This article has been accepted for publication in IEEE Access.

This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3456586

Date of publication xxxx 00, 0000, date of current version xxxx 00, 0000.
Digital Object Identifier 10.1109/ACCESS.2024.Doi Number

Islanded Microgrids Frequency Support Using

Green Hydrogen Energy Storage with AI-Based
Controllers
Helmy M. El Zoghby1, A. Safwat1, Ahmed F. Bendary1, A. Hazem1, Haitham S. Ramadan2,3,
Mahmoud M. Elmesalawy4, Ramy S. A. Afia1*
1
Department of Electrical Power and Machines Engineering, Helwan University, 11792, Cairo, Egypt
2
Electrical Power and Machines Department, Zagazig University, 44519, Sharkia, Egypt
3
ISTHY, Institut International sur le Stockage de l’Hydrogene, 90400, Meroux-Moval, France
4
Department of Electronics and Communication Engineering, Helwan University, 11792, Cairo, Egypt

Corresponding author: Ramy S. A. Afia (e-mail: [email protected]).

“This work was supported by the Science, Technology & Innovation Funding Authority (STIFA) of Egypt under Grant No:47332.”

ABSTRACT Islanded microgrids, powered by renewable energy sources, offer a sustainable electricity
solution for remote areas. However, maintaining frequency stability in these systems remains a challenge due
to the intermittent nature of renewables. This research proposes an approach to enhance microgrid stability
by integrating a green hydrogen energy storage system (GHESS) and employing advanced AI-based control
strategies. The GHESS plays a pivotal role in storing excess renewable energy as hydrogen and then
converting it back to electricity when needed, reducing reliance on traditional backup generators. To optimize
microgrid performance, a hybrid single-neuron PID (SNPID) controller, augmented by machine learning
techniques, is developed and compared against conventional proportional, integral, and derivative PID and
fuzzy self-tuning PID (FSTPID) controllers. The system's performance was evaluated using four realistic
scenarios. In all cases, the SNPID controller significantly outperformed the existing options. It achieved a
58% reduction in frequency fluctuations compared to the fuzzy self-tuning PID (FSTPID) controller and an
impressive 87% reduction compared to the traditional PID controller. The Simulation results underscore the
SNPID controller's exceptional performance in frequency stability, emphasizing the transformative potential
of AI for microgrid management.

INDEX TERMS Artificial intelligence, Fuzzy controller, Green hydrogen, Islanded microgrid, Single
neuron controller.

I. INTRODUCTION transportation of fossil fuels, including the possibility of

Most conventional power plants are powered by fossil spills, leaks, and habitat destruction [4]. Although
fuels, such as coal, oil, and natural gas. However, these conventional power plants offer easily accessible energy,
resources have a high price tag. Because these non- the fact that they depend on fossil fuels for their operation
renewable resources are limited and unsustainable, the contributes significantly to various environmental and
energy needs of future generations are put at risk by their health problems, necessitating a swift transition to more
continued use. When they are burned, dangerous sustainable and cleaner alternatives [5].
greenhouse gases are released into the atmosphere [1]. Microgrids powered by renewable energy present an
These gases have a significant role in climate change and enticing picture of an energy future that is more resilient
all its disastrous effects, including rising sea levels and and sustainable [6]. Utilizing regional resources such as
extreme weather. These emissions also pollute the air, solar, wind, and biomass, these renewable power systems
harming ecosystems and human health by causing lessen dependency on fossil fuels and the environmental
respiratory issues and acid rain [2], [3]. Significant harm they cause [7]. This reduces reliance on unstable
environmental risks are associated with the extraction and centralized grids, improving energy security while

VOLUME XX, 2017 1

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3456586

promoting clean air and addressing climate change [8], [9]. logic to directly update the controller parameters
There are two primary configurations for microgrids: determined by the Harmony Search algorithm [29].
isolated and grid-connected [10]. For isolated communities Similarly, but in a different way, Ref. [30] has used sliding
that are far from the main grid, isolated microgrids are ideal mode control, and [31] has used the particle swarm
options as they run independently and provide continuous algorithm to adjust the PID controller parameters for
power even during blackouts. Grid-connected microgrids frequency regulation in small grids caused by uncertainty.
allow for the smooth integration of energy exchange based Under specific operating conditions, the PID controllers’
on local production and needs with the main grid [11]. Both gains are optimized to yield an optimal response; however,
systems offer improved power quality, reduced reliance on under other operating conditions, they produce a
external energy sources, and more local job opportunities. suboptimal response. A self-tuning PID controller
Nonetheless, while grid-connected systems entail intricate inspection is necessary to address these issues and
interconnection regulations, isolated microgrids must guarantee optimal system performance across all operating
manage fluctuating renewable energy with careful planning conditions [32], [33]. The mathematical model is
and storage solutions [12]. All things considered, unnecessary when using the fuzzy logic controller (FLC), a
renewable microgrids are a significant step toward a future heuristic method of enhancing control system performance.
of decentralized, sustainable, and community-powered It draws from personal experience. Thus, fuzzy logic, first
energy. used in industrial applications a few years ago, has recently
Renewable energy sources like solar and wind are become a handy tool. In order to satisfy a particular set of
variable and do not have intrinsic inertia like traditional closed-loop system performance requirements, the self-
grids with massive, inertia-rich generators [13]. When load tuning PID controller design aims to update the controller
imbalances arise, this variability can lead to frequency gains that are optimized based on optimization techniques
fluctuations, which could cause malfunction of the whole like the genetic algorithm method [34], [35].
system [14]. In these dynamic environments, efficient load One novel method in the field of control systems is the
frequency control (LFC) becomes essential. Energy storage single-neuron PID controller. In contrast to the traditional
is starting to take an active role in LFC [15]. Frequency is approach, this novel technique imitates the functions of
directly impacted by batteries or other storage systems, PID control using a single artificial neuron [36]. This
which absorb excess energy during peak loads and inject it neuron is a flexible tool that modifies its internal weights
back during dips [16], [17]. dynamically in response to the discrepancy between the
Although batteries are now the mainstay of microgrid expected and actual output of the system. This adjustment
energy storage, hydrogen is a newcomer with unique makes the controller more proficient over time, even with
benefits for LFC. Hydrogen is more scalable than batteries, intricate or non-linear systems [37], [38]. The single-
making it an ideal option for more prolonged imbalances or neuron approach has advantages over traditional PID
renewable lulls in larger microgrids that rely heavily on controllers with fixed gains, such as less tuning effort and
intermittent sources [18], [19]. Furthermore, hydrogen fuel better robustness to disturbances. Notwithstanding, certain
cells can quickly inject power due to their fast-lightning- obstacles persist, including the requirement for appropriate
response time compared to batteries [20], [21]. This allows training data and possible computational constraints for
quick frequency adjustments during abrupt load applications with limited resources. The single-neuron PID
fluctuations, improving microgrids' stability and resilience. controller offers a viable path toward intelligent control,
Different controllers have been used in the literature to with learning and adaptability, that could be useful for
reduce frequency fluctuations in isolated microgrids, such various robotic and industrial applications [39], [40]. A
as conventional PID controllers [22], intelligent controllers variety of optimization algorithms have been leveraged to
such as sliding mode controllers, fuzzy controllers, and identify optimal controller gain settings, as evidenced by [41],
many others [23], [24]. Most studies have compared [42], [43], [44], and [45]. Notably, particle swarm
intelligent controllers to conventional controllers to prove optimization (PSO) and genetic algorithms have demonstrated
their efficiency. In [25], an integral (I) controller with a effectiveness in tuning state-dependent proportional-integral-
linear quadratic regulator (LQR) was used. In [26], a derivative (SNPID) and fractional-order PID (FOPID)
conventional PID controller has been used to regulate the controllers, respectively.
frequency of a microgrid containing thermal and The microgrid system suggested in this study has a novel
hydroelectric power units. A fractional order-PI controller design. Excess electrical energy is converted to hydrogen
has been used to deal with the uncertain characteristics of by the Aqua electrolyzer (AE) and stored there. A stable
microgrids, such as the instability of power generated by equilibrium between generation and demand makes it
renewable energy units [27]. It is possible to train the possible to reduce frequency oscillations in power grids.
system to predict load and power generation changes using Robust and appropriate control solutions are needed to
a predictive control model [28]. Uncertainty can be dealt control microgrid frequency caused by the intermittency of
with using conventional PID controllers and type-II fuzzy renewable energy sources. This study's objective is to

VOLUME XX, 2017 7

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3456586

improve the microgrid's structural design and suggest a In the proposed microgrid system, IoT is crucial in
reliable control method for load frequency control. Two AI- optimizing and managing the energy generated and used. A
based controllers are used for controlling microgrid DGs network of sensors can be deployed on photovoltaic (PV)
and hydrogen energy storage systems: the first is the panels, biogas turbine generators, and biodiesel engine
SNPID controller augmented with machine learning, and generators to collect real-time data on power generation and
the second is a FSTPID controller. By incorporating consumption. The collected data are then fed into a central
hydrogen-fuel cells into MGs, the research makes several control system that analyzes it and makes intelligent decisions
noteworthy contributions, such as: [47], [48]. The system can manage the fuel cell and
• The microgrid under study contains several electrolyzer based on these insights. For instance, during peak
renewable energy sources, such as solar, biofuel, and solar production, the system might prioritize using the excess
a green hydrogen energy storage system. They are energy to power the electrolyzer, creating hydrogen for
all environmentally friendly elements, as depicted in storage while reducing reliance on the fuel cell. This data-
Fig. 1. driven approach from IoT ensures efficient energy
• Contrary to the literature where solar power has been management within the microgrid.
considered constant in the simulation system to The structure of the paper is as follows: the islanded
facilitate the operation of the controller [46], this microgrid configuration with GHESS is covered in Section 2,
study uses actual solar data to prove the efficiency the Single-Neuron Proportional-Integral-Derivative (SNPID)
of the proposed controllers. controller augmented with machine learning, and the Fuzzy
• Create and evaluate advanced AI algorithms Self-Tuning PID controller (FSTPID) and their design steps
(FSTPID and SNPID) to better control hydrogen are presented in Section 3, several case studies are examined
energy storage systems and enhance frequency in Section 4 to show how efficient does the proposed controller
stability. perform in comparison with FSTPID controller and
• To underscore the efficacy of the proposed AI conventional PID controller. Finally, the research findings are
controllers relative to conventional controllers, a summarized in Section 5.
diverse range of potential scenarios are modeled.
II. FUEL CELL-POWERED AUTONOMOUS MICROGRID
MODELING
The utilization of a subject region's energy resources in
isolated microgrids should consider the environment and
economy. Because renewable energy sources are intermittent
and variable, using them calls for improved energy

FIGURE 1. Proposed microgrid structure.

VOLUME XX, 2017 7

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3456586

management practices and energy storage methods. Storage the generated hydrogen to electrical power to fulfill the
and management of energy in this situation is greatly aided by required demand when necessary, and the electrolyzer, which
fuel cells (FCs) and aqua electrolyzers (AEs). The microgrid transforms excess energy into hydrogen. This control system
layout in this area was planned to optimize the utilization of guarantees a stable balance between supply and demand. Fig.2
renewable resources available on the nearby farm. The design shows the autonomous power system block diagram under
included PV, FC, AE, Reformer units, a biodiesel engine investigation. This section briefly examines the mathematical
generator (BDEG) system, and a biogas turbine generator equations for each component comprising the model.
(BGTG) system within the self-sufficient renewable
microgrid. The system parameters are detailed in Table 2. A. BIOGAS TURBINE GENERATOR SYSTEM
LFC is achieved in the microgrid by consistently meeting The combustion chamber, fuel system, gas turbine, speed
the load demand. Nevertheless, the time interval during which regulator, and Gas valve actuator make up the BGTG system
sunlight can be received during the day is restricted and [49]. The BGTG system converts animal dung and
depends on the weather. However, the amount of waste and biodegradable wastes into biogas, which is then used to
energy crops that biogas and biodiesel plants can convert into produce electricity. Equation (1) formula expresses the BGTG
energy varies based on the surrounding conditions. These transfer function [47]. The discharge time constants, biogas
power generation systems' intermittent and fluctuating delay, combustion reaction delay, valve actuator, lag time, and
characteristics highlight the importance of energy storage and lead time constants of the BGTG unit are denoted by the
management strategies. The pf1 and pf2 participation variables TBT, TBG, TCR, bB, YCS, and XCS, respectively.
coefficients are utilized to adjust the LFC involvement rates of
the BGTG and BDEG modules in the designed microgrid
1 + X CS S 1 + TCR S 1
based on their capacity for biogas and biodiesel, respectively. GBGTG ( S ) = ( )( )( ) (1)
(1 + YCS S )(1 + bB S ) 1 + TBG S 1 + TBT S
If the system produces excess energy, the AE creates and
stores hydrogen. Furthermore, hydrogen and carbon atoms can
be generated from Methane gas (CH4) if the system has excess B. BIODIESEL ENGINE GENERATOR SYSTEM
biogas through various processes. The hydrogen that is Energy plants are processed chemically to create biodiesel.
produced is then stored. Enhanced sustainability is achieved Equation (2) [48] provides the BDEG system linearized
by using stored hydrogen as fuel in the FC during a power transfer function, which first transforms biodiesel's energy
outage. The system frequency deviates if there is an imbalance into mechanical energy and subsequently into electrical
between the microgrid load and produced power. The power energy. The terms time constant, engine gain, valve actuator
balance between renewable sources is maintained by delay, and valve gain of the BDEG are represented in the
Controller 1. Controller 2 inspects the fuel cell, which converts equation by the letters TBE, KBE, TVA, and KVA.

FIGURE 2. Autonomous power system block diagram under investigation.

VOLUME XX, 2017 7

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3456586

in this situation. Equation (7) provides the net power (PE)

K S K BE
GBDEG ( S ) = ( VA ) ( ) (2) change in the suggested microgrid at any time.
1 + TVA S 1 + TBE S
PE = PPV + PBGTG + PBDEG + PFC − PAE − PL (7)

C. AQUA-ELECTROLYZER Equation (8) displays the transfer function of the fuel cell
The energy storage component of the renewable microgrid microgrid's equivalent generator dynamic model. The
system, which comprises renewable energy resources (RESs) damping constant and the equivalent inertia constant are
with unstable structures, is the aqua-electrolyzer. The energy denoted by D and M in the equation, respectively [54]. The
demand in the grid is quickly met using energy storage units. proposed MG parameters are listed in Table I.
The extra energy is transformed into hydrogen and stored in
the electrolyzer. The fuel cell (FC) system uses the hydrogen 1
GSYS ( S ) = ( ) (8)
stored as fuel if the system needs energy. Equation 3 presents MS + D
the aqua-electrolyzer's transfer function [50], where the time
TABLE I
constant and gain of the AE unit are denoted by TAE and KAE, PARAMETERS OF THE RENEWABLE MG SYSTEM [55].
respectively, in the equation.
Parameter Value Parameter Value
K AE
GAE ( S ) = ( ) (3) TPV, KPV 1.8s, 1 TBE, KBE 0.5s, 0.002
1 + TAE S bB, YCS, XCS 0.05, 1s, 0.6 R, D, M 2.4, 0.012,
0.2s
TBT, TBG, TCR 0.2s, 0.23s, TFC, KFC 4s, 0.01
D. FUEL CELL SYSTEM 0.01s
Excess energy is converted into hydrogen during off-peak TVA, KVA 0.05s, 1 TAE, KAE 0.5s, 1
hours by using the AE unit. To fulfill the load power
requirement, the FC unit converts the stored hydrogen into III. AI-BASED CONTROLLERS DESIGN
electrical power in situations where the demand cannot be Stable control strategies are essential for maintaining
supplied from the microgrid's RES. Equation (4) displays the frequency stability in isolated microgrids powered by
FC system's transfer function. The terms KFC and TFC stand for renewable energy sources. To achieve this, Two AI-based
the FC unit's gain and time constant, respectively [51]. controllers are proposed: the first is the Single-Neuron
Proportional-Integral-Derivative (SNPID) controller
K FC augmented with machine learning, and the second is a Fuzzy
GFC ( S ) = ( ) (4)
1 + TFC S Self-Tuning PID controller (FSTPID).

A. SINGLE-NEURON PROPORTIONAL-INTEGRAL-
E. PHOTOVOLTAIC SYSTEM DERIVATIVE CONTROLLER DESIGN
Arrays of PV panels, a DC/AC power converter, and The SNPID controller is categorized as a type of online
connecting devices make up the PV system, transforming adaptive PID controller because it has the ability to modify the
solar energy into electrical energy. Based on the PV cell's weighting factors in order to perform self-organizing and
surface temperature and radiation intensity, the PV unit's adaptive functions. SNPID controller enhances and optimizes
electrical output power can be expressed mathematically as conventional PID control by fusing neural networks with
[52]: traditional PID control. The SNPID controller's architecture,
PPV =  S 1 − 0.005(25 + Ta ) (5) which utilizes an incremental PID control algorithm, is
depicted in Fig. 3. The desired output is represented by R,
while the system output is defined as Y.
This formula uses η as the solar arrays' conversion
efficiency, ϕ as the irradiance in kW/m2, Ta as the PV surface's 3 (9)
surface temperature in degrees Celsius, and S as the measured u (k ) = u (k − 1) + k  wi (k )X i (k )
surface area. Equation (6) can be used to express the PV i =1

transfer function. The time constant and gain of the PV are 3

denoted by TPV and KPV, respectively [53]. wi (k ) = wi (k ) /  wi (k ) (10)
i =1
K PV
GPV ( S ) = ( ) (6) In (9) and (10), w is the neuron's weight, X is the error
1 + TPV S
signal output from the state converter, K is the proportion
coefficient, and u ( k ) is the controller's output [56]. It is easy
F. SYSTEM DYNAMICS MODEL to obtain the error signal through:
Microgrid stability can be achieved by promptly responding to
• Writing the continuous time PID controller equation is
the load's power demand. A stable control strategy is needed
the first step in proving a discrete PID controller.

VOLUME XX, 2017 7

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3456586

w2 = w2 (k − 1) + i X 2 (k − 1)u (k − 1)e(k − 1) (18)

w3 = w3 (k − 1) +  D X 3 (k − 1)u (k − 1)e(k − 1) (19)

Where e is the error,  P is the proportional learning speed, i
is the integral learning speed, and  D is the derivative learning
speed [49]. To find the optimal values for six critical
parameters: neuron coefficient [ K ] , learning speed
FIGURE 3. Single-Neuron PID Controller.
parameters [ P ,i , D ] , neuron weights [ w1 (k − 1), w2 (k − 1),
w3 ( k − 1)] , and the output u (k − 1) , the SNPID control
t
architecture employs particle swarm optimization (PSO). The
u (t ) = K P e(t ) + Ki  e(t )dt + K d e(t ) (11) used cost function minimizes the integrated square error e(t )
0
. The controller parameters determined by the optimization
Where KP, Ki, and Kd are proportional, integral, and algorithm are presented in Table II.
derivative gains, respectively. Table II
• Discretization can be achieved by differentiating both PARAMETERS OF SNPID CONTROLLER.
sides of (11) Controller 1 Controller 2
Parameter Value Parameter Value
u (t ) = K P e(t ) + K i e(t ) + K d e(t ) (12) 𝑊1 , 𝑊2 , 𝑊3 2.33, 8.99, 0.15 𝑊1 , 𝑊2 , 𝑊3 9.94, 1.75, 1.87
• Using (12) and the backward differentiation method
𝜂𝑃 , 𝜂𝑖 , 𝜂𝐷 0.0026, 0.0016, 𝜂𝑃 , 𝜂𝑖 , 𝜂𝐷 0.0083, 0.0047,
yields: 0.0088 0.0043
𝐾 0.002 𝐾 0.019
u (k ) − u (k − 1) = K p [e(k ) − e(k − 1)] + K i [e(k )] +
(13)
K d [e( k ) − e( k − 1)]
B. FUZZY SELF-TUNING PID (FSTPID) CONTROLLER
u (k ) = u (k − 1) + K p [e(k ) − e(k − 1)] + K i [e(k )] + DESIGN
(14) After nearly 50 years of development and application, PID
K d [e(k ) − e(k − 1)] − [e(k − 1) − e(k − 2)] controllers are still widely used in process control and
industrial automation. The primary causes of this widespread
u (k ) = u (k − 1) + K p [e(k ) − e(k − 1)] + K i [e(k )] + use are the wide range of linear systems' low cost, low
(15) maintenance, simplicity of design, and effectiveness. Under
K d [e(k ) − 2e(k − 1)] + e(k − 2)]
specific operating conditions, the PID controller's gain is
optimized to yield an optimal response; however, under other
u (k ) = u (k − 1) + K p [ X 1 (k )] + Ki [ X 2 (k )] + K d [ X 3 (k )] (16)
operating conditions, it produces a suboptimal response. To
Where X 1 (k ) = e(k ) − e(k − 1) , X 2 (k ) = e(k ) , and X 3 (k ) = address these issues and guarantee optimal system
e(k ) − 2e(k − 1) + e(k − 2) . From (13) to (16), it is evident that performance across all operating conditions, a self-tuning
the proportional error is represented by X 1 ( k ) , the integral PID controller inspection is necessary. Unlike traditional
error by X 2 (k ) , and the derivative error by X 3 (k ) . These control systems, fuzzy logic controllers (FLCs) use heuristic
error signals are inputs to the single neuron PID after being rules based on human expertise rather than precise
multiplied by the appropriate weights. Setting the correct mathematical models to enhance system performance. Thus,
value for the K coefficient is crucial because a higher value fuzzy logic, first used in industrial applications a few years
indicates that the control system will operate more quickly, but ago, has recently become a very useful tool. Fuzzy self-
it also has a significant overshoot that increases the risk of the tuning of PI controller gains has been designed recently
control unit becoming unstable. The way the SNPID controller using fuzzy sets and fuzzy logic. To satisfy a particular set
weights are adjusted has a substantial impact on how efficient of closed-loop system performance requirements, the self-
the controller is. Based on neural network learning theory, tuning PID controller design updates the controller gains,
numerous weight learning algorithms exist, such as supervised which are optimized using the genetic algorithm method
Hebb learning rules, unsupervised Hebb learning rules, [58].
improved Hebb learning rules, and delta-supervised learning The controller comprises a conventional PID controller and
rules [57]. Supervised learning is the most well-known Hebb a self-tuning fuzzy logic control section. As illustrated in Fig.
learning rule algorithm employed in this paper. 4, a system's proportional, integral, and derivative gains (
K P , K i , K d ) can be automatically modified online. The PID
w1 = w1 (k − 1) +  p X1 (k − 1)u (k − 1)e(k − 1) (17) controller's fundamental architecture is:

VOLUME XX, 2017 7

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3456586

The universe of error discourse and the conversion of error

inputs to fuzzy must match its actual values. This can be
achieved by either placing normalization gains before entering
the controller or by using [emin emax ] and [cemin cemax ] as the
input period limits. The values of K e and K e are
determined using genetic algorithm by dividing the boundary
values of the input membership functions of error (e) and
change in error (Δe) by their respective maximum actual
FIGURE 4. Adaptive fuzzy-self-tuning controller structure.
values to standardize them within the limits of the membership
universe of discourse. The controller parameters determined
de(t ) by the optimization algorithm are presented in Table III. The
U (t ) = K P e(t ) + K i  e(t )dt + K d (20) normalizing gain for the error input is denoted by K e , and the
dt
normalizing gain for the change in error input is represented
The following equation is obtained after a fuzzy controller is by K e . Tables IV, V, and VI tabulate the rule bases used to
added to modify the PID controller gains: determine K P1 , Ki1 , and K d 1 , respectively.
TABLE III
de(t ) PARAMETERS OF FSTPID CONTROLLER.
U (t ) = K P 2 e(t ) + K i 2  e(t )dt + K d 2 (21)
dt Controller 1 Controller 2
Parameter Value Parameter Value
Where K P 2 = K P  K P1 , K i 2 = K i  K i1 , and K d 2 = K d  K d 1 . 𝐾𝑒 0.8115 𝐾𝑒 0.08147
The fuzzy controller gain outputs are denoted as
𝐾Δ𝑒 0.5885 𝐾Δ𝑒 0.1211
K P1 , K i1 , and K d 1 .
Fig. 5 illustrates the linguistic labels used to name the input TABLE IV
membership functions for error and change-in-error. {Positive KP1 RULE BASES.
Big ‘PB’, Positive Medium ‘PM’, Positive Small ‘PS’, Zero ce
e PS PM PB ZE NB NM NS
‘Z’, Negative Small ‘NS’, Negative Medium ‘NM’, Negative
Big ‘NB’} are the linguistic labels assigned to the inputs. The PS MB B VB B B B B
outputs are linguistically labeled {Very Big ‘VB,’ Medium PM B MB VB MB MB MB MB
PB VB VB VB VB VB VB VB
Big ‘MB,’ Big ‘B,’ Medium ‘M,’ Small ‘S,’ Medium Small ZE S S S MS ZE ZE ZE
‘MS,’ Zero ‘Z’}. NB VB VB VB VB VB VB VB
The fuzzy output membership function curves are shown NM B MB VB MB MB MB MB
NS MB B VB B B B B
in Fig. 6. No particular analytical procedures must be followed
when designing rule bases and membership functions; instead, TABLE V
it is based on the operator’s experience and skill. Ki1 RULE BASES.
ce
e PS PM PB ZE NB NM NS

PS S S S S S S S
PM M M M M M M M
PB M M M M M M M
ZE MS MS MS ZE MS MS MS
NB M M M M M M M
NM M M M M M M M
NS S S S S S S S

TABLE VI
FIGURE 5. Output membership functions. Kd1 RULE BASES.
ce
e PS PM PB ZE NB NM NS

PS VB VB VB VB MB MB VB
PM VB VB VB VB B MB VB
PB VB VB VB VB VB VB VB
ZE VB VB VB MB M B MB
NB MB B VB M ZE MS S
NM B B VB B MS S M
NS VB VB VB MB S M B

FIGURE 6. Input membership functions of e and ce. The last phase is the defuzzification procedure, which
converts the fuzzy output to a crisp value for use as a non-
fuzzy control action. The most popular defuzzification

VOLUME XX, 2017 7

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3456586

FIGURE 7. Fuzzy inference system output surfaces.

technique is to apply the center of gravity, also called the interactions with the load frequency control mechanism are
center of the area [59]. It is expressed as: meticulously examined across four distinct scenarios. A
r
comparative analysis is conducted between the proposed AI
 u (u )u
i =1
i i
and the conventional PID controllers under each scenario.
Figure 8 visually represents the load fluctuation profiles and
u= (22)
r solar radiation patterns employed throughout the simulation
 u (ui )
i =1
studies.

A. SCENARIO 1
Where u (ui ) is the membership grade (weight) of the element In this baseline scenario, an ideal operating condition is
ui , which is the result of the rule i . The fuzzy inference established wherein all renewable energy sources are
system's output surface is displayed for K P1 , K i1 , and K d 1 presumed to function optimally. To accurately reflect real-
in Fig. 7 (left), 7 (middle), and 7 (right), respectively. The world solar conditions, actual solar radiation data is employed.
output value is calculated based on the current error and the Furthermore, biogas and biodiesel availability are stabilized at
rate of error change. 50% capacity, represented by pf1 and pf2 values of 0.5,
respectively. A thorough examination of Figure 9
IV. RESULTS, ANALYSIS AND CRITICS
unequivocally demonstrates the superior performance of the
This section presents a detailed simulation analysis conducted
within the MATLAB/Simulink environment to evaluate the proposed controllers in effectively damping frequency
performance of the proposed AI controllers in preserving oscillations. The Integral Time Absolute Error (ITAE)
system stability and mitigating frequency deviations within an performance metric is calculated to assess controller efficacy.
isolated microgrid. To replicate real-world operating The proposed SNPID controller exhibits a notably lower ITAE
conditions accurately, the simulations incorporate time- value of 0.1747 compared to the FSTPID controller's 0.5837
varying solar radiation data from the National Renewable and the traditional PID controller's significantly higher value
Energy Laboratory [60]. Given the inherent variability of of 3.2159. These quantitative results reinforce the qualitative
RES, such as BESS, BG, and PV systems, their dynamic observations from Figure 9, confirming the substantial

FIGURE 8. Variations in load and PV power.

VOLUME XX, 2017 7

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3456586

improvement offered by the SNPID controller in regulating These findings showed that, compared to the conventional
system frequency. PID controller and fuzzy self-tuning PID controller, the
suggested SNPID controller reduced the maximum frequency
deviation by roughly 48% and 20%, respectively.
Additionally, it significantly decreased the oscillations in
frequency.
The results of applying the ITAE to the SNPID, FSTPID,
and PID controller performances are 0.0548, 0.1643, and
1.0968, in that order.

C. SCENARIO 3
The solar cells operate normally due to ideal weather, as in
Scenario 1. Participation factor pf1=0.25 due to insufficient
quantities of biogas being available in stocks to meet the load
FIGURE 9. Frequency response of scenario 1. requirements. Participation factor pf2 = 0.75 due to the
presence of large amounts of biodiesel. The data is shown in
Fig. 11, and the frequency responses obtained for this scenario
B. SCENARIO 2 are compared with those obtained from the proposed SNPID,
It was assumed in this scenario that the microgrid had no PV FSTPID, and PID controllers.
support because it was nighttime or inclement weather. Same Upon analyzing Fig. 12, it becomes evident that the
as Scenario 1, given the availability of biogas and biodiesel in suggested controller effectively reduces the maximum
stocks, the participation coefficients for each = 0.5. The frequency deviations and oscillations.
suggested controller's performance in this scenario was (ITAE) is used to get the values 0.1951 for the SNPID
contrasted with that of the PID controller shown in Fig. 10. controller, 0.7968 for the FSTPID controller, and 3.5007 for
the PID controller.

FIGURE 10. Frequency response of scenario 2.

D. SCENARIO 4
Like Scenario 3, the solar cells operate at total capacity. In
contrast, the participation factor pf1=0.75 due to the
availability of biogas, and the participation factor pf2 =0.25
due to the unavailability of biodiesel. A comparison was made
between the SNPID controller's performance in Scenario 4 and
the controllers' respective performances in Fig. 12. Based on
the examination of Fig. 11, it can be concluded that the
suggested controller successfully reduced the frequency
deviations. As with the other three scenarios, the suggested
control for Scenario 4 significantly decreased frequency
FIGURE 11. Frequency response of scenario 3.

VOLUME XX, 2017 7

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3456586

oscillations and settling time compared to the measured System with Integrated Smart Management using Artificial
controllers. Intelligence and IoT.”
When the ITAE performance index is applied, the SNPID,
FSTPID, and PID controllers give values of 0.1617, 0.4614, REFERENCES
and 3.0939, respectively.
Table VII compares the performance of different scenarios. [1] A. Shafiq et al., “Solar PV-Based Electric Vehicle
The results show that SNPID outperforms FSTPID and Charging Station for Security Bikes: A Techno-
traditional PID, exhibiting lower frequency fluctuation. Economic and Environmental Analysis,”
Sustainability 2022, Vol. 14, Page 13767, vol. 14, no.
TABLE VII
DIFFERENT SCENARIOS COMPARISON
21, p. 13767, Oct. 2022, doi: 10.3390/SU142113767.
[2] V. Strezov and H. H. Cho, “Environmental impact
ITAE PID FST-PID SNPID assessment from direct emissions of australian
Scenario 1 3.2159 0.5837 0.1747
thermal power generation technologies,” J Clean
Scenario 2 1.0968 0.1643 0.0548 Prod, vol. 270, p. 122515, Oct. 2020, doi:
Scenario 3 3.5007 0.7968 0.1951 10.1016/J.JCLEPRO.2020.122515.
Scenario 4 3.0939 0.4614 0.1617 [3] A. K. Karmaker, M. M. Rahman, M. A. Hossain, and
M. R. Ahmed, “Exploration and corrective measures
of greenhouse gas emission from fossil fuel power
stations for Bangladesh,” J Clean Prod, vol. 244, p.
118645, Jan. 2020, doi:
10.1016/J.JCLEPRO.2019.118645.
[4] A. Gani, “Fossil fuel energy and environmental
performance in an extended STIRPAT model,” J
Clean Prod, vol. 297, p. 126526, May 2021, doi:
10.1016/J.JCLEPRO.2021.126526.
[5] A. Shamoon et al., “Environmental impact of energy
production and extraction of materials - a review,”
Mater Today Proc, vol. 57, pp. 936–941, Jan. 2022,
doi: 10.1016/J.MATPR.2022.03.159.
[6] O. Elma, S. K. Dash, and S. Chakraborty, “Energy
FIGURE 12. Frequency response of scenario 4. planning for a sustainable transition to a decarbonized
generation scenario,” Sustainable Energy Planning in
Smart Grids, pp. 1–20, Jan. 2024, doi: 10.1016/B978-
V. CONCLUSIONS AND PERSPECTIVES
0-443-14154-6.00017-X.
As demonstrated by this research, Microgrid performance is [7] M. Aurangzeb, A. Xin, S. Iqbal, and M. U. Jan, “An
significantly enhanced through the integration of AI and IoT Evaluation of Flux-Coupling Type SFCL Placement
technologies. A standalone microgrid incorporating biodiesel, in Hybrid Grid System Based on Power Quality Risk
biogas, and solar PV generation was optimized by converting Index,” IEEE Access, vol. 8, pp. 98800–98809, 2020,
excess energy and biogas into hydrogen for fuel cell backup. doi: 10.1109/ACCESS.2020.2996583.
An AI-based frequency controller, refined through genetic [8] M. Khalid, “Smart grids and renewable energy
algorithms and particle swarm optimization, surpassed systems: Perspectives and grid integration
traditional methods by a substantial margin. The system's challenges,” Energy Strategy Reviews, vol. 51, p.
effectiveness was validated across various scenarios, proving 101299, Jan. 2024, doi: 10.1016/J.ESR.2024.101299.
the feasibility of AI-driven, hydrogen-integrated microgrids. [9] P. Pablo-Romero et al., “A Global Assessment: Can
The proposed controller (SNPID) outperforms other Renewable Energy Replace Fossil Fuels by 2050?,”
controllers (FSTPID and PID) in the case studies, achieving Sustainability 2022, Vol. 14, Page 4792, vol. 14, no.
frequency response gains that are 58% to 87% higher based 8, p. 4792, Apr. 2022, doi: 10.3390/SU14084792.
on the ITAE-performance index. Future research should [10] K. Cabana-Jiménez, J. E. Candelo-Becerra, and V. S.
explore advanced AI applications, hybrid energy storage, grid Santos, “Comprehensive Analysis of Microgrids
integration strategies, and comprehensive lifecycle Configurations and Topologies,” Sustainability 2022,
assessments to maximize microgrid potential. Vol. 14, Page 1056, vol. 14, no. 3, p. 1056, Jan. 2022,
doi: 10.3390/SU14031056.
[11] N. Shirzadi, F. Nasiri, and U. Eicker, “Optimal
ACKNOWLEDGMENT
Configuration and Sizing of an Integrated Renewable
This paper is based upon work supported by the Science,
Energy System for Isolated and Grid-Connected
Technology & Innovation Funding Authority (STIFA) of
Microgrids: The Case of an Urban University
Egypt under Grant No:47332 for a funded Research Project
entitled “Sustainable Green Hydrogen-based Energy Eco-

VOLUME XX, 2017 7

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3456586

Campus,” Energies 2020, Vol. 13, Page 3527, vol. 13, [21] A. F. El-Hamalawy, H. E. Z. Farag, and A. Asif,
no. 14, p. 3527, Jul. 2020, doi: 10.3390/EN13143527. “Optimal design of grid-connected green hydrogen
[12] J. Hmad, A. Houari, A. E. M. Bouzid, A. Saim, and plants considering electrolysis internal parameters and
H. Trabelsi, “A Review on Mode Transition battery energy storage systems,” Energy Convers
Strategies between Grid-Connected and Standalone Manag, vol. 302, p. 118127, Feb. 2024, doi:
Operation of Voltage Source Inverters-Based 10.1016/J.ENCONMAN.2024.118127.
Microgrids,” Energies 2023, Vol. 16, Page 5062, vol. [22] K. Nosrati, A. Tepljakov, E. Petlenkov, Y. Levron, V.
16, no. 13, p. 5062, Jun. 2023, doi: Skiparev, and J. Belikov, “Coordinated PI-based
10.3390/EN16135062. frequency deviation control of isolated hybrid
[13] H. Ur Rehman, X. Yan, M. A. Abdelbaky, M. Ullah microgrid: An online multi-agent tuning approach via
Jan, and S. Iqbal, “An advanced virtual synchronous reinforcement learning,” IEEE PES Innovative Smart
generator control technique for frequency regulation Grid Technologies Conference Europe, vol. 2022-
of grid-connected PV system,” International Journal October, 2022, doi: 10.1109/ISGT-
of Electrical Power & Energy Systems, vol. 125, p. EUROPE54678.2022.9960311.
106440, Feb. 2021, doi: [23] V. Skiparev et al., “Virtual Inertia Control of Isolated
10.1016/J.IJEPES.2020.106440. Microgrids Using an NN-Based VFOPID Controller,”
[14] S. Iqbal, S. Habib, N. H. Khan, M. Ali, M. Aurangzeb, IEEE Trans Sustain Energy, vol. 14, no. 3, pp. 1558–
and E. M. Ahmed, “Electric Vehicles Aggregation for 1568, Jul. 2023, doi: 10.1109/TSTE.2023.3237922.
Frequency Control of Microgrid under Various [24] K. Nosrati, A. Tepljakov, E. Petlenkov, V. Skiparev,
Operation Conditions Using an Optimal Coordinated J. Belikov, and Y. Levron, “Constrained Intelligent
Strategy,” Sustainability 2022, Vol. 14, Page 3108, Frequency Control in an AC Microgrid: An Online
vol. 14, no. 5, p. 3108, Mar. 2022, doi: Reinforcement Learning Based PID Tuning
10.3390/SU14053108. Approach,” IEEE Power and Energy Society General
[15] S. Iqbal et al., “The Impact of V2G Meeting, vol. 2023-July, 2023, doi:
Charging/Discharging Strategy on the Microgrid 10.1109/PESGM52003.2023.10252482.
Environment Considering Stochastic Methods,” [25] R. Mandal and K. Chatterjee, “Frequency control and
Sustainability 2022, Vol. 14, Page 13211, vol. 14, no. sensitivity analysis of an isolated microgrid
20, p. 13211, Oct. 2022, doi: 10.3390/SU142013211. incorporating fuel cell and diverse distributed energy
[16] U. Datta, A. Kalam, and J. Shi, “Battery Energy sources,” Int J Hydrogen Energy, vol. 45, no. 23, pp.
Storage System for Aggregated Inertia-Droop Control 13009–13024, Apr. 2020, doi:
and a Novel Frequency Dependent State-of-Charge 10.1016/J.IJHYDENE.2020.02.211.
Recovery,” Energies 2020, Vol. 13, Page 2003, vol. [26] Y. Xu, C. Li, Z. Wang, N. Zhang, and B. Peng, “Load
13, no. 8, p. 2003, Apr. 2020, doi: Frequency Control of a Novel Renewable Energy
10.3390/EN13082003. Integrated Micro-Grid Containing Pumped
[17] U. Akram, M. Nadarajah, R. Shah, and F. Milano, “A Hydropower Energy Storage,” IEEE Access, vol. 6,
review on rapid responsive energy storage pp. 29067–29077, Apr. 2018, doi:
technologies for frequency regulation in modern 10.1109/ACCESS.2018.2826015.
power systems,” Renewable and Sustainable Energy [27] M. T. Özdemir, “The effects of the FOPI controller
Reviews, vol. 120, p. 109626, Mar. 2020, doi: and time delay on stability region of the fuel cell
10.1016/J.RSER.2019.109626. microgrid,” Int J Hydrogen Energy, vol. 45, no. 60,
[18] C. Spataru and P. Bouffaron, “Off-grid energy pp. 35064–35072, Dec. 2020, doi:
storage,” Storing Energy: with Special Reference to 10.1016/J.IJHYDENE.2020.05.211.
Renewable Energy Sources, pp. 731–752, Jan. 2022, [28] J. Hu, Y. Shan, J. M. Guerrero, A. Ioinovici, K. W.
doi: 10.1016/B978-0-12-824510-1.00033-7. Chan, and J. Rodriguez, “Model predictive control of
[19] J. Park, S. Kang, S. Kim, H. S. Cho, S. Heo, and J. H. microgrids – An overview,” Renewable and
Lee, “Techno-economic analysis of solar powered Sustainable Energy Reviews, vol. 136, p. 110422,
green hydrogen system based on multi-objective Feb. 2021, doi: 10.1016/J.RSER.2020.110422.
optimization of economics and productivity,” Energy [29] M. H. Khooban, T. Niknam, F. Blaabjerg, and T.
Convers Manag, vol. 299, p. 117823, Jan. 2024, doi: Dragičević, “A new load frequency control strategy
10.1016/J.ENCONMAN.2023.117823. for micro-grids with considering electrical vehicles,”
[20] R. Agarwal, S. Singh, and A. E. Shalan, “Sustainable Electric Power Systems Research, vol. 143, pp. 585–
Energy Storage Devices and Device Design for 598, Feb. 2017, doi: 10.1016/J.EPSR.2016.10.057.
Sensors and Actuators Applications,” Sustainable [30] A. Bagheri, A. Jabbari, and S. Mobayen, “An
Energy Storage in the Scope of Circular Economy: intelligent ABC-based terminal sliding mode
Advanced Materials and Device Design, pp. 225–290, controller for load-frequency control of islanded
Apr. 2023, doi: 10.1002/9781119817741.CH10. micro-grids,” Sustain Cities Soc, vol. 64, p. 102544,
Jan. 2021, doi: 10.1016/J.SCS.2020.102544.

VOLUME XX, 2017 7

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3456586

[31] A. Safari, F. Babaei, and M. Farrokhifar, “A load [41] C. Zhang, G. Tian, A. M. Fathollahi-Fard, W. Wang,
frequency control using a PSO-based ANN for micro- P. Wu, and Z. Li, “Interval-Valued Intuitionistic
grids in the presence of electric vehicles,” Uncertain Linguistic Cloud Petri Net and Its
International Journal of Ambient Energy, vol. 42, no. Application to Risk Assessment for Subway Fire
6, pp. 688–700, 2021, doi: Accident,” IEEE Transactions on Automation
10.1080/01430750.2018.1563811. Science and Engineering, vol. 19, no. 1, pp. 163–177,
[32] Y. S. Lee and D. W. Jang, “Optimization of Neural Jan. 2022, doi: 10.1109/TASE.2020.3014907.
Network-Based Self-Tuning PID Controllers for [42] R. K. Khadanga, D. Das, A. Kumar, and S. Panda,
Second Order Mechanical Systems,” Applied “Sine augmented scaled arithmetic optimization
Sciences 2021, Vol. 11, Page 8002, vol. 11, no. 17, p. algorithm for frequency regulation of a virtual inertia
8002, Aug. 2021, doi: 10.3390/APP11178002. control based microgrid,” ISA Trans, vol. 138, pp.
[33] N. Bakhtadze et al., “Self-Tuning Controller Using 534–545, Jul. 2023, doi:
Shifting Method,” Mathematics 2023, Vol. 11, Page 10.1016/J.ISATRA.2023.02.025.
4548, vol. 11, no. 21, p. 4548, Nov. 2023, doi: [43] H. Yu et al., “Key technology and application analysis
10.3390/MATH11214548. of quick coding for recovery of retired energy vehicle
[34] S. Mahfoud, A. Derouich, N. E. L. Ouanjli, M. E. L. battery,” Renewable and Sustainable Energy
Mahfoud, and M. Taoussi, “A New Strategy-Based Reviews, vol. 135, p. 110129, Jan. 2021, doi:
PID Controller Optimized by Genetic Algorithm for 10.1016/J.RSER.2020.110129.
DTC of the Doubly Fed Induction Motor,” Systems [44] A. M. Fathollahi-Fard, M. Hajiaghaei-Keshteli, and
2021, Vol. 9, Page 37, vol. 9, no. 2, p. 37, May 2021, R. Tavakkoli-Moghaddam, “Red deer algorithm
doi: 10.3390/SYSTEMS9020037. (RDA): a new nature-inspired meta-heuristic,” Soft
[35] N. Jalali, H. Razmi, and H. Doagou-Mojarrad, comput, vol. 24, no. 19, pp. 14637–14665, Oct. 2020,
“Optimized fuzzy self-tuning PID controller design doi: 10.1007/S00500-020-04812-Z.
based on Tribe-DE optimization algorithm and rule [45] Amir Mohammad Fathollahi-Fard, Mostafa
weight adjustment method for load frequency control Hajiaghaei-Keshteli, Reza Tavakkoli-Moghaddam,
of interconnected multi-area power systems,” Appl “The social engineering optimizer (SEO),”
Soft Comput, vol. 93, p. 106424, Aug. 2020, doi: applications of artificial intelligence, Elsevier,
10.1016/J.ASOC.2020.106424. Accessed: Jul. 19, 2024. [Online]. Available:
[36] C. Napole, O. Barambones, I. Calvo, and J. Velasco, https://www.sciencedirect.com/science/article/pii/S0
“Feedforward Compensation Analysis of 952197618300873.
Piezoelectric Actuators Using Artificial Neural [46] A. Oshnoei, M. Kheradmandi, and S. M. Muyeen,
Networks with Conventional PID Controller and “Robust Control Scheme for Distributed Battery
Single-Neuron PID Based on Hebb Learning Rules,” Energy Storage Systems in Load Frequency Control,”
Energies 2020, Vol. 13, Page 3929, vol. 13, no. 15, p. IEEE Transactions on Power Systems, vol. 35, no. 6,
3929, Aug. 2020, doi: 10.3390/EN13153929. pp. 4781–4791, Nov. 2020, doi:
[37] M. Zhou et al., “Development of a Depth Control 10.1109/TPWRS.2020.2997950.
System Based on Variable-Gain Single-Neuron PID [47] B. Mohanty, S. Panda, and P. K. Hota, “Controller
for Rotary Burying of Stubbles,” Agriculture 2022, parameters tuning of differential evolution algorithm
Vol. 12, Page 30, vol. 12, no. 1, p. 30, Dec. 2021, doi: and its application to load frequency control of multi-
10.3390/AGRICULTURE12010030. source power system,” International Journal of
[38] J. Hernandez-Barragan, J. D. Rios, A. Y. Alanis, C. Electrical Power & Energy Systems, vol. 54, pp. 77–
Lopez-Franco, J. Gomez-Avila, and N. Arana-Daniel, 85, Jan. 2014, doi: 10.1016/J.IJEPES.2013.06.029.
“Adaptive Single Neuron Anti-Windup PID [48] A. K. Barik and D. C. Das, “Expeditious frequency
Controller Based on the Extended Kalman Filter control of solar photovoltaic/biogas/biodiesel
Algorithm,” Electronics 2020, Vol. 9, Page 636, vol. generator based isolated renewable microgrid using
9, no. 4, p. 636, Apr. 2020, doi: grasshopper optimisation algorithm,” IET Renewable
10.3390/ELECTRONICS9040636. Power Generation, vol. 12, no. 14, pp. 1659–1667,
[39] W. Tang, L. Wang, J. Gu, and Y. Gu, “Single Neural Oct. 2018, doi: 10.1049/IET-RPG.2018.5196.
Adaptive PID Control for Small UAV Micro-Turbojet [49] F. Ding et al., “Gain self-adjusting single neuron PID
Engine,” Sensors 2020, Vol. 20, Page 345, vol. 20, no. control method and experiments for longitudinal
2, p. 345, Jan. 2020, doi: 10.3390/S20020345. relative position of harvester and transport vehicle,”
[40] S. Wan, K. Wang, P. Xu, and Y. Huang, “Numerical Comput Electron Agric, vol. 213, p. 108215, Oct.
and experimental verification of the single neural 2023, doi: 10.1016/J.COMPAG.2023.108215.
adaptive PID real-time inverse method for solving [50] D. J. Lee and L. Wang, “Small-signal stability
inverse heat conduction problems,” Int J Heat Mass analysis of an autonomous hybrid renewable energy
Transf, vol. 189, p. 122657, Jun. 2022, doi: power generation/energy storage system part I: Time-
10.1016/J.IJHEATMASSTRANSFER.2022.122657. domain simulations,” IEEE Transactions on Energy

VOLUME XX, 2017 7

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3456586

Conversion, vol. 23, no. 1, pp. 311–320, Mar. 2008, [58] S. Mohamad Ali Tousi, A. Mostafanasab, and M.
doi: 10.1109/TEC.2007.914309. Teshnehlab, “Design of Self Tuning PID Controller
[51] A. K. Barik and D. C. Das, “Active power Based on Competitional PSO,” 2020 4th Conference
management of isolated renewable microgrid on Swarm Intelligence and Evolutionary
generating power from Rooftop solar arrays, sewage Computation, CSIEC 2020, pp. 22–26, Sep. 2020,
waters and solid urban wastes of a smart city using doi: 10.1109/CSIEC49655.2020.9237318.
Salp swarm algorithm,” International Conference on [59] D. Sain and B. M. Mohan, “Modeling, simulation and
Technologies for Smart City Energy Security and experimental realization of a new nonlinear fuzzy PID
Power: Smart Solutions for Smart Cities, ICSESP controller using Center of Gravity defuzzification,”
2018 - Proceedings, vol. 2018-January, pp. 1–6, Jun. ISA Trans, vol. 110, pp. 319–327, Apr. 2021, doi:
2018, doi: 10.1109/ICSESP.2018.8376744. 10.1016/J.ISATRA.2020.10.048.
[52] R. Ben Mansour, M. A. Mateen Khan, F. A. [60] “KRC data:MOST - Appropedia, the sustainability
Alsulaiman, and R. Ben Mansour, “Optimizing the wiki.” Accessed: Feb. 08, 2024. [Online]. Available:
Solar PV Tilt Angle to Maximize the Power Output: https://www.appropedia.org/KRC_data:MOST
A Case Study for Saudi Arabia,” IEEE Access, vol. 9,
pp. 15914–15928, 2021, doi:
10.1109/ACCESS.2021.3052933.
[53] A. Bakeer, G. Magdy, A. Chub, and H. Bevrani, “A
sophisticated modeling approach for photovoltaic
systems in load frequency control,” International
Journal of Electrical Power & Energy Systems, vol.
134, p. 107330, Jan. 2022, doi:
10.1016/J.IJEPES.2021.107330.
[54] M. Rauniyar et al., “Evaluation of Probing Signals for
Implementing Moving Horizon Inertia Estimation in
Microgrids,” 2020 52nd North American Power
Symposium, NAPS 2020, Apr. 2021, doi:
10.1109/NAPS50074.2021.9449811.
[55] S. Yıldız, H. Gunduz, B. Yildirim, and M. T.
Özdemir, “An innovative LFC scheme for multi-area
microgrid incorporating with hydrogen-based
demand response mechanism,” Int J Hydrogen
Energy, vol. 48, no. 99, pp. 39425–39441, Dec. 2023,
doi: 10.1016/J.IJHYDENE.2023.03.278.
[56] Z. Li, P. Xu, H. Huang, Y. Ning, and B. Li, “A novel
variable stiffness actuator based on a rocker-linked
epicyclic gear train,” Mech Mach Theory, vol. 177, p.
105035, Nov. 2022, doi:
10.1016/J.MECHMACHTHEORY.2022.105035.
[57] E. G. Desisa, S. Janbhasha, and G. A. Meru,
“Identifying Learning Standards of Artificial Neural
Networks: A Study,” Smart Innovation, Systems and
Technologies, vol. 248, pp. 225–239, 2022, doi:
10.1007/978-981-16-4177-0_25/FIGURES/3.

Helmy M. El Zoghby was born in Cairo, Egypt in Ayman Safwat Samir was born in Cairo,
1977, He received his B.Sc. degree in Electrical Egypt in 1992. He received his B.S. and M.S.
Engineering in 2000 from Helwan university, degrees in Electrical Power and Machines
faculty of engineering, power department MSc, Engineering from Helwan University, Cairo,
and Ph.D. in control engineering 2007 and 2012 Egypt, in 2016 and 2022, respectively. He is
respectively. He is currently working as an currently working as a lecturer assistant for the
associate professor at faculty of engineering, faculty of engineering at Helwan University. His
Helwan university. His current research interest current research interests include power system
includes smart grids, renewable energy, Electric control, artificial intelligent control techniques,
drives, power electronics and intelligent control smart grids, and power electronics.
engineering. He can be contacted at email:
[email protected].

VOLUME XX, 2017 7

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4
 This article has been accepted for publication in IEEE Access. This is the author's version which has not been fully edited and
content may change prior to final publication. Citation information: DOI 10.1109/ACCESS.2024.3456586

Mahmoud M. Elmesalawy (M’14) was born in

Ahmed F. Bendary is an associate professor in the Cairo, Egypt, in 1981. He received the B.S., M.S.,
Electric Power and Machine Engineering and Ph.D degrees in Electronics and
Department at Helwan University. He received the Communications Engineering from Helwan
B.Sc., M.Sc., and Ph.D. degrees in Electric power University, Cairo, Egypt, in 2002, 2005, and 2010,
and machine, in 2004, 2008, and 2011, respectively. respectively. He is currently working as a
His current research interests are mainly focused professor of Wireless Communications in the
on renewable energy resources and intelligent smart Department of Electronics and Communications
networks. This includes the design of network Engineering and the Dean of Faculty of
controllers as well as energy management strategies. Engineering at Helwan University, Egypt. He led
He also works as a consultant for many Authorities the effort to build the cellular-based
and factories in the field of electric power system planning. His research communication infrastructure of the Egyptian wide area monitoring system
experience is extremely relevant to the project idea, especially in the (EWAMS) that is developed on the Egyptian grid and operated by Helwan
development of training courses and consultation services related to the University. His current research interests are mainly focused on wireless and
technologies of electric vehicles (EVs) and their integration with power IoT communication technologies. This includes Internet of Things (IoT)
system networks. communications, Unmanned Aerial Vehicle (UAV) communications,
Reconfigurable Intelligent Surface (RIS), Smart Grid Communication and
AI-based applications in Communication systems.

Abdelmonem Hazem was born in Giza, Egypt,

in 1999. He received his B.S. degree in Electrical
Power and Machines Engineering from Helwan Ramy S. A. Afia was born in Monofiya, Egypt,
University, Cairo, Egypt, in 2022. He is currently in 1986. He received a B.Sc. and M.Sc. degrees in
pursuing his master’s degree at Helwan University, electrical engineering from Helwan University,
specializing in Renewable Energy, with a specific Cairo, Egypt, in 2008 and 2015, respectively. In
focus on Green Hydrogen. 2022, he obtained a Ph.D. degree in electrical
engineering from Budapest University of
Technology and Economics, Budapest, Hungary.
He has been working as a Teaching Assistant with
the Department of Electrical Power and Machines
Engineering, Faculty of Engineering, Helwan
Haitham S. Ramadan (World’s Top 2% University, Cairo, Egypt since 2010. In March
Scientists’ List – Stanford University 2015, he was promoted as an Assistant Lecturer with the Department of
2020/2021/2022/2023, hindex: 34, 112 Electrical Power and Machines Engineering.
Publications, citations > 3650) obtained his PhD He is currently working as a research associate at the University of
from the Department of Energy and Automatic Manchester, UK. His research topics are directed to aging mechanisms and
Control of École Supérieure d'Électricité condition monitoring of polymer insulating materials, non-destructive
(SUPELEC), University of South Paris XI, France condition monitoring techniques of cables, partial discharge measurement
in March 2012. Since August 2013, Prof. Dr. with Very Low Frequency (VLF), as well as power and high frequencies,
Ramadan did several postdoctoral research condition monitoring of outdoor insulators, asset condition assessment,
missions in the Laboratory of Signal and Systems renewable energy sources, energy storage and management, and energy
(LSS), the Federation of Research of Fuel Cells green hydrogen systems.
(FCLAB), the FEMTO-ST laboratory and the University of Belfort
Montbéliard in France. In 2020, Prof. Dr. Ramadan joined the International
Institute ISTHY, l’Institut international sur le Stockage de l’Hydrogène, the
first French laboratory for testing and certifications of Type IV tanks as
Research and Innovation Chair. Since June 2022, Prof. Dr. Ramadan is
Professor of Control of Power Systems, Zagazig University, Egypt. He is
the author of more than 112 high-ranked journal and conference papers and
one book chapter. He was/is the Principal Investigator and coordinator of
different Egyptian-French projects and international collaborations; and
summer schools. Since 2017, Dr. Ramadan has been the Co-chair of the
annual International Conference of Emerging and Renewable Energy:
Generation and Application (ICEREGA): ICEREGA’20, ICEREGA’19,
ICEREGA’18 and ICEREGA’17, the General Secretary of ICREGA16, the
International Program Committee member of ICAFE’17. He is/was the
Guest and Managing Editor for different Elsevier journals, His fields of
interest include hydrogen reservoirs and hydrogen storage systems, power
systems control and optimization, multi-physical modelling of energy
systems, renewable energy (Solar and Wind), hybrid power systems,
hydrogen economy, hydrogen technologies, Fuel Cells and Batteries,
Energy Management topics, electric and hybrid electric vehicles,
microgrids, smart grids and HVDC.

8 VOLUME XX, 2017

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. For more information, see https://creativecommons.org/licenses/by-nc-nd/4