SERBIAN JOURNAL OF ELECTRICAL ENGINEERING

Vol. 21, No. 1, February 2024, 1-37

UDC: 621.316.1:621.33-835 DOI: https://doi.org/10.2298/SJEE2401001P

Cheetah Optimization Algorithm for Simultaneous

Optimal Network Reconfiguration and
Allocation of DG and DSTATCOM with
Electric Vehicle Charging Station
Arvind Pratap1, Prabhakar Tiwari1,
Rakesh Maurya2, Bindeshwar Singh3
Abstract: The potential of Electric Vehicles (EVs) to decarbonize the transport-
ation industry has attracted a lot of attention in recent years in response to growing
environmental concerns. Electric Vehicle Charging Stations (EVCSs) need to be
properly located for widespread EV integration. The distribution system is facing
additional challenges due to inclusion of EVCS. The adverse impacts of EVCS on
the Radial Distribution Network (RDN) may be minimized using Distributed
Generations (DGs) or Distribution Static Compensators (DSTATCOMs) or by
reconfiguring the network. This paper uses a novel optimization technique to solve
the problem of simultaneous optimal placement of EVCS with network
reconfiguration and optimal planning (siting and sizing) of DGs and
DSTATCOMs. The multiple objective functions are considered in order to
minimize the active power losses, the voltage deviation, the investment costs for
DGs and DSTATCOMs, and to increase the voltage stability of the system. A
novel meta-heuristic Cheetah Optimization Algorithm (COA) is used to solve the
optimization problem. To examine the effectiveness of the suggested strategy on
33-bus and 136-bus networks, several scenarios of simultaneous incorporation of
EVCS, DG, and DSTATCOM installations with network reconfiguration are taken
into consideration. The COA results are also compared to the results of grey wolf
optimization and genetic algorithms.

Keywords: Network Reconfiguration, Distributed Generation, Electric Vehicle

charging Station, Real Power Loss reduction, Distribution Static Compensator.

1
Madan Mohan Malaviya University of Technology, Department of Electrical Engineering, Gorakhpur, UP,
India; E-mails: [email protected]; [email protected]
2
Sardar Vallabhbhai National Institute of Technology, Department of Electrical Engineering, Surat, India;
E-mail: [email protected],
3
Kamla Nehru Institute of Technology, Department of Electrical Engineering, Sultanpur, India;
E-mail: [email protected]

1
A. Pratap, P. Tiwari, R. Maurya, B. Singh

Abbreviations
CB Capacitor Banks
COA Cheetah Optimization Algorithm
GOA Grasshopper Optimization Algorithm
GWO Grey Wolf Optimizer
ICRI Investment Cost Reduction Index
IRPL Real Power Loss Index
ITVD Index for Total Voltage Deviation
MOF Multi Objective Function
NR Network Reconfiguration
PSO Particle Swarm Optimization
RPL Real Power Loss
VD Voltage Deviation
VP Volage Profile
VS Voltage Stability
VSI Voltage Stability Index
WNR With Network configuration
WONR Without Network configuration

Symbols
 DG unit's rate of return on capital investment
br Total branches in network
Cdg , Cdst Capital cost of DG [$/kW] and DSTATCOM [$/kVAr]
d, n Dimension of the search space and population size
Nbus Total nodes in network
N dst , N dg Total number of DSTATCOM and DG units
N D , NS DG and DSTATCOM life spans in years
Pj , Pk Effective active power load at j and k bus
bc wc
P , P
loss loss Real power losses with initial condition and different cases
Pss Real power supplied by feeder
Pcs Total real power demand by EVCSs
k k
Pdg , max , P dg , min Maximum and minimum real power injection limits of DG
Pdg ,i Real power drawn from ith DG unit
Pload Total real power load demand of the system
Qload Total reactive power load demand of the system
Q j , Qk Effective reactive power load at j and k bus
Qloss Reactive power loss
k
Qdst k Minimum and maximum reactive power injection limits of
, min , Qdst ,max
DSTATCOM

2
Cheetah Optimization Algorithm for Simultaneous Optimal Network Reconfiguration…

Qdst ,i Reactive power injected by ith DSTATCOM

Qss Reactive power provided by feeder
R jk , X jk Resistance and reactance of jk-branch
wc bc
Vk ,V k Voltage of kth bus with different cases and initial condition
V j , Vk Voltage of jth and kth bus
Vmin Minimum bus voltage in p.u.
 DSTATCOM's asset rate of return

1 Introduction
One of the major sources of air pollution and greenhouse gas emissions is
the utilization of automobiles driven by internal combustion engines. In such a
scenario, Electric Vehicles (EVs) can play an important role in the
decarbonization of the transportation sector. However, the load demand on the
distribution side has increased due to a new class of electric loads. Furthermore,
it results in additional power loss and voltage fluctuations in the distribution
network. Researchers have focused more emphasis on EVCS to accommodate the
rising demand for EVs throughout the world. Integration of renewable
Distributed Generation (DG) technologies into distribution networks has the
potential to significantly improve system performance. Additionally, Distribution
Static Compensator (DSTATCOM) is also utilized to resolve issues associated
with power quality in distribution networks, such as excessive power loss and
voltage instability. The negative impact of EV Charging Station (EVCS) on RDN
can be reduced by the incorporation of the intelligent alternative scheduling
option provided by the combination of renewable DG technology and
DSTATCOM in the RDN. Several technical and economic advantages may be
achieved by integrating DGs and DSTATCOM with EVCS in distribution
systems at the same time. Furthermore, the optimal device allocation is crucial
for maximizing technical and economic benefits. Numerous metaheuristic
optimization strategies had been used by researchers to ascertain the optimal
location of EVCS in coordination with optimal planning (siting and sizing) of DG
and DSTATCOM units on the RDN.
The authors of [1] proposed a chaotic student psychology-based optimization
(CSPBO) method to determine the optimal site and size of DG units to mitigate
the detrimental impact of EVCS on the RDNs. In [2], the authors employed a
Transient Search Optimization (TSO) algorithm for optimal allocation of EVCS
with DG on 25-bus unbalanced RDN with goals of improving the VP and
minimizing the RPL. In [3], the authors presented a hybridized GWO and PSO
(HGWOPSO) algorithm for optimal allocation of DG in order to mitigate the
detrimental impact of EVCS on the RDN. In addition to this, multiple objective

3
A. Pratap, P. Tiwari, R. Maurya, B. Singh

functions are considered to minimize the RPL, VD, and VSI reduction for optimal
allocation of DG to enhance the performance of the EVCS-loaded RDN. In [4],
the authors proposed a methodology to determine the optimal placement and
capacity of EVCS in an unbalanced RDN using PSO algorithm. Furthermore, the
goal of the research is to mitigate the negative effects of EVCS by strategically
placing extra DGs units on an unbalanced RDN. The study also examined the
impact of EVCS deployment on the RPL and VP of the system. In [5], the authors
employed Genetic Algorithm (GA) for simultaneous optimal planning (placing
and sizing) of EVCSs and DGs by optimizing multiple objectives such as
investment costs, system reliability, RPL, VP, and environmental benefits on the
RDN. In [6], the authors employed the Arithmetic Optimization (AO) algorithm
for the optimal allocation of DG units with EVCS to minimize RPL on the RDN.
The authors of [7] presented a method for optimal planning of different types of
DG units to mitigate negative impact of EVCS on the RDN. In addition to this,
Harries Hawk Optimization (HHO) and Teaching-Learning Based Optimization
(TLBO)algorithms were used to minimize the RPL, VD and VSI reduction for
optimal planning of DG with EVCS on RDNs.
The authors of [8] used the Marine Predators' Algorithm (MPA) for optimum
allocation of DG and CB with EVs to reduce RPL and enhance the VSI of an 83-
bus Taiwan distribution system. In [9], a two-stage GOA-based fuzzy multi-
objective scheme was employed for optimal planning of DG, CB, and EVCSs on
the RDNs. The proposed algorithm was used in the first stage to determine the
optimal site and size of DGs and CB to improve system power factor, VP and
minimize the RPL, while the optimal sites and sizes of EVCS on the distribution
system connected with DGs and CB were determined in the second stage. The
authors of [10] used the HGWOPSO algorithm for optimum allocation of EVCS
with CB on the RDNs to lessen the negative impacts of EVCS on the RDNs with
the aims of minimizing the RPL, maximizing the net profit, and improving the
system's reliability. In [11], the authors employed the African Vulture
Optimization (AVO) algorithm for simultaneous optimal placement of EVCS
with allocation of DG and DSTATCOM units on the RDN. The primary objective
was to minimize RPL, VD, and VSI reductions in the distribution system.
However, the economic aspects for the installation of DG and DSTATCOM units
and the significance of NR were not addressed in this study. The authors of [12]
employed a Cooperative Spiral GA with Differential Evolution (CoSGADE)
algorithm for the optimal allocation of EVCS with DG and CB units to minimize
RPL, VD, and VSI reduction of 15, 69, and 118-bus RDNs.
In [13], the authors proposed a hybridized AVO and Pattern Search
(HAVOPS) algorithm for optimal planning of DG and DSTATCOM with NR to
minimize the RPL and VD while maximizing the VSI of the EVCS loaded RDN.
However, the simultaneous approach for optimum placement of EVCS with DG,
DSTATCOM, and NR was not considered. The authors of [14] employed the
4
Cheetah Optimization Algorithm for Simultaneous Optimal Network Reconfiguration…

PSO algorithm for optimal planning of EVCS with DG and NR in order to

minimize the investment and loss costs of an imbalanced RDN. The authors of
[15] used a hybridized Bacterial Foraging Optimization and PSO (HBFOPSO)
algorithm for optimal allocation of EVCS with photovoltaic-based DG system on
RDN in order to minimize RPL, VD, and VSI reduction. In [16], the authors used
the Honey Badger Algorithm (HBA) to tackle the issue of simultaneous optimum
NR and DG allocation, with the goal of lowering RPL, boosting the system's VP,
and minimizing greenhouse gas emissions while taking EV load penetration into
consideration. In order to address NR issues, the authors of [17] presented a
framework for incorporating radiality constraints into mathematical models of
optimization problems for RDN. The authors of [18] used a Harmonic Search
Algorithm (HSA) to solve the NR issue in the presence of DG units with the
objective of minimizing the RPL and enhancing the VP of the RDNs at different
load levels. The authors of [19] suggested a cooperative coevolutionary GA
method for simultaneous fast charging station allocation and NR with the goal of
minimizing investment and energy loss costs.
Table 1
Taxonomy of previous related works and the author’s contribution.
DG, DSTATCOM & NR

Objectives
incorporation of EVCS,
DSTATCOM/ CB

DG / DSTATCOM
Investment Cost of
Simultaneous

Techniques
Reference

DG

NR

RPL

VSI
VD

[1] √ × × × CSPBO √ √ √ ×
[3] √ × × × HGWOPSO × √ √ √
[4] √ × × × PSO × √ √ ×
[6] √ × × × AO algorithm × √ × ×
[7] √ × × × HHO and TLBO × √ √ √
[8] √ × × × MPA × √ × √
[11] √ √ × × AVO algorithm × √ √ √
[12] √ √ × × CoSGADE × √ √ √
[13] √ √ √ × HAVOPS √ √ √ √
[14] √ × √ × PSO √ √ × ×
[15] √ × × × HBFOPSO × √ √ √
[20] √ × √ × Unified PSO × √ × ×
[21] √ × × × GOA × √ × ×
Simulated
[22] √ × × × × √ √ ×
annealing with PSO
[23] √ √ √ √ Bat algorithm × √ × ×
This
√ √ √ √ COA √ √ √ √
paper

5
A. Pratap, P. Tiwari, R. Maurya, B. Singh

The aforementioned literature describes a number of heuristic and meta-

heuristic approaches for optimal placement of EVCS with different compensating
devices. In addition, a MOF considering RPL, VD, and VSI is used in most work.
As far as the authors are aware, this is the first time that COA [24], a nature-
inspired algorithm inspired by the hunting techniques of cheetahs, has been
applied to the problem of maximizing the technical and economic benefits by
achieving simultaneous optimal placement of EVCS with optimal NR and
planning (siting and size) of DG and DSTATCOM units on RDN. Furthermore,
Table 1 shows that very few research has been conducted on the economic
aspects and simultaneous planning of EVCS, DG, and DSTATCOM with optimal
NR. The main contributions to this article are as follows:
– Simultaneous optimal placement of EVCS in coordination with optimal NR
and planning (siting and sizing) of DGs and DSTATCOMs.
– A MOF is formulated, considering technical factors such as reducing IRPL
and ITVD while maximizing VSI, along with economic aspects related to
investment costs for DG and DSTATCOM.
– The computational efficiency of the COA is examined on the 136-bus
network [25], which is a real part of Brazil's Tres Lagoas distribution system.
– COA’s computational efficiency is compared to GWO's [26] and GA’s [27].
The subsequent sections of this paper are structured as follows: The
mathematical problem formulation is discussed in Section 2. The overview and
implementation of the COA on the specified optimization problem are discussed
in Section 3. The simulation findings, which include a comparison to GWO and
GA, are reported in Section 4, while Section 5 summarizes the research results.
2 Mathematical Problem Formulations
The aim of this section is to develop a multi-objective optimization problem
that includes the optimal NR and allocation of multiple EVCS, DG, and
DSTATCOM units with the aim of minimizing several adopted objective
functions. In this work, the EVCS is modelled as a sink that absorbs the system's
actual power during EV battery charging [28], while the DG and DSTATCOM
are modelled as active and reactive power sources, respectively, and operate
under steady-state conditions. The mathematical modelling of EVCS is based on
the work presented in [11, 13, 28], while the DG and DSTATCOM modelling is
based on the work presented in [11, 13] and [29]. A hypothetical RDN with a
branch jk between buses j and k is depicted in Fig. 1. The NR modelling, MOF,
and system constraints are discussed in Subsections 2.1–2.7, subsequently.
2.1 Network reconfiguration
The process of reconfiguring a network involves opening the sectionalizing
switches and closing the tie switches [18, 30]. This switching is carried out such

6
Cheetah Optimization Algorithm for Simultaneous Optimal Network Reconfiguration…

that the network's radial topology is preserved and all loads are powered on. The
optimal reconfiguration issue involves disconnecting or rerouting feeders to
reduce power losses. A severely loaded line may be relieved by rearranging the
distribution system. In this work, the optimal reconfiguration of the distribution
network is determined using COA. After identifying the fundamental loops in the
network using graph theory, the Bus Incidence Matrix (𝐵𝐼𝑀) is computed. If the
determinant of BIM [30] is 1 or -1, then the radial topology continues to hold;
otherwise, the next possible solution is investigated using COA. The Total
Fundamental Loop ( TFL ) in a RDN can be determined using (1).
TFL = br − ( Nbus − 1) . (1)

Branch jk
Substation

Bus j Rjk+ j Xjk Bus k

Pj + jQj Pk + jQk

Fig. 1 – Hypothetical radial distribution network.

2.2 Formulation of weighting factors based multi-objective functions

In this paper, three performance indices (IRPL, ITVD, and VSI) are taken
into account for enhancing the technical advantages of the RDN and one
performance index (ICRI) for reducing the investment costs associated with
installing the DG and DSTATCOM units [31]. In addition to this, the weighting
factors are used to formulate the multiple objective functions in a single
mathematical equation [32], as given by (2). In this work, the decision variables
in the optimization problem are the location of EVCS, switches to be opened for
optimal NR, and locations of compensating devices along with their optimal
sizes. Furthermore, the decision variables are optimized to achieve the
minimization of multiple objective functions.
MOF = min 1  J1 + 2  J 2 + 3  J 3 +4  J 4  , (2)
where 1 , 2 , 3 and 4 are the weighting factors associated with J1 , J 2 , J3 and
J 4 , respectively. The sum of all the weighting factors must be equal to one. The
various adopted objective functions ( J1 , J 2 , J3 , J 4 ) are discussed in Subsections
2.3–2.6, subsequently.

7
A. Pratap, P. Tiwari, R. Maurya, B. Singh

2.3 Minimization of IRPL (J1)

The first objective function J1 is utilized to minimize the RPL of the network
by minimizing the value of IRPL while satisfying the system constraints [13].
Mathematically, J1 can be defined as:
J1 = Minimize ( IRPL ) . (3)
The IRPL for the RDN can be formulated as the ratio of the actual power
loss at different cases to the power loss at base case [33], and which can be
expressed as provided by (4). The real power loss in the branch jk can be
computed using (5):
wc
Ploss
IRPL = bc
, (4)
Ploss
( Pk 2 + Qk 2 )
Pll ( jk ) = R jk . (5)
Vk 2
The total real power loss of the network can be calculated as the sum of
power losses in all branches using (6).
br
Ploss =  Pll ( jk ) , (6)
jk =1

where Ploss denotes total active power loss in the RDN and Pll denotes active
power loss in jk branch.
2.4 Minimization of ITVD (J2)
The second objective function J 2 is utilized to improve the VP of buses in
the RDN by minimizing the value of the ITVD while satisfying the system
constraints [13, 34]. Mathematically, J 2 can be defined as:
J 2 = Minimize ( ITVD ) . (7)
The ITVD for the RDN can be expressed as provided by (8):

 (1 − V ) .
Nbus wc

ITVD = k =1 k
(8)
 (1 − V )
Nbus bc
k =1 k

2.5 Maximization of VSI (J3)

The third objective function J3 is utilized to maximize the voltage stability
index of the system [35]. Mathematically, J3 can be defined as:
J 3 = Minimize ((VSI (k ))-1 ) . (9)

8
Cheetah Optimization Algorithm for Simultaneous Optimal Network Reconfiguration…

The VSI of the receiving end bus k of a branch jk can be expressed as

provided by (10):
VSI (k ) = V j2 − 2( Pk R jk + Qk X jk ) − 4( Pk2 + Qk2 ) ( R jk 2 + X jk 2 ) .
2
(10)

2.6 Minimization of the ICRI (J4)

The optimal planning of DG and DSTATCOM units in the RDN might
reduce the amount of capital expenditure needed to install such units. In addition
to this, ICRI is employed for optimal sizing of DG and DSTATCOM units to
achieve the most cost-effective deployment while adhering to technical
constraints. A lower ICRI value indicates a more economical integration of DG
and DSTATCOM into the RDN while maximizing technical advantages [5, 13,
36]. Mathematically, J 4 can be defined as:
J 4 = Minimize (ICRI ) . (11)
The ICRI can be computed based on the installation costs of DG and
DSTATCOM and their costs at the maximum penetration limit at the optimal
nodes [31], as given by (12). In addition to this, the total cost of DG and
DSTATCOM is converted to annual costs using the economic life of the devices
and the interest rate [5, 13, 36]:


N dg
(1 +  / 100) N D   
N dst
(1 +  / 100) NS 
 dg  dg ,i
C P   dst  dst ,i
+ C Q 

 i =1 (1 +  / 100) ND − 1   i =1 (1 +  / 100) NS − 1 
ICRI = , (12)
CICDGmax + CICDST
max

DG DST
where CIC max
and CIC max
are the total installation costs of DG and DSTATCOM
units at their maximum rating in [$], respectively.
2.7 Constraints
The MOF is subjected to several system constraints that need to be satisfied
[18]. These inequality and equality constraints are discussed in this section as
follows:
Power balance equations: The limitations on the real and reactive power balances
are specified by (13) and (14), respectively.
Pss + i =dg1 Pdg ,i = Pload + Pcs + Ploss ,
N
(13)

Qss + i =dst1 Qdst ,i = Qload + Qloss .

N
(14)
Limits of bus voltage: The voltage at each bus should be maintained within a
certain range to ensure that the system operates stably. The voltage constraint for
each bus in the RDN is expressed by (15):
0.95  Vk ( p.u.)  1.05 . (15)
9
A. Pratap, P. Tiwari, R. Maurya, B. Singh

VSI limitations: The VSI of each bus must be greater than zero to ensure the stable
operation of the system, as specified by (16).
VSI ( k )  0 for k = 2,3, , N bus . (16)
DG and DSTATCOM rating limits: The capacity of individual DG and
DSTATCOM units is limited by their maximum and minimum rating as specified
by (17) and (18), respectively:
Pdgk ,min  Pdgk  Pdgk ,max , (17)

Qkdst ,min  Qdskt  Qkdst ,max . (18)

The constraints for the total installed capacity of DG and DSTATCOM units are
specified by (19) and (20), respectively:


Ndg
i =1
Pdg ,i  Pload , (19)


Ndst
i =1
Qdst ,i  Qload . (20)
Radiality constraint: The constraint that ensures the existence of radial topology
in a network is provided by (21):
BIM = 1 . (21)

3 Proposed Methodology
The details of the Cheetah Optimization Algorithm (COA) and its
application are discussed in Subsections 3.1–3.2, subsequently.
3.1 Overview of COA
COA is a nature-inspired optimization approach proposed by [24]. COA is
inspired by cheetah hunting techniques. Cheetahs hunt by seeking out their prey,
waiting, and then launching an attack. The leave the prey and return home method
is also used in the hunting process to increase population diversity, convergence
performance, and the robustness of the suggested framework. The strategies [24]
involved in COA are presented in Figs. 2a–2d. The flowchart of suggested
algorithm is depicted in Fig. 3.
A) Search strategy
To obtain food, cheetahs have to look throughout their territories (search
space) and in the surrounding area. The searching strategy of cheetahs may be
mathematically modelled by using the notation Z it, j to express the current
location of cheetah i in arrangement j. Different types of prey are encountered by
each individual cheetah. The different states of cheetah constitute a population,
and each individual prey is a decision variable that corresponds to the optimal

10
Cheetah Optimization Algorithm for Simultaneous Optimal Network Reconfiguration…

option. Thereafter, the new location of cheetah i in each arrangement are updated
based on their previous position and an arbitrary step size, as given in (22):
Zit,+j1 = Zit, j + i−,1j it , j , i = 1, 2,, n, j = 1, 2, , d , (22)

ti , j = 0.0001(t / Tht ) (U b ( j ) − Lb ( j ) ) , (23)

where Zit,+j1 , Z it, j , i−,1j and ti , j represent the next position, the current positions,
randomization parameter and the step length of cheetah i in arrangement j,
respectively. Ub and Lb are the upper and lower limits of search space. The
position of the cheetah (leader) is updated in each cheetah arrangement by
assuming the step length ti , j as provided by (23), while the locations of the
remaining cheetahs are updated by assuming the step length ti , j as given by (24).
Locating the neighboring member of the cheetah involves the following steps: i)
Randomly choose a member of the cheetah population. ii) Iterate over the chosen
members. iii) For each selected individual, check whether it is the last member of
the population. iv) If it's the last member, choose the preceding member as the
neighboring agent. v) Choose the next member as the neighbor agent if it's not
the last.
ti , j = 0.0001(t / Tht )abs( X leader − Zit, j ) + Ap , (24)
where t is the current hunting time, and Tht is the maximum length of hunting
time. X leader is the leader position, and Ap is random number which is equal to
0.001 times of the rounded value of a random number generated between 0 and 1
(i.e., 1 if the random number > 0.9, else 0).
B) Sit-and-wait strategy
Cheetahs come across their prey when they are actively looking for it. Every
action that the cheetah does in this circumstance has the potential to notify the
target animal, causing it to break up the pursuit and run away. To avoid this
concern, the cheetah could try to ambush its prey by hiding behind some shrubs
or lying down on the ground. Therefore, when operating in this mode, the cheetah
waits for its prey to get closer to it (Fig. 2b). This behavior may be modelled as
(25):
Zit,+j1 = Zit, j , (25)

where Zit,+j1 and Z it, j are updated and current locations of cheetah i in arrangement
j, respectively. This technique needs the COA to not alter all the cheetahs in each
group at the same time in order to prevent premature convergence and improve
hunting.

11
A. Pratap, P. Tiwari, R. Maurya, B. Singh

(a) (b)

c
a
b

(c) (d)

Fig. 2 – (a) Searching strategy; (b) Sitting-and-waiting strategy;

(c) Searching strategy; (d) Capturing strategy of Cheetah [24].

C) Attack strategy
Cheetahs are able to successfully hunt because of their speed and their
suppleness. Cheetahs are known to run up to their prey before making an attack.
After a short period of time, the prey becomes aware of the cheetah's attack and
runs away. The cheetah moves very swiftly in pursuit of the prey along the
direction of the interception, as can be shown in Fig. 2c. To put it another way,
the cheetah tracks its prey and modifies its motion such that it can outmaneuvers
the prey at some time. Because the cheetah's next location is close to the prey's
previous position, as illustrated in Fig. 2d, the prey needs to run away and change
its position as rapidly as possible in order to live. During a collective hunt, it is
possible for each cheetah to modify its location in response to the prey and the
leader. These cheetah assault tactics are mathematically stated as (26) – (28):
Zit,+j1 = Z Bt , j + TFi , j it , j (26)

Hit, j = X inb, j − Zit, j , (27)

12
Cheetah Optimization Algorithm for Simultaneous Optimal Network Reconfiguration…

TFi , j = abs(rand )exp ( rand / 2) sin(2rand ) , (28)

where Z Bt , j is the location of the prey in arrangement j; X inb, j is the neighbor
position; TFi , j and ti , j are Cheetah i’s turning factor and interaction factor in
arrangement j, respectively. Cheetahs, while in an attacking mode, utilize their
maximal speed to swiftly narrow the distance between themselves and their prey;
hence, Z Bt , j is employed in (26).

D) Hypotheses
In hunting, the searching or attacking approach is employed randomly,
although the searching method becomes more probable with time owing to
depleting cheetah energy. In other circumstances, the search strategy comes first,
whereas the attack method is used for high values of 𝑡 to produce better results.
If rand2  rand3 , the sit-and-wait approach is chosen; otherwise, one of the
seeking or assaulting strategies is chosen based on a random value R, as given by
(29) and (30):
Eq. (26), if rand 2  rand3 & R  rand 4 ,
t +1 
Z i, j = Eq. (22), if rand 2  rand3 & R  rand 4 , (29)
Eq. (25), if rand  rand ,
 2 3

2(1−t / Tht )
R= e (2rand1 − 1) , (30)
where rand1, rand2 and rand3 are random number from the range [0, 1]. rand4 is
a random number from 0 to 3. Tuning rand3 controls the switching rate between
sit-and-wait and two additional strategies. Higher rand4 values prioritize the
exploitation phase, whereas lower values prioritize the exploration phase.
3.2 Application of COA
In this work, COA is employed to solve the problem of simultaneous optimal
placement of EVCS with NR and planning of DG and DSTATCOM units on the
RDN. During the optimization process, the Cheetah positions are considered as
potential solution for the problem of optimal placement of EVCSs with NR and
planning of DG and DSTATCOM units on the RDN. The algorithm searches for
the optimal combination of these variables that minimizes the MOF, while
satisfying the system constraints. The following are the suggested steps for
solving the optimal allocation problem of EVCS with compensating devices and
NR using COA.
Enter the bus data, branch data, and EVCS load data, as well as algorithm
parameters (d, n and maximum iterations (Itmax)).
Generates an initial population of cheetahs randomly in the search space using
(31), as given by (32).
13
A. Pratap, P. Tiwari, R. Maurya, B. Singh

Zpi = Lb + [rand (1, d )  (Ub − Lb )] , (31)

 z1,1 , z1,2 , z1,3 , z1, d 
z , z , z , z 
Zp = 
2,1 2,2 2,3 2, d 
, (32)
 
 
 zn ,1 , zn ,2 , zn ,3 , zn ,d 
where Zpi represented a cheetah or a solution vector of dimensions d; Zp
represents the initial population of cheetahs in the search space as a matrix of
size ( n  d ) ; rand is a random number generator. Each column of the matrix
Zp represents a decision variable of the optimization problem. Furthermore,
each row of the matrix Zp represents a possible solution (i.e., a cheetah), as
given in (33):
Zpi = [TS1 , TS 2, , TSsw , EV1 , EV2 , , EVKev , D1 , D2 , ..., DNdg ,
Switchesstatus EVCS location DG location
(33)
DS1 , DS2 , , DS Ndg , C1 , C2 , , C Ndst , CS1 , CS 2 , , CS Ndst ],
DG size DSTATCOM location DSTATCOM size

where the subscript Kev and Tsw denote total number of EVCS and opened
switches for optimal NR, respectively; ( D1 , D2 , , DNdg ) are the locations of
DG units, and the corresponding DG sizes are denoted by
( DS1 , DS2 , , DS Ndg ) . The locations of DSTATCOM units are denoted by
(C1 , C2 , , C Ndst ) , and the corresponding DSTATCOM sizes are denoted by
(CS1 , CS2 , , CS Ndst ) . The locations for multiple EVCS installation are
denoted by (EV1 , EV2 , , EVKev ) , and the open switches for optimal NR are
denoted by (TS1 , TS2 , ,TSsw ) .
For each randomly produced solution, compute the fitness value.
Verify the system’s constraints for each initial population solution.
Select the best solution from the initial population as the initial leader position.
Also, initialize the current position of ith cheetah, neighbour position, and the
prey position. The home position represents the best solution found so far, the
leader position represents the best solution among the neighbouring solutions,
and the prey position represents the global best solution.
Initialize the iteration, hunting time counters, and the function evaluation
counter.
Start the main loop: The algorithm starts the main loop that will continue until
the maximum iterations achieved ( FEC  Itmax ).

14
Cheetah Optimization Algorithm for Simultaneous Optimal Network Reconfiguration…

Select a random set of m cheetahs from the population and algorithm iterates
over the selected cheetahs and applies the different strategy to each one of
them.
For each cheetah in the selected set, the algorithm selects a neighbour cheetah.
The COA utilizes three different strategies for the cheetahs to search for prey
and improve their solutions, as given by (34):
 Attack Strategy if rand2  rand3 & R  rand4 ,

Strategy =  Search Strategy if rand 2  rand3 & R  rand4 , (34)
 Sit &Wait Strategy if rand 2  rand3 .

Search Strategy: When cheetah does not detect any prey nearby (i.e.,
rand2  rand3 & R  rand4 ) it uses this strategy to move to a new location in
the search space. The new position is calculated using (22) where a random
number i−,1j ti , j is added to the current position of ith cheetah to introduce
some randomness into the movement as given by (35):
Zit,+j1 = [TS1c , c
, TSTsw , EV1c , c
, EVKev , D1c , , DNc dg ,
(35)
DS1c , , DS Nc dg , C1c , , CNc dst , CS1c , , CS Nc dst ] + i−,1j ti , j ,
where superscript c is used to denote the current position of i th cheetah.
Attack Strategy: When a cheetah detects prey nearby (i.e.,
rand2  rand3 & R  rand4 ), it uses this strategy to move towards the prey’s
location and attempt to catch it. The new position is calculated using (26)
where TFi , j it , j is added to the prey position to introduce some randomness
into the attack as given by (36):
Zit,+j1 = [TS1P , P
, TSTsw , EV1P , , EVKPev , D1P , , DNPdg ,
(36)
DS1P , , DS NPdg , C1P , , CNPdst , CS1P , , CS NPdst ] + TFi , j  ti , j ,
where superscript P is used to denote the prey position in arrangement j.
Sit-and-wait strategy: When a cheetah detects prey nearby but is not close
enough to attack (i.e., rand2  rand3 ), it uses this strategy to stay in its current
position and wait for the prey to come closer as given by (37).
Zit,+j1 = [TS1c , c
, TSTsw , EV1c , c
, EVKev , D1c , , DNc dg ,
(37)
DS1c ,..DS Nc dg , C1c , , CNc dst , CS1c , , CS Nc dst ],
Updates the Function Evaluation Counter (FEC) and the best cheetah in the
population.
Update hunting time counters ( t ).
15
A. Pratap, P. Tiwari, R. Maurya, B. Singh

If t  Tht rand and the leader position does not change for a period of time,
then implement the leave the prey and return home strategy and modify the
leader position (i.e., replace the position of member i with the prey position).
Iterate through steps 7–13 until the maximum allowed number of iterations is
achieved.
Return the best cheetah’s position (i.e., optimal solution).
Generate the initial population of cheetahs
Assign the problem data, d and
Start Zpi (i=1, 2, 3,...,n) and evaluate the fitness of
the initial population size (n)
each cheetah

Initialize the population s

Define t=0, iter=1, itmax, function evaluation counter (FEC)
home, leader, and prey
and Hunting time (Tht)=60*d/10
solutions

No
FEC < itmax

Yes No
Select m members of cheetahs
Set i =0 i<m
randomly between 2 to n
Yes
Assign the neighbour
Set j=0
agent of member i

j<d No
Yes
Choose random numbers
Find ϒ, TF, μ , H, and R
rand2 and rand3

Set rand4 (random number from 0

Yes rand2 rand3
to 3)

Search Strategy Attack Strategy Sit-and-wait Strategy

R rand4 No
No Yes

Determine member i's new Determine member i's new Using equ. (25), determine
position using equ. (26) position using equ. (22) member i's new position

j=j+1

Update member i and leader's

solutions

i=i+1

t=t+1

Leave the prey and Yes No

t > rand * Tht iter = iter+1
go back home

Global best solution Stop

Fig. 3 – COA flowchart.

16
Cheetah Optimization Algorithm for Simultaneous Optimal Network Reconfiguration…

4 Results and Discussions

This research considers renewable energy type DG, which generates power
at unity power factor, while DSTATCOM delivers reactive power. EVCS are
presumed to have 25 charging outlets, and each charger’s power demand is
40 kW. Therefore, 25 EVs can be charged at the same time via charging stations.
The maximum power consumption of an EVCS is 1 MW when all of its charging
points are utilized for charging EVs at the same time. The best location in the
distribution network must be chosen for the placement of charging stations. Due
to the placement of EVCS, the network's real power loss rises, and the voltage
profile of the buses is disrupted. As a result, DGs or DSTATCOMs must be well-
positioned to mitigate the impacts of EVCS on the RDN.
In this study, the power flow analysis on the RDN is done using the
forward/backward sweep approach in this work [37]. Simulations are run on an
Intel i7 computer with 3.0GHz and 8GB of RAM using MATLAB. The suggested
COA is applied to two distribution systems. These systems are the 33-bus [38]
and the 136-bus RDN [25]. The parameters for ICRI calculation are taken from
[4, 5, 39], as shown in Table 2. To achieve a balance between technical and
economic advantages, proper consideration of weighting factors is essential. In
this study, an analytical test on a 136-bus RDN with EVCS and DG units is
conducted to determine the most effective values of weighting factors, enabling
the assessment of the optimal value of MOF with a balanced technical and
economic advantage. The impact of different combinations of weighting factor
values on MOF components are shown in Table 3.
Table 2
Economic data for DG and DSTATCOM allocation.
Parameters Cdg Cdst α ν ND NS
Values 500 $/kW 50$/KVAr 10 10 20 30

4.1 Different case studies

Four operational cases are presented to evaluate the effectiveness of the COA
and study the impact of EVCSs in combination with DGs, DSTATCOMs, and
optimal NR on the system performance.
Base case: RDN without EVCS, DG, and DSTATCOM.
Case 1: Integration of EVCSs into RDN (with and without NR).
Case 2: Simultaneous integration of EVCSs and DSTATCOMs into
RDN (with and without NR).
Case 3: Simultaneous integration of EVCSs and DGs into RDN (with and
without NR).
Case 4: Simultaneous integration of EVCSs, DGs, and DSTATCOMs
into RDN (with and without NR).

17
 Total
Weighting factors
RPL EVCS DG DG cost
IRPL VSI ITVD DG size [MW] and location
[kW] location capacity [$]
β1 β2 β3 β4 [MW]
14, 106, 84, 1.699 (86), 2.77 (44), 1.437 (101),
0.7 0.1 0.1 0.1 0.3466 0.8560 111.03 0.6184 1.642 (100), 1.8 (64), 2.805 (76)
12.152 713709.1
90, 28, 52
90, 9, 82, 1.718 (122), 2.85 (64), 1.175 (7),
0.6 0.2 0.1 0.1 0.3712 0.8643 118.92 0.6583 2.9 (3), 1.393 (100), 1.863 (86)
11.899 698833.5
106, 131, 49
107, 91, 53, 2.798 (100), 1.377 (64), 1.764
0.5 0.2 0.2 0.1 0.3658 0.8714 117.19 0.6178 (122), 1.463 (18), 2.062 (76), 2.176 11.640 683650.1
84, 32, 9
(77)
7, 53, 85, 2.372 (40), 2.727 (18), 0.787 (54),
0.5 0.3 0.1 0.1 0.3869 0.8608 123.96 0.6553 2.316 (76), 1.986 (2), 1.893 (100)
12.080 709433.7
108, 32, 90
84, 106, 109, 2.306 (100), 1.924 (40), 1.046 (43),
A. Pratap, P. Tiwari, R. Maurya, B. Singh

0.4 0.3 0.2 0.1 0.4502 0.8871 144.23 0.6495 2.101 (84), 3 (3), 1.681 (86)
12.058 708173.9
57, 9, 93

18
106, 49, 109, 1.997 (2), 2.816 (41), 1.122 (19),
0.3 0.3 0.3 0.1 0.4562 0.8891 146.14 0.6175 12.037 706932
Table 3

93, 29, 14 1.823 (64), 2.752 (40), 1.528 (18)

111, 32, 84, 0.899 (18), 2.171 (2), 2.093 (84),

0.6 0.1 0.1 0.2 0.4295 0.8719 137.59 0.6645 1.844 (106), 3 (100), 1.83 (19)
11.836 695146.1
16, 106, 48
81, 14, 35, 1.492 (76), 0.894 (64), 1.598 (50), 3
0.5 0.2 0.1 0.2 0.4611 0.8763 147.73 0.6841 (122), 3 (40), 0.2 (101)
10.184 598088.9
132, 107, 49
7, 117, 53, 0.218 (122), 0.586 (53), 3 (131),
0.4 0.2 0.2 0.2 0.4921 0.8887 157.64 0.6925 2.47 (86), 2.343 (2), 2.052 (40)
10.669 626561.5
106, 131, 14
Weighting factor values and various indices.

36, 54, 84, 1.501 (18), 3 (54), 1.232 (100),

0.3 0.3 0.2 0.2 0.4714 0.8894 151.00 0.6281 2.953 (122), 0.589 (64), 1.2 (86)
10.475 615195.6
108, 35, 95
95, 28, 53, 0.959 (40), 1.257 (18), 2.059 (122),
0.4 0.2 0.1 0.3 0.4809 0.8616 154.07 0.7157 0.581 (41), 2.822 (76), 0.838 (64)
8.517 500197.5
84, 106, 134
7, 108, 37, 1.486 (37), 2.89 (100), 0.351 (101),
0.3 0.2 0.2 0.3 0.5335 0.8831 170.93 0.7066 1.985 (40), 0.2 (64), 1.737 (2)
8.649 507930.8
49, 10, 37
Cheetah Optimization Algorithm for Simultaneous Optimal Network Reconfiguration…

According to Table 3, the DG penetration decreases as the 4 value

increases. Consequently, the technical advantage diminishes. Also, there is close
competition between the first six combinations of the weighting factors.
Furthermore, the values of 1, 2, 3 and 4 at 0.7, 0.1, 0.1, and 0.1 have the lowest
MOF but have a lesser impact on ITVD and VSI. The values of 1, 2, 3 and 4
at 0.5, 0.2, 0.2, and 0.1 have a stronger impact on ITVD, IRPL, VSI, and the DG
investment cost. As a result, the values of 1, 2, 3 and 4 considered are 0.5,
0.2, 0.2, and 0.1, respectively.
The case studies for 33 and 136 bus networks are discussed in Subsection
5.2-5.3, subsequently.
4.2 For 33-bus RDN
The one-line diagram of the 33-bus RDN is depicted in Fig. 4 [30, 38]. The
system has a total real power load requirement of 3.715 MW and a reactive power
load requirement of 2.3 MVAr. The network consists of 33 buses, 37 branches,
32 selection switches, 5 tie switches, and 5 fundamental loops. The system
operates at a rated line voltage of 12.66 kV with a base MVA rating of 10.
Furthermore, three EVCS are considered for integration into the 33-bus RDN to
meet customer demand and ensure EVCS availability for a significant number of
EV customers. The different operational cases of the 33-bus RDN are as follows:

Fig. 4 – Standard 33-bus RDN.

4.2.1 Case 1
In this case, EVCSs are integrated into the system without considering DGs
and DSTATCOMs to evaluate the impact of charging station load on the RDN.
Furthermore, technical aspects such as IRPL, ITVD, and Minimum VSI (MVSI)
are taken into consideration for MOF. Additionally, the impacts of EVCS
installation on RDN are analyzed using two scenarios, WONR and WNR.
A) Scenario 1 (WONR)
In this scenario, the MOF is optimized by determining the optimal bus
positions for EVCS installations using the COA. Table 4 presents the simulation
results for this scenario. The potential bus locations considered for EVCS
19
A. Pratap, P. Tiwari, R. Maurya, B. Singh

installations are 2, 19, and 20, as provided in Table 5. The resulting IRPL, ITVD
and MVSI values are 1.1933, 1.05387 and 0.6614, respectively. Furthermore, the
value of RPL and minimum bus voltage are 251.78 kW and 0.9018 p.u.,
respectively. In addition to this, the VS tolerance limit for each bus voltage is not
kept within the permissible range of ±5%.
Table 4
Performance analysis of the 33-bus RDN with four cases.
Total size of
Vmin (p. u.)
DG [MW]/ RPL
Cases Scenario IRPL MVSI ITVD & bus
DSTATCOM [kW]
number
[MVAr]
Base
--- --- --- --- 0.9037 & 18 --- 211
case
Case WONR 1.1933 0.6614 1.05387 0.9018 & 18 --- 251.78
1 WNR 0.8638 0.7679 0.66483 0.9361 & 32 --- 182.27

Case WONR 0.8799 0.7388 0.80340 0.9271 & 18 ---/1.12 185.66

2 WNR 0.6582 0.8377 0.55472 0.9567 &32 ---/1.15 138.87

Case WONR 0.4696 0.9087 0.25053 0.9764 &33 3.180/--- 99.07

3 WNR 0.3496 0.9145 0.24959 0.9779 & 33 3.357/--- 73.77

Case WONR 0.2370 0.9065 0.22467 0.9757 &18 3.034/1.079 50.01

4 WNR 0.1754 0.9446 0.14244 0.9859 & 24 3.160/1.065 37.00

B) Scenario 2 (WNR)
In this scenario, the potential bus locations for the EVCS installation are
determined in coordination with the optimal network reconfiguration by
optimizing the MOF via the COA. In comparison to the WONR scenario, when
the RDN is reconfigured with three EVCSs, RPL decreases to 182.27 kW from
251.78 kW, VSI increases to 0.7679 from 0.6614, and the minimum bus voltage
increases to 0.9361 from 0.9018 p. u., as shown in Table 4. Moreover, scenario
2 further lowers the IRPL value by 27.61 percent in comparison to WONR
scenario. From Fig. 5, it is seen that the VPs of each node do not follow the VS
limit of ± 5%.
4.2.2 Case 2
The integration of EVCSs has an additive influence on RPL and a negative
impact on the VP and VSI, as shown in case 1. The inclusion of DSTATCOM
units at the optimal nodes can compensate for these disruptions. In this case, the
MOF considers both technical factors (IRPL, ITVD, and VSI) and economic
factors (the cost of investing in DSTATCOM installation). Furthermore, the
EVCS are optimally allocated in coordination with DSTATCOM by optimizing
the MOF using COA under two distinct scenarios.

20
Cheetah Optimization Algorithm for Simultaneous Optimal Network Reconfiguration…

A) Scenario 1 (WONR)
In this scenario, the impact of EVCS' integration in coordination with
DSTATCOM units on IRPL, ITVD, VSI and ICRI is studied for the 33-bus RDN,
and results are provided in Table 4. Three DSTATCOMs are installed at buses
31, 30, and 14, with penetrations of 0.3127, 0.4732, and 0.3641 MVAr,
respectively. The potential bus locations for EVCS installation are 2, 20, and 19.
Due to reactive power compensation by DSTATCOM, it can be noticed that RPL
and ITVD are significantly decreased as compared to case 1, whereas VSI is
significantly improved. In addition to this, the resultant values of IRPL, ITVD,
and MVSI are 0.8799, 0.8034, and 0.7388, respectively. The value of RPL and
minimum bus voltage are 185.66 kW and 0.9271 p. u., respectively. As shown in
Fig. 5, the system's voltage profile improves when EVCS and DSTATCOM are
simultaneously placed on RDN. From Table 4, it is seen that installing EVCS,
with or without NR, does not achieve the desired results of minimizing IRPL,
ITVD, and VSI reduction.
B) Scenario 2 (WNR)
In this scenario, the impact of EVCS' integration in coordination with
DSTATCOM units and NR on various performance indices is studied for the 33-
bus RDN, and results are provided in Table 4. The switches S7, S9, S14, S37,
and S32 are opened during NR. The candidate bus locations for EVCS are 2, 3,
and 19. The potential bus locations for DSTATCOM installation are 30, 32, and
17, with penetrations of 0.6431, 0.2591, and 0.2478 MVAr, respectively.
Compared to scenario 1 of case 2, the suggested COA technique reduces RPL to
138.87 kW from 185.66 kW, a reduction of 25.2%. In addition to this, the MVSI
value is improved from 0.7388 to 0.8377, and the minimum bus voltage is
increased from 0.9271 to 0.9567 p. u. From Table 4, it is seen that improvements
in IRPL reduction, ITVD reduction, and MVSI maximization are higher when
compared to scenario 1 of case 2.
4.2.3 Case 3
As observed in case 2, the inclusion of EVCS in coordination with
DSTATCOM units at the optimal nodes does not yield the desired results of
maximizing IRPL reduction, ITVD reduction, VSI improvement and ICRI
reduction. As a result, the EVCS are optimally allocated in coordination with DG
units by optimizing the MOF via COA under two distinct scenarios, WONR and
WNR. In this case, the MOF considers both technical factors (IRPL, ITVD, and
VSI) and economic factors (cost of investing in DG).

21
A. Pratap, P. Tiwari, R. Maurya, B. Singh

A) Scenario 1 (WONR)
In this scenario, the impact of EVCS' integration in coordination with DG
units on IRPL, ITVD, VSI and ICRI is studied for the 33-bus RDN, and results
are provided in Table 4. Three DG units are installed at optimal buses 25, 30, and
13, with penetrations of 0.66, 1.35, and 1.17 MW, respectively. The potential bus
locations for EVCS installation are 2, 3, and 19. As a result of active power
compensation by DGs units, it is seen that IRPL, ITVD, and VSI reduction have
all minimized significantly compared to cases 1 and 2 of the 33-bus RDN. The
values of RPL and minimum bus voltage are 99.07 kW and 0.9764 p. u.,
respectively. In comparison to cases 1 and 2, integrating EVCS with DGs units
significantly improved the VP of the 33-bus network, as shown in Fig. 5.
Furthermore, the power loss in each branch is also reduced by integrating EVCS
with DGs.
Table 5
Optimal site and size of compensating devices for allocation on 33-bus RDN.
DG sizes DSTATCOM
EVCSs
Cases Scenario [MW] and sizes [MVAr] Switches opened
locations
sites and sites
WONR 2, 19, 20 --- --- S33-S37
Case 1
WNR 2, 3, 19 --- --- S9, S7, S14, S32, S28
0.3127 (31),
WONR 2, 20, 19 --- 0.4732 (30), S33-S37
0.3641(14)
Case 2
0.6431 (30),
WNR 2, 3, 19 --- 0.2591 (32), S7, S9, S14, S37, S32
0.2478 (17)
0.66 (25),
WONR 3, 2, 19 1.35 (30), --- S33-S37
1.17 (13)
Case 3
0.72 (14),
WNR 3, 19, 2 0.75 (8), --- S11, S6, S34, S32, S28
1.887 (29)
0.777 (9), 0.2531 (30),
WONR 19, 2, 32 0.580 (10), 0.6984 (31), S33-S37
1.677 (31) 0.1275 (8)
Case 4
1.03 (21), 0.290 (25),
WNR 3, 19, 2 0.72 (18), 0.265 (18), S11, S7, S34, S31, S24
1.41 (29) 0.510 (30)

B) Scenario 2 (WNR)
In this scenario, three EVCS are optimally located with the simultaneous
incorporation of NR and three DG units. Table 4 demonstrates the simulation
result for this scenario. Fig. 6 represents a reconfigured 33-bus RDN with
optimally positioned EVCS and DG units. The candidate bus locations for EVCS
installation are 2, 3, and 19, as provided in Table 5. Three DG units are located

22
Cheetah Optimization Algorithm for Simultaneous Optimal Network Reconfiguration…

at buses 14, 8, and 29, with penetrations of 0.72, 0.75, and 1.887 MW,
respectively. The switches S11, S6, S34, S32, and S28 are opened during NR.
Compared to scenario 1 of case 3, the suggested COA technique reduces RPL to
73.77 kW from 99.07 kW, a reduction of 25.53%. In addition to this, the MVSI
value is improved to 0.9145 from 0.9087 and the minimum bus voltage is
increased from 0.9764 to 0.9779 p. u. In comparison to scenario 1 of case 3,
scenario 2 significantly improved the VP of the 33-bus RDN. From Figs. 7−10,
it is seen that improvements in IRPL reduction, ITVD reduction, and MVSI
maximization are higher compared to cases 1 and 2 due to active power
compensation by three DGs and optimal NR. According to Fig. 11, scenario 2
also minimizes power loss at each of the branches of the 33-bus RDN.

Fig. 5 – VP for 33-bus RDN.

Fig. 6 – Reconfigured 33-bus RDN with multiple EVCSs and DGs units.

4.2.4 Case 4
To assess the impact of EVCSs installation with simultaneous integration of
DG and DSTATCOM units on RDN, the MOF is optimized using COA under
multiple scenarios. The following are the two possible scenarios:

23
A. Pratap, P. Tiwari, R. Maurya, B. Singh

A) Scenario 1 (WONR)
In this scenario, the potential bus locations for the EVCS installation are
determined in coordination with the optimal site and size of the DGs and
DSTATCOMs units by optimizing the MOF using COA. The results of the 33-
bus RDN for this scenario are shown in Table 4. In this scenario, the actual and
reactive power compensation by DGs and DSTATCOMs units considerably
increases the IRPL reduction, ITVD reduction, and VSI maximization when
compared to cases 1, 2, and 3. The resultant values of IRPL, ITVD, and MVSI
are 0.2370, 0.22467, and 0.9065, respectively. Also, the minimum bus voltage is
0.9757 p.u., which is found on bus 18. The improved VPs of each bus for this
scenario is shown in Fig. 5. In addition, simultaneous integration of EVCS, DGs,
and DSTATCOMs lowers power loss in each branch of the network, as shown in
Fig. 11.
B) Scenario 2 (WNR)
In this scenario, the potential bus locations for the EVCS installation are
determined in coordination with the optimal NR as well as the optimal location
and size of the DGs and DSTATCOMs units by optimizing the MOF via COA.
Table 4 shows the results of the 33-bus network based on scenario 2 of case 4.
During NR, the switches S11, S7, S34, S31, and S24 are opened. Also, Table 4
shows that case 4’s scenario 2 is much superior to case 4’s scenario 1 for the
33-bus network in terms of minimizing the IRPL, ITVD, and VSI reduction.
Compared to scenario 1 of case 4, scenario 2 has a further 26.01 % drop in IRPL.
Meanwhile, the minimum bus voltage rises from 0.9757 p.u. to 0.9859 p. u. and
the VSI increases from 0.9065 to 0.9446. The improvement in IRPL reduction,
ITVD reduction, and MVSI is highest for scenario 2 of case 4 compared to other
cases of the 33-bus RDN, as shown in Figs. 7 – 9. Also, the power loss in each
branch is significantly reduced, as shown in Fig. 11.

1,2 0,95
1 0,86
0,8 0,77
IRPL

MVSI

0,6
0,68
0,4
0,2 0,59
0 0,5
WONR
WNR
WONR
WNR
WONR
WNR
WONR
WNR
WONR
WNR
WONR
WNR
WONR
WNR
WONR
WNR

Case 1 Case 2 Case 3 Case 4 Case 1 Case 2 Case 3 Case 4

Fig. 7 – IRPL for 33-bus RDN. Fig. 8 – MVSI for 33-bus RDN.

24
Cheetah Optimization Algorithm for Simultaneous Optimal Network Reconfiguration…

1,2 400
1
300
0,8

RPL
ITVD

0,6 200
0,4
100
0,2
0 0
WONR
WNR
WONR
WNR
WONR
WNR
WONR
WNR

WONR
WNR
WONR
WNR
WONR
WNR
WONR
WNR
Case 1 Case 2 Case 3 Case 4 Case 1 Case 2 Case 3 Case 4

Fig. 9 – ITVD for 33-bus RDN. Fig. 10 – RPL for 33-bus RDN.

Fig. 11 – Branch losses of 33-bus RDN (Case 1-4).

4.3 For 136-bus network

The one-line diagram of 136-bus RDN is shown in Fig. 12 [30, 25]. In
addition, the system’s total real and reactive power load requirements are 18.31
MW and 7.93 MVAr, respectively. The network consists of 136 buses, 156
branches, 135 selection switches, 21 tie switches and 21 fundamental loops. The
rated line voltage of the system is 13.88 kV, and its base MVA rating is 10. To
fulfil customer demand and assure EVCS availability for a significant number of
EV customers, six EVCS have been considered for installation on the 136-bus
RDN. The different operational cases are as follows:
4.3.1 Case 1
In this case, EVCSs are integrated into the system without considering DGs
and DSTATCOMs to evaluate the impact of charging station load on RDN.
Furthermore, technical aspects such as IRPL, ITVD, and VSI are taken into
consideration for MOF. Additionally, the impacts of EVCS installation on RDN
are analyzed using two scenarios, WONR and WNR.

25
A. Pratap, P. Tiwari, R. Maurya, B. Singh

A) Scenario 1 (WONR)
In the same manner as the 33-bus RDN, the impact of EVCS' integration on
IRPL, ITVD, and VSI is studied for the 136-bus RDN, and results are presented
in Table 6. The candidate bus locations for EVCS installations in the 136-bus
RDN are 76, 64, 2, 100, 67, and 40. Also, Table 6 shows that the resultant IRPL,
ITVD, and MVSI values are 1.1342, 1.0388, and 0.7499, respectively. The value
of RPL and minimum bus voltage are 363.35 kW and 0.9306 p. u. Fig. 13 shows
the voltage profiles of each node do not follow the VS tolerance limit of ± 5%.

Fig. 12– Standard 136-bus RDN.

26
Cheetah Optimization Algorithm for Simultaneous Optimal Network Reconfiguration…

B) Scenario 2 (WNR)
In this scenario, the impacts of EVCS installation with optimal NR on IRPL,
ITVD, and VSI are studied for the 136-bus RDN, and the results are presented in
Table 6. The resulting IRPL, ITVD, and MVSI values are 0.9287, 0.9240, and
0.8672, respectively. The percentage reductions in IRPL and ITVD relative to the
WONR scenario are 18.12 percent and 11.04 percent, respectively. The minimum
voltage of 0.9650 p.u. is achieved at bus 61. From Fig. 13, it is seen that the VPs
of each node follow the permissible range of VS limits of ± 5%.
Table 6
Performances analysis of the 136-bus RDN with cases 1 and 2.

Case 1 Case 2
Scenario WONR WNR WONR WNR
Vmin (p. u.)/bus 0.9306/117 0.9650/61 0.9566/117 0.9655/106
IRPL 1.1342 0.9287 1.0168 0.9538
MVSI 0.7499 0.8672 0.8375 0.8689
ITVD 1.0388 0.924 0.7109 0.7017
Total size [MVAr]
--- --- ---/3.918 ---/3.913
of DSTATCOM
RPL [kW] 363.35 297.51 325.74 305.58
76, 64, 2, 86, 64, 100, 122, 40, 86, 100, 43, 3, 18, 65, 64,
EVs locations
100, 67, 40 76, 40 64, 122, 4 122
1.3403 (49), 1.0216 (109),
0.4732 (28), 0.7153 (34),
DSTATCOM sizes 0.7556 (113), 0.5279 (82),
--- ---
[MVAr]/ sites 0.5006 (52), 1.0124 (9),
0.1724 (108), 0.100 (78),
0.6760 (94) 0.5355 (134),
S136, S137, S136, S9, S138,
S138, S38, S51, S38, S140, S141,
Switches opened S136-S156 S141, S54, S143- S136-S156 S54, S143, S144-
S152, S106, S152, S106, S126,
S126, S155, S156 S128, S156

4.3.2 Case 2
Similar to case 2 of the 33-bus network, the EVCS are optimally allocated in
coordination with DSTATCOM by optimizing the MOF using COA under two
distinct scenarios.
A) Scenario 1 (WONR)
The impacts of EVCS' integration in coordination with DSTATCOM units
on various performance indices are studied for the 136-bus RDN. Six
DSTATCOMs are installed at buses 49, 28, 113, 52, 108, and 94, with
penetrations of 1.3403, 0.4732, 0.7556, 0.5006, 0.1724, and 0.6760 MVAr,
respectively, as presented in Table 6. The candidate bus locations for EVCS
27
A. Pratap, P. Tiwari, R. Maurya, B. Singh

installation are 40, 86, 100, 64, 122, and 4. From Table 6, it is observed that the
simultaneous integration of EVCS and DSTATCOM has an incremental impact
on VSI while having a decremental impact on the IRPL and ITVD of the 136-bus
RDN. The resultant IRPL, ITVD, and MVSI values are 1.0168, 0.7109, and
0.8375, respectively. The minimum bus voltage of 0.9566 p. u. is obtained at bus
117. Also, Table 6 shows that improvements in RPL reduction, ITVD reduction,
and MVSI are higher when compared to case 1 of the 136-bus network. Figs. 13
and 14 show that the VPs of each bus is improved and the power loss in each
branch is reduced in scenario 1 of case 2.
B) Scenario 2 (WNR)
The integration of EVCSs with DSTATCOMs and optimal NR resulted in a
reduction in ITVD and IRPL while increasing VSI of the 136-bus system, as
shown in Table 6. Compared to scenario 1 of case 2, the suggested COA
technique reduces the value of RPL from 325.74 kW to 305.58 kW, a reduction
of 6.18%. In addition to this, the MVSI value is improved from 0.8375 to 0.8689,
and the minimum bus voltage is increased from 0.9566 to 0.9655 p. u. According
to Fig. 13, adding EVCS with DSTATCOM units has improved the VPs of the
136-bus RDN in comparison to case 1. Furthermore, the power loss in each
branch is also reduced, as illustrated in Fig. 14.
4.3.3 Case 3
Similar to case 3 of the 33-bus RDN, the EVCS are optimally allocated in
coordination with DG units by optimizing the MOF via COA under two distinct
scenarios, WONR and WNR. In this case, the MOF considers both technical
factors and economic factors.
A) Scenario 1 (WONR)
In this scenario, the impacts of EVCS' integration in coordination with DG
units on various performance indices are studied for the 136-bus RDN. The
potential bus locations for EVCSs are 90, 9, 82, 106, 131, and 49 on 136 bus
RDN. The optimal size and site of DG units are presented in Table 7. The
resulting values of IRPL, ITVD and MVSI are 0.3658, 0.6178 and 0.8714,
respectively. The value of RPL and minimum bus voltage are 117.2 kW and
0.9662 p. u., respectively. In comparison to cases 1 and 2, integrating EVCS with
DGs units significantly improved the voltage profile of the 136-bus network, as
shown in Fig. 13. Furthermore, the power loss in each branch is reduced by
integrating EVCS with DGs, as illustrated in Fig. 14. From the Table 7, it is noted
that this scenario shows significantly higher improvements in terms of RPL
reduction, voltage deviation reduction, and VSI maximization when compared to
cases 1 and 2 of the 136-bus RDN.

28
Cheetah Optimization Algorithm for Simultaneous Optimal Network Reconfiguration…

B) Scenario 2 (WNR)
Similar to the 33-bus network, the integration of EVCSs with DGs and
optimal NR resulted in a reduction in ITVD and IRPL while increasing VSI in
the 136-bus network, as shown in Table 7. The potential bus locations for EVCS
installation on the 136-bus RDN are 86, 92, 76, 18, 122, and 100. The opened
switches for optimal NR and the location of DG units with their optimal sizes are
also provided in Table 7. Compared to scenario 1 of case 2, the suggested COA
technique reduces the value of RPL from 117.2 kW to 98.3 kW, a reduction of
16.12 %. In addition to this, the MVSI value is improved from 0.8714 to 0.9099,
and the minimum bus voltage is increased from 0.9662 to 0.9766 p. u. The
improved VP of each bus of the 136-bus network is shown in Fig. 13. Also,
scenario 2 minimizes power loss at each of the branches of the 136-bus RDN, as
illustrated in Fig. 14.
Table 7
Performances analysis of the 136-bus RDN with cases 3 and 4.
Case 3 Case 4
Scenario WONR WNR WONR WNR
Vmin (p. u.)/bus 0.9662/117 0.9766/117 0.9734/135 0.9797/97
IRPL 0.3658 0.3068 0.2638 0.2294
MVSI 0.8714 0.9099 0.8967 0.9202
ITVD 0.6178 0.5848 0.322 0.309
Total size of
11.5814/ 11.720/
DG/DSTATCOM 11.641/--- 12.151/---
3.7947 3.884
in [MW/MVAr]
RPL [kW] 117.2 98.3 84.5 73.5
90, 9, 82, 86, 92, 76, 19, 2, 64, 54, 86, 100,
EVs locations
106, 131, 49 18, 122, 100 100, 40, 76 64, 65, 76
2.7982(107), 2.310 (92), 2.015(14), 1.148 (108),
1.3774 (91), 1.450 (82), 2.284(106), 2.9 (53),
DG sizes [MW]/ 1.764 (53), 2.359 (52), 1.13(93), 1.285 (89),
sites 1.4625 (84), 2.185 (14), 2.463(49), 1.337(106),
2.0623 (32), 2.176 (106), 1.294(84), 2.228 (74),
2.1762 (9) 1.671 (32) 2.396(29) 2.821 (39)
0.735(32), 0.6434 (14),
0.73(54), 0.5279 (32),
DSTATCOM
0.437(95), 1.0459 (26),
sizes [MVAr]/ --- ---
0.667(107), 1.048 (136),
sites
0.515(16), 0.4983 (74),
0.71(106) 0.120 (50)
S73, S137-S140- S68, S9, S138,
S142, S62, S144, S27, S25, S141-
Switches opened S136-S156 S145, S134, S147- S136-S156 S145, S83, S147-
S152, S107, S126, S152, S106, S126,
S128, S156 S155, S156

29
A. Pratap, P. Tiwari, R. Maurya, B. Singh

4.3.4 Case 4
Similar to case 4 of the 33-bus RDN, the impacts of EVCS' integration in
coordination with DG and DSTATCOM units on various performance indices are
studied for the 136-bus RDN. The following are the two possible scenarios of this
case:
A) Scenario 1 (WONR)
The potential bus locations for EVCS installation on the 136-bus RDN are
19, 30, 64, 100, 56, and 76. The optimal location of DG and DSTATCOM units,
along with the optimal size are provided in Table 7. The resulting values of IRPL,
ITVD, and MVSI are 0.2638, 0.3220, and 0.8967, respectively. Furthermore, the
improved voltage profiles for each bus in this scenario are shown in Fig. 13,
where bus 135 has a minimum voltage of 0.9734 p. u. Also, the active power loss
in each branch is reduced due to the simultaneous integration of EVCS with DGs
and DSTATCOMs units, as illustrated in Fig. 14.

Fig. 13 – Voltage profile of 136-bus RDN (Cases 1-4).

Fig. 14 – Branch losses of 136-bus RDN (Cases 1-4),

30
Cheetah Optimization Algorithm for Simultaneous Optimal Network Reconfiguration…

1,2 1
1 0,8
0,8

MVSI
IRPL

0,6
0,6
0,4
0,4
0,2 0,2
0 0

WNR

WNR

WNR

WNR
WONR

WONR

WONR

WONR
WNR

WNR

WNR

WNR
WONR

WONR

WONR

WONR
Case 1 Case 2 Case 3 Case 4 Case 1 Case 2 Case 3 Case 4
Fig. 15 – IRPL for 136-bus RDN. Fig. 16 – VSI for 136-bus RDN.

B) Scenario 2 (WNR)
The simultaneous optimal integration of EVCS, DG, and DSTATCOM units
with NR has a higher impact on achieving the desired results of minimizing IRPL,
ITVD, and VSI reduction on the 136-bus network, as shown in Table 6.
Moreover, scenario 2 further lowers the IRPL value by 13.01 percent in
comparison to WONR scenario. The minimum voltage is 0.9797 p. u. at bus 97,
while the MVSI value is 0.9202. Also, the simultaneous integration of EVCS with
DGs, DSTATCOMs and optimal NR significantly decreases power loss in each
branch, as shown in Fig. 14. According to Table 6, the improvement in IRPL
reduction, ITVD reduction, and VSI maximization is highest for scenario 2 of
case 4 compared to other cases of the 136-bus network, as shown in Figs. 15 – 18.
Furthermore, the scenario 2 of case 4 achieves a better VP when compared to
other scenarios of the 136-bus RDN.
400
1
RPL (kW)

300
0,75
ITVD

0,5 200

0,25 100
0 0
WONR
WNR
WONR
WNR
WONR
WNR
WONR
WNR

WNR

WNR

WNR

WNR
WONR

WONR

WONR

WONR

Case 1 Case 2 Case 3 Case 4 Case 1 Case 2 Case 3 Case 4

Fig. 17 – ITVD for 136-bus RDN. Fig. 18 – RPL for 136-bus RDN.

4.4 Comparisons of COA results from GWO and GA results

In this section, the results of COA are compared with the results of GWO
[26] and the GA [27]. As the problem's complexity grows, it becomes more likely
that the algorithm will get trapped at the local optimum value. In addition to this,
31
A. Pratap, P. Tiwari, R. Maurya, B. Singh

scenario 2 of case 4 is considered to assess the robustness of the COA for both
the 33-bus and 136-bus RDNs. The dimension of the problem in scenario 2 of
case 4 is 17 for 33-bus RDN (i.e., 5 for NR, 6 for DSTATCOM sites and sizes, 6
for DG sites and sizes, and 3 for EVCS placement) and 51 for 136-bus RDN (i.e.,
21 for NR, 12 for DSTATCOM sites and sizes, 12 for DG sites and sizes, and 6
for EVCS placement). Each algorithm is run 25 times to assess the COA's
effectiveness in solving optimization problems. To assess the robustness of COA,
their performance characteristics are evaluated in terms of the best, mean, and
worst values of objective functions, as illustrated in Fig. 19. The optimal solutions
achieved by each technique for 33 bus and 136 bus networks are shown in Table
8. From Fig. 19 and Table 8 it is observed that COA is a more reliable and
effective approach than GWO and GA algorithms for solving complex
optimization problems. Fig. 20 depicts the convergence curves of the best
solution produced by adopted algorithms for scenario 2 of case 4 of the 136-bus
RDN. In addition to this, it is seen that the COA exhibits a better convergence
rate compared to GWO and GA.

0.28 Case 4: WNR (33-bus RDN)

0.64 Case 4: WNR (136-bus RDN)
0.26
0.60
0.24
MOF
MOF

0.56
0.22
0.52
0.20
0.48
0.18
COA GWO GA COA GWO GA

(a) (b)
Fig. 19 – Comparisons of COA from GWO and GA results.

1
GA GWO COA

0,8
MOF

0,6

0,4
1

16

31

46

61

76

91

106

121

136

151

166

181

196

Iterations
Fig. 20 – Comparisons of convergence curve of COA from GWO and GA.

32
 Switch EVCS DG sizes DSTATCOM RPL
System Algorithm Scenario IRPL ITVD VSI
opened locations [MW]/sites sizes [MVAr]/sites [kW]

S11, S7, 0.897/21, 0.293/25,

GA WNR S14, S31, 3, 19, 2 0.811/18, 0.339/21, 0.2083 0.1633 0.9425 43.96
S24 1.457/29 0.442/31
S11, S7, 0.888/21, 0.367/25,
33-bus
GWO WNR S14, S31, 3, 19, 2 0.761/17, 0.229/22, 0.1815 0.1912 0.9314 38.29
system
S25 1.541/29 0.525/30
S11, S7, 1.03/21, 0.29/25,
COA WNR S34, S31, 3, 19, 2 0.72/18, 0.265/18, 0.1754 0.1424 0.9446 37.00
S24 1.41/29 0.51/30
S70, S137,
S13, S38, 2.209 (107), 0.307 (105),
S51, S141, 1.259 (99), 0.413 (120),
5, 65, 2,
S53, S62, 3.00 (74), 1.035 (36),

33
GA WNR 71, 76, 0.3387 0.4264 0.9007 108.5
S144, S90, 1.861 (126), 0.739 (81),
122
Table 8

S107, S126, 2.254 (84), 0.644 (101),

S155, S156, 1.464 (26) 0.619 (71)
S145-S151,
S136-S139,
2.404 (52), 0.72 (9),
136- S51, S94,
1.507 (69), 0.661 (134),
bus S128, S141- 76, 64, 2,
1.507 (106), 0.45 (81),
system GWO WNR S145, S80, 100, 67, 0.2603 0.3412 0.9108 83.4
1.992 (82), 0.7 (111),
S126, S147- 40
2.404 (11), 0.701 (57),
S151, S105,
2.308 (48) 0.546 (56)
S156
S68, S9, 1.149 (108), 0.643 (14),
S138, S27, 2.90 (53), 0.528 (32),
54, 86,
S25, S141- 1.285 (89), 1.046 (26),
COA WNR 100, 64, 0.2294 0.3090 0.9202 73.5
S145, S83, 1.337 (106), 1.048 (136),
65, 76
S147-S152, 2.228 (74), 0.498 (74),
S126, S155, 2.821 (39) 0.12 (50)
Comparison of COA performance for 33 and 136-bus RDNs (Scenario 2 of case 4).

S156
Cheetah Optimization Algorithm for Simultaneous Optimal Network Reconfiguration…
A. Pratap, P. Tiwari, R. Maurya, B. Singh

5 Conclusions
Cleaner transportation may be possible with the advent of EVs. EVCSs are
developed to promote the development of EVs. However, widespread EV
adoption necessitates a robust and efficient charging infrastructure, which results
in additional burden on the power distribution side. The objective of this research
is to examine the impact of EVCS together with optimal NR and planning (siting
and size) of DGs and DSTATCOMs on the RDN. The main goal is the
maximization of technical and economic advantages of the system. Moreover, the
results of COA are compared to the results of GWO and GA to evaluate its
accuracy. Two distinct distribution systems have been subjected to four operating
cases, including EVCSs, DGs, and DSTATCOMs. The impacts of EVCS
deployment with DGs and DSTATCOM on RDN are also examined in two
scenarios: WONR and WNR. The following is a summary of the key conclusions
from the simulation:
– The integration of DG or DSTATCOM is advantageous in minimizing the
detrimental impact of EVCS on power system performance. However,
incorporating DG and DSTATCOM simultaneously with NR is a more
efficient approach to reducing the negative impact of EVCS on the system
performance.
– Significant improvements in IRPL reduction, ITVD reduction, and VSI
enhancement are achieved when EVCS is optimally allocated together with
DG
– and DSTATCOM, as well as with NR.
– Also, the simultaneous approach for optimal planning of EVCS, DG, and
DSTATCOM with NR results in substantial technical and economic
benefits.
– Better convergence characteristics of the COA have been found in
comparison to the GWO and GA.
– The COA provides more accurate solutions compared to the GWO and GA.

6 References
[1] K. Balu, V. Mukherjee: Optimal Allocation of Electric Vehicle Charging Stations and
Renewable Distributed Generation with Battery Energy Storage in Radial Distribution System
Considering Time Sequence Characteristics of Generation and Load Demand, Journal of
Energy Storage, Vol. 59, March 2023, p. 106533.
[2] J. S. Bhadoriya, A. R. Gupta, M. Zellagui, N. K. Saxena, A. K. Arya, A. K. Bohre: Optimal
Allocation of Electric Vehicles Charging Station in Distribution Network Beside DG Using
TSO, Ch. 29, Planning of Hybrid Renewable Energy Systems, Electric Vehicles and
Microgrid, Springer, Singapore, 2022.

34
Cheetah Optimization Algorithm for Simultaneous Optimal Network Reconfiguration…

[3] M. Bilal, M. Rizwan, I. Alsaidan, F. M. Almasoudi: AI-Based Approach for Optimal

Placement of EVCS and DG with Reliability Analysis, IEEE Access, Vol. 9, November 2021,
pp. 154204 − 154224.
[4] M. S. Kumar Reddy, K. Selvajyothi: Optimal Placement of Electric Vehicle Charging Station
for Unbalanced Radial Distribution Systems, Energy Sources, Part A: Recovery, Utilization,
and Environmental Effects, February 2020, pp. 1 − 15.
[5] S. Pazouki, A. Mohsenzadeh, S. Ardalan, M.- R. Haghifam: Simultaneous Planning of PEV
Charging Stations and DGs Considering Financial, Technical, and Environmental Effects,
Canadian Journal of Electrical and Computer Engineering, Vol. 38, No. 3, Summer 2015, pp.
238 − 245.
[6] K. Kathiravan, P. N. Rajnarayanan: Application of AOA Algorithm for Optimal Placement of
Electric Vehicle Charging Station to Minimize Line Losses, Electric Power Systems
Research, Vol. 214, Part A, January 2023, p. 108868.
[7] V. K. B. Ponnam, K. Swarnasri: Multi-Objective Optimal Allocation of Electric Vehicle
Charging Stations and Distributed Generators in Radial Distribution Systems Using
Metaheuristic Optimization Algorithms, Engineering, Technology & Applied Science
Research, Vol. 10, No. 3, June 2020, pp. 5837 − 5844.
[8] N. Dharavat, S. Kumar Sudabattula, V. Suresh: Optimal Integration of Distributed Generators
(DGs) Shunt Capacitors (SCs) and Electric Vehicles (EVs) in a Distribution System (DS)
using Marine Predator Algorithm, International Journal of Renewable Energy Research, Vol.
12, No. 3, September 2022, pp. 1637 − 1650.
[9] S. R. Gampa, K. Jasthi, P. Goli, D. Das, R. C. Bansal: Grasshopper Optimization Algorithm
Based Two Stage Fuzzy Multiobjective Approach for Optimum Sizing and Placement of
Distributed Generations, Shunt Capacitors and Electric Vehicle Charging Stations, Journal of
Energy Storage, Vol. 27, February 2020, p. 101117.
[10] M. Bilal, M. Rizwan: Integration of Electric Vehicle Charging Stations and Capacitors in
Distribution Systems with Vehicle-to-Grid Facility, Energy Sources, Part A: Recovery,
Utilization, and Environmental Effects, May 2021, pp. 1 − 30
[11] A. Pratap, P. Tiwari, R. Maurya, B. Singh: Minimisation of Electric Vehicle Charging Stations
Impact on Radial Distribution Networks by Optimal Allocation of DSTATCOM and DG
Using African Vulture Optimisation Algorithm, International Journal of Ambient Energy,
Vol. 43, No. 1, August 2022, pp. 8653 − 8672.
[12] K. E. Adetunji, I. W. Hofsajer, A. M. Abu-Mahfouz, L. Cheng: A Novel Dynamic Planning
Mechanism for Allocating Electric Vehicle Charging Stations Considering Distributed
Generation and Electronic Units, Energy Reports, Vol. 8, November 2022, pp. 14658 − 14672.
[13] A. Pratap, P. Tiwari, R. Maurya, B. Singh: A Novel Hybrid Optimization Approach for
Optimal Allocation of Distributed Generation and Distribution Static Compensator with
Network Reconfiguration in Consideration of Electric Vehicle Charging Station, Electric
Power Components and Systems, Vol. 51, No. 13, April 2023, pp. 1302 − 327.
[14] M. S. Kumar Reddy, K. Selvajyothi: Investment Analysis for Optimal Planning of Electric
Vehicle Charging Station on a Reconfigured Unbalanced Radial Distribution System,
Electrical Engineering, Vol. 104, No. 3, June 2022, pp. 1725 − 1739.
[15] W. S. Tounsi Fokui, M. J. Saulo, L. Ngoo: Optimal Placement of Electric Vehicle Charging
Stations in a Distribution Network with Randomly Distributed Rooftop Photovoltaic Systems,
IEEE Access, Vol. 9, September 2021, pp. 132397 − 132411.

35
A. Pratap, P. Tiwari, R. Maurya, B. Singh

[16] S. Thumati, S. Vadivel, M. Venu Gopala Rao: Honey Badger Algorithm Based Network
Reconfiguration and Integration of Renewable Distributed Generation for Electric Vehicles
Load Penetration, International Journal of Intelligent Engineering and Systems, Vol. 15, No.
4, August 2022, pp. 329 − 338.
[17] M. Lavorato, J. F. Franco, M. J. Rider, R. Romero: Imposing Radiality Constraints in
Distribution System Optimization Problems, IEEE Transactions on Power Systems, Vol. 27,
No. 1, February 2012, pp. 172 − 180.
[18] R. S. Rao, K. Ravindra, K. Satish, S. V. L. Narasimham: Power Loss Minimization in
Distribution System Using Network Reconfiguration in the Presence of Distributed Generation,
IEEE Transactions on Power Systems, Vol. 28, No. 1, February 2013, pp. 317 − 325.
[19] A. Pahlavanhoseini, M. S. Sepasian: Scenario-Based Planning of Fast Charging Stations
Considering Network Reconfiguration Using Cooperative Coevolutionary Approach, Journal
of Energy Storage, Vol. 23, June 2019, pp. 544 − 557.
[20] D. Kothona, A. S. Bouhouras: A Two-Stage EV Charging Planning and Network
Reconfiguration Methodology towards Power Loss Minimization in Low and Medium
Voltage Distribution Networks, Energies, Vol. 15, No. 10, May 2022, p. 3808.
[21] S. Kaveripriya, V. Suresh, S. Suresh Kumar, K. Abinaya: Optimal Allocation of DERs in
Distribution System in Presence of EVs, Proceedings of the 2nd International Conference on
Power Engineering Computing and Control (PECCON), Chennai, India, December 2019, pp.
77 − 88.
[22] H. Tang, J. Wu: Multi-Objective Coordination Optimisation Method for DGs and EVs in
Distribution Networks, Archives of Electrical Engineering, Vol. 68, No. 1, February 2019,
pp. 15 − 32.
[23] S. R. Salkuti: Binary Bat Algorithm for Optimal Operation of Radial Distribution Networks,
International Journal on Electrical Engineering and Informatics, Vol. 14, No. 1, March 2022,
pp. 148 − 160.
[24] M. A. Akbari, M. Zare, R. Azizipanah-abarghooee, S. Mirjalili, M. Deriche: The Cheetah
Optimizer: A Nature-Inspired Metaheuristic Algorithm for Large-Scale Optimization
Problems, Scientific Reports, Vol. 12, No. 1, December 2022, p. 10953.
[25] J. R. S. Mantovani, F. Casari, R. A. Romero: Reconfiguracao de Sistemas de Distribuicao
Radiais Utilizando o Criterio de Queda de Tensao, SBA Controle and Automacao, Vol. 11,
No. 3, September 2000, pp. 150 − 159.
[26] S. Mirjalili, S. M. Mirjalili, A. Lewis: Grey Wolf Optimizer, Advances in Engineering
Software, Vol. 69, March 2014, pp. 46 − 61.
[27] D. E. Goldberg, J. H. Holland: Genetic Algorithms and Machine Learning, Machine Learning,
Vol. 3, No. 2-3, October 1988, pp. 95 − 99.
[28] S. Deb, K. Tammi, K. Kalita, P. Mahanta: Impact of Electric Vehicle Charging Station Load
on Distribution Network, Energies, Vol. 11, No. 1, January 2018, p. 178.
[29] V. Janamala, U. Kamal Kumar, T. K. Sai Pandraju: Future Search Algorithm for Optimal
Integration of Distributed Generation and Electric Vehicle Fleets in Radial Distribution
Networks Considering Techno-Environmental Aspects, SN Applied Sciences, Vol. 3, March
2021, p. 464.
[30] T. T. Tran, K. H. Truong, D. N. Vo: Stochastic Fractal Search Algorithm for Reconfiguration
of Distribution Networks with Distributed Generations, Ain Shams Engineering Journal, Vol.
11, No. 2, June 2020, pp. 389 − 407.

36
Cheetah Optimization Algorithm for Simultaneous Optimal Network Reconfiguration…

[31] S. Rao Gampa, D. Das: Optimum Placement and sizing of DGs Considering Average Hourly
Variations of Load, International Journal of Electrical Power and Energy Systems, Vol. 66,
March 2015, pp. 25 − 40.
[32] K. Balu, V. Mukherjee: Optimal Siting and Sizing of Distributed Generation in Radial
Distribution System Using a Novel Student Psychology-Based Optimization Algorithm,
Neural Computing and Applications, Vol. 33, No. 22, November 2021, pp. 15639 − 15667.
[33] K. R. Devabalaji, K. Ravi: Optimal Size and Siting of Multiple DG and DSTATCOM in
Radial Distribution System Using Bacterial Foraging Optimization Algorithm, Ain Shams
Engineering Journal, Vol. 7, No. 3, September 2016, pp. 959 − 971.
[34] A. Mohamed Imran, M. Kowsalya, D. P. Kothari: A Novel Integration Technique for Optimal
Network Reconfiguration and Distributed Generation Placement in Power Distribution
Networks, International Journal of Electrical Power & Energy Systems, Vol. 63, December
2014, pp. 461 − 472.
[35] M. Chakravorty, D. Das: Voltage Stability Analysis of Radial Distribution Networks,
International Journal of Electrical Power & Energy Systems, Vol. 23, No. 2, February 2001,
pp. 129-135.
[36] E. S. Oda, A. M. Abd El Hamed, A. Ali, A. A. Elbaset, M. Abd El Sattar, M. Ebeed: Stochastic
Optimal Planning of Distribution System Considering Integrated Photovoltaic-Based DG and
DSTATCOM Under Uncertainties of Loads and Solar Irradiance, IEEE Access, Vol. 9,
February 2021, pp. 26541 − 26555.
[37] R. D. Zimmerman, C. E. Murillo-Sánchez, R. J. Thomas: MATPOWER: Steady-State
Operations, Planning, and Analysis Tools for Power Systems Research and Education, IEEE
Transactions on Power Systems, Vol. 26, No. 1, February 2011, pp. 12 − 19.
[38] M. A. Kashem, V. Ganapathy, G. B. Jasmon, M. I. Buhari: A Novel Method for Loss
Minimization in Distribution Networks, Proceedings of the International Conference on
Electric Utility Deregulation and Restructuring and Power Technologies, London, UK, April
2000, pp. 251 − 256.
[39] E. S. Oda, A. A. Abdelsalam, M. N. Abdel-Wahab, M. M. El-Saadawi: Distributed Generations
Planning Using Flower Pollination Algorithm for Enhancing Distribution System Voltage
Stability, Ain Shams Engineering Journal, Vol. 8, No. 4, December 2017, pp. 593 − 603.

37