Article

Hyper-FDB-INFO Algorithm for Optimal Placement and Sizing

of FACTS Devices in Wind Power-Integrated Optimal Power
Flow Problem
Bekir Emre Altun 1 , Enes Kaymaz 2 , Mustafa Dursun 2 and Ugur Guvenc 2, *

1 Department of Electricity and Energy, Technical Sciences Vocational School, Amasya University,
Amasya 05100, Türkiye; [email protected]
2 Electrical and Electronics Engineering, Engineering Faculty, Duzce University, Duzce 81620, Türkiye;
[email protected] (E.K.); [email protected] (M.D.)
* Correspondence: [email protected]

Abstract: In this study, firstly, the balance between the exploration and exploitation capabilities of
the weighted mean of vectors (INFO) algorithm was developed using the fitness–distance balance
(FDB) method. Then, the FDB-INFO algorithm was developed with a hyper-heuristic method to
create the beginning optimal population by using Linear Population Reduction Success History-based
Adaptive Differential Evolution (LSHADE) and a novel Hyper-FDB-INFO algorithm was presented.
Finally, the developed Hyper-FDB-INFO algorithm was applied to solve the optimal placement
and sizing of FACTS devices for the optimal power flow (OPF) problem incorporating wind energy
sources. Moreover, determining the placement and sizing of FACTS devices is an additional problem
to minimize the total cost of generation and reducing the power losses of the power system. The
experimental results showed that the Hyper-FDB-INFO algorithm is a more effective solver than
the SHADE-SF, INFO, FDB-INFO and Hyper-INFO algorithms for wind power and FACTS devices
integrating the OPF problem.

Keywords: hyper-heuristic; fitness–distance balance; info; optimal power flow; facts; wind energy

Citation: Altun, B.E.; Kaymaz, E.;

Dursun, M.; Guvenc, U. Hyper-FDB-
INFO Algorithm for Optimal 1. Introduction
Placement and Sizing of FACTS
Today, electrical power systems form the basis of modern life and continue to shape the
Devices in Wind Power-Integrated
future of the energy sector. Especially in modernizing power systems due to technological
Optimal Power Flow Problem.
innovations and globally expanding economies, OPF is one of the main tools of planning
Energies 2024, 17, 6087. https://
doi.org/10.3390/en17236087
and management technologies for power grids that provide high-quality electrical power at
affordable prices. In other words, OPF constitutes the backbone of the planning and analysis
Academic Editor: Frede Blaabjerg of power systems as it increases efficiency and provides economic benefits. OPF is a non-
Received: 5 November 2024
linear, non-convex, constrained, and large-scale optimization problem. It aims to minimize
Revised: 27 November 2024 various objective functions such as generation cost, emission, active power transmission
Accepted: 29 November 2024 loss, voltage deviation, and voltage stability index improvement by considering various
Published: 3 December 2024 equality and inequality constraints and to obtain the optimal control variables of the power
system [1,2]. These control variables may comprise generated active or reactive power,
generator bus voltages, or transformer tap ratios. On the other hand, state variables may
comprise the voltage of the load bus and the reactive power output of the generators [3].
Copyright: © 2024 by the authors. In the recent past, classical OPF analyses were performed primarily for power plants
Licensee MDPI, Basel, Switzerland.
powered by fossil fuels using traditional mathematical methods and metaheuristic search
This article is an open access article
algorithms (MSAs) [4–13]. After that, with the liberalization of energy markets, renewable
distributed under the terms and
energy sources (RESs) have been successfully integrated into power systems, bringing
conditions of the Creative Commons
about significant benefits. These include a reduction in active power losses, an increase
Attribution (CC BY) license (https://
in the reliability and power quality of electrical networks, and a noticeable decrease in
creativecommons.org/licenses/by/
4.0/).
environmental pollution. However, the use of functions that account for the stochastic

Energies 2024, 17, 6087. https://doi.org/10.3390/en17236087 https://www.mdpi.com/journal/energies

Energies 2024, 17, 6087 2 of 25

nature of RESs, such as wind and solar energy, has added complexity to the classical OPF
problem, a key technical challenge in this context. In response, MHS algorithms have
become a crucial tool in overcoming this complexity, thanks to their flexibility and powerful
search capabilities. For instance, OPF results were improved by Roy and Jadhav [14]
compared to the well-established methods used in the past by using the Gbest-guided
artificial bee colony (GABC) algorithm in the IEEE 30-bus test system consisting of thermal
and wind generators. Later on, Mishra et al. [15] presented the modified Cuckoo Search
(CS) algorithm to solve the OPF problem in which wind power costs are included. In [16],
the success history-based adaptation technique of differential evolution with superiority of
feasible solutions (SHADE-SF) algorithm was used by Biswas et al. to solve the optimal
power flow by combining stochastic wind and solar power with conventional thermal
power generators in the system. In the study, Lognormal and Weibull probability density
functions (PDFs) were used to model solar irradiation and wind speed uncertainties. Then,
the OPF problem was solved by Reddy [17] through a genetic algorithm (GA) in the IEEE
30-bus test system, including wind, solar, and energy storage systems. In [18], Khan et al.
focused on a gray wolf optimizer (GWO) algorithm to handle the OPF problem in IEEE 30-
and IEEE 57-bus power systems incorporating renewable energy. In [19], Nusair and Alasali
used the golden ratio optimization method (GROM) for objective functions such as total
cost, emission and power loss minimization, and voltage stability improvement in IEEE-30
and 118 bus power systems with and without wind and solar energy. In [20], a barnacles
mating optimizer (BMO) was presented by Sulaiman and Mustaffa to conduct OPF analysis
for stochastic wind–solar PV–small hydro-power integrated IEEE 30-bus and IEEE 57-bus
test systems for different case studies regarding the minimization of generation cost, total
transmission loss, and emission. In [21], a hybrid particle swarm gray wolf optimizer
(HPS-GWO) was proposed by Riaz et al. to solve the OPF problem in the modified IEEE
30-bus test system containing RES by considering three main objective functions, which
are generation cost, emission, and power losses. In [22], the developed Lévy coyote
optimization algorithm (LCOA) by Kaymaz et al. was tested in solving the OPF problem,
where a wide range of objective functions were taken into account, such as fuel cost
minimization, emission reduction, voltage stability improvement, voltage deviation, and
active power loss minimization for IEEE 30, 57, and 118-bus test systems. In [23], a fitness–
distance balance-based adaptive guided differential evolution (FDB-AGDE) algorithm was
proposed by Guvenc et al. for solving the security-constrained OPF problem with wind and
solar power in the IEEE 30-bus test system. In [24], Souza et al. proposed a gradient-based
optimizer (GBO) to study the impact of wind power on the economic operation of the
electrical system, bus voltages, and transmission power loss. In [25], Farhat et al. proposed
an improved version of the marine predator algorithm in the OPF analysis performed
considering the uncertainty of RESs. In [26], Alghamdi proposed the hybrid Firefly–Jaya
(HFAJAYA) algorithm to minimize the total generation cost and emissions in solving the
OPF problem with and without the inclusion of RES in a power system with IEEE 30 bus.
In [27], Shaneen et al. proposed a novel circle search algorithm (CSA) to minimize the total
generation cost in the RES-integrated OPF analysis conducted on the IEEE 57 and 118 bus
test systems. Due to the stochastic nature of RES in the study, Beta and Weibull PDFs were
used to model solar irradiation and wind speed. In [28], Mouassa et al. proposed the Slime
Mold-inspired Algorithm (SMA) to reduce the overall operating cost of the main grid and
emissions in the solar PV and wind power-integrated OPF analysis performed on the IEEE-
30 bus test system and the Algerian power system DZA-114 bus. In [29], Adhikari et al.
proposed an adaptive lightning assignment procedure optimizer (ALAPO) to solve the
OPF problem in wind and solar-integrated IEEE 57 bus power system in order to minimize
voltage deviation and power loss and improve voltage stability. In [30], Maheshwari et al.
proposed a new flow direction algorithm (FDA) to solve the RES-integrated OPF problem
consisting of solar PV, wind, and small hydropower generators. In [31], Hasanien et al.
proposed the enhanced coati optimization algorithm (ECOA) in the analysis carried out for
IEEE 57 and 118 bus test systems in the presence and absence of wind and solar PV energy
Energies 2024, 17, 6087 3 of 25

in the OPF problem, where reducing the total generation cost is the main objective. In [32],
a new hybrid ACGO algorithm was introduced by combining Chaos Game Optimization
(CGO) with the Artificial Ecosystem-based Optimization (AEO) method. With the proposed
algorithm, the authors focused on minimizing both fuel costs and fuel costs with total
emissions in modified IEEE 30 and IEEE 57 bus test systems connected to RES. In [33], the
Improved Turbulent Flow of Water-Based Optimization (ITFWO) algorithm was introduced
by the authors to minimize the total cost in an IEEE 30-bus power system containing a PV
generator and wind turbine, taking into account emissions, losses, and the valve-point effect.
In [34], the WHO (Wild Horse Optimizer) and EESWHO (elite evolutionary strategy based
on Wild Horse Optimizer) algorithms are presented to reach the best solution for the OPF
by considering the uncertainty modeling of RESs. In [35], an enhanced Growth Optimizer
algorithm with a dynamic fitness–distance balance (dfdb-GO) method is proposed for the
security constraint OPF problem performed in the wind and solar-integrated IEEE 30 and
IEEE 57 bus test systems for a total of 12 different case studies considering various objective
functions such as total cost with valve point effect and prohibited operating zone and total
cost with multi-fuels. In [36], a self-adaptive wild goose algorithm (SAWGA) was used
to optimize an OPF model in different IEEE 30-bus and 118-bus power grids containing
traditional thermal power units, solar PV, and wind power units. In the study, the authors
emphasized that SAWGA provides faster and more effective convergence in reducing
overall fuel consumption costs. In the study presented in [37], an adaptive method-based
MODE algorithm was presented, and OPF analysis was conducted on modified IEEE 30-bus
and IEEE 118-bus systems with intermittent wind and solar power integration to verify
the effectiveness of the proposed approach in real-world scenarios. In [38], Hassan et al.
proposed a modified version of Manta Ray Foraging Optimization based on the Artificial
Hummingbird Algorithm (AHAMRFO) to solve the OPF problem, including wind turbine,
PV, and plug-in electric vehicle (PEV)-integrated IEEE 30 bus power system to minimize
generation cost and emissions.
Although RESs are effective in increasing power quality and system reliability and
provide economic or environmental advantages, the operation of electrical power networks
may require reactive power. A reactive power shortage in a system can cause serious
fault situations, such as voltage fluctuations, instability, and collapse [39]. In the past, tap
changer transformers or capacitor banks were used to address these situations. However,
these devices were not responsive enough to handle unexpected changes in the system.
As a result, power electronics-based FACTS compensators have been preferred more
frequently in recent times. In addition to improving the flexibility of power transmission
and enhancing the dynamic stability of the power system, FACTS devices offer several
benefits, such as minimizing active power losses, meeting reactive power requirements,
reducing operating costs, and regular power flow analysis [40]. However, just like the
inclusion of RESs, using FACTS devices makes obtaining the optimum solution more
difficult than solving the classical OPF problem due to the added mathematical calculations.
At this point, the integrated OPF problem that includes RES and FACTS devices emerges
as a crucial optimization problem. In the study presented by Panda and Tripathy [41], the
security-constrained OPF problem of the wind–thermal generation system was solved,
again using a modified bacteria foraging algorithm (MBFA). In the study, it was seen
that the static synchronous compensator (STATCOM) provided reactive power support
against wind uncertainties. In [42], Elmitwally and Eladl proposed a hybrid method based
on Particle Swarm and Sequential Quadratic Programming (HPS-SQP) to solve the OPF
problem where annual net cost minimization is determined as the objective function. In
the study, OPF analysis was performed on wind power, static VAR compensator (SVC),
thyristor-controlled series capacitor (TCSC), and Unified Power Flow Controller (UPFC)-
integrated IEEE 14-bus and 118-bus test systems. In [43], a modified hybrid particle
swarm optimization and gravitational search algorithm with chaotic maps (CPSOGSA)
method was proposed by Duman et al. to solve the security-constrained OPF problem
of power systems with stochastic wind energy and FACTS devices such as TCSC and
Energies 2024, 17, 6087 4 of 25

thyristor-controlled phase shifter (TCPS). In the study, the performance of the proposed
algorithm was evaluated according to the results of statistical analysis, taking into account
objective functions such as minimization of cost, emission and active power losses, and
improvement of voltage stability in IEEE 30 bus and 57 bus test systems. In [44], the
SHADE-SF method was proposed by Biswas et al. to solve the OPF in which the generation
cost is optimized by incorporating stochastic wind power and various FACTS devices
such as SVC, TCSC, and TCPS. In the study, while creating the objective function, the
cost of thermal generation, the direct cost of scheduled wind power, the penalty cost
for underestimation, and the reserve cost in case of overestimation of wind power were
taken into account. In the study conducted by Nusair et al. [45], the single- and multi-
objective OPF problem was created to minimize generation cost, power losses, and voltage
deviation in an IEEE 30-bus power system integrated with wind and solar energies and
FACTS devices such as SVC, TCSC, and TCPS. The problem was solved with the Slime
Mold Algorithm (SMA), Artificial Ecosystem-based Optimization (AEO), Marine Predators
Algorithm (MPA), and Jellyfish Search (JS) algorithms. In the study by Mohamed et al. [46],
the optimal location and size of FACTS devices such as TCSC, TPSC, and SVC, taking
into account OPF, are presented in a hybrid power system containing stochastic wind and
conventional thermal power plants. In the study, the optimal size and location of FACTS
devices were determined by creating a multi-objective function that includes active power
losses as well as reserve costs for overestimation and penalty costs for underestimation
of intermittent renewable resources. A hybrid method consisting of a gradient-based
optimizer (GBO) and moth flame optimization algorithm (MFO) has been proposed to
minimize this objective function. In [47], a modified version of the Runge Kutta optimizer
(MRUN) based on Cauchy mutation and quasi-oppositional learning techniques was
presented by Ebeed et al. to solve the stochastic OPF problem with optimal integration of
wind turbines and solar PV systems along with TCSC in an IEEE 57 bus system. In [40],
the Chaotic African Vultures Optimization Algorithm (CAVOA) has been proposed for the
analysis carried out to reduce the overall power cost and power loss in the IEEE 30-bus
power system in solving the OPF problem integrated with wind power, SVC, TCSC, and
TCPS. In the study, penalty cost and reserve cost were included in the objective function.
In the study presented in [48], the OPF problem involving renewable energy sources and
FACTS was solved using the Chaos Game Optimization (CGO) algorithm for various
single-objective and multi-objective functions. In the study, wind turbines and PV units
were used to produce renewable energy sources, while FACTS devices preferred SVC,
TCSC, and TCPS.
As seen in the studies mentioned, many different algorithms have been proposed to
solve the optimal power flow problem integrated with RES and FACTS devices. However,
especially based on the “No free launch” theorem [49], it is thought that any optimization
algorithm is insufficient to solve all types of optimization problems. Based on this, re-
searchers continue to design new optimization frameworks or improve existing algorithms
day by day. One of the metaheuristic search algorithms presented recently is the INFO op-
timization algorithm. The INFO optimization algorithm is based on the idea of a weighted
average for a set of vectors proposed by Ahmadianfar et al. [50]. The INFO algorithm
performs the iteration process using three stages: rule update, vector combination, and local
search. The INFO algorithm has shown competitive performance in various benchmark
systems, engineering design problems, and various power system problems [51–54]. Due
to its successful performance, in this study, INFO is preferred for solving the optimal power
flow problem integrated with wind power and FACTS in power systems. However, it has
been observed that INFO presents some disadvantages for this real-world engineering
problem where high dimensions and many parameters are optimized. Especially in the
update phase of the algorithm, an early convergence problem was observed due to the
random selection of individuals in INFO’s search mechanism, which suggested that the
algorithm may be inadequate in terms of exploration strategy. At the same time, the al-
gorithm becoming stuck at local optimum points during the search process also revealed
Energies 2024, 17, 6087 5 of 25

deficiencies in terms of exploitation strategy. As a result, a novel FDB-INFO algorithm is

presented using the FDB selection method [55] to overcome the early convergence problem
of INFO and achieve an effective balance between exploration and exploitation.
Metaheuristic algorithms can provide effective results in solving any optimization
problem. However, if the parameters in the operation of a metaheuristic algorithm are
not set to optimal values or if the proper method is not used in the improvement of
the algorithm, the performance of the algorithm decreases. This problem is one of the
disadvantages of metaheuristic algorithms. In the literature, a hyper-heuristic approach
has been proposed to eliminate this disadvantage. The hyper-heuristic approach basically
consists of two levels. The first one is a low-level structure, a class in which low-level
heuristics (such as the parameters required for the operation of the algorithm) are located.
The other one is a high-level structure, an interface in which the most suitable low-level
heuristic is selected [56,57].
In this study, chaotic maps (CMs), opposition-based learning (OBL) methods, and the
population ratio are used in the low-level structure, and the LSHADE [58] algorithm is used
as a high-level structure in the proposed Hyper-FDB-INFO algorithm. While designing
the proposed algorithm, the LSHADE algorithm creates a random population among the
chaotic maps, OBL methods, and population ratio in the low-level heuristic structures.
Each individual in this created population (low-level heuristic) contains the CMs, OBL
methods, and population ratio combinations used to produce the initial population of
the INFO algorithm. The fitness value obtained as a result of the iterative process of the
INFO algorithm according to the initial population is accepted as the fitness value of the
low-level heuristic structure. After the iterative processes of the LSHADE algorithm, the
most appropriate low-level selection process is performed. As a result of this, the Hyper-
FDB-INFO algorithm was introduced. The proposed algorithm was used to solve the
optimal power flow problem with integrated wind power and FACTS in power systems.
To solve the problem, the results obtained with Hyper-FDB-INFO were compared with
SHADE-SF [44], INFO, FDB-INFO, and Hyper-INFO algorithms.
Accordingly, the main contributions of this study can be briefly summarized as follows:
• A new hyper-heuristic approach, called Hyper-INFO, is proposed to find the initial
population of INFO by obtaining the best combination of chaotic maps, OBL methods,
and population ratio using the LSHADE algorithm. In order to improve the low-level
heuristic structure of the proposed hyper-heuristic approach, the exploration and
exploitation capability of the INFO algorithm in the structure is improved by using
the FDB method. As a result of this, another novel Hyper-FDB-INFO algorithm is
presented to the literature.
• The proposed Hyper-FDB-INFO algorithm was used to solve the optimal power
flow problem integrated with wind power and FACTS in the power system. In
addition to the recommended Hyper-FDB-INFO, INFO, FDB-INFO, and Hyper-INFO
algorithms were used for the first time to solve the problem. The results obtained
with Hyper-FDB-INFO were compared with SHADE-SF [44], INFO, FDB-INFO, and
Hyper-INFO algorithms.
• The optimal power flow problem integrated with wind power is solved considering
various FACTS devices such as SVC, TCSC, and TCPS in an electrical power system. In
the study, for three different cases at constant load, the location of the FACTS devices
and their ratings were optimized by including the uncertain cost of wind power. In
addition, for the fourth case study, optimization was performed for four different
loading conditions for the fourth case study.
Accordingly, the remaining sections of the paper are as follows: Section 2 describes
the cost models of thermal and wind power for optimal power flow. Section 3 introduces
the proposed Hyper-FDB-INFO algorithm. In Section 4, the results obtained with Hyper-
FDB-INFO and competing algorithms for solving the problem are compared. Finally, the
conclusions drawn from the study were evaluated and discussed.
 Energies 2024, 17, x FOR PEER REVIEW 6 of 26

Energies 2024, 17, 6087

the proposed Hyper-FDB-INFO algorithm. In Section 4, the results obtained with Hyper- 6 of 25
FDB-INFO and competing algorithms for solving the problem are compared. Finally, the
conclusions drawn from the study were evaluated and discussed.
2. Materials and Methods
2. Materials and Methods
In this study, the OPF problem is solved over the IEEE-30 bus test system modified
In this et
by Biswas study, the OPF
al. [44]. Theproblem
modified is solved over theincludes
test system IEEE-30 4bus test system
thermal modified2 wind
generators,
by Biswas et al. [44]. The modified test system includes 4 thermal generators,
generators, and 6 FACTS devices. There are two of each of the FACTS devices consisting 2 wind gen- of
erators, and 6 FACTS devices. There are two of each of the FACTS
a thyristor-controlled series compensator (TCSC), thyristor-controlled phase shifter devices consisting of a(TCPS),
thyristor-controlled series compensator (TCSC), thyristor-controlled phase shifter (TCPS),
and static VAR compensator (SVC), and their mathematical models are described in ref. [44].
and static VAR compensator (SVC), and their mathematical models are described in ref.
TCSC and TCPS are the series compensation devices that are used to enhance the loading
[44]. TCSC and TCPS are the series compensation devices that are used to enhance the
capability and power flow of the line. A shunt compensation device SVC supplements
loading capability and power flow of the line. A shunt compensation device SVC supple-
the reactive power of the system. The determination of optimal placement and sizing
ments the reactive power of the system. The determination of optimal placement and siz-
of the FACTS
ing of the FACTS devices
devicesare
arethe
the main issuesofofthe
main issues the
OPFOPF problem.
problem. The addition
The addition of FACTS
of FACTS
devices to the system and the location and rating of FACTS devices
devices to the system and the location and rating of FACTS devices decrease the total gen- decrease the total
generation cost, reduce power losses, and increase the reliability of the
eration cost, reduce power losses, and increase the reliability of the electrical network’s electrical network’s
operating conditions.
operating conditions.
Figure
Figure 11 shows
showsthe themodified
modifiedIEEE IEEE30-bus
30-bus test
test system.
system. In this
In this study,
study, thermal
thermal generators
genera-
were placed
tors were in buses
placed 1, 2, 8,
in buses 1, and
2, 8, 13 of 13
and theofIEEE 30-bus
the IEEE power
30-bus system,
power whilewhile
system, windwind
generators
generators
were were integrated
integrated into busesinto buses
5 and 11.5 The
and 11. The system
system has fourhastransformers
four transformers between
between the buses
the buses 6-9, 6-10, 4-12, and 27-28. Two SVCs from FACTS devices
6-9, 6-10, 4-12, and 27-28. Two SVCs from FACTS devices were placed between buses 21 were placed between
buses
and 24,21two
andTCSCs
24, twowere
TCSCs werebetween
placed placed between
buses 2-5 buses
and2-5 and and
24-25, 24-25,
twoand two TCPSs
TCPSs were placed
were placed between
between buses 3-4 and 9-10. buses 3-4 and 9-10.

29
27 28
25

30 26

23 24

15 19

18

17 20
G
21
13 12 16

14 9

11 22
G Thermal Units

Transformers 10
1

Wind Generator 3 4 8
G
6
G
TCSC
2 5
7
TCPS
G

SVC

Figure 1.
Figure 1. Modified
ModifiedIEEE
IEEE30-bus system
30-bus incorporating
system windwind
incorporating generators and FACTS
generators devicesdevices
and FACTS [44]. [44].

Generally, the cost function of thermal units (C Ti ) by considering the valve-point

loading effect using fossil fuels in OPF problems is defined as
 
2
CTi ( PTGi ) = ai + bi PTGi + ci PTGi + di x sin ei x ( Pmin
TGi − PTGi ) (1)

ai , bi , ci , di and, ei are the cost coefficients, PTGi is the producing power output, and
min is for the ith thermal unit’s minimum power output.
PTGi
Energies 2024, 17, 6087 7 of 25

Wind power generators do not require any fuel, and by using the direct cost coefficient
( gwj ), the cost function of jth wind generators (Cwj ) is identified as a function of the plant’s


scheduled power Pwsj :

Cwj Pwsj = gwj Pwsj (2)
Due to the uncertainty of wind energy, reserve

and penalty costs need to be added to
the wind power cost. The reserve cost CRwj and the penalty cost (CPwj ) component for
underestimation of wind power are defined in Equation (3) and Equation (4), respectively.



CRwj Pwsj − Pwavj = K Rwj Pwsj − Pwavj
RP (3)
= K Rwj 0 wsj Pwsj − Pw f wj ( Pw )dPw

CPwj Pwavj − Pwsj = K Pwj Pwavj − Pwsj
RP (4)
= K Pwj P wrj Pw − Pwsj f wj ( Pw )dPw


wsj

K Rwj and K Pwj represent the coefficient of reserve and penalty cost for the jth wind
power plant. Here, the actual power available from the plant is Pwavj . f wj signifies the
probability density function of the plant and is given in Equation (5) [44]:

v b−1 −(v/a)b
  
b
f wj (v) = e f or 0 < v < ∞ (5)
a a

The cost coefficients of all thermal units are provided in Table 1, and the cost coeffi-
cients and probability density function parameters of all wind generators are provided in
Table 2.

Table 1. Cost coefficients of thermal power units.

Generator Bus a (USD/h) b (USD/MWh) c (USD/MW2 h) d (USD/h) e (rad/MW)

TG1 1 0 2 0.00375 18 0.037
TG2 2 0 1.75 0.0175 16 0.038
TG8 8 0 3.25 0.00834 12 0.045
TG13 13 0 3 0.025 13.5 0.041

Table 2. PDF parameters and cost coefficients for wind power-generating plants.

No. of Rated Power, Weibull PDF Cost Coefficients (USD/MWh)

Windfarm
Turbines Pwr (MW) Parameters Direct, gw Reserve, KRw Penalty, KPw
WG5 (bus 5) 25 75 a = 9, b = 2 1.60 3.0 1.50
WG11 (bus 11) 20 60 a = 10, b = 2 1.75 3.0 1.50

Finally, the total generation cost of the modified system is defined in Equation (6).

NTG NWG 
∑ CTi ( PTGi ) + ∑




Cgen = Cwj Pwsj + CRwj Pwsj − Pwavj + CPwj Pwavj − Pwsj (6)
i =1 j =1

Researchers generally work to minimize the total cost in OPF problems. However, due
to the inherent resistance of the transmission systems, real power losses occur in the system.
In this study, power transmission losses are also taken into account. The mathematical
model of the real power loss due to inherent resistance is given in Equation (7).
h i
nl
Ploss = ∑q=1 Gq(mn) Vm2 + Vn2 − 2Vm Vn cos(δm − δn ) (7)
Energies 2024, 17, 6087 8 of 25

where Vm is the voltage magnitude at bus m, Gq(mn) is the conductance of line m-n, δm , δn
are the voltage angles at bus m, n, and nl is the number of transmission lines.
Also, when solving the OPF problem on a transmission line, we should consider the
system constraints, including the equality and inequality constraints. The power balance
equations in the presence of the FACTS devices are defined in Equations (8) and (9) [44].
NB
PGm + Pms − PDm − Vm ∑n=1 Vn Ymn cos(θmn + δm − δn ) = 0 ∀m ∈ NB (8)

NB
QGm + Qms + QSVCm − Q Dm − Vm ∑ Vn Ymn sin(θmn + δm − δn ) = 0 ∀m ∈ NB (9)
n =1

where PGm is the real power generation at bus m, PDm is the real power demand at bus
m, QGm is the reactive power generation at bus m, Q Dm is the reactive power demand at
bus m, Pms is the real power generation, injected by the TCPS at bus m, Qms is the reactive
power generation, injected by the TCPS at bus m, QSVCm is the injected reactive power at
bus m by the SVC, Ymn is the magnitude of bus admittance element m,n, θmn is the angle of
bus admittance element m,n, and NB is the total number of buses.
There are three inequality constraints in the OPF problem: Generator constraints,
Security constraints, and Transformer constraints. Generator constraints are defined in
Equations (10)–(12).
Pmin
Gi ≤ PGi ≤ P Gi
max
∀i ∈ NG (10)
Qmin max
Gi ≤ Q Gi ≤ Q Gi ∀i ∈ NG (11)
V min
Gi ≤ VGi ≤ V max
Gi ∀i ∈ NG (12)
Security constraints are defined in Equations (13) and (14).

V min max
Lp ≤ VLp ≤ V Lp ∀ p ∈ NL (13)

S Lq ≤ Smax
Lq ∀q ∈ nL (14)
Transformer constraints are defined in Equations (15)–(18).

T min
t ≤ Tt ≤ T max
t ∀t ∈ NT (15)
min max
TCSC : τTCSCm ≤ τTCSCm ≤ τTCSCm ∀m ∈ NTCSC (16)
min max
TCPS : ΦTCPSn ≤ ΦTCPSn ≤ ΦTCPSn ∀n ∈ NTCPS (17)
SVC : Qmin max
SVCj ≤ QSVCj ≤ QSVCj ∀ j ∈ NSVC (18)

3. Proposed Algorithm (Hyper-FDB-INFO)

This paper improves the performance of an innovative optimizer named weIghted
meaN oF vectOrs (INFO) [50]. Generally, The INFO algorithm has good global exploration
capability to find the global optimum solution. The INFO algorithm works by calculating
the average weights of the vector and consists of a population vector sequence. The most
suitable solution calls on a variety of consecutive objects. Each generation has three vectors
updating their position. These are Stage 1: updating rule, Stage 2: vector combination, and
Stage 3: local search [50]. The role of the updating rule in the INFO algorithm is to increase
the population’s diversity. This updating stage is the biggest difference between the INFO
algorithm and other meta-heuristic search algorithms. In this stage, new vectors are created
using the weighted mean of vectors. The new vector is calculated in this stage by using
Equation (19):
g g
z1 = x1 + σ × MeanRule + CA (19)
In the INFO algorithm, the mean rule is used to increase the diversity of the population.
The mean rule is expressed mathematically in the following equations.
Energies 2024, 17, 6087 9 of 25

g g
Meanrule = r × W M1l + (1 − r ) × W M2l l = 1, 2, . . . , Np (20)
g
W M1l is calculated using Equation (21):

g w1 ( x a1 − x a2 ) + w2 ( x a1 − x a3 ) + w3 ( x a2 − x a3 )
W M1l = δ × l = 1, 2, . . . , Np (21)
w1 + w 1+ w1 + ε

The index numbers of a1 ̸= a2 ̸= a3 ̸= 1 are different integers randomly selected from

the range [1, NP ]. Randomly selected a1 , a2 and a3 values were used to increase diversity in
the updating stage. However, the exploration ability of the INFO algorithm in the search
process has still some disadvantages. Kahraman et al. [55] proposed an effective method
called the fitness–distance balance (FDB) method, which provides effective guidance on
meta-heuristic search algorithms in the search process. In FDB, the score vector (SP ) of the
population is given in Equation (22).
 
S1
SP ≡  ...  (22)
 

Sn n ×1

Here, S1 = Sfdb , the solution candidate, new vector, is selected by the FDB method. Sfdb
is used instead of the a1 . Sfdb is the solution candidate with the maximum value in the score
vector of FDB. In this study, firstly, the INFO algorithm is enhanced by using FDB. The
proposed algorithm, which is called FDB-INFO, has the ability to improve the exploration
and local optima avoidance of INFO.
In metaheuristic search algorithms, initial populations are generated randomly, and
this process creates some disadvantages. A generated population may be far from the global
solution and may contain weak diversity solution candidates. In this study, a different
optimization algorithm is considered to generate the most suitable initial population to be
used in INFO and FDB-INFO algorithms. For this purpose, the LSHADE metaheuristic op-
timization algorithm, which has superior performance and a simple structure, was selected.
In other words, a hyper-heuristic method is proposed using the LSHADE algorithm to find
an optimum initial population for INFO and INFO-FDB algorithms.
The LSHADE optimization algorithm has Initialization, Mutation, External Archive,
Parameter Adaption, Crossover, Selection, and Linear population size reduction pro-
cesses [58]. The LSHADE optimization algorithm’s first step is to randomly generate
an initial population within the feasible bounds of the decision variables, which can be
written as follows:

Xi,j = L j + rand × Uj − L j i = 1, 2, . . . ., N (23)
where N is the population size, and Lj and Uj are the lower and upper boundary of the
jth dimension. Mirjalili et al. [56] proposed a hyper-heuristic algorithm. In their study,
there are three variables: the first variable consists of CMs (1-10), the second variable
consists of the OBL method (1-4), and the third variable is the ratio of the population (0-1)
in the LSHADE algorithm. Chaotic maps (CMs) are used to improve the performance
of algorithms and provide systematic stochastic behaviors. The use of CMs provides an
effective ability to search for the best solution by avoiding the problem of stagnation at the
local point and can increase the convergence towards the global solution. CMs are generally
used in the initialization phase of the population in optimization problems. In addition to
the CMs strategy, another strategy is the opposition-based learning (OBL) strategy. The
main purpose of the OBL strategy is to calculate the opposite local solution for each solution
in the initial phase and optimization process in order to improve the exploration ability of
the algorithm. More detailed information about CMs and OBL methods can be found in
Abd Elaziz and Mirjalili’s paper [56].
The LSHADE algorithm first creates a population randomly based on three variables
and the population size. A candidate solution in the LSHADE population represents
Energies 2024, 17, 6087 10 of 25

the initial population of the INFO/FDB-INFO algorithm. Then, the INFO/FDB-INFO

algorithm runs for 50 iterations based on this population. The best fitness value obtained
here becomes the fitness value of the corresponding candidate solution of LSHADE. In other
words, the fitness value of LSHADE’s candidate solution is obtained from the execution of
INFO/FDB-INFO. After LSHADE completes all its iterations, the best candidate solution is
determined. This stage is called the training stage. The best initial value for INFO/FDB-
INFO is now set as the best candidate solution of LSHADE. Based on this initial population
obtained, the INFO/FDB-INFO algorithm is applied to the problem with the maximum
iteration. This stage is called the test stage. In the proposed algorithm, if INFO is used,
its name is defined as Hyper-INFO, and if FDB-INFO is used, its name is defined as
Hyper-FDB-INFO. The proposed algorithm, called Hyper-FDB-INFO, has a better ability to
Energies 2024, 17, x FOR PEER REVIEWimprove the exploration and local optima avoidance of INFO, FDB-INFO, 11 ofand
26 Hyper-INFO.

The framework of the proposed hyper heuristic algorithms is given in Figure 2.

Figure 2. Framework of the proposed algorithm (Hyper-INFO/Hyper-FDB-INFO).

Figure 2. Framework of the proposed algorithm (Hyper-INFO/Hyper-FDB-INFO).
4. 4.
Simulation Results
Simulation Results
In this section, the simulation results obtained with Hyper-FDB-INFO for different
In this section, the simulation results obtained with Hyper-FDB-INFO for different
case studies in the OPF problem are compared with SHADE-SF [37], INFO, INFO-FDB,
case studies in the OPF problem are compared with SHADE-SF [37], INFO, INFO-FDB, and
and Hyper-INFO algorithms and presented in tables. The case descriptions of the simula-
Hyper-INFO
tion algorithms
study are given and
in Table 3 as presented in tables. The case descriptions of the simulation
a summary.
study are given in Table 3 as a summary.
Table 3. Summary of case studies.

Case No Case Description Ref. Equation

Case 1 Minimize cost Equation (6)
Case 2 Minimize power loss Equation (7)
Case 3 Minimize cost and power loss Equation (24)
Energies 2024, 17, 6087 11 of 25

Table 3. Summary of case studies.

Case No. Case Description Ref. Equation

Case 1 Minimize cost Equation (6)
Case 2 Minimize power loss Equation (7)
Case 3 Minimize cost and power loss Equation (24)
Case 4 Minimize cost and power loss with uncertain load demand Equation (24)

There are a total of 27 control variables in the OPF analysis performed. The locations of
FACTS devices are also included in these decision variables. While the locations of FACTS
devices are expressed with branch numbers for TCSC and TCPS, they are expressed with
the bus numbers connected to SVC. Each FACTS device is assigned two control variables,
one representing the location and the other representing the degree of the device. When
integrating FACTS devices into the system, the following points should be considered.
Two FACTS devices cannot be on the same bus, SVC cannot be installed on the bus where
generators are located since reactive power is exchanged, and TCSC and TCPS cannot be
placed on branches with tap changer transformers. The step settings of transformers were
selected between 0.90 and 1.10 p.u. in case studies. The maximum compensation value on
the bus located at TCSC is determined as 50% of the line reactance. In addition, the reactive
power absorption and generation capacity of SVC is up to 10 MVAr. The phase shifter
(TCPS) angle varies between −5 degrees and 5 degrees. The reactive power capacities of
wind generators are determined between −0.4 pu and 0.5 pu [44]. The variable limits used
in the study for each case are given in Tables 4–6.

Table 4. Simulation results of case 1 with fixed loading for the adapted IEEE 30-bus system.

Limits Case 1
Min Max SHADE-SF [44] INFO INFO-FDB Hyper-INFO Hyper-FDB-INFO
PTG2 (MW) 20 80 40.6265 39.8425 39.5558 39.7771 40.6435
PWG5 (MW) 0 75 49.5418 50.3234 50.3841 50.2550 49.6026
PTG8 (MW) 10 35 10.0000 10.0000 10.0000 10.0001 10.0000
PWG11 (MW) 0 60 41.8692 41.9159 42.0934 42.0166 41.7787
PTG13 (MW) 12 40 12.0000 12.0006 12.0000 12.0000 12.0000
V1 (p.u.) 0.95 1.10 1.0752 1.0737 1.0752 1.0727 1.0744
V2 (p.u.) 0.95 1.10 1.0598 1.0589 1.0580 1.0580 1.0596
V5 (p.u.) 0.95 1.10 1.0383 1.0369 1.0396 1.0361 1.0382
V8 (p.u.) 0.95 1.10 1.0372 1.0368 1.0368 1.0348 1.0372
V11 (p.u.) 0.95 1.10 1.0955 1.0829 1.0892 1.0812 1.0924
V13 (p.u.) 0.95 1.10 1.0759 1.0659 1.0732 1.0753 1.0740
Control Variables

T11 (p.u.) 0.90 1.10 1.0200 1.0000 1.0000 0.9800 1.0400

T12 (p.u.) 0.90 1.10 0.9200 0.9800 1.0200 0.9800 0.9200
T15 (p.u.) 0.90 1.10 1.0000 1.0200 1.0000 1.0400 1.0000
T36 (p.u.) 0.90 1.10 0.9600 0.9800 0.9800 0.9800 0.9800
tTCSC1 (%) 0 50% 26.3174 24.9122 13.7618 0.1664 28.5902
tTCSC2 (%) 0 50% 41.3031 21.4662 23.3623 25.9642 29.7363
UTCSC1 (deg.) −5 5 −1.2319 1.0744 2.5641 1.2460 3.0106
UTCSC2 (deg.) −5 5 2.5568 −4.1178 0.8437 2.8004 −1.1295
QSVC1 (MVAr) −10 10 9.8990 5.7117 9.7120 9.9913 9.7432
QSVC2 (MVAr) −10 10 9.5080 9.9670 10.0000 9.4666 9.9943
TCSC1 branch. 1 40 5 40 6 10 5
TCSC2 branch. 1 41 33 13 5 5 7
TCPS1 branch. 1 40 4 1 14 1 14
TCPS2 branch. 1 41 14 13 6 14 2
SVC1 bus no: 3 29 7 19 24 24 24
SVC2 bus no: 3 30 24 24 21 7 21
Energies 2024, 17, 6087 12 of 25

Table 4. Cont.

Limits Case 1
Min Max SHADE-SF [44] INFO INFO-FDB Hyper-INFO Hyper-FDB-INFO
PTG1 (MW) 50 200 134.9079 134.9079 134.9191 134.9079 134.9081
QTG1 (MVAr) −20 150 3.9339 2.0120 6.8386 1.4326 2.4785
Parameters

QTG2 (MVAr) −20 60 15.9236 16.3795 9.3022 16.3100 16.1352

QWG5 (MVAr) −30 35 20.1032 24.4690 26.8940 19.8304 24.2143
QTG8 (MVAr) −15 48.7 30.6543 31.5852 27.9424 30.0214 28.4837
QWG11 (MVAr) −25 30 25.5750 23.0657 22.5092 17.8511 26.0697
QTG13 (MVAr) −15 44.7 20.0273 21.3839 20.7599 30.5872 18.5194
Cgen (USD/h) 807.0166 807.3134 807.2136 807.1923 806.9707
Results

Ploss (MW) 5.5454 5.5903 5.5524 5.5567 5.5329

Cgross (USD/h) 1361.5566 1366.3468 1362.4574 1362.8591 1360.2633
VD (p.u.) 0.8690 0.6656 0.7113 0.5832 0.8458

Table 5. Simulation results of case 2 with fixed loading for the adapted IEEE 30-bus system.

Limits Case 2
Min Max SHADE-SF [44] INFO INFO-FDB Hyper-INFO Hyper-FDB-INFO
PTG2 (MW) 20 80 25.1468 25.1586 25.1567 25.1474 25.2009
PWG5 (MW) 0 75 75.0000 75.0000 75.0000 75.0000 74.9980
PTG8 (MW) 10 35 35.0000 35.0000 35.0000 34.9994 35.0000
PWG11 (MW) 0 60 60.0000 59.9999 60.0000 60.0000 59.9773
PTG13 (MW) 12 40 40.0000 40.0000 39.9998 39.9968 39.9651
V1 (p.u.) 0.95 1.10 1.0579 1.0587 1.0549 1.0578 1.0567
V2 (p.u.) 0.95 1.10 1.0520 1.0526 1.0493 1.0518 1.0514
V5 (p.u.) 0.95 1.10 1.0419 1.0425 1.0396 1.0421 1.0421
V8 (p.u.) 0.95 1.10 1.0471 1.0477 1.0447 1.0472 1.0472
V11 (p.u.) 0.95 1.10 1.0927 1.0879 1.0928 1.0916 1.0983
V13 (p.u.) 0.95 1.10 1.0721 1.0694 1.0831 1.0721 1.0731
Control Variables

T11 (p.u.) 0.90 1.10 1.0400 1.0000 1.0600 1.0600 1.0400

T12 (p.u.) 0.90 1.10 0.9200 0.9800 0.9000 0.9000 0.9400
T15 (p.u.) 0.90 1.10 1.0000 1.0000 1.0200 1.0000 1.0000
T36 (p.u.) 0.90 1.10 0.9800 0.9800 0.9800 0.9800 0.9800
tTCSC1 (%) 0 50% 32.2355 50.0000 49.9898 49.9963 27.4032
tTCSC2 (%) 0 50% 50.0000 25.7413 6.7808 21.7892 25.7246
UTCSC1 (deg.) −5 5 4.6022 4.2554 4.6055 4.6084 −2.3488
UTCSC2 (deg.) −5 5 −2.3660 −0.5604 −2.3864 −2.3985 4.6592
QSVC1 (MVAr) −10 10 9.9476 10.0000 9.9997 10.0000 9.9984
QSVC2 (MVAr) −10 10 9.5046 5.7099 9.7297 9.6377 9.9626
TCSC1 branch. 1 40 38 14 34 34 2
TCSC2 branch. 1 41 37 2 18 5 5
TCPS1 branch. 1 40 33 33 33 33 14
TCPS2 branch. 1 41 14 8 14 14 33
SVC1 bus no: 3 29 21 24 21 21 21
SVC2 bus no: 3 30 24 19 24 24 24
PTG1 (MW) 50 200 50.0000 50.0000 50.0000 50.0000 50.0001
QTG1 (MVAr) −20 150 −2.0427 −0.8129 −4.2870 −1.9729 −3.8958
Parameters

QTG2 (MVAr) −20 60 10.0436 11.0250 7.4974 9.5649 9.4943

QWG5 (MVAr) −30 35 22.0610 22.4438 21.6198 22.0511 21.9420
QTG8 (MVAr) −15 48.7 32.5537 35.7034 28.6637 32.4627 30.5356
QWG11 (MVAr) −25 30 27.1544 23.1180 29.9992 29.2840 29.9561
QTG13 (MVAr) −15 44.7 17.9252 17.0708 26.6233 17.9367 18.6767
Cgen (USD/h) 939.2916 939.3300 939.3225 939.2749 939.1754
Results

Ploss (MW) 1.7467 1.7586 1.7564 1.7436 1.7413

Cgross (USD/h) 1113.9616 1115.1862 1114.9613 1113.6353 1113.3088
VD (p.u.) 0.9256 0.8586 0.9090 0.9180 0.9048
Energies 2024, 17, 6087 13 of 25

Table 6. Simulation results of case 3 with fixed loading for the adapted IEEE 30-bus system.

Limits Case 3
Min Max SHADE-SF [44] INFO INFO-FDB Hyper-INFO Hyper-FDB-INFO
PTG2 (MW) 20 80 38.7639 34.8642 39.0098 38.8890 38.4926
PWG5 (MW) 0 75 75.0000 74.9999 75.0000 75.0000 75.0000
PTG8 (MW) 10 35 35.0000 34.9995 35.0000 35.0000 35.0000
PWG11 (MW) 0 60 60.0000 59.9998 60.0000 60.0000 60.0000
PTG13 (MW) 12 40 26.4970 30.3718 26.2471 26.3755 26.7611
V1 (p.u.) 0.95 1.10 1.0595 1.0578 1.0580 1.0598 1.0596
V2 (p.u.) 0.95 1.10 1.0542 1.0522 1.0536 1.0545 1.0544
V5 (p.u.) 0.95 1.10 1.0437 1.0416 1.0439 1.0439 1.0437
V8 (p.u.) 0.95 1.10 1.0473 1.0454 1.0473 1.0476 1.0472
V11 (p.u.) 0.95 1.10 1.0895 1.0940 1.0834 1.0661 1.0851
V13 (p.u.) 0.95 1.10 1.0725 1.0813 1.0726 1.0726 1.0726
Control Variables

T11 (p.u.) 0.90 1.10 1.0400 1.0200 1.0200 1.0200 1.0200

T12 (p.u.) 0.90 1.10 0.9200 0.9200 0.9400 0.9400 0.9400
T15 (p.u.) 0.90 1.10 1.0000 1.0200 1.0000 1.0000 1.0000
T36 (p.u.) 0.90 1.10 0.9800 0.9800 0.9800 0.9800 0.9800
tTCSC1 (%) 0 50% 20.8723 5.0230 49.9862 49.1607 49.9972
tTCSC2 (%) 0 50% 17.1833 26.1281 47.7258 43.5889 26.1374
UTCSC1 (deg.) −5 5 2.9326 0.5061 −0.5555 −0.5034 0.5941
UTCSC2 (deg.) −5 5 4.3128 3.2059 2.8346 2.9743 3.0436
QSVC1 (MVAr) −10 10 9.9362 2.4191 9.9984 9.9991 9.9981
QSVC2 (MVAr) −10 10 9.5065 10.0000 9.9051 9.8982 9.9998
TCSC1 branch. 1 40 5 3 21 13 30
TCSC2 branch. 1 41 41 5 24 24 2
TCPS1 branch. 1 40 35 1 4 5 9
TCPS2 branch. 1 41 34 33 35 35 33
SVC1 bus no: 3 29 21 26 21 21 21
SVC2 bus no: 3 30 24 24 24 24 24
PTG1 (MW) 50 200 50.0000 50.0000 50.0140 50.0000 50.0000
QTG1 (MVAr) −20 150 −1.8868 −2.0450 −3.7573 −1.9341 −1.7389
Parameters

QTG2 (MVAr) −20 60 10.2193 9.1989 11.4585 11.0148 11.1391

QWG5 (MVAr) −30 35 22.2143 22.1723 23.4287 22.4678 22.4888
QTG8 (MVAr) −15 48.7 32.8832 32.9734 34.9140 34.3944 33.6712
QWG11 (MVAr) −25 30 25.8137 26.4239 21.1274 17.9129 21.6277
QTG13 (MVAr) −15 44.7 17.6613 26.2715 18.0390 17.7730 17.7216
Cgen (USD/h) 918.0000 923.0536 917.7267 917.8658 918.2982
Results

Ploss (MW) 1.8608 1.8351 1.8709 1.8645 1.8537

Cgross (USD/h) 1104.0771 1106.5660 1104.8206 1104.3138 1103.6675
VD (p.u.) 0.9177 0.8815 0.9044 0.9187 0.9155

4.1. Case Studies with Fixed Loading

This subsection consists of three case studies, Case 1, Case 2, and Case 3, where 100%
loading conditions are considered. The optimization problem was run independently
30 times by each algorithm used for the three cases. In the optimization process, the best
suitability values obtained with different algorithms and the parameter values of the control
variables are given in Tables 4–6.
The aim is to obtain the minimum value of the production cost (Cgen) in Equation (5)
for Case 1. According to the determined objective function, as can be seen from Table 4, the
minimum cost value of 806.9707 USD/h is obtained with the proposed Hyper-FDB-INFO
algorithm. The cost values for Case 1 are 807.1923 USD/h with the Hyper-INFO algorithm;
807.2136 USD/h with INFO-FDB; 807.3134 USD/h with INFO; and 807.1923 USD/h with
SHADE-SF [37]. When we compare this value with other algorithms, Hyper-FDB-INFO
provides 0.0275% lower cost than Hyper-INFO, 0.0301% lower cost than INFO-FDB, 0.0425%
lower cost than INFO, and 0.0057% lower cost than SHADE-SF [44]. In addition, Table 4
includes the control parameter variables of all compared algorithms for Case 1 and the
Energies 2024, 17, 6087 14 of 25

optimal locations and ratings of FACTS devices. The branch and bus numbers connected
to TCSC and TCPS are determined as 5-7 and 14-, respectively, with Hyper-FDB-INFO.
In addition, the optimal locations for SVC1 and SVC2 are determined as buses 24 and
21. In this study, the maximum compensation value for TCSC with Hyper-FDB-INFO
(tTCSC1 and tTCSC2) was found to be 28.5902% and 29.7363%, respectively, the angle
value of the phase shifter (TCPS) (UTCPS1 and UTCPS2) was found to be 3.0106 and
−1.1295 degrees, respectively, and the reactive power extracting and generating capacities
of the SVC (QSVC1and QSVC2) were found to be 9.7432 and 9.9943 MVAr, respectively.
The main objective in Case 2 is to minimize the power loss. Minimizing the real power
loss of the grid allows an increase in the loading capacity of the grid by using FACTS
devices. When the control variables and the obtained fitness function value given in Table 5
are examined, it is seen that the power loss is obtained as 1.7413 MW with the Hyper-FDB-
INFO algorithm. The power loss values are obtained as 1.7436 MW with the Hyper-INFO
algorithm; 1.7564 MW with INFO-FDB; 1.7586 MW with INFO; and 1.7467 MW with
SHADE-SF [44] for Case 2. When we compare this value with other algorithms, it is shown
that Hyper-FDB-INFO has a lower power loss of 0.13% than Hyper-INFO, 0.86% than
INFO-FDB, 0.98% than INFO, and 0.31% than SHADE-SF [44]. In addition, Table 5 also
includes the control parameter variables and the locations of the FACTS devices for all
the algorithms compared to Case 2. The branch and bus numbers connected to TCSC and
TCPS are determined as 2-5 and 14-33, respectively, with Hyper-FDB-INFO. In addition,
the optimal locations for SVC1 and SVC2 are determined as buses 21 and 24. In this study,
the maximum compensation values for TCSC with Hyper-FDB-INFO (tTCSC1 and tTCSC2)
are 27.4032% and 25.7246%, respectively, the angle values of the phase shifter (TCPS)
(UTCPS1 and UTCPS2) are −2.3488 and 4.6592 degrees, respectively, and the reactive
power extracting and generating capacities of the SVC (QSVC1 and QSVC2) are found to
be 9.9984 and 9.99626 MVAr, respectively.
When the values obtained in Case 1 and Case 2 are examined, it is seen that a more
effective result is obtained in terms of generation cost for Case 1 compared to Case 2, while
the power loss is seen to be higher in Case 1 compared to Case 2. This situation reveals
the necessity of a new objective function that will minimize both the cost and power loss
at the same time. For this reason, a new cost model was created in line with this objective
function in Case 3. The cost of the energy considered in this study is 0.10 USD/kWh. This
cost equation (Cgross) is given in Equation (24):

Cgross = Cgen + Ploss ∗ 103 ∗ 0.10 (24)

Accordingly, in Case 3, the value of the cost function to be minimized was obtained as
1103.6675 USD/h by the Hyper-FDB-INFO algorithm, considering both cost and power
loss in Table 6. The cost values are 1104.3138 USD/h with the Hyper-INFO algorithm,
1104.8206 USD/h with INFO-FDB, 1106.5660 USD/h with INFO, and 1104.0771 USD/h with
SHADE-SF [44] for Case 3. When we compare this value with other algorithms, it is shown
that Hyper-FDB-INFO provides a cost value that is 0.059% lower than Hyper-INFO, 0.134%
lower than INFO-FDB, 0.262% lower than INFO, and 0.0371% lower than SHADE-SF [44].
The branch and bus numbers connected to TCSC and TCPS are determined as 30-2 and
9-33, respectively, with Hyper-FDB-INFO. Also, the optimal locations for SVC1 and SVC2
are determined as buses 21 and 24. In this study, the maximum compensation values for
TCSC with Hyper-FDB-INFO (tTCSC1and tTCSC2) are 49.9972% and 26.1374%, the angle
values of the phase shifter (TCPS) (UTCPS1 and UTCPS2) are 0.5941 and 3.0436 degrees,
respectively, and the reactive power extracting and generating capacities of SVC (QSVC1
and QSVC2) are found to be 9.9981 and 9.9998 MVAr, respectively.

4.2. A Case Study Considering Uncertainties in Load Demand

In this subsection, Case 4 is considered, where uncertain load demands are assumed.
The uncertain load demand is expressed using the normal PDF. The optimization is carried
Energies 2024, 17, 6087 15 of 25

out by creating various scenarios according to different load demands. The loading and
probabilities for all scenarios are given in Table 7 [44].

Table 7. Loading scenarios and their probabilities.

—
Loading Scenario % Loading, Pd Scenario Probability, ∆SC
sc1 54.749 0.15866
sc2 65.401 0.34134
sc3 74.599 0.34134
sc4 85.251 0.15866

According to Table 7, in a scenario, % loading means that the demands on all buses are
multiplied by the percentage of that scenario. In each scenario, the Cgross objective function
in Equation (23) according to the loading level in the grid is optimized by all algorithms.
The planned power from all generators is optimized in each scenario. It is not practical to
change the locations of FACTS devices under different loading scenarios. Therefore, as in
the reference study [44], the locations of FACTS devices are optimized for scenario 3 (load
level 3), and the same optimized locations are used for other loading scenarios. The ratings
of FACTS devices are optimized for different loading levels (scenarios).
First of all, the analysis results for scenario 1 (sc1 ) are given in Table 8. According to
these results, the minimum cost value of 514.5761 USD/h is obtained with the proposed
Hyper-FDB-INFO algorithm. For Case 4_sc1 , the cost values are 514.6121 USD/h with the
Hyper-INFO algorithm; 514.8226 USD/h with INFO-FDB; 515.3933 USD/h with INFO; and
514.6576 USD/h with SHADE-SF [37]. The value obtained with the proposed algorithm,
when compared to the competing algorithms, provides 0.007% lower cost than Hyper-INFO,
0.0479% lower cost than INFO-FDB, 0.1586% lower cost than INFO, and 0.0158% lower cost
than SHADE-SF [37]. In addition, the branch and bus numbers for Hyper-FDB-INFO with
TCSC and TCPS were determined as 35-2 and 8-14, respectively. The optimal locations for
SVC1 and SVC2 were determined as buses 21 and 24. In the conducted study, the maximum
compensation values for TCSC with Hyper-FDB-INFO (tTCSC1 and tTCSC2) are 48.5856%
and 27.0498%, the angle values of the phase shifter (TCPS) (UTCPS1 and UTCPS2) are
−0.4557 and 1.2485 degrees, respectively, and the reactive power absorption and generation
capacities of SVC (QSVC1 and QSVC2) are found to be 6.6082 and 4.9726 MVAr, respectively.
The results for Case 4_sc2 are given in Table 9. According to results, the minimum cost
value of 626.1899 USD/h is obtained with the Hyper-FDB-INFO algorithm. For Case 4_sc2 ,
the cost values are 626.1903 USD/h with the Hyper-INFO algorithm; 626.2569 USD/h with
INFO-FDB; 626.7105 USD/h with INFO; and 626.1980 USD/h with SHADE-SF [44]. The
value obtained with the proposed algorithm, when compared to the competing algorithms,
provides 0.000064% lower cost than Hyper-INFO, 0.0107% lower cost than INFO-FDB,
0.0831% lower cost than INFO, and 0.0013% lower cost than SHADE-SF [44]. In addition,
the branch and bus numbers of TCSC and TCPS with Hyper-FDB-INFO were determined
as 20-2 and 8-14, respectively. In addition, the optimal locations for SVC1 and SVC2 were
determined as buses 21 and 24, as in the previous scenario. In this study, the maximum
compensation values for TCSC with Hyper-FDB-INFO (tTCSC1 and tTCSC2) were found
to be 0.0315% and 25.4221%, the angle values of the phase shifter (TCPS) (UTCPS1 and
UTCPS2) were found to be −0.3639 and 2.5628 degrees, respectively, and the reactive power
absorption and generation capacities of SVC (QSVC1 and QSVC2) were found to be 7.0620
and 5.6781 MVAr, respectively.
The analysis results for Case 4_sc3 are given in Table 10. According to these results,
the minimum cost value of 740.2716 USD/h is obtained with the proposed Hyper-FDB-
INFO algorithm. For Case 4_sc3 , the cost values are 740.6363 USD/h with the Hyper-
INFO algorithm; 740.6066 USD/h with INFO-FDB; 741.5760 USD/h with INFO; and
740.2894 USD/h with SHADE-SF [44]. The value obtained with the proposed algorithm,
when compared to the competing algorithms, provides 0.05% lower cost than Hyper-INFO,
Energies 2024, 17, 6087 16 of 25

0.045% lower cost than INFO-FDB, 0.1759% lower cost than INFO, and 0.0024% lower
cost than SHADE-SF [44]. In addition, the branch and bus numbers for Hyper-FDB-INFO
with TCSC and TCPS were determined as 20-2 and 8-14, respectively, as in the previous
scenario, and the optimal locations for SVC1 and SVC2 were determined as buses 21 and
24. In the conducted study, the maximum compensation values for Hyper-FDB-INFO and
TCSC (tTCSC1 and tTCSC2) are found to be 0.0002% and 25.1171%, respectively, the angle
values of the phase shifter (TCPS) (UTCPS1 and UTCPS2) are −0.3360 and 2.9835 degrees,
respectively, and the reactive power absorption and generation capacities of the SVC
(QSVC1 and QSVC2) are found to be 8.4660 and 6.7646 MVAr, respectively.

Table 8. Simulation results for scenario_1(sc1 ) of loading in Case 4.

Limits Case 4_sc1

Min Max SHADE-SF [44] INFO INFO-FDB Hyper-INFO Hyper-FDB-INFO
PTG2 (MW) 20 80 20.0000 20.0713 20.0291 20.0015 20.0000
PWG5 (MW) 0 75 36.4100 36.2637 36.2854 36.3348 36.3288
PTG8 (MW) 10 35 10.0000 10.0042 10.0037 10.0003 10.0001
PWG11 (MW) 0 60 27.7150 27.7887 27.7882 27.7877 27.7953
PTG13 (MW) 12 40 12.0000 12.0039 12.0032 12.0001 12.0001
V1 (p.u.) 0.95 1.10 1.0582 1.0561 1.0579 1.0584 1.0582
V2 (p.u.) 0.95 1.10 1.0528 1.0516 1.0526 1.0531 1.0527
V5 (p.u.) 0.95 1.10 1.0448 1.0444 1.0446 1.0448 1.0449
V8 (p.u.) 0.95 1.10 1.0456 1.0461 1.0460 1.0458 1.0455
V11 (p.u.) 0.95 1.10 1.0680 1.0659 1.0669 1.0700 1.0713
V13 (p.u.) 0.95 1.10 1.0554 1.0503 1.0503 1.0494 1.0522
Control Variables

T11 (p.u.) 0.90 1.10 1.0000 1.0000 1.0000 1.0000 1.0000

T12 (p.u.) 0.90 1.10 1.0000 1.0200 1.0200 1.0200 1.0200
T15 (p.u.) 0.90 1.10 1.0000 1.0000 1.0000 1.0000 1.0000
T36 (p.u.) 0.90 1.10 1.0000 1.0000 1.0000 1.0000 1.0000
tTCSC1 (%) 0 50% 0.0000 38.1699 40.0266 49.9560 48.5856
tTCSC2 (%) 0 50% 25.5421 23.9478 25.8194 27.0889 27.0498
UTCSC1 (deg.) −5 5 −0.4679 −0.4508 −0.4372 −0.4334 −0.4557
UTCSC2 (deg.) −5 5 1.1966 1.2562 1.3952 1.3752 1.2485
QSVC1 (MVAr) −10 10 5.6993 5.8404 6.1986 6.6040 6.6082
QSVC2 (MVAr) −10 10 5.2251 4.9686 4.9909 5.0006 4.9726
TCSC1 branch. 1 40 20 37 35 35 35
TCSC2 branch. 1 41 2 2 2 2 2
TCPS1 branch. 1 40 8 8 8 8 8
TCPS2 branch. 1 41 14 14 14 14 14
SVC1 bus no: 3 29 21 21 21 21 21
SVC2 bus no: 3 30 24 24 24 24 24
PTG1 (MW) 50 200 50.0000 50.0004 50.0172 50.0003 50.0001
QTG1 (MVAr) −20 150 −2.5289 −4.8064 −2.7152 −2.6255 −2.4385
Parameters

QTG2 (MVAr) −20 60 3.4474 3.8595 3.4553 3.9621 3.0088

QWG5 (MVAr) −30 35 11.0616 11.4066 10.9112 10.7301 10.9502
QTG8 (MVAr) −15 48.7 15.6185 17.7104 16.2651 15.1358 14.4208
QWG11 (MVAr) −25 30 9.9746 10.5460 10.7020 11.5900 11.8846
QTG13 (MVAr) −15 44.7 7.5287 6.7786 6.4347 5.7831 6.7603
Cgen (USD/h) 417.9276 417.9366 417.9181 417.9058 417.9024
Results

Ploss (MW) 0.9673 0.9746 0.9690 0.9671 0.9667

Cgross (USD/h) 514.6576 515.3933 514.8226 514.6121 514.5761
VD (p.u.) 0.9125 0.8319 0.8461 0.8605 0.8772
Energies 2024, 17, 6087 17 of 25

Table 9. Simulation results for scenario_2(sc2) of loading in Case 4.

Limits Case 4_sc2

Min Max SHADE-SF [44] INFO INFO-FDB Hyper-INFO Hyper-FDB-INFO
PTG2 (MW) 20 80 20.0087 20.2425 20.2163 20.0002 20.0002
PWG5 (MW) 0 75 51.4299 51.0485 51.1107 51.5829 51.5828
PTG8 (MW) 10 35 10.7322 10.7381 10.7264 10.7191 10.7191
PWG11 (MW) 0 60 42.2339 42.2768 42.2832 42.1023 42.1023
PTG13 (MW) 12 40 12.0000 12.0972 12.0664 12.0000 12.0000
V1 (p.u.) 0.95 1.10 1.0581 1.0538 1.0552 1.0581 1.0581
V2 (p.u.) 0.95 1.10 1.0527 1.0483 1.0492 1.0527 1.0527
V5 (p.u.) 0.95 1.10 1.0457 1.0407 1.0417 1.0458 1.0458
V8 (p.u.) 0.95 1.10 1.0450 1.0412 1.0416 1.0451 1.0451
V11 (p.u.) 0.95 1.10 1.0792 1.0821 1.0824 1.0787 1.0790
V13 (p.u.) 0.95 1.10 1.0636 1.0633 1.0643 1.0636 1.0636
Control Variables

T11 (p.u.) 0.90 1.10 1.0200 1.0200 1.0200 1.0200 1.0200

T12 (p.u.) 0.90 1.10 0.9600 0.9600 0.9600 0.9600 0.9600
T15 (p.u.) 0.90 1.10 1.0000 1.0000 1.0000 1.0000 1.0000
T36 (p.u.) 0.90 1.10 0.9800 0.9800 0.9800 0.9800 0.9800
tTCSC1 (%) 0 50% 0.0000 17.2597 19.3445 0.0242 0.0315
tTCSC2 (%) 0 50% 25.3565 24.1883 23.8777 25.4696 25.4221
UTCSC1 (deg.) −5 5 −0.3668 1.4976 1.6299 −0.3686 −0.3639
UTCSC2 (deg.) −5 5 2.5846 1.4099 1.3390 2.5631 2.5628
QSVC1 (MVAr) −10 10 6.9519 7.0656 6.3697 7.0624 7.0620
QSVC2 (MVAr) −10 10 5.7358 5.0362 6.5601 5.6782 5.6781
TCSC1 branch. 1 40 20 5 5 20 20
TCSC2 branch. 1 41 2 2 2 2 2
TCPS1 branch. 1 40 8 14 14 8 8
TCPS2 branch. 1 41 14 33 33 14 14
SVC1 bus no: 3 29 21 24 24 21 21
SVC2 bus no: 3 30 24 21 21 24 24
PTG1 (MW) 50 200 50.0000 50.0080 50.0036 50.0000 50.0000
QTG1 (MVAr) −20 150 −2.5287 −2.9315 −1.5289 −2.5517 −2.5741
Parameters

QTG2 (MVAr) −20 60 4.4639 3.7172 2.6527 4.3796 4.3876

QWG5 (MVAr) −30 35 13.2331 12.2267 12.3793 13.2884 13.2809
QTG8 (MVAr) −15 48.7 18.0931 17.4974 16.5016 18.2190 18.1884
QWG11 (MVAr) −25 30 16.7744 18.9927 18.5914 16.5823 16.6623
QTG13 (MVAr) −15 44.7 10.4344 11.7914 11.6809 10.4506 10.4257
Cgen (USD/h) 520.4780 520.3510 520.3438 520.4947 520.4947
Results

Ploss (MW) 1.0572 1.0636 1.0591 1.0570 1.0570

Cgross (USD/h) 626.1980 626.7105 626.2569 626.1903 626.1899
VD (p.u.) 1.0274 0.9765 1.0047 1.0271 1.0277

The analysis results for Case 4_sc4 are given in Table 11. According to these results,
the minimum cost value of 882.4098 USD/h is obtained with the proposed Hyper-FDB-
INFO algorithm. For Case 4_sc4, the cost values are 882.4101 USD/h with the Hyper-
INFO algorithm; 882.4200 USD/h with INFO-FDB; 882.5805 USD/h with INFO; and
882.4103 USD/h with SHADE-SF [37]. The value obtained with the proposed algorithm,
when compared to the competing algorithms, provides 0.000034% lower cost than Hyper-
INFO, 0.001155% lower cost than INFO-FDB, 0.019% lower cost than INFO, and 0.000056%
lower cost than SHADE-SF [37]. In addition, the branch and bus numbers for Hyper-
FDB-INFO with TCSC and TCPS were determined as 19-2 and 8-14, respectively, and the
optimal locations for SVC1 and SVC2 were determined as busbars 21-24. In the conducted
study, the maximum compensation values for Hyper-FDB-INFO and TCSC (tTCSC1 and
tTCSC2) are found to be 0.0013% and 24.9518%, respectively, the angle values of the phase
shifter (TCPS) (UTCPS1 and UTCPS2) are −0.3315 and 3.4102 degrees, respectively, and
the reactive power absorption and generation capacities of SVC (QSVC1 and QSVC2) are
found to be 9.0528 and 7.8057 MVAr, respectively.
Energies 2024, 17, 6087 18 of 25

Table 10. Simulation results for scenario_3(sc3 ) of loading in Case 4.

Limits Case 4_sc3

Min Max SHADE-SF [44] INFO INFO-FDB Hyper-INFO Hyper-FDB-INFO
PTG2 (MW) 20 80 20.0112 20.0137 20.0012 20.0132 20.0000
PWG5 (MW) 0 75 61.6479 61.6765 61.6428 61.6809 61.6285
PTG8 (MW) 10 35 21.3290 21.2856 21.3031 21.3447 21.3135
PWG11 (MW) 0 60 47.5728 47.5672 47.5685 47.5132 47.6255
PTG13 (MW) 12 40 12.0068 12.0113 12.0122 12.0188 12.0000
V1 (p.u.) 0.95 1.10 1.0584 1.0584 1.0585 1.0589 1.0583
V2 (p.u.) 0.95 1.10 1.0528 1.0530 1.0530 1.0532 1.0529
V5 (p.u.) 0.95 1.10 1.0461 1.0461 1.0463 1.0463 1.0463
V8 (p.u.) 0.95 1.10 1.0466 1.0466 1.0465 1.0464 1.0466
V11 (p.u.) 0.95 1.10 1.0808 1.0802 1.0796 1.0817 1.0803
V13 (p.u.) 0.95 1.10 1.0662 1.0654 1.0649 1.0650 1.0660
Control Variables

T11 (p.u.) 0.90 1.10 1.0200 1.0200 1.0200 1.0200 1.0200

T12 (p.u.) 0.90 1.10 0.9600 0.9400 0.9600 0.9400 0.9600
T15 (p.u.) 0.90 1.10 1.0000 1.0000 1.0000 1.0000 1.0000
T36 (p.u.) 0.90 1.10 0.9800 0.9800 0.9800 0.9800 0.9800
tTCSC1 (%) 0 50% 0.0031 43.6015 28.4747 41.3138 0.0002
tTCSC2 (%) 0 50% 25.2402 19.5952 24.3143 16.0406 25.1171
UTCSC1 (deg.) −5 5 −0.3536 3.0479 0.6440 2.5747 −0.3360
UTCSC2 (deg.) −5 5 2.9971 3.0883 3.0874 0.7541 2.9835
QSVC1 (MVAr) −10 10 8.1690 4.5370 8.7641 4.5730 8.4660
QSVC2 (MVAr) −10 10 6.7203 6.3786 6.8966 7.6967 6.7646
TCSC1 branch. 1 40 20 24 1 33 20
TCSC2 branch. 1 41 2 26 7 25 2
TCPS1 branch. 1 40 8 14 1 14 8
TCPS2 branch. 1 41 14 13 14 35 14
SVC1 bus no: 3 29 21 21 21 19 21
SVC2 bus no: 3 30 24 24 24 24 24
PTG1 (MW) 50 200 50.0000 50.0260 50.0432 50.0001 50.0000
QTG1 (MVAr) −20 150 −1.8919 −2.2364 −2.0838 −1.5837 −2.3085
Parameters

QTG2 (MVAr) −20 60 5.4891 7.3914 5.8820 6.7536 5.8249

QWG5 (MVAr) −30 35 15.3614 15.9120 15.6391 15.8509 15.4973
QTG8 (MVAr) −15 48.7 22.1944 25.0282 21.9585 23.9627 22.1060
QWG11 (MVAr) −25 30 18.0776 18.1357 17.5758 18.5357 17.8350
QTG13 (MVAr) −15 44.7 12.4160 12.4128 11.8131 11.4958 12.2909
Cgen (USD/h) 624.7694 624.7946 624.7491 624.7894 624.7737
Results

Ploss (MW) 1.1552 1.1678 1.1586 1.1585 1.1550

Cgross (USD/h) 740.2894 741.5760 740.6066 740.6363 740.2716
VD (p.u.) 0.9985 0.9719 0.9992 1.0148 1.0020

In the simulation study, the number of iterations and the number of populations were
kept constant for all algorithms, 30 independent runs were performed, and best, worst,
mean, and standard deviation values were obtained. These values are given in Table 12,
and the best results obtained are given in bold. Accordingly, in Case 1, the minimum
fitness value obtained by the Hyper-FDB-INFO was 806.9707. From this result, it can be
seen that the best fitness value for minimum values is obtained with Hyper-FDB-INFO
in Case 1. Although Hyper-FDB-INFO gives the best result in terms of minimum fitness
value, to show the robustness of the algorithm more clearly, the average and standard
deviation values obtained as a result of 30 runs should also be evaluated. Considering
the structure of metaheuristic algorithms, it may not be enough to obtain the best result
only once to make a fairer comparison. When comparing the mean values for Case 1, the
Hyper-FDB-INFO algorithm is seen as the algorithm that gives the most effective result
with 807.1176. Regarding the mean value, SHADE-SF [44] ranked second with a fitness
value of 807.1832, and Hyper-INFO-FDB ranked third with a fitness value of 807.5032. In
terms of standard deviation value, Hyper-INFO-FDB was the most stable algorithm with a
Energies 2024, 17, 6087 19 of 25

value of 0.0383 for Case 1 after 30 independent runs. The second algorithm that gave the
best results in terms of standard deviation value was Hyper-INFO, while the third-ranked
algorithm was SHADE-SF [44]. As can be seen from its success in mean and standard
deviation values, Hyper-FDB-INFO also gave better results than the competing algorithms
in terms of worst value with a value of 807.1689. When the results obtained for Case 2 are
evaluated, Hyper-FDB-INFO is the algorithm that gives the best result with a minimum
fitness value of 1.7413, while Hyper-INFO is in second place with a value of 1.7436, and
SHADE-SF is in third place with 1.7467. For Case 2, the algorithm that gives effective
results in terms of mean values is again Hyper-FDB-INFO with 1.7909, while SHADE-SF
is in second place with 1.7947. The most effective algorithm after these two algorithms is
INFO-FDB, with 1.8765. Moreover, Hyper-FDB-INFO is also superior to its competitors
in terms of standard deviation value with a value of 0.0147. For Case 3, the algorithm
that provides the best result with a value of 1103.6675 in terms of minimum values is
again Hyper-FDB-INFO, while this proposed algorithm has shown that it is more effective
than its competitors with a mean value of 1106.4385 and a standard deviation value of
0.7444. Unlike the fixed load case performed in the first three case studies, an uncertain
load case was considered for Case 4 under four different loading conditions. In the first
scenario of Case 4, in terms of minimum values, while Hyper-FDB-INFO provides the
best solution with 514.5761, Hyper-INFO is in second place with a value of 514.6121, and
SHADE-SF [44] is in third place with a value of 514.6576. When comparing in terms of mean
and standard deviation values, it is seen that Hyper-FDB-INFO is once again the algorithm
that gives the best results with values of 516.3374 and 0.4433, respectively. For all scenario
cases in Case 4, only minimum values are available for SHADE-SF [44] in the referenced
study. Therefore, the worst, mean, and standard deviation values cannot be compared with
this algorithm for Case 4. In the second scenario of Case 4, when the minimum values
are examined, Hyper-FDB-INFO is the algorithm that provides the best solution with
626.1899. This algorithm is followed by Hyper-INFO with 626.1903 and SHADE-SF [44]
with 626.1980. In the second scenario of Case 4, the most effective algorithm in terms of
the worst, mean, and standard deviation values was again Hyper-FDB-INFO with values
of 628.9009, 628.1864, and 0.5235, respectively. In the third scenario of Case 4, when the
minimum values are examined, Hyper-FDB-INFO is the algorithm that provides the best
solution with 740.2716. SHADE-SF [44] is in second place with a value of 740.2894, and
INFO-FDB is third with a value of 740.6066 for this scenario. In this scenario, the best
algorithm in terms of mean values was Hyper-FDB-INFO with 742.4758, while Hyper-INFO
was more effective regarding standard deviation with 0.6172. However, when the worst
value is examined, it is seen that Hyper-FDB-INFO provides a better solution with a value
of 740.2716. In the fourth scenario of Case 4, while Hyper-FDB-INFO is the algorithm that
provides the best solution in terms of minimum fitness value with a value of 882.4098,
Hyper-INFO is in second place with a value of 882.4101, and SHADE-SF [44] is in third
place with a value of 882.4103. In this scenario, the best algorithm for mean values was
Hyper-FDB-INFO, with a value of 884.9147. In addition, Hyper-INFO is in second place
with a value of 885.1692, and INFO-FDB is in third place with a value of 886.6490. When
comparing the worst and standard deviation values, it is seen that Hyper-FDB-INFO is the
algorithm that gives the best results with values of 885.8958 and 0.6451, respectively. When
the results obtained for all cases are examined, it is seen that Hyper-INFO-FDB is superior
to its competitors in solving the problem.
Energies 2024, 17, 6087 20 of 25

Table 11. Simulation results for scenario_4(sc4 ) of loading in Case 4.

Limits Case 4_sc4

Min Max SHADE-SF [44] INFO INFO-FDB Hyper-INFO Hyper-FDB-INFO
PTG2 (MW) 20 80 21.7884 21.6322 22.0530 22.0125 22.0195
PWG5 (MW) 0 75 72.1559 72.3912 72.1409 72.1457 72.1420
PTG8 (MW) 10 35 34.5873 34.8091 34.0717 34.1984 34.1685
PWG11 (MW) 0 60 52.4033 52.0951 52.6722 52.5805 52.6076
PTG13 (MW) 12 40 12.0000 12.0080 12.0002 12.0005 12.0000
V1 (p.u.) 0.95 1.10 1.0584 1.0586 1.0584 1.0584 1.0583
V2 (p.u.) 0.95 1.10 1.0529 1.0533 1.0530 1.0530 1.0529
V5 (p.u.) 0.95 1.10 1.0463 1.0468 1.0462 1.0464 1.0464
V8 (p.u.) 0.95 1.10 1.0480 1.0485 1.0480 1.0480 1.0481
V11 (p.u.) 0.95 1.10 1.0916 1.0817 1.0918 1.0922 1.0923
V13 (p.u.) 0.95 1.10 1.0685 1.0690 1.0684 1.0685 1.0684
Control Variables

T11 (p.u.) 0.90 1.10 1.0400 1.0200 1.0400 1.0400 1.0400

T12 (p.u.) 0.90 1.10 0.9400 0.9600 0.9400 0.9400 0.9400
T15 (p.u.) 0.90 1.10 1.0000 1.0000 1.0000 1.0000 1.0000
T36 (p.u.) 0.90 1.10 0.9800 0.9800 0.9800 0.9800 0.9800
tTCSC1 (%) 0 50% 0.0000 0.0326 0.0020 0.0005 0.0013
tTCSC2 (%) 0 50% 24.8635 24.6008 24.8246 24.8824 24.9518
UTCSC1 (deg.) −5 5 −0.3312 −0.3304 −0.3319 −0.3315 −0.3315
UTCSC2 (deg.) −5 5 3.3590 3.2166 3.4106 3.4067 3.4102
QSVC1 (MVAr) −10 10 9.3297 9.9995 9.0428 9.0826 9.0528
QSVC2 (MVAr) −10 10 7.6378 8.0207 8.0423 7.8017 7.8057
TCSC1 branch. 1 40 20 21 20 20 19
TCSC2 branch. 1 41 2 2 2 2 2
TCPS1 branch. 1 40 8 8 8 8 8
TCPS2 branch. 1 41 14 14 14 14 14
SVC1 bus no: 3 29 21 21 21 21 21
SVC2 bus no: 3 30 24 24 24 24 24
PTG1 (MW) 50 200 50.0000 50.0000 50.0000 50.0000 50.0000
QTG1 (MVAr) −20 150 −2.2110 −2.3121 −2.4902 −2.3845 −2.4628
Parameters

QTG2 (MVAr) −20 60 6.8154 7.6986 7.2178 6.9261 6.8821

QWG5 (MVAr) −30 35 17.8232 18.2264 17.6244 17.8436 17.8772
QTG8 (MVAr) −15 48.7 24.4962 26.7341 24.5297 24.4752 24.5890
QWG11 (MVAr) −25 30 25.3875 18.9936 25.4993 25.6811 25.6991
QTG13 (MVAr) −15 44.7 14.2290 14.5789 14.1184 14.1759 14.1692
Cgen (USD/h) 749.0503 749.2361 748.8366 748.8636 748.8593
Results

Ploss (MW) 1.3336 1.3334 1.3358 1.3355 1.3355

Cgross (USD/h) 882.4103 882.5805 882.4200 882.4101 882.4098
VD (p.u.) 0.9726 0.9697 0.9753 0.9737 0.9740

Table 12. The best, worst, mean, and standard deviation values of results obtained by each algorithm
for all case studies.

Algorithms
Cases Results
SHADE-SF [44] INFO INFO-FDB Hyper-INFO Hyper-FDB-INFO
Best 807.0166 807.3134 807.2136 807.1923 806.9707
Worst 807.3969 810.8664 808.1025 807.5921 807.1689
Case 1
Mean 807.1832 809.9631 807.9348 807.5032 807.1176
Std. dev. 0.1007 0.7005 0.1776 0.0804 0.0383
Best 1.7467 1.7586 1.7564 1.7436 1.7413
Worst 1.8924 1.9376 1.9110 1.9221 1.8124
Case 2
Mean 1.7947 1.8858 1.8765 1.8776 1.7909
Std. dev. 0.0462 0.0356 0.0312 0.0354 0.0147
Energies 2024, 17, 6087 21 of 25

Table 12. Cont.

Algorithms
Cases Results
SHADE-SF [44] INFO INFO-FDB Hyper-INFO Hyper-FDB-INFO
Best 1104.0771 1106.5660 1104.8206 1104.3138 1103.6675
Worst 1112.0792 1116.1043 1111.2180 1108.5136 1107.4616
Case 3
Mean 1107.0184 1113.6073 1109.5461 1107.2813 1106.4385
Std. dev. 2.017 2.0114 1.2499 0.8414 0.7444
Best 514.6576 515.3933 514.8226 514.6121 514.5761
Worst ... 520.9447 518.6774 517.1154 516.9137
Case 4-a
Mean ... 519.6917 517.6121 516.3548 516.3374
Std. dev. ... 1.1373 0.7311 0.4793 0.4433
Best 626.1980 626.7105 626.2569 626.1903 626.1899
Worst ... 633.2367 630.7533 629.0554 628.9009
Case 4-b
Mean ... 631.5133 629.5709 628.3721 628.1864
Std. dev. ... 1.2183 0.9155 0.5969 0.5235
Best 740.2894 741.5760 740.6066 740.6363 740.2716
Worst ... 749.1986 745.4911 743.9313 743.3712
Case 4-c
Mean ... 747.4419 744.1908 743.0216 742.4758
Std. dev. ... 1.5511 0.9223 0.6172 0.6287
Best 882.4103 882.5805 882.4200 882.4101 882.4098
Worst ... 891.0599 888.1738 886.1891 885.8958
Case 4-d
Mean ... 888.7323 886.6490 885.1692 884.9147
Std. dev. ... 1.7676 1.1805 0.7753 0.6451

5. Conclusions
In this paper, a hyper-heuristic optimization algorithm, called Hyper-INFO, is pre-
sented, which can find the optimal initial population of INFO by using LSHADE. In this
way, the convergence speed of INFO has been increased. Also, the INFO algorithm was
improved by using the FDB method with its abilities. As a result, another novel Hyper-
FDB-INFO algorithm is presented. The developed algorithms have been applied to solve
the optimal placement and sizing of FACTS devices for OPF problems incorporating wind
energy sources. In addition to the recommended Hyper-FDB-INFO, INFO, FDB-INFO, and
Hyper-INFO algorithms were used for the first time to solve the problem. The experimental
results showed the merits of the proposed Hyper-FDB-INFO algorithm in finding an opti-
mal initial population for the INFO optimization algorithm. When the results are evaluated
in general, the best fitness values were obtained by the Hyper-FDB-INFO algorithm for
OPF case studies with fixed load demand (Case 1, Case 2, and Case 3). In these three case
studies, the proposed algorithm calculated the best generation cost, gross cost, and active
power loss to be 806.9707 USD/h, 1.7413 MW, and 1103.6675 USD/h. Additionally, the
best fitness values were also achieved by the Hyper-FDB-INFO algorithm for the OPF case
studies with uncertain load demand (Case 4a, Case 4b, Case 4c, and Case 4d). The gross cost
values with the proposed algorithm are 514.5761 USD/h, 626.1899 USD/h, 740.2716 USD/h,
and 882.4098 USD/h, which were obtained for Case 4. All the findings obtained from the
simulation studies show that Hyper-FDB-INFO provides better performance compared to
the competing algorithms in solving complex OPF problems with minimum generation
and gross costs, and this proposed algorithm has significant potential in terms of producing
effective solutions for different real-life engineering problems.

Author Contributions: Conceptualization, B.E.A., M.D., E.K., and U.G.; methodology, B.E.A., M.D.,
E.K., and U.G.; validation, B.E.A., M.D., E.K., and U.G.; formal analysis, B.E.A., M.D., E.K., and U.G.;
investigation, B.E.A., M.D., E.K., and U.G.; resources, B.E.A., M.D., E.K., and U.G.; data curation,
B.E.A., M.D., E.K., and U.G.; writing—original draft preparation, B.E.A., M.D., E.K., and U.G.;
writing—review and editing, B.E.A., M.D., E.K., and U.G.; visualization, B.E.A., M.D., E.K., and U.G.;
Energies 2024, 17, 6087 22 of 25

supervision, M.D. and U.G.; project administration, U.G. All authors have read and agreed to the
published version of the manuscript.
Funding: This research received no external funding.
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.

Nomenclature
CTi cost function of thermal units
PTGi generating power output of thermal units
min
PTGi i’th thermal unit’s minimum power output.
ai , bi , ci , di , ei cost coefficients,
gwj direct cost coefficient
Cwj cost function of jth wind generators
Pwsj plant’s scheduled power
CRwj reserve cost
CPwj penalty cost
KRwj coefficient of reserve cost for the jth wind power plant
KPwj coefficient of penalty cost for the jth wind power plant.
Pwavj active power available from the plant
fwj probability density function of the plant
Pwr rated wind power
Cgen total generation cost
Ploss active power loss
Vm , Vn voltage magnitude at bus m and n
Gq(mn) conductance of line m-n
δm , δn voltage angles at bus m,n
nl number of transmission lines
PGm real power generation at bus m
PDm real power demand at bus m
QGm reactive power generation at bus m
QDm reactive power demand at bus m
Pms real power generation injected by the TCPS at bus m
Qms reactive power generation injected by the TCPS at bus m
QSVCm injected reactive power at bus m by the SVC
Ymn magnitude of bus admittance element m,n
θ mn angle of bus admittance element m,n,
NB total number of buses.
min , Pmax
PGi active power generation limits,
Gi
Qmin
Gi , Q Gi
max reactive power generation limits
VGimin , V max constraint on generator bus voltage
Gi
VLpmin , V max load bus voltage
Lp
SLq , SmaxLp line capacity constraint
Ttmin , Ttmax transformer tap setting ranges
min
τTCSCm max
, τTCSCm limits on TCSC
min
ΦTCPSn , ΦTCPSn max limits on TCPS
min
QSVCj , QSVCj max limits on SVC
g
z1 new vectors in the gth generation
g
x1 first solution among all vectors in the population for the gth generation,
σ scaling rate of a vector
MeanRule mean-based rule
CA convergence acceleration
r random number within the range [0, 0.5]
g
W M1l weighted mean of initial vectors random generation
SP score vector
Energies 2024, 17, 6087 23 of 25

Sfdb solution candidate with the maximum value in the score vector of FDB.
N population size
Xi ,j randomly generating an initial population
Lj, Uj lower and upper boundary of the jth dimension

References
1. Mouassa, S.; Alateeq, A.; Alassaf, A.; Bayindir, R.; Alsaleh, I.; Jurado, F. Optimal power flow analysis with renewable energy
resource uncertainty using dwarf mongoose optimizer: Case of ADRAR isolated electrical network. IEEE Access 2024, 12,
10202–10218. [CrossRef]
2. Attia, A.F.; El Sehiemy, R.A.; Hasanien, H.M. Optimal power flow solution in power systems using a novel sine-cosine algorithm.
Int. J. Electr. Power Energy Syst. 2018, 99, 331–343. [CrossRef]
3. Delgado, J.A.; Baptista, E.C.; Balbo, A.R.; Soler, E.M.; Silva, D.N.; Martins, A.C.; Nepomuceno, L. A primal–dual penalty-interior-
point method for solving the reactive optimal power flow problem with discrete control variables. Int. J. Electr. Power Energy Syst.
2022, 138, 107917. [CrossRef]
4. Sasson, A.M. Combined use of the powell and fletcher-powell nonlinear programming methods for optimal load flows. IEEE
Trans. Power Appar. Syst. 1969, 10, 1530–1537. [CrossRef]
5. Sasson, A.M. Decomposition techniques applied to the nonlinear programming load-flow method. IEEE Trans. Power Appar. Syst.
1970, 1, 78–82. [CrossRef]
6. Maria, G.A.; Findlay, J.A. A Newton optimal power flow program for Ontario Hydro EMS. IEEE Trans. Power Syst. 1987, 2,
576–582. [CrossRef]
7. Ponnambalam, K.; Quintana, V.H.; Vannelli, A. A fast algorithm for power system optimization problems using an interior point
method. IEEE Trans. Power Syst. 1992, 7, 892–899. [CrossRef]
8. Abido, M.A. Optimal power flow using particle swarm optimization. Int. J. Elect. Power Syst. 2002, 24, 563–571. [CrossRef]
9. Ongsakul, W.; Tantimaporn, T. Optimal power flow by improved evolutionary programming. Electr. Power Compon. Syst. 2006,
34, 79–95. [CrossRef]
10. Capitanescu, F.; Glavic, M.; Ernst, D.; Wehenkel, L. Interior-point based algorithms for the solution of optimal power flow
problems. Electr. Power Syst. Res. 2007, 77, 508–517. [CrossRef]
11. Bai, X.; Wei, H.; Fujisawa, K.; Wang, Y. Semidefinite programming for optimal power flow problems. Int. J. Electr. Power Energy
Syst. 2008, 30, 383–392. [CrossRef]
12. Bhattacharya, A.; Chattopadhyay, P.K. Application of biogeography-based optimisation to solve different optimal power flow
problems. IET Gener. Transm. Distrib. 2011, 5, 70–80. [CrossRef]
13. Duman, S.; Güvenç, U.; Sönmez, Y.; Yörükeren, N. Optimal power flow using gravitational search algorithm. Energy Convers.
Manag. 2012, 59, 86–95. [CrossRef]
14. Roy, R.; Jadhav, H.T. Optimal power flow solution of power system incorporating stochastic wind power using Gbest guided
artificial bee colony algorithm. Int. J. Electr. Power Energy Syst. 2015, 64, 562–578. [CrossRef]
15. Mishra, C.; Singh, S.P.; Rokadia, J. Optimal power flow in the presence of wind power using modified cuckoo search. IET Gener.
Transm. 2015, 9, 615–626. [CrossRef]
16. Biswas, P.P.; Suganthan, P.N.; Amaratunga, G.A. Optimal power flow solutions incorporating stochastic wind and solar power.
Energy Convers. Manag. 2017, 148, 1194–1207. [CrossRef]
17. Reddy, S.S. Optimal power flow with renewable energy resources including storage. Electr. Eng. 2017, 99, 685–695. [CrossRef]
18. Khan, I.U.; Javaid, N.; Gamage, K.A.; Taylor, C.J.; Baig, S.; Ma, X. Heuristic algorithm based optimal power flow model
incorporating stochastic renewable energy sources. IEEE Access 2020, 8, 148622–148643. [CrossRef]
19. Nusair, K.; Alasali, F. Optimal power flow management system for a power network with stochastic renewable energy resources
using golden ratio optimization method. Energies 2020, 13, 3671. [CrossRef]
20. Sulaiman, M.H.; Mustaffa, Z. Solving optimal power flow problem with stochastic wind–solar–small hydro power using barnacles
mating optimizer. Control Eng. Pract. 2021, 106, 104672. [CrossRef]
21. Riaz, M.; Hanif, A.; Masood, H.; Khan, M.A.; Afaq, K.; Kang, B.G.; Nam, Y. An optimal power flow solution of a system integrated
with renewable sources using a hybrid optimizer. Sustainability 2021, 13, 13382. [CrossRef]
22. Kaymaz, E.; Duman, S.; Guvenc, U. Optimal power flow solution with stochastic wind power using the Lévy coyote optimization
algorithm. Neural Comput. Appl. 2021, 33, 6775–6804. [CrossRef]
23. Guvenc, U.; Duman, S.; Kahraman, H.T.; Aras, S.; Katı, M. Fitness–Distance Balance based adaptive guided differential evolution
algorithm for security-constrained optimal power flow problem incorporating renewable energy sources. Appl. Soft Comput. 2021,
108, 107421. [CrossRef]
24. Souza, R.R.; Balbo, A.R.; Martins, A.C.; Soler, E.M.; Baptista, E.C.; Sousa, D.N.; Nepomuceno, L. A gradient-based approach
for solving the stochastic optimal power flow problem with wind power generation. Electr. Power Syst. Res. 2022, 209, 108038.
[CrossRef]
25. Farhat, M.; Kamel, S.; Atallah, A.M.; Khan, B. Developing a marine predator algorithm for optimal power flow analysis
considering uncertainty of renewable energy sources. Int. Trans. Electr. Energy 2022, 2022, 3714475. [CrossRef]
Energies 2024, 17, 6087 24 of 25

26. Alghamdi, A.S. A hybrid firefly–JAYA algorithm for the optimal power flow problem considering wind and solar power
generations. Appl. Sci. 2022, 12, 7193. [CrossRef]
27. Shaheen, M.A.; Ullah, Z.; Qais, M.H.; Hasanien, H.M.; Chua, K.J.; Tostado-Véliz, M.; Elkadeem, M.R. Solution of probabilistic
optimal power flow incorporating renewable energy uncertainty using a novel circle search algorithm. Energies 2022, 15, 8303.
[CrossRef]
28. Mouassa, S.; Althobaiti, A.; Jurado, F.; Ghoneim, S.S. Novel design of slim mould optimizer for the solution of optimal power
flow problems incorporating intermittent sources: A case study of algerian electricity grid. IEEE Access 2022, 10, 22646–22661.
[CrossRef]
29. Adhikari, A.; Jurado, F.; Naetiladdanon, S.; Sangswang, A.; Kamel, S.; Ebeed, M. Stochastic optimal power flow analysis of power
system with renewable energy sources using adaptive lightning attachment procedure optimizer. Int. J. Electr. Power Energy Syst.
2023, 153, 109314. [CrossRef]
30. Maheshwari, A.; Sood, Y.R.; Jaiswal, S. Flow direction algorithm-based optimal power flow analysis in the presence of stochastic
renewable energy sources. Electr. Power Syst. Res. 2023, 216, 109087. [CrossRef]
31. Hasanien, H.M.; Alsaleh, I.; Alassaf, A.; Alateeq, A. Enhanced coati optimization algorithm-based optimal power flow including
renewable energy uncertainties and electric vehicles. Energy 2023, 283, 129069. [CrossRef]
32. Hassan, M.H.; Kamel, S.; Alateeq, A.; Alassaf, A.; Alsaleh, I. Optimal power flow analysis with renewable energy resource
uncertainty: A hybrid AEO-CGO approach. IEEE Access 2023, 11, 122926–122961. [CrossRef]
33. Alghamdi, A.S. Optimal power flow of hybrid wind/solar/thermal energy integrated power systems considering costs and
emissions via a novel and efficient search optimization algorithm. Appl. Sci. 2023, 13, 4760. [CrossRef]
34. Hassan, M.H.; Kamel, S.; Hussien, A.G. Optimal power flow analysis considering renewable energy resources uncertainty based
on an improved wild horse optimizer. IET Gener. Transm. Distrib. 2023, 17, 3582–3606. [CrossRef]
35. Ozkaya, B. Enhanced growth optimizer algorithm with dynamic fitness-distance balance method for solution of security-
constrained optimal power flow problem in the presence of stochastic wind and solar energy. Appl. Energy 2024, 368, 123499.
[CrossRef]
36. Trojovský, P.; Trojovská, E.; Akbari, E. Economical-environmental-technical optimal power flow solutions using a novel self-
adaptive wild geese algorithm with stochastic wind and solar power. Sci. Rep. 2024, 14, 4135. [CrossRef]
37. Sallam, K.M.; Hossain, M.A.; Elsayed, S.; Chakrabortty, R.K.; Ryan, M.J.; Abido, M.A. Optimal power flow considering intermittent
solar and wind generation using multi-operator differential evolution algorithm. Electr. Power Syst. Res. 2024, 232, 110377.
[CrossRef]
38. Hassan, M.H.; Kamel, S.; Alateeq, A.; Alassaf, A.; Alsaleh, I. Optimal power flow in hybrid Wind-PV-V2G systems with dynamic
load demand using a Hybrid MRFO-AHA Algorithm. IEEE Access 2024, 12, 174297–174329. [CrossRef]
39. Inkollu, S.R.; Kota, V.R. Optimal setting of FACTS devices for voltage stability improvement using PSO adaptive GSA hybrid
algorithm. Eng. Sci. Technol. Int. J. 2016, 19, 1166–1176. [CrossRef]
40. Mohamed, A.A.; Kamel, S.; Hassan, M.H.; Zeinoddini-Meymand, H. CAVOA: A chaotic optimization algorithm for optimal
power flow with facts devices and stochastic wind power generation. IET Gener. Transm. Distrib. 2024, 18, 121–144. [CrossRef]
41. Panda, A.; Tripathy, M. Security constrained optimal power flow solution of wind-thermal generation system using modified
bacteria foraging algorithm. Energy 2015, 93, 816–827. [CrossRef]
42. Elmitwally, A.; Eladl, A. Planning of multi-type FACTS devices in restructured power systems with wind generation. Int. J. Electr.
Power Energy Syst. 2016, 77, 33–42. [CrossRef]
43. Duman, S.; Li, J.; Wu, L.; Guvenc, U. Optimal power flow with stochastic wind power and FACTS devices: A modified hybrid
PSOGSA with chaotic maps approach. Neural Comput. Appl. 2020, 32, 8463–8492. [CrossRef]
44. Biswas, P.P.; Arora, P.; Mallipeddi, R.; Suganthan, P.N.; Panigrahi, B.K. Optimal placement and sizing of FACTS devices for
optimal power flow in a wind power integrated electrical network. Neural Comput. Appl. 2021, 33, 6753–6774. [CrossRef]
45. Nusair, K.; Alasali, F.; Hayajneh, A.; Holderbaum, W. Optimal placement of FACTS devices and power-flow solutions for a power
network system integrated with stochastic renewable energy resources using new metaheuristic optimization techniques. Int. J.
Energy Res. 2021, 45, 18786–18809. [CrossRef]
46. Mohamed, A.A.; Kamel, S.; Hassan, M.H.; Mosaad, M.I.; Aljohani, M. Optimal power flow analysis based on hybrid gradient-
based optimizer with moth–flame optimization algorithm considering optimal placement and sizing of FACTS/wind power.
Mathematics 2022, 10, 361. [CrossRef]
47. Ebeed, M.; Mostafa, A.; Aly, M.M.; Jurado, F.; Kamel, S. Stochastic optimal power flow analysis of power systems with
wind/PV/TCSC using a developed Runge Kutta optimizer. Int. J. Electr. Power Energy Syst. 2023, 152, 109250. [CrossRef]
48. Mohamed, A.A.; Kamel, S.; Hassan, M.H.; Domínguez-García, J.L. Optimal Power Flow Incorporating Renewable Energy Sources
and FACTS Devices: A Chaos Game Optimization Approach. IEEE Access 2024, 12, 23338–23362. [CrossRef]
49. Wolpert, D.H.; Macready, W.G. No free lunch theorems for optimization. IEEE Trans. Evol. Comput. 1997, 1, 67–82. [CrossRef]
50. Ahmadianfar, I.; Heidari, A.A.; Noshadian, S.; Chen, H.; Gandomi, A.H. INFO: An efficient optimization algorithm based on
weighted mean of vectors. Expert Syst. Appl. 2022, 195, 116516. [CrossRef]
51. Hassan, A.Y.; Ismaeel, A.A.; Said, M.; Ghoniem, R.M.; Deb, S.; Elsayed, A.G. Evaluation of weighted mean of vectors algorithm
for identification of solar cell parameters. Processes 2022, 10, 1072. [CrossRef]
Energies 2024, 17, 6087 25 of 25

52. Snášel, V.; Rizk-Allah, R.M.; Izci, D.; Ekinci, S. Weighted mean of vectors optimization algorithm and its application in designing
the power system stabilizer. Appl. Soft Comput. 2023, 136, 110085. [CrossRef]
53. Farhat, M.; Kamel, S.; Atallah, A.M.; Abdelaziz, A.Y.; Tostado-Véliz, M. Developing a strategy based on weighted mean of vectors
(INFO) optimizer for optimal power flow considering uncertainty of renewable energy generation. Neural Comput. Appl. 2023, 35,
13955–13981. [CrossRef] [PubMed]
54. Abd El-Sattar, H.; Houssein, E.H.; Hashim, F.A.; Kamel, S. Optimal design of hybrid renewable energy sources with battery
storage using an efficient weighted mean of vectors algorithm. J. Energy Storage 2024, 87, 111387. [CrossRef]
55. Kahraman, H.T.; Aras, S.; Gedikli, E. Fitness-distance balance (FDB): A new selection method for meta-heuristic search algorithms.
Knowl.-Based Syst. 2020, 190, 105169. [CrossRef]
56. Abd Elaziz, M.; Mirjalili, S. A hyper-heuristic for improving the initial population of whale optimization algorithm. Knowl.-Based
Syst. 2019, 172, 42–63. [CrossRef]
57. Hinislioglu, Y.; Guvenc, U. A novel hyper-heuristic algorithm: An application to automatic voltage regulator. Neural Comput.
Appl. 2024, 36, 21321–21364. [CrossRef]
58. Tanabe, R.; Fukunaga, A.S. Improving the search performance of SHADE using linear population size reduction. In Proceedings
of the 2014 IEEE Congress on Evolutionary Computation (CEC), Beijing, China, 6–11 July 2014; IEEE: New York, NY, USA, 2014;
pp. 1658–1665.

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.