), and (Fig. 13-d) respectively. Ac­
cording to these graphs, the fast and accurate compensation of nonlinear Finally, to better understand and clearly show the differences be­
load harmonics and reactive power by SAPF can be seen. Based on the tween the proposed method and similar conventional methods, in the
quality factor which is the %THD of the grid current, the tracking per­ same conditions and same system parameters settings for both consid­
formance of this method shows the harmonic reduction and waveform ered methods according to Table 4 and Table 5, a complete comparison
correction of the grid side current. Also, according to (Fig. 13-e) and has been made through various performance indicators, and the com­
(Fig. 13-f), reactive power is fully compensated by creating the coordi­ parison results are given in Table 6. Based on this table, the proposed
nation of sinusoidal current with the grid voltage. All these results fully control method has been compared with the conventional PR method
confirm the optimal performance of the SAPF compensator based on the according to the main control measurement criteria each of which has
proposed control method. different characteristics. Meanwhile, the proposed optimized PR by
Transient response is important during system startup and load providing infinite gain at the desired set frequency and eliminating the
changes. In (Fig. 14-a), the results under the transient response at the steady-state error with a simple structure and implementation capability
system startup and in (Fig. 14-b) the results under the transient response as a perfect current control strategy has a much better performance in
during the system operation with 50 % step change in load are presented the harmonic compensation process than the conventional PR method.
to better evaluate the performance of the proposed method. The tran­ Because it provides better performance for the current control system
sient response is mostly influenced by the performance of the control and SAPF compensation process to better switching by improved
system. Accordingly, the control unit detects the harmonic components switching signals. As a result, it leads to a significant reduction of %THD
and compensates in a few milliseconds, which ensures a fast transient for the grid current. Also, the proposed method emphasizes its high
response of the control unit. In general, in the harmonic detection and reference traceability by providing lower IAE. It should be noted that
control system, the transient response is determined according to the IAE is defined for SAPF output current, but %rip is calculated on the grid
cut-off frequency and the order of the LPF. Due to the high sampling side. In addition, the proposed method offers low %THD and unit power
frequency of the control system and the low order of the LPF, the factor on the grid side, minimal dependence on external parameters,
filtering delay and transient response are very low. good compatibility for digital implementation, has a robust, fast, accu­
Sometimes, changes in load or system requirements may cause rate, stable, and more favorable performance than its conventional type.
changes and fluctuations in the DC-link voltage. For this reason, it is
necessary to check how quickly the proposed controller is able to restore 5. Conclusion
the reference DC voltage and what effect these changes and DC-link
fluctuations have on the compensation process. Therefore, by chang­ Eliminating or reducing harmonics is an important issue in
ing the 50 V step in the reference value of the DC-link voltage, the improving power quality. As a suitable tool among different solutions,
performance of the system has been checked and the results are shown in SAPF is the most widely used and the best choice for compensating
Fig. 15. According to this figure, it is clearly visible that the DC-link current harmonics caused by nonlinear loads. However, the control al­
voltage quickly reaches its new value without overshoot and under­ gorithms of this system are the most important factor affecting the
shoot with very slight fluctuations in a very short time (a few millisec­ performance of SAPF, which is very important and sensitive. Harmonic
onds) and good conditions of settling time. However, according to the current compensation processes must be performed simultaneously, fast
grid current curve, the compensator system has a good performance and and accurately, so proper design and optimal control should be
experiences only a very small amount of current fluctuation in a short considered for them in order to improve the overall compensation per­
moment (about 4 ms), which is completely normal for DC-link capacitor formance. In this paper, while providing an overview of the working
voltage charging moment. principle of SAPF, an optimized PR controller is proposed for current
Sometimes the grid voltage has distortions and non-ideal conditions control. In addition, the parameters of the proposed PR controller to
such as unbalanced conditions or voltage drop. Therefore, in this situ­ reach the most optimal tuning mode by GA are presented. Fast and ac­
ation, to achieve the desired compensation and create a performance curate dynamic response in compensation signal generation, good co­
independent of grid voltage distortion, to generate the suitable reference ordination with the SAPF system, robust performance to uncertainties,
current by the harmonic identification algorithm, the positive sequence and effective dealing with grid current harmonics are the advantages of
components of the main voltage component are used for calculations the proposed method. For the proposed system according to the IEEE-
based on the Positive Sequence Detector (PSD). The experimental results 519 standard, the approach is to reduce the grid current THD to less
of system performance under unbalanced voltage conditions are shown than 5 %. Several different simulations and experimental results were
in Fig. 16. According to the data provided by these figures, it can be seen presented in a real system that has satisfactory performance, in which
that the %THD of the grid current using the proposed method is reduced the %THD of the grid current has been reduced from 20.2 % to 4.2 % for
and the desired performance of compensation is presented. This is even the R-L load and from 56.1 % to 4.5 % for the R-C load. Also, according
though only in the phase where there was more of an unbalanced to the mathematical and stability analysis, the proposed method has
voltage drop, the %THD value increased slightly, which does not have a experienced good conditions in terms of stability. In total, the performed
noticeable effect on the overall result, and the system still again has evaluations based on simulation and experimental results, confirm
standard performance. better reference tracking, proper harmonic compensation, reduction of

16
B. Amini et al. International Journal of Electrical Power and Energy Systems 156 (2024) 109738

the IAE performance index, and high robustness of the proposed Symposium on Power Electronics for Distributed Generation Systems (PEDG).
IEEE; 2018.
method. Therefore, the improved and superior performance of the
[13] Han Y, Xu L, Khan MM, Yao G, Zhou L-D, Chen C. Study on a novel approach to
proposed method is fully confirmed by the performance indexes and the active power filter control using neural network-based harmonic identication
obtained results. scheme. Elect Eng 2010;91(6):313325.
To sum up, the main contributions of this paper can be listed as [14] Eskandarian N, Beromi YA, Farhangi S. “Improvement of dynamic behavior of
shunt active power filter using fuzzy instantaneous power theory.”. J Power
follows: Electron 2014;14(6):1303–13.
[15] Chen D, et al. A harmonics detection method based on triangle orthogonal
• Providing a complete and comprehensive procedure in the design of principle for shunt active power filter. Energy Rep 2021;7:98–104.
[16] Sundaram E, Venugopal M. On design and implementation of three phase three
SAPF systems and their control principles. level shunt active power filter for harmonic reduction using synchronous reference
• Applying PR controller for control of SAPF interfacing grid and frame theory. Int J Electr Power Energy Syst 2016;81:40–7.
nonlinear load based on the current control system for effective [17] Büyük M, et al. Improved instantaneous power theory based current harmonic
extraction for unbalanced electrical grid conditions. Electr Pow Syst Res 2019;177:
compensation of current harmonics and improve system 106014.
performance. [18] Montanari AA, Gole AM. Enhanced instantaneous power theory for control of grid
• Utilization of GA for tuning proposed PR control parameters to reach connected voltage sourced converters under unbalanced conditions. IEEE Trans
Power Electron Nov. 2016;32(8):6652–60.
the most optimal state of control system adjustment. [19] Maurya I, Gupta SK, Maurya P. An efficient harmonic detection approach for shunt
• Mathematic and stability analysis of the proposed PR control used in active filter based on wavelet transform. Ain Shams Eng J 2018;9(4):2833–9.
the SAPF system. [20] Liu H, et al. Fast and flexible selective harmonic extraction methods based on the
generalized discrete Fourier transform. IEEE Trans Power Electron 2017;33(4):
• Practical implementation and experimental test of the SAPF device
3484–96.
based on the proposed PR control method. [21] Pandey A, Singh A. SWRDFT Controller for Single Phase Grid Tied PV system with
Low Voltage Ride through Capability. Electr Pow Syst Res 2023;214:108878.
[22] Guo Z, et al. An impedance identification-based islanding detection method for
CRediT authorship contribution statement
droop control inverters based on RDFT and positive–negative notch filters. Energy
Rep 2022;8:1060–70.
Behnam Amini: Investigation, Methodology, Software, Writing – [23] Zhou J, Yuan Y, Dong H. Adaptive DC-link voltage control for shunt active power
filters based on model predictive control. IEEE Access 2020;8:208348–57.
original draft. Hasan Rastegar: Supervision, Writing – review & edit­
[24] Viji A. Jeraldin, Aruldoss Albert Victoire T. “Enhanced PLL based SRF control
ing. Mohammad Pichan: Conceptualization, Investigation, Methodol­ method for UPQC with fault protection under unbalanced load conditions.” Int J
ogy, Software, Supervision. Elect Power Energy Syst 58 (2014): 319-328.
[25] Escobar G, et al. “Nonlinear stability analysis of the conventional SRF-PLL and
enhanced SRF-EPLL.”. IEEE Access 2021;9:59446–55.
Declaration of competing interest [26] Saxena H, Singh A, Rai JN. Design and Testing of Frequency Adaptive Zero-
Crossing Detector as a Synchronizing Technique. 2020 IEEE 9th Power India
International Conference (PIICON). IEEE; 2020.
The authors declare that they have no known competing financial [27] Hoon Y, Mohd Radzi MA, Mohd Zainuri MA, Zawawi MA. Shunt active power
interests or personal relationships that could have appeared to influence filter: A review on phase synchronization control techniques. Electronics 2019 Jul
the work reported in this paper. 15;8(7):791.
[28] Fallah M, et al. A modified indirect extraction method for a single-phase shunt
active power filter with smaller DC-link capacitor size. Sustain Energy Technol
Data availability Assess 2021;45:101039.
[29] Komurcugil H. Double-band hysteresis current-controlled single-phase shunt active
filter for switching frequency mitigation. Int J Electr Power Energy Syst 2015;69:
Data will be made available on request. 131–40.
[30] Pichan M, Seyyedhosseini M, Hafezi H. A New DeadBeat-Based Direct Power
References Control of Shunt Active Power Filter With Digital Implementation Delay
Compensation. IEEE Access 2022;10:72866–78.
[31] Ouchen S, Steinhart H, Benbouzid M, Blaabjerg F. ‘‘Robust DPC-SVM control
[1] Mikkili, Suresh, and Anup Kumar Panda. Power quality issues : current harmonics.
strategy for shunt active power filter based on H∞ regulators,’’ Int. J. Electr. Power
CRC press, 2018. https://doi.org/10.1201/9781315222479.
Energy Syst., vol. 117, May 2020, Art. no. 105699.
[2] Chilipi R, et al. Third order sinusoidal integrator (TOSSI)-based control algorithm
[32] Ray AS, Bhattacharya A. Improved tracking of shunt active power filter by sliding
for shunt active power filter under distorted and unbalanced voltage conditions. Int
mode control. Int J Electr Power Energy Syst 2016;78:916–25.
J Electr Power Energy Syst 2018;96:152–62.
[33] Thentral TMT, et al. “Development of control techniques using modified fuzzy
[3] Meloni LF, Tofoli FL, Rezek AJ, Ribeiro ER. “Modeling and experimental validation
based SAPF for power quality enhancement.”. IEEE Access 2021;9:68396–413.
of a single-phase series active power filter for harmonic voltage reduction,” IEEE.
[34] Chen J, Shao H, Liu C. An improved deadbeat control strategy based on repetitive
Access 2019;7.
prediction against grid frequency uctuation for active power filter. IEEE Access
[4] Roldan-Perez J, Rodriguez-Cabero A, Prodanovic M. “Harmonic Virtual Impedance
2021;9:2464624657.
Design for Parallel-Connected Grid-Tied Synchronverters”. IEEE J Emerg Sel Top
[35] Çelik D. Lyapunov based harmonic compensation and charging with three phase
Power Electron 2019;7(1):493–503.
shunt active power filter in electrical vehicle applications. Int J Electr Power
[5] Ouchen S, Betka A, Gaubert JP, Abdeddaim S. Simulation and real time
Energy Syst 2022;136:107564.
implementation of predictive direct power control for three phase shunt active
[36] Bajaj M, Flah A, Alowaidi M, Sharma NK, Mishra S, Sharma SK. A Lyapunov-
power filter using robust phase-locked loop. Simul Model Pract Theory 2017;78:
function based controller for 3-phase shunt active power filter and performance
1–17.
assessment considering different system scenarios. IEEE Access 2021;9:66079–102.
[6] Zangeneh Bighash E, Sadeghzadeh SM, Ebrahimzadeh E, Blaabjerg F. Adaptive-
[37] Yepes AG, et al. High-performance digital resonant controllers implemented with
Harmonic Compensation in Residential Distribution Grid by Roof-Top PV Systems.
two integrators. IEEE Trans Power Electron 2010;26(2):563–76.
IEEE J Emerg Sel Top Power Electron Dec. 2018;6(4):2098–108.
[38] Herman, Leopold, Igor Papic, and Bostjan Blazic.“A proportional-resonant current
[7] Ouchen S, Betka A, Gaubert JP, Abdeddaim S, Mazouz F. “Fuzzy-Direct Power
controller for selective harmonic compensation in a hybrid active power
Control of a Grid Connected Photovoltaic System Associate with Shunt Active
filter.”IEEE Transactions on Power Delivery 2014.
Power Filter.” In: International Conference on Artificial Intelligence in Renewable
[39] Lenwari W. Optimized design of modified proportional-resonant controller for
Energetic Systems, 2018, pp. 164–172.
current control of active filters. 2013 IEEE International Conference on Industrial
[8] Das JC. -Passive filters-potentialities and limitations,‖ IEEE Trans. Ind. Appl, 2004.
Technology (ICIT). 2013.
[9] Ouchen S, et al. Direct power control of shunt active power filter using space vector
[40] Parandhaman B, Nataraj SK, Baladhandautham CB. Optimization of DC-link
modulation based on super twisting sliding mode control. IEEE J Emerg Selec Top
voltage regulator using Bat algorithm for proportional resonant controller-based
Power Electron 2020;9(3):3243–53.
current control of shunt active power filter in distribution network. Int Trans Elect
[10] Revuelta, Patricio Salmerón, Salvador Pérez Litrán, and Jaime Prieto Thomas.
Energy Syst 2020;30(6):e12369.
Active power line conditioners: design, simulation and implementation for
[41] Sou W-K, et al. Analysis, Design, and Implementation of Multi-Quasi-Proportional-
improving power quality. Academic Press, 2016. https://doi.org/10.1016/C2014-
Resonant Controller for Thyristor-Controlled LC-Coupling Hybrid Active Power
0-02915-2.
Filter (TCLC-HAPF). IEEE Trans Ind Electron 2021;69(1):29–40.
[11] Hoon Y, Radzi MAM, Hassan MK, Mailah NF, Wahab NIA. A simplified
[42] Institute of Electrical and Electronics Engineers (IEEE). IEEE recommended
synchronous reference frame for indirect current controlled three-level inverter-
practice and requirement for harmonic control in electric power systems. In IEEE
based shunt active power filters. 2016.
Std 519-2014 (Revision of IEEE Std 519-1992); IEEE: Piscataway, NJ, USA, 2014.
[12] Ferrari M. Evaluation of harmonic detection algorithms for active power filter
based on voltage and current detection methods. 2018 9th IEEE International

17
B. Amini et al. International Journal of Electrical Power and Energy Systems 156 (2024) 109738

[43] Teodorescu R, et al. Proportional-resonant controllers and filters for grid-connected [46] Gowtham N, Shobha Shankar. “PI tuning of Shunt Active Filter using GA and PSO
voltage-source converters. IEE Proc-Electric Power Appl 2006;153(5):750–62. algorithm.” 2016 2nd International Conference on Advances in Electrical,
[44] Amini B, et al. Interface converter control of distributed generation in microgrids Electronics, Information, Communication and Bio-Informatics (AEEICB). IEEE,
using fractional proportional—Resonant controller. Electr Pow Syst Res 2021;194: 2016.
107097. [47] Thanh HP, et al. Optimizing parameters of the shunt active power filter using
[45] Michalewicz Z, Michalewicz Z. GAs: Why Do They Work? Berlin Heidelberg: Genetic algorithm. 2017 9th International Conference on Knowledge and Systems
Springer; 1996. Engineering (KSE). 2017.

18