Electrical Engineering

https://doi.org/10.1007/s00202-023-02144-3

ORIGINAL PAPER

Enhancing load frequency control with plug-in electric vehicle

integration in non-reheat thermal power systems
Rakesh Rajan Shukla1 · Man Mohan Garg2 · Anup Kumar Panda3 · Debapriya Das4

Received: 10 August 2023 / Accepted: 9 November 2023

© The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2023

Abstract
As the percentage of renewable energy source within the energy production mix has expanded, it is become gradually difficult
to determine the appropriate values for controller gains and model parameters. Controlling system frequency of interconnected
power system during sudden load disturbance requires careful consideration of the controller gain values and small signal
stability model parameters. This study proposes an innovative methodology for ascertaining controller gain values and model
parameters for plug-in electric vehicles (PEVs) in load frequency control (LFC) applications. Proposed method uses the
root locus (RL) approach to find the suitable controller gain values and model parameters for PEVs. This paper offers a
thorough mathematical description of the proposed RL approach. Routh approximation method is used for reduced-order
modelling (ROM), which comprises thermal and PEV systems, to reduce complexity of higher-order system while designing
controllers. Fractional-order proportional-integral-derivative controller (FO-PID) is proposed, and its parameters are adjusted
using particle swarm optimization (PSO) tool. To validate the efficacy of suggested method, a comprehensive comparison of
time response parameters and performance indices (PI) is carefully carried out. Also, the various PEVs state of charge (SOC)
levels is investigated, and effects of these levels are studied in LFC with robustness analysis of controller. The proposed
method for determining gain value is highly reliable and efficient, outperforming existing methodologies in the literature.

Keywords Fractional-order PID · Load frequency control · Model reduction · Plug-in electric vehicles · Root locus

Abbreviations PEVs Plug-in electric vehicles

PI Performance indices
EC Evolutionary computational PSO Particle swarm optimization
FO-PID Fractional-order PID controller PU Peak undershoot
HOS Higher order system RL Root locus
LFC Load frequency control ROM Reduced-order modelling
LOS Lower order system SCL Secondary control loop
MFO Moth-flame optimization ST Settling time
NRTPS Non-reheat thermal power systems WOA Whale optimization algorithm
PCL Primary control loop

B Man Mohan Garg List of symbols

[email protected]
1 Department of Electrical Engineering, Government College Δf Frequency deviation (Hz)
of Engineering, Keonjhar, India Δf ll Fixed lower limit (Hz)
2 Department of Electrical Engineering, Malaviya National Δf lu Fixed upper limit (Hz)
Institute of Technology, Jaipur, India ΔPD Load demand (MW)
3 Department of Electrical Engineering, National Institute of ΔPi max Power reserve upper limit (MW)
Technology, Rourkela, India ΔPi min Power reserve lower limit (MW)
4 Department of Electrical Engineering, Indian Institute of ΔPPFC Primary frequency control power (MW)
Technology Kharagpur, Kharagpur, India ΔPref Reference power (MW)

123
 Electrical Engineering

ΔPT Turbine power (MW) into the network. Previously, flywheel energy storage (FES)
ΔPtie Tie-line power (MW) units, capacitive energy storage (CES) units, superconduct-
μ Order of the derivative ing magnetic energy storage (SMES) units, and other devices
apf Area participation factor were used [1–3].
B Frequency bias parameter (pu MW/Hz) Plug-in electric vehicles (PEVs) have significantly
D Old area load (MW/Hz) increased in recent years, helping to control deviations in
fo Normal frequency (Hz) grid frequency and tie-line power. The global transportation
GH Hydraulic transfer function sector will use 50 million additional electric vehicles annu-
GP Power system transfer function ally, according to [4]. This encouraged researchers to use it
GT Turbine transfer function to control variations in grid frequency and tie-line power.
H(s) Inertia constant With power electronic converters, PEVs function efficiently
KD Derivative gain of PID controller as mobile energy storage systems, employing vehicle-to-grid
Ki Gain of plug-in electric vehicle technology to meet sudden load demand. Traditional sources
KI Integral gain for PID controller in combination with PEVs are deployed to maintain system
Kp Gain of the power system frequency and tie-line power in range, given that traditional
KP Proportional gain for PID controller power generation exhibits high inertia and can thus accom-
K PEV Overall plug-in electric vehicle gain modate sudden demand. The performance analyses of PEVs
N Number of population size using fuzzy-based robust controllers were highlighted by [5].
R Regulation parameter (Hz/MW) By adjusting the status of charge (SOC) signals, the author
SOC State of charge has suggested adaptive approaches to change the output
SOChigh State of charge higher value power of PEVs [6]. A decentralized charging scheme and an
SOClow State of charge lower value H-infinity-based robust controller are also highlighted for an
SOCmax State of charge maximum value interconnected wind power generation system and PEVs [7].
SOCmin State of charge minimum value A two-layer distributed control strategy is highlighted by [8]
T CONV Converter time constant (s) for PEVs and governor/turbine coordinated power systems.
TH Hydraulic time constant (s) Different types of energy mix, such as wind, solar, and diesel
To Synchronizing coefficient of line (s) generators, have been reported for two-area power system [9,
Tp Time constant of the power system (s) 10]. Effect of communication delay and its detailed analyses
TT Turbine time constant (s) in PEVs integrated power system (IPS) is highlighted by [8].
αi Coefficients of alpha(Routh) table The importance of SOC of PEVs unit in islanded IPS for
λ Order of the integrator controlling grid frequency is highlighted by [11]. Work has
also been reported by [12] for three- and four-area systems,
including PEVs. The impact of high-voltage dc (HVDC),
including PEVs, on the smooth flow of power and regulat-
ing system frequency is studied by [13]. The hierarchical
1 Introduction supervisory control in PEVs for IPS is highlighted by [14].
The dynamic response of the PEVs IPS to the deregulated
Electrical power system represents a grid with several dimen- environment is shown by [15]. The authors also emphasized
sions, incorporating an array of energy sources and demands. overall system stability by varying the various parameters of
The complexity grows even more as load demand varies PEVs in an IPS [16].
round the clock. Power imbalance caused by continuous Researchers and industries have developed and imple-
load variation causes frequency dips/rises and changes in mented numerous control strategies for regulating LFC
scheduled tie-line power. In this case, load frequency control operation, deviations in grid frequency, and tie-line power.
(LFC) plays important role because it keeps frequency and The proportional-integral-derivative (PID) controller is high-
tie-line power within a specific range. Due to the scarcity of lighted for its importance and popularity in the literature
traditional energy sources, primarily thermal energy, renew- [17], along with robust controller in LFC operation [18]. PID
able energy sources have received much attention in the last controller with filter (PIDF) is employed to manage the per-
decade. In recent years, solar and wind energy have captured formance of the system [19], and the fractional-order PID
a significant amount of attention. However, power output (FO-PID) controller for improving stability in LFC opera-
of these sources is highly depends on environmental fac- tion [7]. Low-inertia issue due to integration of renewable
tors. As a result, whether an energy source is conventional energy sources (RES) and its solution is highlighted in [20,
or renewable, the possibility of a power imbalance exists. 21]. FO-PID controller has gained popularity in recent years
To address this, energy storage systems are being integrated thanks to its many demonstrated applications in engineering

123
Electrical Engineering

and science fields. It improves the system’s performance and overdamped, underdamped, and critically damped models of
robustness. An FO-PID controller gives more design versa- PEVs.
tility than a PID controller because there are five parameters Furthermore, dynamic analysis of system with proposed
to pick from instead of three in the former. PEV connected two-area model is analysed with differ-
However, when building the controller for PEVs, the ben- ent controllers. The initial case employs WOA-tuned fuzzy
efits mentioned earlier of the FO-PID controller should be cascade controllers, while the second scenario adopts an
given more thought. Applications for LFC operation uti- MFO-tuned PI-PDF controller. In contrast, the third case uti-
lizing fuzzy logic controllers (FLC) are also highlighted. lizes a novel PSO-tuned FO-PID controller. Subsequently, a
However, its employment in IPS is constrained by the FLC’s comparative analysis of the various SOC levels is carried out.
requirement for more excellent system memory and the FO The subsequent sections in this paper are arranged in fol-
controller’s requirement for extra parameters. Also, many lowing fashion: Sect. 2 provides a detailed description of
authors have proposed different meta-heuristic ways to adjust system modelling, followed by the controller design and
the gain parameters of the controller. These include the objective function (J) in Sect. 3. PSO algorithm is explained
teaching learning-based optimization (TLBO) [22, 23], the in Sect. 4. Section 5 displays simulation results, and Sect. 6
moth-flame optimization (MFO) [24], the ant bee algorithm concludes this investigation.
(ABO) [25], the whale optimization algorithm (WOA) [26],
and water cycle algorithm (WCA) [27]. However, the benefits
of particle swarm optimization (PSO) algorithm, like its sim- 2 System configuration
ple algorithm, new solutions are always accepted, and they
are updated once in the entire iteration cycle are overlooked. Figure 1 depicts transfer function model of two-area power
Furthermore, in the PSO algorithm, acquiring the updated systems, featuring NRTPS integrated with PEVs in both
parameter can be achieved through a single stage [28]. areas. Interconnected electrical power system is intrinsically
A control system designed for a higher-order system is challenging, irregular, and dynamic, with number of genera-
also more likely to be complex. As a result, several strategies tor and load. However, for the sake of modelling and analysis,
for lowering the order of the linear time-invariant system are all the generators are merged into a single generator, as are
highlighted in [29]. Routh approximations are considered in the loads. The data of the examined two-area interconnected
this paper because they are simple and always result in a model of NRTPS are included in Appendix A.
stable transfer function [30]. Nonetheless, the authors have
pinpointed the following constraints and areas for further
2.1 Single-area non-reheat thermal power system
investigation in the previously published literature:
(NRTPS)
• In earlier published work, the controller gain parameter
Schematic representation of single-area NRTPS having both
limit was randomly selected.
primary and secondary control loops and generating power
• Previous research used bode plots to select the gain param-
using a single generator is presented in Fig. 2. Primary control
eter for PEVs, but it overlooked system type and a proper
loop (PCL) comprises two inputs-reference power settings
range of gain values.
(ΔPref ) and load power demand (ΔPD ), and one output-
• Without lowering their order, PEVs linearized models are
frequency deviation (Δf ). From the block diagram as shown
used in LFC operation.
in Fig. 1, using the mathematical relation between turbine
• The PEVs parameter selection criterion for the stability’s
power and load demand, frequency deviation is given as:
nature is not addressed.

This study presents a PSO-optimized FO-PID controller  f o (s)  G P (s)[PT (s) − PD (s)] (1)
for non-reheat thermal power systems (NRTPS) integrated
with PEVs based on the above research gaps. The research where GP (s), PT (s), and PD (s) represent power system,
also takes into account the PEV parameter, SOC, and PEV turbine power, and load power demand, respectively.
number, which have an impact on the stability. Root locus
(RL) technique is proposed to select the PEV gain parameter, KP
G P (s)  (2)
and the nature of the system is explored. The subsequent step 1 + sTP
involves the use of Routh approximation technique to ascer-
2H
tain the reduced-order transfer function of PEV. Once the TP  (3)
controller parameters are identified, they are further refined f 0D
using the PSO technique. Following this, comparative analy- 1
sis is carried out with the help of MATLAB Simulink across KP  (4)
D

123
 Electrical Engineering

Fig. 1 Two-areas interconnected

power system [8]

Fig. 2 Block diagram of

single-area NRTPS

Using transfer function model, a hydraulic valve actuator is equipped with a battery that interfaces with power sys-
is represented as: tems via battery charger. This storage unit accumulates the
necessary grid demand, while charger regulates the power
1 exchange. The dynamic model for the closed-loop system is
G H (s)  (5)
1 + sTH shown by first-order transfer function with a minimal time
constant, transferred into a transfer function form. The con-
In terms of turbine dynamics, the turbine transfer function trol loop comprises an adjustable regulation droop parameter
is: and dead-band unit with predefined boundaries. Regulation
parameter value is affected by primary reserve, charging
1
G T (s)  (6) capacity, and model participation factor. The variable droop
1 + sTT is represented by the ratio Ri /K i . Figure 4 illustrates the
charging and idle mode of PEVs in relation to K i and the
From Fig. 2, through inspection frequency deviation is
SOC.
given as:
Figure 5 illustrates the influence of the PCL power on
   frequency. PCL power reduces from ΔP’PCL to ΔPPCL when
1
Pref −  f G H G T − PD G P   f (7) K i is adjusted to K i ’. The participation factor play crucial
R
role in determining the impact of individual PEVs on LFC
within the electrical grid. It takes into account the distinct
For a constant reference input (ΔPref  0) and unit step
operational states of each PEV, namely, charging, idle, or
load change, frequency deviation Δf(s) is given as:
disconnected mode.
GP 1
 f (s)  − ∗ (8)
1 + (1/R)G P G H G T s 2.3 Model approximation

The development of PEV in LFC operation leads to a higher-

2.2 Plug-in electric vehicle order system (HOS). A load frequency controller designed
for the HOS will likely be more complicated. The natural way
The proposed PEV model for PCL study includes a charging to handle HOS is to approximate it with a lower-order system
block and a PCL control block, as illustrated in Fig. 3. PEV (LOS). Several methods are available to convert HOS to the

123
Electrical Engineering

Fig. 3 Transfer function model of

PEV [8]

array coefficients correspond to that of original model up to

the specified order. Any Routh approximant of a stable sys-
tem is stable using this approach. The Routh approximation
approach includes four distinct steps:
Step 1: Apply reciprocal transformation to the original
transfer function P(s) to obtain P̂(s):

∧ 1
Fig. 4 Droop ratio versus SOC a Charging mode b Idle mode [8] P (s)  P(s) (9)
s
Step 2: Compute alpha and beta coefficients using the α-β
table.
Step 3: Determine the numerator and denominator using
the α-β coefficients.
Step 4: Reduced model is obtained by reciprocating the
transfer function obtained in step 3.
Using step-1 in Eq. (8), single-area reduced model is given
as:

Fig. 5 Effect of participation factor on droop characteristic [8]  f (s) PH (s)PT (s)PP (s)
P(s)   (10)
U (s) 1 + PH (s)PTR(s)PP (s)

LOS, the dominant pole approach, the Pade approximation where,

technique, the Routh approximation strategy, etc. Using par-
1
tial fractions, the system’s transfer function is expanded using PH (s)  (11)
TH s + 1
the dominant pole method, and the approximation keeps the
terms whose poles are closest to the imaginary axis. This 1
PT (s)  (12)
approach has the drawback that solving the characteristic TT s + 1
equation determines the system’s poles. As the system’s order
KP
increases, finding the poles using characteristic equations is PP (s)  (13)
TP s + 1
challenging. In the Pade approximation method, the order of
numerator and denominator is selected for the approximant Using Eqs. (11), (12), and (13), Eq. (10) reduces to:
system. The selection of coefficients in the numerator and
denominator of a rational function is aimed at achieving the KP
⇒ P(s)  (14)
best possible agreement between the Taylor Series expansion a3 s 3 + a2 s 2 + a1 s + a0
of the approximated system and the approximant system. In
other words, the goal is to make the Taylor Series expan- where,
sions of both systems match in as many terms as feasible.
The issue with this method is that the poles of the approxi- a3  TH TT TP (15)
mant transfer function are dependent on both the numerator
and the denominator, which may lead to an unstable sys- a2  TH TT + TT TP + TP TH (16)
tem for a stable system. The Routh approximation method
is derived from the need for the approximant to be stable a1  TH + TT + TP (17)
for an original system. The fundamental concept is to cre-
ate the well-known “Routh Table” for the original system KP
before building the approximant in such a way that Routh a0  1 + (18)
R

123
 Electrical Engineering

Table 1 Alpha values using

Routh criteria a00 =a0 a00 =a0 a00 =a0 …
…
a01 =a1 a01 =a1 a01 =a1 …

α1 =a00 /a01 a02  a20 − α1 a21 a22  a40 − α1 a41 a42  a60 − α1 a61 …
α2 =a01 /a02 a03  a12 − α2 a22 a23  a41 − α2 a42 …
α3 =a02 /a03 a04  a22 − α3 a23 …
… … …

Table 2 Alpha values Denominator AK (s)  (2.5)(3.173)s 2 + (3.173)s + 1

51 7.624
⇒ 7.93s 2 + 3.173s + 1 (24)
20.38 0.48

α1 =51/20.38  2.5 6.423 Step: 04 Using reciprocating once again, reduced model
α2 =20.38/6.423  3.173 0.48 is given as:
α3 =6.423/0.48  13.381
18.869
P ∗ (s)  (25)
s2 + 3.173s + 7.93
On substituting the values of T H , T T , T P , K P, and R from
Time and frequency domain plot of both original plant and
Appendix A, Eq. (14) reduce to:
reduced system is depicted in Fig. 6a, b, and this validates
120 the similarity between reduced system and original system.
P(s)  (19)
0.48s 3 + 7.624s 2 + 20.38s + 51
2.4 Root locus (RL)
Step: 01 Reciprocating P(s) using relation P(s) 
1/s*P(1/s) There are various methods for selecting the controller gain
  for a PEV. In most articles, the Bode plot determines the
1 120
P    controller’s gain [16, 31]. However, there is ambiguity in
s 0.48 s 3 + 7.624 s12 + 20.38 1s + 51
1
picking the K PEV values within a specific range when uti-
  lizing a bode plot. In [8], controller gain is determined by
1 120s 3
⇒P  assessing the parametric value of K PEV from 0 to 20 using
s 51s 3 + 20.38s 2 + 7.624s + 0.48 the Bode plot method. As for K PEV  8.48, the effect of PEVs
∧ 120s 2 is neutralized in the stated range (0–20), indicating that K PEV
⇒ P (s)  (20)  8.48 cannot be treated as PEV gains. This suggests that
51s 3 + 20.38s 2 + 7.624s + 0.48
the defined range needs to be clearly defined. This paper pro-
Step: 02 Calculation of α-β using the α-β table. poses a novel approach using the RL technique for selecting
Using Eq. (20) and Table 1, different alpha coefficients K PEV values, which will overcome the concerns highlighted
are determined in Table 2 in [8]. Furthermore, the range of K PEV can be determined
Using Eq. (20) and Table 3, beta coefficient is determined much more precisely. Figure 7 displays the RL diagram of
in Table 4 the PEVs taken into consideration for frequency control in
Step: 03 Using Routh convergent, for a second-order sys- this study.
tem numerator and denominator coefficients are given as: The RL diagram provides the following information on
the nature of the system for different K PEV values:
Numerator BK (s)  α2 β1 s + β2 (21) Overdamped system: 0 < K PEV < 2.025.
Critically damped system: K PEV  2.025.
Denominator AK (s)  α1 α2 s + α2 s + 1 (22) Under damped system: K PEV > 2.025.
So the range of K PEV has no ambiguity and is clearly
Using α and β values from Tables 2 and 4, numerator and defined using the RL method. The value of K PEV can be
denominator coefficients are given as: selected between 0 < K PEV < 2.025. PEV behaviour in this
region is similar to an overdamped system, and the system
Numerator BK (s)  (3.173)(5.89)s + 0  18.69s (23) does not experience any oscillations.

123
Electrical Engineering

Table 3 Beta values using Routh

criteria b00 =a1 b21 =b3 b41 =b5 …
b02 =b2 b22 =b4 b42 =b6 …
…

β1 =b01 /a01 b03  b21 − β1 a21 b23  b41 − β1 a41 …

β2 =b02 /a02 b04  b22 − β2 a22 b24  b42 − β2 a42 …
β3 =b03 /a03 b05  b23 − β3 a23 …
… … …

Table 4 Beta values using Routh criteria 3 Controller development and objective
function’s role
120 0
0
The research suggests novel FO-PID controller for NRTPS
β1  120/20.38  5.89
integrated with PEVs to enhance dynamic performance indi-
cators such as settling time, peak undershoot, frequency

Fig. 6 a Step response;

b Frequency response of original
and reduced-order system

123
 Electrical Engineering

deviation, and stability. The block diagram of FO-PID con- T

 
troller is shown in Fig. 8 [32]. JITSE  t × ( f )2 + (Ptie )2 dt (32)
The proposed FO-PID controller integrates an integer-
0
order (IO) type PID controller and a fractional-order (FO)
integrator 1/sΨ [29]. FO-PID controller can be represented T

mathematically as: JIAE  { | f | + |Ptie |} dt (33)

   0
1 1
C(s)  K c 1 + + τd s (26) T
τi s sψ
JITAE  t × { | f | + |Ptie |} dt (34)
The general reduced 2nd order transfer function is repre- 0
sented as:
k
P ∗ (s)  , k 0 (27)
d2∗ s 2 + d1∗ s + d0∗
4 Particle swarm optimization (PSO)
The values of the controller parameters represented by
Eq. (26) are determined by comparing with Eq. (27) and PSO is an evolutionary computational (EC) method moti-
they are given as: vated by a swarm’s foraging and social behaviour. So, in
PSO, the natural process mimics the flocking of a bird or a
K c  d1∗ kλ, τi  d1∗ d0∗ and τd  d2∗ d1∗ (28) school of fish. Three key attributes of PSO are:

The proposed controller, represented by (26), has only • Particles’ peak performance.
two unknown parameters (λ, Ψ ). These two parameters can • Swarm’s peak performance.
be calculated using Eqs. (29) and (30). • Particles’ speed and velocity adjustment.

π − fm PSO is better than other EC techniques as it is developed

 −1 (29)
π 2 with simple concepts using primitive operators, which makes
1 it easier to write code for PSO [35]. This paper considers
λ (30) meta-heuristic algorithms’ advantages, such as adaptabil-
wgc+1
ity, ability to handle complex and nonlinear problems, and
Φ m  π /3 is selected mainly, and wgc is determined based robustness in real-world control systems. The solution in
on the concept that the gain crossover frequency phase plot PSO EC techniques is referred as a particle. Various stages
is flat or does not change much. involved in PSO techniques are represented in flow chart
In previous studies, researchers have predominantly relied Fig. 9 [24]. The first two steps, i.e. initialization of random
on performance indices (PI) to develop an appropriate con- initial value and fitness value evaluation, are the generalized
troller [33, 34]. In this study too, to optimize controller framework of EC techniques. Then there is a decision box
parameters, various PIs are explored, encompassing integral based on several generations, assuming current iteration is
square error (ISE), integral time square error (ITSE), integral smaller than the maximum iteration. So, first step is to update
absolute error (IAE), and integral absolute time error (ITAE). local best of each particle (K P , K I , and K D ) and global best
ITAE is used when minimizing settling time is a priority, IAE of swarm. After that, there is another loop concerning the
is used to reduce the overall error in the system regardless number of particles. As a result, the operation is carried out
of response time, ITSE is used where speed of response and particle-by-particle, first updating velocity and then position.
error both factors are required. In this paper settling time The new particle will yield a new column vector where the
and transient performance are considered for evaluation of particle is evaluated and assigned its fitness value again. After
best controller. Hence all the four PIs are considered in this that count is increased by 1, this process is repeated till each
paper so that a better analysis can be presented. Optimal con- particle is updated (loop-1). So, loop-1 gives updated veloc-
troller parameters are selected based on the lowest PI values. ity and updated position. After that counter for the number
The mathematical formulations of different PIs are presented of generations is increased by 1, and then the process repeats
below: till the termination condition satisfy (loop-2). Once it meets,
PSO terminates, and the optimum value is reported. The PSO
T
  generalized algorithm framework is provided as follows:
JISE  ( f )2 + (Ptie )2 dt (31) (i) Create a random initialization for the search space’s
0
controller (K P , K I , and K D ) settings and Δf.

123
Electrical Engineering

Fig. 7 Root locus diagram of

PEV

(ii) Based on the initial controller parameters, determine

the value of the objective function.
(iii) Assign the current solution as the personal best (Pbest )
and the corresponding objective function value as the best
fitness (f best ).
(iv) Find the population-wide solution that has the best
fitness and designate it as the global best (gbest ), with the
best fitness (f best ) as its fitness value.
(v) Determine the velocity and position of the ith param-
eter.
(vi) Based on its current value and the computed frequency
deviation, determine the new value of the ith parameter.
(vii) Set boundaries around the new value of the ith param- Fig. 8 Block diagram of FO-PID controller [33]
eter.
(viii) Apply the modified value to the objective function
value of the ith parameter. simulation cases were conducted to explore various aspects.
(ix) Update the controller parameter by including the new The first case focused on different types of systems, the sec-
value (x i ) and the corresponding objective function value (f i ). ond examined multiple controllers, the third involved varying
(x) Update the personal best (Pbest ) and the best fitness SOC levels of PEVs, and the fourth involves robustness anal-
(f besti ). ysis of the controllers. The ultimate goal of this analysis is
(xi) Update the global best (gbest ) and the best fitness to validate the proposed scheme for two-area configurations,
(f best ). ensuring its effectiveness across a wide range of scenarios.
(xii) Till the stopping criteria are reached, repeat the pre-
vious operation.
This iterative process continues until the algorithm con-
5.1 Analysis of the NRTPS performance using PEV
verges to a near-optimal solution, satisfying the stopping
for systems with various natures
criteria.
In this section, the primary focus is on investigating how
different types of systems respond to sudden or abrupt
5 Results and discussion load fluctuations. A bode plot approach was used by [8] to
choose the gain value. This technique has the disadvantage
As described here, the study thoroughly analyses perfor- of randomly selecting the K PEV range, which ignores the
mance of PEVs in conjunction with an NRTPS. Four different significance of the system’s nature. To evaluate the system’s

123
 Electrical Engineering

Fig. 10 Dynamic performance of the model for different values of K PEV

under 1% step load disturbance applied in area-1 a Frequency deviation
Fig. 9 Flowchart of PSO for tuning controller parameters [24] in area-1 b Frequency deviation in area-2 c Tie-line power deviation

functionality for LFC, K PEV is determined 1.44 with maxi- to area-1 and taking PEV into account in both areas. Based on
mum phase margin condition. But with this desired value of the simulation results depicted in Fig. 10, the underdamped
the K PEV , the system nature is over damped. This research system performs better than the overdamped and critically
employs the RL approach instead of the Bode plot technique damped systems.
to obtain the suitable K PEV value. Using RL methods, the Table 5 compares settling time and peak undershoots
K PEV ’s range is clearly defined. For 0 < K PEV < 2.025, the values. For all the three parameters (Δf 1 , Δf 2 and ΔPtie )
system is overdamped, for K PEV  2.025, system is criti- underdamped system exhibits less peak undershoot, as
cally damped, and for K PEV > 2.025 system is underdamped. shown in Fig. 11. Table 6 highlights the comparison
Simulation is carried out for overdamped, underdamped, and of PI’s values for a stable system. PSO is used to
critically damped systems by providing 1% step load change tune the controller parameters (K P , K I , and K D ) for an

123
Electrical Engineering

Table 5 Comparative statistics

for case (5.1) Type of system T s (in sec) PUs negative (in Hz) PUs negative (in
puMW)

Δf 1 Δf 2 ΔPtie Δf 1 Δf 2 ΔPtie

Overdamped 24.415 24.478 25.129 0.01351 0.003251 0.001359

system
Critically damped 19.000 23.438 24.186 0.01341 0.003158 0.001319
system
Underdamped 18.119 19.180 20.271 0.007809 0.001901 0.0007906
system

0.016 Overdamped system both thermal power and PEVs, frequency deviation of area-
0.014 Critically damped system 1, area-2, and tie-line power deviations are compared with
PUs negative (in Hz)

0.012 Underdamped system proposed scheme, as illustrated in Fig. 12 [30, 32].

0.01
On substituting values of the plant data considered in
0.008
Sect. 2.3, plant model is given by Eq. (19), and its reduced
0.006
order model is given by Eq. (25). The proposed FO-PID con-
0.004
troller using the specification wgc  15 rad/sec and Φ m 
0.002
0
π/4, becomes:
∆f1 ∆f2 ∆Ptie  
2.502 1
Fig. 11 Column graph analysis of peak undershoot
C I (s)  6.33 1 + + 0.3151s 0.33 (35)
s s

The suggested controller is implemented on the model

overdamped, underdamped, and critically damped system under investigation. A comparative analysis between the
which results in minimum overshoot for an underdamped FO-PID controller with the recently proposed PI-PDF con-
system. Furthermore, Table 6 shows that each of the troller [28] and the fuzzy cascade controller [16] is presented.
four PI produces better results for the underdamped sys- The simulation response depicted in Fig. 12 clearly demon-
tem. strates the superior efficiency of the suggested controller
in responding promptly to abrupt changes in load condi-
tions. Furthermore, the controller efficiently steers the system
5.2 Comparative review of proposed controller towards a stable and steady-state within a notably shorter
with fuzzy cascade and PI-PDF controller time frame. Table 7 compares the controllers’ settling time
and peak undershoot while comparative assessments of peak
In this section effects of different controllers are discussed in undershoot for the controllers are given in Fig. 13.
LFC. For the analysis purpose, three different controllers are
considered (Fuzzy cascade, PI-PDF, and proposed FO-PID 5.3 PEVs contribution towards LFC with different
controller). FO-PID controller parameter is tuned using PSO SOC levels
algorithm. The other two controller parameters are tuned with
recently developed WOA and MFO algorithms, respectively. This section discusses how the SOC level of PEVs affects
To explore the outcomes of the proposed FO-PID system for LFC. Participation of PEVs for frequency regulation depends

Table 6 Comparative numerical values of performance indices

Performance indices Overdamped system Critically damped system Under damped system

Δf 1 Δf 2 ΔPtie Δf 1 Δf 2 ΔPtie Δf 1 Δf 2 ΔPtie

ITAE 0.0533 0.0004 0.05656 0.04881 0.00041 0.05287 0.01619 0.00021 0.02163
IAE 0.02 0.00015 0.00998 0.01903 0.00015 0.00937 0.009109 6.864E-05 0.00454
ITSE 0.00012 7.47E-09 4.01E-05 0.00011 6.87E-09 3.633E-05 3.22E-05 1.678E-09 8.655E-06
ISE 7.41E-05 6.86E-09 9.22E-06 7E-05 6.24E-09 8.505E-06 2.33E-05 1.419E-10 2.377E-06

123
 Electrical Engineering

WOA tuned fuzzy MFO tuned PIPDF

cascade controller controller

30.000
25.000

Settling time (s)

20.000
15.000
10.000
5.000
0.000
∆f1 ∆f2 ∆Ptie

Fig. 13 Column chart comparison of peak undershoot

Table 8 Gain value (K PEV ) of PEVs for different SOC levels

SOC level (%) Value of K PEV

30 97.96
60 134.69
90 195.92

entirely on the PEVs battery’s SOC status. Examining the

impact of diverse SOC levels on the secondary frequency
control is essential. This section employs analysis of three
cases with three different SOC levels (30%, 60%, and 90%).
When a SOC level of more than 90% is considered, battery
life suffers. Similarly, if a SOC level of 30% is used, the
influence of PEVs on LFC will be minimal. As a result, SOC
values of the battery are considered between 30 and 90%.
K PEV values for different SOC levels can be calculated as:
  n 
SOC − SOClow(high)
K PEV  K PEV max 1 −
SOCmax(min) − SOClow(high)
(36)

where, SOCmin  10%, SOCmax  90%, SOClow  20%,

SOChigh  80%, and n  2, values are considered. For dif-
ferent SOC level the gain values of PEVs are determined
using Eq. (36) and they are given below in Table 8.
Fig. 12 Dynamic response of the model under 1% step load disturbance
applied in area-1 a Frequency deviation in area-1 b Frequency deviation The comparison of dynamic responses considering SOC
in area-2 c Tie-line power deviation of 30%, 60%, and 90% for LFC and tie-line power fluctua-
tions is presented in Fig. 14. From the simulation response

Table 7 Comparative numerical values of performance indices

Types of controllers and optimisation techniques T s (in sec) PUs negative (in Hz) PUs negative (in puMW)

Δf 1 Δf 2 ΔPtie Δf 1 Δf 2 ΔPtie

WOA tuned fuzzy cascade controller 19.157 26.860 23.664 0.00135

0.00326 0.00136
MFO-tuned PI-PDF controller 14.129 25.516 21.957 0.00804
0.00345 0.00153
Proposed PSO-tuned FO-PID controller 6.586 7.600 5.986 - 0.00308 0.00025 0.00011

123
Electrical Engineering

analysis it is concluded that, the dynamic behaviour improves

around SOCs of 60% and deteriorates around SOCs of 30%
and 90%.

5.4 Sensitivity analysis of the system with varying

step load

This section analyses the robustness of the proposed con-

troller (FO-PID) under variable step load and different SOC
of PEVs. Figure 15 depicts the step load consider for robust-
ness analysis. Figure 16 shows the system’s response with
varying step loads.
Figure 16 shows that the controller’s performance is
stable, and the system reaches steady-state at each inter-
val. Furthermore, a comparable dynamic response is seen
throughout all three intervals. Thus, the proposed controller’s
performance is stable under step load disturbances that
changes quickly.

6 Conclusion

This article presents a novel approach using RL to assess

the performance of PEVs within an interconnected power
system. By applying RL techniques, the study focuses on
determining the gain values for PEVs in different stable
systems, such as overdamped, underdamped, and critically
damped systems. In contrast to earlier research that employed
a Bode plot, the controllers gain parameters utilizing the RL
technique are well specified within a limited range. A com-
prehensive comparative analysis is conducted to identify the
most suitable system dynamics for LFC. The observations
from this study indicate that configuring controller parame-
ters for thermal units and PEVs in systems utilizing a ROM
is a more straightforward task than in systems with higher-
order dynamics.
Furthermore, the study introduces a FO-PID controller
and compares its performance to conventional controllers,
revealing significant improvements in peak deviation and
settling time. The PSO algorithm is employed to fine-tune
Fig. 14 Dynamic response of the model for different values of SOC the controller’s gain parameters. The article also thoroughly
(30%, 60% & 90%) under 1% step load disturbance applied in area-1 examines various SOC values for PEVs to determine the
a Frequency deviation in area-1 b Frequency deviation in area-2 c Tie-
line power deviation
optimal SOC level for effective LFC operation and robust-
ness analysis of the proposed FO-PID controller. Simulation

123
 Electrical Engineering

Fig. 15 Varying step load

Fig. 16 Dynamic response of the

system with varying step load

results highlight that maintaining a SOC level of approx- Data availability The data and materials can be obtained on request.
imately 60% proves beneficial in enhancing the system’s
dynamic behaviour during sudden load changes. Declarations
Author contributions RRS and MMG contributed to the conception and Competing interests The authors declare no competing interests.
design of the study. RRS contributed to drafting the manuscript. AKP
and DD contributed to revising the manuscript critically for important Conflict of interest The authors declare no competing interests.
intellectual content.

Funding Any funding agency does not support the work reported in
this paper.

123
Electrical Engineering

Appendix A 13. M Tavakoli E Pouresmaeil J Adabi R Godina JPS Catalão 2018

Load-frequency control in a multi-source power system connected
to wind farms through multi terminal HVDC systems Comput Oper
Res 96 305 315 https://doi.org/10.1016/j.cor.2018.03.002
PR1  PR2  2000 MW, f o  60 Hz, 14. K Janfeshan MAS Masoum 2017 Hierarchical supervisory control
system for PEVs participating in frequency regulation of smart
R1  R2  2.4 Hz/puMW, grids IEEE Power Energy Technol Syst J 4 84 93 https://doi.org/
B1  B2  0.425, K P  120, TP  20 s, 10.1109/jpets.2017.2740227
15. A Dutta S Debbarma 2018 Frequency regulation in deregulated
a12  −1, apf11  apf12  apf21  apf22  0.5, market using vehicle-to-grid services in residential distribution net-
work IEEE Syst J 12 2812 2820 https://doi.org/10.1109/JSYST.
To  0.0707, TG  0.08 s, TT  0.3 s, 2017.2743779
TBCU  0.01 s, TIC  0.001 s 16. S Izadkhast P Garcia-Gonzalez P Frias P Bauer 2017 Design of
Plug-In electric vehicle’s frequency-droop controller for primary
frequency control and performance assessment IEEE Trans Power
Syst 32 4241 4254 https://doi.org/10.1109/TPWRS.2017.2661241
17. A Behera TK Panigrahi PK Ray AK Sahoo 2019 A novel cas-
References caded PID controller for automatic generation control analysis with
renewable sources IEEE/CAA J Autom Sin 6 1438 1451 https://
1. Chatterjee K, Sankar R, Chatterjee TK (2012) SMES coordinated doi.org/10.1109/JAS.2019.1911666
with SSSC of an interconnected thermal system for load frequency 18. A Oshnoei M Kheradmandi SM Muyeen 2020 Robust control
control. In: 2012 Asia-Pacific Power Energy Engineering Confer- scheme for distributed battery energy storage systems in load fre-
ence, pp. 1–4 https://doi.org/10.1109/APPEEC.2012.6307061 quency control IEEE Trans Power Syst 35 4781 4791 https://doi.
2. Abraham RJ, Das D, Patra A (2005) Effect of capacitive energy org/10.1109/TPWRS.2020.2997950
storage on automatic generation control. In: 2005 International 19. TS Gorripotu RK Sahu S Panda 2015 AGC of a multi-area power
Power Engineering Conference, pp. 1070–1074 Vol. 2. https://doi. system under deregulated environment using redox flow batteries
org/10.1109/IPEC.2005.207066. and interline power flow controller Eng Sci Technol Int J 18 555
3. A Saha LC Saikia 2019 Load frequency control of a wind-thermal- 578 https://doi.org/10.1016/j.jestch.2015.04.002
split shaft gas turbine-based restructured power system integrating 20. A Abazari H Monsef B Wu 2019 Coordination strategies of dis-
FACTS and energy storage devices Int Trans Electr Energy Syst 29 tributed energy resources including FESS, DEG, FC and WTG in
e2756 https://doi.org/10.1002/etep.2756 load frequency control (LFC) scheme of hybrid isolated micro-
4. Chu Y, Cui H (2023) Annual update on the global transition to grid Int J Electr Power Energy Syst 109 535 547 https://doi.org/10.
electric vehicles: 2022 1016/j.ijepes.2019.02.029
5. Y Arya 2020 Effect of electric vehicles on load frequency control in 21. S Nema V Prakash H Pandzic 2022 Adaptive synthetic inertia
interconnected thermal and hydrothermal power systems utilising control framework for distributed energy resources in low-inertia
cf-foidf controller IET Gener Transm Distrib 14 2666 2675 https:// microgrid IEEE Access 10 54969 54979 https://doi.org/10.1109/
doi.org/10.1049/iet-gtd.2019.1217 ACCESS.2022.3177661
6. M Vahedipour-Dahraie H Rashidizaheh-Kermani HR Najafi A 22. A Abaeifar H Barati AR Tavakoli 2022 Inertia-weight local-search-
Anvari-Moghaddam JM Guerrero 2017 Coordination of EVs par- based TLBO algorithm for energy management in isolated micro-
ticipation for load frequency control in isolated microgrids Appl grids with renewable resources Int J Electr Power Energy Syst 137
Sci https://doi.org/10.3390/app7060539 107877 https://doi.org/10.1016/j.ijepes.2021.107877
7. A Daraz SA Malik AT Azar S Aslam T Alkhalifah F Alturise 2022 23. PK Mohanty BK Sahu TK Pati S Panda SK Kar 2016 Design and
Optimized fractional order integral-tilt derivative controller for analysis of fuzzy PID controller with derivative filter for AGC in
frequency regulation of interconnected diverse renewable energy multi-area interconnected power system IET Gener Transm Distrib
resources IEEE Access 10 43514 43527 https://doi.org/10.1109/ 10 3764 3776 https://doi.org/10.1049/iet-gtd.2016.0106
ACCESS.2022.3167811 24. S Mirjalili 2015 Moth-flame optimization algorithm : a novel
8. P Sanki M Basu PS Pal D Das 2021 Application of a novel PIPDF nature-inspired heuristic paradigm Knowl Based Syst https://doi.
controller in an improved plug-in electric vehicle integrated power org/10.1016/j.knosys.2015.07.006
system for AGC operation Int J Ambient Energy https://doi.org/ 25. P Ganesh Kumar C Rani D Devaraj AAT Victoire 2014 Hybrid ant
10.1080/01430750.2021.1920460 bee algorithm for fuzzy expert system based sample classification
9. P Iliadis S Ntomalis K Atsonios A Nesiadis N Nikolopoulos IEEE/ACM Trans Comput Biol Bioinform 11 347 360 https://doi.
P Grammelis 2021 Energy management and techno-economic org/10.1109/TCBB.2014.2307325
assessment of a predictive battery storage system applying a load 26. Y Mousavi A Alfi IB Kucukdemiral 2020 Enhanced fractional
levelling operational strategy in island systems Int J Energy Res 45 chaotic whale optimization algorithm for parameter identification
2709 2727 https://doi.org/10.1002/er.5963 of isolated wind-diesel power systems IEEE Access 8 140862
10. V Prakash H Pandzic 2023 Fast frequency control service provision 140875 https://doi.org/10.1109/ACCESS.2020.3012686
from active neighborhoods: opportunities and challenges Electr 27. H Hasanien M Matar 2018 Water cycle algorithm-based optimal
Power Syst Res https://doi.org/10.1016/j.epsr.2023.109161 control strategy for efficient operation of an autonomous microgrid
11. S Vachirasricirikul I Ngamroo 2014 Robust LFC in a smart grid IET Gener Transm Distrib https://doi.org/10.1049/iet-gtd.2018.
with wind power penetration by coordinated V2G control and fre- 5715
quency controller IEEE Trans Smart Grid 5 371 380 https://doi. 28. H Bevrani F Habibi P Babahajyani M Watanabe Y Mitani 2012
org/10.1109/TSG.2013.2264921 Intelligent frequency control in an AC microgrid: online PSO-based
12. B Sonker D Kumar P Samuel 2019 Dual loop IMC structure for load fuzzy tuning approach IEEE Trans Smart Grid 3 1935 1944 https://
frequency control issue of multi-area multi-sources power systems doi.org/10.1109/TSG.2012.2196806
Int J Electr Power Energy Syst 112 476 494 https://doi.org/10.1016/
j.ijepes.2019.04.042

123
 Electrical Engineering

29. MF Hutton B Friedland 1975 Routh approximations for reducing 35. XM Yu XY Xiong YW Wu 2004 A PSO-based approach to opti-
order of linear time-invariant systems IEEE Trans Automat Contr mal capacitor placement with harmonic distortion consideration
20 329 337 https://doi.org/10.1109/TAC.1975.1100953 Electr Power Syst Res 71 27 33 https://doi.org/10.1016/j.epsr.2004.
30. S Saxena 2019 Load frequency control strategy via fractional-order 01.002
controller and reduced-order modeling Int J Electr Power Energy
Syst 104 603 614 https://doi.org/10.1016/j.ijepes.2018.07.005
31. S Saxena YV Hote 2013 Load frequency control in power systems
Publisher’s Note Springer Nature remains neutral with regard to juris-
via internal model control scheme and model-order reduction IEEE
dictional claims in published maps and institutional affiliations.
Trans Power Syst 28 2749 2757 https://doi.org/10.1109/TPWRS.
2013.2245349
Springer Nature or its licensor (e.g. a society or other partner) holds
32. Suri G, Professor A, Dinesh K (2015) Implementation of fractional
exclusive rights to this article under a publishing agreement with the
order PID controller for an AVR system
author(s) or other rightsholder(s); author self-archiving of the accepted
33. S Hanwate YV Hote S Saxena 2018 Adaptive policy for load fre-
manuscript version of this article is solely governed by the terms of such
quency control IEEE Trans Power Syst 33 1142 1144 https://doi.
publishing agreement and applicable law.
org/10.1109/TPWRS.2017.2755468
34. A Fathy AG Alharbi 2021 Recent approach based movable damped
wave algorithm for designing fractional-order PID load frequency
control installed in multi-interconnected plants with renewable
energy IEEE Access 9 71072 71089 https://doi.org/10.1109/AC
CESS.2021.3078825

123