You are on page 1of 16

Sustainable Energy Technologies and Assessments 60 (2023) 103420

Contents lists available at ScienceDirect

Sustainable Energy Technologies and Assessments


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

Robust variable step P&O algorithm based MPPT for PMSG wind
generation system using estimated wind speed compensation technique
Ilham Toumi a, Amira Boulmaiz b, Billel Meghni c, *, Oussama Hachana d
a
Department of Electronics and Telecommunications, Faculty of New Technologies of Computing and Communication, University of Ouargla, 30000, Algeria
b
Department of Electronics, LERICA Laboratory, University Badji Mokhtar, Annaba, 23000, Algeria
c
Department of Electrical Engineering, LSEM Laboratory, University Badji Mokhtar, Annaba, 23000, Algeria
d
Department of Drilling and Rig Mechanics, Faculty of Hydrocarbons, Renewable Energies, and Earth and Universe Sciences, University of Ouargla, 30000, Algeria

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

Keywords: To expedite the cost recovery of the wind energy conversion system (WECS), maximizing power availability at all
Maximum power point tracking times is of utmost importance. Therefore, the primary objective of Maximum Power Point Tracking (MPPT)
Robust variable step-perturb & observe algorithms is to optimize power output by ensuring the achievement of maximum power under varying wind
Second order sliding mode controller
speeds. The perturbation and observation algorithm are widely recognized as the most well-known technique
Step Size selection normalization and artificial
neural network (ANN)
used in wind energy conversion systems (WECS) due to its versatility. Although this algorithm is valued for its
simplicity and cost-effectiveness, it does have a significant drawback. Peculiarly, it is insensitive to fluctuations
in instantaneous wind speed conditions and lacks accurate step size estimation. To address these limitations, a
novel approach called the Robust Variable Step P&O (RVS-P&O) technique has been proposed. The RVS-P&O
technique aims to overcome the aforementioned drawbacks. The control strategy employed by the proposed
RVS-P&O-MPPT algorithm is based on three fundamental concepts: a new normalization of the selection of the
step size, a novel correction procedure to estimate the power variation and wind speed estimation (WSE) al­
gorithm based on the artificial neural network (ANN) controller. Simulations have been carried out by means of
MATLAB/Simulink environment. The findings indicate the superiority of the proposed RVS-P&O algorithm
compared to several P&O implementations: fixed Large Step (LS), fixed Small Step (SS), Variable Step (VS) and
Modified (M)-P&O. The developed algorithm allows reaching 99.79% of wind energy conversion system (WECS)
efficiency. This allows obtaining an enhancement of 2.19% in comparison to VS-P&O technique which is the
most competing algorithm. Further, the response speed is clearly improved compared to the various P&O ver­
sions, as the settling time (sec) is of 0.0074 while that of VS and M− P&O is of 0.0157 and 0.0079, respectively.
Indeed, to enhance the quality of the injected power into the grid, a Second Order Sliding Mode Controller
(SOSMC) based Super Twisting Algorithm (STA) is proposed. The results show the suggested technique out­
performs the PI and First Order (FO)-SMC for current THD minimization by attenuating the odd harmonics
without chattering phenomenon.

requirements [2]. PMSGs enable the generation of full power across


Introduction varying wind speeds without the need for a multistage gearbox, thereby
increasing the efficiency of the wind energy conversion system (WECS)
Currently, the integration of variable speed wind energy systems by approximately 10% [3].
Over time, control systems have been integrated into wind turbines
(VS-WES) with the grid has significantly enhanced the global power
production contribution of wind turbines (WT) [1]. Compared to to facilitate cost-effective and consistent energy production while
effectively managing wind variations. The primary objective of WT
squirrel cage induction generators (SCIGs) and doubly-fed induction
generators (DFIGs), VS-WES with permanent magnet synchronous gen­ control solutions is to maximize power generation while ensuring a
consistent energy supply to the grid [4]. Achieving these control ob­
erators (PMSGs) offer numerous advantages, including enhanced power
production reliability and elimination of external DC excitation jectives requires proper handling of the Machine and Grid Side

* Corresponding author.
E-mail addresses: toumiilham@yahoo.com (I. Toumi), boulmaiz.ami2012@gmail.com (A. Boulmaiz), maghni_1990@yahoo.fr (B. Meghni), oussama.hachana@
gmail.com (O. Hachana).

https://doi.org/10.1016/j.seta.2023.103420
Received 22 January 2022; Received in revised form 4 July 2023; Accepted 21 August 2023
Available online 31 August 2023
2213-1388/© 2023 Elsevier Ltd. All rights reserved.
I. Toumi et al. Sustainable Energy Technologies and Assessments 60 (2023) 103420

Nomenclature Pg ref , Qg ref Reference grid active and reactive power


ORC Optimal-regimes characteristic
CP , λ, Vk Power coefficient, Tip speed ratio and wind speed WSE/WS Wind speed estimated and Wind speed
ρ, β,S Air density, Pitch angle and Surface VS/WECS Variable speed/ Wind energy conversion system
Fg Grid frequency ANFIS Adaptive neuro fuzzy inference system
K, αL First unknown gain and Weighting factor DFIG Doubly-fed induction generator
L, N Sector and Normalization-index GSC-MSC Grid and Machine-side converter
Pk, PN k Mechanical and Normalized-mechanical power LS, SS, FSS Large step, Small step and Fixed step size
Vdg , Vqg d,q-axis grid voltage PMSG Permanent- magnet synchronous generator
Idg, Iqg d,q-axis grid current PSO Particle swarm optimizer
Pg , Qg Grid active and reactive power SCIG Squirrel-cage induction generator
SP , Sq sliding surface active and reactive power THD Total harmonic distortion
VDC ref Reference DC link voltage SV-PWM Space vector-Pulse-width modulation

Converters (MSC and GSC). The MSC adjusts the generator speed based guaranteed if the change in WS is quick [23]. The system characteristics
on wind speed to capture the maximum power from the wind [5]. and the optimal curve must be calculated before using the ORB tech­
Consequently, WT aerodynamics, rotor speed, and the amount of wind nique. In real-world scenarios, they are complex to quantify and sus­
power collected are heavily influenced by the availability of random ceptible to change [24]. The P&O approach for monitoring optimal rotor
wind conditions. Hence, Maximum Power Point Tracking (MPPT) al­ speed is popular since it doesn’t need WT characteristic curve knowl­
gorithms play a crucial role in VS-WESs [3]. edge (on the optimal regimes characteristic, ORC). It is a mathematical
A wide range of previous studies have addressed various MPPT al­ optimization approach created to find the MPP where the control vari­
gorithms, offering comprehensive control of wind generation systems, as able is the rotor speed. The output mechanical power is monitored until
indicated in Table 1. These algorithms are categorized as direct power the slope hits zero (on the optimal regimes characteristic, ORC) [16].
controllers (DPCs) and indirect power controllers (IPCs) [6]. The classic P&O (CPO) algorithm with FSS has an effect on the general
IPC control techniques have been applied as front-end approaches for performance of the dynamic system. As a result, CPO algorithm includes
MPPT purposes like optimal torque (OT) [17], tip speed ratio (TSR) [14] a variety of difficulties, including failing to get the MPPs during quick
and power signal feedback (PSF) [18]. The TSR algorithm maximizes the wind fluctuations, especially on large-inertia WTs. Further, determining
Cp by maintaining the system in λopt . It is relatively straightforward; a suitable step size remains a complex task that may lead to tracking
however, it requires precise information about the wind speed. An failure and concerns with directionality [25]. A variety of based P&O
anemometer should be implemented to determine the mean real speed variable step-size (VS) algorithms are studied in the literature to solve
throughout the swept surface of the WT blades for defining the MPP the disadvantages of CPO. VS-P&O methods address the problems of
accurately [14]. Meantime, in addition to reducing WECS efficiency, it is speed convergence, oscillation, wind speed measurement and inertia
complicated, costly, and hard to implement. The control goals of PSF effect better [11]. Moreover, P&O algorithms can be classified into
and OT algorithms are based on the generator’s mathematical model, modified and adaptive ones. Ref [26] provides a two-stage modified
particularly the torque constant, which may be determined approxi­ variable step P&O technique for MPPT in WECSs. In this method, an
mately [22]. However, large-inertia of WT track more slowly at low intermediate variable computed from the system parameters has been
WSs, making MPP calculation difficult. The (IPC) MPPT family is not employed. An adaptive sensor-less P&O approach based on the ideal
favored in current wind turbine design, as it requires knowledge of the power curve has been reported in [27]. This solves the P&O algorithm’s
parameters of the controlled system. On the other hand, the use of an­ drift issue. An ORB-P&O method similar to this has been developed in
emometers increases system complexity, as well as initial and mainte­ [28], which can reliably estimate the optimal relationships under
nance costs. In addition, erroneous MPP detection during operation rapidly changing wind speed. However, one of the main drawbacks of
reduces power capture and system efficiency. these methods remain the need to know the parameters of the controlled
By sensing power changes in proportion to the operating WS, the system, which can limit their accuracy. In addition, the small number of
DPC algorithms are employed to achieve the pre-established ideal sectors used can lead to a coarse approximation of the optimum curve.
connection. DPC methods include the perturb and observe (P&O) Ref [19] provides an adaptive and modified P&O algorithm with a fuzzy
[9,10], incremental conductance (INC) [7] and optimum relation-based logic step size controller for WECS-PMSG. The suggested MPPT con­
(ORB) [8]. The INC approach may extract the most power compared to troller’s performance and durability are examined in an experiment
these days’ techniques. Since P/Ω characteristic includes numerous implementation utilizing a WECS emulator and dSpaceDS1104
peaks, its effectiveness is nevertheless restricted and cannot be controller board. Nevertheless, it is essential to measure wind speed, as

Table 1
MPPT Algorithm classification.
Maximum Power Point Tracking Techniques Direct Power Controller Incremental Conductance [7] Conventional P&O [10]
Optimum Relation Based [8] Adaptive P&O [11]
Perturbation & Observation [9] Variable P&O [12]
Modified P&O [13]
Indirect Power Controller Tip Speed Ratio (TSR) [14] Wind Speed Estimation [15]
Optimal Torque [17] Wind Speed Measurement [16]
Power Signal Feedback [18]
Hybrid Controller Conventional + Smart Control [19]
Conventional + Optimizer Control [20]
Smart Controller Particle Swarm Optimization [21]
Fuzzy logic control (FLC) [16]
Artificial Neural network (ANN) [3]

2
I. Toumi et al. Sustainable Energy Technologies and Assessments 60 (2023) 103420

the operating practice depends on time intervals, which can lead to (WT) of different dimensions. Standards have been established for
neglect of variations in the optimum curve as a function of different measuring power and determining the incremental increase in the
wind speeds. A unique Artificial Intelligence based Adaptive (AIA)-P&O reference speed.
for real-time adaptive hybrid MPPT has been described in [29]. This • To enhance robustness and improve observation measurement ac­
method alternates between a considerable forward step and a small curacy, the impact of wind speed variation has been considered.
retrograde step throughout the tracking phase. This method is clear, • The proposed algorithm utilizes an Artificial Neural Network (ANN)
easy to implement, and solves the controller’s initial tracking speed estimated wind speed algorithm to overcome the limitations of
issue [30]. However, the primary limitation of the MPPT-PO technique traditional anemometers. It also determines the most suitable
using artificial intelligence lies in its dependence on training and adaptive step size for the Robust Variable Step (RVS)-P&O algorithm.
learning data. The effectiveness of the AI algorithm is closely linked to • The P/Ω curve has been divided into multiple operating zones based
the quality and representativeness of the data used in the training phase. on a measurement ratio derived from the current and planned
An original P&O based more significant order sliding mode MPPT con­ Maximum Power Point (MPP).
trol for WTs eliminates the rotor inertial effect as proposed by [31]. • The RVS-P&O algorithm addresses the shortcomings of Constant
Experimental results show the suggested control system is 9.55%, it is Perturbation and Observation (CPO) algorithms, thereby enhancing
more efficient than the traditional one. However, the sliding mode the efficiency of variable step versions.
control causes a chattering effect on the MPP turn which results in low • The DPC-SOSMC technique, based on Sliding-Mode Super Twisting
control accuracy, high wear of mechanical parts and MPPT losses. In Algorithm (STA), improves tracking accuracy, grid power quality,
Refs [32,33], authors develop a P&O algorithm with a tunable step size. and reduces chattering levels. Compared to traditional techniques, it
The step size may be estimated by calculating the distance between the offers improved speed and dynamic performance, higher steady-state
operational point and the MPP on the P/Ω curve. A goal function stability, and harmonic attenuation.
dependent on wind speed has been suggested in [6], and the step size is
adjusted accordingly. Under constant air conditions, this approach The paper is organized as follows: Section 2 describes the mathe­
produces remarkable results. However, when the P/Ω curve shows many matical model of the primary components of the Variable Speed Wind
peak points, under random changes of wind speed the performances are Turbine (VSWT). Section 3 discusses the proposed RVS-P&O algorithms.
drastically reduced. Section 4 provides details on the control of the grid tie inverter. To
These techniques offer essential solutions for overcoming the draw­ demonstrate the effectiveness of the suggested RVS-P&O algorithm,
backs of CPO algorithms. Nevertheless, the dependence on wind speed several simulations were conducted using the MATLAB/SIMULINK
measurement, wind turbine parameters and inertia are significant environment. The results are presented, analyzed, and discussed in
weaknesses. In addition, the above-mentioned algorithms involve Section 5. Finally, the conclusions and future perspectives are presented
numerous computational operations and are based on assumptions that in the last section.
increase the complexity of the system. Furthermore, the limits imposed
on the sectors can lead to a less-than-optimal operating power curve, Model of the VSWT and problem formulation
resulting in higher levels of oscillation at low wind speeds.
At the level of GSC, a DPC-SOSMC based super-twisting algorithm This section presents the VS-WES configuration examined for control
(STA) controller is recommended to reach a satisfying extracted power purposes. The typical topology of the considered VS-WES is illustrated in
distribution. Therefore, the generator and the grid demand can ex­ Fig. 1. In the diagram, a three-bladed horizontal axis turbine is illus­
change active and reactive power amounts without interruption during trated, with the rotor directly connected to the shaft of a Permanent
real changes in wind speed [16]. Magnet Synchronous Generator (PMSG) [3]. Two back-to-back AC-DC-
To improve the dynamic performance of current P&O algorithms and AC IGBT bridges are connected to a shared DC bus, connected to the grid
overcome their weaknesses. This study presents a newly developed [34].
Maximum Power Point Tracking (MPPT) algorithm called the Robust
Variable Step (RVS)-P&O, which was inspired by the preceding discus­ Model of the VSWT
sion. The RVS-P&O algorithm has undergone extensive testing to
address the limitations of the Constant Perturbation and Observation Cp (λ, β) and the aerodynamic power delivered to the generator shaft
(CPO) approach, particularly in terms of longer response time, the are expressed by Eq. (1) [35].
inertia effect of the wind turbine (WT), suppression of speed sensor in­ ⎧
fluence, and determination of the appropriate step size. Moreover, the ⎪


⎪ 1
algorithm allows for adaptive adjustment of the set-point speed step in


⎪ Pk = ρ.SCp (λ, β).Vk3

⎪ 2
response to the inaccuracies of the Maximum Power Point (MPP) ⎪


⎪ R × Ωk

determination while adjusting the P/Ω values. To achieve this, a pro­ ⎪


λ=
Vk

portional modification of the step size is proposed. In addition, an ac­ ⎨ ( )
151 (1)
curate wind speed estimation technique based on Artificial Neural
− 18.4
2.14

⎪ C p = 0.073 − 0.058β − 0.002β − 13.2 e λi
⎪ λi
Networks (ANN) is employed to overcome the limitations of traditional ⎪




anemometers and optimize the wind power output. For performance ⎪

⎪ λi =
1

analysis purposes, the RVS-P&O algorithm, along with several P&O ⎪



1
− 2
0.003

variants, was implemented using similar simulation parameters and ⎪
⎩ λ − 0.02β β +1
conditions. The proposed technique has been evaluated in diverse
environmental settings, including scenarios with gradual changes in The aerodynamic efficiency changes according to λ. In other words,
wind speed, experiments with random variations and real-world data when it is optimum (8.1) λopti , the maximum (0.48) Cpmax is attained and
from the Renewable Energy Research Unit in the Saharan Adrar region the turbine operates on the ORC [3,36].
in 2018, as well as the Saharan Ghardaia region in 2006 and 2016.
The following list provides a summary of the main contributions of Problem formulation
this paper:
The classic Perturbation and Observation method (CPO) are one of
• Standardization of speed variation and observation measurement the most fundamental Maximum Power Point Tracking (MPPT) strate­
allows for satisfactory controller performance across wind turbines gies. It provides extensive application in both practical and research

3
I. Toumi et al. Sustainable Energy Technologies and Assessments 60 (2023) 103420

Fig. 1. Description of the wind system.

settings due to its simplicity and ease of implementation [33]. In this is achieved by incorporating energy change corrections and utilizing an
algorithm, changes in power are observed following each perturbation appropriate simulator.
in the reference speed. If the harvested power of the system increases,
the rotor speed is adjusted in the similar direction. Conversely, if the • For the purpose of providing a synthesized technique for estimating
power decreases, the speed is modified in the opposite direction. The the step size set-point, instantaneous WS has been taken into
flowchart of the CPO algorithm is illustrated in Fig. 2. The tracking di­ consideration. The maximum mechanical power Pmax k is given by Eq.
rection depends on the sign of ΔP; the analysis of this algorithm makes it (1) and determined by ideal values of λopt and Cpmax .
possible to identify the following rules:
The controller dynamics under variable sizes could be optimized by
• If ΔPk > 0, the speed set-point is augmented by ΔΩ steps. standardizing the speed set-point increasing and the power measure­
• If ΔPk < 0, the speed set-point is reduced by ΔΩ steps. ment. Eq. (2) defines the normalized mechanical power (PNk ).

Moreover, selecting the appropriate step size remains a challenging Pk


PNk = × 100 (2)
task. A larger step size results in faster response and more oscillations Pmax
k

around the peak point, thereby reducing efficiency. Conversely, a


smaller step size improves efficiency but slows down convergence. • For constant step reference, the controller takes much more time to
These considerations are summarized in Table 2 [37]. track the MPP under important fluctuations of WS. Since it is not an
adaptive behavior, the outcome will remain the same regardless of
Proposed RVS-P&O MPPT technique little changes in WS. To achieve the MPP and prevent slow reaction,
it is suggested changing the step size proportionally to the control
The newly proposed Robust Variable Step Perturbation and Obser­ signal. The level (L − 1) is selected to be delimiter for normalizing the
vation (RVS-P&O) MPPT technique aims to mitigate the impact of high power into several sectors l = (1, ⋯, L).
wind turbulence by standardizing the mechanical energy variables. This

4
I. Toumi et al. Sustainable Energy Technologies and Assessments 60 (2023) 103420

Fig. 2. CPO algorithm.

Considering Plmax is the maximum power in each sector, represented the RVS-P&O approach. By considering only the portion of ΔPk
by βl ratio in the range of [0, 1]. Eq. (3) shows Plmax and based on this, the resulting from the speed modification in the previous step, this
ref disturbance can be eliminated. As a result, the control algorithm
sector step size (ΔΩk ) is adjusted by αl based on Ωopt
k . Where l = 1, ⋯, L
becomes more resilient to disruptions caused by variations in WS.
indicates the index of the sector, αl is in the range of [0, 1].
The power at any given moment depends on both the WS and the

⎪ Plmax = βl .Pmax turbine speed. Thus, the mechanical power variation (ΔPk ) and its

⎪ k

⎨ first-order approximation at instant k can be expressed by Equation
ΔΩref
k = αl × Ωk
opt
(3) (4):



⎪ λopt × Vk ⎧
⎩ Where Ωopt = ΔPk = Pk (Ωk , vk ) − Pk (Ωk− 1 , vk− 1 )
k
R ⎪
⎨ ⃒
∂f ⃒ (4)
It indicates the magnitude of the speed adjustment relative to the ⎩ ΔPk ≃ f (Ωk , vk− 1 ) + ⃒⃒
⎪ Δvk − f (Ωk− 1 , vk− 1 )
∂v (Ωk ,vk− 1 )
optimal speed. The algorithm requires shifting αl from the lower sector
to the higher sector. The key components of the algorithm are the speed The second component in Eq. (4) (Δvk ) represents the wind speed
step increment and the direction of power variation. The set-point speed variation disturbance. In the case of a constant wind speed, this term is
step will be modified based on the performance outcome: if the per­ null. Eq. (5) is used to get the tuned power change (ΔPωk ):
formance is favorable, the step will increase; if it is unfavorable, the step ⃒
will decrease. ΔPωk = ΔPk −
∂f ⃒⃒
Δv (5)
∂v⃒(Ωk ,vk− 1 ) k
• Since the power change (ΔPk ) is influenced by the change in wind
In practice, ΔPk is estimated through the power error calculation
speed (WS), which in turn causes disturbances, the effectiveness of
under successive samples (k and k − 1). The flowchart of this RVS-P&O
the technique is reduced. This is one of the justifications for adopting

5
I. Toumi et al. Sustainable Energy Technologies and Assessments 60 (2023) 103420

Table 2
Scenarios of Cp case studies using CPO algorithm.
Speeds Variation Function point

ΔPw > 0(left) ΔPw < 0(righ)

MPP1 MPP3
+ Δωref , − Δv
P2 P5

MPP2

Mechanical Power (W)

Mechanical Power (W)


MPP2 P3
P1
P3 MPP1
P4
P1

P2

Rotor speed (rad/s) Rotor speed (rad/s)

− Δωref , + Δv MPP1
MPP1
P1 P2
Mechanical Power (W)

Mechanical Power (W)


P2
MPP2 P1
MPP2
P3 P5 P3
MPP3

P4

Rotor Speed (rad/s)


Rotor Speed (rad/s)

MPPT algorithm is shown in Fig. 3 and the algorithm’s setting param­ Grid tie inverter
eters are depicted in Table 3.
The GSC (Grid Side Converter) is primarily utilized to connect the
Wind speed estimation Permanent Magnet Synchronous Generator Wind Turbine (PMSG-WT)
to the utility grid. Its essential function is to facilitate the control of
In the Variable Speed Wind Energy System (VS-WES), the wind is appropriate power injection under various operating conditions by uti­
considered the primary driving force, converting kinetic energy into lizing the DC link [36]. To achieve this, the STA (Space Vector Modu­
mechanical power and eventually into electrical energy, as described by lation) DPC-SVM-SOSMC (Direct Power Control-Space Vector
Equation (1). Therefore, it is crucial to accurately measure the effective Modulation-Sliding Mode Super Twisting Control) technique is
wind speed experienced by the turbine blades. However, incorporating employed to regulate the power injected into the grid. Unlike the vector
an anemometer into the VS-WES configuration presents challenges in technique [38], this approach directly incorporates the grid voltage into
terms of setup, maintenance, and reliability. Inaccurate wind speed the GSC control. Fig. 1 illustrates the control diagram of the GSC, and the
measurements can result in unsuitable reference speed settings for the grid powers are mathematically expressed in the grid d-q plane by Eq (6)
generator, leading to reduced power extraction. To overcome this issue, [39].
wind estimation algorithms are employed to estimate the effective wind ⎧ ⎧
dIdg 3
speed without relying on mechanical sensors. The proposed approach, ⎪

⎨ Vdg = Vdi − Rg Idg − Ldg + Lqg wg Iqg ⎪

⎨ Pg = Vdg Idg
dt 2
based on an Artificial Neural Network (ANN), enables the calculation of Where (6)
⎪ ⎪
wind speed for a wind turbine based on its mechanical power and ⎪
⎩ Vqg = Vqi − Rg Iqg − Lqg dI qg
− Ldg wg Idg ⎩ Qg = 3Vdg Iqg

rotational speed. dt 2
The principle of wind speed estimation based on the Artificial Neural
Network (ANN) is illustrated in Fig. 4. For the neural estimator created DPC-SVM based higher order SMC
using Matlab/Simulink, the hidden layers consist of 30 neurons, while
the output layer consists of 1 neuron with activation functions of the It is prominent that the standard Sliding Mode Control (SMC) tech­
“tansig” and “purelin” types, respectively. The Levenberg-Marquardt nique exhibits chattering in practical operations [40]. To address this
(LM) algorithm, a backpropagation algorithm, is utilized to update the issue, employing SMC with a more significant order represent a viable
weights and biases of this network. The multi-layer perception network approach. This approach significantly reduces the occurrence of unde­
consists of L layers, where each neuron in a layer is fully connected to sired oscillations while preserving the effectiveness of the first-order
the neurons in the subsequent layer. Each neuron, denoted as k, can be controller [41]. To reach a zero-power factor, the reactive power is
described as a generalized linear automaton with its own activation adjusted to Qgref = 0, meanwhile the active power Pgref is determined by
function, fk . the grid demand.
The (SQ ,SP ) and (s̈P ,s̈Q ) are power sliding surfaces and its second
derivatives (See)Eq. (7)

6
I. Toumi et al. Sustainable Energy Technologies and Assessments 60 (2023) 103420

Fig. 3. Detailed flowchart of the RVS-P&O based MPPT technique.

7
I. Toumi et al. Sustainable Energy Technologies and Assessments 60 (2023) 103420



⎪ ⎧

⎪ V̇ id ⎪ 1.5Vdg ( )
⎪ ⎪ GP = Ṗgref −
{ ⎨ s̈P = Ġp − Lg ⎪
⎪ ⎨ Lg
− Vdg − Rg Idg + Lg wg Iqg
sP = Pgref − Pg
⇒ (7)
sQ = Qgref − Qg ⎪ ⎪ V̇ iq ⎪
⎪ 1.5Vqg ( )

⎪ s̈Q = ĠQ − ⎪
⎩ GQ = Q̇gref − − Vqg − Rg Iqg − Lg wg Idg

⎪ Lg
⎩ Lg

The STA introduced by Levant can be determined using Eq. (8) [41]. performed with varying wind speed (WS) profiles across three various
Wile, K and M are unknown parameters owing to the sliding manifolds case studies. In the first scenario, the wind speed fluctuated in distinct
convergence and could be limited as described below [42]. Where: C0 , steps, while the second scenario involved a casual WS profile. The third
Km and KM are positive constants. scenario explored real-world WS behavior. The control parameters of
⎧ √̅̅̅̅̅̅̅ ∫ the system under consideration are presented in Table A1 in the ap­
⎪ ref eq
⎪ Vp = Vp − M |sP | sign(sP ) − K sign(sP )
⎪ pendix. Performance metrics were calculated for the proposed RVS-P&O



⎪ √̅̅̅̅̅̅̅̅ ∫ technique and the compared MPPT algorithms, detailed in Tables 4, 5,



⎪ ref eq and 6, providing insights into the findings.
⎨ VQ = VQ − M |sQ | sign(sQ ) − K sign(sQ )

(8)

⎪ C0




K>
Km
0 < ρ < 0.5 Wind speed variation




⎪ 4C0 KM (K − C0 )

⎩ M2 ≥ if ρ = 0.5 Many variations steps
Km 2 Km (K − C0 ) Fig. 5 illustrates the results of the various methods under a pro­
gressive change in wind speed (WS). In Fig. 5(a), the predicted wind
Results analysis and discussion speed is compared to the actual wind speed, providing a visual com­
parison of the performance of the different algorithms. The WS is dis­
A comprehensive comparison was conducted to evaluate the per­ played according to a wind speed pattern to facilitate a comprehensive
formance and effectiveness of the proposed Robust Variable Step analysis of the competing algorithms. Evaluating the success of the
Perturbation and Observation (RVS-P&O) technique. Four P&O-based proposed strategy primarily involves examining the ideal values of Cp
approaches were implemented as benchmarks for comparison: Standard and λ. The behavior of these values is described in Fig. 5(b) and (c),
Step (SS), Large Step (LS), Variable Step (VS) [32], and Modified P&O respectively.
(M− P&O) [38]. Using MATLAB/SIMULINK, multiple simulations were The RVS-P&O algorithm demonstrates a rapid tracking of Cp in
comparison to the other algorithms. It achieves a 5% response time of
Table 3 0.0074 sec during the initial wind speed change period of 0–5 sec at V =
Proposed MPPT algorithm setting parameters. 6.6 m/sec. This response time is superior to that of the SS, LS, VS, and
M− P&O methods, which yield response times of 0.0667, 0.0184,
Parameters Sector: l = 1 Sector: l = 2 Sector: l = 3 Sector: l = L
0.0157, and 0.0079 sec, respectively.
βL− 1 0.6 0.4 0.01 0 During a sudden variation in the transient response at 15 sec (from
αL 0.03 0.02 0.01 0.0001
9.5 m/sec to 11.4 m/sec), the SS, LS, VS, and M− P&O methods exhibit

Start

(R. t )/Vt Step 1: Taking measurements at different moments

Of wind speed changes k=k+1

Step 2: Read database from excel data file

Step 3: Split the input data into training subsets


1/(Jp+f )
Step 4: Load data ANN

SOSMC Step 5: ANN inputs and outputs

Ɛ
Vest
MPPT Step 6: ANN training program

Update

Wind Speed
Estimator ANN End

(a) WS estimation based on ANN (b) Flowchart of ANN algorithm

Fig. 4. The principle of wind speed estimation based on ANN.

8
I. Toumi et al. Sustainable Energy Technologies and Assessments 60 (2023) 103420

Table 4
Extracted powers under Scenario I.
Parameter WT curves P&O Techniques

SS LS VS M RVS

Wind speed: [0-5sec]V = 6.6m/s Coefficient Settling time (sec) 0.0667 0.0184 0.0157 0.0079 0.0074
Power Undershoot (%) 21,062 21,770 97,916 21,145 2,541
Tip Speed Settling time (sec) 0.093 0.188 0.017 0.011 0.0097
Ration Undershoot (%) 5,061 7,407 79,012 0.0 0.0
Settling time (sec) 0.082 0.022 0.019 0.023 0.0098
Rotor Speed Undershoot (%) 10,214 7,142 78,571 5.123 4.291
Power power loss (W) 530.052 110.007 11.128 03.528 0.353
efficiency (%) 66.43 93.68 97.60 98.92 99.79

Wind speed: [5-10sec]V = 7.5m/s Coefficient Settling time (sec) 0.073 0.010 0.205 0.007 0.004
Power Undershoot (%) 16.451 12,229 85,416 4,645 2,541
Tip Speed Settling time (sec) 0.106 0.02 0.245 0.006 0.003
Ration Undershoot (%) 22,222 19,506 69,135 11,530 8,641
Settling time (sec) 0.029 0.021 0.266 0.017 0.006
Rotor Speed Undershoot (%) 19,354 22,580 68,709 16,129 09.250
Power Power Loss (W) 378.169 106.351 21.386 07.025 0.513
Efficiency (%) 72.11 94.06 96.13 98.36 99.56

Wind speed: [10-15sec]V = 9.5m/s Coefficient Settling time (sec) 0.07 0.012 0.072 0.013 0.010
Power Undershoot (%) 13.729 13,583 27,166 4,895 13,958
Tip Speed Settling time (sec) 0.08 0.02 0.235 0.013 0.0084
Ration Undershoot (%) 4,938 20,987 20,987 20,864 20,493
Settling time (sec) 0.348 0.02 0.21 0.030 0.010
Rotor Speed Undershoot (%) 2,852 19,289 11,315 1,263 1,013
Power Power Loss (W) 150.125 110.861 40.198 13.705 1.698
Efficiency (%) 74.26 93.24 95.22 98.02 99.31

Wind speed: [15-20sec]V = 11.4m/s Coefficient Settling time (sec) 0.09 0.011 0.124 0.009 0.005
Power Undershoot (%) 3,562 17,166 97,083 9,062 8,333
Tip Speed Settling time (sec) 0.06 0.03 0.330 0.015 0.009
Ration Undershoot (%) 6,172 23,456 81,481 16,666 16,172
Settling time (sec) 0.042 0.032 0.314 0.028 0.011
Rotor Speed Undershoot (%) 6,956 22,826 81,086 2,159 0,728
Power Power Loss (W) 532.358 128.469 61.119 09.127 0.986
Efficiency (%) 66.12 93.49 94.36 98.29 99.47

Table 5 Table 6
Comparison results of the MPPT methods under test scenario II. Comparison results of the MPPT methods under test scenario III.
P&O Cp λ Pω Tracking accuracy(%) Power P&O Ghardaia Region- Ghardaia Region Adrar Region
Techniques average average average ripple Techniques 2006 − 2016 − 2018
(%)
Cp λ Cp λ Cp λ
SS 0.3393 5.9856 1.369 e 96.377 57.159 average average average average average average
+3
SS 0.477 8.019 0.476 8.124 0.469 8.120
LS 0.4544 8.5942 1.465e 69.425 95.122
LS 0.361 8.554 0.405 8.154 0.466 8.023
+3
VS 0.462 8.182 0.421 8.961 0.318 11.698
VS 0.2402 11.1890 1.105 e 61.415 49.159
M 0.427 9.295 0.410 9.388 0.476 8.041
+3
RVS 0.479 8.097 0.477 8.096 0.478 8.061
M 0.4748 8.1433 1.332 e 96.989 28.850
+3
RVS 0.4773 8.0685 1.613 e 99.158 21.178
+3 method demonstrates superior performance in terms of rapidity, as
evidenced by the provided settling time of 0.0097 sec.
The zoomed-in section of Fig. 5(d) illustrates the settling time of the
significant continuous oscillations around the MPP. However, the rotor speed, which is approximately 0.0098 sec for RVS-P&O. In
settling times for the M− P&O and RVS-P&O methods, as shown in the contrast, the other methods, namely SS, LS, VS, and M− P&O, exhibit
zoomed-in section of Fig. 5(b), are notably similar. The undershoot settling times of 0.082, 0.022, 0.019, and 0.023 sec, respectively.
percentages (%) achieved by the SS, LS, VS, and M− P&O methods are Further, when compared to the alternative methods, RVS-P&O and
21.062, 21.770, 97.916, and 21.145, respectively. However, the RVS- M− P&O do not demonstrate any significant overshoot in controlling the
P&O technique demonstrates a significant reduction in undershoot rotor speed. Meantime, M− P&O and RVS-P&O achieve lower to un­
(2.541%) even with rapid wind changes, without any overshoot occur­ dershoot values (5.12% and 4.29%) in the wind speed range of 0–5 sec at
ring within the first 5 s of a wind speed change at V = 6.6 m/sec. Further, V = 6.6 m/sec, as shown in Fig. 5(d). Fig. 5(e) displays the trend of
Fig. 5(c) illustrates that the RVS-P&O method effectively maintains the mechanical power to confirm the superior power extraction perfor­
ideal value of λ (8.100). It also achieves optimal operation with minimal mance of RVS-P&O. The power oscillations using SS, LS, VS, and
settling time when confronted with rapid changes in wind speed, as M− P&O are substantial and continuous around the extracted MPP. On
summarized in Table 4. In contrast, the other methods (SS, LS, VS, and the other hand, RVS-P&O exhibits no power fluctuations even with rapid
M− P&O) exhibit undershoots percentages (%) of 5.061, 7.407, 79.012, changes in wind speed. Additionally, RVS-P&O achieves the new MPP
and 0.0, respectively, with some degree of overshoot. The RVS-P&O more directly compared to other algorithms. For instance, it merely

9
I. Toumi et al. Sustainable Energy Technologies and Assessments 60 (2023) 103420

Fig. 5. Machine side results under WS gradually changes.

10
I. Toumi et al. Sustainable Energy Technologies and Assessments 60 (2023) 103420

Fig. 6. Machine side results under WS high fluctuations.

11
I. Toumi et al. Sustainable Energy Technologies and Assessments 60 (2023) 103420

takes 0.014 sec to maintain full power and stability during a sudden shift namely SS (5.985), LS (8.594), VS (11.189), and M− P&O (8.143). In
at 15 sec, as illustrated in Fig. 5(e). This is more fastened than the times Fig. 6(d), the results of rotor speed tracking are displayed. It is note­
(sec) of 0.07, 0.040, 0.39, and 0.035 needed by SS, LS, VS, and M− P&O, worthy RVS-P&O can swiftly adjust the generator speed in situations of
respectively, as shown in the zoomed-in part of Fig. 5(e). The findings of fast fluctuation, with minimal ripples, unlike the other methods. The
RVS-P&O and the other techniques for the first scenario are summarized proposed method achieves convergence more quickly and with less
in Table 4. speed error compared to the competing techniques. The extracted power
for the 12 sec under the similar conditions is estimated to be (1.613e +
High fluctuations of wind speed 3) W using RVS-P&O, while it is (1.369e + 3), (1.465e + 3), (1.105e +
3), and (1.332e + 3) for SS, LS, VS, and M− P&O, respectively. The
Fig. 6 showcases the results of the competing algorithms in the ma­ tracking accuracy (%) using RVS-P&O is 99.158, making it the most
chine side under variable wind speed fluctuations. The system was accurate among the competing methods, with SS (96.377) and M− P&O
simulated with a nominal wind speed of 9 m/sec. In Fig. 6(a), the pre­ (96.989) remain the closest competitors. Fig. 6(e) illustrates the effi­
dicted wind speed is compared to the actual wind speed, demonstrating ciency of the algorithms in tracking the MPP during random changes in
the correlation between the two. The zoomed-in section highlights the wind speed. It is worth noticing that P&O-based methods exhibit per­
ability of RVS-P&O to rapidly reach the ideal power coefficient (0.48) turbing oscillations in mechanical energy waveforms, impacting energy
compared to the other algorithms. It is evident that under these chal­ quality. The predicted power ripple (%) is 21.178 when using RVS-P&O,
lenging operating conditions, all competing algorithms struggle to while it is 57.159, 95.122, 49.159, and 28.850 for SS, LS, VS, and
effectively track the MPP, and the issue of perturb misdirection further M− P&O, respectively.
involves their ability to do so. On the other hand, the proposed RVS-P&O
method efficiently reaches its maximum operating point along the Real fluctuations of wind speed
optimal rotor speed curve (ORC) throughout the 12 sec of variable wind
speed, with an average of 0.4773. The average results of the investigated The proposed approach has been tested in two separate Saharan lo­
approaches are summarized in Table 5. Fig. 6(c) illustrates the optimal cations with real-world WS variations:
TSR value maintained by RVS-P&O. The RVS-based P&O method
effectively maintains the optimal TSR value without any overshoot, with
an average of 8.068. This outperforms the other competing approaches,

(a1) Wind speed (Ghardaia-Algeria-2006) (a2) Wind speed (Ghardaia-Algeria-2016)

(b1) Power coefficient (b2) Power coefficient

Fig. 7. Machine side results under WS real fluctuations.

12
I. Toumi et al. Sustainable Energy Technologies and Assessments 60 (2023) 103420

Fig. 7. (continued).

- Region 1: Wind speed changes from Ghardaia-Algeria (Renewable As demonstrated in Fig. 7(a1, a2 and a3), the discrepancy between
Energy Research Unit) for the years 2006 and 2016 shown in Fig. 7 expected and actual wind speed is very small. Table 6 illustrates how the
(a1, a2) respectively, with a speed of more than 14 m/s. RVS-based P&O outperforms the other techniques. The Cp average value
- Region 2: as shown in Fig. 7(a3), the WS changes are taken from the has been evaluated at 0.479, 0.477 and 0.478 according to various wind
Renewable Energy Research Unit of Adrar-Algeria on January 2018. speeds by using RVS with an interesting rapid reaction as in Fig. 7(b1, b2
and b3). This is due to the suggested corrections to the standard P&O to

13
I. Toumi et al. Sustainable Energy Technologies and Assessments 60 (2023) 103420

Fig. 8. Results for the FOSMC and SOSMC techniques on the grid.

solve the trajectory loss and the tracking miss-direction under realistic 2006).
climatic circumstances. The based P&O algorithms, M− P&O method has
indicated higher oscillations; meanwhile, LS, VS and SS have shown DPC-SVM-SOSMC based STA for GSC
lower oscillations. Further, RVS average is of 8.097. It is better than the
average reached by means of: SS (8.019), LS (8.554), VS (8.182) and M In this strategy, the powers have been estimated from the grid
(9.295) based P&O techniques under wind speed (Ghardaia Region- accessible variables. Direct power control (DPC) has been emerged as a

14
I. Toumi et al. Sustainable Energy Technologies and Assessments 60 (2023) 103420

vector control technique; meanwhile, because of the variable switching behavior, the loss of tracking issue has been avoided as well as the
frequency, it has shown a significant disadvantage of getting high har­ performance of dynamic tracking has been increased. The following
monics of the generated currents. For this purpose, DPC-SVM-SOSMC points highlight the key findings of this study:
based STA has been suggested.
In Fig. 8(a), it can be observed that the converter is designed to • The RVS-P&O algorithm subdivides the P/Ω curve into various
maintain the DC link voltage at 800 V. The DPC-SVM techniques is operating areas by adaptively adjusting their boundary limits and
employed to adapt the generated power into the grid, as depicted in gradually decreasing the step size.
Fig. 8(b) and Fig. 8(c). The control units of FOSMC and SOSMC are • The observation measurement has been corrected for WS disturbance
capable of accurately tracking the target value, but they differ in terms to increase the robustness.
of power calibration. However, as shown in Fig. 8(b) and Fig. 8(c), the • The RVS-PO algorithm requires accurate WS estimation which
proposed SOSMC control strategy outperforms FOSMC. It provides a maintains potential advantages for wind turbines located at complex
smooth intended slip path and higher efficiency without chattering and regions.
oscillations. Fig. 8(d) illustrates the amount of current delivered into • Accurate wind speed estimation based on ANN has proven to be
grid phase A. The current obtained with FOSMC exhibits a higher total beneficial in terms of increased energy extracted.
harmonic distortion (THD) of 1.38%, indicating poor-quality electrical • With a quick reaction time of 0.0098 sec, the RVS based P&O tracks
power being supplied to the grid. However, as shown in Fig. 8(f), the effectively the MPP under rapid climatic fluctuations.
power quality remains unsatisfactory and the injected current is unde­ • The proposed RVS increases the WECS efficiency by 33.36% and
sirable. On the other hand, the superiority of SOSMC is evident in the 2.19% compared to SS and VS methods-based P&O, respectively.
smooth form of the current, as depicted in Fig. 8(g). The SOSMC • Regardless of the operating circumstances, the RVS method exhibits
approach significantly reduces current distortion by 0.980%. Approxi­ an important level of stability with little changes around the MPP
mately, it attenuates 40% of the odd inherent harmonics compared to with an average of the energy loss of 0.353 W.
FOSMC. Therefore, the implementation of DPC-SVM-SOSMC enables the • A specified SOSMC based STA has been introduced to control the
elimination of sinusoidal currents, minimizes harmonics, and achieves a active and reactive powers at the level of GSC. The reached grid
unit power factor. power values indicate lower THD rate (0.980%) and smooth wave­
forms with adequate tracking indices.
Conclusion • The electrical power quality has been significantly raised in the grid.

Effective maximum power point tracking is necessary to keep the


wind turbine system behavior optimal. Therefore, a novel MPPT tech­ Declaration of Competing Interest
nique denominated RVS-P&O has been proposed. In this technique, the
P/Ω curve has been subdivided into various horizontal operating sectors The authors declare that they have no known competing financial
with an introduction of a synthesized ratio allowing to properly pre­ interests or personal relationships that could have appeared to influence
dicting the required variable step size. This is performed for each the work reported in this paper.
operating sector which permits to get the right MPP and then avoiding
the misdirection and the tracking loss issues. The proposed technique Data availability
has been tested and simulated under several scenarios of WS variation. It
has been compared to various based P&O methods to illustrate its su­ No data was used for the research described in the article.
periority. RVS based P&O increases the WECS efficiency by 2.19% and it
follows the MPP quickly (0.0098 sec) with minimum oscillations (0.005 Acknowledgment
rad/sec) to reach ORC. Since the suggested RVS-P&O properly achieve
the MPP with higher performances, either at transient and steady-state The authors received no specific funding for this study.

Appendix

• βl Determines the size of each sector. When the operational point is near of far from the MPP the tracking action should be made. Around the
MPP, a step size modification on a finer scale has to be performed, where the following rules are suggested Eq (A1):

⎪ 1

⎨ l=1
2
βl = (A1)

⎩ 3 βl− 1 1 < l < L

2
This will enable a rapid first reaction in the case of steady-state disturbances.

Table A1
WT setting parameters.
Radius of the turbine Rt = 2m

Air density ρ = 1.225kg.m3


Pitch angle β = 0◦
Optimal tip speed ratio λopti = 8.1
power Coefficient Cp = 0.48

15
I. Toumi et al. Sustainable Energy Technologies and Assessments 60 (2023) 103420

References [22] Alanis AY. Adaptive neural sensor and actuator fault-tolerant control for discrete-
time unknown nonlinear systems. Franklin Open 2022;1:9–16.
[23] Yu KN, Liao CK. Applying novel fractional order incremental conductance
[1] Sridhar S, Joseph J, Radhakrishnan J. Implementation of tubercles on vertical axis
algorithm to design and study the maximum power tracking of small wind power
wind turbines (VAWTs): An aerodynamic perspective. Sustainable Energy Technol
systems. J Appl Res Technol 2015;13(2):238–44.
Assess 2022;52:102109.
[24] Abdullah MA, Al-Hadhrami T, Tan CW, Yatim AH. Towards green energy for smart
[2] Daili Y, Gaubert JP, Rahmani L, Harrag A. Quantitative feedback theory design of
cities: Particle swarm optimization based MPPT approach. IEEE Access 2018;6:
robust MPPT controller for small wind energy conversion systems: design, analysis
58427–38.
and experimental study. Sustainable Energy Technol Assess 2019;35:308–20.
[25] Mousa HH, Youssef AR, Hamdan I, Ahamed M, Mohamed EE. Performance
[3] Kumar R, Agrawal HP, Shah A, Bansal HO. Maximum power point tracking in wind
assessment of robust P&O algorithm using optimal hypothetical position of
energy conversion system using radial basis function based neural network control
generator speed. IEEE Access 2021;9:30469–85.
strategy. Sustainable Energy Technol Assess 2019;36:100533.
[26] Kumar D, Chatterjee K. A review of conventional and advanced MPPT algorithms
[4] Ahmed SD, Al-Ismail FS, Shafiullah M, Al-Sulaiman FA, El-Amin IM. Grid
for wind energy systems. Renew Sustain Energy Rev 2016;55:957–70.
integration challenges of wind energy: A review. IEEE Access 2020;8:10857–78.
[27] Menezes EJN, Araújo AM, Da Silva NSB. A review on wind turbine control and its
[5] Tripathi SM, Tiwari AN, Singh D. Grid-integrated permanent magnet synchronous
associated methods. J Clean Prod 2018;174:945–53.
generator based wind energy conversion systems: A technology review. Renew
[28] Putri RI, Pujiantara M, Priyadi A, Ise T, Purnomo MH. Maximum power extraction
Sustain Energy Rev 2015;51:1288–305.
improvement using sensorless controller based on adaptive perturb and observe
[6] Mousa HHH, Youssef A-R, Mohamed EEM. State of the art perturb and observe
algorithm for PMSG wind turbine application. IET Electr Power Appl 2018;12(4):
MPPT algorithms based wind energy conversion systems: A technology review. Int
455–62.
J Electr Power Energy Syst 2021;126:106598.
[29] Khan MJ. An AIAPO MPPT controller based real time adaptive maximum power
[7] Mei Q, Shan M, Liu L, Guerrero JM. A novel improved variable step-size
point tracking technique for wind turbine system. ISA Trans 2022;123:492–504.
incremental-resistance MPPT method for PV systems. IEEE Trans Ind Electron
[30] Karami N, Moubayed N, Outbib R. General review and classification of different
2010;58(6):2427–34.
MPPT Techniques. Renew Sustain Energy Rev 2017;68:1–18.
[8] Abdullah MA, Yatim AHM, Tan CW, Saidur R. A review of maximum power point
[31] Karabacak M. A new perturb and observe based higher order sliding mode MPPT
tracking algorithms for wind energy systems. Renew Sustain Energy Rev 2012;16
control of wind turbines eliminating the rotor inertial effect. Renew Energy 2019;
(5):3220–7.
133:807–27.
[9] Ramadan H, Youssef AR, Mousa HH, Mohamed EE. An efficient variable-step P&O
[32] Mousa HH, Youssef AR, Mohamed EE. Variable step size P&O MPPT algorithm for
maximum power point tracking technique for grid-connected wind energy
optimal power extraction of multi-phase PMSG based wind generation system. Int J
conversion system. SN Applied Sciences 2019;1(12):1–15.
Electr Power Energy Syst 2019;108:218–31.
[10] Lahfaoui B, Zouggar S, Mohammed B, Elhafyani ML. Real time study of P&O MPPT
[33] Mousa HH, Youssef AR, Mohamed EE. Adaptive P&O MPPT algorithm based wind
control for small wind PMSG turbine systems using Arduino microcontroller.
generation system using realistic wind fluctuations. Int J Electr Power Energy Syst
Energy Procedia 2017;111:1000–9.
2019;112:294–308.
[11] Mousa HH, Youssef AR, Mohamed EE. Study of robust adaptive step-sizes P&O
[34] Meghni B, Dib D, Azar AT, Saadoun A. Effective supervisory controller to extend
MPPT algorithm for high-inertia WT with direct-driven multiphase PMSG.
optimal energy management in hybrid wind turbine under energy and reliability
International Transactions on Electrical Energy Systems 2019;29(10):e12090.
constraints. International Journal of Dynamics and Control 2018;6(1):369–83.
[12] Ali MM, Youssef AR, Ali AS, Abdel-Jaber GT. Variable step size PO MPPT algorithm
[35] Jamil E, Hameed S, Jamil B, Qurratulain. Power quality improvement of
using model reference adaptive control for optimal power extraction. International
distribution system with photovoltaic and permanent magnet synchronous
Transactions on Electrical Energy Systems 2020;30(1):e12151.
generator based renewable energy farm using static synchronous compensator.
[13] Linus RM, Damodharan P. Maximum power point tracking method using a
Sustainable Energy Technol Assess 2019;35:98–116.
modified perturb and observe algorithm for grid connected wind energy
[36] Meghni B, Dib D, Azar AT. A second-order sliding mode and fuzzy logic control to
conversion systems. IET Renew Power Gener 2015;9(6):682–9.
optimal energy management in wind turbine with battery storage. Neural Comput
[14] Castelló J, Espí JM, García-Gil R. Development details and performance assessment
& Applic 2017;28(6):1417–34.
of a wind turbine emulator. Renew Energy 2016;86:848–57.
[37] Belkaid A, Colak I, Kayisli K. Implementation of a modified P&O-MPPT algorithm
[15] Jena D, Rajendran S. A review of estimation of effective wind speed based control
adapted for varying solar radiation conditions. Electr Eng 2017;99:839–46.
of wind turbines. Renew Sustain Energy Rev 2015;43:1046–62.
[38] Meghni B, Ouada M, Saad S. A novel improved variable-step-size P&O MPPT
[16] Meghni B, Saadoun A, Dib D, Amirat Y. Effective MPPT technique and robust
method and effective supervisory controller to extend optimal energy management
power control of the PMSG wind turbine. IEEJ Trans Electr Electron Eng 2015;10
in hybrid wind turbine. Electr Eng 2020;102(2):763–78.
(6):619–27.
[39] Rahimi M. Coordinated control of rotor and grid sides converters in DFIG based
[17] Ghaffari A, Krstić M, Seshagiri S. Power optimization and control in wind energy
wind turbines for providing optimal reactive power support and voltage regulation.
conversion systems using extremum seeking. IEEE Trans Control Syst Technol
Sustainable Energy Technol Assess 2017;20:47–57.
2014;22(5):1684–95.
[40] Rajendran S, Muthu Kumar A, Agnes Idhaya Selvi V. Particle Swarm Optimization
[18] Taveiros FEV, Barros LS, Costa FB. Back-to-back converter state-feedback control of
tuned Hybrid Sliding Mode Controller based static synchronous compensator with
DFIG (doubly-fed induction generator)-based wind turbines. Energy 2015;89:
LCL filter for Power Quality improvement. Sustainable Energy Technol Assess
896–906.
2022;53:102653.
[19] Gouabi H, Hazzab A, Habbab M, Rezkallah M, Chandra A. Experimental
[41] Ali N, Liu Z, Armghan H, Armghan A. Super-twisting sliding mode controller for
implementation of a novel scheduling algorithm for adaptive and modified P&O
maximum power transfer efficiency tracking in hybrid energy storage based
MPPT controller using fuzzy logic for WECS. Int J Adapt Control Signal Process
wireless in-wheel motor. Sustainable Energy Technol Assess 2022;52:102075.
2021;35(9):1732–53.
[42] Fazli E, Rakhtala SM, Mirrashid N, Karimi HR. Real-time implementation of a super
[20] Sitharthan R, Karthikeyan M, Sundar DS, Rajasekaran S. Adaptive hybrid
twisting control algorithm for an upper limb wearable robot. Mechatronics 2022;
intelligent MPPT controller to approximate effectual wind speed and optimal rotor
84:102808.
speed of variable speed wind turbine. ISA Trans 2020;96:479–89.
[21] Zeddini MA, Pusca R, Sakly A, Mimouni MF. PSO-based MPPT control of wind-
driven Self-Excited Induction Generator for pumping system. Renew Energy 2016;
95:162–77.

16

You might also like