Article

Comparative Evaluation of Traditional and Advanced Algorithms

for Photovoltaic Systems in Partial Shading Conditions
Robert Sørensen and Lucian Mihet-Popa *

Faculty of Information Technology, Engineering and Economics, Østfold University College,

1671 Fredrikstad, Norway; [email protected]
* Correspondence: [email protected]

Abstract: The optimization of photovoltaic (PV) systems is vital for enhancing efficiency and economic
viability, especially under Partial Shading Conditions (PSCs). This study focuses on the development
and comparison of traditional and advanced algorithms, including Perturb and Observe (P&O),
Incremental Conductance (IC), Fuzzy Logic Control (FLC), Grey Wolf Optimization (GWO), Particle
Swarm Optimization (PSO), and Artificial Neural Networks (ANN), for efficient Maximum Power
Point Tracking (MPPT). Simulations conducted in the MATLAB/Simulink software package evaluated
these algorithms’ performances under various shading scenarios. The results indicate that, while
traditional methods like P&O and IC are effective under uniform conditions, advanced techniques,
particularly ANN-based MPPT, exhibit superior efficiency and faster convergence under PSCs. This
study concludes that integrating Artificial Intelligence (AI) and Machine Learning (ML) into MPPT
algorithms significantly enhances the reliability and efficiency of PV systems, paving the way for a
broader adoption of solar energy technologies in diverse environmental conditions. These findings
contribute to advancing renewable energy technology and supporting green energy transition.

Keywords: photovoltaic systems; partial shading conditions; maximum power point tracking;
artificial neural networks; fuzzy logic control; grey wolf optimization; particle swarm optimization

Citation: Sørensen, R.; Mihet-Popa, L.

Comparative Evaluation of 1. Introduction
Traditional and Advanced Algorithms
The transition toward Renewable Energy Sources (RESs) is crucial for addressing
for Photovoltaic Systems in Partial
climate change adoption and achieving zero carbon emissions. Among renewable technolo-
Shading Conditions. Solar 2024, 4,
gies, photovoltaic (PV) systems have emerged as a leading solution due to their ability to
572–594. https://doi.org/10.3390/
convert solar energy into electricity [1]. PV power generation is associated with emissions
solar4040027
solely during the production of its components. Once installed, PV panels harness solar
Academic Editor: Jürgen Heinz irradiation to generate electricity without emitting greenhouse gases. Over their estimated
Werner lifespan of 25 years, PV panels are designed to yield more energy than consumed during
Received: 6 June 2024
their production [2]. In 2022, the PV industry celebrated a significant milestone, surpassing
Revised: 26 September 2024 a total global capacity of 1 terawatt (TW), reaching 1183 gigawatts (GWs) of PV installations.
Accepted: 27 September 2024 Over half of this capacity was added in the last four years alone. The year 2022 also saw a
Published: 8 October 2024 record-setting annual growth, with 236 GW of new PV capacity installed worldwide, a 35%
increase from the previous years [3].
Three primary factors influence the performance of a PV plant: the efficiency of the PV
panels, the conversion efficiency of the power electronic converters, and the effectiveness of
Copyright: © 2024 by the authors.
the Maximum Power Point Tracking (MPPT) algorithm. Enhancing the efficiency of the PV
Licensee MDPI, Basel, Switzerland.
panels and the converters are a complex task that relies on the current state of technology
This article is an open access article
and may necessitate the use of superior components and materials, potentially elevating
distributed under the terms and
the system cost significantly. However, optimizing the MPPT by implementing new control
conditions of the Creative Commons
algorithms is a more feasible approach. This improvement is cost-effective and can be
Attribution (CC BY) license (https://
readily applied to existing plants through a control algorithm update, promptly boosting
creativecommons.org/licenses/by/
4.0/).
the efficiency of PV power generation and reducing costs. However, the efficiency of

Solar 2024, 4, 572–594. https://doi.org/10.3390/solar4040027 https://www.mdpi.com/journal/solar

Solar 2024, 4 573

MPPT algorithms is significantly affected by Partial Shading Conditions (PSCs), which can
occur due to clouds, dust, snow, nearby buildings, trees, the aging effect of the panels [4],
and other obstructions. This creates non-uniform irradiance on the PV panels, leading
to multiple Local Maximum Power Points (LMPPs) on the Power–Voltage (P-V) curve,
which poses a challenge for MPPT techniques in finding the Global Maximum Power
Point (GMPP). Traditional techniques like Perturb and Observe (P&O) and Incremental
Conductance (IC) have been foundational in the development of MPPT methodologies.
However, their performance under PSC is often limited by the inability to consistently
identify the GMPP, leading to suboptimal power extraction from PV arrays [5,6]. Recent
advancements in MPPT technologies have introduced several algorithms to address the
complexities introduced by PSCs, including Fuzzy Logic Control (FLC) [7,8], Grey Wolf Op-
timization (GWO) [9,10], Particle Swarm Optimization (PSO) [11–13], and Artificial Neural
Networks (ANNs) [14–17]. While the discussed MPPT algorithms are well documented
in the literature, there is a noticeable lack of diversity in ANN approaches, with many
studies following similar methodologies or hybrids. This research investigates a novel
ANN approach, aiming to find simpler and more effective ways to implement ANN MPPT
algorithms that perform well under PSCs and can be utilized in real-world applications,
like building-integrated PV or agrivoltaic PV systems.
The motivation behind this research stems from the increasing prevalence of PV
systems in the global energy mix, as shown in Figure 1, along with its growing trend
highlighted in Figure 2. Additionally, there is a need for more resilient and efficient MPPT
strategies to cope with the variable environmental conditions exacerbated by climate change.
The exploration of ANNs in this context is not merely an academic pursuit but a step toward
advancing a technology that stands at the forefront of innovation. By understanding and
contrasting the capabilities and limitations of these MPPT algorithms, this research aims
to contribute to the evolution of PV system efficiency and reliability, harnessing the allure
of Artificial Intelligence (AI) and Machine Learning (ML) to illuminate a path toward a
sustainable future.

Figure 1. Global energy production in 2022 [3].

By exploring the strengths and weaknesses of each MPPT technique, this study seeks
to contribute to the ongoing efforts to enhance the reliability and performance of PV sys-
tems, paving the way for a broader adoption of solar energy as a cornerstone of sustainable
development. As the field of PVs continues to evolve with the introduction of new materials
(perovskite, thin-film [3], GaAs [18]) and innovative applications (building-integrated PV,
agrivoltaics, floating PV [3]), the significance of effective MPPT techniques becomes even
more pronounced. These developments not only offer the potential for higher efficiency so-
lar cells, but also expand the versatility of PV systems in various environmental conditions
and architectural designs. Therefore, understanding the implications of partial shading
and optimizing MPPT algorithms are crucial steps toward maximizing the potentials of
solar energy in the pursuit of a cleaner and more sustainable future.
Solar 2024, 4 574

Figure 2. The yearly PV installation, module PV production, and module production capacity during
2012–2022 (GW) [3].

This research advances the body of knowledge in renewable energy technology with
a focus on optimizing PV systems under PSCs. This study provides a comprehensive
comparative analysis of six different MPPT techniques using the MATLAB/Simulink envi-
ronment. Through detailed simulations of various partial shading scenarios, it elucidates
the performance, efficiency, and reliability of both traditional and advanced MPPT methods,
including P&O, IC, FLC, GWO, PSO, and ANN. This research also seeks to investigate how
an ANN-based MPPT algorithm, utilizing only the irradiance as input, can be developed
to enhance the performance and reliability of PV systems under PSCs, while also reducing
system complexity compared to traditional MPPT algorithms.
The structure of this paper is as follows: Section 2 discusses the PV system modeling,
detailing the equivalent circuit models and the impact of partial shading. Section 3 describes
the simulation models and scenarios developed for this study. Section 4 provides an
overview of the MPPT algorithms evaluated, including their operational principles and
implementation. Section 5 presents the results and discussion, focusing on the performance
analysis of each MPPT algorithm under various conditions. Section 6 concludes the study
with a summary of findings and suggestions for future research.

2. PV System Modeling
PV system modeling is crucial for understanding and predicting solar cell behavior
under various conditions. Key methods include single-diode models, double-diode models,
and empirical models. The single-diode model balances simplicity and accuracy, making
it suitable for practical applications, while the double-diode model offers better represen-
tation at the cost of complexity. Empirical models fit experimental data but lack physical
insight. This paper used the equivalent circuit-based model (single-diode) for its balance of
simplicity and accuracy, particularly under PSCs. This model can be effectively integrated
with system components like MPPT controllers, providing comprehensive system-level
analysis. It helps understand losses and real-world conditions on solar cell output, offering
insights into PV system behavior under partial shading, which is crucial for optimizing
MPPT techniques. Figure 3 provides an overview of a stand-alone PV system and its key
components.
Solar 2024, 4 575

Figure 3. Block diagram of the PV system components.

2.1. Equivalent Circuit of a Solar Cell

Ideally, a PV cell lacks parallel and series resistances. However, in practical appli-
cations, these resistances are incorporated due to leakage current and ohmic resistances,
as is shown in Figure 4. The primary cause of the shunt resistance Rsh is the suboptimal
performance of the p-n junction in the PV diode, while the series resistance Rs originates
from the bulk resistance of the semiconductor material and its interconnections [19,20].

Figure 4. Equivalent circuit of a PV cell [20].

When a PV cell is exposed to solar irradiance, the output current can be calculated
using Kirchhoff’s law, which is detailed in Equation (1):

IPV = IL − Id − Ish , (1)

where IL represents the current generated by light, as defined in Equation (2):

IL = GISC [1 + k a ( T − TSTC )], (2)

where G signifies the solar irradiation, T is the ambient temperature under specific climate
conditions, ISC represents the short-circuit current of the PV cell, k a is the temperature
coefficient, TSTC denotes the temperature of the PV cell under standard test conditions
(STCs), and Id is the diode current of the PV, as defined by Equation (3):
   
qVd
Id = I0 exp −1 , (3)
nkT

where I0 denotes the saturation current of the PV diode, Vd is the terminal voltage across the
PV diode, q represents the electrical charge (1.69 × 10−19 C), and k stands for the Boltzmann
constant (1.3807 × 10−23 J/K), while n is the ideality factor of the PV diode. The universal
formula outlining the Current–Voltage (I-V) characteristics of the PV cell is presented in
Equation (4):
     
q(VPV + IPV Rs ) V + IPV Rs
IPV = IL − I0 exp − 1 − PV , (4)
nkT Rsh
where IPV represents the current generated by the PV cell and VPV is the voltage across
the PV cell. To achieve the desired voltage and current, the PV cells are arranged in a
combination of series, parallel, or series and parallel configurations to create an optimized
Solar 2024, 4 576

PV module. These modules are then connected in different configurations to form the
required PV array.
The main performance attributes of a PV panel are typically delineated by the Power–
Voltage (P-V) curve, as illustrated in Figure 5, which is generated under STCs. At the STCs,
the solar irradiance is set to 1000 W/m2 and the ambient temperature is maintained at
25 ◦ C. The power output of the PV is determined by multiplying the voltage (V) by the
current (I), the latter of which is a function of the voltage. For optimal performance, the
operating current of a PV cell is maintained near the maximum power point (MPP), where
the product of current and voltage is maximized (Pmax ). This current is typically less than
the short-circuit current (Isc ) and corresponds to the voltage at the maximum power point
(Vmp ), as is illustrated in Figure 5. The MPP represents the condition where the PV cell
generates its maximum power output, optimally balancing both voltage and current.

Figure 5. Performance characteristics of a PV cell or module, as represented by the Current–Voltage

(I-V) curve and the Power–Voltage (P-V) curve.

2.2. Maximum Power Point

The characteristics of PV panels are illustrated through I-V and P-V curves, indicating
an optimal operating point where the power output peaks, which is known as the Maximum
Power Point (MPP), as highlighted in Figure 6. In these curves, the MPP is crucial because
of the panel’s relatively low efficiency, necessitating operation near this point for maximal
energy collection [4]. The PMPP can be represented as detailed in Equation (5):

PMPP = VMP IMP . (5)

Figure 6. The I-V curve, P-V curve, and MPP [4].

Solar 2024, 4 577

2.3. Partial Shading

Partial shading occurs when an object obstructs sunlight from reaching a portion
of a PV cell, module, or array. Unlike full shading, partial shading does not cover the
entire surface but affects only part of the PV system, as illustrated in Figure 7. This
phenomenon is significant in the context of solar power because even small shadows can
lead to considerable power losses and efficiency degradation in solar PV systems.

Figure 7. Illustration of three-series PVs with bias diodes under PSC.

Under uniform sunlight distribution, the peak output power of a PV system equals
the aggregate of the peak power outputs from each cell, module, or array. However,
the shading from clouds, trees, or buildings can severely affect PV system performances.
Shaded modules experience a drop in voltage, leading them to act as loads rather than
power generators, which can induce the risk of overheating and is known as the “hot spot”
issue. To mitigate this, PV modules incorporate bypass diodes, which, while preventing
hot spots, introduce multiple peaks in the P-V curve, including one GMPP and several
LMPPs. This can be observed in Figure 8, where the effect of partial shading is present, as
well as the effect of bias diodes.

Figure 8. P-V curves for the PV under PSC with LMPP and GMPP showing the effect of a bias diode.
Solar 2024, 4 578

This complexity poses a challenge for traditional MPPT technologies, which may
erroneously lock onto a LMPP, consequently diminishing the overall power yield of the PV
array. Hence, it is critical to employ a sophisticated MPPT strategy capable of accurately
identifying and following the GMPP, especially under PSCs, to optimize the power output
of the solar power system.
If we were to think of three PVs connected in series, we could represent them with use
of the single-diode model, as in Figure 9.

Figure 9. Equivalent circuit of a three-series-connected PV with bias diodes.

Here, we could calculate the total current generated by the three-series-connected PV

modules, as derived in Equation (6).
 h   V +I R  i V + I R
 Iph1 ( G1 ) − ∑iu=1 I0i1 exp pv1ni1 V pv1 s1
− 1 − pv1 pv1 s1
 V +I t R  i V +RIp1 R


 h 
I = Iph2 ( G2 ) − ∑iu=1 I0i2 exp pv2ni2 V pv2 s2
− 1 − pv2 Rp2pv2 s2 (6)
 h   V +I t R  i V + I R
 Iph3 ( G3 ) − ∑u I0i3 exp
 pv3 pv3 s3 pv3 pv3 s3
−1 −

i =1 ni3 Vt Rp3

Partial shading significantly diminishes the efficiency of PV systems. In order to

maximize power output under such varying environmental conditions, implementing
strategies specifically designed to mitigate the effects of partial shading is crucial, regardless
of whether the PV system operates in a grid-connected or standalone mode. Traditional
MPPT methods, which typically employ fixed step sizes, often fall short as they tend to
lock onto local maxima instead of the optimal MPP. This limitation hinders their ability to
adaptively track the MPP amidst the fluctuating irradiance levels that are characteristic of
partial shading scenarios. The most popular algorithms for MPPT control in PV systems
are P&O, IC, and Constant Voltage (CV). They have the advantage of being very easy
to implement and work very well under uniform radiation and temperature conditions.
Nevertheless, as all techniques have disadvantages, they are impossible to adapt to constant
changes in the environment and fail to find the MPP. From these, several improvements are
Solar 2024, 4 579

derived to optimize the MPP calculation, such as those using the variable step principle
to detect the area under the MPP curve using the derived dP/dV values and a scaling
factor. However, failures have also been detected in finding the MPP at different radiation
levels [17].

3. Simulation Models and Scenario Development

This section describes the simulation model developed to evaluate the performance
of the MPPT algorithms. It includes the description of the PV model, boost converter,
load configurations, PSC model, and the specifics of each MPPT algorithm simulated. The
primary objectives are to present the methodology for simulating the MPPT algorithms
and to demonstrate the setup for comparative analysis under controlled PSCs.

3.1. Simulation Environment

The simulation model was developed within the MATLAB/Simulink (2023b) software
environment, and a schematic representation of it is provided in Figure 10. The stand-alone
model comprises PV arrays subjected to PSC, a boost converter with a Pulse-Width Modu-
lation (PWM) control strategy governed by the different MPPT algorithms for evaluative
comparison, and an assigned load profile.
The model simulates three-series-connected modules, each receiving unique irradiance
and temperature inputs to replicate PSCs. A boost converter regulates the voltage output
from the PV array. It includes an inductor, diode, capacitor, and MOSFET switch, stepping
up the voltage to match the load requirements. The duty cycle is controlled by the MPPT
algorithm. Implemented as a MATLAB function block, the controller executes algorithms
such as P&O, IC, FLC, GWO, PSO, or ANN, adjusting the duty cycle to track the MPP.
Modeled as a variable resistive element, the load configuration allows for testing under
different scenarios, such as battery charging or grid connection.
The simulation dynamically adjusts the boost converter’s duty cycle to maintain the
MPP under various shading conditions. MPPT algorithms utilize the feedback signal
from the PV array’s voltage and current to determine the optimal operating point. The
simulation evaluates performance metrics, including the maximum power output, average
power, settling time, and MPPT efficiency.

Figure 10. Block diagram of the simulation model.

3.2. PV Model
For this study, the SunPower SPR-X20-250-BLK PV panel was chosen, primarily due
to its high efficiency and suitability. This panel is characterized by its robust response to
low-light conditions and its high-power output consistency, making it an ideal candidate
for detailed performance analysis in MATLAB/Simulink.
The simulation model of the Sun Power SPR-X20-250-BLK used in this study was
derived from the Simulink/Simscape library, where a comprehensive set of tools within
the Simulink package simulation tool that facilitates the modeling of physical systems,
including electrical systems such as PV arrays, was used. Simscape provides a physically
Solar 2024, 4 580

modeled library, which allows for the simulation of real-world physical phenomena with
high accuracy and detailed behavioral analysis.
The detailed specifications of the PV modules are presented in Tables 1 and 2.

Table 1. Module data for Sunpower SPR-X20-250-BLK.

Parameters Symbols Values

Maximum Power (W) PMPP 249.952
Cells Per Module Ncell 72
Open-Circuit Voltage (V) VOC 50.93
Short-Circuit Current (A) ISC 6.2
Voltage at Maximum Power Point (V) VMP 42.8
Current at Maximum Power Point (A) IMP 5.84
Temperature Coefficient of VOC (%/◦ C) KOC −0.291
Temperature Coefficient of ISC (%/◦ C) KSC 0.013306

Table 2. Model parameters for Sunpower SPR-X20-250-BLK.

Parameters Symbols Values

Light-Generated Current (A) IL 6.2119
Diode Saturation Current (A) I0 1.3593 × 10−11
Diode Ideality Factor n 1.0262
Shunt Resistance (Ω) Rsh 420.5449
Series Resistance (Ω) Rs 0.37748

As can be seen in Figure 11, the PV model consists of three PV modules connected
in series with bias diodes. Each module experienced individual solar irradiance and
temperature inputs. The three PV modules were simulated using a PV array block from
the Simscape/Electrical/Specialized Power Systems/Sources library in MATLAB Simulink
(2023b). The blocks support both predefined PV modules from the National Renewable
Energy Laboratory (NREL) System Advisor Model (2018) and user-defined PV modules.

Figure 11. Three-series-connected PV modules with bias diodes connected to a converter.

3.3. Boost Converter and Load

Table 3 presents the specific parameters of the boost converter and the load utilized
within the simulation framework.
Solar 2024, 4 581

Table 3. Model parameters for the boost converter and load.

Parameters Symbols Values

Input Filter Capacitor (mF) Cin 1
Output Filter Capacitor (µF) Cout 32.27
Boost Inductor (mH) L 1.15
Switching Frequency (kHz) f sw 5
Resistive Load (Ω) R 100

3.4. Partial Shading and Scenario Development

To simulate the PSCs in this study, the series-connected PVs were segmented into three
groups. Each group was subjected to distinct irradiance levels to facilitate the analysis of
different MPPT techniques under PSC. The irradiance levels applied were also segmented
into five shading scenarios, as denoted in Table 4, where all PVs were maintained at a
constant temperature of 25 °C, reflecting STCs.

Table 4. Solar irradiance scenarios.

Shading Irradiance 1 Irradiance 2 Irradiance 3

Scenario (W/m2 ) (W/m2 ) (W/m2 )
1 1000 1000 1000
2 100 500 1000
3 1000 700 200
4 400 1000 600
5 1000 600 300

Figure 12 displays the I-V and P-V characteristics under uniform irradiance (Shading
Scenario 1). The curves demonstrate that the MPP closely corresponds to the theoretical
values noted in the datasheet of the SunPower SPR-X20-250-BLK panels, as documented
in Table 1. This alignment validates the accuracy of the simulation model in replicating
the expected electrical behavior under standard operating conditions. The values obtained
in Figure 12 were used as a comparison for the MPPT algorithms to determine how they
performed under Shading Scenario 1.

Figure 12. The I-V and P-V characteristics for the PV system illustrating the MPP under full solar
irradiance.
Solar 2024, 4 582

In contrast, Figure 13 displays the I-V and P-V characteristics for Shading Scenarios 2–5,
where each PV is subjected to different irradiance levels. These scenarios were designed
to mimic the PSC effects. We can also observe that, in these scenarios, we experienced
multiple MPPs resulting in two LMPPs and one GMPP. The GMPP was highlighted and
used for comparisons of the performance of the MPPT algorithms where (a) shows the
GMPP for Shading Scenario 2, (b) the GMPP for Shading Scenario 3, (c) the GMPP for
Shading Scenario 4, and (d) shows the GMPP for Shading Scenario 5.

(a) (b)

(c) (d)

Figure 13. The I-V and P-V characteristics illustrating the GMPP for the PV system under Shading
Scenario 2 (a), 3 (b), 4 (c), and 5 (d).

4. MPPT Algorithm Development: Modeling Techniques and Performance

This section provides an overview of the MPPT algorithms evaluated in this study,
including the traditional P&O and IC, as well as the advanced algorithms such as FLC,
GWO, PSO, and ANN. Figure 14 shows a block diagram of the inputs and outputs of the
algorithms, which were implemented using MATLAB function blocks to operationalize
their respective MPPT methods.
Solar 2024, 4 583

Figure 14. Block diagram of the MPPT algorithms using MATLAB function blocks.

4.1. Perturb & Observe (P&O)

The P&O MPPT algorithm operates by periodically perturbing (adjusting) the operat-
ing voltage of the PVs and observing the resulting change in power output to track the MPP.
The main principle behind the P&O algorithm is that, if an increase in the PV voltage leads
to an increase in power, the operating point is moving toward the MPP. Conversely, if an
increase in voltage leads to a decrease in power, the operating point is moving away from
the MPP. The algorithm starts by measuring the current and voltage of the PVs to calculate
the initial power. It then perturbs the voltage by a small increment and measures the new
power output. By comparing the new power output with the previous power output, the
algorithm determines the direction of the perturbation. If the new power is greater than
the previous power, the perturbation continues in the same direction. If the new power is
less than the previous power, the direction of the perturbation is reversed. This process is
repeated continuously. The algorithm keeps adjusting the voltage in small steps and observ-
ing the changes in power output, thereby dynamically tracking the MPP. The perturbation
can be either an increase or a decrease in voltage, and the size of the perturbation step can
affect the algorithms responsiveness and stability. One of the main advantages of the P&O
algorithm is its simplicity and ease of implementation. However, it can oscillate around the
MPP under steady-state conditions and may struggle with rapid changes in environmental
conditions. This is because the algorithm might continue perturbing even after reaching
the MPP, causing the operating point to fluctuate around the optimal value. Despite these
limitations, it is widely used due to its straightforward implementation and effectiveness in
many scenarios. The algorithm was implemented using a “MATLAB Function” block with
the inputs and outputs that are shown in Figure 14. This block operationalized the P&O
method through embedded MATLAB code, following the structure of the P&O algorithm
shown in [21].

4.2. Incremental Conductance (IC)

The core principle of the IC algorithm is based on the fact that the derivative of the
power with respect to the voltage (dP/dV) is zero at the MPP, positive on the left of the MPP,
and negative on the right. The algorithm uses this principle to track the MPP by measuring
the instantaneous conductance (I/V) and the incremental conductance (∆I/∆V) of the PVs.
The IC algorithm begins by measuring the PV’s current (IPV ) and voltage (VPV ). It then
calculates the incremental changes in the current (∆I) and voltage (∆V) from their previous
values. Using these measurements, the algorithm computes the instantaneous conductance
(I/V) and the incremental conductance (∆I/∆V). At each iteration, the algorithm evaluates
the condition of the PVs to determine whether it is operating to the left of or to the right of
the MPP. If ∆I/∆V equals I/V, the operating point is at the MPP and no adjustment to the
duty cycle is needed. If ∆I/∆V is greater than I/V, then the operating point is to the left of
the MPP, indicating that the voltage should be increased to reach the MPP. Conversely, if
∆I/∆V is less than I/V, the operating point is to the right of the MPP, indicating that the
voltage should be decreased. Based on these evaluations, the algorithm adjusts the duty
cycle of the boost converter accordingly. If the operating point is to the left of the MPP, the
algorithm increases the duty cycle, which decreases the voltage and increases the current. If
the operating point is to the right of the MPP, the algorithm decreases the duty cycle, which
increases the voltage and decreases the current. The IC algorithm continues this process
iteratively, constantly measuring the current and voltage, recalculating the conductance
values, and adjusting the duty cycle to maintain operation at or near the MPP. This method
Solar 2024, 4 584

allows the PV system to dynamically adapt to changes in environmental conditions, such

as variations in irradiance and temperature, ensuring optimal power extraction from the
PV array. The algorithm was implemented using a “MATLAB Function” block with the
inputs and outputs that are depicted in Figure 14. This block operationalized the IC method
through embedded MATLAB code, adhering to the structure of the IC algorithm shown
in [22].

4.3. Fuzzy Logic Controller (FLC)

Figure 15 shows a block diagram of the FLC MPPT used in the simulation model.
The core functionality of the model revolves around the FLC, which utilizes a Type-1
Sugeno Fuzzy Inference System (FIS) with Gaussian membership functions and is adept at
managing the non-linearity and uncertainty inherent in PV systems subject to PSC. The
FLC operates by encoding expert knowledge into a set of “if-then” rules, where each input
variable is associated with Gaussian functions characterized by their means and spreads.
These functions are chosen for their smooth and continuous properties, which allow for
effective handling of the gradual transitions in variable states.

Figure 15. Block diagram of the FLC algorithm.

In the simulation model, the solar irradiance values were processed through a calcu-
lation system. Here, the input irradiance values were normalized and scaled to generate
a “FuzzyIn” signal. This signal feeds into the FLC, where it undergoes several processing
stages: fuzzification, where the degree of truth for each input variable is determined against
the Gaussian functions; rule-based inference, which applies the “prod” and “probor” meth-
ods for conjunctions and disjunctions; and defuzzification, which integrates the outputs of
all active rules through a weighted average method to produce a duty cycle value for the
boost converter.
The fuzzy rules, as specified in Table 5, dictate the system’s response to different
scenarios, ensuring accurate tracking of the MPPT. Inputs are first fuzzified, then processed
through the rule-based inference engine, with the outputs aggregated and defuzzified to
yield a precise duty cycle. This duty cycle subsequently drives the PWM signal controlling
the MOSFET in the boost converter, effectively modulating the energy conversion to
maintain optimal efficiency and align the PV output with the MPP.

Table 5. Rules table.

Rule Number Rule Weight Name

If in1 is in1cluster1 and in2 is in2cluster1 and in3 is
1 1 rule1
in3cluster1 then out1 is out1cluster1
If in1 is in1cluster2 and in2 is in2cluster2 and in3 is
2 1 rule2
in3cluster2 then out1 is out1cluster2
If in1 is in1cluster3 and in2 is in2cluster3 and in3 is
3 1 rule3
in3cluster3 then out1 is out1cluster3
If in1 is in1cluster4 and in2 is in2cluster4 and in3 is
4 1 rule4
in3cluster4 then out1 is out1cluster4
If in1 is in1cluster5 and in2 is in2cluster5 and in3 is
5 1 rule5
in3cluster5 then out1 is out1cluster5
Solar 2024, 4 585

4.4. Grey Wolf Optimization (GWO)

The GWO MPPT algorithm works by mimicking the social hierarchy and hunting
behavior of gray wolves in nature to find the optimal duty cycle for a boost converter in
the PV system. The goal of the algorithm is to maximize the power output from the PVs by
continuously adjusting the duty cycle to operate at the MPP. The GWO algorithm starts
by initializing a population of solutions, called grey wolves, which represent different
possible duty cycle values. These wolves are divided into three categories: alpha, beta, and
delta. These correspond to the best, second-best, and third-best solutions, respectively. The
remaining wolves are considered omega and follow the top three wolves in the hierarchy.
In each iteration of the algorithm, the positions of the wolves are updated based on
the positions of the alpha, beta, and delta wolves. This process involves calculating the
distances between each wolf and the alpha, beta, and delta wolves are then updating
their positions using specific equations that simulate the encircling, hunting, and attacking
strategies observed in gray wolves. The fitness of each wolf is evaluated based on the power
output of the PV, which is calculated as the product of the PV voltage and current. The
algorithm compares the power output for each duty cycle value and updates the positions
of the alpha, beta, and delta wolves if better solutions are found. This process of position
updating and fitness evaluation continues iteratively until a stopping criterion is met,
such as reaching a maximum number of iterations or achieving a convergence threshold.
The positions of the alpha wolf at the end of the iterations represents the optimal duty
cycle for the boost converter. The optimal duty cycle is then sent to the PWM generator,
which adjusts the duty cycle of the boost converter accordingly. By doing so, the GWO
MPPT algorithm ensures that the PV operates at or near its MPP, continuously adapting
to changes in environmental conditions such as irradiance and temperature. The GWO
algorithm was implemented using a “MATLAB Function” block, as in Figure 14. This
block operationalized the method through embedded MATLAB code, following the GWO
structure shown in [10].

4.5. Particle Swarm Optimization (PSO)

The PSO MPPT algorithm is a technique that mimics the social behavior of birds
flocking or fish schooling to find the optimal solution in a search space. In the context
of this PV system, the PSO algorithm aims to find the optimal duty cycle for the boost
converter to ensure that the PV operates at its MPP. The algorithm begins by initializing a
swarm of particles, where each particle represents a potential solution, which, in this case,
is a possible duty cycle for the boost converter. These particles are randomly distributed
across the search space. Each particle has a position, which corresponds to a duty cycle
value, and a velocity, which determines the direction and magnitude of the particle’s
movement through the search space. At each iteration, the particles update their positions
and velocities based on two primary factors: their personal best position and the global best
position. The personal best position is the highest power output a particle has achieved
thus far, while the global best position is the highest power output achieved by any particle
in the swarm. The particles adjust their velocities by considering their current velocity,
the distance to their personal best position, and the distance to the global best position.
The influence of these distances are moderated by random coefficients, which introduce a
stochastic element to the algorithm, helping the particles explore the search space more
thoroughly. Once the velocities are updated, the particles move to new positions in the
search space. The power output at these new positions is evaluated by calculating the
product of the PV voltage and current. If a particle’s new position yields a higher power
output than its previous best position, the personal best is updated. Similarly, if any particle
achieves a power output higher than the current global best, the global best is updated. This
process of updating velocities, moving particles, and evaluating power outputs continues
iteratively. Over time, the swarm of particles converges toward the optimal duty cycle
that maximizes the power output of the PVs. The particle at the global best position
represents the optimal duty cycle, which is then used to control the boost converter. The
Solar 2024, 4 586

PSO algorithm is particularly effective because it combines exploration and exploitation.

The particles explore the search space to find the global optimum, and they exploit the best
solutions found by guiding the swarm toward these solutions. This balance helps the PSO
algorithm avoid local maxima and ensures that it finds the true MPP. The PSO algorithm
was implemented using a “MATLAB Function” block with inputs and outputs that are
shown in Figure 14. This block utilized embedded MATLAB code to operationalize the
PSO method shown in [21].

4.6. Artificial Neural Networks (ANNs)

Two ANN MPPTs were developed, each with a different approach. After the successful
training of each ANN, the optimized model was exported into the Simulink environment,
as depicted in Figure 16. This was accomplished by creating a dedicated Simulink block that
encapsulates the functionality of the ANN. Named the “Function Fitting Neural Network”,
this block clearly delineates the inputs and outputs, providing a streamlined interface for
simulation tasks.

Figure 16. Generated ANN Simulink block.

Delving deeper into the structure of this ANN block, Figure 17 outlines the sequential
arrangement of its processing elements. The “Process Input 1” block first conditions the
input data, preparing it for processing through the neural layers. The data then passes
through “Layer 1”, which consists of neurons equipped with sigmoid activation functions
to introduce necessary non-linearity for effective pattern recognition. Following this, “Layer
2” acts as the output layer, applying a linear transformation to produce the final predictions.

Figure 17. Layers inside the ANN Simulink block.

ANN (1) and ANN (2) both employ a two-layer feedforward network with sigmoidal
activation functions in the hidden layers and a linear function in the output neuron, which
is suitable for regression tasks. The ANN development, training, validation, and testing
processes were conducted using MATLAB’s Deep Learning Toolbox, with the dataset split
into 70% for training, 15% for validation, and 15% for testing, ensuring comprehensive
evaluation of the models’ performance. Both networks were trained using the Levenberg–
Marquardt backpropagation algorithm, which is known for its effectiveness in minimizing
the Mean Squared Error (MSE) by precisely adjusting the weights (w) and biases (b)
across the network, thus enhancing the models’ ability to accurately predict continuous
voltage values.
ANN (1) features a twenty-neuron hidden layer, which allows the network to capture
the nonlinear relationships between the inputs and the output more effectively. ANN (2) on
the other hand, processes three solar irradiance inputs through a ten-neuron hidden layer,
Solar 2024, 4 587

each computing a weighted sum followed by a sigmoid activation function. The output
layer consists of a single neuron for continuous voltage prediction.
The ANN (1) model was designed to predict the maximum voltage output of the PV
system. Using historical real-world data, the network is trained to estimate this maximum
voltage based on the irradiance and temperature of each PV module. Consequently, the
neural network has six inputs, i.e., the irradiance and temperature for each PV, as shown in
the block diagram depicted in Figure 18.

Figure 18. Block diagram of the ANN (1) MPPT.

The ANN (2) MPPT model, as illustrated in Figure 19, was designed to predict the
maximum voltage output from the PV system based on variable solar irradiance inputs,
and it was specifically trained for PSCs. In this ANN, the inputs were the irradiance values
from each of the PVs, and the output was the maximum voltage level of the PV system.
This approach is not commonly found in the literature and will be investigated in this study
as a possibly simpler ANN algorithm than using real historical data for training.

Figure 19. Block diagram of the ANN (2) MPPT.

Figure 20 shows the real-world solar irradiance (a) and the temperature (b) data, ob-
tained from NASA Power [23], which were used for training the ANN (1) MPPT. These data
are presumably structured in a way that each entry corresponds to the hourly observations
from Fredrikstad, Norway, covering the period from March to August 2023.

(a) (b)
Figure 20. Solar irradiance (a) and temperature (b) data used for training ANN (1).
Solar 2024, 4 588

To create the dataset for training the ANN (2) such that it can handle an MPPT under
PSCs, 1000 samples were generated simulating various irradiance levels (0 to 1000 W/m2 )
on three PV panels. A snippet of the training data is shown in Table 6.

Table 6. Training data for the ANN (2) model.

Irradiance Panel 1 Irradiance Panel 2 Irradiance Panel 3 Maximum Voltage Output

(W/m2 ) (W/m2 ) (W/m2 ) (V)
723.44 979.45 258.54 83.95
264.02 213.74 271.58 32.07
393.86 958.12 721.58 88.75
818.94 494.94 589.55 81.47
622.95 208.07 82.33 71.19
942.37 751.32 933.01 112.42
543.00 984.43 510.69 87.23
895.85 972.50 117.22 108.34
559.16 328.75 600.56 63.71
983.39 344.99 985.29 99.02
967.87 652.43 687.56 106.36
463.31 709.84 329.90 99.99
908.74 791.53 403.47 90.04
878.08 344.84 883.57 74.80
386.54 967.20 412.38 87.82
699.43 710.99 366.99 76.07
258.14 877.67 89.50 89.50
869.37 632.05 215.44 73.48

5. Results and Discussion

In this section, the outcomes of the simulations are presented and analyzed, focusing
on the effectiveness of the MPPT algorithms in maximizing the power output under PSCs.
The objectives were to interpret the simulation results, compare the efficiency of each MPPT
technique, and discuss the findings.

5.1. Shading Scenario 1

Figure 21a presents the maximum power outputs of the MPPT algorithms under
uniform irradiance. P&O achieved a 749.3 W, IC achieved 749.4 W, FLC achieved 749.7 W,
and GWO achieved 749.4 W. PSO showed a lower performance at 735.3 W. ANN (1) and
ANN (2) demonstrated exceptional performance with outputs of 749.6 W. The P&O, IC,
FLC, GWO, and both ANN models exhibited superior performance, each achieving around
749 W. PSO fell behind, indicating the need for further optimization. These results highlight
the robustness and potential of ANN-based MPPT techniques in PV systems.

(a) (b)
Figure 21. Comparison of the maximum (a) and the average (b) power for Shading Scenario 1.
Solar 2024, 4 589

Further, Figure 21b compares the average power output of the MPPT algorithms. P&O
achieved 747.9 W, IC achieved 735.7 W, and FLC achieved 744.9 W. GWO stood out with
748.9 W. PSO had the lowest output at 734.7 W. ANN (1) excelled with 749.6 W, closely
followed by ANN (2) at 747.2 W. The P&O, GWO, and both ANN models performed
best, exceeding 747 W. These results underscore the robustness of ANN-based techniques,
especially ANN (1) and ANN (2), under uniform irradiance, with GWO also providing
reliable results.
Figure 22a clearly shows that the P&O algorithm had the longest settling time at
0.374 s, indicating slower response to changing conditions. The IC algorithm had a shorter
settling time of 0.087 s, balancing speed, and accuracy. FLC performed well with a settling
time of 0.055 s, demonstrating quick adaptability. GWO also showed a good efficiency
with a settling time of 0.062 s. PSO, while faster than P&O, had a settling time of 0.116 s,
still lagging behind more sophisticated algorithms. The ANN (1) and ANN (2) models
exhibited the fastest settling times at 0.055 s, along with FLC; however, they combined high
accuracy with swift adaptation. ANN (1), ANN (2), FLC, and GWO were the most efficient
in terms of settling time, with IC showing moderate performance. PSO showed longer
settling times than most but was faster than P&O. This analysis highlighted the importance
of selecting MPPT algorithms based on response times, with ANN-based models being
highly effective for applications requiring rapid stabilization.

(a) (b)
Figure 22. Comparison of the settling time (a) and MPPT efficiencies (b) for Shading Scenario 1.

Further, Figure 22b provides a comparative analysis of the MPPT efficiencies. The
P&O algorithm showed a high MPPT efficiency of 99.81%, indicating effective energy
capture despite longer settling time. The IC algorithm had a lower efficiency of 98.18%,
suggesting compromised accuracy. FLC exhibited an efficiency of 99.40%, balancing quick
response and accurate tracking. GWO achieved one of the highest efficiencies at 99.93%,
demonstrating superior capability. PSO showed a lower efficiency of 98.04%, indicating less
effectiveness in tracking the MPP. ANN (1) performed exceptionally well with an efficiency
of 99.99%, the highest among the algorithms, indicating superior adaptability. ANN (2)
also performed strongly at 99.70%, reinforcing the robustness of ANN-based approaches.

5.2. Shading Scenarios 2–5

Figure 23 presents a comparative analysis of both the maximum (a) and average (b)
power outputs for the MPPT algorithms under Shading Scenarios 2 through 5. In Shading
Scenario 2, IC, FLC, GWO, and ANN (2) achieved the highest maximum power outputs
(265.9 W to 266.0 W). ANN (1) had lower average power, and P&O and PSO showed the
lowest outputs. In Shading Scenario 3, IC, FLC, GWO, and both ANN models performed
well, achieving a maximum power around 367 W. ANN (2) and GWO maintained the
highest average power. In Shading Scenario 4, IC, FLC, and ANN (2) achieved the highest
outputs around 328 W, with ANN (1) showing variability in average power. PSO continued
Solar 2024, 4 590

to struggle with lower average power. In Shading Scenario 5, ANN (2), IC, and FLC
performed strongly with maximum outputs around 317 W, while ANN (1) showed a lower
average power. P&O and PSO exhibited the lowest outputs. The IC, FLC, GWO, and ANN
(2) demonstrated high power outputs and robust performance, while ANN (1), P&O, and
PSO showed variability and lower performance, indicating a need for optimization.

Figure 23. Comparison of the maximum power (a) and the average power (b) for Shading
Scenarios 2–5.

Figure 24 compares the settling times of the MPPT algorithms under Shading Scenarios
2 to 5, assessing the responsiveness and adaptation speed. In Shading Scenarios 2, P&O
and IC had longer settling times, with P&O at 0.170 s and IC slightly lower. FLC and
GWO showed significantly shorter settling times, while PSO was slightly higher than FLC
and GWO. ANN (1) and ANN (2) had the fastest response times, at 0.009 and 0.065 s,
respectively. In Shading Scenario 3, P&O and IC again showed higher settling times, with
IC at 0.315 s. FLC, GWO, and ANN models maintained low settling times, with PSO
showing moderate performances. In Shading Scenario 4, IC and P&O had the longest
settling times at 0.315 s, while FLC, GWO, and ANN models maintained low settling
times. In Shading Scenario 5, which was characterized by severe shading, P&O had a
higher settling time; IC performed better but still lagged behind; and FLC, GWO, and
ANN models maintained low settling times, with ANN (1) and ANN (2) demonstrating
the fastest responses. ANN (1) and ANN (2) showed superior rapid stabilization, FLC and
GWO performed consistently well, and IC and P&O exhibited longer settling times. PSO
showed moderate performance. ANN-based models were the most efficient under varying
irradiance conditions, significantly enhancing energy capture and yield.
Figure 25 compares the MPPT efficiencies of the algorithms under Shading Scenarios
2 to 5. In Shading Scenario 2, IC, FLC, GWO, and ANN (2) exceeded 99% efficiency,
while ANN (1) showed 35.75%, P&O achieved 90.31%, and PSO struggled with 41.46%.
In Shading Scenario 3, IC, FLC, GWO, and ANN (2) maintained around 99% efficiency,
ANN (1) achieved 41.46%, and P&O and PSO showed efficiencies of 65.39% and 97.28%,
respectively. In Shading Scenario 4, IC and ANN (2) were near 99.99%, with GWO and FLC
performing well, and ANN (1) improved to 96.60%. In Shading Scenario 5, IC, FLC, and
ANN (2) maintained around 99% efficiency, and ANN (1) improved to 74.94%. The IC, FLC,
GWO, and ANN (2) showed high efficiency across all scenarios, while ANN (1) showed
variability. P&O and PSO exhibited lower efficiencies, indicating a need for optimization.
Solar 2024, 4 591

Advanced algorithms like ANN (2) and GWO handled variability well, while simpler
algorithms like P&O and PSO may not perform consistently under complex conditions.

Figure 24. Comparison of the settling time for Shading Scenarios 2–5.

Figure 25. Comparison of the MPPT efficiencies for Shading Scenarios 2–5.

The ANN (2) model demonstrated not only superior efficiency, but also significantly
faster response times when compared to other MPPT algorithms under various PSCs. The
settling times for the different algorithms were recorded and analyzed, providing insights
into their adaptability and speed in stabilizing at the GMPP. In Shading Scenario 2, the
ANN (2) algorithm achieved a settling time of 0.070 s, while the PSO algorithm had a
settling time of 0.105 s and the P&O algorithm took 0.170 s. This indicates that ANN (2)
was approximately 33% faster than PSO and 59% faster than P&O. In Shading Scenario 3,
the ANN (2) algorithm maintained a settling time of 0.065 s, significantly outperforming
the IC algorithm, which had the highest settling time of 0.315 s. ANN (2) was thus about
79% faster than IC in stabilizing under these conditions. In Shading Scenario 4, ANN
(2) achieved a settling time of 0.038 s, while the IC algorithm again showed a longer
Solar 2024, 4 592

settling time of 0.315 s. This represents an impressive improvement, with ANN (2) being
approximately 88% faster than IC. In Shading Scenario 5, ANN (2) demonstrated a settling
time of 0.044 s, outperforming PSO with a settling time of 0.054 s and P&O with 0.171 s.
ANN (2) was about 19% faster than PSO and 74% faster than P&O. Based on these findings,
the performance of the MPPT algorithms could be evaluated, as shown in Table 7.

Table 7. Performance evaluation of the MPPT algorithms.

MPPT Algorithm Settling Time Response Time Efficiency

P&O Slow Moderate Low
IC Very Slow Moderate High
FLC Fast Fast High
GWO Fast Fast High
PSO Moderate Moderate Moderate
ANN (1) Fast Fast Variable
ANN (2) Fast Very Fast Very High

The ANN (2) model proved to be highly effective in not only maximizing the power
output and efficiency, but also in significantly reducing the settling times compared to
other traditional and advanced MPPT algorithms. Its rapid response and ability to quickly
stabilize under various PSCs make it an ideal choice for PV system applications, where
dynamic environmental changes are common. These findings reinforce the potential of
ANN-based approaches to enhance the performance and reliability of solar energy systems,
contributing to more efficient and adaptable renewable energy solutions and supporting
green energy transition. The results indicate that, while ANN (1) performed exceptionally
well under uniform solar conditions, it requires further training with more data or the
integration of a hybrid approach to ensure robust performance under PSCs. Although ANN
(1) was marginally the most efficient algorithm in stable environments, it exhibited poor
performance under certain PSC scenarios. In contrast, ANN (2) demonstrated superior
performance across various PSCs, highlighting its effectiveness and reliability. Due to
its lack of temperature input, it is not considered to be relevant for applications such as
BIPV unless it could be further developed, but it could be interesting for locations and
applications that experience stable temperature environments (such as cold climates or
possibly applications such as FPVs, which would ensure good and sufficient cooling).

6. Conclusions
This paper presented a comprehensive analysis and evaluation of various Maximum
Power Point Tracking (MPPT) algorithms under Partial Shading Conditions (PSCs) for
photovoltaic (PV) systems. This study focused on traditional methods such as Perturb
and Observe (P&O) and Incremental Conductance (IC), as well as advanced techniques
including Fuzzy Logic Control (FLC), Grey Wolf Optimization (GWO), Particle Swarm
Optimization (PSO), and two Artificial Neural Network (ANN) models (ANN (1) and ANN
(2)). The simulations were conducted in a controlled MATLAB/Simulink environment
to ensure accurate and reliable comparisons. The findings indicate that while ANN (1)
performed exceptionally well under uniform solar conditions, it struggled with PSCs due
to its reliance on stable environmental inputs. On the other hand, ANN (2) demonstrated
superior performance across various PSC scenarios, outperforming traditional and ad-
vanced MPPT techniques in terms of efficiency and convergence speed. Specifically, ANN
(2) was, on average, 72% faster than P&O, 83% faster than IC, and 26% faster than PSO.
ANN (2) also showed significant advantages over GWO and FLC in complex scenarios.
The efficiency improvements were also notable, with ANN (2) being, on average, 23.41%
more efficient than P&O, 0.61% more efficient than IC, 4.28% more efficient than FLC, 1.68%
more efficient than GWO, 24.61% more efficient than PSO, and 29.76% more efficient than
ANN (1). These results underscore the potential of ANN-based approaches in enhancing
the performance and reliability of solar energy systems under dynamic shading conditions.
Solar 2024, 4 593

This study also highlighted the importance of integrating more adaptive and intelligent
systems like Artificial Intelligence (AI) and Machine Learning (ML) into PV optimization
processes. The ANN (2) model, which has only used the irradiance as an input signal,
simplified the MPPT process, reducing the complexity and potential cost of implementation
while maintaining high performance. This makes it interesting to further investigate it as a
practical solution for real-world PV systems, especially in environments with frequent and
complex shading patterns with stable temperature environments.

Author Contributions: Conceptualization, R.S. and L.M.-P.; methodology, R.S. and L.M.-P.; software,
R.S.; validation, R.S. and L.M.-P.; formal analysis, L.M.-P.; investigation, R.S.; writing—original draft
preparation, R.S.; writing—review and editing, R.S. and L.M.-P.; visualization, R.S.; supervision,
L.M.-P. All authors have read and agreed to the published version of the manuscript.
Funding: This research received no external funding.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: The original contributions presented in the study are included in the
article, further inquiries can be directed to the corresponding author.
Conflicts of Interest: The authors declare no conflicts of interest.

References
1. Kabeyi, M.J.B.; Olanrewaju, O.A. Sustainable Energy Transition for Renewable and Low Carbon Grid Electricity Generation and
Supply. Front. Energy Res. 2021, 9, 28. [CrossRef]
2. MacKay, D.J.C. Sustainable Energy–Without the Hot Air; UIT: Cambridge, UK, 2009.
3. Trends in PV Applications 2023. IEA-PVPS. Available online: https://iea-pvps.org/trends_reports/trends-2023/ (accessed on 31
March 2024).
4. Bhukya, L.; Kedika, N.R.; Salkuti, S.R. Enhanced Maximum Power Point Techniques for Solar Photovoltaic System under Uniform
Insolation and Partial Shading Conditions: A Review. Algorithms 2022, 15, 365. [CrossRef]
5. El Iysaouy, L.; Lahbabi, M.; Bhagat, K.; Azeroual, M.; Boujoudar, Y.; Saad El Imanni, H.; Aljarbouh, A.; Pupkov, A.; Rele, M.; Ness,
S. Performance enhancements and modelling of photovoltaic panel configurations during partial shading conditions. Energy Syst.
2023, 2–3. [CrossRef]
6. Youssef, A.R.; Hefny, M.M.; Ali, A.I.M. Investigation of single and multiple MPPT structures of solar PV-system under partial
shading conditions considering direct duty-cycle controller. Sci. Rep. 2023, 13, 19051. [CrossRef] [PubMed]
7. Rajavel, A.; Rathina Prabha, N. Fuzzy logic controller-based boost and buck-boost converter for maximum power point tracking
in solar system. Trans. Inst. Meas. Control 2021, 43, 945–957. [CrossRef]
8. Zou, Y.; Yan, F.; Wang, X.; Zhang, J. An efficient fuzzy logic control algorithm for photovoltaic maximum power point tracking
under partial shading condition. J. Frankl. Inst. 2020, 357, 3135–3149. [CrossRef]
9. Janandra Krishna Kishore, D.; Mohamed, M.R.; Sudhakar, K.; Peddakapu, K. Grey wolf optimization and differential evolution-
based maximum power point tracking controller for photovoltaic systems under partial shading conditions. Energy Sources Part
A Recovery Util. Environ. Eff. 2022, 44, 6286–6302. [CrossRef]
10. Motahhir, S.; Chtita, S.; Chouder, A.; El Hammoumi, A. Enhanced energy output from a PV system under partial shaded
conditions through grey wolf optimizer. Clean. Eng. Technol. 2022, 9, 100533. [CrossRef]
11. Xu, L.; Cheng, R.; Xia, Z.; Shen, Z. Improved Particle Swarm Optimization (PSO)-based MPPT Method for PV String under
Partially Shading and Uniform Irradiance Condition. In Proceedings of the 2020 Asia Energy and Electrical Engineering
Symposium (AEEES), Chengdu, China, 29–31 May 2020; pp. 771–775. [CrossRef]
12. Eltamaly, A.M.; Al-Saud, M.S.; Abo-Khalil, A.G. Performance Improvement of PV Systems’ Maximum Power Point Tracker Based
on a Scanning PSO Particle Strategy. Sustainability 2020, 12, 1185. [CrossRef]
13. Obukhov, S.; Ibrahim, A.; Zaki Diab, A.A.; Al-Sumaiti, A.S.; Aboelsaud, R. Optimal Performance of Dynamic Particle Swarm
Optimization Based Maximum Power Trackers for Stand-Alone PV System Under Partial Shading Conditions. IEEE Access 2020,
8, 20770–20785. [CrossRef]
14. Olabi, A.G.; Abdelkareem, M.A.; Semeraro, C.; Radi, M.A.; Rezk, H.; Muhaisen, O.; Al-Isawi, O.A.; Sayed, E.T. Artificial neural
networks applications in partially shaded PV systems. Therm. Sci. Eng. Prog. 2023, 37, 101612. [CrossRef]
15. Allahabadi, S.; Iman-Eini, H.; Farhangi, S. Fast Artificial Neural Network Based Method for Estimation of the Global Maximum
Power Point in Photovoltaic Systems. IEEE Trans. Ind. Electron. 2022, 69, 5879–5888. [CrossRef]
16. Kurniawan, A.; Shintaku, E. A Neural Network-Based Rapid Maximum Power Point Tracking Method for Photovoltaic Systems
in Partial Shading Conditions. Appl. Sol. Energy 2020, 56, 157–167. [CrossRef]
Solar 2024, 4 594

17. Villegas-Mier, C.G.; Rodriguez-Resendiz, J.; Álvarez Alvarado, J.M.; Rodriguez-Resendiz, H.; Herrera-Navarro, A.M.; Rodríguez-
Abreo, O. Artificial Neural Networks in MPPT Algorithms for Optimization of Photovoltaic Power Systems: A Review.
Micromachines 2021, 12, 1260. [CrossRef]
18. Al-Ezzi, A.S.; Ansari, M.N.M. Photovoltaic Solar Cells: A Review. Appl. Syst. Innov. 2022, 5, 67. [CrossRef]
19. Al-Majidi, S.D.; Abbod, M.F.; Al-Raweshidy, H.S. A novel maximum power point tracking technique based on fuzzy logic for
photovoltaic systems. Int. J. Hydrogen Energy 2018, 43, 14158–14171. [CrossRef]
20. Al-Majidi, S.D.; Abbod, M.F.; Al-Raweshidy, H.S. Design of an Efficient Maximum Power Point Tracker Based on ANFIS Using
an Experimental Photovoltaic System Data. Electronics 2019, 8, 858. [CrossRef]
21. Sagonda, A.F.; Folly, K.A. A comparative study between deterministic and two meta-heuristic algorithms for solar PV MPPT
control under partial shading conditions. Syst. Soft Comput. 2022, 4, 200040. [CrossRef]
22. Tourqui, D.E.; Betka, A.; Smaili, A.; Allaoui, T.; Tourqui, D.E.; Betka, A.; Smaili, A.; Allaoui, T. Improved Performance of a
Photovoltaic Panel by MPPT Algorithms. In Recent Developments in Photovoltaic Materials and Devices; IntechOpen: Rijeka, Croatia,
2019. [CrossRef]
23. POWER|Data Access Viewer. NASA. Available online: https://power.larc.nasa.gov/data-access-viewer/ (accessed on 13
May 2024).

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.