0% found this document useful (0 votes)
16 views22 pages

Optimal Power Scheduling of Microgrid Considering Renewable Sources

Uploaded by

Amit Kumar
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
16 views22 pages

Optimal Power Scheduling of Microgrid Considering Renewable Sources

Uploaded by

Amit Kumar
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 22

Cluster Computing

https://doi.org/10.1007/s10586-024-04669-z (0123456789().,-volV)(0123456789().,-volV)

Optimal power scheduling of microgrid considering renewable sources


and demand response management
Abhishek Srivastava1 • Dushmanta Kumar Das1 • Siseyiekuo Khatsu1

Received: 7 May 2024 / Revised: 25 June 2024 / Accepted: 6 July 2024


Ó The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2024

Abstract
Generation planning in the power system has been a complex and challenging multi-objective optimization problem.
Numerous methodologies have been developed and applied to solve this problem. Still, researchers look forward to further
improved methods that can further enhance the outcomes for this problem. This article suggests a hybrid FPA-PPSO
scheme that combines the flower pollination algorithm (FPA) and phasor particle swarm optimization (PPSO) algorithm to
solve this problem. The suggested FPA-PPSO scheme includes a high search efficiency, balanced local and global search
capabilities, etc. which can help in more efficient planning of the generation in a microgrid. To analyze the supremacy of
the suggested FPA-PPSO scheme over FPA and PPSO methods, a comparative study using twenty-three benchmark
functions is presented. From the analysis conducted from the Friedman test, the suggested method gets a rank of 1.59 which
is better compared to PPSO and FPA which get ranks of 2.61 and 3.04 for the twenty-three benchmark functions. Further,
the suggested method is tested using four test cases to show its effectiveness in solving the scheduling problem of a
microgrid considering renewable energy sources, energy storage systems, and demand response management. The obtained
results conclude that the suggested methods show promising performance in planning the generating unit. Using the DRM
scheme further helps to efficiently schedule the generation as the utilities save up to $ 25545 more when DRM planning is
considered.

Keywords Dynamic economic emission dispatch problem (DEED)  Flower pollination algorithm (FPA) 
Phasor particle swarm optimization (PPSO)  Renewable energy sources (RES)  Demand response management (DRM)

1 Introduction environmental effects can be achieved by the use of


renewable energy sources (RESs) [2]. The power generated
1.1 Motivation and purpose by these sources has an advantage in remote places where
the supply of grid power is a difficult task. However,
The rapid growth of the economy in this modern world of microgrids with small-scale power generation and distri-
highly capitalized society led to environmental concerns bution have various challenges in the efficient power sys-
for which renewable energy resources are a hot issue in the tem operation [3]. One of the major challenges which is
power system [1]. The generation of energy with more encountered is the unpredictable nature of power generated
reliability, flexibility, less cost of generation, and fewer by the RESs. RESs such as solar, wind, etc. are weather
dependent and thus the power delivered from them is
unpredictable [4]. Therefore, to match the demand and
& Dushmanta Kumar Das
[email protected] supply, it becomes an important task to schedule the other
generating sources optimally.
Abhishek Srivastava
[email protected] The scheduling problem in the power system is well-
known among researchers and is commonly known as the
Siseyiekuo Khatsu
[email protected] economic dispatch (ED) problem. The objective of the ED
problem is to minimize the cost of power generation and
1
Department of Electrical and Electronics Engineering, emission rates along with the fulfillment of power demand
National Institute of Technology Nagaland, Dimapur, India

123
Cluster Computing

considering various constraints of the system [5]. Owing to flower pollination algorithm for solving ED problems with
the objectives and constraints encountered in solving, the a time-varying fuzzy selection mechanism is presented. In
ED problem is considered to as a complex, multi-objective [26], a dynamic emission economic dispatch (DEED)
optimization problem [6–9]. Traditional optimization problem is solved for an integrated microgrid. In [27], a
methods such as Newton method [10], gradient method modified harmony search algorithm is used for the com-
[11], lambda-iteration method [10], etc. fail to provide an bined economic emission dispatch (CEED) problem of
efficient and reliable solution for such complex optimiza- microgrids. In [28], the semi-definite programming
tion problems. Therefore, researchers have looked at meta- approach is used to solve the ED problems for energy
heuristic optimization methods as an alternative method to storage systems in microgrids. In [29], an improved
solve ED problems. quantum particle swarm optimization is used to solve the
Meta-heuristic optimization methods are iterative ED problem for a microgrid. In [30], a coronavirus herd
methods that use random search techniques to obtain a immunity algorithm is developed to solve the DEED
solution for an optimization problem by minimizing or problem in the presence of RESs for a microgrid. In [4], a
maximizing a given objective function. Although these bottlenose dolphin optimization (BDO) algorithm is
optimization methods have shown supremacy over the developed to solve DELD and DEED problems in the
traditional optimization scheme, some serious issues are presence of RESs for a microgrid. In [31], authors devel-
still present. The majority of these approaches have one oped a search and rescue optimization algorithm to solve
flaw: they are sensitive to user-defined parameters. Another ELD and CEED problems. In [32], the authors investigate
disadvantage is that meta-heuristic algorithms may not different off-grid RES options available and determine the
always achieve the global optimum, instead converging to most economical one the most economical for remote
a local optimum. Finally, these methods have a high pro- locations.
cessing complexity, which increases the time it takes to In the process of designing an efficient optimization
solve an optimization problem. [12]. To address these scheme, hybrid optimization methods have gained a lot of
drawbacks of meta-heuristic optimization methods, attention among researchers. Hybrid optimization methods
researchers aim to develop new and efficient optimization are formulated by combining two or more optimization
schemes to solve optimization problems for further methods with the specific goals to improve the exploration
improved results. Some of the recently developed methods and exploitation capabilities [33]. This makes them more
developed and applied to solve different optimization efficient and stable in providing a global optimal result
problems include single candidate optimizer [13], player compared to traditional optimization methods. In the past
unknown’s battlegrounds ranking-based optimization [14], few years, many hybrid optimization techniques have been
artificial hummingbird algorithm (AHA) [15], prairie dog investigated and presented in different literature. Some of
optimization algorithm [16], etc. In the case of the ED these hybrid optimizations include a hybrid genetic algo-
problem, the recently developed methodologies are dis- rithm (GA) and particle swarm optimization (PSO), GA-
cussed in the following section. PSO [34], a hybrid differential evolution (DE), chaos
sequences and sequential quadratic programming (SQP),
1.2 Literature review DEC-SQP [35], a hybrid PSO and bacterial foraging
algorithm (BFA), PSO-BFA [36], a hybrid bat algorithm
Many researchers have done tremendous work in the field (BA) and firefly algorithm (FA), BA-FA [37], etc. In the
of generation planning. In [17], the authors suggested the field of generation planning, in [38], a genetic algorithm
use of the leader white shark optimization algorithm to and mixed-integer linear programming are used to optimize
solve the economic load dispatch (ELD) problem. In [18] the unit commitment and ED problem in microgrids. In
and [19], the authors investigated the performance of [39], a multi-objective spotted hyena and emperor penguin
enhanced social network search and two-archive Harris optimizer is used for solving basic ED problems and
hawk optimization respectively to solve the ELD problem. microgrid power dispatch problems. In [40], the authors
The authors in [20] and [21], used many-objective marine introduced an aggrandized class topper optimization to
predators algorithm and multi-agent system respectively to solve the ELD and CEED problems. In [41], the authors
solve multi-objective ED problem. In [22], a reduced gra- investigated the performance of quantum class topper
dient method is used to solve the ED problem for a optimization to solve the CEED problem. In [3], the
microgrid considering RESs. In [23], a genetic algorithm is authors investigated the planning of an interconnected
used to solve the ED problem for an integrated microgrid. microgrid using an interactive class topper optimization.
In [24], the dynamic model of the microgrid having Optimization of resources in a microgrid is also possible
uncertain factors such as power fluctuations of renewable by optimizing the load profile of a microgrid [3]. For this
energy and load fluctuations is solved. In [25], a hybrid purpose, demand response management (DRM) and

123
Cluster Computing

demand-side management (DSM) schemes have been PPSO scheme is investigated with four test cases for the
introduced in microgrids for optimizing the load demand scheduling problem. The results obtained from the inves-
profile. Generally, the load profile optimization is per- tigation confirm the supremacy of the suggested FPA-
formed using different approaches which are load shed- PPSO scheme in exploring and exploiting the search space
ding, peak clipping, strategic load conservation, strategic for an optimal solution.
load building, valley filling, flexible load shape, etc. [42]. To conclude the main objectives and contribution of this
DSM and DRM schemes not only benefit the customers but study:
also benefit the utilities by reducing utility bills, generation
– A hybrid FPA-PPSO method is developed to overcome
costs, emissions, peak loads, etc. Many works have been
the drawbacks of the FPA optimization scheme.
presented to incorporate these schemes into microgrids. In
– For the validation of the supremacy of the suggested
[42], a heuristic-based evolutionary algorithm has been
FPA-PPSO scheme with respect to existing FPA, and
used to optimize the load profile for a microgrid. In [26],
PPSO schemes, a comparative analysis is presented
authors solved a DEED problem considering wind power
using twenty-three benchmark functions.
uncertainties, and energy storage systems. In [43], the
– The suggested optimization method is applied to the
authors used a robust optimization approach with a bidding
scheduling problem in a microgrid along with the
strategy to optimize the cost of a microgrid along with the
integration of renewable energy sources to show its
optimization of the load profile using demand response
supremacy in solving complex optimization problems.
programming. In [44], a genetic algorithm is used to solve
– The impact of demand response management on
DSM and ED problems for a microgrid. In [45], particle
generation planning when renewable energy sources
swarm optimization is used to optimize the cost for DSM in
are considered and when eliminated is studied.
a smart grid. In [46], an artificial fish swarm algorithm is
used for scheduling generation and storage in a microgrid The rest of the article is arranged as follows. Section 2
considering DSM. In [47], an optimal generation schedul- presents a discussion of the problem formulation and
ing for a hydrothermal system with DSM is presented. In mathematical modeling of the optimization problem. Sec-
[48], authors solve a day-ahead scheduling ED problem tion 3 presents a discussion on the motivation, mathemat-
considering combined heat and power units and uncertainty ical modeling, and validation of the suggested FPA-PPSO
in the load demand response using co-evolutionary PSO. scheme. Section 4 shows the simulation results and dis-
The authors in [49] investigated the performance of the cussions. Finally, section 5 presents the conclusion and
modified differential evolution (DE) algorithm by works that can be taken up in the near future.
scheduling CHP units considering RESs and electric
vehicles.
2 Problem formulation
1.3 Contribution and paper organization
2.1 Overview of isolated microgrid
With the advancement in the field of renewable generation
and rising concern towards environmental problems, opti- A dynamic ED problem for an isolated microgrid is for-
mal planning of the generating units has gained interest mulated using thermal generators, diesel generators, wind
among researchers in recent years. Although this problem generation, solar generation, and energy storage devices. A
is not new and various optimization schemes have already layout of this system is shown in the Fig. 1. The model of
been applied to obtain efficient results, the researcher each generating unit is briefly described in the following
investigates new methods for better and improved results. section.
In this regard, this study investigates a hybrid optimization
method developed by combining the existing flower pol-
lination scheme (FPA) and phasor particle swarm opti- Unidirectional power flow
Bidirectional power flow
Data Communication
DG (Solar)

mization (PPSO). The developed hybrid FPA-PPSO Load


Bus

method solves the problem of low search efficiency faced PCC Load
Bus

by the FPA scheme. The integration of the PPSO Load


DG (Diesel)

Load Load

scheme helps to improve the efficiency of the optimizer to Load


Microgrid

solve large variable optimization problems. For the vali- Load

DG (Diesel) DG (Biomass)

dation of supremacy with respect to FPA and PPSO, the Generation


PCC Load
Load

suggested FPA-PPSO scheme is tested with twenty-three DG (Wind)


MCC Microgrid

benchmark functions for its exploration and exploitation


capabilities. Next, the performance of the suggested FPA- Fig. 1 Layout of the microgrid system

123
Cluster Computing

2.1.1 Thermal generator 24 X


X n
ED ¼ ðaD þ bD PD;t þ cD P2D;t Þ; ð5Þ
For typical thermal generators, the fuel cost and emissions t¼1 D¼1

associated with generating units are mathematically where ED is the total emission, aD , bD , cD are emission
described as: coefficients of the Dth generating unit.
24 X
X n A multi-objective DEED issue, similar to that of con-
CT ¼ ðaT þ bT PT;t þ cT P2T;t Þ; ð1Þ ventional generators, is constructed by taking total fuel cost
t¼1 T¼1
and emissions into account. By taking weights into
24 X
X n account, this problem is reduced to a single aim. This
ET ¼ ðaT þ bT PT;t þ cT P2T;t Þ; ð2Þ single objective problem can be phrased as follows:
t¼1 T¼1
FD ¼ x3  CD þ x4  ED ð6Þ
where CT denotes the generation cost, aT , bT , cT shows the
fuel coefficients of the T th generating unit, and PT;t shows where FD is the total cost to be minimized, x3 and x4 are
the generated power by the T th generating unit at tth time the weights. Here, the weight used works in the similar way
interval, ET denotes the total emission, aT , bT , cT shows the as the weights used for (3). For equal contribution of both
emission coefficients of the T th generating unit. the objectives, x3 is set to 0.5, then x4 also becomes 0.5.
Taking total fuel cost and emissions into account for-
mulates a multi-objective DEED problem. By factoring in 2.1.3 Wind-power generation
weights, this multi-objective DEED problem is reduced to
a single objective. This single objective problem can be The total cost associated with the power generation from
phrased as follows: the wind farms is represented as follows:
!
FT ¼ x1  CT þ x2  ET ; ð3Þ X24
r
t
Fw ¼ Pw   Ip þ G e ; ð7Þ
t¼1 1  ð1 þ rÞN
where FT denotes the objective function, x1 and x2 are the
weights. Here, the weight x1 can take values from [0, 1], where Fw is the total cost, Ptw is the power output from the
whereas, x2 = 1 - x1 . The two weight used helps to reg- wind farm at tth time instant, Ip is the ratio of investment
ulate the contribution of the respective objective, CT and cost to unit established power, Ge is the operational and
ET , in the final objective function (FT ). If x1 is set to 1, maintenance cost, r is the interest scale and N is the
then x2 becomes 0 and the objective function (FT ) will investment duration.
optimize the generating units such that only the fuel cost is By considering the parameters Ip , Ge , r and N as
minimized. For equal contribution of both the objectives in 1400$=kw, 0.016$=kw, 0.09 and 20, (7) can be written as
the objective function, generally x1 is set to 0.5, which [51]:
makes x2 also 0.5 [50]. With this consideration the gen-
X
24
erating units are optimized such that a tradeoff is main- Fw ¼ ð153:381  Ptw Þ: ð8Þ
tained while both fuel cost and emission are minimized. t¼1

2.1.2 Diesel generator The power Ptw obtained from a wind farm considering
uncertainty is computed as follows [52]:
8
Diesel generators are widely used at airports, power sta- >
> 0; if WSt \Sci ;
>
>  
tions, and manufacturing plants. They are suitable for long > t
< P  WS  Sci ; if S \WSt \S ;
operations and generally work with less fuel consumption t n ci cr
Pw ¼ Sr  Sci ð9Þ
and emission rates. The total fuel cost associated with the >
>
>
> P n ; if S cr \WS t
\S co ;
diesel generators while generating PD;t units of power is >
:
0; if WSt [ WSco ;
mathematically represented as follows:
24 X
X n where Pn is the nominal power of wind turbine; WSt is the
CD ¼ ðaD þ bD PD;t þ cD P2D;t Þ; ð4Þ speed of wind at tth instant; Sci denotes the cut-in speed, Scr
t¼1 D¼1 shows the rated speed and Sco shows the cut-out speed.
where CD denotes cost of generation, aD , bD , cD shows the
cost coefficients of the Dth diesel unit. 2.1.4 Solar-power generation
Further, the total emission associated with the diesel
generators while generating PD;t units of power is mathe- The total cost associated with a solar farms generating Ps
matically represented as follows: units of power is represented as follows:

123
Cluster Computing

!
r where Fb represents the day-ahead cost, r, Q, FC , MC and
Fs ¼ Pts  N  Ip þ Ge ; ð10Þ
1  ð1 þ rÞ CE represents the interest rate, lifetime, fixed cost, main-
tenance cost and size of storage device.
where Fs is the total cost, Pts is the power output from the
solar farm at tth time instant, 2.2 Objective function
By considering the ratio for investment cost as 5000
$=kw, the operational and maintenance cost as 0.016 $=kw, Using the diesel, solar, wind and ESS, the objective func-
interest scale 0.09 and investment duration as 20 years as tion of the microgrid is formulated as follows:
presented in [51], (10) can be written as:
minðCf Þ ¼ minðFT þ FD þ Fs þ Fw þ Fb Þ; ð16Þ
X
24
Fs ¼ ð547:7483  Pts Þ: ð11Þ where Cf is the objective which has to be minimized.
t¼1

The power Pts obtained from a solar farm considering 2.3 Constraints
uncertainty is computed as follows [53]:
   While optimizing the generating units, it is important that
t StR m t t NT  20 all the generation and power constraints are satisfied. The
Ps ¼ 0  P s þ l  T þ S R   Ts ;
SR 800 constraints considered in this study are discussed in this
ð12Þ section.
where the total irradiance is represented by StR ; irradiance
2.3.1 Load balance
value under standard conditions is represented by S0R ; rated
power under standard conditions is shown as Pm s ; tempera- It is important that the total power demand is equal to the
ture and module temperature is shown as T t and Ts respec- total power generation. This is mathematically expressed as
tively; NT and l represents the normal cell operating follows:
temperature and temperature power coefficient respectively.
X
24 X
24 X
24 X
24 X
24 X
24
PtT þ PtD þ Ptw þ Pts þ Ptb ¼ Ptdem ;
2.1.5 Energy storage system (ESS) t¼1 t¼1 t¼1 t¼1 t¼1 t¼1
ð17Þ
ESS improves the reliability of the power system network
by dealing with the unpredictability generated by the where Ptdem is the total demand at tth time instant.
inclusion of renewable energy sources. ESS units, includ-
ing NaS, Li-Ion, and Lead-Acid batteries, serve as backup 2.3.2 Generator limits
power sources. The ESS state of charge (SoC) and state of
discharge (SoD) are determined by the power of diesel, It is necessary that all the generating units are operated
RES, and load demand. If PD þ Ps þ Pw [ Pden , the ESS is within prescribed operating limits. This is mathematically
charging, as described mathematically below: represented as follows:
Ptb ¼ Pt1 t t
b þ ðPG  PL Þgb ; ð13Þ Pmin max
T  P T  PT ; ð18Þ

where Ptb is the energy available at tth , Pt1 is the energy Pmin max
w  P w  Pw ; ð19Þ
b
available at ðt  1Þth time instant, PtG is the generated Pmin max
ð20Þ
D  PD  PD ;
energy by diesel and RES together at tth , PtL is load demand
at tth and gb is the efficiency of energy conversion. Pmin max
s  P s  Ps ; ð21Þ
Whereas, if PD þ Ps þ Pw \Pden , the ESS is in the state Pmin max
ð22Þ
b  Pb  Pb ;
of discharging which is mathematically expressed as
follows: where Pmin min min min
T , Pw , PD , Ps , and Pb
min
represents the min-
imum generation limits for thermal generators, wind farms,
Ptb ¼ Pt1 t t
b  ðPG  PL Þgb ; ð14Þ
diesel generators, solar farms, and ESS repeatedly.
The cost associated with the ESS is computed as [54]: Whereas, Pmax max max max
T , Pw , PD , Ps , and Pb
max
are the maxi-
(" # ) mum generation limits.
1 r ð 1 þ RÞ Q
Fb ¼  FC  CE þ ðCE  MC Þ ;
365 ð1 þ RÞQ 1
ð15Þ

123
Cluster Computing

2.4 Demand response program pollinators. The mathematical model for local pollination
as per [55], is represented as follows:
In a microgrid, load profile management helps the utilities  
Ptþ1
i ¼ P t
i þ  t
P t
j  P t
k ; ð28Þ
to maximize their profits. Therefore, to optimize the load
profile, in this article, a demand response program (DRP)
where Ptþ1
i is the updated pollen or solution vector at ðt þ
scheme is considered. DRP scheme is generally based on th
1Þ instant Pti is the pollen or solution vector at tth instant,
energy pricing where prices are varied according to the
load demand. Furthermore, to improve the load profile, Ptj and Ptk are pollen from flowers j and k at tth instant and t
shiftable loads are moved from high-price peaks to low- is a constant value between 0 and 1 at tth instant.
price off-peak hours. This helps to reduce the cost of In the case of global pollination which is generally
operating the microgrid. The mathematical model for the referred to as a biotic or cross-pollination process, a pollen
DRM technique, provided in [43], is as follows: carrier is required. The mathematical model for global
pollination as per [55], is represented as follows:
DtL ¼ ð1  DtR Þ  LtB þ Lts ; ð23Þ  
Ptþ1 ¼ P t
þ L f P t
 P t
i ; ð29Þ
where t time period, DtL , DtR , LtB , and Lts represents the load i i g

demand considering DRP, base load percentage which where Ptg is the best solution found among all solutions at
participates in DRP, base load without considering DRP
tth instant and Lf is the strength of pollination, which is
and shiftable load respectively.
essentially a step size and is computed using a Levy dis-
tribution function which is given as:
2.4.1 Constraints
 
kCðkÞsin Pk 1
Lf ¼ 2
; ðs [ 0Þ; ð30Þ
As in a DRM approach, the loads are only shifted from the P s1þk
peak periods to off peak periods, the total load demand
where CðkÞ represents a standard gamma function and
remains unchanged. Mathematically, this can be expressed
s represents the size of the steps.
as follows:
Pollens can pollinate using either of the methods. To
X
T X
T
switch between the local pollination and global pollination,
Lts ¼ DtR  LtB ; ð24Þ
t¼1 t¼1
a switch probability (SP ) is used. If SrP is less than SP , then
agents are updated using the local pollination rule.
Further, it is important that the increment and decrement of Whereas, if SrP is greater than SP , then agents are updated
the load should follow some criterion values which are using the global pollination rule. SrP is a switch probability
given as follows: function which generally represents a random number
Ltinc  inct  LtB ; ð25Þ between 0 and 1.

DtR  Dmax
R ; ð26Þ 3.2 Overview of PPSO technique
t max
inc  inc ; ð27Þ
Particle Swarm Optimization (PSO), presented in [56], is a
where Ltinc is the incremented load, inct represents the popular and most commonly used optimization method.
incremented load percentage, Dmax
R represents the maxi- PSO has been used to solve a variety of optimization
mum loads participates in DRP and incmax shows the problems related to different fields. In the past few decades,
maximum load that will be incremented. many modified and improved versions of PSO have been
developed which are claimed to have a better performance
than the basic PSO technique. One such method is a Phasor
3 Hybrid FPA-PPSO technique Particle Swarm Optimization (PPSO) presented in [57].
The PPSO method uses a diverse strategy which not only
3.1 Overview of FPA technique helps in improving the convergence ability but also
increases the optimization efficiency. In PPSO, each par-
The flower pollination algorithm (FPA), presented in [55], ticle (search agent) is equipped with a scalar phasor angle
is a technique that is inspired by the pollination process of (h) and a magnitude (X) which are together presented as
flowering plants. In FPA, pollens (search agents) are pol- X\h. In order to update the particles, both the magnitude
linated (updated) using either local pollination or global and phasor angles of each particle are updated in each
pollination. Local pollination is generally referred to as iteration. This is mathematically represented as follows:
abiotic or self-pollination which does not require any

123
Cluster Computing

V t ¼pðht Þ  ðPb  X t Þ þ gðht Þ  ðGb  X t Þ; ð31Þ is used for all control variables. This makes the PPSO
scheme a self-adaptive nonparametric meta-heuristic opti-
Ptþ1 ¼Pt þ V t ; ð32Þ mization scheme which is quite simple in implementation.
Further, in case the problem dimensions are increased, the
where the global and position best particle is represented
optimization efficiency of the PPSO scheme is better
by Gb and Pb respectively; the magnitude of the present
compared to other optimization algorithms [59].
and updated particle position is presented by Pt and Ptþ1 In order to implement the suggested technique, the two
respectively; To evaluate pðht Þ and gðht Þ following math- optimization methods, FPA and PPSO, are cascaded with
ematical models are used. each other. In the suggested method, the suggested algo-
2sinðht Þ rithm is split into two stages. In the first stage, the search
pðht Þ ¼jcosðht Þj ; ð33Þ
agents follow the updating rule of local and global polli-
2cosðht Þ
gðht Þ ¼jsinðht Þj ; ð34Þ nation from the FP algorithm. After obtaining the updated
solutions from the first stage, i.e., the updated agents are
The phasor angles for each particle, in every iteration are used in the second stage. In the second stage, the updated
updated as follow: agents follow the position and phase update rule from the
htþ1 ¼ ht þ jcosðht Þ þ sinðht Þj  2  p; ð35Þ PPSO technique. Further, the agents are tested with the
boundary conditions. Then they are feedback for the first
stage. This process is repeated until the termination con-
3.3 Proposed hybrid FPA-PPSO algorithm dition is satisfied. Further to explain the workflow, Figs. 2
and 3 presents the pseudo-code and flow chart for the
As discussed in the previous section 1, for any optimization hybridization process of the FPA and PPSO algorithms
scheme to efficiently explore the search space for an respectively.
optimal solution, it is important that the optimizer is
equipped with good exploration and exploitation capabili- 3.3.1 Time complexity analysis
ties. In the case of the FPA optimization scheme, the
optimizer explores the search space when global pollina- The time complexity of the suggested FPA-PPSO algo-
tion is performed. Whereas, in the case of local or self- rithm depends on the population size of search agents, the
pollination, the optimizer exploits the search space. To maximum number of iterations (T), and the assessment cost
switch between the exploration and exploitation stages, a of the objective function (C). Further, the population of
switch probability is used as discussed in Sect. 3.1. This search agents comprises population size (P), dimension of
mentioned strategy may help the optimizer to solve the problem (D), and phasor angles (A). Based on these
numerous optimization problems, but still, some loopholes parameters, during the initializing stage, the time com-
that may result in its poor searching efficiency exist. plexity of the FPA-PPSO is given by OðP  D  AÞ.
Firstly, at a given time interval the optimizer can either During the assessment stage of the objective function, the
perform exploration or exploitation with a randomized time complexity is given as OðP  C  TÞ. The assessment
switching probability. With such a strategy, there may be a of objective function is done twice, once in the case of the
chance that only exploration or exploitation of the search FPA scheme and second in the case of PPSO. With this, the
space is carried out. As a result, there are high chance that total time complexity for this stage is 2  OðP  C  TÞ.
the optimizer may miss out on the optimal solution, For updating the position of search agents, for the FPA
resulting in poor searching efficiency. Secondly, during the scheme, the time complexity is given by OðP  D  TÞ.
local search, there is unavailability of a self-guide which Whereas, in the case of the PPSO scheme it is given by
again may lead the optimizer to stick to a local optimal OðP  D  A  TÞ. Combining all the cases, the time
solution and result in a low searching efficiency [58]. complexity for FPA-PPSO is given as OðP  D  A þ 2 
To overcome these drawbacks, in this study, a hybrid P  C  Tþ P  D  T þ P  D  A  TÞ. The time
FPA-PPSO is suggested which combines the FPA opti- complexity in the case of FPA and PPSO is given as OðP 
mizer with the PPSO scheme to overall improve the D þ P  C  T þ P  D  TÞ and OðP  D  A þ P 
searching ability of the suggested hybrid optimizer. Owing C  T þ P  D  A  TÞ respectively. From this analysis,
to the popularity of the PSO scheme in the field of opti- it is clear that the time required to solve any optimization
mization, for the development of the suggested hybrid problem is more of FPA-PPSO and is generally of the order
method, PSO is selected. Further, among the different FPA\PPSO\FPA  PPSO.
variants of the PSO scheme, the PPSO scheme is selected.
PPSO scheme includes the concept of phasor angles which

123
Cluster Computing

Fig. 2 Pseudo code

3.3.2 Validation using benchmark functions the functions, it is observed that the suggested FPA-PPSO
shows superior performance. In some cases, the method
This section examines twenty-three benchmark functions shows an identical performance in comparison to PSO,
in order to test the exploration, exploitation, and conver- FPA, and PPSO. Considering the deviation values, it can be
gence capabilities of the suggested FPA-PPSO technique. concluded that the suggested optimizer can converge close
Among these twenty three benchmark functions presented to the optimal solution. As a result, we might conclude that
in [50], functions F1 -F7 are the uni-modal benchmark it is more stable. Finally, the hybridization of the FPA and
functions. Uni-modal functions have only one local opti- PPSO is observed to affect the run time performance the
mum (global optimal point), which makes it easier to FPA-PPSO scheme. For all circumstances, the suggested
evaluate the exploitation capabilities of any optimization technique has the longest run time. Lastly, by analyzing the
strategy. Whereas, functions, F8 -F13 and F14 -F23 , comprise convergence plots present for functions F1 , F6 , F10 , F13 ,
multi-modal and fixed dimension multi-modal functions. F18 and F23 in Fig. 4, one may conclude that the suggest
Multi-modal functions have several local optima but a FPA-PPSO approach has a faster convergence rate with
single global optimal point which makes them useful for respect to PPSO and FPA simulated under same conditions.
exploration analysis. To further analyze the results obtained for the bench-
To evaluate the performance of the suggested FPA- mark functions and show the supremacy of the suggested
PPSO technique, all twenty-three functions are simulated FPA-PPSO, in this study, the Wilcoxson test and Friedman
using the LabVIEWÓ2015 platform which is installed on test are also conducted. The results for the two tests are
an Intel (R) core (TM) i7, 2:0GHz processor, 64 bit com- presented in Table 2. From analyzing the Friedman test, it
puter. Each function is simulated 30 times with a popula- is observed that the suggested FPA-PPSO has a mean rank
tion size of 50 and iteration counts of 500. The results of 1.59 which is the smallest in comparison to the other
obtained post 30 simulation runs are utilized to calculate methods PSO, PPSO, and FPA used for comparison. In the
key parameters: average value, standard deviation, and case of the Wilcoxson rank test, the suggested algorithm
computing time. Table 1 shows the results for the sug- shows a significant difference in the results obtained for
gested method and PSO, FPA, and PPSO used for a com- most of the functions as the obtained p-value is less than
parative study. By analyzing the average value obtained for 0.05. Whereas, in some of the cases it shows a similar

123
Cluster Computing

Fig. 3 Flowchart for hybrid


FPA-PPSO START

Initialize a random population of search agents


Define phasor angle for each search agents between (0, 2 )
Define the parameters t max , s,

False
while (t < t max ) Return: Optimal solution

True
STOP
Evaluate the fitness of each search agent us
Define the current best solution
Update S Pr

False if (S Pr < S P ) True

Agents update themselves Agents update themselves


using Eq. (29) using Eq. (28)

Evaluate each search agents and Check if all agents are within the
Boundary conditions

Evaluate the fitness of each search agent


Define the current P b and G b

Update the position of each agents using Eq. (31) and (32)
Update theta using Eq. (35)

Evaluate each search agents and Check if all agents are within the
Boundary conditions

t =t+1

behavior and has a p-value greater than 0.05. From these – From the analysis with the uni-modal function it is
two tests, it can be concluded that the suggested FPA- found that for almost all the functions the suggested
PSSO performs better than the existing PSO, PPSO, and FPA-PPSO scheme provides improved results. This
FPA schemes. confirms that the suggested has better exploitation
From the analysis with the benchmark functions, it is capabilities with respect to FPA and PPSO.
observed that the suggested FPA-PPSO scheme has some – From the analysis with the multi-model function it is
advantages and disadvantages over the FPA & PPSO observed that the suggested method has improved
schemes. These advantages and disadvantages are listed exploration capabilities with respect to FPA and PPSO
below: which help it to provide improved results.

123
123
Table 1 Comparative results analysis for benchmark functions using PSO, PPSO, FPA and FPA-PPSO schemes
Method PSO PPSO FPA FPA-PPSO
BF Average Deviation Time Average Deviation Time Average Deviation Time Average Deviation Time

F1 0.000136 0.000202 3.56 1.237E-22 9.16E-23 4.71 0.019362 0.015973 3.77 3.95E-122 6.83E-122 7.82
F2 0.042144 0.045421 3.72 1.846E-12 2.99E-13 5.16 0.669356 0.285141 4.11 3.00E-65 3.18E-65 8.31
F3 70.12562 22.11924 3.99 1.208E-20 1.02E-20 5.75 0.007454 0.009744 4.68 5.78E-111 1.00E-110 8.88
F4 1.086481 0.317039 4.23 2.837E-11 1.61E-11 6.12 25.77333 4.249736 5.54 1.105E-21 1.91E-21 9.29
F5 96.71832 60.11559 4.78 2.870E?01 0.007081 6.65 23.2983 4.940419 6.10 6.013 7.951 9.91
F6 0.000102 8.28E-05 5.21 2.545E-22 3.93E-22 7.97 0.20181 0.339907 6.78 5.14E-33 6.41E-33 10.47
F7 0.122854 0.044957 5.89 1.139E-03 0.000284 8.36 3.319087 0.549097 7.55 3.9971E-05 1.62E-05 10.99
F8 4841.29 1152.814 86.26 - 9.469E?03 978.46 112.71 - 7120.16 1261.15 99.11 2 3861.29 576.56 160.58
F9 46.70423 11.62938 86.75 3.546E?01 16.151 113.69 22.561 19.156 102.16 16.591 7.451 161.35
F10 0.276015 0.50901 87.52 1.184E-15 2.05E-15 113.99 9.12361 2.61605 102.75 0 0 161.89
F11 0.009215 0.007724 88.11 0 0 114.36 0.671899 0.487123 103.24 0.255724 0.443 162.56
F12 0.006917 0.026301 89.03 - 6.732E-01 1.66E-01 114.82 3.651653 3.252136 103.98 - 0.25721 0.255 162.97
F13 0.00675 0.008907 90.71 1.350E-32 0.00E?00 115.65 17.33197 4.641035 104.57 1.34978E-32 0 163.56
F14 3.627168 2.560828 16.64 3.277E?00 1.047702 29.16 0.998004 0 22.96 0.998004 0 42.3
F15 0.000577 0.000222 17.46 3.221E-02 0.052634 31.32 0.000752 0.000372 23.5 0.0009697 2.033E-05 43.14
F16 - 1.03163 6.25E-16 17.98 - 5.729E-01 0.429824 31.81 - 1.03136 0 25.12 - 1.03136 0 43.86
F17 0.397887 0 18.52 6.182E-01 0.201742 32.45 0.397887 6.8E-17 26.57 0.397887 1.36E-17 44.34
F18 3 1.33E-15 19.1 4.907E?01 12.72001 35.12 3 0 28.03 3 0 44.97
F19 - 3.86278 2.58E-15 19.87 - 2.947E?00 0.606562 36.69 2 3.87943 5.44E-16 30.11 - 3.5941 2.56E-16 45.56
F20 - 3.26634 0.060516 20.57 - 2.163E?00 0.172855 38.15 2 3.28249 0.068774 31.91 - 3.24279 0.06520 45.98
F21 - 6.8651 3.019644 22.48 - 8.469E?00 2.917022 39.06 - 4.27961 1.383014 32.69 2 10.1532 0 46.26
F22 - 8.45653 3.087094 23.31 - 5.286E?00 4.389415 40.16 - 6.09455 3.915841 33.13 2 10.4029 0 46.75
F23 - 9.95291 1.782786 24.66 - 5.426E?00 4.425494 41.65 - 5.38322 4.462785 36.59 2 10.5364 0 46.99
Best results indicate in bold
Cluster Computing
Cluster Computing

Fig. 4 Convergence behavior for benchmark functions F1 , F6 , F10 , F13 , F18 and F23

– From the time complexity analysis and the computa- 4 Simulation and results
tional time presented, the computational requirement of
the suggested scheme is increased with respect to the To evaluate the performance of the suggested hybrid FPA-
existing FPA and PPSO schemes. PPSO technique to solve the scheduling problem, four
– From the convergence plots it observed that the rate of different test cases have been considered in this study. All
convergence of the objective function with respect to the test cases are simulated five times using the
the iteration counts is enhanced. This can compensate LabVIEWÓ2015 platform which is installed on an Intel
for the increased execution time if the population size (R) core (TM) i7, 2:0GHz processor, 64 bit computer. The
and iteration count are optimally set. population size and iteration counts are set as 100 and 500
respectively. The results obtained are then used to compute
statistical parameters such as best value, worst value,

123
Cluster Computing

Table 2 Wilcoxson and


Methods PSO PPSO FPA FPA-PPSO
Friedman rank test results for
benchmark functions using Parameter p-value Rank p-value Rank p-value Rank p-value Rank
PSO, PPSO, FPA and FPA-
PPSO schemes F1 0.036078 3 0.021481 2 0.0090746 4 – 1
F2 0.001398 3 0.030114 2 0.01094813 4 – 1
F3 0.019416 4 0.00677 2 0.01726412 3 – 1
F4 0.020701 3 0.006786 2 0.02777374 4 – 1
F5 0.059961 4 0.049222 2 0.02022949 2 – 2
F6 0.010623 3 0.042838 2 0.02926749 4 – 1
F7 0.016411 3 0.021124 2 0.02937839 4 – 1
F8 0.961749 3.5 0.691532 1 0.8115022 3.5 – 2
F9 0.443971 3 0.325931 3 0.21165431 3 – 1
F10 0.077932 3 0.064887 2 0.21300019 4 – 1
F11 0.067831 2 0.604887 1 0.222182 4 – 3
F12 0.951572 2 0.658813 1 0.00873817 4 – 3
F13 0.018456 3 1 2 0.22552089 4 – 1
F14 0.070875 4 0.306314 3 1 1.5 – 1.5
F15 0.867867 1 0.054649 4 0.90736122 3 – 2
F16 0.999869 3 0.599823 4 1 1.5 – 1.5
F17 1 1 0.539948 4 1 2.5 – 2.5
F18 1 3 0.248306 4 1 1.5 – 1.5
F19 0.295671 2.5 0.369646 4 0.29576447 1 – 2.5
F20 0.995208 1.5 0.797706 4 0.99404078 1.5 – 3
F21 0.69976 4 0.791422 2 0.59620917 3 – 1
F22 0.776739 2 0.572315 4 0.6263637 3 – 1
F23 0.894502 2 0.576534 3 0.573522 4 – 1
Mean Rank - 2.76 - 2.61 - 3.04 – 1.59

average value, and standard deviation. The results are also (MABC/D/Log) [50], modified ABC with disruptive cat map
compared with the results obtained using some other (MABC/D/C) [50], pareto differential evolution (PDE) [50],
existing techniques to show the supremacy of the suggested non-dominating sorting genetic algorithm II (NSGA-II) [50],
scheme. For a fair comparison, the results for the existing Strength Pareto evolutionary algorithm (SPEA-2) [50], FPA
method are also obtained under the same environmental [50], Kho-kho optimization (KKO) [50], quasi-oppositional
conditions i.e. with a population size of 100 and an itera- teaching learning based optimization (QOTLBO) [61], PSO-
tion count of 500. Further, the objectives and other system gravitational search algorithm (PSOGSA) [61], optimization
specifications considered for the three test cases have been without penalty-based optimization by morphological filter
presented briefly in the following section. algorithm (OWP-based OMF) [61].
From the analysis presented in Table 4, it is found the
4.1 Test case - I suggested FPA-PPSO provides the minimum fuel cost. The
average and deviation values indicate that the suggested FPA-
To assess the performance of the suggested FPA-PPSO PPSO has good accuracy and stability compared to the
strategy in solving the scheduling problem, a large system existing PSO, PPSO and FPA scheme. Further, a comparison
with forty generating units supplying a demand of 10500 MW of the generation cost (CT ) and emission rate (ET ) is also
is considered. While solving this test system the objective included. From the result, it is found that the suggested opti-
function considered is (3) considering t = 1. Further, the fuel mizer maintains a tradeoff between the two objectives. In the
and emission coefficients for generating units were obtained case of minimizing generation cost, the KKO scheme provides
from [60]. Post five simulation runs the obtained best power the minimum result. Whereas, the emission cost is minimum
distribution across the generating units which is presented in for the suggested FPA-PPSO scheme. In the case of the
Table 3. Whereas, Table 4 reports the computed statistical computation time, the suggested optimizer has the largest
parameters. Table 4 also displays the results for existing computation time. Finally, Fig. 5 shows the convergence of
methods such as modified ABC with disruptive logistic map the objective function based on iteration count.

123
Cluster Computing

Table 3 Scheduling of
Unit Power (MW) Unit Power (MW)
generating units for test case I
P1 113.959 P21 437.727
P2 114 P22 437.022
P3 119.758 P23 436.799
P4 177.852 P24 438.033
P5 96.9861 P25 436.548
P6 128.506 P26 437.922
P7 299.635 P27 24.1173
P8 297.509 P28 24.237
P9 297.195 P29 24.4608
P10 131.219 P30 97
P11 306.922 P31 173.582
P12 306.868 P32 174.557
P13 433.594 P33 175.062
P14 408.207 P34 200
P15 411.999 P35 199.983
P16 411.062 P36 200
P17 452.293 P37 103.866
P18 450.987 P38 102.731
P19 436.952 P39 103.753
P20 437.413 P40 439.683
Fuel Cost ($/hr) 128965 Emission Cost (tons/hr) 178038
Total Cost ($/hr) 153501.5

Table 4 Comparative analysis for Test Case I


Algorithm Best ($/hr) Worst ($/hr) Average ($/hr) Deviation CT ($/hr) ET (tons/hr) Time (sec)

PDE [50] 168750 – – – 125730 211770 –


NSGA-II [50] 168390 – – – 125830 210950 –
SPEA-2 [50] 168455 – – – 125810 211100 –
MABC/D/C [50] 190525.6 – – – 124490.9 256560.3 –
MABC/D/Log [50] 190525.8 – – – 124491.2 256560.3 –
KKO [50] 164732 – – 123034 206430 –
QOTLBO [61] 165825.7 - – – 125161 206490.4 –
PSOGSA [61] 153651.3 – – – 128710.9 178591.7 –
OWP-based OMF [61] 153576.6 – – – 128596.0 178557.2 0.078
PSO [56] 153717 153786 153725 83.19 128713 178703 10.14
PPSO [57] 153659 153702 153678 47.23 128749 178570 25.11
FPA [55] 153677 153764 153714 63.15 128803 178551 12.96
FPA-PPSO 153501.5 153582.1 153543.3 40.39 128965 178038 33.76
Best results indicate in bold

Kindly note that the best value are computed based on the CT and ET using the objective (3)

coefficients and emission coefficients along with the


dynamic load profile are taken from [8]. The suggested
4.2 Test case II hybrid FPA-PPSO is tested considering two different
objectives i.e. DELD and DEED. The solar and wind
For the next test case, a basic microgrid model with three uncertainties are not considered for this test case. The
conventional thermal generating unit systems along with a dynamic power output from the wind and solar farm is
solar and wind farm has been considered. The fuel taken from [8].

123
Cluster Computing

Considering the DELD objective, FPA-PPSO is simu- study, a statistical analysis of the results obtained using the
lated under two case scenarios (i) without considering RES suggested method and other existing methods such as grey
and (ii) considering the RES. The behavior of generating wolf optimization (GWO), whale optimization algorithm
units and the obtained best fuel cost for these two test (WAO), symbiotic organism search (SOS) algorithm, PSO,
scenarios are presented in Table 5. For a comparative PPSO, etc. are presented in Table 6. It is observed from
Table 6 that the suggested method optimizes the generating
105
units more efficiently as the obtained cost function is
1.539 minimum. Further, from the comparison of the average
Convergence for FPA-PPSO scheme
Convergence for PSO scheme value and standard deviation, it can be concluded that the
Convergence for PPSO scheme
suggested method provides a more promising and
Objective ($/hr)

1.538 Convergence for FPA scheme


stable result in comparison to the existing FPA and PPSO
1.537
techniques. Further, to study the convergence rate and
behavior of the objective function, Fig. 7 is presented. By
observing Fig. 7, it can be concluded that FPA-PPSO has a
1.536 faster convergence rate compared to FPA and PPSO.
Further, considering the DEED objective, FPA-PPSO is
1.535 also simulated for two case scenarios (i) without consid-
0 100 200 300 400 500
ering RES and (ii) considering RES. For these two cases,
Iteration count
the power distribution among the generating units along
Fig. 5 Convergence for FPA-PPSO for test case I with the best-obtained fuel cost is presented in Table 5. To

Table 5 Behavior of generating units for test case II under the two case scenarios
Objectives DELD DEED
Sources Without RES With RES Without RES With RES
Units P1T P2T P3T P1T P2T P3T P1T P2T P3T P1T P2T P1T

H1 37.5211 45.4687 57.0101 37.2583 44.9087 56.133 50 40 50 48.2999 40.0001 50


H2 40.8139 48.2363 60.9498 38.0463 45.8669 57.5868 60 40 50 51.4951 40.0045 50.0004
H3 42.534 49.6393 62.8267 39.3672 47.0607 59.302 63.1882 41.8117 50.0001 55.73 40 50
H4 44.2339 50.9827 64.7835 38.6049 46.3938 58.3413 64.8169 45.1828 50.0003 53.34 40 50
H5 45.8769 52.4311 66.692 43.4043 50.4471 63.9287 66.3135 48.6865 50 64.19 43.59 50
H6 47.5748 53.7988 68.6264 45.8452 52.4895 66.7253 67.3892 51.1313 51.4796 66.3114 48.7476 50.001
H7 49.1848 55.2664 70.5489 42.1822 49.4164 62.4714 68.0992 52.7741 54.1267 62.9671 41.1028 50
H8 50.9115 56.6183 72.4702 37.0584 44.3778 55.8238 68.8075 54.4046 56.7879 47.26 40 50
H9 61.0583 64.9904 83.9513 46.0145 52.5564 66.7991 73.0538 64.2148 72.7313 66.419 48.9498 50.0012
H10 67.6734 70.5671 91.7595 48.5147 54.6534 69.6118 75.8786 70.7862 83.3352 67.7741 52.0298 52.9762
H11 71.2077 73.2703 95.5219 64.3093 67.6865 87.7942 77.318 74.0496 88.6324 74.4591 67.4253 77.9057
H12 74.6026 76.0886 99.3088 66.9853 69.9006 90.8141 78.7461 77.3602 93.8937 75.5851 70.0246 82.0904
H13 71.0479 73.3461 95.606 55.5061 60.4412 77.7627 77.3222 74.0119 88.6659 70.7436 58.8971 64.0693
H14 64.3887 67.7546 87.8567 51.8687 57.4474 73.5238 74.4702 67.5386 77.9911 69.2237 55.312 58.3043
H15 57.6334 62.221 80.1457 51.4584 57.1172 73.0844 71.638 60.9622 67.3999 69.047 54.954 57.659
H16 50.8284 56.6979 72.4738 44.5694 51.2946 65.1261 68.8041 54.4287 56.7673 65.132 45.8574 50.0006
H17 47.5397 53.806 68.6543 43.1845 50.2019 63.6036 67.3796 51.0749 51.5456 63.9864 43.003 50.0006
H18 52.6128 57.9972 74.39 51.1515 56.8908 72.7776 69.5076 56.063 59.4295 68.9148 54.6679 57.2373
H19 57.6387 62.2032 80.1581 57.4042 62.0287 79.8171 71.6497 60.9518 67.3985 71.5659 60.7143 66.9698
H20 71.0918 73.3303 95.5779 70.9701 73.3219 95.538 77.3234 74.0294 88.6472 77.287 73.9844 88.5587
H21 66.0851 69.1807 89.7343 66.0369 69.1414 89.6717 75.195 69.1589 80.6462 75.1833 69.0746 80.5921
H22 54.2796 59.3891 76.3312 54.1267 59.3486 76.2146 70.2235 57.7044 62.0721 70.1853 57.5705 61.9342
H23 44.164 51.0657 64.7703 43.8154 50.7457 64.3689 64.765 45.235 50 64.4413 44.4886 50
H24 39.1372 46.8756 58.9872 38.917 46.7005 58.8025 55 40 50 54.42 40 50
Cost ($) 176165.8 299893.4 202871.3 325349.2

123
Cluster Computing

Table 6 Comparative analysis of the results for test case II considering DELD objective
Techniques Without RES With RES
Best ($) Worst ($) Average ($) Deviation Best ($) Worst ($) Average ($) Deviation Time (sec)

PSO [8] 176177.9175 – – – 299919.4357 – – – –


DE [8] 176169.0719 – – – 299916.0487 – – – –
SOS [8] 176168.04244 – – – 299906.3846 – – – –
GWO [8] 176167.827 – – – 399896.6562 – – – –
WOA [8] 176166.5662 – – – 299895.531 – – – –
BDO [4] 176166 – – – 299895.1 – – – –
FPA [8] 176172 176187 176175 10.8167 299900 299908 299904 11.9304 135.19
PPSO [8] 176168 176178 176172 5.1316 299897 299905 299899 5.2915 144.74
FPA-PPSO 176165.7 176168.8 176166.5 4.16333 299893.4 299903 299898.1 4.04145 189.16
Best results indicate in bold

Table 7 Comparative analysis of the results for test case II considering DEED objective
Techniques Without RES With RES
Best ($) Worst ($) Average ($) Deviation Best ($) Worst ($) Average ($) Deviation Time (sec)

PSO [8] 202886.6496 – – – 325377.3173 – – – –


DE [8] 202884.8852 – – – 325371.3072 – – – –
SOS [8] 202882.0837 – – – 325369.7976 – – – –
GWO [8] 202882.6042 – – – 325368.4448 – – – –
WOA [8] 202881.7751 – – – 325364.4919 – – – –
BDO [4] 202874.7 – – – 325351.6 – – – –
FPA [8] 202881.5 202895 202891 9.602 325366 325402 325384 28.155 139.42
PPSO [8] 202880 202893 202887 6.957 325359 325400 325376 27.124 147.63
FPA-PPSO 202871.3 202891 202884 5.504 325349.2 325392 325356 20.2662 195.06
Best results indicate in bold

show the effectiveness of the suggested FPA-PPSO in Whereas, the dynamic load profile is considered to be the
solving this problem, a comparative study using statistical same as that used for the second test case. Further, an
data obtained using the suggested method and other uncertainty model for solar generation and wind generation
existing methods such as PSO, PPSO, etc. have been pre- as presented in [52, 53] has been considered in this study.
sented in Table 7. From the comparative study presented in The optimal power allocations for the generating units
Table 7, it is observed that the suggested FPA-PPSO not along with the total fuel cost are presented in Table 10. For
only provides the best fuel cost but also the results obtained a comparative analysis and to show the effectiveness of the
are more stable compared to the existing FPA and PPSO. suggested FPA-PPSO method to solve this problem, results
Next, to analyze the convergence behavior of the objective with the existing optimization techniques have also been
function, Fig. 7 is presented. It is observed from Fig. 7, presented. The techniques used for this comparative study
that FPA-PPSO has a faster convergence rate than FPA and are PSO, GWO, PPSO, and FPA. All the comparative
PPSO. results are presented in Table 9. From Table 9, it is
observed that the suggested method provides the best fuel
4.3 Test case III cost. Further, by observing the average value and standard
deviation, it can be concluded that the suggested method
For the third test case, a conventional thermal generator, can provide effective results with minimum deviation.
three diesel generators, a solar farm, a wind farm, and ESS Hence, the method is more stable. Further, to analyze the
units have been considered. The fuel and emission coeffi- convergence rate and behavior of objective function with
cients along with ESS capacity are presented in Table 8. time (iteration counts), Fig. 7 is presented. It is observed

123
Cluster Computing

Table 8 Generating units cost and emission coefficients data for test case III [8, 62]
Units Gmin (MW) Gmax (MW) ai ($=MW 2 h) bi ($=MWh) ci ($=h) ai (Kg=MW 2 h) bi (Kg/MWh) ci ($=h) gb

T1 37 190 0.021 20.4 600 0.012 0.555 30 -


D1 0 1.5 0.00024 0.21 15.3 0.000105 0.01355 0.6 -
D2 0 1 0.000425 0.3 14.88 8.00E-05 0.006 0.45 –
D3 0 1 0.000315 0.306 9 0.00012 0.00555 0.3 –
ESS 0 300 – 250 – – – – 90%

Table 9 Comparative analysis


Technique Best ($) Worst ($) Average ($) Deviation ($) Time (sec)
of the results for test case III
PSO [56] 140568 140575 140573 4.04145 162.26
GWO [63] 140600 140610 140604 5.2915 178.26
FPA [55] 140569 140582 140574 6.80686 168.14
PPSO [57] 140559 140569 140563 5.2915 176.36
FPA-PPSO 140543 140562 140554 3.84886 210.11
Best results indicate in bold

260 and other parameters such as load profile, ESS units, etc.
Actual load profile
Load profile after 10% DRM are considered similar to that used in test case III.
240
The optimal power allocation to the generating units and
220 the optimal fuel cost are presented in Table 12. Further, the
Load demand (MW)

impact of DRM on the generation side can be observed


200
from Fig. 6. From Table 10 and 12, it is observed that the
180 total fuel cost after considering DRM is reduced by $
25545. The load distribution among the generating units is
160
improved which also reduces the stress on the ESS units.
140 ESS units feed 166.5731 MW of power which is less than
120
274.6371 MW in the case when DRM is not considered.
Further, load profile optimization also improves the power
100
2 4 6 8 10 12 14 16 18 20 22 24
delivery as the overall load shedding is reduced from
Hour 30.51923 MW to 18.51923 MW.
Next, to show the impact of the suggested FPA-PPSO
Fig. 6 Load profile with and without DRM
scheme in solving this problem, a comparative study has
been presented. For this comparison, PSO, FPA, GWO, and
from Fig. 7 that the suggested method has a good con-
PPSO have been considered. All the comparative results
vergence rate as compared to other methods such as PSO,
are presented in Table 11. It is observed from Table 11 that
GWO, PPSO, and FPA used for comparison.
the overall performance of the suggested FPA-PPSO is
better compared to the existing methods used for the
4.4 Test case IV
comparative study. The suggested method provides the best
fuel cost with good stability as the average value and
This test case is an extension of Test Case III and presents a
standard deviations are minimum. Further, Fig. 7 is pre-
study on the effect of DRM on the generation side. To
sented to show the behavior of the objective function
implement a DRM scheme, like [47, 64], 10 percent of the
obtained using different methods. It is observed from Fig. 7
loads from peak locations of 4, 5, 6, and 7 h are shifted to
that suggested FPA-PPSO converges faster than PPSO,
the off-peak location of 8, and 9, 10, and 11 h. The
PSO, GWO, and FPA.
advantage of DRM is that the peak load of the load
demands is reduced in comparison to the actual day-ahead
load demand. The generating units are then optimized
using the new optimized load profile. The generating units

123
Cluster Computing

Table 10 Behavior of generating units for Test case III


Hours PT P1D P2D P3D PW PS PB Load sheading Load demand ESS Behavior

H1 136.564 1.36983 0.556201 0.391 1.11917 0 0 0 140 300


H2 146.042 1.14466 0.819128 0.479697 1.51417 0 0 0 150 300
H3 150.072 1.49647 0.857349 0.993506 1.58 0 0 0 155 300
H4 155.443 1.14815 0.962832 0.875715 1.58 0 0 0 160 300
H5 161.286 1.32897 0.181593 0.524215 1.58 0.101816 0 0 165 300
H6 165.827 1.00561 0.800061 0.55603 1.58 0.232417 0 0 170 300
H7 169.655 1.47783 0.877018 0.9327 1.58 0.477168 0 0 175 300
H8 174.393 1.49821 0.936305 0.96 1.58 0.633548 0 0 180 300
H9 190 1.5 1 1 1.2245 0.757286 13.0664 1.45182 210 286.934
H10 190 1.5 1 1 1.01383 0.870068 31.1545 3.46161 230 255.779
H11 190 1.5 1 1 0.658333 1.01464 40.3443 4.4827 240 215.435
H12 190 1.5 1 1 0.5135 0.959335 49.5244 5.50272 250 165.91
H13 190 1.5 1 1 0.9085 0.811376 40.3021 4.47801 240 125.608
H14 190 1.5 1 1 0.9875 0.665094 22.3627 2.48474 220 103.246
H15 190 1.5 1 1 1.106 0.589812 4.32377 0.480419 200 98.9218
H16 190 1.5 1 1 0.895333 0.265628 0 0 180 112.117
H17 190 1.5 1 1 0.6715 0.223167 0 0 170 134.072
H18 190 1.5 1 1 0.8295 0.162946 0 0 185 142.615
H19 190 1.5 1 1 1.04017 0.068275 4.8524 0.539156 200 137.763
H20 190 1.5 1 1 0.724167 0 41.1983 4.57758 240 96.5644
H21 190 1.5 1 1 0.895333 0 27.5442 3.06047 225 69.0202
H22 190 1.5 1 1 1.027 0 0 0 190 73.0945
H23 190 1.5 1 1 0.948 0 0 0 160 104.098
H24 190 1.5 1 1 0.842667 0 0 0 145 148.506

Table 11 Comparative analysis


Technique Best ($) Worst ($) Average ($) Deviation ($) Time (sec)
of the results for test case IV
PSO [56] 115039 115062 115050 11.5036 163.61
GWO [63] 115017 115039 115025 11.9304 181.01
FPA [55] 115021 115029 115025 4 170.96
PPSO [57] 115005 115035 115023 15.695 178.09
FPA-PPSO 114988 115011 115002 2.50555 211.96
Best results indicate in bold

4.5 Analysis on the performance of FPA-PPSO optimizer produces a value that is close to the best values
scheme obtained. This observation confirms that the suggested
methods show more accuracy and stability when compared
From the results presented in Tables 4, 6, 7, 9 and 11 for to other methods used for the comparisons. A similar
the four test cases, it is observed that the suggested FPA- conclusion is already drawn from the analysis with the
PPSO obtains the best-optimized result. The objective benchmark functions. Hence, from the analysis, it can be
function is optimized to provide the minimum cost. This concluded that the suggested method is superior in terms of
confirms that the suggested method shows a superior per- efficiency, accuracy, and stability in comparison to the
formance in comparison to the other methods used for methods used for comparative study.
comparisons. Based on the average and deviation values To further investigate and confirm the supremacy of the
obtained for test cases, it is observed that the suggested suggested optimizer, the Wilcoxson and Friedman test is
conducted for the results obtained for the four test cases.

123
Cluster Computing

Fig. 7 Convergence behavior of objective function for test case II, III & IV

The results for the two tests are presented in Table 13 along This confirms that the suggested method is more efficient
with the result obtained for PSO, PPSO, and FPA. By than PPSO, PSO, and FPA.
observing the results for the Friedman test, it is observed
that the suggested FPA-PPSO scheme gets a mean rank of
1. Whereas, the PPSO gets a rank of 2.25 followed by FPA 5 Conclusions and future research
with a rank of 3.125, and PSO with a rank of 3.625. As the
rank obtained by FPA-PPSO is the smallest, it is regarded This article presents a hybrid version of the flower polli-
as the winner. In the case of the Wilcoxson test, it is nation algorithm and phasor particle swarm optimization
observed that in all the cases, the suggested FPA-PPSO technique i.e. FPA-PPSO scheme to solve complex opti-
scheme gets a score less than the significant level of 0.05. mization problems. The supremacy of the suggested opti-
mizer with respect to existing FPA and PPSO schemes is

123
Cluster Computing

Table 12 Behavior of generating units for test case IV


Hours PT P1D P2D P3D PW PS PB Load sheading Load demand ESS Behavior

H1 161.819 0.587504 0.759304 0.714897 1.11917 0 0 0 165 300


H2 169.734 1.34698 0.691374 0.713819 1.51417 0 0 0 174 300
H3 174.889 1.4286 0.780776 0.321528 1.58 0 0 0 179 300
H4 178.652 1.26445 0.691205 0.812426 1.58 0 0 0 183 300
H5 160.08 1.43638 0.999171 0.8035 1.58 0.101816 0 0 165 300
H6 165.082 1.25732 0.861442 0.986761 1.58 0.232417 0 0 170 300
H7 169.748 1.48608 0.737791 0.971172 1.58 0.477168 0 0 175 300
H8 174.487 1.49709 0.943193 0.868 1.58 0.633548 0 0 180 300
H9 190 1.5 1 1 1.2245 0.757286 13.0664 1.45182 210 286.934
H10 190 1.5 1 1 1.01383 0.870068 10.4545 1.16161 207 276.479
H11 190 1.5 1 1 0.658333 1.01464 18.7443 2.0827 216 257.735
H12 190 1.5 1 1 0.5135 0.959335 27.0244 3.00272 225 230.71
H13 190 1.5 1 1 0.9085 0.811376 18.7021 2.07801 216 212.008
H14 190 1.5 1 1 0.9875 0.665094 22.3627 2.48474 220 189.646
H15 190 1.5 1 1 1.106 0.589812 4.32377 0.480419 200 185.322
H16 190 1.5 1 1 0.895333 0.265628 0 0 180 198.517
H17 190 1.5 1 1 0.6715 0.223167 0 0 170 220.472
H18 190 1.5 1 1 0.8295 0.162946 0 0 185 229.015
H19 190 1.5 1 1 1.04017 0.068275 4.8524 0.539156 200 224.163
H20 190 1.5 1 1 0.724167 0 19.5983 2.17758 216 204.564
H21 190 1.5 1 1 0.895333 0 27.5442 3.06047 225 177.02
H22 190 1.5 1 1 1.027 0 0 0 190 181.095
H23 190 1.5 1 1 0.948 0 0 0 160 212.098
H24 190 1.5 1 1 0.842667 0 0 0 169 234.906

Table 13 Wilcoxson and


Method FPA-PPSO PPSO FPA PSO
Friedman rank test results for
test case I, II, III and IV Parameter Rank p-value Rank p-value Rank p-value Rank p-value

Test Case 1 1 – 2.75 0.00587 2.5 0.00141 3.75 0.0081


Test Case II 1 – 2.25 0.00407 4 0.00596 2.75 0.0006
Test Case III 1 – 2 0.00617 3 0.00559 4 0.0072
Test Case IV 1 – 2 0.00789 3 0.00951 4 0.0041
Mean Rank 1 – 2.25 - 3.125 - 3.625 –

validated using twenty-three benchmark functions. Based load shedding. The comparative study presented for the test
on the results obtained it is found that the suggested cases proves that the suggested FPA-PPSO method has
methods show a better performance with respect to FPA, good stability and the probability of obtaining an optimal
PPSO, and PSO methods. Next, the suggested FPA-PPSO solution is high. The results also show that the suggested
method is applied to solve the generation scheduling method provides a superior result with respect to some
problem of a microgrid, including renewable generating well-known existing methods such as PSO, FPA, PPSO,
sources and energy storage systems. The suggested method WOA, GWO, DE, etc.
is tested using three cases which include different para- To conclude, the result obtained confirms that the sug-
metric conditions. Further, an additional test case is also gested FPA-PPSO scheme shows superior performance in
considered which presents a study on the impact of demand comparison to the existing optimization scheme. Addi-
response management on the generation side. It is observed tionally, it is also observed that by the use of a demand
that DRM improves the overall generation cost and reduces management scheme, the scheduling problem can be

123
Cluster Computing

efficiently solved. In the near future, the works that can be 5. Basu, M.: Economic environmental dispatch using multi-objec-
taken up to further enhance the contributions of this study tive differential evolution. Appl. soft Comput. 11(2), 2845–2853
(2011)
include: 6. Ali, E., Elazim, S.A.: Mine blast algorithm for environmental
economic load dispatch with valve loading effect. Neural Com-
– Testing and analyzing the performance of the suggested
put. Appl. 30(1), 261–270 (2018)
optimizer in scheduling the generating units by con- 7. Bhattacharjee, K., Bhattacharya, A., nee Dey, S.H.: Backtracking
sidering more practical constraints such as transmission search optimization based economic environmental power dis-
losses, valve point effect, ramp rate limits, prohibited patch problems. Int. J. Electr. Power & Energy Syst. 73, 830–842
(2015)
operating zones, etc. encountered in the real world
8. Dey, B., Roy, S.K., Bhattacharyya, B.: Solving multi-objective
which were not included in this study. economic emission dispatch of a renewable integrated microgrid
– Development and implementation of a more rigorous using latest bio-inspired algorithms. Eng. Sci. Technol. Int. J.
demand management scheme which can further help in 22(1), 55–66 (2019)
9. Nguyen, T.T., Vo, D.N.: The application of one rank cuckoo
scheduling the load profile and generating units.
search algorithm for solving economic load dispatch problems.
– Population size and iteration counts used in the Appl. Soft Comput. 37, 763–773 (2015)
suggested FPA-PPSO method can be made adaptive 10. Wood, A.J., Wollenberg, B.F., Sheblé, G.B.: Power generation,
to improve the computational time to solve the operation, and control. Wiley, New York (2013)
11. Lee, K., Park, Y., Ortiz, J.: Fuel-cost minimisation for both real-
optimization problems as in most of the cases the
and reactive-power dispatches. In: IEE Proceedings C (Genera-
suggested optimizer convergence at a faster rate. tion, Transmission and Distribution), vol. 131, pp. 85–93. IET
– Lastly, based on the performance shown by the (1984)
suggested FPA-PPSO scheme, it can be applied to 12. Srivastava, A., Das, D.K.: Criminal search optimization algo-
rithm: a population-based meta-heuristic optimization technique
solve other offline optimization problems.
to solve real-world optimization problems. Arabian J. Sci. Eng.
47(3), 3551–3571 (2022)
13. Shami, T.M., Grace, D., Burr, A., Mitchell, P.D.: Single candi-
date optimizer: a novel optimization algorithm. Evol. Intell.
Author contributions Abhishek Srivastava: Conceptualization,
17(2), 863–887 (2024)
Methodology, Visualization, Investigation, Writing- Original draft
14. Bodha, K.D., Mukherjee, V., Yadav, V.K.: A player unknown’s
preparation. Dushmanta Kumar Das: Writing- Reviewing and Editing
battlegrounds ranking based optimization technique for power
and Supervision. Siseyiekuo Khatsu: Conceptualization, Methodol-
system optimization problem. Evol. Syst. 14(2), 295–317 (2023)
ogy, Visualization, Investigation.
15. Zhao, W., Wang, L., Mirjalili, S.: Artificial hummingbird algo-
rithm: a new bio-inspired optimizer with its engineering appli-
Funding The authors have not disclosed any funding.
cations. Comput. Methods Appl. Mech. Eng. 388, 114194 (2022)
16. Ezugwu, A.E., Agushaka, J.O., Abualigah, L., Mirjalili, S.,
Data availability The data used to support the finding are cited within
Gandomi, A.H.: Prairie dog optimization algorithm. Neural
the article.
Comput. Appl. 34(22), 20017–20065 (2022)
17. Hassan, M.H., Kamel, S., Selim, A., Shaheen, A., Yu, J., El-
Sehiemy, R.: Efficient economic operation based on load dispatch
Declarations of power systems using a leader white shark optimization algo-
rithm. Neural Comput. Appl. 36(18), 10613–10635 (2024)
Conflict of interest The authors declare no potential Conflict of 18. Hassan, M.H., Kamel, S., Jurado, F., Desideri, U.: Global opti-
interest. mization of economic load dispatch in large scale power systems
using an enhanced social network search algorithm. Int. J. Electr.
Power Energy Syst. 156, 109719 (2024)
References 19. Khunkitti, S., Premrudeepreechacharn, S., Siritaratiwat, A.: A
two-archive Harris Hawk optimization for solving many-objec-
tive optimal power flow problems. IEEE Access 11,
1. Roy, C., Das, D.K.: A hybrid genetic algorithm (ga)-particle 134557–134574 (2023)
swarm optimization (pso) algorithm for demand side manage- 20. Khunkitti, S., Siritaratiwat, A., Premrudeepreechacharn, S.: A
ment in smart grid considering wind power for cost optimization. many-objective marine predators algorithm for solving many-
Sādhanā 46(2), 101 (2021) objective optimal power flow problem. Appl. Sci. 12(22), 11829
2. Alvarez, E., Campos, A.M., Arboleya, P., Gutiérrez, A.J.: (2022)
Microgrid management with a quick response optimization 21. Cai, N., Nga, N.T.T., Mitra, J.: Economic dispatch in microgrids
algorithm for active power dispatch. Int. J. Electr. Power Energy using multi-agent system. In: 2012 North American Power
Syst. 43(1), 465–473 (2012) Symposium (NAPS), pp. 1–5. IEEE (2012)
3. Srivastava, A., Das, D.K.: An interactive class topper optimiza- 22. Augustine, N., Suresh, S., Moghe, P., Sheikh, K.: Economic
tion with energy management scheme for an interconnected dispatch for a microgrid considering renewable energy cost
microgrid. Electr. Eng. 106(2), 2069–2086 (2024) functions. In: 2012 IEEE PES Innovative Smart Grid Technolo-
4. Srivastava, A., Das, D.K.: A bottlenose dolphin optimizer: an gies (ISGT), pp. 1–7. IEEE (2012)
application to solve dynamic emission economic dispatch prob- 23. Arif, A., Javed, F., Arshad, N.: Integrating renewables economic
lem in the microgrid. Knowl. Based Syst. 243, 108455 (2022) dispatch with demand side management in micro-grids: a genetic
algorithm-based approach. Energy Effic. 7(2), 271–284 (2014)

123
Cluster Computing

24. Liu, X., Ding, M., Zhang, Y., Xu, N.: Dynamic economic dis- 41. Srivastava, A., Das, D.K., Gupta, P.K.: A quantum class topper
patch for microgrids. In: Zhongguo Dianji Gongcheng Xue- optimization algorithm to solve combined emission economic
bao(Proceedings of the Chinese Society of Electrical dispatch problem. Evolutionary Intelligence, pp. 1–15 (2020)
Engineering), vol. 31, pp. 77–84. Chinese Society for Electrical 42. Logenthiran, T., Srinivasan, D., Shun, T.Z.: Demand side man-
Engineering (2011) agement in smart grid using heuristic optimization. IEEE Trans.
25. Dubey, H.M., Pandit, M., Panigrahi, B.: Hybrid flower pollination Smart Grid 3(3), 1244–1252 (2012)
algorithm with time-varying fuzzy selection mechanism for wind 43. Mehdizadeh, A., Taghizadegan, N.: Robust optimisation
integrated multi-objective dynamic economic dispatch. Renew. approach for bidding strategy of renewable generation-based
Energy 83, 188–202 (2015) microgrid under demand side management. IET Renew. Power
26. Alham, M., Elshahed, M., Ibrahim, D.K., El Zahab, E.E.D.A.: A Gener. 11(11), 1446–1455 (2017)
dynamic economic emission dispatch considering wind power 44. Mellouk, L., Boulmalf, M., Aaroud, A., Zine-Dine, K., Benhad-
uncertainty incorporating energy storage system and demand side dou, D.: Genetic algorithm to solve demand side management
management. Renew. Energy 96, 800–811 (2016) and economic dispatch problem. Procedia Comput. Sci. 130,
27. Elattar, E.E.: Modified harmony search algorithm for combined 611–618 (2018)
economic emission dispatch of microgrid incorporating renew- 45. Roy, C., Das, D.K., Srivastava, A.: Particle swarm optimization
able sources. Energy 159, 496–507 (2018) based cost optimization for demand side management in smart
28. Gil-González, W., Montoya, O.D., Holguı́n, E., Garces, A., Gri- grid. In: 2019 International Conference on Electrical, Electronics
sales-Noreña, L.F.: Economic dispatch of energy storage systems and Computer Engineering (UPCON), pp. 1–6. IEEE (2019)
in dc microgrids employing a semidefinite programming model. 46. Kumar, K.P., Saravanan, B.: Day ahead scheduling of generation
J. Energy Storage 21, 1–8 (2019) and storage in a microgrid considering demand side management.
29. Xin-gang, Z., Ze-qi, Z., Yi-min, X., Jin, M.: Economic-environ- J. Energy Storage 21, 78–86 (2019)
mental dispatch of microgrid based on improved quantum parti- 47. Basu, M.: Optimal generation scheduling of hydrothermal system
cle swarm optimization. Energy 195, 117014 (2020) with demand side management considering uncertainty and out-
30. Ali, Z.M., Aleem, S.H.A., Omar, A.I., Mahmoud, B.S.: Eco- age of renewable energy sources. Renew. Energy 146, 530–542
nomical-environmental-technical operation of power networks (2020)
with high penetration of renewable energy systems using multi- 48. Hosseini, S.E., Najafi, M., Akhavein, A., Shahparasti, M.: Day-
objective coronavirus herd immunity algorithm. Mathematics ahead scheduling for economic dispatch of combined heat and
10(7), 1201 (2022) power with uncertain demand response. IEEE Access pp. 1–1
31. Said, M., Houssein, E.H., Deb, S., Ghoniem, R.M., Elsayed, (2022). 10.1109/ACCESS.2022.3168306
A.G.: Economic load dispatch problem based on search and 49. Eskandari, H., Kiani, M., Zadehbagheri, M., Niknam, T.: Optimal
rescue optimization algorithm. IEEE Access pp. 1–1 (2022). scheduling of storage device, renewable resources and hydrogen
10.1109/ACCESS.2022.3168653 storage in combined heat and power microgrids in the presence
32. Hassan, M.H., Kamel, S., Safaraliev, M., Kokin, S.: Improved plug-in hybrid electric vehicles and their charging demand.
techno-economic optimization of hybrid solar/wind/fuel cell/ J. Energy Storage 50, 104558 (2022)
diesel systems with hydrogen energy storage (2024) 50. Srivastava, A., Das, D.K.: A new kho-kho optimization algo-
33. Srivastava, A., Das, D.K.: An adaptive chaotic class topper rithm: an application to solve combined emission economic dis-
optimization technique to solve economic load dispatch and patch and combined heat and power economic dispatch problem.
emission economic dispatch problem in power system. Soft Eng. Appl. Artif. Intell. 94, 103763 (2020)
Computing, pp. 1–22 (2022) 51. Bhoye, M., Pandya, M., Valvi, S., Trivedi, I.N., Jangir, P., Par-
34. Sharma, D., Gaur, P., Mittal, A.: Comparative analysis of hybrid mar, S.A.: An emission constraint economic load dispatch
gapso optimization technique with GA and PSO methods for cost problem solution with microgrid using jaya algorithm. In: 2016
optimization of an off-grid hybrid energy system. Energy Tech- International Conference on Energy Efficient Technologies for
nol. Policy 1(1), 106–114 (2014) Sustainability (ICEETS), pp. 497–502. IEEE (2016)
35. Coelho, L.S., Mariani, V.C.: Combining of chaotic differential 52. Abbaspour, M., Satkin, M., Mohammadi-Ivatloo, B., Lotfi, F.H.,
evolution and quadratic programming for economic dispatch Noorollahi, Y.: Optimal operation scheduling of wind power
optimization with valve-point effect. IEEE Trans. Power Syst. integrated with compressed air energy storage (CAES). Renew.
21(2), 989–996 (2006) Energy 51, 53–59 (2013)
36. Jayabarathi, T., Bahl, P., Ohri, H., Yazdani, A., Ramesh, V.: A 53. Nguyen, D.T., Le, L.B.: Optimal bidding strategy for microgrids
hybrid BFA-PSO algorithm for economic dispatch with valve- considering renewable energy and building thermal dynamics.
point effects. Front. Energy 6(2), 155–163 (2012) IEEE Trans. Smart Grid 5(4), 1608–1620 (2014)
37. Liang, H., Liu, Y., Li, F., Shen, Y.: A multiobjective hybrid bat 54. Elsakaan, A.A., El-Sehiemy, R.A., Kaddah, S.S., Elsaid, M.I.:
algorithm for combined economic/emission dispatch. Int. Optimal economic-emission power scheduling of RERs in mgs
J. Electr. Power Energy Syst. 101, 103–115 (2018) with uncertainty. IET Gener. Transm. Distrib. 14(1), 37–52
38. Nemati, M., Braun, M., Tenbohlen, S.: Optimization of unit (2019)
commitment and economic dispatch in microgrids based on 55. Yang, X.S.: Flower pollination algorithm for global optimization.
genetic algorithm and mixed integer linear programming. Appl. In: International conference on unconventional computing and
Energy 210, 944–963 (2018) natural computation, pp. 240–249. Springer (2012)
39. Dhiman, G.: Moshepo: a hybrid multi-objective approach to solve 56. Kennedy, J., Eberhart, R.: Particle swarm optimization (pso). In:
economic load dispatch and micro grid problems. Appl. Intell. Proceedings IEEE International Conference on Neural Networks,
50(1), 119–137 (2020) Perth, Australia, pp. 1942–1948 (1995)
40. Srivastava, A., Das, D.K.: A new aggrandized class topper opti- 57. Gholamghasemi, M., Akbari, E., Asadpoor, M.B., Ghasemi, M.:
mization algorithm to solve economic load dispatch problem in a A new solution to the non-convex economic load dispatch
power system. IEEE Transactions on Cybernetics, pp. 1–11 problems using phasor particle swarm optimization. Appl. Soft
(2020). 10.1109/TCYB.2020.3024607 Comput. 79, 111–124 (2019)

123
Cluster Computing

58. Jia, Y., Wang, S., Liang, L., Wei, Y., Wu, Y.: A flower polli- Dushmanta Kumar Das received
nation optimization algorithm based on cosine cross-generation his B.Tech. in Electronics and
differential evolution. Sensors 23(2), 606 (2023) Instrumentation Engineering
59. Ghasemi, M., Akbari, E., Rahimnejad, A., Razavi, S.E., Ghavi- from Biju Patnaik University of
del, S., Li, L.: Phasor particle swarm optimization: a simple and Technology, Odisha, India. He
efficient variant of PSO. Soft Comput. 23, 9701–9718 (2019) received M.Tech. and Ph.D.
60. Abdelaziz, A., Ali, E., Elazim, S.A.: Implementation of flower degrees in Control System
pollination algorithm for solving economic load dispatch and Engineering from National
combined economic emission dispatch problems in power sys- Institute of Technology Rour-
tems. Energy 101, 506–518 (2016) kela, India in 2010 and 2015
61. Zaoui, S., Belmadani, A.: Solution of combined economic and respectively. Since 2013, he has
emission dispatch problems of power systems without penalty. been with National Institute of
Appl. Artif. Intell. 36(1), 1976092 (2022) Technology Nagaland, India as
62. Trivedi, I.N., Purohit, S.N., Jangir, P., Bhoye, M.T.: Environment an Assistant Professor. His
dispatch of distributed energy resources in a microgrid using jaya research interests include time-
algorithm. In: 2016 2nd international conference on advances in delay system, control system, robotics, soft computing, machine
electrical, electronics, information, communication and bio-in- learning, meta-heuristic optimization techniques and application of
formatics (AEEICB), pp. 224–228. IEEE (2016) optimization techniques.
63. Mirjalili, S., Mirjalili, S.M., Lewis, A.: Grey wolf optimizer.
Adv. Eng. Softw. 69, 46–61 (2014) Siseyiekuo Khatsu received his
64. Basu, M.: Dynamic economic dispatch with demand-side man- B.E in Electrical and Electron-
agement incorporating renewable energy sources and pumped ics Engineering from Govern-
hydroelectric energy storage. Electr. Eng. 101(3), 877–893 ment College of Engineering
(2019) Salem, Tamil Nadu in the year
2018. He received his M.Tech
Publisher’s Note Springer Nature remains neutral with regard to in Electrical and Electronic
jurisdictional claims in published maps and institutional affiliations. Engineering form National
Institute of Technology, Naga-
Springer Nature or its licensor (e.g. a society or other partner) holds land. His research interest
exclusive rights to this article under a publishing agreement with the includes system optimization,
author(s) or other rightsholder(s); author self-archiving of the demand side management,
accepted manuscript version of this article is solely governed by the demand response management,
terms of such publishing agreement and applicable law. microgrids, renewable energy
sources, meta-heuristic opti-
mization schemes.

Abhishek Srivastava received his


B.Tech degree in Electrical and
Electronics Engineering from
Sam Higginbottom Institute of
Agriculture, Technology and
Sciences, Allahabad in 2013,
the M.Tech and Ph.D. degree in
Power System Engineering
from National Institute of
Technology Nagaland, India in
2018 and 2022 respectively. His
area of interest includes eco-
nomic load dispatch, microgrid,
meta-heuristic techniques, opti-
mization techniques.

123

You might also like