10808 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 34, NO.

11, NOVEMBER 2019

Performance Investigation of Multifunctional

On-Grid Hybrid Wind–PV System With
OASC and MAF-Based Control
Subarni Pradhan , Student Member, IEEE, Shadab Murshid , Member, IEEE, Bhim Singh, Fellow, IEEE,
and Bijaya Ketan Panigrahi, Senior Member, IEEE

Abstract—An observer-based adaptive speed control (OASC) CMS Converter of machine side.
and a multistage adaptive filter (MAF) based control structure are MAF Multistage adaptive filter.
proposed for an on-grid hybrid wind–photovoltaic (PV) system to MPPT Maximum power point tracking.
deal with the two major issues such as peak wind power extraction
and mitigation of power quality problems, respectively. The OASC OASC Observer-based adaptive speed control.
is an adaptive control alongside exhibiting robustness against un- PI controller Proportional and integral controller.
certainties (structured and unstructured). The objective of peak P&O Perturb and observe.
wind power extraction is met through cascaded control, which in- PMSG Permanent magnet synchronous generator.
cludes an inner hysteresis current control and an outer speed con- PV Photovoltaic.
trol based on OASC. The proposed OASC includes a disturbance
observer loop with backstepping control. Moreover, it incorporates PVECS Photovoltaic energy conversion system.
a discontinuous projection law-based adaptive parameter estima- RES Renewable energy sources.
tion, thereby resulting in a sense of hybrid control implementation. SMC Sliding mode control.
The outer speed control loop provides the reference stator current SOGI Second-order generalized integrator.
to the inner loop retaining maximum power point tracking. For VSC Voltage source converter.
enhancement of system reliability, another renewable source (so-
lar PV array) is integrated at the dc link. The connection between WECS Wind energy conversion system.
the hybrid system and the grid is established through an inverter WT Wind turbine.
incorporating the proposed MAF-based control. Herein, the effects
of nonlinear balanced and unbalanced loads are addressed through I. INTRODUCTION
the incorporation of vectorial approach for extracting fundamental
HE concerns for pollution and the very realness of global
positive sequence components from nonlinear load currents to real-
ize harmonic-free fundamental reference currents for the grid. The
proposed controls are simulated and compared with the conven-
T warming have pushed the renewables to be branded as the
energy sources of the future. The consistent decline of polar ice
tional techniques. The control implementation and performance cover over the years, for instance, calls for urgency of appropri-
testing are carried out on the hybrid system built in the laboratory.
ate climate policy action and nations across the world have been
Index Terms—Adaptive filter, disturbance observer, power qual- responding to the challenge. Furthermore, the proof that the
ity, projection law, solar photovoltaic (PV) array, total harmonic market is getting ready to embrace the RES is evident from the
distortion (THD), wind energy conversion system (WECS). drastic decline of cost per unit generation from renewables like
wind and solar in recent times. Due to the fluctuating nature of
NOMENCLATURE
these renewables, it is quite reliable to implement hybrid gener-
AVF Adaptive vectorial filter. ating systems [1], operated as grid coupled or autonomous with
CCP Common coupling point. energy management methods. Power electronics interfaces are
CGS Converter of grid side. the reliable means to extract effective power from these renew-
ables and integrate them to the grid. Peak power extraction from
Manuscript received November 7, 2018; revised January 11, 2019; accepted these renewables under climatic and geographical conditions is
February 6, 2019. Date of publication February 21, 2019; date of current ver- the primary matter of concern [2]. Furthermore, integration of
sion August 29, 2019. This work was supported in part by the Department of these RES-based distributed generations to the utility grid faces
Science and Technology, Government of India, under Grant RP03357 and Grant
RP03391 and in part by J. C. Bose Fellowship under Grant RP03128. Recom- technical problems such as instability and power quality issues
mended for publication by Associate Editor L. Chang. (Corresponding author: [3], associated with the grid or uncertain loads [4]. This very
Subarni Pradhan.) circumstance demands implementation of novel and improved
The authors are with the Department of Electrical Engineering, In-
dian Institute of Technology Delhi, New Delhi 110016, India (e-mail:, control techniques for assured system reliability, stability, and
[email protected]; [email protected]; bhimsinghiitd62@gmail. enhanced performance [5]. The aforesaid problems have been
com; [email protected]). addressed previously with modification and improvement in cir-
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. cuit arrangements and introduction of linear, nonlinear, robust,
Digital Object Identifier 10.1109/TPEL.2019.2900759 and composite controls.

0885-8993 © 2019 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications standards/publications/rights/index.html for more information.

Authorized licensed use limited to: Indian Institute of Information Technology Design & Manufacturing. Downloaded on November 21,2021 at 15:17:10 UTC from IEEE Xplore. Restrictions apply
 PRADHAN et al.: PERFORMANCE INVESTIGATION OF MULTIFUNCTIONAL ON-GRID HYBRID WIND–PV SYSTEM 10809

Commonly, controllers for wind energy extraction are de- variation is assumed as bounded disturbance to the WECS. The
signed using P&O, intelligent, adaptive, and robust controls [1]. disturbances are estimated and included in the speed control
In the light of feedback linearization scheme, a control logic has loop as compensated feedforward term. The parameter adap-
been developed to deal with optimum wind power extraction tation logic based on discontinuous projection law is adopted
from variable-speed WT and to provide active/reactive power inclusive of the disturbance observer for dealing with the param-
support to the grid [6]. Its performance has also been tested for eter uncertainties and external disturbances. The proposed con-
system faults and low-voltage ride through. Usually, PI con- trol ensures the accuracy in speed tracking and convergence of
troller is used in speed control loop, which is widely acceptable tracking error without sacrificing the robustness and other con-
in the industries. However, its transient performance in the pres- straints during system transients. The proposed OASC is used in
ence of disturbances is not promising [7]. Exploring the similar the speed control loop of PMSG-based WECS that regulates the
problem, robust controls like H∞ and SMC are proposed in [7] PMSG speed according to the reference speed satisfying MPPT.
and [8]. A robust high-order SMC is presented in [9], which has The output of this loop is treated as the reference q-axis stator
the potential to perform under disturbances and model uncer- current, which is to be controlled. Although the PMSG torque
tainties. In this case, the sliding mode gain must be chosen to and output power are dependent on the q-axis stator current,
maintain tradeoff between robustness and chattering. This prob- its regulation helps in achieving regulated output power at the
lem of SMC has been addressed by introducing an additional PMSG stator terminals analogous to variable wind speed. The
observer loop [10]. Herein, disturbance is estimated separately d-axis stator current is taken as zero for implementing vector
and added as a feedforward term, which alleviates the need of control.
high gain selection and helps in chattering reduction. The con- The design of MAF is based on AVF that utilizes vectorial
cept of additional disturbance observer loop has been used for a approach for extracting fundamental positive and/or negative
novel PI-type current control design [11]. Moreover, observer- sequence components from polluted signals. It is a different
based adaptive techniques such as feedback linearization have approach for extraction of selective harmonics, where multiple
been used for designing robust speed control of electrical numbers of AVF structures are used for eliminating lower order
machines [12]. harmonics in selective way. The MAF estimates the fundamen-
Numerous linear, nonlinear, adaptive, and robust techniques tal positive sequence components from nonlinear load currents.
have been proposed for grid synchronization, harmonics miti- The fundamental components are then used for balanced ref-
gation, grid currents balancing with power factor improvement, erence grid current generation. This control technique is used
reactive power, and load compensation in grid-connected op- for dealing with harmonics mitigation and load compensation.
eration of individual or hybrid distributed generating systems Balanced grid currents with improved power factor under steady
[13], [14]. Power quality improvement of on-grid WECS us- as well as dynamic conditions are achieved for the on-grid hy-
ing conservative power theory under nonlinear load is proposed brid system with proposed control. Moreover, controlled active
in [4] with insight knowledge of harmonics mitigation and ac- power flow to the grid is achieved without sacrificing the grid
tive/reactive power flow control. Similar objectives have been code requirements. The control implementation and the system
addressed for grid-connected solar PV system in [15]. Kalman performance evaluation are carried out on the hybrid system
estimator has been proposed for estimating fundamental in- built in the laboratory and verified under varying wind speed
phase and quadrature components from distorted and flickered and solar insolation change, balanced and unbalanced nonlinear
grid voltage signals [16]. However, it encounters complexity loads.
and burden during control implementation due to the presence
of derivative terms. Over the decades, the concept of general-
ized integrator (GI) has been extensively adopted for designing II. MODEL DESCRIPTION
adaptive filters to estimate fundamental and other harmonics Fig. 1 presents the adopted on-grid hybrid wind–PV system,
components from distorted and harmonics polluted voltage sig- where the renewables (wind and solar) are considered as the
nals. Some of them are SOGI, damped SOGI [17], SOGI with sources of power generation. The dc link is the common point
quadrature signal generator (SOGI-QSG), second-order SOGI- of power conversion, where RES are connected and it is to
QSG, multiple SOGI (MSOGI) [18], and multilayered fifth- be maintained for adequate power conversion. PMSG-based
order GI (MFOGI) [19]. A different approach for fulfilling the WECS is connected at the dc link though a fully controlled
similar objectives, called AVF, has been proposed in [20], which three-phase VSC known as the CMS. The solar PVECS consists
follows vectorial representation. of a solar PV array and its terminals are connected to the dc link
An investigation of proposed control is performed on the without the use of any other power conversion stage; hence,
hybrid PMSG-based WECS and PVECS, functioning under on- single-stage configuration is realized. The typical method of
grid mode. The objectives of control design include wind en- PMSG speed regulation is considered to have control over the
ergy extraction from WT at variable wind speed by varying the wind power generation at varying wind speed, ensuring the
PMSG speed according to the reference speed retaining MPPT typical speed–power characteristics of the WT [2]. An OASC-
and power quality improvement. These objectives are met by based control is implemented on CMS for harvesting effective
generating switching sequence using OASC for the CMS and a wind power through PMSG speed regulation. Extraction of peak
MAF-based control for the CGS. solar power is ensured by applying a P&O MPPT technique [13].
The structured and unstructured uncertainties are dealt with The output of MPPT algorithm is the reference dc link voltage,
the OASC-based control used for CMS. The wind speed which is used for dc link control. During the night time, the

Authorized licensed use limited to: Indian Institute of Information Technology Design & Manufacturing. Downloaded on November 21,2021 at 15:17:10 UTC from IEEE Xplore. Restrictions apply
 10810 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 34, NO. 11, NOVEMBER 2019

Fig. 1. Schematic of an on-grid hybrid wind–PV system.

solar power generation goes to zero and the reference dc link quite crucial. Owing to the vector control implementation, the
voltage is set at the nominal value. Owing to the advantages of d-axis current is chosen as zero.
single-stage arrangement, the voltage at the PV array terminal is The presence of external disturbances and parametric uncer-
kept equal to that of dc link [14]. The aggregated wind and solar tainties is the means of varying pulsating shaft toque and leads
power is processed through a power conversion unit (known as to instability. It becomes indispensable to reduce their adverse
the CGS), interfacing inductor Li and fed to the load and the grid effect through introduction of additional loops in the control
at the CCP. The CGS is triggered through the proposed control design. The OASC includes backstepping speed control loop
to achieve dc link control and power quality improvement. with a parameter adaption based on discontinuous projection
The use of interfacing inductors reduces the ripples in the law that has the proficiency to reduce the negative impact of
currents flowing to and from the grid. The ripple filters are parametric uncertainties [12]. It is also equipped with a dis-
used to mitigate the high-frequency switching transients. An turbance observer that avoids the need of high feedback gain.
uncontrolled diode bridge rectifier connected with RL load is The goal of speed control design is to track the reference speed
used as a nonlinear load. ωr with high precision and nearly zero steady-state error. The
actual speed of the PMSG is estimated rather than sensed by a
costly mechanical sensor [2]. The estimated parameter ω est is
III. CONTROL TECHNIQUES considered as the PMSG speed for further analysis.
The primary objectives of the proposed controls are addressed The motion equation of PMSG is given as [2], [3]
by generating switching sequence for CMS and CGS using
ω̇m = (Tw t − Te − F ωm ) /J (1)
proposed OASC-based PMSG speed control and MAF-based
control. The P&O MPPT technique is adopted for extracting where ωm is the mechanical speed of rotor, J is the moment of
effective solar power from a solar PV array. It provides the inertia, and Tw t , Te , and F ωe are the turbine torque, electrical
reference dc link voltage that is to be controlled by CGS. In torque, and frictional torque, respectively; Te = (3/2)pϕm iq =
the absence of solar irradiation or during the night time, the Cte iq , where p, ϕm , iq , and Cte are number of pole pair, perma-
reference dc link voltage is set as nominal value. nent magnet flux, q-axis stator current, and the constant asso-
ciated with electromagnetic torque, respectively. The PMSG is
mechanically coupled to the WT that is driven by variable wind
A. Signal Generation for CMS speed.
The effective wind power from the WT is extracted by reg- Simplifying and rearranging the motion equation (1), the
ulating the PMSG speed according to the speed reference gen- speed dynamics of PMSG is given as
erated by the MPPT technique that is independent of turbine Cte F Tw t
dynamics. The CMS is controlled through vector control, which ω̇est = − iq − ωest + . (2)
J J J
is a cascaded control with inner hysteresis current control and
outer speed control. For a surface PMSG, the electrical power Replacing iq with its reference value iq r and assuming Tw t
is a function of torque, which depends on the q-axis stator cur- as external disturbance to the system, which is bounded, state
rent. Therefore, the basic idea is to regulate the PMSG torque space form is reframed from (2) as
by means of stator current regulation, which in turn regulates ẋ1 = ω̇est = −ϑ1 iq r − ϑ2 ωest + d(ωest , t) (3)
the generated power. The reference stator current of q-axis is
the output of speed control loop, which proves the control over where ϑ = [ϑ1 , ϑ2 ]T = [Cte /J, F/J]T represents the param-
electrical power through speed regulation. Hence, precise speed eter set and d(ωest , t) = Tw t /J is the lumped disturbance
control with fast control and accurate convergence of error is to the system, which is unstructured uncertainty. A large

Authorized licensed use limited to: Indian Institute of Information Technology Design & Manufacturing. Downloaded on November 21,2021 at 15:17:10 UTC from IEEE Xplore. Restrictions apply
 PRADHAN et al.: PERFORMANCE INVESTIGATION OF MULTIFUNCTIONAL ON-GRID HYBRID WIND–PV SYSTEM 10811

variation in the system parameters (Cte , J, F ) subjects to struc- Taking derivative of (8) and applying (9) yields
tured uncertainty.
ż1 = ω̇est − ω̇r
1) Introduction of Projection-Based Parameter Adaptation:
The negative impact of parametric uncertainties is dealt with ż1 = −ϑ1n iq r − ϑ2n ωest + x2 − ω̇r . (10)
introducing parameter adaptation technique. An assumption is
made for the parameters that satisfies By considering a positive feedback gain σ 1 , the control input
is derived as
Δ
ϑ ∈ Ωϑ = {ϑ : ϑm in ≤ ϑ ≤ ϑm ax } (4) 1
iq r = − (ω̇r + ϑ2n ωest − σ1 z1 − x2 ) (11)
ϑ1n
where ϑm in = [ϑ1m in , ϑ2m in ]T and ϑm ax = [ϑ1m ax , ϑ2m ax ]T
are lower and upper bounds of parameter vector ϑ, respectively. and the speed error dynamic equation is defined as
The estimate of parameter vector ϑ and estimated error ż1 = −σ1 z1 . (12)
are denoted as ϑ̂ and ϑ̃ = ϑ̂ − ϑ, respectively. A projection-
based parameter adaptation law is used for estimated parameter It is not always possible to get the exact idea about the uncer-
vector as tainties and disturbances defined in (9). Hence, x2 is observed
with the help of an observer and (11) is redefined as
˙
ϑ̂ = Projϑ̂ (Γη) , with ϑm in ≤ ϑ̂(0) ≤ ϑm ax (5)
1
iq r = − (ω̇r + ϑ2n ωest − σ1 z1 − x̂2 ) . (13)
where η is the adaptation function to be derived later, Γ > 0 is ϑ1n
the adaptation rate matrix, and Projϑ̂ (•n ) is the discontinuous By considering the estimated system parameters and their
projection law, which is defined as errors, (3) is written as
⎧
⎪
⎨0, if ϑ̂n = ϑn m ax ; •n > 0 ω̇est = −ϑ̂1 iq − ϑ̂2 ωest + x2
Projϑ̂ n (•n ) = 0, if ϑ̂i = ϑn m in ; •n < 0 (6)
⎪
⎩ ẋ2 = a(t)
•n otherwise
x2 = ϑ̃1 iq + ϑ̃2 ωest + d(ωest , t) (14)
where •n signifies the nth component of vector • with n = 1, 2.
The parameter adaptation law defined in (5) holds the following where a(t) is the derivative of x2 with respect to time. The
properties for any adaptation function η as designed disturbance observer is given as
  ω̂˙ est = −ϑ̂1 iq − ϑ̂2 ωest + x̂2 − 2θ0 (ω̂est − ωest )
Δ
(P1 ) ϑ̂ ∈ Ω̄ϑ̂ = ϑ̂ : ϑm in ≤ ϑ̂ ≤ ϑm ax
 x̂˙ 2 = −θ02 (ω̂est − ωest ) (15)
(P2 ) ϑ̃T Γ−1 Proj(Γη) − η ≤ 0 ∀η. (7)
where ω̂est and x̂2 are denoted as the estimates of ω est and
2) OASC Design: The parametric uncertainties and unstruc- x2 . These estimates are further used for designing the required
tured external disturbances can be estimated together as in [12]. control law. The parameter θ0 is the observer bandwidth, which
However, it deals with poor tracking problem and reduced sys- is chosen positive.
tem performance [12]. Here, the structured and unstructured Lemma 1 : For this lemma, a(t) is assumed as bounded,
uncertainties are dealt explicitly to have enhanced system per- which proves boundedness of estimated states and there exists
formance with better tracking. The adaptive law is implemented a positive constant τ 1 and a finite time Ts such that
to estimate the former uncertainty, whereas the later problem
is dealt with employing a disturbance observer, which is then |x̃j | ≤ τj , τj = O (1/θ0n ) , j = 1, 2 ∀t ≥ Ts (16)
feedforward compensated. The adaptive control law is used for for a positive integer n, where x̃j = xj − x̂j . The positive inte-
estimating parametric uncertainties that is based on projection. ger τ 1 is considered as a preset range of x̃j . This lemma specifies
The primary objective of the control deign is to derive a that a high value of observer bandwidth θ0 results in a small es-
control law such that PMSG speed ω est tracks the reference timated error that is confined to the value τ 1 . To avoid failure
speed ωr . A switching function is introduced as in the control application, the value of τ 1 should not be chosen
too high [12].
z1 = ωest − ωr (8)
The proposed control law is designed by gathering the sense
where z1 is the tracking error of speed, which should be as small of robust and adaptive control, thereby realizing a hybrid control
as possible. law. It avoids the use of high feedback gain for retaining robust-
Taking the nominal values of ϑ1 and ϑ2 as ϑ1n and ϑ2n in ness. By incorporating both the logics in (13), the modified
(3), it is rewritten as control law for OASC is derived as

ẋ1 = ω̇est = −ϑ1n iq r − ϑ2n ωest + x2 iq r = −(iq a + iq s1 + iq s2 ) /ϑ̂1 (17)

x2 = (ϑ1n − ϑ1 ) iq r + (ϑ2n − ϑ2 ) ωest + d(ωest , t) (9) iq a = ω̇r + ϑ̂2 ωest − x̂2 (18)

where x2 is the lumped disturbance of the system. iq s1 = −σ1 z1 . (19)

Authorized licensed use limited to: Indian Institute of Information Technology Design & Manufacturing. Downloaded on November 21,2021 at 15:17:10 UTC from IEEE Xplore. Restrictions apply
 10812 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 34, NO. 11, NOVEMBER 2019

The final control law includes three significant parts con-

tributing substantial modification to meet the control objective.
The first term iq a is the model compensation part, which ensures
speed tracking, iq s1 specifies the feedback control implemen-
tation, and iq s2 specifies the robust control implementation to
deal with the issues of uncertainties.
Implementation of the control laws to (10) gives

ż1 = ω̇est − ω̇r = −ϑ̂1 iq r − ϑ̂2 ωest +x2 − iq a − ϑ̂2 ωest + x̂2

ż1 = −ϑ̂1 iq r − iq a + x̃2 . (20)

Incorporating (17) in (20) gives
ż1 = iq s1 + iq s2 + x̃2 = −σ1 z1 + iq s2 + x̃2 . (21)
The remaining part iq s2 is designed to exhibit robustness and Fig. 2. Switching signal generation for CMS.
accuracy and it should satisfy the following condition:
z1 [iq s2 + x̃2 ] ≤ ετ22 (22)
where ε represents the level to which the model uncertainties
can be attenuated, which is chosen as a small value.
The following example is chosen, which satisfies (22) as
iq s2 = −z1 /4ε. (23)
The projection-based parametric adaptation law holds P1 in
(7) and the estimated parameters follows (4).
For implementing the projection-based parameter adaptation
law, η is chosen as [12]
η = ϕz1 (24) Fig. 3. Switching signal generation for CGS.

where the matrix ϕ is taken as

It justifies that the derived control input is bounded and the
Δ  T
ϕ = iq r −ωest . (25) designed OASC is exponentially stable. Moreover, it is con-
cluded that the rate of convergence is 2σ 1 with the steady-state
3) Stability Analysis: An energy like Lyapunov function is tracking error derived as
assumed for stability analysis, defined as 
V = z12 /2. (26) V (∞) ≤ ετ22 /2σ1 . (31)

Taking derivative of (26) and applying (21) gives Hence, it is worth saying that the important performance
parameters of the control design such as rate of convergence
V̇ = z1 ż1 = z1 (−σ1 z1 + iq s2 + x̃2 )
and the steady-state tracking error are adjusted with the proper
= −σ1 z12 + z1 (iq s2 + x̃2 ) . (27) selection of constants ε, τ 2 , and σ 1 .
The designed control law is implemented for speed tracking
For time t ≥ Ts , based on the chosen lemma defined in problem in speed control loop of vector-controlled CMS. Fig. 2
(16), the chosen function (23) holds the condition given in (22). describes the control law implementation and switching signal
Applying (22) in (27) gives generation for the CMS. The reference three-phase stator cur-
V̇ ≤ −σ1 z12 + ετ22 . (28) rents are generated from reference dq-axis currents, where the
output of speed control loop is considered as reference q-axis
By considering k = 2σ 1 and from (26) and (28), it is synthe- current. A hysteresis current controller is used for generating
sized that controlled switching sequence for CMS that takes the reference
V̇ ≤ −kV + ετ22 . (29) and sensed stator currents.
In Lemma 1, it has been considered as a(t) is bounded and
B. Signal Generation for CGS
the projection-based parameter estimation law is employed for
this purpose. Then, the integration of (29) from Ts to t gives the As mentioned previously, the MAF-based grid synchroniza-
boundary condition of V as tion technique is adopted for generating switching signals for
  CGS, as shown in Fig. 3. It is a PLL-less synchronization tech-
V (t) ≤ V (Ts )e−k (t−T s ) + ετ22 /k 1−e−k (t−T s ) ∀t ≥ Ts . nique that is used with unit templates generated from grid line
(30) voltages (vpab and vpbc ). The phase voltages (vpa , vpb , and vpc )

Authorized licensed use limited to: Indian Institute of Information Technology Design & Manufacturing. Downloaded on November 21,2021 at 15:17:10 UTC from IEEE Xplore. Restrictions apply
 PRADHAN et al.: PERFORMANCE INVESTIGATION OF MULTIFUNCTIONAL ON-GRID HYBRID WIND–PV SYSTEM 10813

(κn ) of reference grid currents as

κn = κL av +κldc − κw f − κpv f (37)
where κpv f and κw f are computed as
κw f = (2/3) (Pw /Vpt ) (38)
κpv f = (2/3) (Ppv /Vpt ) . (39)
Then, three-phase balanced sinusoidal reference currents
(irg a , irg b , irg c ), synchronized with the grid, are determined by
multiplying in-phase unit templates and the net component as
irg a = κn × pa , irg b = κn × pb , irg c = κn × pc .
(40)
The CMS is triggered with the gate signals generated from
Fig. 4 MAF structure. hysteresis current controller by comparing these reference grid
currents and sensed grid currents (ig a , ig b , ig c ).
1) Basic Structure of AVF With Adaptive Frequency Estima-
at CCP and its peak (Vpt ) are calculated as
tion: The basic structure of AVF is constructed in αβ form,
vpa = (2/3) [vpab +0.5vpbc ] , vpb = (2/3) [−0.5vpab +0.5vpbc ] where the fundamental α component takes part in determining
fundamental β component and vice versa. Primarily, the load
vpc = (2/3) [−0.5vpab − vpbc ] (32) current of each phase is converted to αβ form with the assump-
  tion that the angular frequency ω is a slowly varying parameter,
Vpt = 2 + V2 + V2
2 Vpa pb pc /3 . (33) considered as
 T
The unit templates of grid voltage are obtained as iα β = Lm sin ωt cos ωt . (41)

pa = vpa /Vpt , pb = vpb /Vpt , pc = vpc /VP t , The time derivative of (41) gives the vectorial form as
 
−2pb + 2pc 3pa + pb − pc 0 −1
qa = √ , qb = √ , (diα β /dt) = ωAiα β , where A = . (42)
1 0
2 3 2 3
−3pa + pb − pc Considering the output of AVF is the fundamental component
qc = √ (34) of iα β , as iα β f , the product of input and output in scalar and
2 3
vectorial nature is synthesized as
where pa , pb , and pc are denoted as in-phase unit templates and
qa , qb , and qc are quadrature unit templates. sL = iTα β f iα β (43)
The nonlinear load current of each phase passes through vL = iTα β f Liα β . (44)
MAF structure (in αβ form) to extract its fundamental in-
phase and quadrature components (using the block diagram of Time derivative of (43) and (44) results as follows:
 
MAF shown in Fig. 4). The corresponding in-phase components dsL diTα β f
(ipf L a , ipf L b , ipf L c ) are managed through the sample and hold cir- = + ωiα β f A iα β
T
(45)
dt dt
cuit, processed by zero cross detection block with the quadrature
unit templates (qa , qb , and qc ) as inputs. This process gives the  
dvL diTα β f
active fundamental load current components (κL a , κL b , and = + ωiα β f A AiL α β .
T
(46)
κL c ). Its average component (κL av ) is computed as dt dt

κL av = (κL a + κL b + κL c ) /3. (35) The following conditions must be satisfied for meeting the
aim of iα β reconstruction from iα β f :
The dc link voltage is maintained by using a PI controller that
sL → s∗L = iTα β iα β (47)
takes the reference and sensed dc link voltages (Vdcr and Vdc ,
respectively) as inputs. Owing to the single-stage configuration vL → vL∗ = 0. (48)
of solar PV integration, the reference dc link voltage is generated
Δ Δ
from the P&O MPPT method. During availability of low solar Considering s̃L = sL − s∗L and ṽL = vL − vL∗ , (45) and (46)
insolation or at night, the reference value is chosen as a nominal are solved as
 
value (here 360 V). The output of PI controller is denoted by ds̃L ds∗L diTα β f
the dc link loss component (κldc ), which is expressed as + ps s̃L + − ps s̃L = + ωiTα β f A iα β (49)
dt dt dt
κldc = {ap + (ai /s)} (Vdcr − Vdc ) . (36)  
dṽL dvL∗ diTα β f
For transient performance enhancement, two feedforward + pv ṽL + − pv ṽL = + ωiTα β f A Aiα β .
dt dt dt
compensation (FFC) terms, PV FFC (κpv f ) and wind FFC
(κw f ), are included for determining the net peak component (50)

Authorized licensed use limited to: Indian Institute of Information Technology Design & Manufacturing. Downloaded on November 21,2021 at 15:17:10 UTC from IEEE Xplore. Restrictions apply
 10814 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 34, NO. 11, NOVEMBER 2019

ds ∗L Δ di Tα β f
By choosing dt − ps s̃L = ( dt + ωiα β f A)iα β
T
and
Δ
(dvL∗ /dt) − pv s̃L = ((diTα β f /dt) + ωiα β f A)Aiα β , the follow-
T

ing condition has arrived:

ds̃L dṽL
+ ps s̃L = 0 and + pv ṽL = 0. (51)
dt dt
The first-order linear system representations (50) prove that
s̃L and ṽL tend to zero exponentially with ps , pv > 0, and
constant value of s∗L and vL∗ .
By considering the above factors and equations, the dynamics
of iL α β f is derived as
diα β f
= ωAiα β f + ωpf (iα β − iα β f ) (52)
dt
where the filter gain pf = ps = pv is to be adjusted to get re-
quired performance.
The angular frequency ω is the adaptive parameter, which is to Fig. 5. Performance comparison between PI controller, SMC, ASC, and
be adjusted for accurate fundamental extraction from nonlinear OASC.
harmonic-rich load current. The concept of frequency-locked
loop is adopted for calculating ω. Assuming the estimated ω as gain for the positive and negative sequences of fundamental
ω, the dynamic equation (52) is written as component (pf ) is chosen as 0.78 and for other components it
diα β f is chosen as 0.075.
= ωAiα β f + ωpf (iα β − iα β f ) . (53)
dt
IV. SIMULATION RESULTS
Using (53) in (50), the dynamic equation is modified as
dṽL The proposed controls are implemented on the on-grid hybrid
Δ
+ pf ωṽL = sL ω̃, where ω̃ = ω − ω. (54) wind–PV system (as per Fig. 1). The system is simulated using
dt
MATLAB/Simulink toolbox. The system parameters chosen for
The following Lyapunov function is taken with design con- simulation purpose are given in the Appendix. A comparative
stant ζ > 0 as study describing the performance analysis of OASC in compar-

V = (1/2) ṽL2 + ω̃ 2 /ζ . (55) ison with PI controller, SMC, and adaptive speed control (ASC)
without observer is presented in Fig. 5. The OASC parameters
Taking derivative of (55) and solving it, the expression for ω are chosen as σ 1 = 12, θ0 = 60, and ε = 1. The PI controller
is derived as parameters are chosen as proportional constant (kp ) = 2 and in-
tegral constant (kI ) = 50. For SMC implementation, the sliding
dω/dt = −ζsL ṽp . (56)
mode gain is chosen as kSM C = 80. The ASC without observer
For the MAF structure, where positive, negative, and har- is designed by considering the feedback gain as σ 2 = 20. A wind
monic components are calculated, only one ω estimation block speed profile is chosen for performance comparison, in which
is used. the initial wind speed is set as 8.4 m/s, wind speed variation in
2) Structure of MAF: The basic block of MAF structure is step from 8.4 to 12 m/s is applied at 2 s, from 12 to 9.6 m/s
the AVF with dynamics given in (53). A single AVF module is at 2.5 s, and then the wind speed is reduced to 7.2 m/s. The
unable to filter out all the harmonics sequence. Considering the cut-in and rated wind speeds are set as 6 and 12 m/s, respec-
cross-feedback logic, as in the MSOGI structure, the structure tively. According to the wind speed, the reference PMSG speed
of MAF is designed to estimate unwanted lower order harmonic is obtained from P&O MPPT. Tracking speed and accuracy with
components (third, fifth, and seventh) individually, as presented each control are compared. Moreover, comparison of chattering
in Fig. 4. The inputs to each AVF structure are modified by in the control input with all four controls is presented. As ob-
using cross-feedback terms to cancel out the effect of unwanted served, the proposed OASC shows better results in terms of
components from the signal to be filtered. The AVF tuned to accuracy, error, and transient response than the other three tech-
frequency −ω is capable of estimating fundamental negative niques. Tracking accuracy is better in case of OASC than others
sequence component. Likewise, for estimating fundamental and in the presence of sudden change of wind speed. It is achieved
each harmonic behavior, each AVF is tuned to obtain positive because of the implementation of additional parameter adapta-
and negative sequence components at fundamental, third, fifth, tion technique and disturbance observer, which deals with the
and seventh harmonics. The AVF gains are chosen wisely for structured and unstructured uncertainties. It facilitates to choose
achieving accurate and fast result. a feedback gain of low value. The speed tracking error in case of
Fig. 4 represents the structure of MAF to extract fundamental PI controller is observed more with poor transient performance
component from polluted nonlinear load currents. A frequency than the others. In case of SMC, the speed tracking and transient
estimation block is used to make it adaptive. Here, the AVF performance are better than PI controller. In case of ASC, the

Authorized licensed use limited to: Indian Institute of Information Technology Design & Manufacturing. Downloaded on November 21,2021 at 15:17:10 UTC from IEEE Xplore. Restrictions apply
 PRADHAN et al.: PERFORMANCE INVESTIGATION OF MULTIFUNCTIONAL ON-GRID HYBRID WIND–PV SYSTEM 10815

Fig. 6. Performance comparison between SRF, SOGI, MSOGI, and MAF.

Fig. 7. Hardware prototype of a hybrid system.
performance is better than PI controller and SMC. However, it
is required to choose a high feedback gain to deal with the un-
certainties in case of ASC. The chattering in control input or the configuration of solar PV array is established. The PV curves are
reference q-axis current iq r is more in case of SMC implemen- observed on the screen of host PC with the application software
tation than the others. Chattering in iq r waveform is observed installed in it. The nonlinear load is established by considering
as minimum when OASC is applied. The chattering in iq r with an uncontrolled diode bridge rectifier with RL. For realizing the
PI controller is less in comparison with SMC, but the response unbalanced load condition, one phase of three-phase balanced
is slow. Hence, implementation of OASC gives reduced speed load is disconnected by using a mechanical switch. Real-time
tracking error, better transient response, and reduced chattering application of the controls is made through a digital signal pro-
in the control input. cessor board (dSPACE-1103), equipped with digital to analog
The filtering capability of proposed MAF is compared with converter and analog to digital converters (ADCs). The appli-
the synchronous reference frame (SRF), SOGI, and MFOGI. cation software is installed on a host PC, through which the
For analyzing their performance, the average load components signals are observed in real time. The required voltage and
(κL av ) are considered under nonlinear unbalanced load, which current signals at required points are sensed through Hall ef-
is realized by disconnecting the phase “a” of three-phase bal- fect sensors (LV-25P, LA-25A). The dc link voltage (Vdc ), CCP
anced load. Fig. 6 depicts the comparison between four filters. voltages (vpab , vpbc ), and PV voltage (Vpv ) are sensed using
As seen, during the disconnection of phase “a” load, it takes LV-25P. The PMSG stator currents (im a , im b , im c ), PV current
more time to reach the steady-state value of κL av , which is ob- (Ipv ), currents at ac side of CGS (ipa , ipb , ipc ), load currents
tained by using SOGI. Using MFOGI, the performance is better (iL a , iL b , iL c ), and grid currents (ig a , ig b , ig c ) are sensed us-
than SOGI in terms of speed and accuracy. Upon the discon- ing LA-25A. The sensed voltages and currents are connected to
nection of phase “a” load, the behavior of κL av is satisfactory the ADC points of DSP, which become accessible for control
in terms of reaching time and accuracy. Hence, the steady-state implementation through the PC. The MPPT technique required
and dynamic performances of the system with the proposed for the solar power extraction is implemented to get the refer-
MAF-based control are improved. ence voltage to maintain dc link. During night time, the solar
power generation goes to zero and the reference dc link voltage
is set at the nominal value (360 V). The real-time signals are
V. CONTROL IMPLEMENTATION ON PROTOTYPE
observed on the dashboard of the host PC. Moreover, real-time
These controls are verified on a prototype built in the lab- wind speed variation command is provided here through vari-
oratory. It is designed by considering the schematic of hybrid able array. The gate signals for VSCs are obtained from DSP
system depicted in Fig. 1. To realize the WECS, a 5-hp PMSG and passed through isolation circuits. The system performance
coupled to an induction motor (IM) is considered. The variable- is recorded by using a power quality analyzer (Fluke 43B) and a
speed WT is realized on IM, which is run by a variable frequency digital storage oscilloscope. The parameters of developed pro-
drive (VFD075E43A of 7.5 kW). A MATLAB/Simulink file is totype are given in the Appendix. The prototype is tested under
made to feed the variable speed command to VFD to get the WT different cases like rated wind speed with rated solar insolation,
characteristics. The stator terminals of PMSG are connected to and loading variable speed and solar insolation, zero insolation,
VSC. By keeping the dc link as a common dc point, another VSC balanced and unbalanced nonlinear loads, and no load. A pic-
is connected followed by interfacing inductors, whose output is ture of hardware prototype highlighting the generator, VSCs,
connected at the CCP, and at this point, the grid and nonlin- sensors, dSPACE-1103, solar PV simulator, and some other im-
ear load are connected. The ripple filters are connected at CCP portant parts is presented in Fig. 7.
for switching ripple elimination. Three-phase supply connected
with three-phase autotransformer and grid impedance is consid-
ered as the grid. A solar PV simulator is used for constructing a A. Test Results at Rated Condition
solar array of required value. The terminals of PV simulator are The system is tested under rated wind speed of 12 m/s and
directly connected at the common dc point; hence, a single-stage rated solar insolation of 1000 W/m2 . At this rated wind speed and

Authorized licensed use limited to: Indian Institute of Information Technology Design & Manufacturing. Downloaded on November 21,2021 at 15:17:10 UTC from IEEE Xplore. Restrictions apply
 10816 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 34, NO. 11, NOVEMBER 2019

Fig. 8. Test results of a hybrid system at rated wind speed (12 m/s) and solar insolation (1000 W/m2 ).

solar insolation, balanced nonlinear load is applied at CCP. At TABLE I

TEST RESULT AT RATED CONDITION
this rated condition of wind speed and solar insolation, the sys-
tem performance is evaluated and test results are shown in Fig.
8. The CCP voltage vpab , nonlinear load current iL a , load power
PL , and load current THD are given in Fig. 8(a)–(c). As seen,
iL a is quasi-square wave of 4.289 A confirming its nonlinear
nature. The load current is containing harmonics and its THD is
recorded as 23.7%. The power demand of the local load is mea-
sured as 1.56 kW. The nonlinearity of load introduces harmonics
to the grid. By using the CGS operated with the proposed MAF-
based control, it works as a harmonics compensator. Moreover,
it helps in active power flow from the renewables, grid current
balancing, and grid side power factor correction. The vpab with
CGS current ipa and power Ppa is given in Fig. 8(d) and (e).
The ipa is the resultant current of 13.38 A that flows from CSG
to the load and the grid. The Ppa is the aggregated renewable written as
power of 5.08 kW. The natures of grid current ig a and grid Ppv + Pw t − Ploss = PL + Pg (57)
power Pg with its harmonics spectrum are presented in Fig.
8(f)–(h). The proposed MAF is used to extract the fundamen- where Ppv , Pw t , Ploss , PL , and Pg are the generated PV power,
tal component of load current and to generate the sinusoidal generated wind power, power loss, load power, and grid power,
balanced grid currents. Hence, the grid current is observed as respectively. The total generated power [left-hand side of (57)]
sinusoidal of 9.19 A. The surplus active power flowing to the at the ac side of CGS is 5.08 kW. The total load demand is
grid is observed as 3.52 kW. The power factor at the grid side 1.56 kW and the remaining power of 3.52 kW is fed to the grid.
is obtained nearer to unity. In Table I, the currents and powers The PL is shown positive to demonstrate its positive power flow
at load, CGS, and grid terminals are given. From these results, from the sources. The Pg is shown negative that shows the power
it is verified that ig a is sinusoidal with THD 3.7%, where iL a flow of surplus renewable power to the grid after meeting the
is nonlinear with THD of 23.7%. The THD of ig a is coming need of local loads.
under the satisfied condition of the IEEE-519 standard for grid
B. Test Results Under Variable Wind Speed
current THD limits. The power sharing occurs at CCP effec-
tively from the renewables (wind and PV) to the load and the The system is tested under variable wind speed vw s to
grid. A simple equation satisfying the power balance theory is verify the OASC-based control and effective wind power

Authorized licensed use limited to: Indian Institute of Information Technology Design & Manufacturing. Downloaded on November 21,2021 at 15:17:10 UTC from IEEE Xplore. Restrictions apply
 PRADHAN et al.: PERFORMANCE INVESTIGATION OF MULTIFUNCTIONAL ON-GRID HYBRID WIND–PV SYSTEM 10817

Fig. 9. Test results of a hybrid system at different wind speeds and wind speed variation.

extraction under variable wind speed. Wind speed variation The wind speed change is applied in step from 12 to 7.2 m/s;
command in step is given on the screen of host PC through the currents at stator terminal (im a ), PV terminal (Ipv ), load
a variable array. It is considered that the system is operated at (iL a ), and grid (ig a ) are captured and depicted in Fig. 10(a).
the wind speed of 12 m/s. At this condition, the estimated PMSG The load demand remains unaltered and solar insolation is kept
speed ω est and three-phase stator currents (im abc ) are recorded constant at 1000 W/m2 . With the reduction of wind speed, am-
and shown in Fig. 9(a). The stator currents are observed as si- plitude of im a decreases and magnitudes of Ipv and iL a remain
nusoidal, justifying the PMSG operation. The PMSG speed and same as before proving no variation in solar insolation and load
stator currents at the wind speed of 7.2 m/s are recorded and demand. However, a reduction in current flowing to the grid is
presented in Fig. 9(b). The PMSG speed is reduced with reduced noticed, as the total generation from the renewables decreases.
wind speed command and stator currents are seen as sinusoidal The PMSG stator current and grid current are maintained sinu-
with reduced amplitude. Fig. 9(c) depicts the wind speed change soidal. The power flowing through the system is observed from
command in step, PMSG speed variation, stator current in phase the nature of wind power (Pw t ), PV power (Ppv ), load power
“a”, and the extracted peak wind power Pw t . When the speed (PL ), and grid power (Pg ) under wind speed change from 12
variation from rated to 7.2 m/s is applied, reference PMSG to 7.2 m/s, as shown in Fig. 10(b). Herein, it is seen that with
speed is reduced; therefore, reduced estimated PMSG speed is wind speed change, the Pw t decreases, and Ppv and PL are
recorded, the amplitude of stator current is reduced, and hence unchanged due to the application of fixed solar insolation and
the extracted wind power is reduced. Fig. 9(d) presents the vari- loading. Hence, the total generated power that is measured at
ation of PMSG speed, stator current amplitude, and extracted the ac side of CGS is decreased. As the load demand is fixed, the
wind power with the step change in wind speed from 7.2 m/s available active power that flows to the grid decreases. Owing
to its rated value. When the wind speed is increased, the refer- to the power flow toward the grid, Pg is presented with nega-
ence PMSG speed is increased; thus, increased estimated PMSG tive value. Hence, the total power generated from the renewable
speed is obtained. The amplitude of stator current is increased sources is fed to the load and the grid, showing the balanced
and hence increased extracted peak wind power is achieved. power flow in the system. Fig. 10(c) presents the power flow
Moreover, the stator currents are observed as sinusoidal at each in the system under the condition of wind speed below cut-in.
case, which is noticed from highlighted portions. It verifies When the wind speed goes below cut-in speed, the Pw t be-
that effective wind power Pw t is extracted at variable wind comes zero. At this condition, the load demand is transferred to
speeds. the PVECS and the remaining power is fed to the grid. In this

Authorized licensed use limited to: Indian Institute of Information Technology Design & Manufacturing. Downloaded on November 21,2021 at 15:17:10 UTC from IEEE Xplore. Restrictions apply
 10818 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 34, NO. 11, NOVEMBER 2019

Fig. 10. Test results of a hybrid system at variable wind speed.

case, very less amount of active power is to be fed to the grid. ipf L a (κL a ), and κL av during balanced loading and disconnec-
Parameters referring to the control implementation such as wind tion of phase “a” load (unbalanced load). Herein, expected si-
FFC term (κw f ), term referring loss at dc link (κldc ), average nusoidal waveforms of ipf L a are obtained using the proposed
load term (κL av ), and the net component (κn ) are presented MAF and the peak of fundamental active component κL a is
in Fig. 10(d). The κn decides the magnitude of reference grid then obtained under unbalanced load. During the disconnection
currents. The κw f describes the effect of Pw t variation, which of phase “a” load, the current flowing through that phase goes to
results from variable wind speed. The loss at dc link is dealt zero, while the currents still flow through the other two phases.
with the PI controller, which is noticed from κldc . The κL av This effect is observed from κL a that during phase disconnec-
specifies the average active weight component of load. A wind tion attains zero value. However, κL av has some value due to
speed profile is applied to the system, in which it is reduced the currents flowing in other two phases. Reconnection of phase
from 12 to 7.2 m/s and then again increased to 12 m/s. Due to “a” load makes the load balanced and the normal operating
the reduction of wind speed, the Pw t is reduced and its effect is condition is retrieved.
incorporated through κw f reduction. Due to the application of Fig. 11(c) depicts the nature of currents im a , Ipv , iL a , and ig a
fixed load demand, κL av is observed as constant. This change under unbalanced load at rated wind speed and solar insolation.
is reflected on the grid through the net parameter κn , which This condition is achieved by disconnecting phase “a” load. The
serves as the amplitude of grid current. These results prove the load variation is observed from the zero current flowing in this
effective wind power extraction at rated PV generation and load particular phase, when im a and Ipv are maintained at the rated
under variable wind speed. value. The load appears as a single-phase load on the system
and the effective loading on the system is reduced. As the total
generated renewable power is at rated condition, the unbalanced
C. Test Results Under Unbalance Load
condition of load increases the peak of ig a . Due to the implemen-
To verify the working of MAF-based control implemented tation of MAF-based control, an optimum value of fundamental
for the CGS, the system is tested under balanced and un- active weight component is achieved, which helps in achieving
balanced nonlinear loads. The primary aim of MAF is to the sinusoidal reference grid currents synchronized to the grid.
extract the fundamental component from the nonlinear load Hence, the sensed grid current ig a is sinusoidal. Considering the
current. Fig. 11(a) and (b) show the extracted fundamental previous condition of unbalanced loading, the power flow from
component (ipf L a ) from phase “a” load current (iL a ), peak of the renewables to the load and the grid is depicted in Fig. 11(d). It

Authorized licensed use limited to: Indian Institute of Information Technology Design & Manufacturing. Downloaded on November 21,2021 at 15:17:10 UTC from IEEE Xplore. Restrictions apply
 PRADHAN et al.: PERFORMANCE INVESTIGATION OF MULTIFUNCTIONAL ON-GRID HYBRID WIND–PV SYSTEM 10819

Fig. 11. System performance under balanced, unbalanced, and no load cases at rated wind speed and solar insolation.

describes that during unbalanced load, the Pg increases with the D. Test Results Under Varying Solar Insolation
decrease in PL at rated Pw t and Ppv . Fig. 11(e) presents the pa- The final case study is performed under variable solar inso-
rameters referring to the control action (κL av , κw f , κpv f , κn )
lation at rated wind speed and fixed balanced nonlinear load.
under load disconnection. At this stage, the load demand at the
Fig. 12(a) shows the performance under insolation change from
CCP goes to zero and the total generated power is fed to the zero to rated condition (1000 W/m2 ). At zero insolation, Ipv
grid. Here, κn increases to more negative with zero κL av at
is zero, and im a and iL a are at their rated values. At zero in-
rated κw f and κpv f . It shows the flow of grid currents of higher
solation, the generated Ppv goes to zero and WECS meets the
magnitude. The reconnection of balanced load at CCP restores load demand. Hence, due to the reduced total power generation,
the previous results and this is observed from the behavior of pa-
the peak of ig a is reduced. Under the solar insolation change
rameters referring to the control action (κL av , κw f , κpv f , κn ). from 1000 to 500 W/m2 , generated power from RES and its
Test results describe the satisfactory action of the MAF-based distribution within the load and grid are illustrated in Fig. 12(b).
control.
When the solar insolation decreases to 500 W/m2 , Ppv is re-
duced. However, Pw t is maintained at its rated value, because

Authorized licensed use limited to: Indian Institute of Information Technology Design & Manufacturing. Downloaded on November 21,2021 at 15:17:10 UTC from IEEE Xplore. Restrictions apply
 10820 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 34, NO. 11, NOVEMBER 2019

Fig. 12. System performance under vaiable solar insolation at rated wind speed and balanced nonlinear load.

Fig. 13. MPPT operation of solar PV array at 1000 and 500 W/m2 with highlighted point of operation.

wind speed is maintained at its rated condition. Therefore, Pg reduction in κpv f reflects on κn with its reduced magnitude.
decreases with reduction in total generated power due to de- Reduction in κn proves the current flow toward the grid is of re-
crease in Ppv . Under the same condition of insolation change, duced amplitude. Fig. 13 shows the MPPT performance of solar
κpv f , κldc , κL av , and κn are recorded and illustrated in Fig. PV array at 1000 and 500 W/m2 , where the operating points are
12(c). The insolation variation is reflected through κpv f with heighted. PV array operating at 1000 W/m2 draws 7.036 A at a
its reduced value and its effect on the grid is presented through PV voltage of 368 V with MPPT of 99.51%. At solar insolation
decrement of κn . Because of the reduction in solar insolation, of 500 W/m2 solar array draws 3.549 A with a PV voltage of
κpv f is reduced, dc link is maintained through κldc , and κL av 362 V with MPPT of 98.60%. It verifies that solar power is
remains unaltered verifying fixed loading on the system. The extracted effectively at acceptable percentage of MPPT.

Authorized licensed use limited to: Indian Institute of Information Technology Design & Manufacturing. Downloaded on November 21,2021 at 15:17:10 UTC from IEEE Xplore. Restrictions apply
 PRADHAN et al.: PERFORMANCE INVESTIGATION OF MULTIFUNCTIONAL ON-GRID HYBRID WIND–PV SYSTEM 10821

VI. CONCLUSION [5] M. Fatu, F. Blaabjerg, and I. Boldea, “Grid to standalone transition motion-
sensorless dual-inverter control of PMSG with asymmetrical grid voltage
The proposed OASC- and MAF-based controls have been sags and harmonics filtering,” IEEE Trans. Power Electron., vol. 29, no. 7,
designed and implemented on a hybrid wind–PV system under pp. 3463–3472, Jul. 2014.
[6] F. Delfino, F. Pampararo, R. Procopio, and M. Rossi, “A feedback lin-
on-grid condition. The design of OASC included a disturbance earization control scheme for the integration of wind energy conversion
observer loop for unstructured uncertainty, which has been esti- systems into distribution grids,” IEEE Syst. J., vol. 6, no. 1, pp. 85–93,
mated and compensated by considering it as feedforward term. Mar. 2012.
[7] R. Errouissi and A. Al-Durra, “A novel PI-type sliding surface for PMSG-
It incorporated a discontinuous projection law-based parametric based wind turbine with improved transient performance,” IEEE Trans.
uncertainties estimation loop to enhance its performance. The Energy Convers., vol. 33, no. 2, pp. 834–844, Jun. 2018.
backstepping control with additional features has proved its per- [8] S. Das and B. Subudhi, “A Hin fi n ity robust active and reactive power
control scheme for a PMSG-based wind energy conversion system,” IEEE
formance under steady and dynamic conditions by extracting Trans. Energy Convers., vol. 33, no. 3, pp. 980–990, Sep. 2018.
effective power from the variable-speed WT. The performance [9] F. Valenciaga and R. D. Fernandez, “Multiple-input–multiple-output high-
of OASC has been compared with conventional techniques and order sliding mode control for a permanent magnet synchronous generator
wind-based system with grid support capabilities,” IET Renewable Power
simulation results have been provided. Satisfactory results have Gener., vol. 9, no. 8, pp. 925–934, Nov. 2015.
been achieved with the proposed OASC in terms of chattering, [10] S. Li, M. Zhou, and X. Yu, “Design and implementation of terminal sliding
reaching time, and accuracy. The hardware results have been mode control method for PMSM speed regulation system,” IEEE Trans.
Ind. Inform., vol. 9, no. 4, pp. 1879–1891, Nov. 2013.
presented under variable wind speed and satisfactory perfor- [11] R. Errouissi, A. Al-Durra, and M. Debouza, “A novel design of PI current
mance of OASC has been observed. Moreover, by using the controller for PMSG-based wind turbine considering transient perfor-
MAF-based control, harmonic contents from the grid currents mance specifications and control saturation,” IEEE Trans. Ind. Electron.,
vol. 65, no. 11, pp. 8624–8634, Nov. 2018.
have been rejected effectively, resulting in 3.7% of THD in grid [12] J. Yao, Z. Jiao, and D. Ma, “Adaptive robust control of DC motors with ex-
current, obeying the IEEE-519 standard. The performance of tended state observer,” IEEE Trans. Ind. Electron., vol. 61, no. 7, pp. 3630–
this technique has been tested under unbalanced loading, vari- 3637, Jul. 2014.
[13] M. M. R. Singaravel and S. A. Daniel, “MPPT with single DC–DC
able solar insolation, and wind speed. It was observed that the converter and inverter for grid-connected hybrid wind-driven PMSG–
satisfactory results have been obtained. PV system,” IEEE Trans. Ind. Electron., vol. 62, no. 8, pp. 4849–4857,
Aug. 2015.
[14] Y. Chen, Y. Liu, S. Hung, and C. Cheng, “Multi-input inverter for grid-
APPENDIX connected hybrid PV/wind power system,” IEEE Trans. Power Electron.,
vol. 22, no. 3, pp. 1070–1077, May 2007.
A. Simulation and Hardware Parameters [15] M. Singh and A. Chandra, “Application of adaptive network-based fuzzy
inference system for sensorless control of PMSG-based wind turbine with
PMSG: 5 hp, p = 2, stator resistance (Rs ) = 0.2747 Ω, d- nonlinear-load-compensation capabilities,” IEEE Trans. Power Electron.,
axis stator impedance (Xsd ) = 4.669 Ω, q-axis stator impedance vol. 26, no. 1, pp. 165–175, Jan. 2011.
(Xsq ) = 5.573 Ω, permanent magnet flux (ϕm ) = 0.59582 V · [16] S. Reza, M. Ciobotaru, and V. G. Agelidis, “Accurate estimation of single-
phase grid voltage fundamental amplitude and frequency by using a fre-
s, friction coefficient (F) = 0.00015248 Nms/rad, moment of quency adaptive linear Kalman filter,” IEEE J. Emerg. Sel. Topics Power
inertia (J) = 0.00133 kgm2 . Electron., vol. 4, no. 4, pp. 1226–1235, Dec. 2016.
Wind turbine: Air density (ρ) = 1.225 kg/m2 , wind speed [17] B. Singh, S. Kumar, and C. Jain, “Damped-SOGI-based control algorithm
for solar PV power generating system,” IEEE Trans. Ind. Appl., vol. 53,
(rated and cut-in) = (12 m/s, 6 m/s), pitch angle (βt ) = 0◦ , no. 3, pp. 1780–1788, May/Jun. 2017.
optimum value of power coefficient (Cpopt ) = 0.48, optimum [18] S. Golestan, J. M. Guerrero, J. Vasquez, A. M. Abusorrah, and Y. A.
value of tip speed ratio (λtopt ) = 8.1, c1 = 0.5176, c2 = 116, Al-Turki, “A study on three-phase FLLs,” IEEE Trans. Power Electron.,
vol. 34, no. 1, pp. 213–224, Jan. 2019.
c3 = 0.4, c4 = 5, c5 = 21, c6 = 0.0068. [19] N. Kumar, I. Hussain, B. Singh, and B. K. Panigrahi, “Implementation
Solar PV array: Ppv = 2.6 kW, PV open-circuit voltage of multilayer fifth-order generalized integrator-based adaptive control for
(Vo c ) = 420 V, Vpv = 360 V, PV short-circuit current (Isc ) = grid-tied solar PV energy conversion system,” IEEE Trans. Ind. Inform.,
vol. 14, no. 7, pp. 2857–2868, Jul. 2018.
7.6 A, sampling period (Ts ) = 30 μs. [20] S. Vazquez, J. A. Sanchez, M. R. Reyes, J. I. Leon, and J. M. Carrasco,
Grid parameter: Vpab = 220 V, Rg = 0.2 Ω, Lg = 1 mH, “Adaptive vectorial filter for grid synchronization of power converters un-
Li = 3 mH, Rf = 5 Ω, Cf = 10 μF, Vdc = 360 V, Cdc = der unbalanced and/or distorted grid conditions,” IEEE Trans. Ind. Elec-
tron., vol. 61, no. 3, pp. 1355–1367, Mar. 2014.
2200 μF.

REFERENCES
[1] M. Rezkallah, A. Hamadi, A. Chandra, and B. Singh, “Design and im-
plementation of active power control with improved P&O method for
wind-PV-battery-based standalone generation system,” IEEE Trans. Ind. Subarni Pradhan (S’15) was born in Odisha, India,
Electron., vol. 65, no. 7, pp. 5590–5600, Jul. 2018. in 1989. She received the B.Tech. degree in electrical
[2] S. Pradhan, B. Singh, B. K. Panigrahi, and S. Murshid, “A composite engineering from Siksha ‘O’ Anushandhan Univer-
sliding mode controller for wind power extraction in remotely located sity, Bhubaneswar, India, in 2011, and the M.Tech.
solar PV-wind hybrid system,” IEEE Trans. Ind. Electron., vol. 66, no. 7, degree in power control and drives from the National
pp. 5321–5331, 2019. Institute of Technology, Rourkela, India, in 2013. She
[3] S. M. Mozayan, M. Saad, H. Vahedi, H. Fortin-Blanchette, and M. Soltani, is currently working toward the Ph.D. degree at the
“Sliding mode control of PMSG wind turbine based on enhanced exponen- Department of Electrical Engineering, Indian Insti-
tial reaching law,” IEEE Trans. Ind. Electron., vol. 63, no. 10, pp. 6148– tute of Technology Delhi, New Delhi, India.
6159, Oct. 2016. Her research interests include power electronics,
[4] A. S. Bubshait, A. Mortezaei, M. G. Simões, and T. D. C. Busarello, renewable energy, microgrid, power quality, electri-
“Power quality enhancement for a grid connected wind turbine energy cal machines and drives, and application of adaptive and robust control tech-
system,” IEEE Trans. Ind. Appl., vol. 53, no. 3, pp. 2495–2505, May/Jun. niques in power systems.
2017.

Authorized licensed use limited to: Indian Institute of Information Technology Design & Manufacturing. Downloaded on November 21,2021 at 15:17:10 UTC from IEEE Xplore. Restrictions apply
 10822 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 34, NO. 11, NOVEMBER 2019

Shadab Murshid (M’15) was born in Patna, Bihar, Bijaya Ketan Panigrahi (SM’06) received the Ph.D.
India, in 1991. He received the B.E. degree in elec- degree in power systems from Sambalpur University,
trical engineering from Aligarh Muslim University, Sambalpur, India, in 2004.
Aligarh, India, in 2013, and the M.Tech. degree in Since 2005, he has been an Associate Professor
power electronics electrical machines and drives from with the Department of Electrical Engineering, In-
the Indian Institute of Technology Delhi, New Delhi, dian Institute of Technology (IIT) Delhi, New Delhi,
India, in 2013, where he is currently working toward India, where he became a Professor in 2017. Prior
the Ph.D. degree in electrical engineering in 2014. to joining IIT, he was a Lecturer with the Univer-
His research interests include electric drives, solar sity College of Engineering, Sambalpur, for 13 years.
water pumping, wind energy conversion systems, and His research interests include intelligent control of
applications of adaptive and intelligent control for re- flexible ac transmission system devices, digital sig-
newable power generation and power quality improvement in grid interactive nal processing, power quality assessment, and application of soft-computing
renewable energy systems. techniques to power system planning, operation, and control.

Bhim Singh (SM’99–F’10) was born in Rahamapur,

Bijnor, UP, India, in 1956. He received the B.E. de-
gree in electrical engineering from the University of
Roorkee, Roorkee, India, in 1977, and the M.Tech.
degree in power apparatus and systems and the Ph.D.
degree in electrical engineering from the Indian Insti-
tute of Technology Delhi, New Delhi, India, in 1979
and 1983, respectively.
In 1983, he joined, as a Lecturer, the Department of
Electrical Engineering, University of Roorkee (now
IIT Roorkee), where he became a Reader in 1988. In
December 1990, he joined, as an Assistant Professor, the Department of Elec-
trical Engineering, IIT Delhi, where he became an Associate Professor in 1994
and a Professor in 1997. He was an ABB Chair Professor from 2007 to 2012.
He was a CEA Chair Professor from 2012 to 2017. He was the Head of the De-
partment of Electrical Engineering, IIT Delhi, from 2014 to 2016. Since 2016,
he has been the Dean of Academics with IIT Delhi. He has been a J. C. Bose
Fellow of DST, Government of India, since December 2015. He has guided
74 Ph.D. dissertations and 167 M.E./M.Tech./M.S.(R) theses. He has filed 37
patents. He has executed more than 80 sponsored and consultancy projects.
He has coauthored a textbook on power quality: Power Quality Problems and
Mitigation Techniques (Wiley, 2015). His research interests include solar PV
grid interface systems, microgrids, power quality monitoring and mitigation,
solar PV water pumping systems, improved power quality ac–dc converters,
power electronics, electrical machines, drives, flexible alternating transmission
systems, and high-voltage direct current systems.
Prof. Singh is a Fellow of the Indian National Academy of Engineering,
India, the Indian National Science Academy, India, the National Academy of
Sciences, India, the Indian Academy of Sciences, India, The World Academy
of Sciences, Institute of Electrical and Electronics Engineers, the Institute of
Engineering and Technology, Institution of Engineers (India), and Institution of
Electronics and Telecommunication Engineers, India, and a Life Member of the
Indian Society for Technical Education, System Society of India, and National
Institution of Quality and Reliability. He was the recipient of Khosla Research
Prize of the University of Roorkee in 1991. He was the recipient of J. C. Bose
and Bimal K. Bose Awards of the Institution of Electronics and Telecommu-
nication Engineers for his contribution in the field of power electronics. He
was also the recipient of Maharashtra State National Award of Indian Society
for Technical Education in recognition of his outstanding research work in the
area of power quality. He was the recipient of PES Delhi Chapter Outstanding
Engineer Award in 2006, the Khosla National Research Award of IIT Roorkee
in 2013, Shri Om Prakash Bhasin Award in 2014 in the field of engineering
including energy and aerospace, the IEEE PES Nari Hingorani Custom Power
Award in 2017, and the “Faculty Research Award as a Most Outstanding Re-
searcher” in the field of engineering in 2018 of Careers-360, India. He was
the General Chair of the 2006 IEEE International Conference on Power Elec-
tronics, Drives, and Energy Systems, and General Co-Chair of the 2010 IEEE
International Conference on Power Electronics, Drives, and Energy Systems,
the 2015 IEEE India Council International Conference, the 2016 IEEE Interna-
tional Conference on Power Systems, New Delhi, and the 2017 National Power
Electronics Conference, Pune. He was the Chair of PES-IAS Delhi Chapter
for 2005–2010 (PES-IAS Delhi Chapter won Outstanding Chapter Award-2005
Large and High-Performance Chapter Award Every Year), the Chair of PELS-
IES Delhi Chapter 2007–2010, and a Founder Chair of PELS-IES Delhi Chapter.
He has been the Chair of IEEE Delhi Section for 2012–2014.

Authorized licensed use limited to: Indian Institute of Information Technology Design & Manufacturing. Downloaded on November 21,2021 at 15:17:10 UTC from IEEE Xplore. Restrictions apply