Received February 9, 2021, accepted March 6, 2021, date of publication March 17, 2021, date of current version March

25, 2021.
Digital Object Identifier 10.1109/ACCESS.2021.3066329

Recent Methodology-Based Gradient-Based

Optimizer for Economic Load Dispatch Problem
SANCHARI DEB1 , DIAA SALAMA ABDELMINAAM 2,3 , MOKHTAR SAID 4,

AND ESSAM H. HOUSSEIN 5

1 VTT Technical Research Centre, 02100 Espoo, Finland
2 Faculty of Computers and Artificial Intelligence, Benha University, Benha 12311, Egypt
3 Faculty of Computers Science, Misr International University, Cairo 11311, Egypt
4 Electrical Engineering Department, Faculty of Engineering, Fayoum University, Fayoum 63514, Egypt
5 Faculty of Computers and Information, Minia University, Minia 61519, Egypt

Corresponding author: Diaa Salama Abdelminaam ([email protected])

This work was supported in part by the Misr International University, under Grant DSA28211231302952.

ABSTRACT Economic load dispatch (ELD) in power system problems involves scheduling the power
generating units to minimize cost and satisfy system constraints. Although previous works propose solutions
to reduce CO2 emission and production cost, an optimal allocation needs to be considered on both cost
and emission—leading to combined economic and emission dispatch (CEED). Metaheuristic optimization
algorithms perform relatively well on ELD problems. The gradient-based optimizer (GBO) is a new
metaheuristic algorithm inspired by Newton’s method that integrates both the gradient search rule and
local escaping operator. The GBO maintains a good balance between exploration and exploitation. Also,
the possibility of the GBO getting stuck in local optima and premature convergence is rare. This paper
tests the performance of GBO in solving ELD and CEED problems. We test the performance of GBO
on ELD for various scenarios such as ELD with transmission losses, CEED and CEED with valve point
effect. The experimental results revealed that GBO has been obtained better results compared to eight
other metaheuristic algorithms such as Slime mould algorithm (SMA), Elephant herding optimization
(EHO), Monarch butterfly optimization (MBO), Moth search algorithm (MSA), Earthworm optimization
algorithm (EWA), Artificial Bee Colony (ABC) Algorithm, Tunicate Swarm Algorithm (TSA) and Chimp
Optimization Algorithm (ChOA). Therefore, the simulation results showed the competitive performance of
GBO as compared to other benchmark algorithms.

INDEX TERMS Gradient-based optimizer (GBO), economic load dispatch (ELD), combined economic and
emission dispatch (CEED), metaheuristics, optimization.

ABBREVIATIONS
ELD Economic Load Dispatch SMA Slime mould algorithm
PSO Particle Swarm Optimization MBO Monarch butterfly optimization
QBA Quantum Bat Algorithm CEED Economic and Emission Dispatch
MBA Mine Blast Algorithm TCO Termite Colony Optimization
MFO Moth Flame Optimizer ACS Artificial Cooperative Search
ALO Ant Lion Optimization SHO Spotted Hyena Optimizer
SSA Salp Swarm Algorithm
GWO Grey Wolf Optimizer
GA Genetic Algorithm
ACCS Adaptive Charged System Search
ADFA Ameliorated Dragonfly Algorithm
ISFS Improved Stochastic Fractal Search
GBO Gradient-Based Optimizer
LFA Lighting Flash Algorithm
ACTO Aggrandized CTO
MSSA Modified Social Spider Algorithm
MODE Multi-objective DE
TSA Tunicate Swarm Algorithm
The associate editor coordinating the review of this manuscript and NSGA Non Dominated Sorting GA
approving it for publication was Qingli Li . PDE Pareto DE

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/
44322 VOLUME 9, 2021
S. Deb et al.: Recent Methodology-Based GBO for ELD Problem

EHO Elephant herding optimization Metaheuristics have proved to be useful in various problems,
EWA Earthworm optimization algorithm including this one [11]–[13].
CTO Class Topper Optimization The Economic Load Dispatch (ELD) plays an important
BA Bat Algorithm role in powering the electrical loads and reducing emissions
FA Firefly Algorithm with the guarantee of meeting the equality and inequality
WOA Whale Optimization Algorithm constraints [14]–[17]. Metaheuristic optimization algo-
AEFA Artificial Electric Field Algorithm rithms (MHs) performs relatively well on the ELD problem.
ABC Artificial Bee Colony This work tests the performance of Gradient Based Opti-
MFO Moth Flame Optimization mizer (GBO) in solving ELD and CEED. GBO is a new
SOS Symbiotic Organism Search metaheuristic inspired by the gradient based Newton method
FPA Flower Pollination Algorithm involving Gradient search rule (GSR) and local escaping
ChOA Chimp Optimization Algorithm operator (GEO). GBO has good balance between exploration
DE Differential evolution and exploitation.
WCA Water Cycle Algorithm This work uses a new met-heuristic algorithm called
MSA Moth search algorithm Gradient-Based Optimizer (GBO) [18] was developed by
TLBO Teaching Learning Based Optimization Ahmadianfar et al. in 2020, which was one of the most
promising algorithms for solving different variants of ELD.
GBO is a metaheuristic inspired by the gradient based New-
I. INTRODUCTION ton method involving Gradient search rule (GSR) and local
Engineers of 21st century are curious about the increasing escaping operator (GEO). To evaluate various characteris-
complexity of the societal and technological challenges tics of the GBO, 28 mathematical test functions were first
such as the parameter extraction problem in photovoltaic used and then six engineering problems were optimized
(PV) [1] and the problem of the Economic Load Dispatch by the GBO. Moreover, the exploitative, exploratory, and
(ELD) [2]. The ELD involves involves minimizing produc- local optima avoidance of GBO was also investigated using
tion costs by allocating power produced by each power unimodal, multi-modal and composition problems. Finally,
system unit economically [3]. Although some solutions are the results show that GBO was capable of finding excel-
proposed to reduce emissions and production costs [4] by lent solutions compared to other well-regarded optimizers.
considering both cost and emission optimally, leading to com- Also, the possibility of getting stuck in local optima and
bined economic and emission dispatch (CEED), the ELD and premature convergence is rare in GBO. The performance
CEED solutions require efficient optimization algorithms. of GBO on ELD is tested for various scenarios such as
Several metaheuristic optimization algorithms are ELD with transmission losses, CEED, CEED with valve
proposed to solve a wide range of real-life problems. For point effect and for various test networks. The perfor-
example, nature-inspired algorithms mimic the biological, mance of GBO is compared with compared with eight other
physical, or environmental processes [5]–[7]. The meta- metaheuristic algorithms such as Slime mould algorithm
heuristic algorithms’ versatility and gradient-free features (SMA) [19], Elephant herding optimization (EHO) [20],
consider black-box problems in addition to the theoreti- Monarch butterfly optimization (MBO) [21], Moth search
cal developments and significant advantages. The resulting algorithm (MSA) [22], Earthworm optimization algorithm
search space is not limited, making the algorithms scalable (EWA) [23], Artificial Bee Colony (ABC) Algorithm [24],
for solving different problems. Real problems are solved Tunicate Swarm Algorithm (TSA) [25] and Chimp Optimiza-
more effectively since solutions are not restricted to locally tion Algorithm (ChOA) [26].
optimal approaches. The metaheuristic algorithms are applied The rest of the paper is organized as follows. Section III
in various fields and proved helpful [8]–[12]. elaborates the ELD problem, then an overview for the
Several metaheuristic optimization algorithms are Gradient-Based Optimizer (GBO) is presented in Section IV.
proposed to solve a wide range of real-life problems. For The obtained findings and discussion is introduced in
example, nature-inspired algorithms mimic the biological, Section V. Finally, Section VI concludes the work.
physical, or environmental processes [5], [6]. Nature-inspired
algorithms are designed to mimic the biological, physical, II. RELATED WORK
or environmental processes they are modeled after [7], [8]. An overview of metaheuristics used for solving the different
The metaheuristic algorithms’ versatility and gradient-free variants of ELD is as shown in Table 1. In [27], authors have
features consider black-box problems in addition to the used CTO and ACTO to solve ELD as well as CEED. It was
theoretical developments and significant advantages. The observed that CTO performs better than other metaheuris-
resulting search space is not limited, making the algorithms tics such as TLBO, DE, GA, PSO etc. Dynamic ELD was
scalable for solving different problems. Real problems are solved by hybrid PSO TCO [28]. The hybridization of PSO
solved more effectively since solutions are not restricted to and TCO favoured faster convergence and produced better
locally optimal approaches. The metaheuristic algorithms quality of solutions. The authors performed ELD considering
are applied in various fields and proved helpful [9], [10]. renewable resources [29]. A modified version of BA was used

VOLUME 9, 2021 44323

 S. Deb et al.: Recent Methodology-Based GBO for ELD Problem

for solving the problem. In [30], a quantum inspired BA was CEED in [70]. An improved version of DE was used to solve
solved for using ELD. It was observed that the premature con- ELD with and without valve point effect in [71].
vergence was avoided by the modified quantum inspired BA. From Table 1, it is observed that metaheuristics such as
Dynamic ELD considering renewable sources was solved by PSO, GA, BA, GWO are widely used by researchers in
multiswarm PSO [31]. Quantum inspired BA was used for solving different variants of ELD. Despite the availability and
solving ELD considering valve point effect [32]. In [33], use of different metaheuristics for solving ELD, researchers
the authors utilized ACO to solve different variants of ELD. are still proposing new and novel algorithms for its solution.
Non-convex ELD was solved by ACS in [17]. The authors The prime motivation behind this is the No Free Lunch (NFL)
have hybridized BA and FA for solving ELD and CEED theorem [72]. NFL theorem states that a single algorithm does
in [34]. A self-adaptive version of Jaya algorithm was used not perform equally well on all the optimization problems.
for solving ELD in [34]. It was observed that the modified Hence, it is justified to propose new more efficient algorithms
version of Jaya algorithm performed better that the basis Jaya and improve the existing algorithms.
algorithm and TLBO in solving the ELD problem. In [35],
the authors proposed MBS for solving CEED with valve point III. ECONOMIC LOAD DISPATCH PROBLEM
effect. A drift mechanism in the self-adaptive version of PSO ELD is one of the prime and complex problems of mod-
was introduced and used for solving the ELD problem [36]. ern power system planning and operation. The objective of
In [37], the authors proposed a hybrid ED SHO for ELD is to maximize the economic welfare of the power
solving ELD. system subject to certain operational constraints thereby opti-
In the same context, in [38], [39], WOA, and MFO were mally allocating each production units and reducing the net
used to solve different variants of ELD. In [40], ELD with fuel cost consumption. The different variants of ELD is elab-
valve point effect was solved by using modified GWO. orated in this section.
In [41], ELD was solved by an improved version of
PSO with inertia weights factor. In [42], [43], AEF, and A. ECONOMIC LOAD DISPATCH (ELD) WITH LOSSES
ALO were used for solving ELD for a small-scale power The mathematical formulation of ELD with losses is
system. In [44] adaptive charged system search algorithm explained in this section. For operating n generators, the over-
is applied to solve EED of power systems. In [45], [46] all fuel cost is:
the authors have utilized improved version of PSO for solv-
ing ELD. Authors solved the ELD problem for 24-hour Min(F) = F1 (P1 ) + · · · Fn (Pn ) (1)
load pattern by GWO [47]. An enhanced version of BA where F is the overall fuel cost, F1 is the fuel cost of 1st
for solving ELD was proposed in [48], [49]. In [50], mod- generator and Fn is the fuel cost of nth generator.
ified ABC algorithm was used for solving non-smooth The fuel cost function is further approximated in quadratic
dynamic ELD. A modified version of cultural algorithm hav- form as:
ing a local search component for solving ELD and CEED was Xn Xn
proposed in [51]. A chaotic firefly algorithm having muta- Min(F) = Fi (Pi ) = ak P2k + bk Pk + ck (2)
tion operator is used for solving ELD in case of large scale k=1 k=1
power system having valve point effect and multiple fuel where a, b, c are the weight constants of the fuel cost The
options [52]. minimization of fuel cost is performed subject to constraints
In [53], the dynamic ELD problem was solved by mod- given by equation (3) and (5)
ified TLBO. In [54], smooth as well as non-smooth ELD
n
was solved by WCA. The ELD problem in presence of X
Pk − PD − PL = 0 (3)
wind power was attacked by hybrid BA [55]. In [56]–[58],
k=1
multi-area ELD was solved by SSA, FSA, and MFO
algorithm. In [59], the authors used an improved ver- where PD represents net demand of the network and
sion of TLBO for solving ELD problem considering dis- PL represents the transmission losses of the network
tributed generation. A modified version of GWO was used n X
X n
for solving non-convex ELD for current power system PL = Pi Bij Pj (4)
scenario in [60]. In [61], dynamic ELD was solved by i=1 j=1
hybrid GA PSO. LFA and SSA was used for solving where Bij is the loss coefficient, Pi is the power generated at
large scale dynamic ELD in [62], [63]. In [64], GWO was ith generator, and Pj is the power generated at jth generator
used for solving ELD and CEED with valve point effect.
An improved version of DFA was used for solving ELD Pmin max
k ≤ Pk ≤ Pk (5)
considering demand response and renewable resources [65].
In [66]–[68], chaotic bat, modified social spider and FPA B. COMBINED ECONOMIC LOAD DISPATCH (CEED)
algorithm was used for solving ELD. In [69] Levy flight Stress has been laid on reduction of emission also in addition
Moth-Flame optimizer is proposed to solve the ELD. A novel to production cost. Hence, optimal allocation is done consid-
parallel hurricane algorithm was used to solve ELD as well as ering both cost and emission leading to Combined Economic

44324 VOLUME 9, 2021

S. Deb et al.: Recent Methodology-Based GBO for ELD Problem

TABLE 1. Recent research studies on ELD.

VOLUME 9, 2021 44325

 S. Deb et al.: Recent Methodology-Based GBO for ELD Problem

and Emission Dispatch (CEED). In CEED, both economic B. GRADIENT SEARCH RULE (GSR) PROCESS
and emission dispatch are taken into consideration. In GBO algorithm, a significant factor ρ is employed to
The emission dispatch problem is concerned with mini- achieve balanced exploration of significant search space
mization of the gases from power plants. Mathematically, regions while still achieving near optimum and global
the emission factor is given by: points.The ρ is employed as follows:
n
X n
X ρ1 = 2 × rand × α − α (11)
Min(E) = Ei (Pi ) = αk P2k + βk Pk + γk (6)
  
3π 3π
α = β × sin + sin β ×
 
k=1 k=1
 (12)
2 2 
The objective function of the CEED problem is:
  m 3 2
β = βmin + (βmax − βmin ) × 1 − (13)
n n
! M
where; βmin is a constant value of 0.2 and βmax is a con-
X X
objective function = Min Ei (Pi ) + he Fi (Pi ) (7)
k=1 k=1 stant value of 1.2, while m represents the current iteration
number, while M represents the total number of iterations.
where he is the price penalty factor as in equation 8: The parameter ρ1 is responsible to balance the exploration
Fi (Pimax ) and exploitation based on the sine function. This parameter
he = (8) value changes during optimization iterations, beginning at a
Ei (Pimax )
large value to facilitate wider variety, then decreasing over
The optimization is performed subject to constraints given by the iterations to speed up convergence. The parameter value
equations (3) and (5). increases through defined iterations within a range. This
increases diversity in solutions, and allows the algorithm to
a: CEED WITH VALVE POINT EFFECT: explore multiple solutions to the problem. Described above,
In modern era, steam turbines have multiple valves causing the GSR can be calculated as follows:.
valve point effect. The valves make the cost function nonlin- 21x × xn
GSR = randn × ρ1 × (14)
ear as in equation 9: (xworst − xbest + ε)
n n The GBO algorithm uses a random behavior to create
a randomized exploration mechanism that includes finding
X X
Min(F) = Fi = ak P2k + bk Pk + ck
k=1 k=1
local optima. In Equation (14), it is specified the random
+ |ek sin(fk × (Pkmin − Pk ))| (9) offset that deals the difference between the best solution
m The meaning 1x
(xbest ) and a randomly selected solution xr1
where ek and fk are the coefficients reflecting valve point of the variable is altered by iterations due to the following
effect of kth generator. The optimization is concerned with equation. (17). Additionally another random number (randn)
minimization of both cost and emission subject to constraints is included to allow for exploration as follows:
as in equations (3) and (5). 1x = rand(1 : N ) × | step | (15)
m +δ


xbest − xr1
IV. GRADIENT-BASED OPTIMIZER (GBO) step = (16)
Researchers have invented the GBO metaheuristic algo- 2 
m + xm + xm + xm
 xr1

δ = 2 × rand ×  r2 r3 r4 m

rithm which mimics the population-based and gradient-based  − xn  (17)
methods [18]. To explore the search space for a set of search 4
metrics, Newton’s approach is utilized. The main steps of where rand(1 : N ) is a random vector of N elements in the
GBO are as follows: range of ∈ [0, 1].
The four randomly selected integers are r1, r2, r3, and r4
A. THE INITIALIZATION PROCESS such that (r1 6= r2 6= r3 6= r4 6= n). step represents a phase
scale, which is quantified by xbest and xr1m.
In GBO, the control parameters (α) and probability rate
are used to balance and switch from the exploration to To achieve convergence, directional movement is
exploitation. Population and iterations numbers are related employed in order to converge across the solution field xn .
to the problem’s complexity. In GBO, the vector of In order to provide a convenient local search tendency with a
N vectors in D-dimensional space can be described. The major effect on GBO convergence, the term DM uses the best
initial vectors for the GBO are usually randomly generated vector from a set of candidate vectors and transfers the current
in the D-dimensional search space. vector (xn ) in the direction of the best vector (xbest − xn ) and
is computed as follows:
Xn = Xmin + rand(0, 1) × (Xmax − Xmin ) (10) DM = rand × ρ2 × (xbest − xn ) (18)
where Xmin , and Xmax are the bounds of decision variables X, where, rand is a uniform distributed number within range ∈
and rand(0, 1) is a random number in [0, 1]. [0, 1], a function of two parameters, and ρ2 is a random

44326 VOLUME 9, 2021

S. Deb et al.: Recent Methodology-Based GBO for ELD Problem

TABLE 4. Statistical results for fitness function of case 1.

FIGURE 1. Friedman rank test for case 1.

TABLE 2. Parameter settings of GBO and all other competed algorithms.

TABLE 5. Best costs of case 1 at various demand setting in $ per hour.

where ypm n , yqn are equal to yn + 1x, and yn − 1x, yn vector

m
TABLE 3. Test cases.
is equal to the average of two vectors: current solution xn ,
and zn+1 .
21x × xn
zn+1 = xn −randn × (22)
(xworst − xbest + ε)
although xn is current solution vector, randn is a random
solution vector of dimension n, xworst and xbest represent best
and worst solutions, and 1x is given by equation.(15).
parameter employed to adjust phase size of each vector agent,
Taking the previous formula and substituting current solu-
besides the ρ2 parameter considers of significant parameters
tion vector xnm with new solution vector xbest in it, we get
of the GBO exploration process. This is how the ρ2 parameter
current solution vector X 2m
n.
is computed:
21x × xnm
ρ2 = 2 × rand × α − α. (19) n = xbest − randn × ρ1 ×
X 2m
n − yqn + ε


ypm m
Finally, based on the words GSR and DM, Eq. 20 and 21 + randn × ρ2 × xr1
m m
− xr2


(23)
are modified based on the current vector location (xnm ).
The GBO algorithm aims to improve discovery and
X 1m m
n = xn −GSR + DM (20)
exploitation by making each more profitable than the next
where, X 1m n is the modified vector resulting from
using using Eq. (21). To help the global search on the dis-
modifying X 1mn . According to (13), (18), the transformation
covery process and to boostEq. (23) is used to boost the
of X 1m
n can be defined as:.
exploitation process of the local search. Finally, a new version
of the solution is found as follows
21x × xnm
n = xn − randn × ρ1 ×
X 1m m
n − yqn + ε xnm+1 = ra × rb × X 1m n + (1 − rb ) × X 2n
m



ypm m

+ randn × ρ2 × xbest − xnm + (1 − ra ) × X 3m

(21) n (24)

VOLUME 9, 2021 44327

 S. Deb et al.: Recent Methodology-Based GBO for ELD Problem

TABLE 6. Allocation vector at best fitness function using all algorithms for case 1 at 700 MW load demand.

TABLE 7. Allocation vector at best fitness function using all algorithms for case 1 at 1000 MW load demand.

TABLE 8. Allocation vector at best fitness function using all algorithms for case 1 at 1200 MW load demand.

FIGURE 2. Robustness curves for all algorithms for case 1 at 700 MW load demand.

where ra , and rb are random numbers determined in range the procedure is performed based on a scheme that is as
[0, 1], and X 3m
n is defined as: follows:.
If rand < pr
X 3m m+1
n = Xn − ρ1 × (X 2m m
n − X 1n ) (25)  m+1
+ f1 u1 xbest − u2 xkm



 Xn
+ f2 ρ1 u3 X 2m m



n
− X 1n


C. THE LOCAL ESCAPING OPERATOR (LEO) PROCESS 

m m
/2, if rand < 0.5

The LEO is implemented to add extra power to an optimiza- m
 + u2 xr1 − xr2
XLEO = m+1 m


tion algorithm by helping to solve tricky engineering prob- 

 Xn + f1 u1 xbest − u2 xk
+ f2 ρ1 u3 X 2m m



lems. The LEO operator helps the algorithm to quickly switch 


 n
− X 1n
out of local optima points to speed up the convergence of the 

+ u2 xr1m m
− xr2 /2, otherwise
algorithm. To build a new solution with a superior efficiency,
m ), by many solutions (X (26)
the LEO operator targets (XLEO best
n , X 1n are randomly selected
best solution, the solutions X 1m m End
from population, Xr1 m , X 1m randomly generated solutions), where pr is a probability value, pr = 0.5, the values f1 ,
r2
so that the current solution can effectively be modified, and f2 are uniform distribution random numbers ∈ [−1, 1],

44328 VOLUME 9, 2021

S. Deb et al.: Recent Methodology-Based GBO for ELD Problem

FIGURE 3. Convergence curves for all algorithms for case 1 at 700 MW load demand.

FIGURE 4. Robustness curves for all algorithms for case 1 at 1000 MW load demand.

FIGURE 5. Convergence curves for all algorithms for case 1 at 1000 MW load demand.

(
and u1 , u2 , u3 are random values generated as following: rand if µ1 < 0.5
u2 = (28)
1 otherwise
(
if µ1 < 0.5
(
2 × rand if µ1 < 0.5 rand
u1 = (27) u3 = (29)
1 otherwise 1 otherwise

VOLUME 9, 2021 44329

 S. Deb et al.: Recent Methodology-Based GBO for ELD Problem

FIGURE 6. Robustness curves for all algorithms for case 1 at 1200 MW load demand.

FIGURE 7. Convergence curves for all algorithms for case 1 at 1200 MW load demand.

where rand is a random number between 0 and 1, while mu1

is an arbitrary value from the interval [0, 1].
The previous equations for u1 , u2 , u3 , can be explained as
follow:
u1 = L1 × 2 × rand + (1 − L1 ) (30)
u2 = L1 × rand + (1 − L1 ) (31)
u3 = L1 × rand + (1 − L1 ) (32)
where L1 is a binary parameter take value 0, 1, such as if
parameter µ1 < 0.5, then value of L1 = 1, otherwise L1 = 0.
where the solution xkm is generated as follow:
(
m xrand if µ2 < 0.5
xk = (33)
xpm otherwise
xrand is a random generated solution according to following FIGURE 8. Friedman rank test for case 2.
formula:
xrand = Xmin + rand(0, 1) × (Xmax − Xmin ) (34)
V. EXPERIMENTAL RESULTS AND NUMERICAL ANALYSIS
and xpm is a random selected solution from population, The performance of GBO on different variants of GBO is
µ2 is a random number ∈ [0, 1]. For more details about compared with other metaheuristics such as; Slime mould
GBO see [18]. algorithm (SMA) [19], Elephant herding optimization (EHO)

44330 VOLUME 9, 2021

S. Deb et al.: Recent Methodology-Based GBO for ELD Problem

FIGURE 9. Robustness curves for all algorithms for case 2 at 700 MW load demand.

FIGURE 10. Convergence curves for all algorithms for case 2 at 700 MW load demand.

FIGURE 11. Robustness curves for all algorithms for case 2 at 1000 MW load demand.

[20], Monarch butterfly optimization (MBO) [21], Moth Chimp Optimization Algorithm (ChOA) [26]. The compar-
search algorithm (MSA) [22], Earthworm optimization algo- ison results are reported in this section.
rithm (EWA) [23], Artificial Bee Colony (ABC) Algo- The obtained results of the proposed GBO as well as
rithm [24], Tunicate Swarm Algorithm (TSA) [25] and the considered competitor algorithms to solve the Economic

VOLUME 9, 2021 44331

 S. Deb et al.: Recent Methodology-Based GBO for ELD Problem

FIGURE 12. Convergence curves for all algorithms for case 2 at 1000 MW load demand.

FIGURE 13. Robustness curves for all algorithms for case 2 at 1200 MW load demand.

FIGURE 14. Convergence curves for all algorithms for case 2 at 1200 MW load demand.

Load Dispatch (ELD) problems.All the experiments are A. PARAMETER SETTINGS

implemented by coded through MATLAB R2016a and exe- For fair comparison, the algorithms are tested with the fol-
cuted on Intel(R) Core i7 CPU- 2.80 GHz with 8 GB RAM lowing settings: population size N = 30, maximum iterations
and operating system (Windows 10). tmax = 1000, and number of independent runs is 30 for all the

44332 VOLUME 9, 2021

S. Deb et al.: Recent Methodology-Based GBO for ELD Problem

FIGURE 15. Convergence curves for all algorithms for case 3 at 2000 MW load demand.

FIGURE 16. Robustness curves for all algorithms for case 3 at 2000 MW load demand.

TABLE 9. Statistical results for fitness function of case 2. TABLE 10. Costs at the best fitness function for case 2 at all various
demand.

B. COMPARISON OF GBO WITH TSA AND CHOA

The comparison results of GBO with TSA and ChOA is
reported in this section. The analysis is done for the test
cases as shown in Table3. The ELD problem is solved for
all the test cases reported in Table 3. Table 4 reports the
best, worst, average fitness function obtained by the three
aforesaid algorithms. Table 5 reports the best fuel cost for
case 1 at various demand setting. The superior performance
of GBO is clear from the obtained result. Figure 1 shows
the Friedman rank of the three algorithms. It is observed that
experiments. Table 2 reported the parameters setting of each GBO has obtained the best rank followed by TSA for case 1.
algorithm. Table 6 to Table 8 reports the allocation vector obtained by

VOLUME 9, 2021 44333

 S. Deb et al.: Recent Methodology-Based GBO for ELD Problem

TABLE 11. Allocation vector at best fitness function using all algorithms for case 2 at 700 MW load demand.

TABLE 12. Allocation vector at best fitness function using all algorithms for case 2 at 1000 MW load demand.

TABLE 13. Allocation vector at best fitness function using all algorithms for case 2 at 1200 MW load demand.

TABLE 14. Statistical results for fitness function of case 3. three algorithms. It is observed that GBO maintains a good
balance between cost and emission. Figure 8 shows the
Friedman rank obtained for case 2. It is observed that GBO
has obtained the best rank followed by ChOA. Table 11,
Table 13, and Table 13 reports the allocation vector for
demand 700 MW, 1000 MW, and 1200 MW respectively.
Further, the convergence and robustness of GBO is compared
with TSA and ChOA. Figure 9, Figure 11, Figure 13 shows
the robustness curve of case 2 with load demand 700 MW,
TABLE 15. Costs at the best fitness function for case 3 at 2000 MW load
demand. 1000 MW, and 1200 MW respectively. Figure 10, Figure 12,
Figure 14 shows the convergence curve of case 2 with load
demand 700 MW, 1000 MW, and 1200 MW respectively. It is
observed that in case of GBO a faster convergence towards the
optima is favoured.
The best, worst, average value of fitness function for case 3
is as shown in Table 14. The superior performance of GBO
is prominent from the results reported in Table 14. Table 15
reports the fuel cost and emission for case 3 obtained by the
the aforesaid algorithm for demand 700 MW, 1000 MW,
three algorithms with and without valve point effect. Table 16
and 1200 MW respectively. Figure 2, Figure 4, Figure 6
reports the allocation vector for case 3. Further, the conver-
shows the robustness curve of case 1 for demand 700 MW,
gence and robustness of GBO is compared with TSA and
1000 MW, and 1200 MW respectively. It is observed that
ChOA. Figure 15 and Figure 16 shows the convergence and
GBO produces uniform solutions for different runs as com-
robustness curve for case 3 respectively.
pared to other algorithms. Figure 3, Figure 5, Figure 7
shows the convergence curve for case 1 in case of different
demand. C. COMPARISON OF GBO WITH CTO AND ITS VARIANTS
The best, worst, average value of fitness function for The performance of GBO is compared with CTO and its
case 2 is reported in Table 9. It is observed that GBO variants in solving case 3 of Table 3. The results of CTO and
performs well as compared to other algorithms. Table 10 its variants are taken from ref [27] and the ELD problem is
reports the fuel cost and emission for case 2 obtained by the solved by GBO with the same general parameter setting as

44334 VOLUME 9, 2021

S. Deb et al.: Recent Methodology-Based GBO for ELD Problem

TABLE 16. Allocation vector at best fitness function using all algorithms for case 3 at 2000 MW load demand.

TABLE 17. Comparison of GBO with CTO and its variants. VI. CONCLUSION AND FUTURE WORK
ELD is one of the complex problems of power system.
This work tests the performance of gradient based opti-
mization (GBO) in solving different variants of ELD such
as ELD with losses, CEED, and CEED considering valve
point effect. GBO is a metaheuristic inspired by the gradient
TABLE 18. Comparison of GBO with DE, PDE and MODE. based Newton method involving Gradient search rule (GSR)
and local escaping operator (GEO). GBO has good balance
between exploration and exploitation. Also, the possibility
of getting stuck in local optima and premature conver-
gence is rare in GBO. The performance of GBO is com-
pared with eight other metaheuristic algorithms such as
Slime mould algorithm (SMA), Elephant herding optimiza-
TABLE 19. Comparison of GBO with NSGA-II and SPEA-2. tion (EHO), Monarch butterfly optimization (MBO), Moth
search algorithm (MSA), Earthworm optimization algorithm
(EWA), Artificial Bee Colony (ABC) Algorithm, Tunicate
Swarm Algorithm (TSA) and Chimp Optimization Algo-
rithm (ChOA). In addition, GBO is evaluated against other
existing studies in the literature such as Differential evolution
(DE), Class Topper Optimization (CTO), Non Dominated
in ref [27]. The comparison results are as shown in Table 17. Sorting GA (NSGA-II), and Strength pareto evolutionary
It is observed GBO is the second-best performing algorithm. algorithm 2 (SPEA-2) for different demands. It is observed
that GBO performs relatively well as compared to the afore-
D. COMPARISON OF GBO WITH DE AND ITS VARIANTS said algorithms. Further, it is seen that GBO has good balance
The performance of GBO is compared with DE and its vari- between exploration and exploitation and the possibility of
ants in solving case 3 of Table 3. The results of DE and getting stuck in local optima and premature convergence is
its variants are taken from ref [76] and the ELD problem is rare in GBO. Our future work will focus on:
solved by GBO with the same general parameter setting as in • Performance of GBO on dynamic ELD considering
ref [76]. The comparison results are as shown in Table 18. It is renewable resources
observed that GBO performs better than PDE and MODE in • Performance of GBO on other power system problems
terms of cost. And, GBO performs better than DE and MODE such as unit commitment, optimal load flow
in terms of emission. • Hybridization of GBO with other metaheuristics for
solving power system optimization problems
E. COMPARISON OF GBO WITH NSGA-II AND SPEA-2
The performance of GBO is compared with NSGA II and In the future studies, the GBO can be a good candidate to
SPEA 2 in solving case 3 of Table 3. The results of NSGA-II solve the problems in renewable energy for instance solar cell
and SPEA-2 are taken from ref [76] and the ELD problem is systems. Due to the great performance of the GBO, future
solved by GBO with the same general parameter setting as work may extend to solve various single and multi-objective
in ref [76]. The comparison results are as shown in Table 19. optimization problems in different fields.
It is observed that GBO performs better than SPEA 2 and
NSGA II in terms of cost. Also, GBO performs better than CONFLICT OF INTEREST
NSGA II in terms of emission. The authors declare that there is no conflict of interest.

VOLUME 9, 2021 44335

 S. Deb et al.: Recent Methodology-Based GBO for ELD Problem

CREDIT AUTHOR STATEMENT [18] I. Ahmadianfar, O. Bozorg-Haddad, and X. Chu, ‘‘Gradient-based opti-
All authors contributed equally to this paper, where; Sanchari mizer: A new Metaheuristic optimization algorithm,’’ Inf. Sci., vol. 540,
pp. 131–159, Nov. 2020.
Deb: Formal analysis, Resources, Writing - review & edit- [19] S. Li, H. Chen, M. Wang, A. A. Heidari, and S. Mirjalili, ‘‘Slime mould
ing.Diaa Salama AbdElminaam: Resources,Methodology, algorithm: A new method for stochastic optimization,’’ Future Gener.
Conceptualization, Data curation, Writing - review & editing. Comput. Syst., vol. 111, pp. 300–323, Oct. 2020.
[20] G.-G. Wang, S. Deb, and L. D. S. Coelho, ‘‘Elephant herding optimiza-
Mokhtar Said: Software, Formal analysis, Resources, Writing tion,’’ in Proc. 3rd Int. Symp. Comput. Bus. Intell. (ISCBI), Dec. 2015,
- original draft.Essam H. Houssein: Supervision, Methodol- pp. 1–5.
ogy, Conceptualization, Formal analysis, Writing - review & [21] G.-G. Wang, S. Deb, and Z. Cui, ‘‘Monarch butterfly optimization,’’ Neural
Comput. Appl., vol. 31, no. 7, pp. 1995–2014, 2019.
editing. All authors read and approved the final paper.
[22] G.-G. Wang, ‘‘Moth search algorithm: A bio-inspired Metaheuristic algo-
rithm for global optimization problems,’’ Memetic Comput., vol. 10, no. 2,
REFERENCES pp. 151–164, Jun. 2018.
[23] G. G. Wang, S. Deb, and L. D. S. Coelho, ‘‘Earthworm optimisation
[1] A. A. K. Ismaeel, E. H. Houssein, D. Oliva, and M. Said, ‘‘Gradient-based algorithm: A bio-inspired Metaheuristic algorithm for global optimisation
optimizer for parameter extraction in photovoltaic models,’’ IEEE Access, problems,’’ Int. J. Bio-Inspired Comput., vol. 12, no. 1, p. 1, 2018.
vol. 9, pp. 13403–13416, 2021. [24] D. Karaboga and B. Basturk, ‘‘An artificial bee colony (ABC) algorithm
[2] M. H. Hassan, E. H. Houssein, M. A. Mahdy, and S. Kamel, ‘‘An improved for numeric function optimization,’’ in Proc. IEEE Swarm Intell. Symp.,
manta ray foraging optimizer for cost-effective emission dispatch prob- 2006, pp. 181–184.
lems,’’ Eng. Appl. Artif. Intell., vol. 100, Apr. 2021, Art. no. 104155. [25] S. Kaur, L. K. Awasthi, A. L. Sangal, and G. Dhiman, ‘‘Tunicate swarm
[3] A. Sheta, H. Faris, M. Braik, and S. Mirjalili, ‘‘Nature-inspired meta- algorithm: A new bio-inspired based metaheuristic paradigm for global
heuristics search algorithms for solving the economic load dispatch prob- optimization,’’ Eng. Appl. Artif. Intell., vol. 90, Apr. 2020, Art. no. 103541.
lem of power system: A comparison study,’’ in Applied Nature-Inspired [26] M. Khishe and M. R. Mosavi, ‘‘Chimp optimization algorithm,’’ Expert
Computing: Algorithms and Case Studies. Singapore: Springer, 2020, Syst. Appl., vol. 149, Jul. 2020, Art. no. 113338.
pp. 199–230.
[27] A. Srivastava and D. K. Das, ‘‘A new aggrandized class topper opti-
[4] Y. A. Gherbi, H. Bouzeboudja, and F. Z. Gherbi, ‘‘The combined economic mization algorithm to solve economic load dispatch problem in a power
environmental dispatch using new hybrid Metaheuristic,’’ Energy, vol. 115, system,’’ IEEE Trans. Cybern., early access, Nov. 6, 2020, doi: 10.
pp. 468–477, Nov. 2016. 1109/TCYB.2020.3024607.
[5] A. LaTorre, S. Muelas, and J.-M. Peña, ‘‘A comprehensive comparison of [28] D. Santra, A. Mukherjee, K. Sarker, and S. Mondal, ‘‘Dynamic economic
large scale global optimizers,’’ Inf. Sci., vol. 316, pp. 517–549, Sep. 2015. dispatch using hybrid metaheuristics,’’ J. Electr. Syst. Inf. Technol., vol. 7,
[6] M. N. A. Wahab, S. Nefti-Meziani, and A. Atyabi, ‘‘A comprehensive no. 1, p. 3, Dec. 2020.
review of swarm optimization algorithms,’’ PLoS ONE, vol. 10, no. 5, [29] F. Tariq, S. Alelyani, G. Abbas, A. Qahmash, and M. R. Hussain, ‘‘Solv-
May 2015, Art. no. e0122827. ing renewables-integrated economic load dispatch problem by variant of
[7] E. H. Houssein, Y. Mina, and E. Aboul, ‘‘Nature-inspired algorithms: metaheuristic bat-inspired algorithm,’’ Energies, vol. 13, no. 23, p. 6225,
A comprehensive review,’’ in Hybrid Computational Intelligence: Nov. 2020.
Research and Applications. Boca Raton, FL, USA: CRC Press, 2019, p. 1. [30] F. X. Rugema, G. Yan, S. Mugemanyi, Q. Jia, S. Zhang, and C. Bananeza,
[8] G.-G. Wang, A. H. Gandomi, A. H. Alavi, and D. Gong, ‘‘A comprehensive ‘‘A cauchy-Gaussian quantum-behaved bat algorithm applied to solve the
review of krill herd algorithm: Variants, hybrids and applications,’’ Artif. economic load dispatch problem,’’ IEEE Access, vol. 9, pp. 3207–3228,
Intell. Rev., vol. 51, no. 1, pp. 119–148, Jan. 2019. 2021.
[9] E. H. Houssein, M. R. Saad, F. A. Hashim, H. Shaban, and M. Hassabal- [31] R. Keswani, H. K. Verma, and S. K. Sharma, ‘‘Dynamic economic load
lah, ‘‘Lévy flight distribution: A new Metaheuristic algorithm for solving dispatch considering renewable energy sources using multiswarm statisti-
engineering optimization problems,’’ Eng. Appl. Artif. Intell., vol. 94, cal particle swarm optimization,’’ in Proc. IEEE Int. Conf. Comput., Power
Sep. 2020, Art. no. 103731. Commun. Technol. (GUCON), Oct. 2020, pp. 405–410.
[10] F. A. Hashim, E. H. Houssein, M. S. Mabrouk, W. Al-Atabany, and [32] P. Vasant, F. P. Mahdi, J. A. Marmolejo-Saucedo, I. Litvinchev,
S. Mirjalili, ‘‘Henry gas solubility optimization: A novel physics- R. R. Aguilar, and J. Watada, ‘‘Quantum-behaved bat algorithm for solving
based algorithm,’’ Future Gener. Comput. Syst., vol. 101, pp. 646–667, the economic load dispatch problem considering a valve-point effect,’’ Int.
Dec. 2019. J. Appl. Metaheuristic Comput., vol. 11, no. 3, pp. 41–57, Jul. 2020.
[11] F. A. Hashim, K. Hussain, E. H. Houssein, M. S. Mabrouk, and [33] A. Srivastava and S. Singh, ‘‘Implementation of ant colony optimization in
W. Al-Atabany, ‘‘Archimedes optimization algorithm: A new metaheuris- economic load dispatch problem,’’ in Proc. 7th Int. Conf. Signal Process.
tic algorithm for solving optimization problems,’’ Appl. Intell., vol. 51, Integr. Netw. (SPIN), Feb. 2020, pp. 1018–1024.
pp. 1531–1551, Sep. 2020. [34] Y. A. Gherbi, F. Lakdja, H. Bouzeboudja, and F. Z. Gherbi, ‘‘Hybridization
[12] E. H. Houssein, A. G. Gad, Y. M. Wazery, and P. N. Suganthan, ‘‘Task of two metaheuristics for solving the combined economic and emission
scheduling in cloud computing based on meta-heuristics: Review, taxon- dispatch problem,’’ Neural Comput. Appl., vol. 31, no. 12, pp. 8547–8559,
omy, open challenges, and future trends,’’ Swarm Evol. Comput., vol. 62, Dec. 2019.
Apr. 2021, Art. no. 100841. [35] E. S. Ali and S. M. Abd Elazim, ‘‘Mine blast algorithm for environmental
[13] E. H. Houssein, M. A. Mahdy, M. G. Eldin, D. Shebl, W. M. Mohamed, economic load dispatch with valve loading effect,’’ Neural Comput. Appl.,
and M. Abdel-Aty, ‘‘Optimizing quantum cloning circuit parameters based vol. 30, no. 1, pp. 261–270, Jul. 2018.
on adaptive guided differential evolution algorithm,’’ J. Adv. Res., early [36] W. T. Elsayed, Y. G. Hegazy, M. S. El-bages, and F. M. Bendary,
access, Oct. 17, 2020, doi: 10.1016/j.jare.2020.10.001. ‘‘Improved random drift particle swarm optimization with self-
[14] J. N. Kuk, R. A. Gonçalves, L. M. Pavelski, S. M. G. S. Venske, adaptive mechanism for solving the power economic dispatch
C. P. D. Almeida, and A. T. R. Pozo, ‘‘An empirical analysis of con- problem,’’ IEEE Trans. Ind. Informat., vol. 13, no. 3, pp. 1017–1026,
straint handling on evolutionary multi-objective algorithms for the envi- Jun. 2017.
ronmental/economic load dispatch problem,’’ Expert Syst. Appl., vol. 165, [37] G. Dhiman, S. Guo, and S. Kaur, ‘‘ED-SHO: A framework for solving
Mar. 2021, Art. no. 113774. nonlinear economic load power dispatch problem using spotted hyena opti-
[15] D. Singh and J. S. Dhillon, ‘‘Ameliorated grey wolf optimization for eco- mizer,’’ Mod. Phys. Lett. A, vol. 33, no. 40, Dec. 2018, Art. no. 1850239.
nomic load dispatch problem,’’ Energy, vol. 169, pp. 398–419, Feb. 2019. [38] A. Kumar, V. Bhalla, P. Kumar, T. Bhardwaj, and N. Jangir, ‘‘Whale opti-
[16] F. Mohammadi and H. Abdi, ‘‘A modified crow search algorithm (MCSA) mization algorithm for constrained economic load dispatch problems—
for solving economic load dispatch problem,’’ Appl. Soft Comput., vol. 71, A cost optimization,’’ in Ambient Communications and Computer Systems.
pp. 51–65, Oct. 2018. Singapore: Springer, 2018, pp. 353–366.
[17] S. H. A. Kaboli and A. K. Alqallaf, ‘‘Solving non-convex economic load [39] A. Bhadoria, V. K. Kamboj, M. Sharma, and S. Bath, ‘‘A solution to non-
dispatch problem via artificial cooperative search algorithm,’’ Expert Syst. convex/convex and dynamic economic load dispatch problem using moth
Appl., vol. 128, pp. 14–27, Aug. 2019. flame optimizer,’’ INAE Lett., vol. 3, no. 2, pp. 65–86, 2018.

44336 VOLUME 9, 2021

S. Deb et al.: Recent Methodology-Based GBO for ELD Problem

[40] M. A. Al-Betar, M. A. Awadallah, and M. M. Krishan, ‘‘A non-convex [60] J. Paramguru and S. K. Barik, ‘‘Modified grey wolf optimization applied
economic load dispatch problem with valve loading effect using a to non-convex economic load dispatch in current power system sce-
hybrid grey wolf optimizer,’’ Neural Comput. Appl., vol. 32, no. 16, nario,’’ in Proc. Int. Conf. Recent Innov. Electr., Electron. Commun. Eng.
pp. 12127–12154, Aug. 2020. (ICRIEECE), Jul. 2018, pp. 2704–2709.
[41] T. Yalcinoz and K. Rudion, ‘‘Economic load dispatch using an improved [61] A. E. Fergougui, A. A. Ladjici, A. Benseddik, and Y. Amrane, ‘‘Dynamic
particle swarm optimization based on functional constriction factor and economic dispatch using genetic and particle swarm optimization algo-
functional inertia weight,’’ in Proc. IEEE Int. Conf. Environ. Electr. rithm,’’ in Proc. 5th Int. Conf. Control, Decis. Inf. Technol. (CoDIT),
Eng. IEEE Ind. Commercial Power Syst. Eur. (EEEIC / I CPS Europe), Apr. 2018, pp. 1001–1005.
Jun. 2019, pp. 1–5. [62] M. Kheshti, X. Kang, Z. Bie, Z. Jiao, and X. Wang, ‘‘An effective lightning
[42] A. Yadav and N. Kumar, ‘‘Application of artificial electric field algorithm flash algorithm solution to large scale non-convex economic dispatch
for economic load dispatch problem,’’ in Proc. Int. Conf. Soft Comput. with valve-point and multiple fuel options on generation units,’’ Energy,
Pattern Recognit. Cham, Switzerland: Springer, 2019, pp. 71–79. vol. 129, pp. 1–15, Jun. 2017.
[43] V. K. Kamboj, A. Bhadoria, and S. K. Bath, ‘‘Solution of non-convex [63] Y. Sonmez, H. T. Kahraman, M. K. Dosoglu, U. Guvenc, and S. Duman,
economic load dispatch problem for small-scale power systems using ant ‘‘Symbiotic organisms search algorithm for dynamic economic dispatch
lion optimizer,’’ Neural Comput. Appl., vol. 28, no. 8, pp. 2181–2192, with valve-point effects,’’ J. Experim. Theor. Artif. Intell., vol. 29, no. 3,
Aug. 2017. pp. 495–515, May 2017.
[44] P. Zakian and A. Kaveh, ‘‘Economic dispatch of power systems using an [64] V. K. Kamboj, S. K. Bath, and J. S. Dhillon, ‘‘Solution of non-convex eco-
adaptive charged system search algorithm,’’ Appl. Soft Comput., vol. 73, nomic load dispatch problem using grey wolf optimizer,’’ Neural Comput.
pp. 607–622, Dec. 2018. Appl., vol. 27, no. 5, pp. 1301–1316, Jul. 2016.
[45] A. Lin and W. Sun, ‘‘Multi-leader comprehensive learning particle swarm [65] V. Suresh, S. Sreejith, S. K. Sudabattula, and V. K. Kamboj, ‘‘Demand
optimization with adaptive mutation for economic load dispatch prob- response-integrated economic dispatch incorporating renewable energy
lems,’’ Energies, vol. 12, no. 1, p. 116, Dec. 2018. sources using ameliorated dragonfly algorithm,’’ Electr. Eng., vol. 101,
[46] Q. Qin, S. Cheng, X. Chu, X. Lei, and Y. Shi, ‘‘Solving non-convex/non- no. 2, pp. 421–442, Jun. 2019.
smooth economic load dispatch problems via an enhanced particle swarm [66] B. R. Adarsh, T. Raghunathan, T. Jayabarathi, and X.-S. Yang, ‘‘Eco-
optimization,’’ Appl. Soft Comput., vol. 59, pp. 229–242, Oct. 2017. nomic dispatch using chaotic bat algorithm,’’ Energy, vol. 96, pp. 666–675,
Feb. 2016.
[47] K. Kadali, R. Loganathan, M. Veerasamy, and V. Jawalker, ‘‘Cost-effective
[67] W. T. Elsayed, Y. G. Hegazy, F. M. Bendary, and M. S. El-bages, ‘‘Modified
dispatch using grey wolf optimization algorithm: Solution with diverse
social spider algorithm for solving the economic dispatch problem,’’ Eng.
load pattern,’’ in Proc. Int. Conf. Syst., Comput., Autom. Netw. (ICSCAN),
Sci. Technol., Int. J., vol. 19, no. 4, pp. 1672–1681, Dec. 2016.
Jul. 2020, pp. 1–5.
[68] A. Y. Abdelaziz, E. S. Ali, and S. M. A. Elazim, ‘‘Implementation of
[48] G. Pradhan and P. D. Dewangan, ‘‘Solving optimal load dispatch problem
flower pollination algorithm for solving economic load dispatch and com-
using enhanced BAT optimization algorithm,’’ in Proc. Innov. Power Adv.
bined economic emission dispatch problems in power systems,’’ Energy,
Comput. Technol. (I-PACT), Apr. 2017, pp. 1–6.
vol. 101, pp. 506–518, Apr. 2016.
[49] F. P. Mahdi, P. Vasant, M. Abdullah-Al-Wadud, V. Kallimani, and [69] I. N. Trivedi, A. Kumar, A. H. Ranpariya, and P. Jangir, ‘‘Economic load
J. Watada, ‘‘Quantum-behaved bat algorithm for many-objective combined dispatch problem with ramp rate limits and prohibited operating zones
economic emission dispatch problem using cubic criterion function,’’ Neu- solve using levy flight moth-flame optimizer,’’ in Proc. Int. Conf. Energy
ral Comput. Appl., vol. 31, no. 10, pp. 5857–5869, Oct. 2019. Efficient Technol. Sustainability (ICEETS), Apr. 2016, pp. 442–447.
[50] I. Marouani, A. Boudjemline, T. Guesmi, and H. H. Abdallah, ‘‘A modified [70] R. M. Rizk-Allah, R. A. El-Sehiemy, and G.-G. Wang, ‘‘A novel parallel
artificial bee colony for the non-smooth dynamic economic/environmental hurricane optimization algorithm for secure emission/economic load dis-
dispatch,’’ Eng., Technol. Appl. Sci. Res., vol. 8, no. 5, pp. 3321–3328, patch solution,’’ Appl. Soft Comput., vol. 63, pp. 206–222, Feb. 2018.
Oct. 2018. [71] D. Zou, S. Li, G.-G. Wang, Z. Li, and H. Ouyang, ‘‘An improved dif-
[51] C. A. O. D. Freitas, R. C. L. D. Oliveira, D. J. A. D. Silva, J. C. Leite, and ferential evolution algorithm for the economic load dispatch problems
J. D. A. Brito, ‘‘Solution to economic–emission load dispatch by cultural with or without valve-point effects,’’ Appl. Energy, vol. 181, pp. 375–390,
algorithm combined with local search: Case study,’’ IEEE Access, vol. 6, Nov. 2016.
pp. 64023–64040, 2018. [72] D. H. Wolpert and W. G. Macready, ‘‘No free lunch theorems for optimiza-
[52] Y. Yang, B. Wei, H. Liu, Y. Zhang, J. Zhao, and E. Manla, ‘‘Chaos firefly tion,’’ IEEE Trans. Evol. Comput., vol. 1, no. 1, pp. 67–82, Apr. 1997.
algorithm with self-adaptation mutation mechanism for solving large-scale [73] J.-T. Yu, C.-H. Kim, A. Wadood, T. Khurshaid, and S.-B. Rhee, ‘‘Jaya
economic dispatch with valve-point effects and multiple fuel options,’’ algorithm with self-adaptive multi-population and Lévy flights for solving
IEEE Access, vol. 6, pp. 45907–45922, 2018. economic load dispatch problems,’’ IEEE Access, vol. 7, pp. 21372–21384,
[53] B. M. Alshammari, ‘‘Teaching-Learning-Based optimization algorithm 2019.
for the combined dynamic economic environmental dispatch prob- [74] M. J. Mokarram, J. P. S. Catalao, J. Aghaei, M. Shafie-khah, and
lem,’’ Eng., Technol. Appl. Sci. Res., vol. 10, no. 6, pp. 6432–6437, T. Niknam, ‘‘Hybrid optimization algorithm to solve the nonconvex
Dec. 2020. multiarea economic dispatch problem,’’ IEEE Syst. J., vol. 13, no. 3,
[54] A. Dihem, A. Salhi, D. Naimi, and A. Bensalem, ‘‘Solving smooth pp. 3400–3409, Sep. 2019.
and non-smooth economic dispatch using water cycle algorithm,’’ in [75] P. P. Biswas, P. N. Suganthan, B. Y. Qu, and G. A. J. Amaratunga, ‘‘Mul-
Proc. 5th Int. Conf. Electr. Eng.-Boumerdes (ICEE-B), Oct. 2017, tiobjective economic-environmental power dispatch with stochastic wind-
pp. 1–6. solar-small hydro power,’’ Energy, vol. 150, pp. 1039–1057, May 2018.
[55] H. Liang, Y. Liu, Y. Shen, F. Li, and Y. Man, ‘‘A hybrid bat algorithm for [76] M. Basu, ‘‘Economic environmental dispatch using multi-objective dif-
economic dispatch with random wind power,’’ IEEE Trans. Power Syst., ferential evolution,’’ Appl. Soft Comput., vol. 11, no. 2, pp. 2845–2853,
vol. 33, no. 5, pp. 5052–5061, Sep. 2018. Mar. 2011.
[56] V. Chaudhary, H. M. Dubey, M. Pandit, and J. C. Bansal, ‘‘Multi-area eco-
nomic dispatch with stochastic wind power using salp swarm algorithm,’’
Array, vol. 8, Dec. 2020, Art. no. 100044.
[57] J. Lin and Z.-J. Wang, ‘‘Multi-area economic dispatch using an
improved stochastic fractal search algorithm,’’ Energy, vol. 166, pp. 47–58,
SANCHARI DEB received the Master of Engi-
Jan. 2019. neering degree in power systems and the Ph.D.
[58] M. A. Ali, H. M. Dubey, and M. Pandit, ‘‘Moth-flame optimization for degree from the Centre for Energy, Indian Insti-
multi area economic dispatch: A novel heuristic paradigm,’’ in Proc. tute of Technology, Guwahati, India. She is cur-
Int. Conf. Energy, Commun., Data Analytics Soft Comput. (ICECDS), rently working as an ERCIM Fellow with the VTT
Aug. 2017, pp. 1068–1073. Technical Research Centre, Finland. Her research
[59] P. M. Joshi and H. K. Verma, ‘‘An improved TLBO based economic interests include power systems, energy, elec-
dispatch of power generation through distributed energy resources consid- tric vehicles, charging infrastructure, optimization,
ering environmental constraints,’’ Sustain. Energy, Grids Netw., vol. 18, and evolutionary algorithms. She is a member of
Jun. 2019, Art. no. 100207. the IEEE PES.

VOLUME 9, 2021 44337

 S. Deb et al.: Recent Methodology-Based GBO for ELD Problem

DIAA SALAMA ABDELMINAAM received the ESSAM H. HOUSSEIN received the Ph.D. degree
Ph.D. degree in information system from the Fac- in Computer Science Wireless Networks based
ulty of Computers and Information, Menufia Uni- on Artificial Intelligence, in 2012. He is cur-
versity, Egypt, in 2015. Since 2011, he has been rently working as an Associate Professor with the
an Assistance Professor with the Information Sys- Faculty of Computers and Information, Minia Uni-
tems Department, Faculty of Computers and Infor- versity, Egypt. He is also the Founder of the Com-
mation, Benha University, Egypt. He has worked puting and Artificial Intelligence Research Group
on several research topics. He has contributed (CAIRG), Egypt. He has more than 80 scientific
more than 40 technical articles in the areas of research articles published in prestigious inter-
wireless networks, wireless network security, national journals in the topics of optimization,
information security and Internet applications, cloud computing, mobile machine learning, image processing, and the IoT and its applications. His
cloud computing, the Internet of Things, and machine learning in interna- research interests include wireless sensor networks, the IoT, bioinformatics
tional journals, international conferences, local journals, and local confer- and biomedical, cloud computing, soft computing, image processing, artifi-
ences. He majors in cryptography, network security, the IoT, big data, cloud cial intelligence, data mining, optimization, and metaheuristics techniques.
computing, and deep learning. He serves as a Reviewer of more than 30 journals, such as Elsevier, Springer,
and IEEE.

MOKHTAR SAID received the B.Sc. degree

(Hons.) in electrical engineering, the M.Sc. degree,
and the Ph.D. degree in renewable energy from
the Faculty of Engineering, Fayoum Univer-
sity, Egypt, in June 2009, September 2013, and
2018, respectively. His major research interests
include modeling and simulation of electrical sys-
tems, electrical drives control, and optimization of
renewable energy systems.

44338 VOLUME 9, 2021