energies

Article
Multi-Objective Optimal Scheduling of a Microgrid Using
Oppositional Gradient-Based Grey Wolf Optimizer
Arul Rajagopalan 1 , Karthik Nagarajan 2 , Oscar Danilo Montoya 3,4 , Seshathiri Dhanasekaran 5, * ,
Inayathullah Abdul Kareem 1 , Angalaeswari Sendraya Perumal 1 , Natrayan Lakshmaiya 6
and Prabhu Paramasivam 7

1 School of Electrical Engineering, Vellore Institute of Technology, Chennai 600127, Tamil Nadu, India
2 Department of Electrical & Electronics Engineering, Hindustan Institute of Technology & Science,
Chennai 601301, Tamil Nadu, India
3 Grupo de Compatibilidad e Interferencia Electromágnetica, Facultad de Ingeniería, Universidad Distrital
Francisco José de Caldas, Bogotá 110231, Colombia
4 Laboratorio Inteligente de Energía, Universidad Tecnológica de Bolívar, Cartagena 131001, Colombia
5 Department of Computer Science, UiT The Arctic University of Norway, 9037 Tromsø, Norway
6 Department of Mechanical Engineering, Saveetha School of Engineering, SIMATS,
Chennai 602107, Tamilnadu, India
7 Department of Mechanical Engineering, College of Engineering and Technology, Mattu University,
Mettu 318, Ethiopia
* Correspondence: seshathiri.dhanasekaran@uit.no

Abstract: Optimal energy management has become a challenging task to accomplish in today’s
advanced energy systems. If energy is managed in the most optimal manner, tremendous societal
Citation: Rajagopalan, A.; benefits can be achieved such as improved economy and less environmental pollution. It is possible
Nagarajan, K.; Montoya, O.D.; to operate the microgrids under grid-connected, as well as isolated modes. The authors presented a
Dhanasekaran, S.; Kareem, I.A.; new optimization algorithm, i.e., Oppositional Gradient-based Grey Wolf Optimizer (OGGWO) in the
Perumal, A.S.; Lakshmaiya, N.; current study to elucidate the optimal operation in microgrids that is loaded with sustainable, as well
Paramasivam, P. Multi-Objective as unsustainable energy sources. With the integration of non-Renewable Energy Sources (RES) with
Optimal Scheduling of a Microgrid microgrids, environmental pollution is reduced. The current study proposes this hybrid algorithm
Using Oppositional Gradient-Based to avoid stagnation and achieve premature convergence. Having been strategized as a bi-objective
Grey Wolf Optimizer. Energies 2022,
optimization problem, the ultimate aim of this model’s optimal operation is to cut the costs incurred
15, 9024. https://doi.org10.3390/
upon operations and reduce the emission of pollutants in a 24-h scheduling period. In the current
en15239024/
study, the authors considered a Micro Turbine (MT) followed by a Wind Turbine (WT), a battery
Academic Editor: Antonio unit and a Fuel Cell (FC) as storage devices. The microgrid was assumed under the grid-connected
Cano-Ortega mode. The authors validated the proposed algorithm upon three different scenarios to establish
Received: 27 October 2022 the former’s efficiency and efficacy. In addition to these, the optimization results attained from the
Accepted: 23 November 2022 proposed technique were also compared with that of the results from techniques implemented earlier.
Published: 29 November 2022 According to the outcomes, it can be inferred that the presented OGGWO approach outperformed
other methods in terms of cost mitigation and pollution reduction.
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
Keywords: microgrid; multi-objective optimization; optimal scheduling; gradient-based grey wolf
published maps and institutional affil-
iations.
optimizer; renewable energy

Copyright: © 2022 by the authors. 1. Introduction

Licensee MDPI, Basel, Switzerland. 1.1. Literature Review
This article is an open access article
The demand for power has drastically increased in recent years, whereas fossil fuels
distributed under the terms and
are depleting at a fast rate. In this scenario, engineers, researchers, and scientists across the
conditions of the Creative Commons
globe are looking for promising alternative energy resources that are not only renewable,
Attribution (CC BY) license (https://
but also environmentally friendly. Environmental pollution and the resultant global warm-
creativecommons.org/licenses/by/
ing are two major concerns raised around the globe today. Fossil fuel-based traditional
4.0/).

Energies 2022, 15, 9024. https://doi.org/10.3390/en15239024 https://www.mdpi.com/journal/energies

Energies 2022, 15, 9024 2 of 24

power plants that use coal, gas and other non-renewable energy sources are identified as the
primary environmental pollutants. In this background, hybrid energy systems that combine
wind, solar, hydro, and other renewable energy sources have gained attention in the past
two decades, owing to its renewable nature and pollution-less energy production practices.
Further, these hybrid energy systems exhibit high efficiency and reliability. However,
fluctuation in terms of both quality as well as quantity, as a result of environmental condi-
tions, remains the primary disadvantage in renewable energy sources such as Photovoltaic
solar cells, wind energy systems, etc. Since it is challenging to predict the fluctuations in
PV and small Wind Turbines (WT) at the time of power generation, it becomes inevitable to
integrate these sources with other reliable energy sources such as Fuel Cell (FC), Microtur-
bine (MT), battery storage and so on, to ensure a constant supply of quality power. These
hybrid systems can be relied upon as sustainable energy systems. In this background, the
microgrid is one of the promising candidates that is not only energy-efficient, but also can
integrate multiple number of pilot-level Distributed Energy Resources (DERs), such as
microturbine (MT), Wind Turbine (WT), battery units, Fuel Cell (FC), photovoltaic (PV), and
other energy sources. Though it is an advanced concept that was introduced in recent times,
Microgrid (MG) has gained immense popularity owing to multiple advantages provided
by the system. A microgrid can integrate different types of distributed RES to fulfil the
local load requirements and connect or disconnect from the utility grid [1,2].
In literature, a PV-based 17-bus Low Voltage microgrid, connected with the grid, was
analysed in the literature. This microgrid had a Wind Turbine as well as a Fuel Cell (FC) for
the on-site production of hydrogen. The researchers optimally scheduled the storage of
energy so as to provide energy at the right time and to ensure reserve provision [3]. In [3],
the Harmony Search Algorithm was utilized to schedule the storage of energy during a
24-h horizon to mitigate the costs incurred upon operations. Further, the study also widely
leveraged the hydrogen system under all operational scenarios, to balance the uncertainties
that tend to arise from loads and renewable energy sources.
In [4], multi-objective PSO algorithm was proposed to elucidate multi-objective energy
management problem in a grid connected microgrid. A simulation was carried out in a
microgrid test system comprising a wind turbine, solar cell, battery, microturbine, and
a diesel generator. Operating cost and pollutant emission were considered as objective
functions. In order to investigate the performance of MOPSO, three different scenarios
were tested in the simulation. In the study conducted earlier [5], NSGA II algorithm was
applied to implement multi-objective optimal operation of a grid connected microgrid test
system by considering operation cost and pollutant emission as objective functions. In [5],
the uncertainty of the generation capacity of solar and wind power generating units and
the uncertainty of load was considered to elucidate the multi-objective energy management
problem. In literature [6], the researchers evaluated the bi-level optimal operation of a grid
connected microgrid that contains loads, storage devices, and distributed energy resources.
The upper level optimal model for distribution network dispatch was solved by using
a multi-objective optimization approach by considering the microgrid’s power loss and
voltage profile as objective functions. A self-adaptive genetic algorithm was proposed
to implement the optimal operation of the microgrid. The lower-level optimal model
was developed by applying non-linear programming technique to seek the optimal daily
operating scheme of different DGs in the microgrid. The microgrid operating cost was taken
as objective function. In the study conducted by Xinhui Lu, the authors developed a multi-
objective optimal dispatch model to be applied in the microgrid in grid-connected mode.
This model holistically considered the costs incurred upon operations and environmental
protection through the microgrid system [7]. The distribution generators, considered in
this microgrid system, included Electric Vehicles (EVs), diesel engines, WTs, PV arrays,
microturbines, etc. An improved PSO algorithm was developed to overcome the multi-
objective optimal dispatch problem. In literature [8], a two-stage stochastic p-robust
optimal energy trading management system was proposed to be applied in the microgrid.
This system included a microturbine, diesel engine, PV and WT systems. Based on the
Energies 2022, 15, 9024 3 of 24

components involved, the authors implemented a multi-scenario tree approach to develop

different scenarios for unsure parameters such as WTs, loads, PV, and market-clearing prices.
The discretization of all the probability density functions was accomplished via definite
intervals. Then, the authors made use of a differential evolution clustering algorithm to
mitigate the number of scenario sets to accomplish the general scenario.
A hybrid algorithm, integrating Gradient Boosting Decision Tree (GBDT) and Sand-
piper Optimization Algorithm (SOA), was proposed by Prakash Arumugam [9] to achieve
optimal energy management in MG systems under the grid-connected mode. This grid-
connected microgrid had MT, WT, battery, and photovoltaic (PV) systems. The objective of
the proposed technique was to mitigate the cost incurred upon fuel and other areas during
operations and maintenance of grid-connected microgrid system and to reduce the hourly
variations in power generation of the microgrid. The researchers listed state-of-charge for
storage elements, power demand and renewable energy sources as some of the constraints
faced in this research work. In the study conducted earlier [10], the authors proposed an
energy management system on the basis of Grasshopper Optimization Algorithm (GOA)
to find the optimal power generated by Distributed Generators in microgrid. This is crucial
information when it comes to cost mitigation from total power generation. In this study,
the researcher applied the proposed unit upon microgrid that had five generating units:
industrial, commercial, feeding, and residential loads. Further, the study also made use
of the Muddy Soil Fish Optimization Algorithm (MSFOA), a novel algorithm developed
on the basis of the foraging pattern of fishes. The main aim of this algorithm is to reduce
the overhead incurred upon production and mitigate the costs spent during the import of
energy from the grid. Further, the study also considered the system constraints [11]. The
researchers implemented the presented energy management system in a microgrid frame-
work that had energy storage units, solar PV, diesel generators, and wind units. Further,
numerous case scenarios for instance, isolation, and grid-tied scenarios were investigated
under different aspects to prove the potentials of the presented MSFOA approach. In the
study conducted earlier [12], a promising energy dispatch strategy was devised by the
researchers for both the individual microgrid and the grid-connected ones. These micro-
grids were also inclusive of battery Energy Storage Systems (BESS), MT, WT, PV, FC, and a
diesel generator. In this study, the researchers formulated Microgrid Energy Management
(MGEM) mechanism as a mixed-integer linear programming technique. To achieve energy
management in the microgrid, a novel multi-objective solution was proposed along with
a demand response program. In [13], the authors incorporated a multi-objective Particle
Swarm Optimization (PSO) technique to ensure energy resource management and optimal
distribution of the resources in the proposed microgrid system. In this study, emission
and operation cost were considered as objective functions. The researchers calculated
the overall operational cost incurred from the microgrid and pollutant-induced emissions
under three different scenarios. The microgrid was analyzed in detail earlier [14] as an
energy management system when connected with primary power system. The study took
parameters such as WT, PV, and load demand into consideration under probabilistic and
deterministic conditions. Further, an Equilibrium Optimizer (EO) was also utilized to
overcome the energy management problem. In this study, the authors considered operation
cost, voltage deviation, and voltage stability index as objective functions. In the study
conducted earlier [15], an optimization framework was proposed with three objectives for
microgrid energy management in the case of smart homes and Demand Response (DR)
programs. The proposed model was tested using an 83-bus distribution system with 11
microgrids. This study considered load demand variations and uncertainties present in
the output power of Renewable Energy Sources (RESs). Further, a max–min fuzzy method
was utilized to model the objective function as bi-objective and tri-objective models. In
addition to these, the study also considered Peak-to-Average Ratio (PAR), emissions and
operations cost as objective functions. Recently several optimization techniques were
proposed to eliminate energy management problems in microgrids [16,17]. In [18], ANN-
based backtracking search algorithm (ANN-BBSA) and ANN-based binary practical swarm
Energies 2022, 15, 9024 4 of 24

optimization (ANN-BPSO) algorithm was implemented to elucidate multi-objective energy

management problem in a virtual power plant, which was realized as a modified IEEE
14-bus system with five identical microgrids. In the study conducted earlier by the current
study authors, the Local Energy Management System (LEMS) was proposed and imple-
mented under Generalized Power Prediction Model (GPPM) to overcome the performance
uncertainties for varying distributed generations of a microgrid [19]. In [20], the interior
search algorithm was applied to solve the economic load dispatch problem in a microgrid
which comprises diesel generators, fuel cells, and wind turbine. In this study [20], total
operating cost was taken as the objective function. Table 1 shows the contributions made
by research investigations in microgrid energy management. This table covers different
aspects such as optimization algorithm, test system, Renewable Energy Sources (RESs),
and storage technology.

Table 1. A review of optimization techniques in the energy management of a microgrid.

Reference System Storage RES Year

No. Objectives Control Algorithm Description Technology
Hybrid AC and DC
[21] Operation cost Modified bat algorithm Battery PV, WT, tidal 2021
microgrid
Multi-Objective Particle
[22] Operation cost and Swarm Optimization Grid connected Battery PV, WT 2017
Emission microgrid
(MOPSO)
Operation cost, Binary Orientation Search Grid connected
[23] Emission Battery PV, WT 2022
Algorithm microgrid

[24] Operation cost Θ-modified krill algorithm Grid connected Battery WT, PV 2021
microgrid

[25] Decentralized energy HOMER Software Islanded DC Hydrogen, PV 2020

management microgrid Battery
Hierarchical energy HOMER pro Islanded DC
[26] management Hydrogen, BESS PV 2019
Strategy Software microgrid

Energy Management PV, WT, 2020

[27] System Hybrid automata algorithm Islanded Microgrid Hydrogen, BESS biomass
Preference-based
[28] Multi-microgrids multi-objective reinforcement Grid connected Battery PV, WT, tidal 2022
energy management Microgrid
learning (PMORL) technique
Optimal power flow
with reactive power Generalized reduced-gradient Grid connected
[29] BESS PV, WT 2020
cost of MG as (GRG) algorithm Microgrid
objective function
Stochastic optimal
energy management
[30] of MG with operation GAMS using CPLEX solver Grid connected BESS, PHEV, TES PV 2018
Microgrid
cost and emission as
objectives
Energy Management NSGA-II Grid connected Battery PV, WT 2019
[31] System Microgrid
Energy Management Grid connected BESS PV, WT 2020
[32] System Modified bat algorithm (MBA) Microgrid

Energy Management Particle swarm optimization Grid conncted and

[33] System autonomous BESS, PEV PV, WT 2021
(PSO) Microgrid
Mixed integer linear Grid connected
Energy Management
[34] programing technique using Microgrid with PV, Battery PV, WT 2019
in Microgrid
CPLEX solver FC, MT, battery
Energy Management Branch and reduce Microgrid with PV,
[35] System of DC optimization navigator Battery PV, WT 2021
(BARON) algorithm FC, MT, DE, battery
microgrid
Stochastic energy
[36] management of smart Quantum Particle Swarm Grid connected Battery PV, WT 2021
Optimization (QPSO) Microgrid
microgrid
Energies 2022, 15, 9024 5 of 24

Table 1. Cont.

Reference System Storage RES Year

No. Objectives Control Algorithm Description Technology
meta-dynamic-approach-
Energy management Grid connected
[37] based multiobjective flower Battery PV, WT 2021
in microgrid Microgrid
pollination algorithm
3-unit
Economic Emission Hybrid Modified version of RES integrated low
[38] Dispatch in Microgrid - PV, WT 2021
GWO voltage microgrid
system

Energy management 3,5,6,7,11, 38-units

[39] System Crow Search Algorithm microgrid test BESS PV, WT 2019
system
Islanded Microgrid
Energy management Modified version of artificial with stationary and
[40] System BESS PV, WT 2020
bee colony algorithm the dynamic energy
Bid

1.2. Research Gap and Objective of the Paper

The authors conducted an extensive review of literature from which knowledgeable
insights have been gained with regards to energy management challenges in dynamic
systems and different test systems and entities. However, most of the research works
conducted earlier emphasized the inclusion of specific multi-objective optimization algo-
rithm to achieve energy management in dynamic test systems. These research studies
were conducted on the basis of the Pareto front, utilizing multi-objective techniques and
attained energy management. From the review of literature and comparative analysis
conducted among two or more than two multi-objective energy management methods,
the authors found a research gap that needs to be fulfilled. In the literature [41], the mi-
crogrid test considered single and multi-objective optimization to be implemented for
the first case study. However, for the second and third case study, only single objective
optimization was done for the optimal scheduling of the microgrid. In this current research
study, we implemented the single and multi-objective optimization using the proposed
optimization algorithm for the optimal scheduling of the microgrid for all the case studies
considered. A comparison was made between the optimization results obtained by the
proposed method and four other different optimization algorithms coded by us to find the
better-compromised solution for less pollutant emissions and minimum generation cost.
In this research paper, we have proposed the Oppositional Gradient-based Grey Wolf
Optimizer (OGGWO) for elucidating the multi-objective optimal scheduling of a grid con-
nected microgrid. Opposition based learning is incorporated with the proposed algorithm
in order to enhance the solution quality and convergence speed. The incorporation of the
opposition-based learning encompasses the instantaneous consideration of an estimate
and its equivalent opposite estimate so as to accomplish an enhanced approximation for
the present candidate solution. Our proposed optimization algorithm utilized the advan-
tages of the gradient feature that offers valuable information about the solution search
space. This incorporation of the gradient feature explores the solution search space more
perceptively with the consideration of the gradient direction in the problem search process
and aids in attaining the global optimal solution. In order to enhance the exploration
and exploitation capabilities of the proposed OGGWO algorithm, Gaussian walk and
Levy flight techniques were incorporated, and the results were compared with the results
obtained using EDNSGA-II, NSGA-II, GGWO, GWO, CSA, and PSO to prove the effec-
tiveness of the proposed algorithm. GGWO, GWO, CSA, and PSO algorithms were coded
by us for comparison purposes. The results of EDNSGA-II and NSGA-II were taken from
the literature.

1.3. Contributions
The following is the list of contributions made by current research work for the state-
of-the-art energy management system proposed in this study.
Energies 2022, 15, 9024 6 of 24

i. At first, OGGWO was implemented as an optimization tool to overcome the problem.

Further, the efficiency and robustness of the proposed model were determined and
contrasted against GGWO and conventional GWO.
ii. The authors compared the outcomes of the proposed method under three different
scenarios to identify the optimization algorithm that produces the best compromised
solution between generation cost and the emission of pollutants.
iii. Multi-objective optimization was carried out under all the case studies considered
so as to find the best-compromised solution for the multi-objective energy manage-
ment problem.

2. Problem Formulation
The primary objective behind the optimal operations of the microgrid, with Distributed
Energy Sources (DERs) that include non-renewable and renewable energy sources, is to
identify the optimal operating points of DERs to generate power during the scheduled
horizon. The key aim of the optimal operations of the microgrid that is incorporated with
DERs such as RESs and non-RESs is to find out the optimal operating points of DERs so
that the power can be produced at the time of scheduled horizon. This objective should be
met to reduce the cost incurred upon operations and reduce the emission rate in parallel.
Thus, the current study aims at achieving two different and conflicting objectives together,
i.e., emission reduction and mitigation of cost incurred upon operations.
In the current study, the authors propose a precise mathematical model to achieve
short-term energy management so as to mitigate the costs spent upon operations and
reduce the emission of pollutants in microgrid operations.

2.1. Objective Functions

As mentioned above, the current study considered the following objective functions
such as operating costs and costs incurred upon reducing the pollution emissions.

2.1.1. Mitigation of Operating Cost

In Equation (1), the authors calculate the total operating costs incurred in a microgrid.
In this equation, T corresponds to the total time taken, by the study, in terms of hours,
whereas Ng denote the generation of energy and Ns correspond to storage units. Here,
Ui (t) corresponds to the status of ith unit at time t (either it is switched on or off) PGi (t)
denote the output power amount of ith storage unit whereas in case of jth at time, t. The
price of the energy rendered for ith storage unit is denoted by BGi (t) while it was BSj (t) for
jth storage unit at time, t. SGi (t) and Ssj (t) denotes the costs involved during start-up or
shut-down functions spent upon ith storage unit and jth storage unit respectively. Further,
PGrid (t) and BGrid (t) denote the quantities of power that was exchanged with the offered
market, at time, t [13].

Ng
( )
T ∑i=1 [Ui (t) PGi (t) BGi (t) + SGi |Ui (t) − Ui(t + 1)|] − . . .
Min f 1 ( X ) = ∑ i =1 Ns 
∑ j=1 j (t) Psj (t) Bsj (t) + Ssj Uj (t) − Uj (t + 1) − ( PGrid (t) BGrid (t))
U
(1)

2.1.2. Mitigation of Emission

For the next objective function, i.e., emission mitigation, a few components were
considered such as carbon dioxide (CO2 ), nitrogen oxides (NOx ), and sulphur dioxide
(SO2 ), i.e., atmospheric pollutants. Equation (2) shows the mathematical formulation for
overall pollutant emission in kg [13]:

n o
T Ng Ns
Min f 2 ( X ) = ∑t=1 ∑i=1 [Ui (t) PGi (t)EGi (t)] + ∑ j=1 [Uj (t) Psj (t)Esj (t)] + ( PGrid (t)EGrid (t)) (2)
Energies 2022, 15, 9024 7 of 24

EGi (t) denotes the volume of pollution emitted by ith generation unit, while for the jth storage
unit, it was denoted by Esj (t). In the case of volume of pollution at market, it was denoted by EGrid (t)
at time, t. All the values are denoted in kg/MWh, respectively.

2.1.3. Constraints and Limitations

(a) Power Balance constraints

Nk Ng Ns
∑ PLK (t) = ∑ [ PGi (t)] + ∑ [ Psj (t)] + ( PGrid (t)) (3)
k =1 i =1 j =1

Here, PLK corresponds to K amount at load level, whereas Nk denotes the whole count of load
levels available in the grid.
(b) Ramp Rate Constraints
This constraint occurs as a result of fluctuations in output power. It can be either an increasing
fluctuation or vice versa in DGs. The following is the equation used to describe the constraint [13]:

Ridown .∆t ≤ P(h)i − P(h − 1)i ≤ Riup .∆t (4)

Here, Ridown denotes the ramp-down of ith DG output power whereas Riup correspond to ramp-
up of ith DG output power. Further, ∆t denotes the time step taken, in terms of hours.
(c) Inequality constraints
All the units considered in this study have both upper, as well as lower thresholds for their
power generation capacity. The units include DGs, storage units, and the market [13].

PGi,min (t) ≤ PGi (t) ≤ PGi,max (t)

Psj,min (t) ≤ Psj (t) ≤ Psj,max (t) (5)
PGrid,min (t) ≤ PGrid (t) ≤ PGrid,max (t)

Equation (6) covers the limitations of charging and discharging rates of the storage unit.

SOCsj (t) = SOCsj (t − 1) + Pchg/Dchg (t)

(6)
0 ≤ | Pchg/Dchg (t)| ≤ PCDSj,max

Here, SOCsj (t) denotes the charging amount of the storage unit at current time and SOCsj (t − 1)
denotes the charging amount of the storage unit at previous time. Pchg/Dchg (t) corresponds to the
charging (discharging) amount at tth hour while the maximum charging (discharging) rate is denoted
by PCDSj,max .

3. Oppositional Gradient-Based Grey Wolf Optimizer

Mirjalili et al. (2014) proposed GWO inspired from the hunting behavior of grey wolves [41].
In general, there are four types of wolves, such as the Alpha, followed by Beta, Delta, and finally
Omega [42]. In a pack of wolves, Alpha is the dominant wolf and is the decision-maker for the whole
pack, while the rest of the members in the pack must obey the alpha’s commands and decisions.
Further, only the alphas breed in the pack. The second position is occupied by Beta wolves, which
act as a mediator between the alpha wolf and the rest of the wolves. It helps the alpha wolf and
becomes the first choice of nominee for alpha, in case an Alpha wolf dies or is too old for swarm
management. Though beta wolves control other wolves of the swarm, it fulfils the order of Alpha
wolf. Hierarchically, Omega is the least ranked grey wolf [43] and follows the rest of the high-ranking
wolves. Those wolves that are not categorized as the other three, i.e., Alpha, Beta, or Omega fall
under the category Delta Wolf. Deltas tend to manage the Omega wolves, while at the same time it
also helps the Alpha wolf and Beta wolves.
It is important for an algorithm to ensure there is equilibrium between exploration and ex-
ploitation skills so that the exploration can be achieved. GWO continues to look for solution space
through constant upgradation of the position of dominated wolves such as Alpha, Beta, and Delta.
Two new features have been added to the traditional GWO algorithm in this study, which resulted in
Oppositional Gradient-Based Grey Wolf Optimizer (OGGWO). The first feature is the adoption of a
novel procedure that uses gradient data to improve the algorithm’s exploitation as well as exploration
skills. In most of the optimization challenges, the gradient tends to render the critical information
Energies 2022, 15, 9024 8 of 24

with regards to the shape of the solution space. This is done by measuring the steepest slope at
every point in the solution space. Afterwards, the particles are transferred to a nearby local optima
using gradient information, while at the same time, an appropriate exploration capability is also
maintained. GWO is heavily empowered by such updating operators. It results in the efficient
searching of solutions in solution space and enhances the exploration potentials of the algorithm
to side-step local optima. Opposition Based Learning (OBL) is one of the powerful optimization
tools to enhance the convergence speed of numerous heuristic optimization techniques [44]. The
successful implementation of OBL involves the evaluation of population at both the opposite site and
current site in the same generation so that a better candidate solution can be obtained for the given
problem. The concept of OBL has successfully been applied in various meta-heuristics to enhance
the convergence speed. The concept of opposite number needs to be defined to explain OBL [44]. A
novel procedure is incorporated to update the formulations for Omega wolves in OGGWO algorithm
as shown in (7) and (8).
Both Gaussian walks as well as Levy Flight (LF) are added as second features to the existing
traditional GWO. These random walks enhance the random nature of the presented OGGWO model
and empower it for exploration. Though self-similar clusters (or trajectories) are created by Levy
Flight and Gaussian walks, significant differences are found between their structures [45]. Gaussian
walk produces a minor, but dense cluster that contains numerous small steps and executes the same
number of iterations, such as with clusters [46]. Having been randomly chosen, these methods
improve the proposed OGGWO model’s exploration capability by helping the algorithm avoid local
optima. In this case, OGGWO switches randomly between Gaussian walk, as well as LF and make use
of both the methods [46]. Figure 1 shows the flowchart of the proposed Oppositional Gradient-based
Grey Wolf Optimization algorithm. The following formulations are used in OGGWO to upgrade the
Energies 2022, 15, x FOR PEER REVIEW
Omega wolves’ position in the solution space at the time of closing every iteration11asofshown
28 in (12)
and (13).

Figure 1. A flowchart of the Oppositional Gradient-based Grey Wolf Optimizer.

Figure 1. A flowchart of the Oppositional Gradient-based Grey Wolf Optimizer.

4. Fuzzy Logic-Based Collection of the Finest Compromise Solution

In general, it is important to select the optimal compromise solution out of the
available options prior to making any decisions. In this study, a fuzzy membership ap-
proach was used by the author to identify the optimal compromised solution [48]. In 𝑗 𝑡ℎ
objective function, 𝑓𝑗 of individual k is characterized by a membership function µ𝑘𝑗 ,
Energies 2022, 15, 9024 9 of 24

Step 1: The position of grey wolves (search agents) is randomly initialized in the search space.
This step further fixes the number of iterations, followed by population size (number of wolves).
Step 2: For each search agent, the fitness value is determined as it denotes the distance between
prey and the wolf.
Step 3: The opposite points are initialized and used to generate the opposite population so as to
calculate the fitness values of every individual population.
Step 4: Sorting is executed for both current and opposite populations (pop and opop respec-
tively), in line with their fitness values
Step 5: The n P number of the fittest solution is selected out of a combination of current and the
corresponding opposite population.
Step 6: Based on fitness values, three solutions are found such as best (a), second-best (b) and
finally, the third best (d). These solutions correspond to alpha, beta and delta category wolves,
respectively.
Step 7: Based on Equation (7), the position of the grey wolves is modified.

i ∂ f Min ∂f ∂ f Max
XW ( t + 1) = γ . ≤ ( t ) < γ f or i = 1, . . . , m and w = 1, . . . , n, (7)
∂X i ∂Xw i ∂X i
 
i ∂f
XW (t) − rand(0, 1)λi (t) i
( t ) , Otherwise, f or i = 1, . . . , m and w = 1, . . . , n, (8)
∂Xw
Here, i corresponds to the index of decision variables in an optimization problem. On the
∂ f Max ∂ f Min
contrary, n denotes the count of grey wolves. ∂Xi and ∂Xi correspond to the highest positive and
the least negative slopes, at every dimension, during every iteration of the algorithm. Further, γ
corresponds to a continuous parameter that is calculated as (0,1]. As per the formulation given above,
λi is updated using Equation (9) as given herewith.

0.1[Ubi − Lbi ]
λi ( t ) =  Min (9)
∂ f Max

∂f
max ∂Xi , ∂Xi

Here, LB corresponds to Lower Bound, whereas UB denotes the Upper Bound of the problem.
This results in the following equation straightforwardly.

∂f
λi ( t ) i
(t) ≥ 10(Ubi − Lbi ) (10)
∂Xw

The gradient of the problem in few optimization problems may remain unknown. This is
attributed to the non-differentiability of objective function or discrete features possessed by decision
variables. In order to overcome these issues, the following equation is presented.

∂f [ f (t) − f (t − 1)]
= (11)
∂x [ X (t) − X (t − 1)]

Step 8: The fitness value is updated based on the modified position of grey wolves.
Step 9: Alpha, beta and delta values are updated
Step 10: The position of omega wolves is updated using the Equations (12) and (13) that employ
Gaussian walk and Levy flight.
i
XW,new i + KGaussian (| ϑ |, σ ) − ( ξ × ϑ − ξ 0 × X i ),
= XW i i W (12)
f or i = 1, . . . , m and w = 1, . . . , n,

i
XW,new = XW i + X i Levy ( η ),
W (13)
f or i = 1, . . . , m and w = 1, . . . , n,
Here, the best solution is denoted by ϑi whereas |σ| correspond to the standard deviation
of Gaussian distribution. The Gaussian parameter is changed by OGGWO as σ = |K × ( xi − BP)|,
log(l )
while it also reduces the length of steps at the time of iterations by fixing = l . Here, l corresponds
i
to the number of iteration. Further, XW,new corresponds to the new position of the wolves, whereas
Energies 2022, 15, 9024 10 of 24

i
XW denotes the current position. In addition to these, ξ 0 and ξ denote random numbers in (0,1].
Equation (14) denotes the calculation for Levy flight.

[0.01 × σ × r1 ]
Levy( x ) = 1
(14)
|r2 | β
Here, the random numbers are denoted by r1 and r2 between (0, 1]. β corresponds to constant,
which is equal to 1.5. In Equation (13), σ is computed with the help of following equation.
1/β
Γ (1 + β )

πβ ( β −1)
σ = Γ(1 + β) sin( )/( β[2 2 ]) (15)
2 2

Step 11: The opposite population is created out of the current population using the jumping rate.
Step 12: The n P number of the fittest solution is selected after integrating the current and the
opposite populations.
Step 13: The steps 3–7 are repeated until the maximum number of iterations is achieved.
Step 14: The best solution is achieved as the output.
As per the steps discussed above, the leading framework remains the same in both GGWO and
OGGWO. However, significant changes are made in the latter. For example, OGGWO leverages a mix
of original and gradient-based operators to update the position of wolves more than the conventional
GWO operators. Further, a new function is also added in OGGWO to determine the gradient’s
objective function at every point in the solution space [47]. In addition to the above, Gaussian walk
and Levy Flight have been incorporated in OGGWO to increase randomness during the closure of
every iteration. This phenomenon significantly increases both exploration and exploitation capability,
as long steps and short steps shift the particles in solution space.

4. Fuzzy Logic-Based Collection of the Finest Compromise Solution

In general, it is important to select the optimal compromise solution out of the available options
prior to making any decisions. In this study, a fuzzy membership approach was used by the author
to identify the optimal compromised solution [48]. In jth objective function, f j of individual k is
characterized by a membership function µkj , owing to the indefinite characteristic of the conclusion
of the decision maker as denoted herewith.
min

 max 1 f j ≤ f j

 f − f j
µkj = f max − f min f jmin < f j < f jmax
j
(16)
 j j
max


j0f ≥ f j

Here, f jmax corresponds to the highest value of the jth fitness function. In addition to this, its
least value is denoted by f jmin among the rest of the non-dominated solutions. The calculation is done
for normalized membership function µk that is applicable for each non-dominated solution k.

∑N k
j =1 µ j
µk = (17)
∑rk=1 ∑ N k
j =1 µ j

Here, r corresponds to the whole number of non-dominated solutions. The highest value, i.e., µk
is achieved by the best compromise solution.

5. Modeling of the Microgrid

In the case of applying DGs individually, it creates multiple problems. So, it is suggested to
follow a systematic approach so that generation, as well as associated loads, is considered as either
the microgrid or sub-system to avoid the challenges involved in it and leverage the true potentials
of DGs. In parallel, the DGs in the microgrid can be combined together and analyzed about the
exploitation of renewable energy sources in huge quantities regarding economy, technology, and
environment within the target system. These insights help in making knowledge-based decisions
and manage the operations in a better manner. In addition to this, Distribution Generation is a
compendium of a vast range of prime mover technologies that include fuel cells, gas turbines, micro
turbines, solar PV, IC engines, and wind power. Such RES-based technologies are cost-efficient and
emit fewer pollutants, contradicting a huge economy [17]. For instance, Fuel Cells generate electricity
Energies 2022, 15, 9024 11 of 24

out of oxygen and hydrogen, whereas water vapor is the only emission from this source. Having
said so, natural gas and other fuel reformation emit NOx and CO2 . Fuel cells are comparatively more
efficient than microturbines, in terms of emission, which incurs heavy costs. In the current research
work, the researchers considered a commonly-known LV microgrid based on the study conducted
earlier that included various types of DGs such as PV, WT, MT, low-temperature Fuel Cell (PAFC),
and storage devices such as lead-acid batteries [38]. The authors assumed that the active power,
produced by whole DG sources, stands as unity power, while at the same time, it does not request
or produce reactive power. Furthermore, a power exchange link is also used between utility (LV
network) and the mentioned microgrid for the purpose of energy trading at different hours in a day,
in line with the decisions taken by the MGCC (Microgrid Central Controller).

6. Results and Discussion

As per the literature [41], the current study considered a microgrid test system with a distributor
and a few DGs such as Wind Turbine (WT), Fuel Cell (FC), Photovoltaic panel (PV), Microturbine (MT),
and battery. The model, suggested in this study, has two objective functions, such as whole emission
of the pollutants and microgrid costs, which include the costs incurred upon power generation and
start-up/shut-down costs of units. The chosen problem was rectified through three different scenarios
such as main case, secondary case, and third case. In the main issue, the whole set of units were
dispatched based on their original drawbacks. In the second case, photovoltaic (PV), as well as Wind
Turbine (WT) functions were assumed to be at its maximum output (Max-Ren). For the final case,
the utility was considered as an unconstraint unit which can exchange energy with the microgrid
without any limits (WLE). The microgrid was expected to meet the total load demands as predicted
in [41]. One prime residential area, one feeder with light usage commercial consumers, along with
one industrial feeder that supplies power to a small workshop in a typical day was considered. The
total energy demand of the day considered was 1695 kWh.
In addition to these, the energy price difference for a real-time market was sourced from the
literature [41] for the day considered. In order to ensure flawless microgrid operations, three DG units
with WT, PV, and MT were used. The optimization algorithm has the ability to assign these units with
on and off states, when there is a power dispatch problem, considering both objectives. Likewise,
the on/off state was regarded as 1 for the utility in all cases, since the microgrid was operated in
grid-connected mode. To analyze the overall impact of PAFC and battery upon grid operations and
to leverage the optimal benefits out of the resources, the on state was selected with an intention for
relevant units. From the study conducted earlier, both minimum as well as maximum generation
limits were taken for DG sources [41]. Further, the same source was also used to provide the bid
coefficients, in terms of Euro cents per kWh. In order to perform the data analysis in a rapid fashion,
all the units in the current research work were considered to function under electricity mode only.
The microgrid is preferred for installation since it has a salient feature, i.e., the ability to combine
multiple sorts of renewable energy sources together.
Microgrid forecasts are highly important for actual operations since this information can be
utilized to prepare the systems to be flexible and appropriate. Since it is impossible to execute
the renewable energy operations in a conventional manner, it is easy to predict its behavior. Thus,
microgrid system efficiency can be enhanced with the help of forecasted information. An expert
prediction model was used in the current research work to evaluate the power outputs from PV
and WT. From the literature [41], the daily maximum power output from PV and WT values were
extracted. Further, the authors used the same study to source information on both daily load power,
as well as energy market price in a typical microgrid.
In order to validate the efficacy of the proposed optimization algorithm, simulation was con-
ducted for 50 trial runs to migrate the cost incurred upon operations. The current study considered
population size (50) and 1000 maximum iterations as its controlling parameters. Tables 1 and 2 show
the characteristics of 5 DG sources which generate electricity in the microgrid. Through the common
coupling point, the microgrid either shares surplus energy or demands additional energy from the
utility. The whole set of units that include the macrogrid (utility), are able to function within their
power limits, while it can meet the requirements in parallel.
Energies 2022, 15, 9024 12 of 24

Table 2. An optimal generation schedule for Case-I.

Hour MT FC PV WT ESS Grid

1 16.17 30 0 8.84 −30 36.99
2 19.74 26.98 0 10.62 −4.79 7.45
3 6 22.01 0 8.01 −20.05 44.03
4 6.75 30 0 12.44 −12.38 24.19
5 8.68 23.02 0 12.01 −19.98 43.27
6 25.28 0 0 13.56 −18.13 54.29
7 25.02 23.9 0 12.03 −25.57 48.62
8 0 30 0 13.37 2.34 44.29
9 27.39 27.03 0.17 15.19 −27.23 48.45
10 29.99 30 2.04 19.34 29.05 −14.42
11 30 29.51 8.03 24.26 26.18 −24.98
12 29.18 28.03 9.72 22.03 11.03 −11.99
13 30 29.86 11.03 19.2 −29.98 25.89
14 0 30 8.91 24.09 29.13 −6.13
15 30 30 8.44 24.99 27.24 −29.67
16 22.48 24.66 3.99 19.98 −20.23 45.12
17 30 30 1.98 23.47 −29.98 46.53
18 30 0 0 18.98 9.81 46.21
19 30 22.19 0 19.01 1.02 35.78
20 30 30 0 22.31 −0.59 22.28
21 17.99 19.79 0 12.99 6.94 35.29
22 1.02 28.93 0 21.08 27.98 5.99
23 22.05 0 9 13.09 −8.16 42.02
24 0 0 0 20.11 29.41 17.48

6.1. Case I: Operation of Distributed Energy Sources with Specified Limits

The first test case, considered in this study, was the operation of all DGs and grids within the
specified limits as represented in [41]. Table 2 shows the optimal generation schedule for 24 h to
minimize both cost and emission levels. From Table 3, it can be understood that the majority of the
load demands, during the first few hours of the day, was met by fuel cells present inside the grid and
utility through the common coupling point. This is attributed to fewer bids of relevant units than the
rest of the others for a specific period. Since the demand increased for high number of utility bids in
the next few hours, the DG units enhanced their output power based on the priorities such as low
cost and low emission level.
Further, the microgrid regulatory controller sequentially supplied the units and exported the
energy from the microgrid, which in turn increased the revenue and reduced the net emission for
a specific period under study. On the other hand, the battery was charged for the first few hours a
day, especially when the low cost is charged during this period. When the load curve achieved the
highest value in the later hours of the day, the discharge action was executed. When RESs such as
solar and wind are deployed, it emits less pollutants, whereas the operating cost increases. In other
terms, these energy sources are economically unviable due to which the energy generated from these
sources needing to be confined to a limit based on emission/economic limitations.
Table 3 shows the analytical outcomes of optimization algorithms and compares the results
achieved for the main case. Based on the results attained, in terms of best and worst solutions, it can
be inferred that the proposed optimization algorithm accomplished an excellent outcome for cost and
emission objectives. Further, it also attained faster convergence characteristics. In addition to this,
the proposed algorithm proved to have another advantage for optimization, based on the statistical
indices of average and standard deviation.
Table 3 shows the SD values attained by the proposed algorithm for cost and emission objectives,
which were 2.55 and 4.72, respectively, which infers that the proposed algorithm is an excellent
choice for optimization. In order to showcase the performance of the proposed OGGWO algorithm,
Figures 2 and 3 portray the convergence characteristics held by OGGWO, as well as the rest of the
optimization algorithms for the best solution and for every single objective. The illustrations show that
the proposed method achieved the minimum value for cost objective function after 705 iterations and
remained unchanged thereafter. On the other hand, the GWO algorithm converged at 713 iterations.
With regards to emission objective function, the proposed method reached the minimum value at
734 iterations, whereas GWO algorithm required 745 iterations for convergence.
By applying the fuzzy logic based approach as discussed in Section 4, the global best compromise
solution was attained by the proposed algorithm to mitigate the operating cost and emission.
Figure 4 shows the pareto-fronts of respective trade-off objectives, obtained by various meth-
ods, used for comparison and the best-compromised solution. The figure infers that the proposed
algorithm was effective in elucidating the non-linear multi-objective optimization problem since the
Pareto optimal fronts of non-dominated solutions were distrusted in a good manner. In addition to
Energies 2022, 15, x FOR PEER REVIEW 15 of 28

Energies 2022, 15, 9024 Min Emission 163.97 1175.9913 of 24

BCS 158.81 1212.7
Std Dev 2.9 5.08
Min Cost 165.49 1236.5
this, the computational time consumed by the proposed algorithm to mitigate both emission, as well
NSGA-II
as operating Min Emission
cost, was lesser. This reveals 173.77
the quality of the proposed algorithm. The1199.08
optimization
results amply [41]
corroborate the ability of BCS
the proposed algorithm to168.27 the issues associated
tackle 1216.54 with
equality and inequality that tend to happen Std in energy management3.15
Dev challenges. 7.08
Min Cost
Table 3. A statistical comparison of simulation results for case-I. 168.15 1241.29
Min Emission 176.04 1203.12
PSO
Optimization Parameter Cost Euro Emission Kg
Algorithm BCS 171.38 1220.07
Min CostStd Dev 148.67 4.28 8.14
1230.66
Min Emission 163.97 1175.99
EDNSGA-II [41] BCS Min Cost 158.81 167.38 1239.86
1212.7
Std Min
Dev Emission 2.9 175.42 1202.18
5.08
CSA Min Cost 165.49 1236.5
Min EmissionBCS 173.77 170.46 1219.14
1199.08
NSGA-II [41] BCS 168.27 1216.54
Std DevStd Dev 3.15 4.06 7.99
7.08
Min Cost
Min Cost 168.15 161.14 1241.29
1234.08
Min Emission 176.04 1203.12
PSO Min Emission
BCS 171.38 170.78 1195.64
1220.07
GWO Std Dev 4.28 8.14
Min Cost BCS 167.38 164.26 1214.31
1239.86
Min Emission
Std Dev 175.42 3.09 1202.18
6.87
CSA BCS 170.46 1219.14
Min Cost
Std Dev 4.06 147.39 1228.14
7.99
Min Cost 161.14 1234.08
Min
Min EmissionEmission 170.78 162.04 1173.67
1195.64
GGWO
GWO BCS BCS 164.26 157.13 1214.31
1212.03
Std Dev 3.09 6.87
Min CostStd Dev 147.39 2.71 4.88
1228.14
Min Emission 162.04 1173.67
GGWO BCS Min Cost 157.13 146.44 1226.71
1212.03
Std Min
Dev Emission 2.71 161.28 4.88
1172.48
OGGWO Min Cost 146.44 1226.71
OGGWO
Min EmissionBCS 161.28 156.31 1211.28
1172.48
BCS 156.31 1211.28
Std DevStd Dev 2.55 2.55 4.72
4.72

Figure 2. The cost convergence characteristic for Case-I.

Energies 2022, 15, x FOR PEER REVIEW 16 of 28
Energies 2022, 15, x FOR PEER REVIEW 16 of 28

Energies 2022, 15, 9024

Figure 2. The cost convergence characteristic for Case-I. 14 of 24
Figure 2. The cost convergence characteristic for Case-I.

Figure 3. The emission convergence characteristic for Case-I.

Figure 3. The emission convergence characteristic for Case-I.
Figure 3. The emission convergence characteristic for Case-I.

Figure 4. The emission cost trade-off characteristic for Case-I.

Figure 4. The emission cost trade-off characteristic for Case-I.
Figure 4. The emission cost trade-off characteristic for Case-I.
Energies 2022, 15, 9024 15 of 24

6.2. Case II: Operation of Renewable Energy Sources at Maximum Limits

Both PV, as well as WT, was considered to operate at maximum power in this case for every
hour, whereas the rest of the sources considered such as grid, battery, MT, and FC were operated at
permitted levels, as per the literature [41]. Table 4 shows the comparison of statistical performances of
OGGWO and the rest of the optimization algorithms for each objective. When the results achieved by
GWO were compared for each objective with the proposed OGGWO algorithm, the outcomes inferred
that the GWO algorithm was deficient and its performance was inconsistent. For instance, Table 5
shows the performance of the proposed algorithm with regards to the multi-objective optimization
and allocation of optimal power to units. The table implies that heavily-performing units must be
leveraged to attain economic and emission objectives simultaneously. As per this inference, both
the battery, as well as PAFC, is suggested to be extensively used, whereas the rest of the units are
to be used as and when required. In the case of light load requirements, i.e., during the first few
hours of the day, surplus energy can be exported to the macrogrid. However, if the load requirement
is high, the reverse option is preferred. So, utility-supplied energy plays an important role in
sustainable operations.
In the second scenario, an assumption was made that RESs such as WT and PV produce the
highest amount of power for every hour in a day. On the other hand, the other generators such as
a battery, PAFC, MT, and utility function in normal mode, similar to their behavior found in the
primary case. In this scenario, the proposed algorithm was again incorporated and validated for the
multi-objective optimization problem. The results were attained and inferred through a comparison
with that of the other algorithms. Table 5 compares the statistical performances of the algorithms
discussed earlier. In this comparison, every single objective underwent 50 trials.
From the table, it can be inferred that the multi-objective optimization problem was appro-
priately handled by the proposed algorithm while it also ensured a minimum diversity for every
objective and trial, compared with other optimal solutions. The total operation cost of the microgrid
was increased in the second scenario when compared with the main case. However, the net emission
was reduced compared to the previous case. This is attributed to the fact that when RESs penetrate
heavily into the grid network, it reduces the net emission. However, it incurs high costs upon the
operations for a specific period under study.

Table 4. A statistical comparison of simulation results for case-II.

Optimization
Parameter Cost Euro Emission Kg
Algorithm
Min Cost 190.6 NA
Min Emission NA 1162.4
EDNSGA-II [41]
BCS NA NA
Std Dev NA NA
Min Cost 194.24 1205.29
Min Emission 203.15 1168.12
PSO
BCS 198.57 1184.89
Std Dev 5.98 7.63
Min Cost 193.38 1203.14
Min Emission 201.08 1166.58
CSA
BCS 196.78 1183.28
Std Dev 5.79 7.58
Min Cost 189.73 1200.23
Min Emission 198.29 1160.25
GWO
BCS 192.04 1179.27
Std Dev 4.76 6.64
Min Cost 187.13 1197.38
Min Emission 202.87 1158.03
GGWO
BCS 190.48 1177.49
Std Dev 4.43 4.63
Min Cost 186.27 1210.24
Min Emission 201.89 1156.79
OGGWO
BCS 188.23 1175.82
Std Dev 4.26 4.46
Energies 2022, 15, 9024 16 of 24

Table 5. An optimal generation schedule for Case-II.

Hour MT FC PV WT Battery Grid

1 29.98 0 0 9.15 −28.13 51
2 17.22 16.89 0 10.84 −16.93 31.98
3 6.75 21.56 0 11.56 −18.31 38.44
4 23.15 3 0 12.68 −20.89 43.06
5 6.23 19.23 0 13.29 −29.98 58.23
6 26.99 3.73 0 13.41 −25.12 55.99
7 26.86 24.91 0 14.78 −24.89 42.34
8 6.91 25.99 0 13.95 −14.98 58.13
9 0 30 0.45 19.12 24.31 17.12
10 30 30 2.29 19.25 27.45 −12.99
11 30 30 8.02 24.11 30 −29.13
12 29.98 0 10.23 22.43 29.92 −4.56
13 0 21.56 10.75 18.79 −5.99 40.89
14 30 30 10.18 24.12 23.11 −31.41
15 0 0 9.78 24.87 19.23 37.12
16 30 30 5.09 24.92 −29.98 35.97
17 24.35 28.15 0.56 23.12 −26.99 52.81
18 11.98 20.03 0 20.54 18.22 34.23
19 29.95 30 0 20.79 −29.98 57.24
20 29.95 3.56 0 22.59 −8.34 56.24
21 30 30 0 22.34 −30 40.66
22 27.98 0 0 23.89 29.73 3.4
23 19.98 19.45 0 22.57 −19.23 35.23
24 8.78 20.12 0 20.61 −20.48 37.97

The convergence characteristics for operating cost and emission, attained by OGGWO, GGWO,
GWO, CSA, DE, and PSO are shown in Figures 5 and 6, respectively. These figures infer that the
proposed OGGWO algorithm possesses stable and quick convergence characteristics to find the
optimal solution. Figure 7 showcases the distribution of non-dominant solutions, which infers the
Pareto optimal fronts of OGGWO and GGWO algorithms. The figure implies that the proposed
OGGWO algorithm is a promising candidate to overcome the multi-objective energy management
problem compared to the rest of the optimization techniques. In addition to the above, the proposed
Energies 2022, 15, x FOR PEER REVIEW 19 of 28
OGGWO algorithm ensured the distribution of non-dominated solutions in the best manner. It also
assured that the solutions are feasible for large-scale standard test systems.

Figure5.5.The
Figure Thecost
costconvergence
convergence characteristic forCase-II.
characteristic for Case-II.
Energies 2022, 15, 9024 17 of 24

Figure 5. The cost convergence characteristic for Case-II.

Energies 2022, 15, x FOR PEER REVIEW 20 of 28

Figure6.6.The
Figure Theemission
emissionconvergence
convergencecharacteristic
characteristicfor
forCase-II.
Case-II.

Figure 7.
Figure 7. The
The emission
emission cost
cost trade-off
trade-off characteristic
characteristic for
for Case-II.
Case-II.

6.3.
6.3. Case III: Unlimited
Unlimited Power
Power Exchange
Exchange between
between LV
LV and
and MV
MV
The
Theoverall
overallset
setofofDG
DGunits
unitsfunction as as
function perper
their power
their limits
power in the
limits thirdthird
in the scenario. In this
scenario. In
case, the utility behaves as an unconstraint unit whereas energy exchange occurs
this case, the utility behaves as an unconstraint unit whereas energy exchange occurs between utility
and microgrid irrespective of the limitations. The data required to overcome the multi-objective
between utility and microgrid irrespective of the limitations. The data required to over-
optimization issue such as real-time market prices, technical specifications of DGs, and load curve
come the multi-objective optimization issue such as real-time market prices, technical
specifications of DGs, and load curve remains unaltered. Like the earlier cases, evolu-
tionary-based optimization algorithms were utilized in this case as well to resolve oper-
ation management issues found in the microgrid. The simulation results were collected
from the experimentation.
Energies 2022, 15, 9024 18 of 24

remains unaltered. Like the earlier cases, evolutionary-based optimization algorithms were utilized
in this case as well to resolve operation management issues found in the microgrid. The simulation
results were collected from the experimentation.
Table 6 shows the simulation results for the third case, which was attained to mitigate the
operation cost and emission. As per the results, both operating costs, as well as the emission
rate of the microgrid, were reduced more than in than the basic scenario, under a few conditions
such as allowing unlimited power exchange and economic and environmental objectives. When
the second and third scenarios were compared for optimization, in terms of costs and emission,
the results infer that the total cost of operation was reduced in the third scenario, while the net
emission rate was reduced in the second scenario. Table 7 compares the statistical results of OGGWO
and other optimization algorithms, pertaining to every single objective. The comparative analysis
results infer that GGWO was deficient and its results were more inconsistent than for the proposed
OGGWO algorithm.
One of the significant observations in the third scenario is that during the first few hours of the
day, the utility occupied the top position in meeting load demands inside the grid. On the contrary,
the energy was purchased from the microgrid in bulk quantities at peak times. Regarding both
objectives, when the grid experienced a shortage in power generation, both WT and PV were started;
otherwise, both were triggered in the case of requirement for more energy to be exported to the
macrogrid. Other DGs such as battery and FC produced the maximum electricity for the majority of
the hours in a day under study, whereas MT generated the highest power between 9.00 to 17.00 h.
Figures 8 and 9 show the convergence characteristics of OGGWO, GGWO, GWO, CSA, DE,
and PSO for operating cost and emission. The proposed OGGWO achieved a fast and smooth
convergence characteristic curve towards reaching the optimal solution. The proposed algorithm
successfully overcame the optimization problem for the third time, whereas minor differences can
be found in optimal solutions for both the objective functions. Figure 10 represents the trade-off
characteristics between operating cost and emission for OGGWO and GGWO which revealed the
efficiency of OGGWO in elucidating multi-objective energy management problems. Further, the
comparative analysis conducted between the Pareto fronts of the proposed OGGWO algorithm
and GGWO algorithm showcases that the former afforded the least and was well-distributed than
the latter one. The simulation outcomes established the superiority of the proposed OGGWO in
managing equality and inequality issues that occur in energy management problems.

Table 6. A statistical comparison of simulation results for case-III.

Optimization
Parameter Cost Euro Emission Kg
Algorithm
Min Cost 127.3 NA
Min Emission NA 1506.9
EDNSGA-II [41]
BCS NA NA
Std Dev NA NA
Min Cost 133.12 1552.48
Min Emission 142.39 1512.83
PSO
BCS 137.03 1530.25
Std Dev 3.78 9.83
Min Cost 132.28 1549.28
Min Emission 140.59 1511.35
CSA
BCS 135.25 1528.94
Std Dev 3.59 9.64
Min Cost 126.74 1545.85
Min Emission 135.38 1505.37
GWO
BCS 132.06 1526.12
Std Dev 2.61 8.57
Min Cost 125.13 1556.29
Min Emission 140.56 1503.32
GGWO
BCS 130.27 1522.48
Std Dev 2.34 6.69
Min Cost 124.46 1556.73
Min Emission 139.59 1502.68
OGGWO
BCS 129.06 1520.83
Std Dev 2.18 6.52
Energies 2022, 15, 9024 19 of 24

Table 7. An optimal generation schedule for Case-III.

Energies 2022, 15, x FOR PEER REVIEW 22 of 28
Hour MT FC PV WT Battery Grid
1 6 0 0 4.97 −13.99 65.02
2 12.23 3.18 0 0.34 −10.99 55.24
37 00 3.25
0 0
0.27 12.99
10.98 −−25.12
29.23 92.78
77.98
48 06 00 00 5.65
8.28 −−17.23
13.99 69.34
92.95
5 0 0 0 11.98 −30 85.02
9 30 30 0.13 18.12 −15.29 28.04
6 0 3 0 13.21 −28.23 87.02
710 030 30
3.25 1.68
0 19.23
12.99 −29.76
25.12 −14.67
92.78
811 630 29.99
0 7.13
0 23
8.28 30
−17.23 −27.12
92.95
912 3030 30
30 0.13
9.89 18.12
22.23 −15.29
30 28.04
−34.12
10 30 30 1.68 19.23 29.76 −14.67
13 30 0 11.03 16.02 27.03 1.92
11 30 29.99 7.13 23 30 −27.12
1214 9.56
30 9.32
30 2.99
9.89 7.94
22.23 9.89
30 46.3
−34.12
1315 3030 30
0 8.07
11.03 24.29
16.02 18.99
27.03 −20.35
1.92
1416 9.56
11.23 9.32
11.28 2.99
1.86 7.94
8.75 9.89
8.76 46.3
53.12
1517 30
29.74 30
3.93 8.07
1.03 24.29
22.27 18.99
−30 −75.03
20.35
16 11.23 11.28 1.86 8.75 8.76 53.12
17
18 0
29.74
0
3.93
0
1.03
18.99
22.27 −−30
30
116.01
75.03
1819 6.93
0 15.24
0 00 12.96
18.99 −20.12
−30 92.99
116.01
1920 7.12
6.93 3
15.24 00 15.38
12.96 −−21.11
20.12 99.61
92.99
2021 7.12
30 3
30 00 15.38
20.03 −29.89
21.11 99.61
−16.92
21 30 30 0 20.03 29.89 −16.92
22 0 0 0 22.34 −29.99 92.65
22 0 0 0 22.34 −29.99 92.65
2323 00 3.23
3.23 00 8.03
8.03 12.02
12.02 54.72
54.72
2424 00 00 00 20.01
20.01 −−30
30 76.99
76.99

Figure8.8.The
Figure Thecost
costconvergence
convergencecharacteristic
characteristicfor
forCase-III.
Case-III.
EnergiesEnergies
2022, 15, 9024
2022, 15, x FOR PEER REVIEW 23 of 2820 of 24

Energies 2022, 15, x FOR PEER REVIEW 24 of 28

Figure 9. The emission convergence characteristic for Case-III.

Figure 9. The emission convergence characteristic for Case-III.

Figure 10.The
10.
Figure The emission cost
emission cost trade-off
trade-off characteristic
characteristic for Case-III.
for Case-III.

7. Conclusions
Environmental pollution is a significant threat to humanity and other living beings
on earth. Across the globe, there is a tremendous increase observed in environmental
Energies 2022, 15, 9024 21 of 24

7. Conclusions
Environmental pollution is a significant threat to humanity and other living beings on earth.
Across the globe, there is a tremendous increase observed in environmental pollution due to unsus-
tainable and environmentally-harmful human activities. The very existence of human beings on earth
is being questioned these days, whereas the measures taken to curb the environmental pollution will
only decide the fate of human beings. Traditional power plants, fueled by fossil fuels to produce
energy remain the primary source of environmental pollution. Different corrective measures are being
taken these days, such as the utilization of advanced technologies, the prevalent implementation
of renewable energy sources into the power sector, and the appropriate utilization of energy. These
initiatives may curb environmental pollution in the coming years. Microgrids with different RESs
such as PV, WT, battery, and so on enact an important role in curbing the environmental pollution
caused during conventional energy generation methods. In this background, a novel algorithm has
been proposed in this study, i.e., the Oppositional Gradient-based Grey Wolf Optimizer (OGGWO).
The proposed algorithm was validated under three different scenarios to establish its effectiveness. A
typical microgrid was used as a test system loaded with a battery, FC, WT, MT, and PV. The microgrid
was connected with the utility grid, as well as for power exchange as decided by MGCC. Further, the
researchers proposed a multi-objective formulation encompassing both emission and operational
costs under equality and inequality constraints. The application results were retrieved on a 24-h
horizon, whereas the results were compared with other existing techniques. The results establish that
the proposed technique outperformed other algorithms and produced excellent results. Further, the
results of the proposed algorithm were also compared against GGWO and different optimization
algorithms. Based on the comparison results, it can be inferred that the OGGWO algorithm is capable
of avoiding premature convergence. The proposed optimization algorithm can also be applied to
solve the optimal sizing and placement of FACTS devices in distribution systems.

Author Contributions: Conceptualization, A.R. and K.N.; Methodology, K.N. and S.D.; Software,
O.D.M., S.D. and P.P.; Validation, S.D. and I.A.K.; Formal analysis, A.R.; Investigation, S.D.; Resources,
A.S.P.; Data curation, A.R., K.N. and S.D.; Writing—original draft, A.R., S.D., N.L. and P.P.; Writing—
review & editing, A.R., S.D., N.L. and P.P.; Visualization, O.D.M., N.L. and P.P.; Supervision, O.D.M.,
S.D. and I.A.K.; Project administration, K.N., O.D.M. and A.S.P. All authors have read and agreed to
the published version of the manuscript.
Funding: This research received no external funding.
Data Availability Statement: The data presented in this study are available on request from the
corresponding author.
Conflicts of Interest: The authors declare no conflict of interest.

Abbreviations
ANN Artificial Neural Network
BARON Branch and Reduce Optimization Navigator
BBSA Backtracking search algorithm
BESS Battery Energy Storage System
BPSO Binary Particle Search Algorithm
CSA Cuckoo Search Algorithm
DER Distributed Energy Resources
DG Distributed Generator
DR Demand Response
DSM Demand Side Management
EDNSGA-II Economic Dispatch based Non-Dominated Sorting Genetic Algorithm
EO Equilibrium Optimizer
EV Electric Vehicle
FC Fuel Cell
GAMS General Algebraic Modeling System
GBDT Gradient Boosting Decision Tree
GGWO Gradient based Grey Wolf Optimizer
GOA Grasshopper Optimization Algorithm
GPPM Generalized Power Prediction Model
GRG Generalized reduced-gradient
Energies 2022, 15, 9024 22 of 24

GWO Grey Wolf Optimizer

LEMS Local Energy Management System
LF Levy Flight
LV Low Voltage
MBA Modified Bat Algorithm
MG Microgrid
MGCC Microgrid Central Controller
MGEM Microgrid Energy Management
MOPSO Multi-Objective Particle Search Algorithm
MSFOA Muddy Soil Fish Optimization Algorithm
MT Microturbine
MV Medium Voltage
NA Not Available
NSGA-II Non-Dominated Sorting Genetic Algorithm
OBL Opposition Based Learning
OGGWO Oppositional Gradient based Grey Wolf Optimizer
PAFC Phosphoric Acid Fuel Cell
PAR Peak-to-Average Ratio
PMORL Preference-based multi-objective reinforcement learning
PSO Particle Search Algorithm
PV Photovoltaic
QPSO Quantum Particle Search Algorithm
RES Renewable Energy Sources
SOA Sandpiper Optimization Algorithm
WT Wind Turbine

References
1. Nagarajan, K.; Rajagopalan, A.; Angalaeswari, S.; Natrayan, L.; Mammo, W.D. Combined Economic Emission Dispatch of
Microgrid with the Incorporation of Renewable Energy Sources Using Improved Mayfly Optimization Algorithm. Comput. Intell.
Neurosci. 2022, 15, 1–22. [CrossRef] [PubMed]
2. Karthik, N.; Parvathy, A.K.; Arul, R. A review of optimal operation of microgrids. Int. J. Electr. Comput. Eng. 2020, 10, 3. [CrossRef]
3. Konstantinopoulos, S.A.; Anastasiadis, A.G.; Vokas, G.A.; Kondylis, G.P.; Polyzakis, A. Optimal management of hydrogen storage
in stochastic smart microgrid operation. Int. J. Hydrogen Energy 2018, 43, 1. [CrossRef]
4. Aghajani, G.; Yousefi, N. Multi-objective optimal operation in a micro-grid considering economic and environmental goals. Evol.
Syst. 2019, 10, 239–248. [CrossRef]
5. Kamarposhti, M.A.; Colak, I.; Shokouhandeh, H.; Iwendi, C.; Padmanaban, S.; Band, S.S. Optimum operation management of
microgrids with cost and environment pollution reduction approach considering uncertainty using Multi-objective NSGAII
algorithm. IET Renew. Power Gener. 2022, 16, 1–13. [CrossRef]
6. Lv, T.; Ai, Q.; Zhao, Y. A bi-level multi-objective optimal operation of grid-connected microgrids. Electr. Power Syst. Res. 2016, 131,
60–70. [CrossRef]
7. Lu, X.; Zhou, K.; Yang, S. Multi-objective optimal dispatch of microgrid containing electric vehicles. J. Clean. Prod. 2017, 165,
1572–1581. [CrossRef]
8. Kim, H.J.; Kim, M.K.; Lee, J.W. A two-stage stochastic p-robust optimal energy trading management in microgrid operation
considering uncertainty with hybrid demand response. Int. J. Electr. Power Energy Syst. 2021, 124, 106422. [CrossRef]
9. Arumugam, P.; Kuppan, V. A GBDT-SOA approach for the system modelling of optimal energy management in grid connected
micro-grid system. Int. J. Energy Res. 2020, 45, 6765–6783. [CrossRef]
10. Gad, Y.; Diab, H.; Abdelsalam, M.; Galal, Y. Smart Energy Management System of Environmentally Friendly Microgrid Based on
Grasshopper Optimization Technique. Energies 2020, 13, 5000. [CrossRef]
11. Veluchamy, K.; Veluchamy, M. A new energy management technique for microgrid system using muddy soil fish optimization
algorithm. Int. J. Energy Res. 2021, 45, 14824–14844. [CrossRef]
12. Jain, D.K.; Tyagi, S.K.S.; Neelakandan, S.; Prakash, M.; Natrayan, L. Metaheuristic optimization-based resource allocation
technique for cybertwin-driven 6G on IoE environment. IEEE Trans. Ind. Inform. 2021, 18, 4884–4892. [CrossRef]
13. Aghajani, G.; Ghadimi, N. Multi-objective energy management in a micro-grid. Energy Rep. 2018, 4, 218–225. [CrossRef]
14. Ahmed, D.; Ebeed, M.; Ali, A.; Alghamdi, A.S.; Kamel, S. Multi-Objective Energy Management of a Micro-Grid Considering
Stochastic Nature of Load and Renewable Energy Resources. Electronics 2021, 10, 403. [CrossRef]
15. Mansouri, S.A.; Ahmarinejad, A.; Nematbakhsh, E.; Javadi, M.S.; Jordehi, A.R.; Catalao, J.P. Energy management in microgrids
including smart homes: A multi-objective approach. Sustain. Cities Soc. 2021, 69, 102852. [CrossRef]
16. Tabar, V.S.; Jirdehi, M.A.; Hemmati, R. Energy management in microgrid based on the multi objective stochastic programming
incorporating portable renewable energy resource as demand response option. Energy 2017, 118, 827–839. [CrossRef]
Energies 2022, 15, 9024 23 of 24

17. Mandal, S.; Mandal, K.K. Optimal energy management of microgrids under environmental constraints using chaos enhanced
differential evolution. Renew. Energy Focus 2020, 34, 129–141. [CrossRef]
18. GM Abdolrasol, M.; Hannan, M.A.; Hussain, S.M.S.; Ustun, T.S.; Sarker, M.R.; Ker, P.J. Energy Management Scheduling for
Microgrids in the Virtual Power Plant System Using Artificial Neural Networks. Energies 2021, 14, 6507. [CrossRef]
19. Majumder, I.; Dash, P.K.; Dhar, S. Real-time Energy Management for PV–battery–wind based microgrid using on-line sequential
Kernel Based Robust Random Vector Functional Link Network. Appl. Soft Comput. 2021, 101, 107059. [CrossRef]
20. Karthik, N.; Parvathy, A.K.; Arul, R.; Jayapragash, R.; Narayanan, S. Economic load dispatch in a microgrid using Interior Search
Algorithm. In 2019 Innovations in Power and Advanced Computing Technologies (i-PACT); IEEE: Piscataway, NJ, USA, 2019; pp. 1–6.
21. Esapour, K.; Abbasian, M.; Saghafi, H. Intelligent energy management in hybrid microgrids considering tidal, wind, solar and
battery. Int. J. Electr. Power Energy Syst. 2021, 127, 106615. [CrossRef]
22. Aghajani, G.R.; Shayanfar, H.A.; Shayeghi, H. Demand side management in a smart micro-grid in the presence of renewable
generation and demand response. Energy 2017, 126, 622–637. [CrossRef]
23. Jasim, A.M.; Jasim, B.H.; Kraiem, H.; Flah, A. A Multi-Objective Demand/Generation Scheduling Model-Based Microgrid Energy
Management System. Sustainability 2022, 14, 10158. [CrossRef]
24. Yin, N.; Abbassi, R.; Jerbi, H.; Rezvani, A.; Müller, M. A day-ahead joint energy management and battery sizing framework based
on θ-modified krill herd algorithm for a renewable energy-integrated microgrid. J. Clean. Prod. 2021, 282, 124435. [CrossRef]
25. Han, Y.; Yang, H.; Li, Q.; Chen, W.; Zare, F.; Guerrero, J.M. Mode-triggered droop method for the decentralized energy
management of an islanded hybrid PV/hydrogen/battery DC microgrid. Energy 2020, 199, 117441. [CrossRef]
26. Han, Y.; Zhang, G.; Li, Q.; You, Z.; Chen, W.; Liu, H. Hierarchical energy management for PV/hydrogen/battery island DC
microgrid. Int. J. Hydrogen Energy 2019, 44, 11. [CrossRef]
27. Kafetzis, A.; Ziogou, C.; Panopoulos, K.D.; Papadopoulou, S.; Seferlis, P.; Voutetakis, S. Energy management strategies based
on hybrid automata for islanded microgrids with renewable sources, batteries and hydrogen. Renew. Sustain. Energy Rev. 2020,
134, 110118. [CrossRef]
28. Xu, J.; Li, K.; Abusara, M. Preference based multi-objective reinforcement learning for multi-microgrid system optimization
problem in smart grid. Memetic Comp. 2022, 14, 225–235. [CrossRef]
29. Jordehi, A.R.; Javadi, M.S.; Catalão, J.P. Energy management in microgrids with battery swap stations and var compensators.
J. Clean. Prod. 2020, 272, 122943. [CrossRef]
30. Sedighizadeh, M.; Esmaili, M.; Mohammadkhani, N. Stochastic multi-objective energy management in residential microgrids
with combined cooling, heating, and power units considering battery energy storage systems and plug-in hybrid electric vehicles.
J. Clean. Prod. 2018, 195, 301–317. [CrossRef]
31. Yaghi, M.; Luo, F.; El Fouany, H.; Junfeng, L.; Jiajian, H.; Jun, Z. Multi-Objective optimization for Microgrid Considering
Demand Side Management. In Proceedings of the 2019 Chinese Control Conference (CCC), Guangzhou, China, 27–30 July 2019;
pp. 7398–7403. [CrossRef]
32. Luo, L.; Abdulkareem, S.S.; Rezvani, A.; Miveh, M.R.; Samad, S.; Aljojo, N.; Pazhoohesh, M. Optimal scheduling of a renewable
based microgrid considering phototaic system and battery energy storage under uncertainty. J. Energy Storage 2020, 28, 101306.
[CrossRef]
33. Farinis, G.K.; Kanellos, F.D. Integrated energy management system for Microgrids of building prosumers. Electr. Power Syst. Res.
2021, 198, 107357. [CrossRef]
34. Kakran, S.; Chanana, S. Operation management of a renewable microgrid supplying to a residential community under the effect
of incentive-based demand response program. Int. J. Energy Environ. Eng. 2019, 10, 121–135. [CrossRef]
35. Mosa, M.A.; Ali, A.A. Energy management system of low tage dc microgrid using mixed-integer nonlinear programing and a
global optimization technique. Electr. Power Syst. Res. 2021, 192, 106971. [CrossRef]
36. Hasankhani, A.; Hakimi, S.M. Stochastic energy management of smart microgrid with intermittent renewable energy resources in
electricity market. Energy 2021, 219, 119668. [CrossRef]
37. De, M.; Das, G.; Mandal, K.K. An effective energy flow management in grid-connected solar–wind-microgrid system incorporating
economic and environmental generation scheduling using a meta-dynamic approach-based multiobjective flower pollination
algorithm. Energy Rep. 2021, 7, 2711–2726. [CrossRef]
38. Dey, B.; Bhattacharyya, B.; Márquez, F.P.G. A hybrid optimization-based approach to solve environment constrained economic
dispatch problem on microgrid system. J. Clean. Prod. 2021, 307, 127196. [CrossRef]
39. Dey, B.; Bhattacharyya, B.; Srivastava, A.; Shivam, K. Solving energy management of renewable integrated microgrid systems
using crow search algorithm. Soft Comput. 2020, 24, 10433–10454. [CrossRef]
40. Paliwal, N.K.; Singh, A.K.; Singh, N.K. Energy scheduling optimisation of an islanded microgrid via artificial bee colony guided
by global best, personal best and asynchronous scaling factors. Int. J. Sustain. Energy 2020, 39, 6. [CrossRef]
41. Sarshar, J.; Moosapour, S.S.; Joorabian, M. Multi-objective energy management of a micro-grid considering uncertainty in wind
power forecasting. Energy 2017, 139, 680–693. [CrossRef]
42. Mirjalili, S.; Mirjalili, S.M.; Lewis, A. Grey Wolf Optimizer. Adv. Eng. Softw. 2014, 69, 46–61. [CrossRef]
43. Abdel-Basset, M.; El-Shahat, D.; El-henawy, I.; de Albuquerque, V.H.C.; Mirjalili, S. A new fusion of grey wolf optimizer algorithm
with a two-phase mutation for feature selection. Expert Syst. Appl. 2020, 139, 112824. [CrossRef]
Energies 2022, 15, 9024 24 of 24

44. Dhargupta, S.; Ghosh, M.; Mirjalili, S.; Sarkar, R. Selective opposition based grey wolf optimization. Expert Syst. Appl. 2020,
151, 113389. [CrossRef]
45. Pradhan, M.; Roy, P.K.; Pal, T. Grey wolf optimization applied to economic load dispatch problems. Int. J. Electr. Power Energy
Syst. 2016, 83, 325–334. [CrossRef]
46. Khalilpourazari, S.; Doulabi, H.H.; Çiftçioğlu, A.Ö.; Weber, G.W. Gradient-based grey wolf optimizer with Gaussian walk:
Application in modelling and prediction of the COVID-19 pandemic. Expert Syst. Appl. 2021, 177, 114920. [CrossRef] [PubMed]
47. Pahnehkolaei, S.M.A.; Alfi, A.; Sadollah, A.; Kim, J.H. Gradient-based water cycle algorithm with evaporation rate applied to
chaos suppression. Appl. Soft Comput. 2017, 53, 420–440. [CrossRef]
48. Karthik, N.; Parvathy, A.K.; Arul, R.; Padmanathan, K. Multi-objective optimal power flow using a new heuristic optimization
algorithm with the incorporation of renewable energy sources. Int. J. Energy Environ. Eng. 2021, 12, 641–678. [CrossRef]