Electrical

Engineering
https://doi.org/10.1007/s00202-024-02518-1

O R I G I N A L PAPER

Frequency control of an islanded microgrid with multi-

stage PID control approach using moth-flame optimization
algorithm
Gotte Vikram Raju1 · Nandiraju Venkata Srikanth1

Received: 17 December 2023 / Accepted: 30 May 2024

© The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2024

Abstract
Microgrids (MGs) are installed with renewable energy sources (RES) to meet the dynamic load demands. The frequency
control of MG operating in an islanded mode is more difficult than in grid-connected mode. This is due to the low inertia
of MG system, intermittency in RES output powers, and stochastic loads. These factors amplify the frequency deviations
which affect the system’s operating frequency creating stability problems that may weaken the MG or result in a blackout
situation. In the present work, investigations on frequency control of an islanded Bella Coola MG having DEG, RES,
and PHEVs are presented. To stabilize the MG frequency oscillations during different system operating conditions, a
multi-stage PID controller is proposed whose parameters are optimized with the moth-flame optimization (MFO)
algorithm. Some popular meta-heuristic-based PID controllers (PSO-PID, TLBO-PID, and GOA-PID) are investigated for
a step load change to assess the effectiveness of MFO algorithm in optimizing the PID parameters. The MATLAB
simulation results have shown that the MFO-PID controller has shown better performance with improvements of at least
35.48%, 33.33%, and 18.18% in undershoot, overshoot, and settling time compared to other PID controllers. The robust
performance of the proposed MFO-optimized multi-stage controller is evaluated by considering the simultaneous power
fluctuations of RES and load along with system parametric uncertainties. The frequency deviations and integral time
absolute error (at least 24%) are reduced significantly during different disturbance conditions which are confirmed through
MATLAB/SIMULINK simulation results. The stability and comparison analyses prove that the proposed method works
efficiently.

Keywords Microgrid · Frequency control · Multi-stage PID controller · Moth-flame optimization · Renewable energy
sources · Parametric uncertainties

Abbreviations GWO Gray wolf optimization

HSA Harmony search algorithm
AGC Automatic generation control ITAE Integral time absolute
AI Artificial intelligence error LFC Load frequency
ANN Artificial neural network control MFO Moth-flame
CSA Cuckoo search optimization MG Microgrid
algorithm DEG Diesel engine MPID Multi-stage PID controller
generator ESS Energy storage MPPT Maximum power point
systems FFA Firefly algorithm tracking NFL No free lunch
FPD Filter-type proportional derivative PD Proportional derivative
FSA Future search algorithm PFC Primary frequency control
GA Genetic algorithm PHEVs Plug-in hybrid electric vehicles
GOA Grasshopper optimization algorithm PI Proportional–integral
PID Proportional–integral–derivative
B Gotte Vikram Raju POS Peak overshoot
[email protected] PSO Particle swarm
1 optimization PUS Peak
Electrical Engineering Department, National Institute
of Technology, Warangal, India undershoot
PV Photovoltaic
1
3
 Electrical
Engineering
RES Renewable energy system for only limited changes in the operating conditions.
source ST Settling time
SFC Secondary frequency
control SOC State of charge
SSO Social spider optimization
TLBO Teaching learning-based optimization
WTG Wind turbine generator

1 Introduction

The supply of power from traditional generating units to

distant areas and islands is expensive, unreliable, and dam-
aging to the environment under the current situation of
contemporary power systems [1]. To limit the fossil fuel
consumption and reduce carbon emissions for generating
electricity, the installation of RES is actively encouraged
in different nations. The risk of depletion of fossil fuels
and environmental effects like global warming is alarming
globally. Under these circumstances, microgrids (MGs) offer
reliable and cost-effective power solutions for supplying
the required excess power demand to remote areas.
Microgrid is a combination of diesel engine generators
(DEGs), loads, renewable energy sources (RES), and
various energy stor- age systems. An MG acts like a
single controllable unit that contains a group of distributed
energy sources to sup- ply the excess demand from the
different connected loads (residential, commercial, or
small industrial loads). It could be regarded as a small-
scale local grid with control capabil- ities at low-voltage
distribution levels [2].
In traditional power systems, the frequency regulation
job is simple as the disturbance arises due to the stochastic
loads only. But in an islanded MG; functional complexity,
vari- able structure, and diversity in the generation are some
of the intrinsic attributes. These attributes introduce rapidly
chang- ing operational points of the system [3]. Moreover,
the rapid growth in the increased penetration levels of RES
into MG will result in low inertia of the MG system. The
low inertia of system, stochastic loads, and
intermittent/discontinuous generation of renewables will
create complications in the frequency control of an
islanded MG. Massive frequency deviations will cause
stability and reliability problems and sometimes may lead
to MG blackout. Although the PID controllers are simple in
structure, reliable, and better at per- formance with fair
cost, the classical PI/PID controllers fail to provide a
satisfactory response for MG frequency con- trol under
these rapid changes in MG operating conditions [4]. The
manual tuning which is tedious and hectic task is the
general practice to tune PI/PID controllers. The param-
eters tuned for a particular operating point can control the

1
3
Electrical
Engineering
Therefore, an intelligent and robust controller is required parameters while TLBO is independent of algorithm-specific
to regulate the frequency of an islanded MG for wide parameters. In cases, where single opti- mization technique
range of operating conditions. cannot provide better performance of PID

1.1 Literature review and

research gaps

To address the job of LFC problem of an islanded MG,

var- ious artificial intelligent-based PI/PID controllers
have been proposed by different authors. The GA-PID
controller was proposed by Das et al. in [5], in [6], Shashi
et al. proposed the PSO-PID controller, in [7], Fergany et
al. proposed the SSO- PID controller, in [8], Sreenivas et
al. proposed GWO-PID controller, in [9], Ray et al.
proposed FFA-PID controller, in [10], Shankar et al.
proposed HSO-PID controller, in [11], Sreenivas et al.
proposed TLBO-PID controller, in [12], Anil et al.
proposed GOA-PID controller, and in [13], Latif et al.
proposed CSO-PID controller for MGs with various
RES and ESSs. Similarly, a combination of fuzzy and
var- ious optimization techniques is proposed for a
stand-alone MG frequency control [14–16]. In [14],
Bevarani et al. pro- posed PSO-optimized adaptive fuzzy
PI controller, in [15], Khooban et al. proposed a
modified HSA based adaptive fuzzy PI controller. A PID
control model for MG frequency control based on ANNs
was developed by Bevarani et al. in [17]. Similarly,
several authors have proposed a combination of ANN and
PID controllers for the MG frequency control. The GA
and PSO are the basic optimization algorithms that are
based on the survival of the fittest and swarm- ing
behavior of birds. The GA requires several generations of
evaluations to find the elite population through a com-
putationally complex process of involving GA operators
viz. selection, mutation, and cross of population. The
PSO is one of the simplest algorithms which find the
optimal solutions using the particles’ global and personal
best per- formances. However, this algorithm is highly
dependent on its algorithm-specific parameters viz. W,
C1, and C2 that decide the movement, personal, and
social performance of the particles. The SSO algorithm
can be viewed as a better version of ant colony
optimization. This algorithm is compu- tationally costly
and cannot offer a better trade-off between exploration
and exploitation of search space. The GWO algo- rithm is
based on the hunting behavior of wolves where the pack
of wolves are divided into four groups based on their
strength to hunt the prey. Although, this algorithm
provides better solutions compared to other algorithms, it
is sensitive to initial population as the next position is
decided by encir- cling the prey based on the distance
vectors of wolves. Both TLBO and GOA algorithms
have two stages that are used to explore and exploit the
search space but GOA depends on its algorithm-specific

1
3
 Electrical
Engineering
controller for MG systems with high nonlinearity and para- plus (1 + PI) and type-2 fuzzy cascaded to PI frac- tional
metric uncertainties, hybrid algorithms are also implemented order controllers are optimized with gray wolf and
for tuning the PID controller. For instance, in [9], firefly
and PSO algorithms are used for global and local search of
the variable space to counter the limitations of each
algorithm. Similarly, few authors have proposed PI/PID
controllers with fuzzy logic or ANN to have greater flexibility
of the controller for wide operating conditions of the system.
In fuzzy adaptive PI controllers, the scaling factors of
membership functions are optimized. In [16], PSO
algorithm is used for tuning the scaling factors of
membership functions, and GWO is used to optimize the
weights of the fuzzy logic rule base.
In real-time control studies, the AI-optimized PID con-
trollers are well suited for MG frequency control applica-
tions. However, for certain operating scenarios, it has been
observed that these PID controllers have underperformed due
to the difficulty in finding an optimal operating state, i.e.,
accounted for PID control structure. This
underperformance is mostly caused by an improper offset
between the derivative and integral gain components. A PI
controller can minimize the system’s steady-state error. If
the operator has increased the integral gain component, the
steady-state error is min- imized. However, the system
exhibits undesirable behavior during the transient state.
The system’s stability and speed are reduced by the
dominant integral component. Similarly, the dominant PD
component improves the system’s tran- sient response.
But, the system’s steady-state response is disturbed. To
overcome this problem and obtain optimal per- formance
from the PID controller, this work proposes an optimal
MPID controller for MG frequency control.
Recently, the concept of multi-stage controllers has
been implemented by several authors due to their robust
perfor- mance in regulating the frequency of an islanded
MG or multi-MG with RES and ESSs [18–23]. In [18],
Khadanga et al. proposed a multi-stage PID (MPID)
controller using the hybridization of two algorithms (whale
optimization and pattern search) to tune its parameters.
Bhuvnesh et al. pro- posed multi-stage fuzzy PD-TID
controller with chaotic crow search algorithm [19] and an
MPID cascade controller (PI- PD) with the help of salp
swarm optimization [20] to tune the controller parameters.
Anil et al. implemented the fuzzy- tuned MPID and
fractional order MPID controllers where its parameters are
optimally tuned by the grasshopper opti- mization [21] and
future search algorithms [22], respectively. The authors in
[23] presented an MPID controller optimized with an atom
search algorithm toward the multi-MG system. Similarly,
this multi-stage controller concept for frequency control is
also used in multi-area power systems contain- ing FACTS
devices, hydro, thermal, and gas power plants using multi-
verse optimization and grasshopper optimiza- tion
techniques [24–26]. The multi-stage controllers viz. FPD

1
3
Electrical
Engineering
crow search optimization algorithms [27, 28]. These con- powerful MFO is presented to optimize the MPID
trollers are implemented to get robust performance in the controller gain constants. Considering the above
automatic generation control or frequency control actions discussion, the present work investigates the frequency
of multi-source or RES penetrated multi-area power dynamics of an islanded MG comprising of DEG, PV,
systems. WTG, and PHEVs.
One of the most popular nature-inspired swarm intelli-
gence approaches is the moth-flame optimization (MFO)
methodology. It is straightforward to understand and easy
to implement. The MFO algorithm successfully solved
sev- eral optimization problems relating to different
domains viz. medical applications, marine engineering,
engineer- ing design, image processing, and electrical
power systems [29, 30]. The controller gain constants of
PI or PID are optimally refined using the MFO algorithm
presented in [31–35] for AGC/LFC of single-area/multi-
area power sys- tems with/without incorporating FACTS
devices. In wind energy conversion systems, it is utilized
to develop blade pitch controllers [36]. The MFO
algorithm is successfully implemented in the above-
mentioned works and has shown better performance
compared to other swarm intelligent tech- niques in those
works.
From the literature, it is understood that the
regulation of frequency of MG either having RES or ESS
in islanded mode is achieved with different controller
structures whose parameters are optimized with a variety
of swarm intelli- gent techniques. The performance of these
techniques mainly depends on the choice of specific
parameters of the algorithm, and the improper choice may
roll the solution toward local optimum values. The
controller’s approach based on fuzzy logic or fractional
order calculus is exhibiting better perfor- mance but at the
cost of optimizing the structure of the fuzzy inference
system or the fractional order system parameters along
with controller gain constants. The MPID controller is
preferred over a normal PID controller due to its benefits.
Also, from the literature, it was observed that controlling
the frequency deviations of an islanded MG having RES
and PHEVs with an MPID controller is rarely
investigated.

1.2 Motivation and

contribution

The selection of appropriate optimal

algorithms/techniques for modifying the control
parameters is crucial to improvise the MG’s dynamic
response. A perfectly optimized controller could enhance the
system’s performance in terms of low over- shoots, reduced
error values, and fast settling times. Also, the NFL theorem
states that no single meta-heuristic approach can solve all
engineering optimization issues, and there is always an
opportunity for improvement in the performance [37, 38].
Therefore, in the present work, a newly devel- oped and
1
3
 Electrical
Engineering
The coordinated control of DEG and PHEV with the MPID controller. Table 1 contains the parameters of the MG test
controller is studied using the MFO algorithm. system.
This paper’s main contributions are as follows: From Fig. 1, the generation–load balance equation is:

1. The performance of differently tuned PID controllers ∆P DEG + ∆P PH EV + ∆PWT G + ∆PPV = ∆PL (1)
(GOA, TLBO, PSO, and MFO) is demonstrated and
compared through the MATLAB/SIMULINK simulation The expression for deviation in the MG frequency (∆f )
results to show the superiority and feasibility of the MFO caused by the RES and load disturbances is:
algorithm over the other powerful and well-known 1
meta- heuristic techniques. ∆f = (∆PDEG + ∆PWT G + ∆PPV − ∆P PH EV
2. An MFO-tuned MPID controller has been proposed for Ms + D
−β∆f − ∆PL ) (2)
mitigating the oscillations in the frequency dynamics
of the modeled Bella Coola MG incorporating RES and The proposed coordinated control’s aim is to minimize the
PHEVs. The MFO algorithm is applied to optimize the frequency deviations of MG under critical operating
MPID controller’s gain constants. scenar- ios with the proposed MPID controller. In this
3. The MPID controller’s robustness is evaluated by work, only DEG and PHEVs are responsible for power
taking into consideration the simultaneous changes in balance. There- fore, the PV and WTG sources are
RES, load dynamics, and MG and PHEV uncertainties modeled as a disturbance in LFC analysis.
in a single controller framework.
4. The effect of PHEVs in the secondary and primary fre- 2.1 DEGs
quency control loops (SFC and PFC) of an islanded
MG has been analyzed with RES and load disturbances Since the power generation from the solar PV and WTG
to choose the better control action. sys- tems is stochastic, DEG forms the best means of
supplying reliable and quality power to the essential loads
in a stand- alone MG. A speed governing mechanism and
1.3 Formation of the paper diesel engine combination forms the diesel engine
generator system. The DEG mathematical model is given
Apart from the introduction in Sect. 1, the following four sec- in Fig. 2 [40]. The lim- its for GRC (generation rate
tions are formed to detail the contents of the paper. Section constraint) and GDB (governor dead band) ± for the DEG
2 depicts the modeling of stand-alone MG compatible with are considered as 10% per min and 0.06%.
LFC studies. The mathematical modeling of the MFO
algo- rithm and its application to the suggested MPID 2.2 WTGs
controller are shown in Sect. 3. Section 4 examines and
evaluates the performance of the proposed controller under The mechanically produced output power of a WTG is highly
diverse oper- ating scenarios. Section 5 summarizes the inconsistent since it depends on the stochastic wind speeds
outcomes of the proposed work. and is given by:

Pwp = 0.5ρ AVw3 C p (β, λ) (3)

2 Mathematical modeling where Pwp = power output of windmill, ρ = air density

of the investigated MG (kg/m3), V w = speed of wind (m/s), A = rotor swept area
(m2), and Cp = power coefficient (function of tip speed
Figure 1 shows the simplified form of the Bella Coola hybrid ratio, λ and blade pitch angle, β (deg)). R = radius of the
MG mathematical model under investigation. The present = and ω angular velocity of the blade (rad/s). The
blade (m)
model comprises DEG, PHEV aggregator, and RES (wind Cp can be expressed as follows:
and solar power output) [39]. This study considers WTG
and PV powers to be uncontrolled power sources (as dis- 25.52 −125
/γ
turbances). The advantages of using renewable energy are C p (β, λ) = γ − 1.1 − 0.088β exp
hindered, as frequency deviations limit the MPPT output ,
power when they participate in frequency control. So, in γ = 1 1 0.035 (4)
−
this
work, based on load changes and available PV/WTG powers, λ+0.08β 1+β3
the diesel engine generator and PHEVs manage the power Rω
balance in the MG that is achieved through the suggested λ=
Vw

1
3
Electrical
Engineering

Fig. 1 MG mathematical model

Table 1 MG system parameters

[22] Parameter Metric value Parameter Metric value Parameter Metric value

M (s) 0.1667 TPHEV (s) 0.1 NEV 600

D (puMW) 0.015 TWTG(s) 2 Rav(Hz/puMW) 2.4
T 1(s) 0.025 TPV (s) 1.8 Pmax(kW) 5
T 2(s) 2 KWTG 1 Δf u(Hz) 0.1
T 3(s) 3 KPV 1 Δf l (Hz) 0.1

Fig. 2 DEG mathematical model

The WTG linear model can be represented as follows [41]: The ∆PPV with respect to ∆ϕ can be computed using
∆P K Eq. (7):
WT G WT G
T FWT G = P = (5) Psolar
W Pout 1 + STWT G ∆PPV =
put
ϕ ST C
Figure 3 depicts the model for the generation of wind (∆ϕ + Kt [∆ϕTa + ϕ∆Ta + 0.0512ϕ∆ϕ − TST C ∆ϕ]) (7)
output power variations mathematically.
The PV system’s first-order model can be represented as
2.3 PV model follows [42]:

The extraction of electrical energy from photon energy ∆PPV K PV (8)

takes T FPV = =
place in the PV cells that are made from a semiconductor ∆ϕ 1 + sT P V
material. The possibility of the desired level of installation
capability of PV sources is the key benefit of Figure 4 depicts the mathematical model for PV output
photovoltaics. The power generated from a PV source power with random solar irradiation pattern generation
depends on the amount of solar irradiation and temperature. [20]. The power data used for the WTG power and PV
If 25 °C ambient temperature is maintained, thenPPV varies power can be found in [42].
linearly with the irradiation constant (ϕ) only [41].
2.4 PHEVs
ϕ
PPV Psolar (1+ Kt [Ta + 0.0256ϕ∆ϕ TST C Figure 5 illustrates the PHEV aggregator’s mathematical
]) ϕST C
model used for LFC studies. The controller’s command
= − (6)
sig- nal (∆Uc) is used to determine whether the ∆P PH EV
1
3
 Electrical
Engineering
will be

1
3
Electrical
Engineering
Fig. 3 Mathematical model for
wind output power variations

Fig. 4 Mathematical model for

PV output power with random
solar irradiation pattern
generation

Fig. 5 PHEV aggregator model

for frequency control studies

used for charging/discharging [16]. The slow discharge the KEV, i =

0. At a given instance, different EVs have dif-
rate, quick response time, and dispersed availability of ferent SOC of batteries. The EV’s participation factor may
PHEVs made them a great energy storage alternative for vary depending on the SOC of EV battery. An EV having
LFC needs. The output power from the PHEV aggregator considerable amount of SOC will have a better
based on fre- quency deviation is as follows (∆P PH EV ): participation in the frequency control. Figure 6 depicts the
⎧
⎪⎨ K E V , i ∆ f ; . K E V , i ∆ f . ≤ Pmax typical graphs for the participation of an EV in frequency
control based on the SOC of EV battery during discharge
and idle modes
Pmax; K EV , i ∆f > Pmax
∆P PH EV , i (9) ±
⎪⎩
= − Pmax ; K E V , i ∆ f < −
Pmax [43]. However, in load frequency control (LFC) studies, the
EVs having battery with an average SOC of 0.85 0.05 are
assumed to be more controllable to enhance the LFC per-
∆P PH EV , AG = N EV ∗ ∆P PH EV , i (10) state of charge (SOC) of battery. If an EV is disconnected
Pmax is the maximum power available from an from the aggregator, then
individual EV. KEV,i represents the single EV’s
participation gain fac- tor or availability to participate in
the frequency control. The KEV, i value depends on the EV’s

1
3
 Electrical
Engineering
formance of microgrid [44, 45]. This is because, if SOC of
battery is maintained nearly 100%, then life of battery decays
faster, and the SOC of battery is kept above 80% considering
EV owner’s comfort for the next trip. Therefore, an
average value of participation gain factor (KEV, i 0.55) is =
assumed in the present study that falls in the band of
average SOC of battery. NEV denotes the number of
EVs, Rav denotes the

1
3
Electrical
Engineering
Fig. 6 KEV,i vs. SOC a
Discharge mode and b IDLE
mode

droop characteristics of PHEV aggregator, and TEV , i is the This algorithm can be divided into two components, called
battery time constant. moths and flames, either of them is regarded to be the solu-
tion. The moths and flames are viewed as search agents
and elite positions explored during the iterations in this
algorithm. Thus, flames can be regarded as flags that moths
3 Moth-flame optimization-based MPID drop when exploring the search space. So that flames can
controller never lose their best position during the population update
and while finding the global best solution. Below is an
The escalating complexity of power systems requires a quick illustration of the math- ematical model used to replicate
and precise refining of the controller’s gain constants to the swarming behavior of moths:
achieve a better LFC response. To address this problem, many The moths are initialized as follows for the first iteration:
swarm intelligent strategies for PI/PID controller tuning ⎡ ⎤
have been suggested in the literature. The fundamental m1, 1 · · · m1, d
character- ⎢ . .. .
istics of these algorithms are the absence of derivatives and M= ⎣ (11)
. . .⎥
non-dependence on the plant model. A few of them are PSO, ⎦
m n,1 · · · mn, d
TLBO, GA, CSO, HSO, and GOA. So, for fine-tuning the
MPID controller, a recently developed and powerful MFO where M is the total population (no. of moths),d is the dimen-
is used. Mirjali et al. developed MFO in 2015, which imi- sion of search space which depends on the no. of optimization
tates the social behavior of moth flames. This algorithm variables, and n represents the total no. of moths.
was benchmarked on various standard test problems and Based on fitness function and its value, the moths can be
verified its performance quantitatively and qualitatively sorted in ascending/descending order as shown in Eq. (12)
with differ- ent optimization approaches in the literature.
This method was used extensively to solve different OM = (OM 1 , OM 2 · · · OM n ) T (12)
engineering problems because of its merits viz. quick
convergence, few controlling factors, straightforward Similarly, the flame matrix can be updated as:
implementation structure, and sim- plicity [29, 46, 47]. ⎡ ⎤
f1, 1 · · · f1, d
. .. . (13)
3.1 Mathematic ⎢=
F
. ⎥
al Modeling of ⎣ . . ⎦
MFO
moonlight. Once it comes too close to a light source, they fly
This algorithm mimics the flying characteristics of moths. spirally around it and finally reaches convergence after
Moths are flying insects that belong to the same class as certain modifications.
butterflies. The swarming behavior is a unique feature of
the moths that are found at both stages of moth’s evolution
(nymph/larval and adult stages). Moreover, moths exhibit
special navigation paths at night by nature called
transverse orientation. The moths follow a specific path
with respect to the moonlight during their travel in the
night. When moths are far away from any light, they travel
long distances in a straight line (single direction) based on

1
3
 Electrical
Engineering
f n,1 · · · fn, d

Based on fitness function and its value, the flames can

be sorted in ascending/descending order as shown in Eq.
(14).

OF = (O F1, O F2 · · · O Fn)T (14)

The typical structure of MFO can be expressed with

three approximations as follows:

MFO = (I , P, T ) (15)

where I: initialization, P: position update, and T : termination.

1
3
Electrical
Engineering
In Eq. (15), I denotes the initialization of the population, its input, and the PI controller in the final stage receives input
which can be defined as: from the output of the initial stage [18]. The MPID controller
output (ΔUc) is a reference power command signal for
Mi, j = lb(i) + randomno 1, N pop (ub(i) − lb(i)) (16) DEG and PHEVs. The key benefit of an MPID controller
is that the system can utilize the best features of both
where Npop denotes the population size, ub is the upper controllers. At first, PD controller is used produce constant
limit boundary, and lb is the lower limit boundary. output to improve the transient response, and later, the
In Eq. (15), P denotes the position update of moths. The steady-state error is reduced with the integral component of
position update of the moth with respect to flames can be the PI con- troller. So, in this controller, the integral term is
updated by using Eq. (17). absent during the transient state. This overcomes the
restriction of a tradi- tional PID controller, which includes
Mi = S Mi , Fj (17) the incorporation of an integral component during the
transient state.
The logarithmic spiral function is used as the key updat- The control signal fed to DEG and PHEV through the
ing strategy because the algorithm is based on the moth’s MPID controller can be expressed as [18]:
transverse orientation around a flame and can be written as
follows:
sN KI
∆Uc = KP + KD s + N 1 + KPP + ∆f (22)
S Mi , Fj = Fj + Di ebt cos(2π t) (18)

where Fj denotes the position of jth flame and Mi indicates 3.3 Application of MFO algorithm to
the position of ith moth, ‘t’ indicates a random number in optimize the MPID controller
[− 1, 1], and ‘b’ for defining the shape of the logarithmic
spiral. Di is the distance of the ith moth from the jth flame, The following steps are an explanation of the process for
that is defined as follows: optimizing the proposed the MPID controller with MFO:
Step 1: Random initialization In the first iteration, gen-
Di = . Fj − Mi. (19) erate a random population by using Eq. (11) and Eq. (16).
Since the proposed controller has (K P , K D , K P P , K I , N )
The moth’s position is updated with respect to flame as 5 controller parameters, the population size is taken to be
dur- ing exploration and exploitation [48]. The exploitation 100
is taken care by the logarithmic spiral function, and it takes × 5. Here, 100 stands for the population of moths and 5 for
place if the subsequent point is located between the moth the dimensional size of the search space.
and flame. Exploration is taken care by the distance term, Step 2: Objective/fitness assessment Simulate the devel-
Di. In any swarm intelligence algorithm, an optimal trade- oped model to evaluate the population’s performance with
off between exploitation and exploration is needed to the fitness function (Integral time absolute error (ITAE))
obtain global optimization. In this algorithm, the number given in Eq. (23).
of flames reduces from iteration to iteration to have equalized
effect on exploitation and exploration, as stated in Eq. (20). ∫tsim
IT AE = min t |∆f |dt (23)

flame_ no = 0
J−1
round J− (20)
T Subjected to the optimization of 0.01 ≤ K P , K P P ,K I ,
i
dem.≤Figure 8 illustrates
≤ the structure of an MPID controller.
i ≤ in the initial stage receives an error (∆f )
The PD controller
t = 1 − randomno 2 + T (21)
as
where J specifies maximum no. of flames, T is the maximum
count of iterations, and i is the present iteration. A
flowchart of MFO algorithm is shown in Fig. 7.

3.2 Multi-stage PID (MPID) controller

An MPID controller is a union of PD and PI controllers in tan-

1
3
 Electrical
Engineering
K D 2 and 0 N 200 [22]. Where ‘tsim’ denotes the
total simulation time. The ∆f value will be obtained from
the Simulink model shown in Fig. 1.
Step 3: Selection Based on the fitness value (ITAE),
flag the best positions by flames.
Step 4: Update the MFO algorithm-specific parameters
At each iteration, calculate the value of D for the
corresponding moth and update the number of flames,
and value of t using Eqs. (19) and (20).
Step 5: Population update Based on the values of D, F
and t estimate the updated position of moth with respect
to the best flame’s position using Eq. (18).

1
3
Electrical
Engineering

Fig. 7 Flowchart of MFO algorithm

Fig. 8 Mathematical model of

MPID controller

1
3
 Electrical
Engineering
Step 6: Stopping rule When the current iteration exceeds
the maximum iteration count, the optimal parameters for
the controller are the most optimal solution achieved till
the last iteration that corresponds to the optimal value of
ITAE. Show the optimized K P , K P P ,K I , K D , N, and ITAE
values.

4 Results and discussion

The simulation studies are performed on the Bella Coola MG

having different renewable energy sources and PHEVs that
are modeled in the MATLAB Simulink platform. This
study is implemented in MATLAB R2020a version
Fig. 9 ITAE performance characteristics
software, and Intel(R) Core(TM) i5-10210U CPU @ 2.11-
GHz processor with 8-GB RAM Windows 11 operating
system is utilized for the present work. An extensive Table 2 Optimized gains of various
investigation of the sim- ulation outcomes of an islanded
MG’s frequency deviations is provided in this section. The controllers Methods Optimized Gains
MG frequency deviations are
analyzed with variations in the load (∆PL ), RES power dis-
turbances (solar power (∆Pϕ), and wind power (∆PWT G)) KP KI KD N
along with the parametric uncertainties of MG and PHEV.
The performance of the MFO-optimized MPID controller PSO-PID 1.0329 1.6248 1.1992 100.2266
is presented for the aforementioned disturbances. Initially, TLBO-PID 1.1922 0.8718 1.8374 119.4526
to examine the supremacy of the MFO algorithm over GOA-PID 1.39556 1.42816 1.90277 59.623
other popular meta-heuristic algorithms (PSO, TLBO, and MFO-PID 1.8230 1.5632 1.7258 28.5346
GOA), the performance of different PID controllers is
assessed with step load and multiple step load
perturbations. Later, the performance of MFO-optimized
PID and MPID controllers is evaluated for several
operating scenarios that have been created and tested on
Bella Coola MG. Further, the MG fre- quency deviations
are analyzed with the presence of PHEVs in primary and
secondary frequency control loops is pre- sented. Finally,
the stability analysis and comparison results are presented.

4.1 Initial case: Dynamic response evaluation

of various meta-heuristic-based PID
controllers against step load and multi-step
load disturbances (1 P L)
controllers. Table 3 represents the comparative performance
The initial case is meant to demonstrate the effectiveness
of MFO-based PID controller above the other well-known
meta-heuristic optimized PID controllers in the literature.
In the present operating scenario, a 10% step change in
load demand is considered, and the PID controllers’
parameters are optimized using the meta-heuristic
algorithms consider- ing the ITAE as the fitness function.
The ITAE characteristics of differently tuned PID
controllers are shown in Fig. 9. Table 2 denotes the
optimized parameters of PID controller with various meta-
heuristic techniques. Figure 10 displays the frequency
perturbation response of the MG for the oper- ating
conditions of step load change case using various PID
1
3
Electrical
Engineering
Fig. 10 MG frequency perturbation response for 10% step load
change

analysis of differently tuned PID controllers, considering

Fig. 10 with regard to ITAE value, settling time, and
over- shoot/undershoots.
From the above results, it is evident that the elite
perfor- mance of the MFO-PID is visible from MG dynamic
response following a step load perturbation. The MFO-
PID controller minimizes the frequency fluctuations and
improves the set- tling times far better than other PID
controllers. Further, multi-step load disturbances depicted
in Fig. 11a are con- sidered to compare the MG
frequency deviation response (Fig. 11b) of the differently
tuned PID controllers. In this case

1
3
 Electrical
Engineering
Table 3 Comparative analysis of
differently optimized PID Optimizing Performance analysis terms
controllers for step load change method-PID
case of operating conditions Peak undershoot Peak overshoot Settling time (s) Integral time
(Hz) (Hz) absolute error

PSO-PID — 0.12 0.1 21 0.000043

TLBO-PID — 0.095 0.05 13 0.000038
GOA-PID — 0.093 0.045 11 0.000032
MFO-PID — 0.06 0.03 9 0.000021

Fig. 11 Dynamic response of various controllers: a multi-step load disturbances in MG and b frequency perturbation response of MG

also, the MFO-PID controller offers superior performance

than the other controllers. It is evident that the adopted
MFO algorithm is well suited for this problem; hence, the
MFO algorithm is employed to optimize the proposed
MPID’s gain constants to have better control of MG
frequency deviations. From the next case onwards, the
proposed MFO-MPID con- troller and the MFO-PID
controller are contrasted to provide a better perspective of
the results view.
Case-1 Dynamic response evaluation of MFO-
optimized multi-stage PID controller considering multiple
power dis- turbances (concurrent changes in ∆PL, ∆Pϕ, and
∆PWT G ) and parametric uncertainties of MG and PHEV.
This case is meant to demonstrate the effectiveness of
multi-stage PID controller over PID controller using MFO Fig. 12 Concurrent power fluctuations (multiple disturbances) in MG
algorithm. The parameters of MPID and PID controllers
are optimized considering the simultaneous disturbances in
The purpose of this case is to examine how well an
load, solar, and wind powers along with parametric
MPID controller outperforms the PID controller in the con-
uncertainties of MG and PHEV using ITAE as the fitness
text of various disturbances, MG, and PHEV uncertainties.
function. Figure 12 shows the multiple power fluctuations,
Figure 14 depicts the frequency perturbation response of both
and Table 4 presents the percentage of parametric
the controllers in this context. It can be observed that there
uncertainties considered for the MG and PHEV aggregator.
is improved performance of the proposed multi-stage PID
Figure 13 shows the ITAE characteristics of the proposed
con- troller than the PID in minimizing the frequency
MFO-MPID and MFO-PID controllers. Table 5 presents the
deviations of the MG.
optimal gain constants of the two controllers.

1
3
Electrical
Engineering
Table 4 Percentage uncertainties in MG and PHEVs

Parameters Percentage variation Actual

KEV − 30% 1
Rav(Hz/puMW) − 50% 2.4
TEV (s) + 50% 0.1
M (s) − 20% 0.1667
D (puMW) − 20% 0.015
R (Hz/puMW) + 20% 2.4

Fig. 15 MG frequency perturbation response for multiple power distur-

bances

In case-2, the changes in load, solar, and wind power

dis- turbances are considered concurrently in MG. The
concurrent power changes in MG are depicted in Fig. 12,
and the MG frequency perturbation response is shown in
Fig. 15.
Case-3 Response evaluation of the proposed controller
under solar power disturbances (∆Pϕ).
In case-3, only solar disturbances are considered in MG.
Figure 16a depicts the random change in solar irradiation
power, and the respective MG frequency deviations are
Fig. 13 MFO-PID and -MPID controllers’ ITAE performance
evalua- tion with case-1 conditions shown in Fig. 16b.
Case-4 Response evaluation of the proposed controller
to wind power fluctuations (∆PWTG).
Table 5 MFO PID & MPID controllers’ optimized
For this case-4, only ∆PWT Gis considered in MG. The
gains ControllerController parameters random fluctuations in wind power are shown in Fig. 17a
and b which depicts the corresponding response of frequency
deviation in the MG.
From the results of the above of the cases-1, 2, 3, and
KP KI KPP KD N 4, it is understood that the proposed MFO-optimized multi-
stage PID controller is performing better in mitigating the
MFO-PID 1.1334 1.3196 – 1.9169 144.3923 MG frequency deviations than the PID controller. Table 6
is
MPID presented the
for controllers
the dynamic
MFO- 1.4922 1.6656 0.0774 2 134.2390
in performance
terms of peakevaluation of peak
undershoot, both
overshoot, and ITAE value. Table 6 result conveys that
the suggested MFO-MPID controller enhances the
performance of MG over the MFO-PID controller when
subjected to RES/load disturbances. The case-1 result also
reveals that the suggested controller showed a good level of
resilience toward the con- sidered parametric uncertainties.
Also, it is clear from ITAE performance characteristics that,
in comparison with the PID controller, the ITAE value is
still reduced with proper use of the PID controller in
multi-stages. Therefore, the MFO- MPID controller can be
utilized for the control of modern power systems
Fig. 14 Frequency perturbation response of MG against multiple power incorporating high renewable energy content. Case-5 Impact
disturbances and parametric uncertainties of PHEVs in the SFC and PFC loops (pri-
mary frequency control).
Case-2 Response evaluation of the proposed controller
This case objective is to investigate the impact of
under multiple power disturbances (concurrent changes in
PHEVs on the dynamic frequency control in the SFC loop
∆PL, ∆Pϕ, and ∆PWT G).
with a control signal from the MPID controller versus the
1
3
 Electrical
Engineering
impact of PHEVs in the PFC loop. This can be
illustrated with

1
3
Electrical
Engineering

Fig. 16 a Fluctuations of solar power in MG and b MG frequency perturbation response for case-3

Fig. 17 a Fluctuations of wind power in MG and b MG frequency perturbation response for case-4

Table 6 Dynamic performance

evaluation of MFO PID & MPID Cases Peak undershoot (Hz) Peak overshoot (Hz) ITAE
controllers against the cases 1, 2,
3, and 4 operating conditions MFO-PID MFO MPID MFO-PID MFO MPID MFO-PID MFO MPID

Case-1 — 0.1 — 0.0357 0.09 0.032 0.0003623 0.0002384

Case-2 — 0.058 — 0.035 0.056 0.03 0.000283 0.000212
Case-3 — 0.055 — 0.032 0.052 0.028 0.000250 0.000190
Case-4 — 0.039 — 0.011 0.042 0.012 0.000216 0.000152

the help of PHEV aggregator model (in Fig. 5). In the

suggested method, the control signal for PHEV aggregator
output power is obtained from the MFO-optimized MPID
controller. Whereas in the PFC loop, there will be no
control signal for PHEV output power, and the PPHEV is
obtained from droop characteristics of the PHEV
aggregator (Rav). The MG frequency perturbation response
under case-5 con- ditions is depicted in Fig. 18. The input
disturbances for this case are same as case-1 conditions. As
shown in Fig. 18, with the suggested approach, the MG
frequency deviations are significantly reduced over the
PHEV aggregator in the
Fig. 18 Impact of PHEV aggregator due to proposed method on the
PFC loop. This due to proper coordination is established by frequency response of MG over the generalized methods in the literature
[43]

1
3
 Electrical
Engineering
the proposed controller between DEGs and PHEVs in the Table 7 Eigenvalues of MG system for different operating conditions
SFC loop.
Case-1 Cases-2, 3, 4
4.2 Observations from
— 69.50 + j2949.60 — 71.20 + j3231.40
simulation results
— 69.50 + j2949.60 — 71.20 − j3231.40
The following observations have been noticed from the var- — 40 + j0 — 40 + j0
ious operating cases. — 1 + j0.40 — 1 + j0.40
— 1 − j0.40 — 1 − j0.40
1. From initial case conditions, it is noticed that the MFO- — 0.30 + j0 — 0.30 + j0
PID controller exhibits an improvement in undershoot — 0.50 + j0 − 0.50 + j0
(50%, 36.84%, and 35.48%), overshoot (70%, 40%,
and 33.33%), and settling times (57.14%, 30.77%, and
18.18%) over PSO-PID, TLBO-PID, and GOA-PID s-plane thereby confirming the stable operation of the
controllers, respectively. The above quantitative results MG system. If the gain of the linearized transfer
confirm that the proposed MFO algorithm is precisely func- tion model of MG system is greater than 4.3 (k
adaptable to the present MG frequency control > 4.3), then two closed loop poles traverse to the
problem. positive half of complex s-plane making the system
2. Upon comparison with the MFO-PID controller, it has unstable which can be seen in the zoomed view of
been noticed that the suggested MFO-MPID controller Fig. 19.
reduces the ITAE value by at least 24% for the
operating conditions of cases 1, 2, 3, and 4.
4.4 Comparison with earlier
3. Finally, from the conditions of case-5, it is observed
reported works
that the MG frequency response is improved with the
pres- ence of PHEVs in the SFC loop (DEGs and
The comparison of the suggested approach with ear-
PHEVs are well coordinated with an MPID controller)
lier reported works for islanded MG frequency
rather than in the PFC loop.
control employing an MPID controller regarding
settling time, ITAE, and controller optimization
technique is pre- sented in Table 8. Table 8 reveals
4.3 Stability analysis of MG
that the performance of the proposed controller is
better and nearly on par with the other reported
The linear analysis tool from the Simulink Control
methods, demonstrating opti- mal results in both
Design toolbox of MATLAB/Simulink is used in the
ITAE value and settling time. A qualitative
present work to demonstrate the stability analysis of
comparison on computationally burden can be
the MG for various operating situations together with
assessed as high/low depending on the count of
root locus plots [19] and eigenvalues of the system’s
controller parameters optimized, the count of optimiza-
lin- earized transfer function model [20]. Table 7
tion techniques employed, and their algorithm-specific
displays the eigenvalues for various cases. Figure 19
parameters. The computational burden required for
displays the root locus plots for cases 1 and 2. It can
optimizing more controller parameters with one or
be observed that from the eigenvalues and root locus
more techniques would be comparatively more.
plots, all the closed loop poles lie to the left half of
the complex

1
3
Electrical
Engineering

Fig. 19 Root locus plots for case-1 and case-2

5 Conclusion extended level of robustness for MG and PHEV paramet-

ric uncertainties. Furthermore, the proposed MFO
In this paper, an MFO-optimized MPID controller is pro- algorithm had shown its supremacy over the well-established
posed to regulate/control the MG frequency deviations optimiza- tion techniques in the literature viz. GOA, TLBO,
during different disturbance conditions of an islanded Bella and PSO. The MFO-PID controller has shown a better
Coola MG comprising of DEG, RES (wind and solar), and performance with an improvement in undershoot (50%,
PHEVs. The parameters and performance of the proposed 36.84%, and 35.48%), overshoot (70%, 40%, and
controller are evaluated through MATLAB simulations by 33.33%), and settling
subjecting the stand-alone MG to different disturbance times (571.14%, 30.77%, and 18.18%) over PSO, TLBO,
con- ditions (viz. simultaneous power perturbations of RES and GOA tuned PID controllers. On the other hand, the
and load) along with system parametric uncertainties. The MG frequency deviations are examined with the impact of
MG frequency deviations and ITAE value of at least 24% PHEVs in the PFC and LFC loops. It is found that MG
are predominantly reduced for different operating attains bet- ter frequency regulation when PHEVs are
scenarios of the system with the proposed multi-stage present in SFC loops, i.e., PHEVs and DEGs are well
controller over MFO-PID controller, i.e., confirmed coordinated through the MFO-MPID controller. Further,
through the simulation results. Moreover, the proposed the stability of the MG is analyzed root locus plots and
MFO-optimized MPID con- troller offers enhanced eigenvalues. Therefore, the suggested controller stands as a
frequency dynamic response (in terms of less viable solution for dynamic frequency control applications
under/overshoots and fast settling times ) and an in MG.

1
3
 Electrical
Engineering
Table 8 Comparison with earlier reported works

Comparison term Reference

[19] [21] [22] Proposed

Multiple-step load – 0.000596, {5 s} – , {4 s} 0.000190,

perturbation
(ITAE, response
{ST (s)}) {5 s}
Simultaneous changes in
RES power, load, and 0.7276 0.000135492 0.006 0.0002384
parametric uncertainties
(ITAE)
Controller and optimization MPID and hybrid whale Fuzzy-tuned MPID and Adaptive fuzzy-tuned MPID and
techniques optimization and grasshopper optimization fractional order MFO
pattern search algorithm MPID and future
algorithms search algorithm
No. of controller parameters 4 11 35 5
optimized
No. of algorithm-specific 5 3 0 1
parameters
PHEVs considered (Yes/No) No No No Yes
Stability analysis (Yes/No) No No No Yes

Appendix Declarations
Optimization data Conflict of interest The authors declare that they have no
competing interests.

S. No. Optimization algorithm Data

1 MFO [48] No. of moths = 100, No. of iterations = 50

2 GOA [49] Population size = 100, No. of iterations = 50, Adaptation factor (c) =
[0.0001 – 4], Attraction factor (f) = 0.5, Attractive length scale (l) = 1.5
3 TLBO [50] Population size = 100, No. of iterations = 50, TF = 2−1
4 PSO [6] Population size = 100, No. of iterations = 50, Inertia weight (w) = 0.65,
Cognitive factors (C1, C2) = 2

References
PHEV data
1. Dekker J, Nthontho M, Chowdhury S, Chowdhury SP (2012)
Eco- nomic analysis of PV/diesel hybrid power systems in
SOC 1 = 0.1, SOC 2= 0.2, SOC 3=0.8, SOC 4=0.9, N EV
different climatic zones of South Africa. Int J Electr Power
= 600, T PH EV =0.1 s, Rav=2.4 Hz/puMW, PMax (Individual Energy Syst 40(1):104–112.
EV) = 5 kw, ∆f L = 0.1 Hz, ∆f U = 0.1 Hz. https://doi.org/10.1016/j.ijepes.2012.02.010
2. "IEEE Guide for Design, Operation, and Integration of
Author Contributions Gotte Vikram Raju helped in Distributed Resource Island Systems with Electric Power
conceptualiza- tion, methodology, software, investigation, and writing Systems", In: IEEE Std 1547.4–2011, pp. 1–54, 20 July 2011.
—original draft preparation and Nandiraju Venakata Srikanth worked 3. Obaid ZA, Cipcigan LM, Abrahim L, Muhssin MT (2019) Fre-
in supervision, conceptualization, validation, and writing—reviewing quency control of future power systems: reviewing and
and editing. evaluating challenges and new control methods. J Modern Power
Syst Clean Energy. 7(1):9–25. https://doi.org/10.1007/s40565-
Funding Not applicable. 018-0441-1
4. Bevrani H, Feizi MR, Ataee S (2016) Robust frequency control
Data availability There are no new data generated in this work. in an islanded microgrid: H∞ and µ synthesis approaches. IEEE
What- ever data used are available openly and cited properly. Transact Smart Grid 7(2):706–717. https://doi.org/10.1109/TSG.
Moreover, the new findings in this work are kept openly. 2015.2446984
5. Das DC, Roy AK, Sinha N (2012) GA based frequency
1
3
Electrical
Engineering
controller for solar thermal–diesel–wind hybrid energy
generation/energy

1
3
 Electrical
Engineering
storage system. Int J Electr Power Energy Syst 43(1):262–279.
Syst 48(6–7):711–726. https://doi.org/10.1080/15325008.2020.17
https://doi.org/10.1016/j.ijepes.2012.05.025
97936
6. Pandey SK, Mohanty SR, Kishor N, Catalão JP (2014)
21. Annamraju A, Nandiraju S (2020) A novel fuzzy tuned
Frequency regulation in hybrid power systems particle swarm
multistage PID approach for frequency dynamics control in an
optimization and linear matrix inequalities based robust
islanded micro- grid. Int Transact Electr Energy Syst
controller design. Int J Electr Power Energy Syst 63:887–900.
30(12):e12674. https://doi. org/10.1002/2050-7038.12674
https://doi.org/10.1016/j.ij epes.2014.06.062
22. Annamraju A, Bhukya L, Nandiraju S (2021) Robust frequency
7. El-Fergany AA, El-Hameed MA (2017) Efficient frequency con-
control in a standalone microgrid: an adaptive fuzzy based frac-
trollers for autonomous two-area hybrid microgrid system using
tional order cascade PD-PI approach. Advanced Control for Appl
social-spider optimiser. IET Gener Transm Distrib 11(3):637–648.
Eng Industrial Syst 3(3):1–18. https://doi.org/10.1002/adc2.72
https://doi.org/10.1049/iet-gtd.2016.0455
23. Mishra S, Nayak PC, Prusty RC, Panda S (2022) Performance anal-
8. Sreenivasaratnam C, Yammani C, Maheswarapu S (2019) Load
ysis of multistage PID controller for frequency regulation of
frequency control of multi-microgrid system considering renew-
multi microgrid system using atom search optimisation. Int J
able energy sources using grey wolf optimization. Smart Sci
Ambient Energy 43(1):7371–7386.
7(3):198–217. https://doi.org/10.1080/23080477.2019.1630057
https://doi.org/10.1080/01430750.2022. 206318
9. Ray PK, Mohanty A (2019) A robust firefly–swarm hybrid opti-
24. Nayak PC, Prusty RC, Panda S (2021) Grasshopper optimization
mization for frequency control in wind/PV/FC based microgrid.
algorithm optimized multistage controller for automatic genera-
Appl Soft Comput 85:105823. https://doi.org/10.1016/j.asoc.2019.
tion control of a power system with FACTS devices. Prot
105823
Control Mod Power Syst 6(8):2021.
10. Shankar G, Mukherjee V (2016) Load frequency control of an
https://doi.org/10.1186/s41601-021- 00187-x
autonomous hybrid power system by quasi oppositional harmony
25. Sahoo PK, Mohapatra S, Gupta DK, Panda S (2022) Multi verse
search algorithm. Int J Electr Power Energy Syst 78:715–734.
optimized fractional order PDPI controller for load frequency
https://doi.org/10.1016/j.ijepes.2015.11.091
con- trol. IETE J Res 68(5):3302–3315.
11. SC, C Yammani and S Maheswarapu (2019) Frequency con-
https://doi.org/10.1080/0377 2063.2020.175693
trol of Autonomous Hybrid Multi-Microgrid System. In: 2019
26. DE Abah, AS Abubakar, GS Shehu (2022) Load frequency
National Power Electronics Conference (NPEC), Tiruchirappalli,
control of two-area interconnected network using multi-stage
India, 2019, pp. 1-6. https://doi.org/10.1109/NPEC47332.2019.90
con- troller. Nigerian J Eng, 29(1), ISSN (print): 0794 – 4756,
34751.
ISSN (online):2705–3954.
12. A Annamraju, S Nandiraju (2018) Frequency Control in an
27. Sahu PC, Prusty RC, Panda S (2019) A gray wolf optimized FPD
Autonomous Two-area Hybrid Microgrid using Grasshopper Opti-
plus (1+PI) multistage controller for AGC of multisource non-
mization based Robust PID Controller.In: 2018 8th IEEE India
linear power system. World J Eng 16(1):1–13. https://doi.org/10.
International Conference on Power Electronics (IICPE),
1108/WJE-05-2018-0154
JAIPUR, India, 2018, pp. 1–6.
28. Sahu PC, Samantaray SR (2023) Resilient frequency stability of
https://doi.org/10.1109/IICPE.2018.8709428.
a PV/wind penetrated complex power system with CSA tuned
13. Latif A, Pramanik A, Das DC, Hussain I, Ranjan S (2018) Plug-
robust Type-2 fuzzy cascade PIF Controller. Electr Power Syst Res
in hybrid vehicle-wind-diesel autonomous hybrid power system:
225:109815. https://doi.org/10.1016/j.epsr.2023.109815
frequency control using FA and CSA optimized controller. Inter-
29. Mirjalili S (2015) Moth-flame optimization algorithm: a
national J Syst Assurance Eng Manag 9(5):1147–1158. https://doi.
novel nature-inspired heuristic paradigm. Knowledge-Based Syst
org/10.1007/s13198-018-0721-1
89:228–249. https://doi.org/10.1016/j.knosys.2015.07.006
14. Bevrani H, Habibi F, Babahajyani P, Watanabe M, Mitani Y
30. Sahoo SK, Saha AK, Ezugwu AE, Agushaka JO, Abuhaija B,
(2012) Intelligent frequency control in an AC microgrid: online
Alsoud AR, Abualigah L (2023) Moth flame optimization:
PSO based fuzzy tuning approach. IEEE Transact Smart Grid
theory, modifications, hybridizations, and applications. Arch
3(4):1935–1944. https://doi.org/10.1109/TSG.2012.2196806
Computa- tional Methods Eng 30:391–426.
15. Khooban MH, Niknam T, Blaabjerg F et al (2016) A robust
https://doi.org/10.1007/s11831- 022-09801-z
adaptive load frequency control for microgrids. ISA Transact
31. Acharyulu BVS, Mohanty B, Hota PK (2019) Comparative
65:220–229. https://doi.org/10.1016/j.isatra.2016.07.002
perfor- mance analysis of pid controller with filter for automatic
16. Annamraju A, Nandiraju S (2018) Robust frequency control in
generation control with moth-flame optimization algorithm. In:
an autonomous microgrid: a two-stage adaptive fuzzy approach.
Malik H, Srivastava S, Sood Y, Ahmad A (eds) Applications of
Electr Power Components Syst 46(1):83–94.
artificial intelligence techniques in engineering advances in
https://doi.org/10.1080/15 325008.2018.1432723
intelligent sys- tems and computing. Springer, Singapore
17. H. Bevrani, F. Habibi, S. Shokoohi (2013) ANN-based self-tuning
32. Lal DK, Barisal AK (2019) Combined load frequency and
frequency control design for an isolated microgrid. Meta-Heuristics
terminal voltage control of power systems using moth fame
Optimization Algorithms in Engineering, Business, Economics,
optimization algorithm. J Electr Syst Inf Technol 6(8):2019.
and Finance. P. Vasant (Ed), Chapter 12, IGI Global, USA,
https://doi.org/10. 1186/s43067-019-0010-3
pp. 357–385, 2013.
33. D Mohanty ,S Panda (2021) Moth flame optimization based frac-
18. Khadanga RK, Padhy S, Panda S, Kumar A (2018) (2018)
tional order controller for frequency regulation of hybrid power
Design and analysis of multi-stage PID controller for frequency
system with UC and UPFC. In: 1st International Conference on
control in an islanded micro-grid using a novel hybrid whale
Power Electronics and Energy (ICPEE), Bhubaneswar, India,
optimization- pattern search algorithm. Int J Numerical
2021, pp. 1–5,
Modelling 31(5):e2349. https://doi.org/10.1002/jnm.2349
https://doi.org/10.1109/ICPEE50452.2021.9358618.
19. Khokhar B, Dahiya S, Parmar KPS (2021) (2021) A novel
34. Peddakapu K, Mohamed MR, Srinivasarao P, Leung PK (2022)
hybrid fuzzy PD-TID controller for load frequency control of a
Simultaneous controllers for stabilizing the frequency changes in
standalone microgrid. Arab J Sci Eng 46:1053–1065.
deregulated power system using moth flame optimization.
https://doi.org/10.1007/ s13369-020-04761-7
Sustain Energy Technol Assess 51:101916.
20. Khokhar B, Dahiya S, Singh Parmar KP (2020) A robust cascade
https://doi.org/10.1016/j.seta. 2021.101916
controller for load frequency control of a standalone micro-
35. Hameed JBS, Mahabuba A, Jayasree R (2023) A novel LA-MFO
grid incorporating electric vehicles. Electr Power Components
technique based vehicle to grid control mechanism for automatic

1
3
Electrical
Engineering
generation control in deregulated power system. Int J Intell Syst
45. Khokhar B, Parmar KS (2023) Utilizing diverse mix of energy
Appl Eng 11(2):772–783
storage for LFC performance enhancement of a microgrid: a
36. Ebrahim MA, Becherif M, Abdelaziz AY (2018) Dynamic per-
novel MPC approach. Appl Energy 333:120639.
formance enhancement for wind energy conversion system using
https://doi.org/10.1016/ j.apenergy.2023.120639
moth-flame optimization based blade pitch controller. Sustain
46. Debbarma S, Dutta A (2017) Utilizing electric vehicles for LFC
Energy Technol Assess 27:206–212.
in restructured power systems using fractional order controller.
https://doi.org/10.1016/j.seta. 2018.04.012
IEEE Trans Smart Grid 8(6):2554–2564.
37. Khokar B, Dahiya S, Parmar KPS (2021) A novel fractional
https://doi.org/10.1109/TSG. 2016.2527821
order proportional integral derivative plus second-order deriva-
47. Huang L, Yang B, Zhang X, Yin L (2019) Optimal power
tive controller for load frequency control. Int J Sustain Energy
tracking of doubly fed induction generator-based wind turbine
40(3):235–252. https://doi.org/10.1080/14786451.2020.1803861
using swarm moth–flame optimizer. Trans Inst Meas Control
38. Wolpert DH, Macready WG (1997) No free lunch theorems for
41(6):1491–1503. https://doi.org/10.1177/0142331217712091
optimization. IEEE Trans Evol Comput 1(1):67–82.
48. Reddy S, Panwar L, Panigrahi B, Kumar R (2018) Solution to
https://doi.org/ 10.1109/4235.585893
unit commitment in power system operation planning using binary
39. AH Hajimiragha and MRD Zadeh (2013) Research and develop-
coded moth flame optimization algorithm: a flame selection
ment of a microgrid control and monitoring system for the
based computational technique. J Comput Sci 25:298–317.
remote community of Bella Coola: challenges, solutions,
https://doi. org/10.1016/j.jocs.2017.04.011
achievements and lessons learned. In: 2013 IEEE International
49. Saremi S, Mirjalili S, Lewis A (2017) Grasshopper optimisa-
Conference on Smart Energy Grid Engineering (SEGE), pp. 1–6.
tion algorithm: theory and application. Advances Eng Software
https://doi.org/ 10.1109/SEGE.2013.6707898.
105:30–47. https://doi.org/10.1016/j.advengsoft.2017.01.004
40. Papathanassiou SA, Papadopoulos MP (2001) Dynamic char-
50. Savsani P, Jhala RL, Savsani VJ (2016) Comparative study of
acteristics of autonomous wind-diesel systems. Renew Energy
dif- ferent metaheuristics for the trajectory planning of a robotic
23(2):293–311. https://doi.org/10.1016/S0960-1481(00)00180-4
arm. IEEE Syst J 10(2):697–708.
41. Dong-Jing L, Li W (2008) Small-signal stability analysis
https://doi.org/10.1109/JSYST.2014. 2342292
of an autonomous hybrid renewable energy power genera-
tion/energy storage system part I: time domain simulations. IEEE
Trans Energy Convers 23(1):311–320. https://doi.org/10.1109/
TEC.2007.914309 Publisher’s Note Springer Nature remains neutral with regard to
42. El-Hameed MA, El-Fergany AA (2016) Water cycle algorithm- juris- dictional claims in published maps and institutional affiliations.
based load frequency controller for interconnected power sys-
tems comprising non-linearity. IET Gener Transm Distrib Springer Nature or its licensor (e.g. a society or other partner) holds
10(15):3950–3961. https://doi.org/10.1049/iet-gtd.2016.0699 exclusive rights to this article under a publishing agreement with the
43. Izadkhast S, Garcia-Gonzalez P, Frías P (2015) An aggregate author(s) or other rightsholder(s); author self-archiving of the
model of plug-in electric vehicles for primary frequency control. accepted manuscript version of this article is solely governed by the
IEEE Trans Power Syst 30(3):1475–1482. https://doi.org/10.1109/ terms of such publishing agreement and applicable law.
TPWRS.2014.2337373
44. Khokhar B, Parmar KPS (2022) A novel adaptive intelligent
MPC scheme for frequency stabilization of a microgrid
considering SoC control of EVs. Appl Energy 309:118423.
https://doi.org/10.1016/ j.apenergy.2021.118423

1
3