International Journal of Applied Power Engineering (IJAPE)

Vol. 13, No. 1, March 2024, pp. 52~65

ISSN: 2252-8792, DOI: 10.11591/ijape.v13.i1.pp52-65  52

Optimization of controllers using soft computing technique for

load frequency control of multi-area deregulated power system

Dharmendra Jain, M. K. Bhaskar

Department of Electrical Engineering, M.B.M Engineering College, M.B.M. University, Jodhpur, India

Article Info ABSTRACT

Article history: Given the changing nature of power systems, it is challenging to optimize
the controller for controlling load frequency problems. Distributed power
Received Feb 19, 2023 generating sources and power system reorganization with multi-sources and
Revised May 30, 2023 multi-stakeholders make traditional load frequency control approaches
Accepted Jun 24, 2023 unsuitable for current power systems. This research provides the
comparative analysis of regulation of the load frequency in a multiple-area
deregulated electricity system with the help of soft computing. In a
Keywords: reorganized electrical system, the major objectives of load frequency control
(LFC) are to set up system frequency into acceptable limit, swiftly return the
Deregulated frequency to the setpoint, reduce tie-line power flow fluctuations across
Genetic algorithm adjacent control zones, and track load demand agreements. To achieve
Load frequency control LFC's goals, proportional integral derivative (PID) gain values must be
Particle swarm optimization tuned, for optimization purpose, soft computational methods are used in this
Proportional integral derivative present work. MATLAB/Simulink simulation results show that soft
Restructured computing controllers can keep tie line power interchange within contracted
constraints and frequency variation within the allowed range. This article
compares auto tuned PID, genetic algorithm (GA), and particle swarm
optimization (PSO) controllers in unregulated circumstances, load frequency
regulation of two-area power systems.
This is an open access article under the CC BY-SA license.

Corresponding Author:
Dharmendra Jain
Department of Electrical Engineering, M.B.M Engineering College, M.B.M. University
Ratanada, Jodhpur, Rajasthan, India
Email: [email protected]

1. INTRODUCTION
The term "interconnected power system" refers to the tie line connection of many control zones.
Interconnected network offers numerous advantages. But, whenever abrupt alteration of load occurs in any of
the connected zone, tie-line power as well as frequency deviate from their predefined amount. In order to
retain the frequencies and comparative power angles with respect to the particular values with acceptable
boundaries for both dynamic and static circumstances, the generators inside a control region constantly
change their speed in concert, either speeding up or slowing down. Whenever a load element is initially
introduced, momentarily drawing on the system's kinetic energy, the necessary energy is obtained.
Interconnected grids enable the distribution of power from various sources to meet varying demands. This
load balancing helps stabilize the grid and ensures that power supply matches consumption, reducing the risk
of blackouts due to sudden spikes in demand. There, should be some methods of control in the power system
to restore the deviated tie-line power and frequency to desired values otherwise it may degrade the system
performance and large changes may eventually result in system failure.
In the traditional electrical system, an individual firm known as a vertically integrated utility (VIU)
owns the power generating, transmission, and distribution facilities. Consumers get power at a rate specified

Journal homepage: http://ijape.iaescore.com

Int J Appl Power Eng ISSN: 2252-8792  53

by VIU. Power system became more complex now a days due to introduction of distributed generation,
restructuring of power system and uncertain environment. As the power system was restructured, many
market participants, including generating companies (GENCOs), transmission companies (TRANSCOs),
distribution companies (DISCOs), and independent system operators, assumed the role of VIU. In the
deregulated electrical system, each control region is responsible for supplying its own requirement and
planned interchange electricity. A shift in frequency can be detected when any imbalance between the power
supply and the load occurs.
In a multiple-area competitive electrical system, the regulation of the load frequency often includes
an appropriate control system that allows the zone frequency and tie-line power to be reset to their
established values, whensoever a significant shift in load takes place in an interconnected power system. For
the complete power system to operate economically, securely, and steadily, innovative and better control is
necessary. The restructured power system's load frequency can be greatly aided by the application of new
control techniques. Current advancements demonstrate the usage of soft computing techniques, which are
highly adaptable to changing environments and also able to make quick judgements.
Many researchers are working in this area. Several investigations have been conducted for various
load frequency control (LFC) issues in a liberalized power system to overcome these situations. Many of the
researchers used proportional integral derivative (PID) controllers to solve load frequency control problem,
because of its accuracy and high speed. The fine-tuning of the settings of a PID controller determines its
effectiveness unswervingly. Therefore, many researchers used soft computing-based techniques like neural
networks, fuzzy logic honey bee algorithm, firefly algorithm or other methods for tuning of parameters in
order to significantly improve controller gains.
Donde et al. [1] discussed evaluation and optimizations in an automatic generation control (AGC)
system post liberalization. The fundamental concept of autonomous generation regulation in a deregulated
electricity market for a linked power system is also been discussed by Kothari et al. [2]. Tan et al. [3] has
described decentralised load frequency regulation in restructured contexts. Jain et al. [4]–[6] discussed the
use of PID controller is used to control for load frequency control of multi area conventional and restructured
power network. Optimal firefly algorithm for LFC of electrical system in restructured scenario has been
explained by Sekhar et al. [7]. Babahajiani et al. [8] described intelligent demand response for LFC. Sahoo
[9] explained the use of neural network for line congestion study. Cohn [10] presented various tie-line bias
control issues for multi area connected power systems. Fosha and Elgerd [11], [12] used optimal control
theory for megawatt frequency control problem. Current operating problems associated with AGC have been
described in [13]. Abedinia et al. [14] used fuzzy PID using honey-bee mating optimization (HBMO) for
LFC. Abd-Elazim and Ali [15] constructed a load frequency controller for a two-zone system that consists of
a photovoltaic (PV) grid and a thermal generator using the firefly algorithm. For power quality
enhancements, Mishra et al. [16] used a particle swarm optimization-grey wolf optimization (PSO-GWO)
optimised fractional order PID (FOPID) based hybrid shunt active power filter.
Suid and Ahmad [17] discussed how to tune a sigmoid PID controller for an autonomous voltage
regulator using a nonlinear sine-cosine algorithm. PSO-PID controller was utilised by Dhanasekaran et al. [18]
for the LFC of a standalone multi-source power system. Concordia and Kirchmayer [19] has discussed the
frequency regulation and tie line power of electrical power systems. Sood [20] proposed optimal power flow
based on evolutionary programming and its confirmation for deregulated electrical systems. Method of
reduced order observer is used by Rakhshani and Sadeh [21] for AGC of two power zone system.
Pathak et al. [22] given the realistic model of centralized AGC. Reinforced learning neural network
controller has been used for LFC by Saikia et al. [23] and Pal et al. [24]. Shree and Kamaraj [25] suggested a
hybrid neuro fuzzy technique for automatic generation control in a restructured electrical supply system.
Bhateshvar and Mathur [26] employed a fuzzy logic-controlled SMES unit to stabilise frequency for a
thermal-hydropower system in a restructured environment. Arya and Kumar [27] proposed fuzzy gain
scheduling controller. Shankar et al. [28] have discussed impression of energy storing system on load
frequency control. The detailed study of thyristor-controlled series compensator for load following in
liberalized power system has been done by Deepak and Abraham [29]. Adaptive decentralized LFC of multi-
zone power systems has been presented by Zribi et al. [30]. According to a memoizable and smoothed
functional algo, Mok and Ahmad [31] explained how to tune a fractional order PID controller for an
automatic voltage regulator (AVR) system to its best possible settings. Computational analysis of PID and
PSO-PID optimization for MIMO process control system was covered by Shaikh et al. [32].
Genetic algorithm (GA) optimization technique and PSO technique are used to optimize the
parameters of PID controller to solve LFC of two area interconnected deregulated electrical power system in
this work with the objective to minimize the fitness function ITAE. Comparative analysis has also been made
for different contractual conditions in deregulated power system which shows the superiority of the proposed
PSO based controller over other controllers used in this work.

Optimization of controllers using soft computing technique for load … (Dharmendra Jain)
54  ISSN: 2252-8792

2. DEREGULATED POWER SYSTEM

A deregulated power system is one in which independent system operators (ISO), generating
businesses (GENCOs), transmission companies (TRANSCOs), and distribution companies (DISCOs) take
the place of vertically integrated utilities (VIUs). Though distribution companies (DISCOs) are permitted to
enter into agreements with any of the GENCOs within their own or other parts of the deregulated electricity
system, generation firms (GENCOs) are not required to engage in the AGC mission. As a result, there are
numerous different contract scenario combinations that might be used by DISCOs and GENCOs. The two-
area deregulated electricity system model is used to express different potential contracts using the distribution
participation matrix (DPM) concept. DPM is a matrix having a row count that corresponds to the number of
GENCOs in the system and a column count that corresponds to the number of DISCOs. Every element in the
DPM, referred to as a contract participation factor (CPF), is a percentage of the total contracted load needs of
a DISCO that a GENCO is meeting. As a result, the ijth entry cpfij represents the percentage of the total load
power that DISCO j purchased from a GENCO i. Addition of all the elements of a column in a disco
participation matrix is equal to one. Think about a two-region deregulated electricity system with two
GENCOs and two DISCOs in each area. As illustrated in Figure 1, GENCO1, GENCO2, DISCO1, and
DISCO2 are placed in Area I and GENCO3, GENCO4, DISCO3, and DISCO4 are placed in Area II.

Figure 1. Power system with two areas in a newly organized setting

The associated DPM will be represented as (1).

𝑐𝑝𝑓11 𝑐𝑝𝑓12 𝑐𝑝𝑓13 𝑐𝑝𝑓14

𝑐𝑝𝑓21 𝑐𝑝𝑓22 𝑐𝑝𝑓23 𝑐𝑝𝑓24
DPM = [ ] (1)
𝑐𝑝𝑓31 𝑐𝑝𝑓32 𝑐𝑝𝑓33 𝑐𝑝𝑓34
𝑐𝑝𝑓41 𝑐𝑝𝑓42 𝑐𝑝𝑓43 𝑐𝑝𝑓44

A local load is reflected whenever a load needed by a DISCO shifts within the vicinity in which the DISCO
corresponds. At the point of input to the power system block, this should be recorded in the deregulated AGC
system block diagram as the local loads PL1 and PL2. Each area has a large number of GENCOs, thus the
area control error (ACE) signal must be divided among them proportionately to their involvement in the
AGC. ACE participation factors are coefficients that allocate ACE to numerous GENCOs. Noteworthy is the
fact that:

∑𝑚
𝑖=1 𝑎𝑗𝑖 = 1 (2)

where, participation factor of i-th GENCO in j-th area is represented as aji and the number of GENCOs in j-th
area is given by m. On the tie line, the estimated steady state power flow is indicated as in (3).

∆𝑃𝑡𝑖𝑒1−2 𝑠𝑐ℎ𝑒𝑑𝑢𝑙𝑒𝑑 =
(𝑅𝑒𝑞𝑢𝑒𝑠𝑡𝑒𝑑 𝑝𝑜𝑤𝑒𝑟 𝑜𝑓 𝐷𝐼𝑆𝐶𝑂𝑠 𝑖𝑛 𝑆𝑒𝑐𝑜𝑛𝑑 𝑎𝑟𝑒𝑎 𝑡ℎ𝑟𝑜𝑢𝑔ℎ 𝐺𝐸𝑁𝐶𝑂𝑠 𝑖𝑛 𝑓𝑖𝑟𝑠𝑡 𝑎𝑟𝑒𝑎) −
(𝑅𝑒𝑞𝑢𝑒𝑠𝑡𝑒𝑑 𝑝𝑜𝑤𝑒𝑟 𝑜𝑓 𝐷𝐼𝑆𝐶𝑂𝑠 𝑖𝑛 𝑓𝑖𝑟𝑠𝑡 𝑎𝑟𝑒𝑎 𝑓𝑟𝑜𝑚 𝐺𝐸𝑁𝐶𝑂𝑠 𝑖𝑛 𝑠𝑒𝑐𝑜𝑛𝑑 𝑎𝑟𝑒𝑎) (3)
𝑗=4 𝑗=2
∆𝑃𝑡𝑖𝑒1−2𝑠𝑐ℎ𝑒𝑑𝑢𝑙𝑒𝑑 = ∑𝑖=2 𝑖=4
𝑖=1 ∑𝑗=3 𝐶𝑃𝐹𝑖𝑗 ∆𝑃𝐿𝑗 − ∑𝑖=3 ∑𝑗=1 𝐶𝑃𝐹𝑖𝑗 ∆𝑃𝐿𝑗 (4)

The tie line power error ∆𝑃𝑡𝑖𝑒1−2,𝑒𝑟𝑟𝑜𝑟 is characterized at any specified instant as in (5).

∆𝑃𝑡𝑖𝑒1−2,𝑒𝑟𝑟𝑜𝑟 = ∆𝑃𝑡𝑖𝑒1−2,𝑎𝑐𝑡𝑢𝑎𝑙 − ∆𝑃𝑡𝑖𝑒1−2,𝑠𝑐ℎ𝑒𝑑𝑢𝑙𝑒𝑑 (5)

Int J Appl Power Eng, Vol. 13, No. 1, March 2024: 52-65
Int J Appl Power Eng ISSN: 2252-8792  55

As the real tie line power flow meets the designed power flow, the ∆𝑃𝑡𝑖𝑒1−2,𝑒𝑟𝑟𝑜𝑟 disappears in the
steady state. In the conventional case, this error value is used to create the corresponding ACE signals.

𝐴𝐶𝐸1 = 𝐵1 ∆𝑓1 + ∆𝑃𝑡𝑖𝑒1−2,𝑒𝑟𝑟𝑜𝑟 (6)

𝐴𝐶𝐸2 = 𝐵2 ∆𝑓2 + ∆𝑃𝑡𝑖𝑒2−1,𝑒𝑟𝑟𝑜𝑟 (7)

Where:

∆𝑃𝑡𝑖𝑒2−1,𝑒𝑟𝑟𝑜𝑟 = − (𝑃𝑟1 /𝑃𝑟2 ) ∆𝑃𝑡𝑖𝑒1−2,𝑒𝑟𝑟𝑜𝑟 (8)

and 𝑃𝑟1 , 𝑃𝑟2 are the rated powers of Areas I and II, respectively. Therefore:

𝐴𝐶𝐸2 = 𝐵2 ∆𝑓2 + 𝛼12 ∆𝑃𝑡𝑖𝑒1−2,𝑒𝑟𝑟𝑜𝑟 (9)

where:

𝛼12 = − (𝑃𝑟1 /𝑃𝑟2 ) (10)

for a two-area system, the contractual power provided by i-th GENCO is stated as:

∆𝑃𝑖 = ∑𝑛𝑗=1
𝑑𝑖𝑠𝑐𝑜=4
𝐶𝑃𝐹𝑖𝑗 ∆𝑃𝐿𝑗 (11)

For 𝑖 = 1, ∆𝑃1 = 𝐶𝑃𝐹11 ∆𝑃𝐿1 + 𝐶𝑃𝐹12 ∆𝑃𝐿2 + 𝐶𝑃𝐹13 ∆𝑃𝐿3 + 𝐶𝑃𝐹14 ∆𝑃𝐿4 (12)

Similarly, ∆P2, ∆P3, and ∆P3 can be calculated easily.

Figure 2 depicts the Simulink diagram for the bilaterally deregulated LFC in two area (with reheat
turbine) system. It is structurally built on the notion of [1]. Demand signals are displayed on dashed lines.
Areas I and II's local loads are identified by the symbols ∆𝑃1𝐿𝑂𝐶 and ∆𝑃2𝐿𝑂𝐶 , respectively. ∆𝑃𝑢𝑐1 and ∆𝑃𝑢𝑐2
are uncontracted power. Also note that:

∆𝑃1𝐿𝑂𝐶 = ∆𝑃𝐿1 + ∆𝑃𝐿2 (13)

∆𝑃2𝐿𝑂𝐶 = ∆𝑃𝐿3 + ∆𝑃𝐿4 (14)

Figure 2. An illustration of a two-area restructured power system

Optimization of controllers using soft computing technique for load … (Dharmendra Jain)
56  ISSN: 2252-8792

3. PARAMETER TUNING OF PID CONTROLLER

PID controllers are most common and very effective controller. There are three separate parameters
in PID controller and accordingly entitled as 3-element control: the proportional, the integral and derivative
gains, denoted KP, KI, and KD. The interpretation of these values in terms of error is that K P depends on the
current error, KI on the cumulative of previous errors, and KD is a forecast of future errors based on the
current rate of change. Through a control element, the process is modified using the weighted total of these
three parameters. PID controller tuning is the process of determining the appropriate settings for K P, KI, and
KD. Although having only three parameters, PID tuning is a tough challenge.
In every system, fast response and good stability are the prime requirements. Unfortunately, faster
the response, worse the stability, and better the stability, slower the response. So, for the proper control of the
system, compromise has to be made between acceptable stability and fastness of response. By using preset
tunings, PID controllers frequently offer enough control, but deliberate tuning can usually boost performance.
The PID controller's parameters can be adjusted using a variety of techniques. The majority of
efficient techniques often entail the creation of some sort of process model, followed by the selection of KP,
KI, and KD depending on the characteristics of the dynamic model. Manual tuning techniques may not be
very effective, because it takes very long time. Z-N method and IMC methods are used by many researchers,
are manual tuning methods. PID controller parameter adjustment can be done using soft computing
techniques. For determining the correct values of K P, KI, and KD, these techniques are highly useful.

3.1. Tuning by particle swarm optimization technique

Eberhart and Kennedy (1995) created the well-known stochastic optimization method known as
particle swarm optimization (PSO). It takes its cues from social interactions between flocks of birds or
schools of fish. It is used in this work to investigate the search space for a certain topic to find the optimal
parameter values of controller required to satisfy the load frequency control objectives. The PSO commences
with a collection of random particles (solutions), after which it updates the solutions to look for the best one.
Every particle is represented by two vectors, position ‘xi’ and velocity ‘vi’.
Every particle's position at a specific moment is taken into account as a potential solution to the
issue at hand. The particles move throughout the search region while varying their speed and position to
always be in the ideal position. Each and every one of the particles contain fitness values that the fitness
function evaluates in order to optimize, and they all have velocities that control how they fly. The particles
follow the current optimal particles as they move through the problem space. The following vectors serve as
representations for the position and speed of each ith particle in a practical d-dimensional solution space

xi = [xi1 , xi2 , . . . , xid ] (15)

vi = [vi1 , vi2 , . . . , vid ] (16)

There are two best value, first best outcome, or the fitness it has attained till the iteration, or 𝑝𝑏𝑒𝑠𝑡
(particular best), and the second best value that has been attained yet by any member of the population, or
𝑔𝑏𝑒𝑠𝑡 (global best)—are used to update each particle. 𝑃𝑏𝑒𝑠𝑡 is the optimal location that produces the ith
particle's best fitness value, and gbest is the optimal position over the entire swarm population. The following
is a representation of the ith particle's best performances:

𝑝𝑏𝑒𝑠𝑡i = [𝑝𝑏𝑒𝑠𝑡𝑖1 , 𝑝𝑏𝑒𝑠𝑡𝑖2 , . . . , 𝑝𝑏𝑒𝑠𝑡𝑖𝑑 ] (17)

𝑔𝑏𝑒𝑠𝑡i = [𝑔𝑏𝑒𝑠𝑡𝑖1 , 𝑔𝑏𝑒𝑠𝑡𝑖2 , . . . , 𝑔𝑏𝑒𝑠𝑡𝑖𝑑 ] (18)

The (19), is used by the PSO algorithm to modify its velocity and location. The velocity updating
equation is:

𝑣𝑖𝑑 (𝑗 + 1) = 𝑤(𝑗)𝑣𝑖𝑑 (𝑗) + 𝑐1 𝑟1 [𝑝𝑏𝑒𝑠𝑡𝑖𝑑 (𝑗) − 𝑥𝑖𝑑 (𝑗)] + 𝑐2 𝑟2 [𝑔𝑏𝑒𝑠𝑡𝑖𝑑 (𝑗) − 𝑥𝑖𝑑 ] (19)

the velocity of the 'i'th particle in the 'd'th dimension and during the jth iteration is represented by 𝑣𝑖𝑑 (𝑗).
The (20) states that when the velocity for every particle has been determined, its position will be
updated by adding the new velocity to its prior position. PSO flow chart is given in Figure 3.

𝑣𝑖𝑑 (𝑗 + 1) = 𝑥𝑖𝑑 (𝑗) + 𝑣𝑖𝑑 (𝑗 + 1) (20)

Int J Appl Power Eng, Vol. 13, No. 1, March 2024: 52-65
Int J Appl Power Eng ISSN: 2252-8792  57

In order to investigate and emphasize the successful implementation of PSO to optimize the PID gains for
LFC in a deregulated electricity system that runs under bilateral contractual conditions, performance index-
based analysis is used. The performance index is given in (21) and performance index optimization curve is
shown in Figure 4.
𝑇
𝑓(𝐼𝑇𝐴𝐸) = ∫0 𝑇|∆𝑃𝑡𝑖𝑒−𝑖 + 𝐵𝑖 ∆𝑓𝑖 | 𝑑𝑇 (21)

Figure 3. PSO flow chart Figure 4. Optimization curve

There are only a few steps in the PSO algorithm, and they are continued unless a terminating
requirement is fulfilled. The following are the steps:
- Step 1: the first step is to initialize the particles. Set the iteration number k to zero. Create n particles
randomly, Xi, i = 1, 2,..., n, with xi = [xi1, xi2,...,xid] and starting velocities Vi =[Vi1, Vi2,...,Vid].
- Step 2: second step is to update the iteration counting as k=k+1.
- Step 3: use velocity (12) to update velocity of the particle as a next step.
- Step 4: in next step, use position (13) to update position of the particle.
- Step 5: now particle best can be updated:

𝐼𝑓 𝑒𝑣𝑎𝑙𝑖 (𝑥𝑖𝑘 ) > 𝑒𝑣𝑎𝑙𝑖 (𝑝𝑏𝑖𝑘−1 ) 𝑡ℎ𝑒𝑛 𝑝𝑏𝑖𝑘 = 𝑥𝑖𝑘 𝐸𝑙𝑠𝑒 𝑝𝑏𝑖𝑘 = 𝑝𝑏𝑖𝑘−1

- Step 6: as a next step global best can be updated as:

𝑒𝑣𝑎𝑙 (𝑔𝑏 𝑘 ) = 𝑚𝑎𝑥 (𝑒𝑣𝑎𝑙𝑖 (𝑝𝑏𝑖𝑘−1 ))

𝐼𝑓 𝑒𝑣𝑎𝑙 (𝑔𝑏 𝑘 ) > 𝑒𝑣𝑎𝑙(𝑔𝑏 𝑘−1 ) 𝑡ℎ𝑒𝑛 𝑔𝑏 𝑘 = 𝑔𝑏 𝑘 𝐸𝑙𝑠𝑒 𝑔𝑏 𝑘 = 𝑔𝑏 𝑘−1

- Step 7: last step is to reach decision to stop the algorithm: If the overall value of iterations reaches the
highest number of iterations or the total reporting is 100%, halt; else, return to step 2.
These procedures result in the controllers' optimal values with the minimizing ITAE. Parameters are listed in
the Table 1.

Table 1. Controller optimal tuned values by PSO algorithm

Controller-I Controller-II Controller-III
KP KI KD KP KI KD KP KI KD
0.3862 0.8755 0.3145 0.4989 0.8802 0.2414 0.0316 0.5102 0.1957

3.2. Tuning by genetic algorithm

GAs are a type of soft computing technique. GA is a commonly used optimization algorithm that
evolves solutions to issues using ideas from natural genetics. Darwin's idea of evolution serves as a major
inspiration for genetic algorithms. Several scientists and engineers have successfully used GA to solve a wide
Optimization of controllers using soft computing technique for load … (Dharmendra Jain)
58  ISSN: 2252-8792

range of difficulties in their fields, including optimization, machine learning, automatic programming,
transportation issues, and adaptive control.
One type of adaptive system that primarily aims to learn from, adopt from, and behave like
biological or natural entities is the genetic algorithm. By using mathematical techniques to extract, construct,
and explain a number of fundamental components, behaviors, and methods of biological processes and
adaptation, GAs was presented as an alternative optimization strategy. The fundamental mechanism is
described in flowchart shown in Figure 5.

Figure 5. Flow chart of genetic algorithm

Tuning of the PID controller has been done using GA by minimizing the time multiplied absolute
error. The various steps in finding the parameters of a PID controller are:
- Step 1: recognize the plant's transfer function or identify the system model.
- Step 2: initialize controller gains KP, KI, and KD, and calculate ITAE.
- Step 3: obtain the values, particle finest value (pbest) and the global best value (gbest).
- Step 4: apply mutation to calculate new population.
- Step 5: obtain updated particle best (pbest1) and updated global best (gbest1) values.
- Step 6: compare previous and updated particle best values (pbest and pbest1).
- Step 7: compare previous and updated global best values (gbest and gbest1).
- Step 8: obtain the updated controller gains KP, KI, and KD, and find out the response for the system.
- Step 9: if stopping criteria reached or maximum number of iterations reached: stop otherwise go to step 4.
Optimized parameters of the GA based controllers are obtained and given in Table 2, with the minimizing ITAE.

Table 2. Controller parameters tuned by GA algorithm

Controller-I Controller-II Controller-III
KP KI KD KP KI KD KP KI KD
9.6488 7.2247 1.5497 7.9835 5.1655 1.6694 8.9965 2.2321 9.081

4. SIMULATION AND RESULTS

4.1. Case-I
Case-I is the base case. Required load requirement across all DISCOs is 0.005 pu MW. Relative
responses for frequencies of both the area and responses of GENCO 1 to 4 (all the GENCO in both areas)
obtained using PID Controller, Figures 6 to 9 illustrate GA-based and PSO-based controller responses.
Figures 6 and 7 give the comparison of time response of area frequency f-1 and f-2. Transient response due to
sudden change die out in few seconds and change in frequencies of both the area settle to zero and
frequencies of both the areas set back to desired or predefined values. Similarly, GENCO of bath the areas
supply power as per the contractual conditions after the transient period.

Int J Appl Power Eng, Vol. 13, No. 1, March 2024: 52-65
Int J Appl Power Eng ISSN: 2252-8792  59

Figure 6. Area-1 frequency response (F-1)

Figure 7. Area-2 frequency response (F-2)

Figure 8. Outputs of GENCO of Area-1

Optimization of controllers using soft computing technique for load … (Dharmendra Jain)
60  ISSN: 2252-8792

Figure 9. Outputs of GENCO of Area-2

4.2. Case-II
Case-II: at t=25 sec., Area-1 raises a new load requirement of 0.0025 pu-MW, but only GENCO-1
of Area-1 can meet it. This case is known as contract violation case. Relative responses for frequencies of
both the area and responses of GENCO 1 to 4 (all the GENCO in both areas) obtained using PID controller,
Figures 10 to 13 depict GA-based and PSO-based controllers responses. Figures 10 and 11 give the
comparison of time response of area frequency f-1 and f-2. Transient response due to sudden change die out
in few seconds due to controller action. And change in frequencies of both the area settle to zero and
frequencies of both the areas set back to desired or predefined values. GENCO-1 of Area-1 supply additional
power demanded by the disco of Area-1 after contract violation and all other GENCOs maintain supply as per
the contractual conditions.

4.3. Case-III:
Case-III: at t=25 sec., Area-1 and Area-2 enhance their load demand by 0.0025 pu-MW, which is
met only by the GENCOs in those two areas. GENCO-1 of Area-1 supply additional load demand od Area-1
and GENCO-3 of Area-2 supply the additional load demand of Area-2. Figures 14 to 17 compare the
response characteristics of the system utilising a PID controller, a PSO and GA based controllers.

Figure 10. Comparative analysis of frequency F-1 of Area-1

Int J Appl Power Eng, Vol. 13, No. 1, March 2024: 52-65
Int J Appl Power Eng ISSN: 2252-8792  61

Figure 11. Comparative analysis of frequency F-2 of Area-2

Figure 12. Outputs of GENCO of Area-1

Figure 13. Outputs of GENCO of Area-2

Optimization of controllers using soft computing technique for load … (Dharmendra Jain)
62  ISSN: 2252-8792

Figure 14. Comparative analysis of frequency F-1 of Area-1

Figure 15. Comparative analysis of frequency F-2 of Area-2

Figure 16. Outputs of GENCO of Area-1

Int J Appl Power Eng, Vol. 13, No. 1, March 2024: 52-65
Int J Appl Power Eng ISSN: 2252-8792  63

Figure 17. Outputs of GENCO of Area-2

4.4. Analysis in terms of time response parameters

Time-response parameters for change in frequency of Area-1, f1 for Case-1, are given in Table 3.
Time-response parameters of change in frequency of Area-2 for Case-1, are shown in Table 4. Time response
specifications for f1 are represented in Table 5 and f2 are in Table 6 for Case-II. Time response
specifications for f1 and Δf2 for Case-III, are listed in Tables 7 and 8 respectively.

Table 3. Comparison in terms of time-response parameters for frequency change of Area-1 for Case-1
S. No. Controller MP TP (Seconds) TR (Seconds) TS (Seconds) Remark
1 GA -0.71×10-4 0.27 s 0.22 s 7.82 s Stable
2 PID -3.05×10-4 1.58 s 1.17 s 8.44 s Stable
3 PSO -2.51×10-4 1.86 s 1.57 s 6.68 s Stable

Table 4. Comparison in terms of time-response parameters for frequency change of Area-2 for Case-1
S. No. Controller MP TP (Seconds) TR (Seconds) TS (Seconds) Remark
1 GA -0.73×10-4 0.28 s 0.23 s 6.95 s Stable
2 PID -1.15×10-4 0.12 s 10.26 s 11.59 s Stable
3 PSO -4.5×10-4 1.95 s 1.58 s 8.49 s Stable

Table 5. Comparison in terms of time-response parameters for frequency change of Area-1 for Case-II
S. No. Controller MP TP (Seconds) TR (Seconds) TS (Seconds) Remark
1 GA -0.51*10-4 0.28 s 0.24 s 7.71 s Stable
2 PID -3.1*10-4 1.57 s 1.19 s 14.05 s Stable
3 PSO -2.5*10-4 0.45 s 6.55 s 7.41 s Stable

Table 6. Comparison in terms of time-response parameters for frequency change of Area-2 for Case-II
S. No. Controller MP TP (Seconds) TR (Seconds) TS (Seconds) Remark
1 GA -0.73*10-4 0.28 s 0.24 s 6.2 s Stable
2 PID -1.2*10-4 0.16 s 10.0 s 11.10 s Stable
3 PSO -4.46*10-4 1.95 s 1.56 s 6.26 s Stable

Table 7. Comparison in terms of time-response parameters for frequency change of Area-1 for Case-III
S. No. Controller MP TP (Seconds) TR (Seconds) TS (Seconds) Remark
1 GA -0.72*10-4 0.27 s 0.22 s 6.83 s Stable
2 PID -3.05*10-4 1.19 s 1.58 s 8.58 s Stable
3 PSO -2.52*10-4 0.44 s 6.61 s 6.67 s Stable

Table 8. Comparison in terms of time-response parameters for frequency change of Area-2 for Case-III
Optimization of controllers using soft computing technique for load … (Dharmendra Jain)
64  ISSN: 2252-8792

S. No. Controller MP TP (Seconds) TR (Seconds) TS (Seconds) Remark

1 GA -0.72*10-4 0.28 s 0.24 s 6.84 s Stable
2 PID -1.2*10-4 0.16 s 8.08 s 8.08 s Stable
3 PSO -4.46*10-4 1.90 s 1.58 s 6.68 s Stable

5. CONCLUSION
The most important objective of the LFC to maintain connected area tie-line power and network
frequency as near to a planned values as possible. So, appropriate control strategy is required in the
interconnected deregulated power system. In a deregulated context, a system model of a two-area
interconnected electrical network for simulation has been created. in order to apply the controllers and find
their responses. The soft computing technique-based controllers like GA based controller and PSO based
controller have been developed with minimizing ITAE. Comparative responses with PID, GA and PSO based
controllers have been obtained with different contractual conditions and shown in figures. The comparative
analysis with respect to time response specifications has also been done. It has been observed that the usage
of a PID controller can enhance the system's ability to perform if parameter optimization is done properly.
Comparative analysis shows that soft computing techniques are capable of providing the proper tuning of
controller and can optimize the controller gains for load frequency control issues of liberalized power system.
Also, the PSO based controller delivers the finest response for two area restructured electricity network as
compared to other controllers used in this work.

REFERENCES
[1] V. Donde, M. A. Pai, and I. A. Hiskens, “Simulation and optimization in an AGC system after deregulation,” IEEE Transactions
on Power Systems, vol. 16, no. 3, pp. 481–489, 2001, doi: 10.1109/59.932285.
[2] M. L. Kothari, N. Sinha, and M. Rafi, “Automatic generation control of an interconnected power system under deregulated
environment,” in Proceedings of 1998 Power Quality Conference, PQ 1998, 1998, pp. 95–102, doi: 10.1109/PQ.1998.710361.
[3] W. Tan, H. Zhang, and M. Yu, “Decentralized load frequency control in deregulated environments,” International Journal of
Electrical Power & Energy Systems, vol. 41, no. 1, pp. 16–26, Oct. 2012, doi: 10.1016/j.ijepes.2012.02.013.
[4] D. Jain, M. K. Bhaskar, and M. Parihar, “Comparative Analysis of Load Frequency Control Problem of Multi Area Deregulated
Power System Using Soft Computing Techniques,” Mathematical Statistician and Engineering Applications, vol. 71, no. 4, pp.
10713–10729, 2022.
[5] D. Jain, M. K. Bhaskar, S. K. Joshi, and D. Bohra, “Analysis of Load Frequency Control Problem for Interconnected Power
System Using PID Controller,” International Journal of Emerging Technology and Advanced Engineering, vol. 4, no. 11, pp.
276–283, 2014.
[6] D. Jain, M. K. Bhaskar, and M. Kumar, “Comparative analysis of different methods of tuning the PID controller parameters for
load frequency control problem,” International Journal of Advanced Research in Electrical, Electronics and Instrumentation
Engineering, vol. 3, no. 11, pp. 13144–13154, 2014.
[7] G. T. C. Sekhar, R. K. Sahu, A. K. Baliarsingh, and S. Panda, “Load frequency control of power system under deregulated
environment using optimal firefly algorithm,” International Journal of Electrical Power & Energy Systems, vol. 74, pp. 195–211,
Jan. 2016, doi: 10.1016/j.ijepes.2015.07.025.
[8] P. Babahajiani, Q. Shafiee, and H. Bevrani, “Intelligent demand response contribution in frequency control of multi-area power
systems,” IEEE Transactions on Smart Grid, vol. 9, no. 2, pp. 1282–1291, Mar. 2018, doi: 10.1109/TSG.2016.2582804.
[9] P. K. Sahoo, P. K. Satpathy, and S. Patnaik, “Application of soft computing neural network tools to line congestion study of
electrical power systems,” International Journal of Information and Communication Technology, vol. 13, no. 2, pp. 219–226,
2018, doi: 10.1504/IJICT.2018.090559.
[10] N. Cohn, “Some Aspects of Tie-Line Bias Control on Interconnected Power Systems,” Transactions of the American Institute of
Electrical Engineers. Part III: Power Apparatus and Systems, vol. 75, no. 3, pp. 1415–1436, Jan. 1956, doi:
10.1109/AIEEPAS.1956.4499454.
[11] O. I. Elgerd and C. E. Fosha, “Optimum Megawatt-Frequency Control of Multiarea Electric Energy Systems,” IEEE Transactions
on Power Apparatus and Systems, vol. PAS-89, no. 4, pp. 556–563, Apr. 1970, doi: 10.1109/TPAS.1970.292602.
[12] C. E. Fosha and O. I. Elgerd, “The Megawatt-Frequency Control Problem: A New Approach Via Optimal Control Theory,” IEEE
Transactions on Power Apparatus and Systems, vol. PAS-89, no. 4, pp. 563–577, Apr. 1970, doi: 10.1109/TPAS.1970.292603.
[13] A. J. Connor, F. I. Denny, J. R. Huff, T. Kennedy, and C. J. Frank, “Current operating problems associated with automatic
generation control,” IEEE Transactions on Power Apparatus and Systems, vol. 98, no. 1, pp. 88–96, 1979, [Online]. Available:
http:/iaes/.
[14] O. Abedinia, M. S. Naderi, and A. Ghasemi, “Robust LFC in deregulated environment: Fuzzy PID using HBMO,” in 2011 10th
International Conference on Environment and Electrical Engineering, May 2011, pp. 1–4, doi: 10.1109/EEEIC.2011.5874843.
[15] S. M. Abd-Elazim and E. S. Ali, “Load frequency controller design of a two-area system composing of PV grid and thermal
generator via firefly algorithm,” Neural Computing and Applications, vol. 30, no. 2, pp. 607–616, Jul. 2018, doi: 10.1007/s00521-
016-2668-y.
[16] A. K. Mishra, S. R. Das, P. K. Ray, R. K. Mallick, A. Mohanty, and D. K. Mishra, “PSO-GWO Optimized Fractional Order PID
Based Hybrid Shunt Active Power Filter for Power Quality Improvements,” IEEE Access, vol. 8, pp. 74497–74512, 2020, doi:
10.1109/ACCESS.2020.2988611.
[17] M. H. Suid and M. A. Ahmad, “Optimal tuning of sigmoid PID controller using Nonlinear Sine Cosine Algorithm for the
Automatic Voltage Regulator system,” ISA Transactions, vol. 128, pp. 265–286, Sep. 2022, doi: 10.1016/j.isatra.2021.11.037.
[18] B. Dhanasekaran, J. Kaliannan, A. Baskaran, N. Dey, and J. M. R. S. Tavares, “Load Frequency Control Assessment of a PSO-
PID Controller for a Standalone Multi-Source Power System,” Technologies, vol. 11, no. 1, p. 22, Jan. 2023, doi:
10.3390/technologies11010022.

Int J Appl Power Eng, Vol. 13, No. 1, March 2024: 52-65
Int J Appl Power Eng ISSN: 2252-8792  65

[19] C. Concordia and L. K. Kirchmayer, “Tie-Line Power and Frequency Control of Electric Power Systems [includes discussion],”
Transactions of the American Institute of Electrical Engineers. Part III: Power Apparatus and Systems, vol. 72, no. 3, pp. 562–
572, Jun. 1953, doi: 10.1109/AIEEPAS.1953.4498667.
[20] Y. R. Sood, “Evolutionary programming based optimal power flow and its validation for deregulated power system analysis,”
International Journal of Electrical Power and Energy Systems, vol. 29, no. 1, pp. 65–75, Jan. 2007, doi:
10.1016/j.ijepes.2006.03.024.
[21] E. Rakhshani and J. Sadeh, “Simulation of two-area AGC system in a competitive environment using reduced-order observer
method,” in 2008 5th International Conference on the European Electricity Market, May 2008, pp. 1–6, doi:
10.1109/EEM.2008.4579082.
[22] N. Pathak, I. Nasiruddin, and T. S. Bhatti, “A More Realistic Model of Centralized Automatic Generation Control in Real-time
Environment,” Electric Power Components and Systems, vol. 43, no. 19, pp. 2205–2213, Nov. 2015, doi:
10.1080/15325008.2015.1076540.
[23] L. C. Saikia, S. Mishra, N. Sinha, and J. Nanda, “Automatic generation control of a multi area hydrothermal system using
reinforced learning neural network controller,” International Journal of Electrical Power & Energy Systems, vol. 33, no. 4, pp.
1101–1108, May 2011, doi: 10.1016/j.ijepes.2011.01.029.
[24] A. K. Pal, P. Bera, and K. Chakraborty, “AGC in two-area deregulated power system using reinforced learning neural network
controller,” in 2014 First International Conference on Automation, Control, Energy and Systems (ACES), Feb. 2014, pp. 1–6, doi:
10.1109/ACES.2014.6808004.
[25] S. Baghya Shree and N. Kamaraj, “Hybrid Neuro Fuzzy approach for automatic generation control in restructured power system,”
International Journal of Electrical Power and Energy Systems, vol. 74, pp. 274–285, Jan. 2016, doi:
10.1016/j.ijepes.2015.05.029.
[26] Y. K. Bhateshvar and H. D. Mathur, “Frequency stabilization for thermal-hydro power system with fuzzy logic controlled smes
unit in deregulated environment,” in International Conference on Advanced Computing and Communication Technologies,
ACCT, Feb. 2014, pp. 536–540, doi: 10.1109/ACCT.2014.58.
[27] Y. Arya and N. Kumar, “Fuzzy Gain Scheduling Controllers for Automatic Generation Control of Two-area Interconnected
Electrical Power Systems,” Electric Power Components and Systems, vol. 44, no. 7, pp. 737–751, Apr. 2016, doi:
10.1080/15325008.2015.1131765.
[28] R. Shankar, K. Chatterjee, and R. Bhushan, “Impact of energy storage system on load frequency control for diverse sources of
interconnected power system in deregulated power environment,” International Journal of Electrical Power and Energy Systems,
vol. 79, pp. 11–26, Jul. 2016, doi: 10.1016/j.ijepes.2015.12.029.
[29] M. Deepak and R. J. Abraham, “Load following in a deregulated power system with Thyristor Controlled Series Compensator,”
International Journal of Electrical Power & Energy Systems, vol. 65, pp. 136–145, Feb. 2015, doi: 10.1016/j.ijepes.2014.09.038.
[30] M. Zribi, M. Al-Rashed, and M. Alrifai, “Adaptive decentralized load frequency control of multi-area power systems,”
International Journal of Electrical Power & Energy Systems, vol. 27, no. 8, pp. 575–583, Oct. 2005, doi:
10.1016/j.ijepes.2005.08.013.
[31] R. Mok and M. A. Ahmad, “Fast and optimal tuning of fractional order PID controller for AVR system based on memorizable-
smoothed functional algorithm,” Engineering Science and Technology, an International Journal, vol. 35, p. 101264, Nov. 2022,
doi: 10.1016/j.jestch.2022.101264.
[32] H. Shaikh, N. Kulkarni, and M. Bakshi, “Computational Analysis of PID and PSO-PID Optimization for MIMO Process Control
System,” in Computational Intelligence, Singapore: Springer, 2023, pp. 663–676.

BIOGRAPHIES OF AUTHORS

Dharmendra Jain is a research scholar at M.B.M University, Jodhpur,

Rajasthan, India since 2017. He is former Head of Department of Electrical Engineering,
Examination in-charge and TPO at Mahaveer Institute of Technology and Science, Jadan,
Rajasthan. He has also been Dean Academics at Mahaveer Institute of Technology and
Science, Jadan, Pali, Rajasthan. He received his Bachelor of Engineering and Master of
Engineering degree from M.B.M Engineering College, Jai Narain Vyas University, Jodhpur,
Rajasthan, India. He has published over twelve research papers on his area of expertise. His
areas of interest are power system, control system and soft computing techniques i.e. genetic
algorithm, particle swarm optimization technique. He can be contacted at email:
[email protected].

Dr. M. K. Bhaskar is currently working as Professor in Department of Electrical

Engineering, M.B.M. Engineering College, M.B.M. University, Jodhpur, Rajasthan, India. He
is having more than 31 years of teaching and research experience. He has guided many Ph.D.
and PG students. He has published more than fifty research papers. He has done B.E. with
honors in Electrical Engineering and M.Tech. in electrical engineering and Ph.D. from
M.B.M. Engineering college, JNVU, Jodhpur, Rajasthan. His main research directions include
power system, control system, soft computing techniques i.e. neural networks, fuzzy logic,
genetic algorithm, particle swarm optimization technique. He can be contacted at email:
[email protected].

Optimization of controllers using soft computing technique for load … (Dharmendra Jain)