BET (HONS) IN ELECTRICAL

STUDY ON NATURE INSPIRED OPTIMIZATION TECHNIQUES FOR

LOAD FREQUENCY CONTROL (LFC) USING FRACTIONAL
ORDER (FOPID) CONTROLLER

NOR FATIN BINTI ABD LATIFF

51211216066

UNIVERSITI KUALA LUMPUR

2018

JULY 2018
STUDY ON NATURE INSPIRED OPTIMIZATION TECHNIQUES
FOR LOAD FREQUENCY CONTROL (LFC) USING
FRACTIONAL ORDER (FOPID) CONTROLLER

NOR FATIN BINTI ABD LATIFF

51211216066

Report Submitted to Fullfill the Partial Requirements

For the Bachelor of Engineering Technology (Hons.) in
Electrical
Universiti Kuala Lumpur

JULY 2018
i
 DECLARATION

I declare that this report is my original work and all references have been
cited adequately as required by the university.

Date: ___________ Signature: …………………….

(Hardcover submission date) Full Name: Nor Fatin Binti Abd Latiff
ID Number: 51211216066

ii
 APPROVAL PAGE

I have supervised and examined this report and verify that it meets the
program and university’s requirements for the Bachelor of Engineering
Technology (Hons.) in Electrical.

Date: ___________ Signature: …………………….

(Hardcover submission date) Supervisor: Dr Kanendra Naidu S/O
Vijyakumar

Official Stamp:

iii
 ACKNOWLEDGEMENT

First and foremost, it is my sincerest delight to those who are involved in

guiding and assisting me to finish this research project. I would like to give
my sincerest gratitude to the lecturers of Electrical Section University Kuala
Lumpur British Malaysian Institute for providing excellent environment and
equipment for me to finish this project

Special gratitude to my project supervisor, Dr Kanendra Naidu S/O

Vijyakumar for his dedication and endless efforts in giving advice and
guidance throughout the entirety of this project. His dedication and motivation
as a lecturer lead to the success of this project.

Last but not least I would like to thank my family and my fellow friends
for their motivational words, moral and financial support during the process of
completing this project. Finally, my finest gratitude goes to those who are
involve direct or indirectly in the completion of this project. Without their
support this project would not have been done successfully.

iv
 ABSTRACT

This research presents a comparative study on nature inspired optimization

technique for Load Frequency Control (LFC) in a reheat thermal system
using Firefly Algorithm (FA) and Cuckoo Search (CS). Load Frequency
Control (LFC) is an important control element in the synchronous generator
which is tasked to damp frequency oscillation during load changes. LFC
plays a very important role in providing quality power in the case of isolated
as well as interconnected power systems. The objective of this project is to
design a single area of reheat thermal system model and to apply the nature
inspired optimization techniques into the system in order to find the optimal
solution for the controller used to get the least error by observing the Integral
Time Squared Error (ITSE) of the system. This research covers single area
of reheat thermal system which uses a Fractional order controller (FOC) and
also uses two well-known nature inspired optimization techniques. The
Fractional order controller (FOC) possess important properties such as,
eliminating steady state error, robustness toward plant gain variations and
also good disturbance rejection. Thus, the FOC based PID controller
(FOPID) will be beneficial to the power system as it will ensure good system
performance. The fitness function Integral Time of Squared Error is chosen to
evaluate the error of an area because it shows more advantageous
performance compared to the other fitness function such as ITAE and IAE. In
this study, a comparative analysis is carried out using the Cuckoo Search
(CS) algorithm and Firefly Algorithm (FA) technique to tune the FOPID
controller of the LFC for a single area interconnected reheat thermal system.
The frequency response and tie line power should be observed in terms of
settling time and overshoot/undershoot to evaluate the performance of the
tuned FOPID controller and to give the better performance.

v
 TABLE OF CONTENTS

Contents Page No
DECLARATION ii
APPROVAL PAGE iii
ACKNOWLEDGEMENT iv
ABSTRACT v
TABLE OF CONTENTS vi
LIST OF FIGURES ix
LIST OF TABLES xiv
LIST OF ABBREVIATIONS xvii

CHAPTER 1: INTRODUCTION
1.1 Background of the Project 1
1.2 Problem Statement 2
1.3 Objectives of the Project 3
1.4 Significance of the Project 3
1.5 Scopes and Limitations of the Project 4
1.6 Chapter Summary 4

CHAPTER 2: LITERATURE REVIEW

2.1 Introduction 5
2.2 Power System 5
2.2.1 Frequency 6
2.2.2 The Changes Frequency in Power System 6
2.3 Control System 6
2.3.1 Load Frequency Control 7
2.4 Classical Control Techniques 9
2.4.1 LQR Based Techniques 9

vi
 2.4.2 Fractional Calculus and Fractional Order PIDµλ 10
(FOPID) Controller Based Techniques

2.5 Nature Inspired Optimization Techniques 12

2.5.1 Firefly Algorithm 14
2.5.2 Cuckoo Search Algorithm 15
2.5.2.1 Cuckoo Breeding Behavior 16
2.5.2.2 Levy Flights 16
2.6 Gantt Chart 17
2.6.1 Final Year Project 1 17
2.6.2 Final Year Project 2 18
2.7 Chapter Summary 19

CHAPTER 3: METHODOLOGY
3.1 Introduction 20
3.2 Research Plan 21
3.3 MATLAB and Simulink Software 22
3.3.1 Power System Modelling 23
3.3.2 Modelling of the System 26
3.4 Load Frequency Control using FOPID Controller 27
3.4.1 Implementing FOPID Controller for Load 27
Frequency Control
3.5 Optimal Tuning of FOPID Controller using Firefly 28
Algorithm and Cuckoo Search Algorithm
3.5.1 Firefly Algorithm 29
3.5.2 Cuckoo Search Algorithm 32
3.6 Integral Time Square Error (ITSE) 35
3.7 Chapter Summary 35

CHAPTER 4: RESULTS AND DISCUSSIONS

4.1 Introduction 36
4.2 Case 1: Application of Load 37
4.3 Case 2: Application of Governor and Turbine 57

vii
4.4 Case 3: Application of Fitness Function 73
4.5 Comparison of Performance Between Firefly Algorithm 89
and Cuckoo Search Algorithm
4.6 Discussions 92
4.7 Chapter Summary 97

CHAPTER 5: CONCLUSIONS AND RECOMMENDATIONS

5.1 Introduction 98
5.2 Conclusions 98
5.3 Recommendations 99

REFERENCES 100

APPENDICES
Appendix A 103
Appendix B 104
Appendix C 107

viii
 LIST OF FIGURES

Figure No Descriptions Page No

Figure 2-1 Model Concept of Automatic Generation 7
Control (AGC)
Figure 2-2 Block Diagram of FOPID Controller 11
Figure 2-3 Formulation of an Optimization Problem 12
Figure 2-4 Illustration of Firefly Algorithm 15
Figure 3-1 Research Plan 21
Figure 3-2 Modelling of a Single Area Reheat Thermal 26
System
Figure 3-3 Example of Step Response Waveform 27
Figure 3-4 Block Diagram of FOPID Controller 28
Figure 3-5 Flowchart of Firefly Algorithm 31
Figure 3-6 Flowchart of Cuckoo Search Algorithm 34
Figure 4-1 Firefly Algorithm Frequency Response for Load 37
Value = 0.1
Figure 4-2 Firefly Algorithm Convergence graph for Load 38
Value = 0.1
Figure 4-3 Firefly Algorithm Frequency Response for Load 39
Value = 0.05
Figure 4-4 Firefly Algorithm Convergence Graph for Load 40
Value = 0.05
Figure 4-5 Firefly Algorithm Frequency Response for Load 41
Value = 0.15
Figure 4-6 Firefly Algorithm Convergence Graph for Load 42
Value = 0.15
Figure 4-7 Firefly Algorithm Frequency Response for Load 43
Value = 0.075
Figure 4-8 Firefly Algorithm Convergence graph for Load 44

ix
 Value = 0.075
Figure 4-9 Firefly Algorithm Frequency Response for Load 45
Value = 0.125
Figure 4-10 Algorithm Convergence Graph for Load value 46
= 0.125
Figure 4-11 Cuckoo Search Algorithm Frequency 47
Response for Load = 0.1
Figure 4-12 Cuckoo Search Algorithm Convergence Graph 48
for Load Value = 0.1
Figure 4-13 Cuckoo Search Algorithm Frequency 49
Response for Load Value = 0.05
Figure 4-14 Cuckoo Search Algorithm Convergence Graph 50
for Load Value = 0.05
Figure 4-15 Cuckoo Search Algorithm Frequency 51
Response for Load Value = 0.15
Figure 4-16 Cuckoo Search Algorithm Convergence Graph 52
for Load Value = 0.15
Figure 4-17 Cuckoo Search Algorithm Frequency 53
Response for Load Value = 0.075
Figure 4-18 Cuckoo Search Algorithm Convergence Graph 54
for Load Value = 0.075
Figure 4-19 Cuckoo Search Algorithm Frequency 55
Response for Load Value = 0.125
Figure 4-20 Cuckoo Search Algorithm Convergence Graph 56
for Load Value = 0.125
Figure 4-21 Firefly Algorithm Frequency Response with 57
+50% Efficiency of Governor and Turbine
Figure 4-22 Firefly Algorithm Convergence Graph with 58
+50% Efficiency of Governor and Turbine
Figure 4-23 Firefly Algorithm Frequency Response with 59
+25% Efficiency of Governor and Turbine
Figure 4-24 Firefly Algorithm Convergence Graph with 60
+25% Efficiency of Governor and Turbine

x
Figure 4-25 Firefly Algorithm Frequency Response with - 61
50% Efficiency of Governor and Turbine
Figure 4-26 Firefly Algorithm Convergence Graph with - 62
50% Efficiency of Governor and Turbine
Figure 4-27 Firefly Algorithm Frequency Response with - 63
25% Efficiency of Governor and Turbine
Figure 4-28 Firefly Algorithm Convergence Graph with - 64
25% Efficiency of Governor and Turbine
Figure 4-29 Cuckoo Search Algorithm Frequency 65
Response with +50% Efficiency of Governor
and Turbine
Figure 4-30 Cuckoo Search Algorithm Convergence Graph 66
with +50% Efficiency of Governor and Turbine
Figure 4-31 Cuckoo Search Algorithm Frequency 67
Response with +25% Efficiency of Governor
and Turbine
Figure 4-32 Cuckoo Search Algorithm Convergence Graph 68
with +25% Efficiency of Governor and Turbine
Figure 4-33 Cuckoo Search Algorithm Frequency 69
Response with -50% Efficiency of Governor
and Turbine
Figure 4-34 Cuckoo Search Algorithm Convergence Graph 70
with -50% Efficiency of Governor and Turbine
Figure 4-35 Cuckoo Search Algorithm Frequency 71
Response with -25% Efficiency of Governor
and Turbine
Figure 4-36 Cuckoo Search Algorithm Convergence Graph 72
with -25% Efficiency of Governor and Turbine
Figure 4-37 Firefly Algorithm Frequency Response with 73
ITSE Fitness Function
Figure 4-38 Firefly Algorithm Convergence Graph with 74
ITSE Fitness Function
Figure 4-39 Firefly Algorithm Frequency Response with 75

xi
 ITAE Fitness Function
Figure 4-40 Firefly Algorithm Convergence Graph with 76
ITAE Fitness Function
Figure 4-41 Firefly Algorithm Frequency Response with IAE 77
Fitness Function
Figure 4-42 Firefly Algorithm Convergence Graph with IAE 78
Fitness Function
Figure 4-43 Firefly Algorithm Frequency Response with ISE 79
Fitness Function
Figure 4-44 Firefly Algorithm Convergence Graph with ISE 80
Fitness Function
Figure 4-45 Cuckoo Search Algorithm Frequency 81
Response with ITSE Fitness Function
Figure 4-46 Cuckoo Search Algorithm Convergence Graph 82
with ITSE Fitness Function
Figure 4-47 Cuckoo Search Algorithm Frequency 83
Response with ITAE Fitness Function
Figure 4-48 Cuckoo Search Algorithm Convergence Graph 84
with ITAE Fitness Function
Figure 4-49 Cuckoo Search Algorithm Frequency 85
Response with IAE Fitness Function
Figure 4-50 Cuckoo Search Algorithm Convergence Graph 86
with IAE Fitness Function
Figure 4-51 Cuckoo Search Algorithm Frequency 87
Response with ISE Fitness Function
Figure 4-52 Cuckoo Search Algorithm Convergence Graph 88
with ISE Fitness Function
Figure 4-53 Cuckoo Search Algorithm Frequency 93
Response with Load Value = 0.05
Figure 4-54 Cuckoo Search Algorithm Frequency 95
Response with -50% Efficiency of Governor
and Turbine

xii
Figure 4-55 Cuckoo Search Algorithm Frequency 97
Response with ITSE Fitness Function

xiii
 LIST OF TABLES

Table No Descriptions Page No

Table 2-1 Comparisons Between PID Controller and 11
FOPID Controller
Table 2-2 Nature Inspired Optimization Techniques 13
Table 2-3 Gantt Chart FYP 1 17
Table 2-4 Gantt Chart FYP 2 18
Table 3-1 Simulink Blocks 24
Table 4-1 System Parameter Values 37
Table 4-2 Step Info 37
Table 4-3 System Parameter Values 39
Table 4-4 Step Info 39
Table 4-5 System Parameter Values 41
Table 4-6 Step Info 41
Table 4-7 System Parameter Values 43
Table 4-8 Step Info 43
Table 4-9 System Parameter Values 45
Table 4-10 Step Info 45
Table 4-11 System Parameter Values 47
Table 4-12 Step Info 47
Table 4-13 System Parameter Values 49
Table 4-14 Step Info 49
Table 4-15 System Parameter Values 51
Table 4-16 Step Info 51
Table 4-17 System Parameter Values 53
Table 4-18 Step Info 53
Table 4-19 System Parameter Values 55
Table 4-20 Step Info 55
Table 4-21 System Parameter Values 57

xiv
Table 4-22 Step Info 57
Table 4-23 System Parameter Values 59
Table 4-24 Step Info 59
Table 4-25 System Parameter Values 61
Table 4-26 Step Info 61
Table 4-27 System Parameter Values 63
Table 4-28 Step Info 63
Table 4-29 System Parameter Values 65
Table 4-30 Step Info 65
Table 4-31 System Parameter Values 67
Table 4-32 Step Info 67
Table 4-33 System Parameter Values 69
Table 4-34 Step Info 69
Table 4-35 System Parameter Values 71
Table 4-36 Step Info 71
Table 4-37 System Parameter Values 73
Table 4-38 Step Info 73
Table 4-39 System Parameter Values 75
Table 4-40 Step Info 75
Table 4-41 System Parameter Values 77
Table 4-42 Step Info 77
Table 4-43 System Parameter Values 79
Table 4-44 Step Info 79
Table 4-45 System Parameter Values 81
Table 4-46 Step Info 81
Table 4-47 System Parameter Values 83
Table 4-48 Step Info 83
Table 4-49 System Parameter Values 85
Table 4-50 Step Info 85
Table 4-51 System Parameter Values 87
Table 4-52 Step Info 87
Table 4-53 System Parameter Values Case 1 89
Table 4-54 System Parameter Values Case 2 90

xv
Table 4-55 System Parameter Values Case 3 91
Table 4-56 Step Info for Cuckoo Search Algorithm with 93
Load Value = 0.05
Table 4-57 Step Info for Cuckoo Search Algorithm with the 95
Governor and Turbine -50%
Table 4-58 Step Info for Cuckoo Search Algorithm with 96
Fitness Function ITSE

xvi
 LIST OF ABBREVIATIONS

Abbreviation Descriptions
LFC - Load Frequency System
AVR - Automatic Voltage Regulator
FA - Firefly Algorithm
CSA - Cuckoo Search Algorithm
FOPID - Fractional Order PID
MATLAB - Matrix Laboratory

xvii
 CHAPTER 1

INTRODUCTION

1.1 Background of the Project

An electric power system is a system that implement a district's homes and

industry with electrical power for a sizeable region. This power system is well
known as the grid and can be extensively separated into three categories,
generators that provides the power, the transmission system that carries the
power between the generating centers and the load centers and the
distribution system that delivers the power to neighbouring homes and
industries. Mini power systems are likewise discovered in hospitals, industry,
homes and commercial buildings. In particular, majority of these systems
depends upon three-phase AC power which is the criterion for big scale
power transmission and distribution globally. Some particular power systems
that does not often depend on three-phase AC power are usually found in
aircraft, electric rail systems, ocean liners and automobiles.

Power generation is a mission to create electric power from sources of

main energy. For electric utilities, it is the primary method in the distribution of
power to consumers. Different procedures, for example, transmission,
distribution, energy storage and recovery utilizing pumped storage process
are typically done by the electrical power industry. Power is mostly generated
at a power station by electromechanical generators, chiefly controlled by heat
engines fuelled by combustion or nuclear fission additionally by various goal
for instance, kinetic energy of flowing water and wind. The problem emerges
from controlling the output of the real power of a generating component due
to fluctuating system frequency and exchanging of tie-line power inside

1
particularize boundary, is known as load frequency control (LFC). In a power
system, LFC is known to be the key part as it will determine the performance
of the power system. Most of the time an Integral Controller can be seen
used in common LFC. However, the major downside of using the Integral
Controller is that the integral gain of the controller restraints the dynamic
operation of the system. Substantial instability and oscillations might happen
from the high gain which declines the system's performance. Along this line,
the integral gain must be of a value which gives a bargain amidst a wanted
temporary recovery and small overshoot in the dynamic performance of the
whole system.
Many methodologies to attune the gain of the fixed parameters of the
PI controller were done in research and along the inclining in magnitude and
advanced power systems, there is a possibility that the system oscillation will
cultivate within wide-area which could result into an increased broad-area
blackout. Thus, modern control strategies were used in LFC such as optimal
control and adaptive control. Nonetheless, to utilize these type of control
strategies, information on the condition of the systems or systematic online
identifier is needed. It is probably difficult to use these methods if there are
shortage of info on these methods.

1.2 Problem Statement

Generally, it is known that major scale power systems are operate by using
the interconnected subsystems. The tie lines required to be set to connect
the linking of control areas. Every area obtaining its own particular generator.
These areas are required to be able to sustain their own load besides
planned interchanges with the nearby areas. Generally, every power system
works under the condition that Generation plus Transmission losses must
equal to the connected load at rated frequency. Theoretically, if there are any
changes of the real power, it can affect the frequency of the system together
with the mechanical input power. The mechanical input power act to
generates the use of control the frequency of output electrical power. Hence,
Load Frequency Control (LFC) is a standout amongst the main issues in
electrical power system design/operation and is becoming much more

2
 noteworthy lately with expanding in size, changes in structure and complexity
in interconnected power system. Due to the loading of a given power system
is inconsistent from time to time. In order to ensure the quality of power
supply, it is crucial to improve the system performance and stability. To
achieve optimum performance of power system when there are sudden area
load changes, the existence of a load frequency controller is essential to
keep the system frequency at the desired nominal value. Here also, using the
conventional PID controller is not very efficient to ensure the system stability.
PID controller can only come out with gain Kp, Ki and Kd. So, the number of
parameters that need to be evaluated is low and also the PID controller is not
very flexible due to it.

1.3 Objectives of the Project

The point of this exploration is to build up an investigation on nature inspired

optimization techniques for Load Frequency Control (LFC) utilizing Fractional
Order PID (FOPID) controller:
i. To model the single area interconnected reheat thermal system with
FOPID controller in a power system.
ii. To integrate the nature inspired optimization technique in the FOPID for
LFC model.
iii. To complete a relative examination utilizing the Cuckoo Search
Algorithm (CSA) and Firefly Algorithm (FA) strategy to tune the FOPID
controller of the LFC.

1.4 Significance of the Project

For this field, this project is much important for the stability of the electric
power system. The optimal tuning of Load Frequency Controller can ensure
the power system to run efficiently. Optimal tuning of Load Frequency
Controller can also be use in future for further analysis. Moreover, since this
project only focused on a single area, it actually can be further extended to
two, three or more areas of multiple generators on Load Frequency Control.

3
1.5 Scopes and Limitations of the Project

i. Analysis of Load Frequency Control is based on MATLAB and Simulink

software.
ii. Optimization of Load Frequency Control is focused on a single area of
reheat thermal system.
iii. Covers two common nature inspired optimization technique which is
Firefly Algorithm and Cuckoo Search Algorithm.
iv. Optimization based on FOPID controller tuning.

1.6 Chapter Summary

Overall in this chapter, a brief explanation on the problem that triggered the
need to this study is explained thus leading to the objective of this project to
solve the problem arose. The scope and limitations are also explained that is
helpful to further point on the disadvantage and advantages of this study for
further research.

4
 CHAPTER 2

LITERATURE REVIEW

2.1 Introduction

Generally, a power system is the interconnection of more than one control

areas through tie lines. The generators in a control area always vary their
speed together (speed up or slow down) for maintenance of frequency and
the relative power angles to the predefined values in both static and dynamic
conditions. If there is any sudden load change occurs in a control area of an
interconnected power system then there will be frequency deviation as well
as tie line power deviation.

2.2 Power System

Power system is a combination of generation, transmission and distribution

networks and loads. Every system has active and reactive power
requirements because every system has different loads. The loads also differ
continuously and randomly. In power system, there are two major parts which
are the frequency and voltage. Changes in real power can affected the
frequency while changes in reactive power can affected the voltage. Making
and stabilize the frequency and voltage is the main constraint on a system. In
a power system, the load varies from time to time. As the load changes, the
frequency and voltage also differ [1].

5
2.2.1 Frequency

In power plant, frequency is a crucial existence along with the voltage. They
play a major role. By adjusting active power at generation side, the frequency
can be controlled. In crucial condition, from the load side, the active power
can be controlled. When the frequency at one end of the system and the
other end of the system is the same, frequency is said to remains the same
throughout the system.

2.2.2 The Changes of Frequency in Power System

The frequency is stated to be inversely proportional to the active power

demand at the load side.

i. The active power at the generation station need to be increased when

the total active power demand at the load side is greater than the total
active power of the generation. It is to prevent frequency from
decreases at generation side.
ii. The active power at the generation station need to be reduced when
the total active power demand at the load side is lesser than the total
active power of the generation. It is to prevent frequency from
increases at the generation side.
iii. If the frequency is constant, there is no need to change the active
power [2].

2.3 Control System

Because the load varies from time to time, the value of power generation and
load consumption are imbalance. The high load variation can lead to power
blackout and power outages in interconnected power system. It is very
important to keep the stability of the power system by regulating the active
power in order to maintain the frequency. This is where control system is
needed.

6
2.3.1 Load Frequency Control

In a common type power system, majority typical sources for power produces
are like thermal, nuclear power, hydro, diesel, natural gas etc. The main
function of a power system is to counterpart the real and reactive power to
generation to the load including losses. Automatic Generation Control (AGC)
is the term used to define the problem of controlling the real power output of
each generating units with the respect to the system frequency deviation and
tie line power exchange within prescribed limits. The problem of controlling
the reactive power balance in the system is often referred to as Excitation
Control [3].

Figure 2- 1: Model concept of Automatic Generation Control (AGC)

One of the elements of AGC is load frequency control and this

segment of AGC is a critical control issue to look into. Frequency deviation
from its typical esteem is primary factor which decides the reliability and
stability of a power framework. For this situation of coordinated power frame
work the issue turns out to be worse. In LFC issue frequency and tie-line
control trade are factors that should be pay attention to. Conserving
frequency and power trade between associated control areas at planned
quantities are the significant key objectives of LFC. Measuring a control error
signal, called the area control error (ACE) which goes about as the instable

7
true power amongst the generation is a direct incorporation of net exchange
and frequency variations can lead to accomplishing the objectives of LFC.

The extending number of a critical power and framework power

outages that have been educated about late years demonstrates these days,
the accessible working force framework require more wary idea of a wide
range of framework precariousness and control issues. The system
framework power outages exhibit that to upgrade the general power
framework control reaction, it is basic to give all the more convincing and
robust control methodologies with a particular true objective to achieve
another trade of productivity and powerful robustness. [4]

Load Frequency Control has grabbed significance with the

improvement of power systems. Electrical power systems end up being
progressively complex because of increment in request, in this way reliable
and steady power is required. Standard working state is portrayed by
enduring frequency and voltage layout with certain power system reliability.
Incompatibility during deviant conditions will be adjusted by applying LFC
which are basic in the working of electrical power system. Interconnected
power system believes to be control areas which are related by methods for
tie-lines, with gathering of all generators. In the midst of neighbourhood
varieties, frequency of the power system should be predictable which
depends on active power. Variety in active power ask for in one system will
impact the entire power system [5].

The first plan to control the frequency was by methods for the flywheel
governor of the synchronous machine and was observe to be inadequate. By
then a supplementary control was connect to the governor by methods for a
signal particularly proportional to the frequency deviation plus its integral.
This proportional plus integral control scheme shapes the conventional
method to manage the automatic generation control of power system.

In contemplation to bring back the frequency and tie-power to their

specific scheduled values, most by far of the utilities needs to utilize FOPID
controller in their respective power system. In complex deregulated structure,
classical control methods are lacking to control the sudden varieties in this

8
kind of power system. In like manner this particular circumstance, LFC issues
in the unstable power system is a difficult mission and researchers are as yet
finding the best solution in this sector. [3]

2.4 Classical Control Techniques

Standard control procedures for LFC issues are ones that take the integral of
a control error as the control signal. In the established control methods, to
procure the wanted gain and phase margins. Bode and Nyquist diagrams
and furthermore root locus are usually utilized. Hence, the power system
process of classic techniques for LFC issues are basic, direct and agreeable
for application. Regardless, the examination done by using these
methodologies reveal that they indicate poor dynamic execution, especially
inside seeing other destabilizing impacts, for instance, parameter varieties
and nonlinearities. [6]

2.4.1 LQR Based Techniques

A special type of optimal control that deals with linear system (in state and in
control) and minimization of fitness function that are quadratic or the
quadratic performance is Linear Quadratic Regulator (LQR) problem [6]. LQR
control in 2011 is said by Singh and Gupta to be considered in scalar system
when the sensor, controller and actuator all have their own clocks that may
drift apart from each other. Both an affine and a quadratic clock model also
considered as LQR control. An analyzation was made by Singh and Gupta
regarding the loss of performance incurred as a function of how
asynchronous the clocks are for a quadratic cost function. After a while, the
guidelines on how often to utilize the communication resources to
synchronize the clocks were obtained by them [7].

9
2.4.2 Fractional calculus and Fractional order PI𝛌𝑫µ (FOPID) Controller
Based Techniques

Fractional order PID (FOPID) controller was proposed for load frequency
control (LFC) in an interconnected power system. This controller had five
parameters to be tuned; thus, it provided two more degrees of freedom in
comparison with the conventional PID [8]. This is not a new thing nor new
concept. It is probably the earliest (19th century) systematic studies have
been done by Liouville, Riemann, and Holmgren. Since then, a few numbers
of definitions have been introduced in the literature for it, from which one is
selected to be presented here. There are few properties of FOPID that could
help to get better performance which are, robustness toward plant gain
variations, eliminating steady state error, and also a good disturbance
rejection. Here, it can be concluded that using Fractional Order PID (FOPID)
controller is actually quite good and suitable for the Load Frequency Control
(LFC) [14].

Podlubny was the first person who proposed the concept of FOPID
controller [9] [10]. Between FOPID and the conventional PID controller, the
difference is that in FOPID, the order of derivative and integral is not integer.
Because of this characteristics, FOPID controller provides extra degrees of
freedom in getting the optimal tuning of the controller. From here, it can also
lead to better performance in comparison with conventional PID [11]. As the
interest towards improving the conventional PID controller has increase,
FOPID controller has received attention in the last few years [12]. Below
shows the formulation of FOPID controller:
𝐾
𝐾(𝑆) = 𝐾𝑝 + 𝑆 𝐼 + 𝐾𝑑 𝑆 µ [13]
𝜆

where Kp, KI, and Kd are proportional, integral, and derivative gain,
respectively.While λ and μ are orders of integral and derivative, respectively.

10
 Figure 2- 2: Block Diagram of FOPID Controller

Table 2- 1: Comparisons between PID controller and FOPID controller

PID Controller FOPID Controller

• The parameters of PID include • The parameters of FOPID
the gains of the proportional, include proportionality gain,
integral and derivative terms. integral gain, derivative gain, λ,
and µ.
• Design of conventional PID • The design of (PIλDµ) controller
controller needs only three needs to optimize five
parameters to optimize. parameters.
• Task of designing PID controller • Task of designing FOPID
is not very challenging. controller is more challenger.
• The conventional PID controller • The new and good performance
is less efficient and not very extension for FOPID can be
good in performance. provided by Fractional calculus
because of the flexibility order of
fractional calculus.

11
2.5 Nature Inspired Optimization Technique

Numerous computational insight-based procedures have been utilized in the

design of Load Frequency Control (LFCs) also. Worldwide advancement
method using developmental and swarm knowledge has been utilized to tune
the PID controller parameters in different writings. Nature-Inspired
Optimization Algorithms gives a deliberate prologue to all significant
nature-propelled calculations for improvement. The book's bound together
approach, adjusting calculation presentation, hypothetical foundation and
useful execution, supplements broad writing with well- picked case studies to
show these calculations function (Xin Che Yang, 2014). Utilizing optimization
procedure is fundamentally to scan for optimality. The fundamental errand is
to discover the ideal arrangements by some arrangement technique utilizing
the privilege scientific techniques.

Figure 2- 3: Formulation of an optimization problem

12
 Table 2- 2: Nature inspired optimization techniques

Types of Nature Inspired Description

Optimization Technique

Ant and Bee Algorithm Combines the characteristics of Ant Colony

and Artificial Bee Colony (ABC) algorithms to
optimize feature selection.

Particle Swarm Enlivened by the running and tutoring

Optimization (PSO) examples of feathered creatures and fish,
Particle Swarm Optimization (PSO) was
developed by Russell Eberhart and James
Kennedy in 1995.

Firefly Algorithm (FA) Idealize by the flashing characteristics of

fireflies as to develop firefly-inspired
algorithms.

Bat Algorithm (BA) Based on the echolocation behaviour of

micro bats with varying pulse rates of
emission and loudness.

Cuckoo Search (CS) The most recent nature-enlivened

metaheuristic calculations. In light of the
brood parasitism of some cuckoo species.

Genetic Algorithm (GA) Genetic Algorithm (GA) are likely the most
famous developmental calculations regarding
the assorted variety of their application. A
larger part of surely understood enhancement
issues have been attempted by genetic
algorithm.

13
2.5.1 Firefly Algorithm

The flashing light of fireflies is a stunning sight in the summer sky in the
tropical and mild regions. There are around two thousand firefly species, and
majority fireflies create short and rhythmic flashes. The pattern of flashes is
frequently extraordinary for specific species. The flashing light is created by a
process of bioluminescence, and the verifiable purpose of such signaling
systems are still debated. Meta heuristic optimization algorithm which is
biologically inspired by the behavior of fireflies is the Firefly Algorithm. Be that
as it may, two essential elements of such flashes are to draw in mating
partners (communication), and to lure potential prey. Likewise, flashing may
also assist as a defensive cautioning instrument. The rhythmic flash, the rate
of flashing and the quantity of time from some part of the signal system that
unites both genders together. Females react to a male's uncommon pattern
of flashing in the same species, while in other type of species such as
photuris, female fireflies can copy the mating flashing pattern of other
different species in order to bait and eat the male fireflies who may
mistakenly consider the flashes as a potential suitable mate. The flashing
light can be formulated in such a way that it is related with the objective
function to be optimized, which makes it conceivable to plan a new
optimization algorithm [16].

A description about Firefly Algorithm is, it uses the fireflies as the

nature based interacting agents [15]. There are few rules that need to abide
so that this algorithm is established. The rules are based on the following:

i. This creature fireflies are unisex, they do not have gender male or
female. Because of that, fireflies are attracted to one another.
ii. As the brightness is increases, the attractiveness of the fireflies also
increases. They are directly proportional to each other. Thus, all these
fireflies easily attracted to move to the place with higher level of
brightness. While for distance and brightness, they are inversely
proportional to each other.
iii. Here, the objection function value or fitness function value is
represented by the level of the brightness.
Function of distance(r) is represented as follow:
14
 β(r) = β0 𝑒 −𝛾𝑟2 [17]

where β0 = attractiveness, γ = fixed light absorption coefficient and r = 0.

All these are presented as the parameters of the Firefly Algorithm [17]. In
addition, all the fireflies are characterized by its light intensity which can
be directly expressed as inverse of a cost function. Two factors are taken
into account, which is the first one the attractiveness of other members
with higher light intensity and fixed random step vector. If only such
randomized step is being used, no brighter firefly can be found [17].

Figure 2- 4: Illustration of Firefly Algorithm

2.5.2 Cuckoo Search Algorithm

Cuckoo Search (CS) was presented in 2009, and it has pulled in awesome
attention because of its promising proficiency in taking care of numerous
advancement issues and real-world applications. Over the latest couple of
years, numerous papers have been appropriated with respect to cuckoo
search, and the relevant writing has broadened basically. This part abridges
quickly most of the writing about cuckoo look in peer-reviewed diaries and
meetings discovered up until now. These references can be systematically
grouped into fitting classes, which can be utilized as a reason for further
inquire about.

15
2.5.2.1 Cuckoo Breeding Behavior

The Cuckoo Search was described as an obligate brood parasitism. This

parasite are some cuckoo species that lay their eggs in the nests of host
birds. Female parasite cuckoos can change colours and patterns of the eggs
of some of the chosen host species. They evolved in such a way that the
female parasitic cuckoos can end up do that. Here, it can decrease the
probability of their eggs getting lost and also at the same time can increases
their system of reproduction [19]. Then, some of the host birds that involved
has direct conflict with the intruding cuckoos. For this case, if let say the host
birds found that the eggs in their nests are not actually theirs, the host birds
will choose either to put them away or just abandon their nests and started
new ones at anywhere else. Mostly these parasitic cuckoos choose a nest
where the nest just got eggs laid by the host bird. Generally, to compare with
host eggs, the cuckoo eggs hatch slightly earlier than their host eggs. His first
instinct action is to evict the host eggs by blindly propelling the eggs out of
the nest right after the first cuckoo female is hatched. This resulting on
increasing of the cuckoo chick’s share of food that is provided [19]. In
addition, a study on Cuckoo Search shows that a cuckoo chick can copy the
call of the host chicks. Here, it can gain access to more feeding opportunity.
This Cuckoo Search model can be used to many types of optimization
problems. Research by Yang and Deb shows that using Lévy Flights can
improve the performance better compare with a simple random walk [20]
[21].

2.5.2.2 Lévy Flights

Creatures look for sustenance in an arbitrary or semi irregular way naturally.

By and large, the searching way of a creature is viably an arbitrary walk in
light of the fact that the following move depends on both the momentum
area/state and the progress likelihood to the following area. The picked
bearing verifiably relies upon a likelihood, which can be displayed
scientifically. Different investigations have demonstrated that the flight
conduct of numerous creatures and creepy crawlies shows the common

16
qualities of Lévy flights [28]. A Lévy flight is an arbitrary stroll in which the
progression lengths are conveyed by a substantial followed likelihood
dissemination. After a substantial number of steps, the separation from the
birthplace of the arbitrary walk watches out for a steady dispersion.

2.5 Gantt Chart

2.5.1 Final Year Project 1

Table 2- 3: Gantt Chart FYP 1

Activities Weeks

1 2 3 4 5 6 7 8 9 1 1 1 1 14 1 1 1
0 1 2 3 5 6 7

Free FYP
Activities

Meet FYP
SV

Introductory
FYP 1
Briefing

Meet FYP
Supervisor

Workshop
(Chapter 4 &
5)

Meet FYP
Supervisor

Free FYP
Activities

Meet FYP
Supervisor

Workshop
(Abstract
Writing &
Skills)

Free FYP
Activities

Meet FYP
Supervisor

Final FYP 2
Briefing

17
Meet FYP
Supervisor

Industry Day

Submit
complete
draft &
Check
Plagiarism

Submit
complete
draft &
Check
Plagiarism

Submission
of FYP
Thesis

2.5.2 Final Year Project 2

Table 2- 4: Gantt Chart FYP 2

Activities Weeks

1 2 3 4 5 6 7 8 9 10 11 12 13 1 15 16 17
4

Meet FYP
Supervisor

Free FYP
Activities

Introductory
FYP 2
Briefing

Free FYP
Activities

Workshop
(Chapter
4&5) in
Thesis
Writing

Free FYP
Activities

Meet FYP
Supervisor

Meet FYP
Supervisor

18
Workshop
(Abstract
Writing &
Poster)

Meet FYP
Supervisor

Meet FYP
Supervisor

Final FYP 2
Briefing

Meet FYP
Supervisor

Industry
Day

Submit
complete
draft &
Check
Plagiarism

Submit
complete
draft &
Check
Plagiarism

Final Week
for
Hardcover
Bound

2.6 Chapter Summary

This section diagrams the past concentrated identified with this examination
venture. Numerous past studies have been done identified with optimization
technique to solve optimization problem. This part abridged on what can be
expected toward the finishing of this research study. This chapter is
additionally useful with a specific end goal to pick up data on what are the
best strategies to use to increase satisfactory results for this research project.

19
 CHAPTER 3

METHODOLOGY

3.1 Introduction

This chapter shall explain and justify the research methods that have chosen
to meet all the objectives in Chapter 1. The methodology is the overall
approach that underpins the research. Within this chapter also, it will expose
in details about the research plan which is the MATLAB and Simulink
software, Load Frequency Control using FOPID controller, implementing the
FOPID Controller for Load Frequency Control, optimal tuning of FOPID
controller using the Firefly Algorithm and Cuckoo Search Algorithm, and the
fitness function using the Integral Time Square Error (ITSE). All the results
and questions was presented and analyzed at Chapter 4.

20
3.2 Research Plan

MATLAB and SIMULINK

Software

Power System Modelling

Modelling of the System

Load Frequency Control Using

FOPID Controller

Implementing FOPID Controller

for LFC

Optimal tuning of FOPID

controller using Firefly
Algorithm and Cuckoo Search

Firefly Algorithm

Cuckoo Search

Integral Time Square Error

Figure 3- 1: Research Plan

21
3.3 MATLAB and SIMULINK Software

This chapter clearly defines the research methods used to conduct the study
for this paper. MATLAB is a high-performance language for technical
computing. It integrates computation, visualization, and programming in an
easy-to-use environment where problems and solutions are expressed in
familiar mathematical notation. Typical uses include:

i. Math and computation

ii. Algorithm and development
iii. Modelling, simulation, and prototyping
iv. Data analysis, exploration, and visualization
v. Scientific and engineering graphics.
vi. Application development, including Graphical User Interface building

MATLAB is an interactive system whose basic data element is an

array that does not require dimensioning. This allows you to solve many
technical computing problems, especially those with matrix and vector
formulations, in a fraction of the time it would take to write a program in a
scalar no interactive language such as C or Fortran. The name MATLAB
stands for matrix laboratory. MATLAB was originally written to provide easy
access to matrix software developed by the LINPACK and EISPACK
projects, which together represent the state-of-the-art in software for matrix
computation. MATLAB has evolved over a period of years with input from
many users. In university environments, it is the standard instructional tool for
introductory and advanced courses in mathematics, engineering and science.
In industry, MATLAB is the tool of choice for high-productivity research
development, and analysis. MATLAB features a family of application-specific
solutions called toolboxes. Very important to most users of MATLAB,
toolboxes allow you to learn and apply specialized technology. Toolboxes are
comprehensive collections of MATLAB functions (M-files) that extend the
MATLAB environment to solve particular classes of problems. Areas in which
toolboxes are available include signal processing, control systems, neural
networks, fuzzy logic, wavelets, simulation, and many others.

22
 By using the MATLAB (Matrix Laboratory) of Simulink, the model block
of single area interconnected reheat thermal system with the additional of
FOPID Controller can be designed. Then, to observe the frequency response
and tie line power in terms of settling time and overshoot/undershoot.
Moreover, a comparative analysis using the Cuckoo Search (CS) algorithm
and Firefly Algorithm (FA) technique to tune the FOPID controller of the LFC.
In addition, a Graphical-user interface (GUI) will be designed to show how
the scheme works and for users’ ease.

3.3.1 Power System Modelling

In this project, the system under study is a single area interconnected reheat
thermal system. To achieve the reliable and safety operation of power
system, it is necessary that one controller has been considered for each
generation unit. Each power plant consists of the speed governing system,
turbine and generator that for simplicity at analysis in the frequency domain,
the transfer functions are used for modelling the power system components.
The Load Frequency Control (LFC) is utilized to keep up sensibly uniform
recurrence, to separate the heap between the generators, and to control the
tie-line exchange plans. The adjustment in recurrence is detected when the
rotor edge δ is changed. The blunder signals are changed into genuine
power command signal, which is sent to prime mover, to require an
augmentation in the torque. The prime mover at that point acquires change
the generator output by a sum which will change the estimations of inside the
predetermined resistance.

Up to six steps are required to model a system:

a) Initializing the system

When displaying a dynamic system, it is first expected to totally instate
the system. When displaying a substantial system which can be snapped
into parts, it is expected to demonstrate the model's each sub-component
independently. By then, subsequent to developing each component, it can
be incorporated into a whole system. A Simulink piece is a model

23
 component that characterizes a numerical connection between its input
and output. To make this basic model, it required four Simulink blocks.

Table 3-1: Simulink Blocks

Block Purpose Model Purpose

Block name
Produce an input Simulate the
Pulse Generator
signal for the model accelerator pedal

Multiply the input Simulate how pressing

Gain
signal by a factor the accelerator affects
the car's acceleration

Second- Integrate input signal Obtain position from

Integrator,
twice acceleration
Order

Designate a signal as Designate the position

Outport
an output from the as an output from the
model model

b) Pinpoint System Components

In defining a system, there are three sorts of categories:
i. Parameters – System values that stay consistent unless it is adjusted.
ii. States – Internal values, representing the dynamics of the block,
computed during simulation
iii. Signals – Input and output values that adjusted dynamically during
simulation.

c) Modelling the System with Equations

The following stage in displaying the system is devising the numerical
conditions which pictures the system. The model could contain:
i. Algebraic equations
ii. Difference equation (Equations used to create the block diagram in
Simulink for discrete systems)

24
d) Constructing Simulink Block Diagram
Afterward, the block diagram can be constructed in Simulink. The block
diagram for every element must be constructed independently. Later, the
subcomponent is ready to be integrated into a whole system.
e) Running the Simulation
The last part in modelling using Simulink is to run the simulation and
examine the outcomes. Simulink allows us to initialize system inputs,
simulate a model, and see alteration of behaviour which enables us to
rapidly assesses the model.
f) Validating Simulation Results
After the model have been simulated, the model must be validated to see
whether it precisely replicates the attributes of the system. The
linearization and trimming instruments accessible from the command line
of MATLAB can be used to examine and validate the model. Outside of
Simulink before beginning to construct the model, MATLAB performs the
first three steps stated.

25
3.3.2 Modelling of the System

Using the Simulink Software of MATLAB, a single area of reheat thermal

power system for Load Frequency Control (LFC) is designed. In this
research, the input to the controller are Area Control Error and it is derivative.
The system parameters are as shown.

Figure 3- 2: Modelling of a Single Area Reheat Thermal System

26
3.4 Load Frequency Control Using FOPID Controller

3.4.1 Implementing FOPID Controller for LFC

The idea of FOPID controller was first proposed by Podlubny. The difference
amongst FOPID and PID controller is that, in FOPID, the request of
derivative and essential is not whole number. This trademark gives additional
degrees of freedom in tuning the controller and can prompt better powerful
execution in comparison with ordinary PID. Because of expanding
enthusiasm for enhancing the response of regular PID, FOPID controller has
gotten extensive consideration in the most recent couple of years. The
definition of FOPID controller is demonstrated as follows:
𝐾
K (S) = 𝐾𝑝 + 𝑆𝜆𝐼 + 𝐾𝑑 𝑆𝜇

where Kp, Ki, and Kd proportional, integral, and derivative gain, respectively.

Figure 3- 3: Example of Step Response Waveform

27
 Additionally, λ and μ are requests of basic and derivative gain,
separately. It is demonstrated that implementing FOPID controller for LFC
enhances the power framework reaction in terms of settling time, overshoots,
and undershoots. Also, this controller is vigorous to watch the recurrence
reaction and tie line control regarding settling time and overshoot/undershoot.
The gains of the FOPID controller can be acquired by trial and error method.

3.5 Optimal tuning of FOPID controller using Firefly Algorithm and

Cuckoo Search

Each FOPID controller has five parameters to be tuned, and along these
lines, in the contemplated single area reheat thermal system, there were five
parameters to be ideally tuned. Figure 7 demonstrates the flowchart of tuning
the parameters utilizing Firefly Algorithm (FA) and Cuckoo Search Algorithm.

Figure 3- 4: Block Diagram of FOPID Controller

28
3.5.1 Firefly Algorithm

The firefly algorithm (FA) is a metaheuristic optimization algorithm which is

biologically origin from the behavior of fireflies. The firefly algorithm is created
by portraying the fireflies as the nature based interacting agents. The
algorithm is set up using following rules:

i. The fireflies are unisex thus the fireflies are attracted to each other.
ii. The attractiveness of the fireflies is proportional to their brightness.
Thus, the fireflies are attracted and move depending on the higher
level of brightness.
iii. The level of brightness however is inversely proportional to the
distance between fireflies.
iv. The level of brightness represents the objective function value.

The attractiveness 𝛽 of the fireflies as a function of distance(r) is defined as:

[3]

where 𝛽0 is the attractiveness at r = 0 and γ is the fixed light absorption

coefficient. The distance between any two fireflies (𝑟𝑖𝑗) is based on the
Euclidean distance:

[4]

where 𝑥𝑖 and 𝑥𝑗 denotes two different fireflies and the motion of the firefly
is conveyed as:

[5]

where α is the randomization parameter and the rand is a random generated

in the range of [0,1]. The scale is utilized as the scaling parameter in which
the value ranges between the upper and lower bound. The usual values used
as part of this study are as per following: 𝛽0 = 1, 𝛾 = 1 and 𝛼 = 0.8.

In the controller optimization problem, the fireflies are used to represent the
FOPID parameters Kp, Ki, Kd, Mu and Lambda. The error criterion of the

29
controller is utilized as the objective function. In this investigation the integral
time weighted squared error (ITSE) is used. The upper bound and lower
bound of the FOPID controller’s parameters are set in the range of [0.2,2].

The program pseudo code is as per following.

30
Figure 3- 5: Flowchart of Firefly Algorithm

31
3.5.2 Cuckoo Search Algorithm

The cuckoo search algorithm is defined as a nature-inspired algorithm

developed by referring on reproduction of cuckoo birds. During the on-going
process through CS algorithms, the cuckoo eggs solution shall play a role to
associate potential solutions. Cuckoos lay their fertilised eggs in other
cuckoos’ nests with the hope of their off-springs being raised by proxy
parents. The idea of the algorithms inspired by the cuckoos discover that the
eggs in their nests do not belong to them, during such occasion, the foreign
eggs shall be eliminated or whole nests are abandoned. There are three
factors that need to be abide through the CS optimisation algorithm:

i. Each cuckoo selects a nest randomly and lays one egg in it.

ii. The best nests with high quality of eggs will be carried over to the next
generation.

iii. For a fixed number of nests, a host cuckoo can discover a foreign egg
with a probability 𝑝𝛼 є [0,1]. In this case, the host cuckoo can either
throw the egg away or abandon the nest and build a new one
somewhere else.

However, the third rule achievable through substituting a fraction pa of

the n host nests with new nests obtaining new random solutions. The
standard of solution is the result of the value of the objective being
equivalent. Based on the view of implementing point, the idea is every egg in
a nest function as a solution. For every cuckoo it only could lay only by an
egg, hence representing a solution. It is important to differ up all the three of
the egg, nest or the cuckoo. The purpose is to exploit and utilize the newly
improve solution of the cuckoo egg to change the bad solution within the
nest.

The cuckoo search algorithm known to be essential towards global

optimisation problems. This due to the function to maintains a balance
between local random walk and the global random walk. The switching
parameter is the factor that oversee between local and global random walks.

32
The basic steps of the Cuckoo search algorithm based on three rules can be
summarized as pseudo code shown:

33
Figure 3- 6: Flowchart of Cuckoo Search Algorithm

34
3.6 Integral Time Square Error

ITSE incorporates the squared error multiplied by the time over time. The
function of ITSE is to weight errors which exist after quite a while. A much
more heavily than those towards the beginning of the response. ITSE tuning
makes systems which settle rapidly. The drawback of ITSE is that the tuning
likewise makes systems to have more oscillations.

𝑡
𝐼𝑇𝑆𝐸 = ∫0 𝑡. (𝐴𝐶𝐸𝑖 )2dt [23]

3.7 Chapter Summary

In this chapter, methodology of this research has been explained in detail

into a few sections such as introduction to research methodology, power
system modelling of the load frequency control, observation of the frequency
response, and optimal tuning of FOPID controller using Firefly Algorithm
and Cuckoo Search Algorithm. Besides, flowchart of nature optimization
technique Firefly Algorithm (FA) process also has been explained in order to
get more understanding in this process.

35
 CHAPTER 4

RESULTS AND DISCUSSIONS

4.1 Introduction

In the simulation the performance of the optimized controller was analyzed by

observing the frequency response of the system. In this paper, two different
optimization technique is used which are Firefly Algorithm and Cuckoo
Search Algorithm. The performance of the optimization technique will then be
compared to see which of the algorithms gives a better performance. Each
simulation was run for 100 iterations and the average value of the ITSE for
each technique will be recorded. Three different case was investigated in this
study.

The first case is the application of load with different number of values
to both of the optimization technique. For each area the value of load will
start at 0.1, the load will then be increased and decreased by 25% and 50%
respectively. The load value that will be tested is 0.1, 0.075, 0.05, 0.125, and
0.5. The second case is the efficiency of the governor and turbine applied for
both of the optimization technique. Here, the value of the governor and
turbine will then be increased and decreased also at 25% and 50%. The third
case is the application of the different fitness function which is IAE, ITAE and
ISE. The parameters of the FOPID controller is recorded to see the
performance of each algorithm.

36
4.2 Case 1: Application of Load

Firefly Algorithm Performance for Load applied in the Area Load Value = 0.1

Table 4- 1: System Parameter Values

Controller Kp Ki Kd µ λ
Values
1.4653 1.4350 1.0653 0.9493 1.0400

Table 4- 2: Step Info

Settling time (s) Overshoot (Hz) Undershoot Objective

Function

16.6798 2.7555e-04 8.9887e-06 0.001708

Figure 4- 1: Firefly Algorithm Frequency Response for Load Value = 0.1

37
Figure 4- 2: Firefly Algorithm Convergence graph for Load Value = 0.1

38
Firefly Algorithm Performance for Load applied in the Area Load Value = 0.05

Table 4- 3: System Parameter Values

Controller Kp Ki Kd µ λ
Values
1.3503 1.2416 1.1884 0.8514 1.0325

Table 4- 4: Step Info

Settling time (s) Overshoot (Hz) Undershoot Objective

Function

17.9294 1.3907e-04 5.2106e-06 0.0005236

Figure 4- 3: Firefly Algorithm Frequency Response for Load Value = 0.05

39
Figure 4- 4: Firefly Algorithm Convergence Graph for Load Value = 0.05

40
Firefly Algorithm Performance for Load applied in the Area Load Value = 0.15

Table 4- 5: System Parameter Values

Controller Kp Ki Kd µ λ
Values
1.4635 1.3073 1.2480 0.9292 1.0686

Table 4- 6: Step Info

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

17.3217 4.9234e-04 1.1712e-05 0.004463

Figure 4- 5: Firefly Algorithm Frequency Response for Load Value = 0.15

41
Figure 4- 6: Firefly Algorithm Convergence Graph for Load Value = 0.15

42
Firefly Algorithm Performance for Load applied in the Area Load Value =
0.075

Table 4- 7: System Parameter Values

Controller Kp Ki Kd µ λ
Values
1.0078 1.3660 1.4719 1.1924 0.9310

Table 4- 8: Step Info

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

21.3171 3.8958e-04 1.3557e-05 0.002022

Figure 4- 7: Firefly Algorithm Frequency Response for Load Value = 0.075

43
Figure 4- 8: Firefly Algorithm Convergence graph for Load Value = 0.075

44
Firefly Algorithm Performance for Load applied in the Area Load Value =
0.125

Table 4- 9: System Parameter Values

Controller Kp Ki Kd µ λ
Values
1.2552 1.5024 0.5088 1.0933 1.1676

Table 4- 10: Step Info

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

11.9303 0.0012 2.2952e-06 0.00528

Figure 4- 9: Firefly Algorithm Frequency Response for Load Value = 0.125

45
Figure 4- 10: Firefly Algorithm Convergence Graph for Load value = 0.125

46
Cuckoo Search Algorithm Performance for Load applied in the Area Load
Value = 0.1

Table 4- 11: System Parameter Values

Controller Kp Ki Kd µ λ
Values
2 2 1.6290 0.8734 1.0047

Table 4- 12: Step Info

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

15.6356 1.9336e-04 7.5357e-06 0.0008918

Figure 4- 11: Cuckoo Search Algorithm Frequency Response for

Load = 0.1

47
Figure 4- 12: Cuckoo Search Algorithm Convergence Graph for Load Value =
0.1

48
Cuckoo Search Algorithm Performance for Load applied in the Area Load
Value = 0.05

Table 4- 13: System Parameter Values

Controller Kp Ki Kd µ λ
Values
2 2 1.6380 0.8730 1.0037

Table 4- 14: Step Info

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

17.2898 9.6420e-05 1.2247e-07 0.0002229

Figure 4- 13: Cuckoo Search Algorithm Frequency Response for Load Value
= 0.05

49
Figure 4- 14: Cuckoo Search Algorithm Convergence Graph for Load Value =
0.05

50
Cuckoo Search Algorithm Performance for Load applied in the Area Load
Value = 0.15

Table 4- 15: System Parameter Values

Controller Kp Ki Kd µ λ
Values
2 2 1.6093 0.8808 1.0025

Table 4- 16: Step Info

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

17.3106 2.8678e-04 3.8523e-07 0.002007

Figure 4- 15: Cuckoo Search Algorithm Frequency Response for Load Value
= 0.15

51
Figure 4- 16: Cuckoo Search Algorithm Convergence Graph for Load Value =
0.15

52
Cuckoo Search Algorithm Performance for Load applied in the Area Load
Value = 0.075

Table 4- 17: System Parameter Values

Controller Kp Ki Kd µ λ
Values
2 2 1.6307 0.8742 1.0059

Table 4- 18: System Parameter Values

Settling time (s) Overshoot (Hz) Undershoot Objective

(Hz) Function

17.2156 1.4612e-04 1.6921e-07 0.0005016

Figure 4- 17: Cuckoo Search Algorithm Frequency Response for Load Value
= 0.075

53
Figure 4- 18: Cuckoo Search Algorithm Convergence Graph for Load Value =
0.075

54
Cuckoo Search Algorithm Performance for Load applied in the Area Load
Value = 0.125

Table 4- 19: System Parameter Values

Controller Kp Ki Kd µ λ
Values
2 2 1.6493 0.8740 1.0046

Table 4- 20: Step Info

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

17.2994 2.4327e-04 2.9694e-07 0.001393

Figure 4- 19: Cuckoo Search Algorithm Frequency Response for Load Value
= 0.125

55
Figure 4- 20: Cuckoo Search Algorithm Convergence Graph for Load Value =
0.125

56
4.3 Case 2: Application of Governor and Turbine

Firefly Algorithm Performance for Governor and Turbine

+50% Governor = 0.3s + 1

+50% Turbine = 0.75s + 1

Table 4- 21: System Parameter Values

Controller Kp Ki Kd µ λ
Values
1.7350 1.7950 1.9100 1.0350 1.1450

Table 4- 22: Step Info

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

13.4041 8.3723e-04 2.6180e-06 0.00397

Figure 4- 21: Firefly Algorithm Frequency Response with +50% Efficiency of

Governor and Turbine

57
Figure 4- 22: Firefly Algorithm Convergence Graph with +50% Efficiency of
Governor and Turbine

58
Firefly Algorithm Performance for Governor and Turbine

+25% Governor = 0.25s + 1

+25% Turbine = 0.625s + 1

Table 4- 23: System Parameter Values

Controller Kp Ki Kd µ λ
Values
1.6252 1.1955 1.6556 1.0764 0.9352

Table 4- 24: Step Info

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

21.0217 2.1812e-04 2.5393e-05 0.004348

Figure 4- 23: Firefly Algorithm Frequency Response with +25% Efficiency of

Governor and Turbine

59
Figure 4- 24: Firefly Algorithm Convergence Graph with +25% Efficiency of
Governor and Turbine

60
Firefly Algorithm Performance for Governor and Turbine

-50% Governor = 0.15s + 1

-50% Turbine = 0.375s + 1

Table 4- 25: System Parameter Values

Controller Kp Ki Kd µ λ
Values
1.2992 1.3249 0.9466 0.6417 1.1268

Table 4- 26: Step Info

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

15.0486 4.3462e-04 5.1211e-06 0.003039

Figure 4- 25: Firefly Algorithm Frequency Response with -50% Efficiency of

Governor and Turbine

61
Figure 4- 26: Firefly Algorithm Convergence Graph with -50% Efficiency of
Governor and Turbine

62
Firefly Algorithm Performance for Governor and Turbine

-25% Governor = 0.1s + 1

-25% Turbine = 0.25s + 1

Table 4- 27: System Parameter Values

Controller Kp Ki Kd µ λ
Values
0.7419 1.7725 1.4378 0.6920 1.0037

Table 4- 28: Step Info

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

18.4435 3.9618e-04 1.1410e-05 0.00234

Figure 4- 27: Firefly Algorithm Frequency Response with -25% Efficiency of

Governor and Turbine

63
Figure 4- 28: Firefly Algorithm Convergence Graph with -25% Efficiency of
Governor and Turbine

64
Cuckoo Search Performance for Governor and Turbine

+50% Governor = 0.3s + 1

+50% Turbine = 0.75s + 1

Table 4- 29: System Parameter Values

Controller Kp Ki Kd µ λ
Values
2 2 2 0.9815 0.9730

Table 4- 30: Step Info

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

15.9763 2.2221e-04 4.9969e-07 0.00108

Figure 4- 29: Cuckoo Search Algorithm Frequency Response with +50%

Efficiency of Governor and Turbine

65
Figure 4- 30: Cuckoo Search Algorithm Convergence Graph with +50%
Efficiency of Governor and Turbine

66
Cuckoo Search Performance for Governor and Turbine

+25% Governor = 0.25s + 1

+25% Turbine = 0.625s + 1

Table 4- 31: System Parameter Values

Controller Kp Ki Kd µ λ
Values
2 2 2 0.9814 0.9730

Table 4- 32: Step Info

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

17.7987 2.0397e-04 5.0660e-07 0.0009253

Figure 4- 31: Cuckoo Search Algorithm Frequency Response with +25%

Efficiency of Governor and Turbine

67
Figure 4- 32: Cuckoo Search Algorithm Convergence Graph with +25%
Efficiency of Governor and Turbine

68
Cuckoo Search Performance for Governor and Turbine

-50% Governor = 0.1s + 1

-50% Turbine = 0.25s + 1

Table 4- 33: System Parameter Values

Controller Kp Ki Kd µ λ
Values
2 2 1.1395 0.8328 1.0215

Table 4- 34: Step Info

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

20.3857 1.6693e-04 9.2766e-08 0.0007054

Figure 4- 33: Cuckoo Search Algorithm Frequency Response with -50%

Efficiency of Governor and Turbine

69
Figure 4- 34: Cuckoo Search Algorithm Convergence Graph with -50%
Efficiency of Governor and Turbine

70
Cuckoo Search Performance for Governor and Turbine

-25% Governor = 0.15s + 1

-25% Turbine = 0.375s + 1

Table 4- 35: System Parameter Values

Controller Kp Ki Kd µ λ
Values
2 2 2 0.9147 0.9918

Table 4- 36: Step Info

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

21.4052 1.9016e-04 3.3816e-07 0.000856

Figure 4- 35: Cuckoo Search Algorithm Frequency Response with -25%

Efficiency of Governor and Turbine

71
Figure 4- 36: Cuckoo Search Algorithm Convergence Graph with -25%
Efficiency of Governor and Turbine

72
4.4 Case 3: Application of Fitness Function

Firefly Algorithm Performance for Fitness Function

Integral Time Square Multiplied by Error (ITSE)

Table 4- 37: System Parameter Values

Controller Kp Ki Kd µ λ
Values
1.0078 1.3660 1.4719 1.1924 0.9310

Table 4- 38: Step Info

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

21.3168 6.4930e-04 2.2848e-05 0.005618

Figure 4- 37: Firefly Algorithm Frequency Response with ITSE Fitness

Function

73
Figure 4- 38: Firefly Algorithm Convergence Graph with ITSE Fitness
Function

74
Firefly Algorithm Performance for Fitness Function

Integral of Time Absolute Error (ITAE)

Table 4- 39: System Parameter Values

Controller Kp Ki Kd µ λ
Values
1.1848 1.5498 0.5512 0.8917 1.1408

Table 4- 40: System Parameter Values

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

11.7268 0.0013 3.4799e-06 0.364

Figure 4- 39: Firefly Algorithm Frequency Response with ITAE Fitness

Function

75
Figure 4- 40: Firefly Algorithm Convergence Graph with ITAE Fitness
Function

76
Firefly Algorithm Performance for Fitness Function

Integral of Absolute Error (IAE)

Table 4- 41: System Parameter Values

Controller Kp Ki Kd µ λ
Values
0.5504 1.5540 0.8025 0.4963 1.2951

Table 4- 42: Step Info

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

7.5925 0.0021 -0.0051 0.182

Figure 4- 41: Firefly Algorithm Frequency Response with IAE Fitness

Function

77
Figure 4- 42: Firefly Algorithm Convergence Graph with IAE Fitness Function

78
Firefly Algorithm Performance for Fitness Function

Integral of Square Error (ISE)

Table 4- 43: System Parameter Values

Controller Kp Ki Kd µ λ
Values
1.6216 1.7812 1.3241 1.2240 0.7920

Table 4- 44: Step Info

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

19.6820 1.3322e-04 -2.9942e-05 0.003227

Figure 4- 43: Firefly Algorithm Frequency Response with ISE Fitness

Function

79
Figure 4- 44: Firefly Algorithm Convergence Graph with ISE Fitness Function

80
Cuckoo Search Algorithm Performance for Fitness Function

Integral of Time multiplied with Square Error (ITSE)

Table 4- 45: System Parameter Values

Controller Kp Ki Kd µ λ
Values
2 1.9999 1.6290 0.8734 1.0047

Table 4- 46: Step Info

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

17.2409 2.4170e-04 2.9589e-07 0.001393

Figure 4- 45: Cuckoo Search Algorithm Frequency Response with ITSE

Fitness Function

81
Figure 4- 46: Cuckoo Search Algorithm Convergence Graph with ITSE
Fitness Function

82
Cuckoo Search Algorithm Performance for Fitness Function

Integral of Time Absolute Error (ITAE)

Table 4- 47: System Parameter Values

Controller Kp Ki Kd µ λ
Values
1.8100 1.9993 0.9411 1.0676 1.0001

Table 4- 48: Step Info

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

16.2377 2.5336e-04 3.5415e-07 0.0451

Figure 4- 47: Cuckoo Search Algorithm Frequency Response with ITAE

Fitness Function

83
Figure 4- 48: Cuckoo Search Algorithm Convergence Graph with ITAE
Fitness Function

84
Cuckoo Search Algorithm Performance for Fitness Function

Integral of Absolute Error (IAE)

Table 4- 49: System Parameter Values

Controller Kp Ki Kd µ λ
Values
2 2 0.9090 1.1050 1.0086

Table 4- 50: Step Info

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

16.3300 2.4956e-04 2.5806e-07 0.06262

Figure 4- 49: Cuckoo Search Algorithm Frequency Response with IAE

Fitness Function

85
Figure 4- 50: Cuckoo Search Algorithm Convergence Graph with IAE Fitness
Function

86
Cuckoo Search Algorithm Performance for Fitness Function

Integral of Square Error (ISE)

Table 4- 51: System Parameter Values

Controller Kp Ki Kd µ λ
Values
2 2 2 0.9438 0.9747

Table 4- 52: Step Info

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

19.5869 2.4512e-04 6.6166e-07 0.00197

Figure 4- 51: Cuckoo Search Algorithm Frequency Response with ISE

Fitness Function

87
Figure 4- 52: Cuckoo Search Algorithm Convergence Graph with ISE Fitness
Function

88
4.5 Comparison of performance between Firefly Algorithm and
Cuckoo Search Algorithm

Case 1: Application of Load

Table 4- 53: System Parameter Values for Case 1

Value Kp Ki Kd µ λ Objective
Function

0.1 1.4653 1.4350 1.0653 0.9493 1.0400 0.001708

0.05 1.3503 1.2416 1.1884 0.8514 1.0325 0.0005236

FA
0.15 1.4635 1.3073 1.2480 0.9292 1.0686 0.004463

0.075 1.0078 1.3660 1.4719 1.1924 0.9310 0.002022

0.125 1.2552 1.5024 0.5088 1.0933 1.1676 0.00528

0.1 2 2 1.6290 0.8734 1.0047 0.0008918

0.05 2 2 1.6380 0.8730 1.0037 0.0002229

CS
0.15 2 2 1.6093 0.8808 1.0025 0.002007

0.075 2 2 1.6307 0.8742 1.0059 0.0005016

0.125 2 2 1.6493 0.8740 1.0046 0.001393

89
Case 2: Application of Governor and Turbine

Table 4- 54: System Parameter Values for Case 2

% Kp Ki Kd µ λ Objective
Function

FA +50% 1.7350 1.7950 1.9100 1.0350 1.1450 0.00397

+25% 1.6252 1.1955 1.6556 1.0764 0.9352 0.004348

-50% 1.2992 1.3249 0.9466 0.6417 1.1268 0.003039

-25% 0.7419 1.7725 1.4378 0.6920 1.0037 0.00234

+50% 2 2 2 0.9815 0.9730 0.00108

CS
+25% 2 2 2 0.9814 0.9730 0.0009253

-25% 2 2 2 0.9147 0.9918 0.000856

-50% 2 2 1.1395 0.8328 1.0215 0.0007054

90
Case 3: Application of Fitness Function

Table 4- 55: System Parameter Values for Case 3

Type Kp Ki Kd µ λ Objective
Function

FA ITSE 1.0078 1.3660 1.4719 1.1924 0.9310 0.005618

ITAE 1.1848 1.5498 0.5512 0.8917 1.1408 0.364

IAE 0.5504 1.5540 0.8025 0.4963 1.2951 0.182

ISE 1.6216 1.7812 1.3241 1.2240 0.7920 0.003227

ITSE 2 1.9999 1.6290 0.8734 1.0047 0.001393

CS
ITAE 1.8100 1.9993 0.9411 1.0676 1.0001 0.0451

IAE 2 2 0.9090 1.1050 1.0086 0.06262

ISE 2 2 2 0.9438 0.9747 0.00197

91
4.6 Discussions

The Firefly Algorithm and Cuckoo Search Algorithm are applied to optimize
the gains of the FOPID controller. The tuning method is used to optimize the
proportional (Kp), integral (Ki), derivative (Kd) gains, µ and λ of the controller.
The modelling and optimization process is carried out using MATLAB
software version R2016a. The model of the single area reheat thermal power
system is designed in Simulink and the controller optimization is executed
through an M-file script.

This project consists of three cases. Case 1 is where the load is set
accordingly to the different value of load in the simulation block. The load
demand of the first value was set to 0.1. Figure 4-1 till Figure 4-20 shows the
frequency response of the both optimization technique used which are Firefly
Algorithm and Cuckoo Search Algorithm. Table 4-53 shows the system
parameter values for each load in Case 1. Here, it can be seen that using the
Cuckoo Search Algorithm as the optimization technique has better
performance compared to the Firefly Algorithm. In terms of the settling time,
Firefly Algorithm with the load value of 0.125 as referred to the Table 4-10
and the frequency response in Figure 4-9 settles the system better
compares to the Cuckoo Search Algorithm in Figure 4.19 with the time
11.9303s. The oscillations for Cuckoo Search Algorithm less than Firefly
Algorithm. The overshoot value for Cuckoo Search Algorithm is slightly
higher compared to the Firefly Algorithm. Comparing the error for both
techniques, Cuckoo Search Algorithm has a lower ITSE compared to the
Firefly Algorithm. Moreover, the value of the load was decreased and
increased to 25% and 50% in from the initial load value which is 0.1, as the
value of load increased, it seems that the frequency response for both Firefly
Algorithm and Cuckoo Search Algorithm has a greater undershoot compared
to the initial load value of 0.1. It is also noticed that the ITSE for both
algorithms are slightly different but only by small margin.

As the load is further decreased to 50% of the initial load, which is

0.05, in Figure 4-3 it resulting the undershoot to slightly decrease however, in
terms of the value of the overshoot, it is much bigger. Comparing the load
demand of 0.125 with increment of 25% of the initial load, it shows that the
92
value of overshoot and undershoot is slightly much lower. The load was then
decreased by 25% and it was observed that the decrement occurs to
undershoot due to decreased in load and the system settles slower. As the
load was further decreased by 50%, hence undershoot decreases compares
to the initial load. For both the algorithms used, the change in frequency
response is almost the same with the increasing or decreasing of
undershoot. Overall in Case 1, as referred to step info at Table 4-56 and
frequency response at Figure 4-53, the Cuckoo Search Algorithm performed
better compares to the Firefly Algorithm as the value of ITSE is the lowest.
Cuckoo Search Algorithm with value of load decreased to 50% from the initial
load which is 0.05 has the lowest value of ITSE.

Table 4- 56: Step Info for Cuckoo Search with Load = 0.05

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

17.2898 9.6420e-05 1.2247e-07 0.0002229

Figure 4- 53: Cuckoo Search Algorithm Frequency Response with Load

Value = 0.05

93
 The algorithms then further tested in Case 2 whereby the value of the
governor and turbine were altered by decreasing and increasing to the value
of 25% and 50% accordingly. The case 2 focuses on whether the efficiency
of both governor and turbine would affect the performance of the system.
Table 4-54 shows the system parameter values for Case 2. This test is done
in order to further verify the performance of the algorithms used which are
Firefly Algorithm and Cuckoo Search Algorithm. The value of governor and
turbine is increased to 50%, thus resulted the value of the overshoot in Firefly
Algorithm to be higher than the value of overshoot in Cuckoo Search
Algorithm. In terms of undershoot, Cuckoo Search Algorithm has lower value
of undershoot compared to Firefly Algorithm. Here, the tuning of the
controller for Firefly Algorithm settled faster compare to the Cuckoo Search
Algorithm. Besides, the value of ITSE in Cuckoo Search Algorithm is lesser
than the value of ITSE in Firefly Algorithm.

After that, the efficiency of governor and turbine is further increased

about 25% from the initial value. The value of overshoot in Cuckoo Search
Algorithm is slightly lower than the Firefly Algorithm while the value of
undershoot in Firefly Algorithm is much higher than the Cuckoo Search
Algorithm. In terms of settling time, Cuckoo Search Algorithm settled at
17.7987s while Firefly Algorithm at 21.0217s. Moreover, the value of ITSE in
Cuckoo Search Algorithm is lower than the Firefly Algorithm.

As the value of governor and turbine is further decreased to 50% from

the initial value, the value of the overshoot in Firefly Algorithm is higher than
the Cuckoo Search Algorithm. In terms of the undershoot, after went through
the simulation, it shows that the value of the undershoot in Cuckoo Search
Algorithm is lower than the Firefly Algorithm. Also, in terms of the settling
time, Firefly Algorithm settled faster than the Cuckoo Search Algorithm. The
value of the ITSE in Cuckoo Search Algorithm is lower than the Firefly
Algorithm. After that, the value of the governor and turbine is decreased to
25% of the initial value. From here, it shows that the value of overshoot in
Cuckoo Search is lower than the value of overshoot in Firefly Algorithm. In
terms of undershoot, Cuckoo Search Algorithm has lower value than the
Firefly Algorithm. Then, the value of the ITSE in Cuckoo Search Algorithm is

94
lower than the Firefly Algorithm. Overall, we can say that in Case 2 as
referring to step info at Table 4-57 and frequency response at Figure 4-54,
Cuckoo Search Algorithm for the value of governor and turbine decreased to
50% has the best performance.

Table 4- 57: Step Info of Cuckoo Search Governor and Turbine -50%

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

20.3857 1.6693e-04 9.2766e-08 0.0007054

Figure 4- 54: Cuckoo Search Algorithm Frequency Response with -50%

Efficiency of Governor and Turbine

Next, the simulation is then run to determine the best algorithm for the
Case 3. Here, by using different types of fitness function, it shows that every
fitness function results by different value of parameters. In terms of
overshoot/undershoot and settling time also different from each other. Table
4-55 shows the system parameter values after running all the simulations for
Case 3. The fitness function used in case three are Integral of Time
multiplied with Square Error (ITSE), Integral of Time Absolute Error (ITAE),
Integral of Absolute Error (IAE), and Integral of Square Error (ISE). The
usage of ITSE shows that the value of the fitness function in Cuckoo Search

95
Algorithm is lower than the Firefly Algorithm. Whereas, in terms of overshoot,
the value of the overshoot in Firefly Algorithm is higher than the Cuckoo
Search Algorithm while the value of undershoot in Cuckoo Search Algorithm
obtaining lower value than Firefly Algorithm. For ITSE, Cuckoo Search
Algorithm settled faster than the Firefly Algorithm.

Based on the approach of ITAE, Cuckoo Search Algorithm has better

performance. The value of overshoot in Cuckoo Search Algorithm is higher
than the Firefly Algorithm. For undershoot, it shows that Cuckoo Search has
lower than the Firefly Algorithm. In terms of settling time, Firefly Algorithm
settled faster than the Cuckoo Search Algorithm.

Next, by using the IAE as the fitness function, it shows that Cuckoo
Search Algorithm has better performance compared to Firefly Algorithm. In
terms of overshoot and undershoot, the value of overshoot in Firefly
Algorithm has lower value than Cuckoo Search Algorithm while the value of
undershoot in Cuckoo Search Algorithm is higher than the Firefly Algorithm.
Here, Firefly Algorithm settled faster than the Cuckoo Search Algorithm.

Finally using the ISE, it proved that Firefly Algorithm has better
performance compared to the Cuckoo Search Algorithm. The value of
overshoot in Cuckoo Search Algorithm is higher and the value of undershoot
in Cuckoo Search Algorithm is lower than the Firefly Algorithm. In terms of
settling time, Cuckoo Search Algorithm settled faster than the Firefly
Algorithm. In this case as referred to step info at Table 4-58 and frequency
response at Figure 4-55, Cuckoo Search Algorithm with ITSE has the best
performance.

Table 4- 58: Step Info Cuckoo Search Algorithm with Fitness Function ITSE

Settling time (s) Overshoot (Hz) Undershoot (Hz) Objective

Function

17.2409 2.4170e-04 2.9589e-07 0.001393

96
 Figure 4- 55: Cuckoo Search Algorithm Frequency Response with ITSE
Fitness Function

Lastly the convergence graph which shows the value of error for all
100 runs of the algorithms used is observed. From the observation, it is seen
that for all three cases tested, Firefly Algorithm has faster rate of searching
for the best optimal solution compare to Cuckoo Search Algorithm. However,
the last value of error for both algorithms is compared and it can be seen that
Cuckoo Search Algorithm has a lower value of error for majority of the test
compared to Firefly Algorithm. This shows that Cuckoo Search Algorithm has
a better performance compared to Firefly Algorithm.

4.7 Chapter Summary

This chapter shows the final results obtained in the project. This chapter is
important in order to know whether the project reach its objective or not.
From the results obtained in this chapter it can be say that the project is a
success as it achieved the objective of comparing and identifying the
optimization algorithm that has the better performance.

97
 CHAPTER 5

CONCLUSIONS AND RECOMMENDATIONS

5.1 Introduction

This chapter shall properly assess the project on how the project is achieved
and being done. Within the chapter also, obtaining the recommendations of
how the project can be improve for the future.

5.2 Conclusions

In this study, two different optimization technique to tune the parameters of

the FOPID controller was used which is Firefly Algorithm and Cuckoo Search
Algorithm. The study was done on a simulation of a single area reheat
thermal power system. The ITSE error criterion is selected as the objective
function of the controller to evaluate the performance of the optimization
algorithms. The Firefly Algorithm and Cuckoo Search Algorithm is compared
to see which algorithm provide a better performance. The performance of the
algorithms to tune the FOPID parameters was evaluated based on the
observation of the frequency response and ITSE error criterion. The study
was tested in three different cases.

In the first case, the Cuckoo Search Algorithm shows a better

performance compared to Firefly Algorithm which has a lower value of ITSE.
Then, in the second case the Firefly Algorithm shows a slightly better
performance than Cuckoo Search Algorithm in terms of the ITSE. In the last
case, it was observed that the Cuckoo Search Algorithm has a better
performance than Firefly Algorithm both in terms of the settling time and
ITSE. Thus, it can be concluded that the Cuckoo Search Algorithm has a
better performance compared to Firefly Algorithm as Cuckoo Search

98
Algorithm shows a better performance in all three out of the three cases
tested. This research can further be investigated by using different variations
of Cuckoo Search Algorithm and Firefly Algorithm available.

5.2 Recommendations

There are few recommendations that can be further done in this project in
future. Firstly, this project can further be done for other type of optimization
algorithms. In further studies, other optimization algorithms to be tested
compared such as the Grey Wolves Optimization (GWO) technique, Particle
Swarm Optimization (PSO) technique which are popular algorithms that are
said to have a good performance in terms of the overshoot/undershoot, and
settling time. As an addition, there are other variation of Firefly Algorithm
such as the Levy Flight Optimization Algorithm and Chaotic Firefly Algorithm
that can be further study in future. Next, this project can also be further
extended on two, three and more areas so that the scope and limitation of
this project can get wider. Moreover, in future this project can be further study
on how a single FOPID controller can be uses to control more than one area.

To make this study a more reliable research, the values of the FOPID
parameters of the controller should be tested using a real FOPID controller to
a power plant to validate the results obtained.

99
 REFERENCES

[1] E.Rakhshani, 2010, “Multi-area load frequency control in a

deregulated power system using optimal output feedback method”.

[2] Satyanarayana, Sharma, & Sappa, 2014. “Automatic generation

control in power plant using PID, PSS and Fuzzy-PID controller”.

[3] Ibraheem, P. Kumar, and D. P. Kothari, "Recent philosophies of

automatic generation control strategies in power systems," IEEE
Transactions on Power Systems, vol. 20, pp. 346-357, 2005.

[4] R. Umrao, S. Kumar, M. Mohan, and D. K. Chaturvedi, "Load

Frequency Control methodologies for power system," in 2012 2nd
International Conference on Power, Control and Embedded Systems,
2012, pp. 110.

[5] E. Venkatraman, S. Velusami, M. M. T. Ansari, and S. Ganapathy,

"Design of type-II fuzzy logic load frequency controller for two area
thermal reheat power systems," in 2012 International Conference
on Computing, Electronics and Electrical Technologies (ICCEET),
2012, pp. 180-185.

[6] Hans P. Geering (2007): “Optimal Control with Engineering

Applications”, 1st Edition, Zurich, Switzerland, Spinger Verlag
Inc.Berlin, Heidelberg, pp.: 3-15.

[7] Rahul Singh and Vijay Gupta (2011): “On LQR Control with
Asynchronous Clocks” Department of Electrical Engineering,
University of Notre Dome, Notre Dome, IN 46556, Dec. 2011.

[8] Ain Shams Engineering Journal, Volume 5, Issue 1, March 2014,

Pages 121-135 “Fractional order PID controller design for LFC in
electric power systems using imperialist competitive algorithm”.

[9] I. Podlubny Fractional-order systems and PIλDμ controllers IEEE

Trans Automatic Control, 44 (1) (1999), pp. 208-214.

[10] Podlubny I, Dorcak L, Kostial I. On fractional derivatives, fractional-

order dynamic systems and PIλDμ controllers. In: Proceedings of the

100
 36th conference on decision and control. San Diego, California,
USA; 1997.

[11] K. Akbari Moornani, M. HaeriRobustness in fractional proportional–

Integral–derivative-based closed-loop systems” IET Control Theory
Appl, 4 (10) (2010), pp. 1933-1944.

[12] M.A.S. Aboelela, M.F. Ahmed, H.T. Dorrah Design of aerospace

control systems using fractional PID controller. Adv
Res, 3 (3) (2012), pp. 225-232.

[13] R. Roy, P. Bhatt, S.P. Ghoshal Evolutionary computation based three-

area automatic generation control Int J Electr Power Energy
Syst, 37 (8) (2010), pp. 5913-5924.

[14] Swati Sondhi & Yogesh Hote Energy Conversion and Management,
Volume 85, September 2014, Pages 343-353. “Fractional order PID
controller for load frequency control”.

[15] X.-S. Yang, "Firefly algorithms for multimodal optimization," in

Stochastic algorithms: foundations and applications, ed: Springer,
2009, pp. 169-178.

[16] M. R. Shakarami, I. Faraji, I. Asghari, and M. Akbari, "Optimal PID

tuning for load frequency control using Levy-flight firefly algorithm,"
in 2013 3rd International Conference on Electric Power and Energy
Conversion Systems, 2013, pp. 1-5.

[17] D. Karaboga and B. Basturk, "A powerful and efficient algorithm for
numerical function optimization: artificial bee colony (ABC)
algorithm," Journal of global optimization, vol. 39, pp. 459-471,
2007.

[18] Yang, X.S.: Nature-Inspired Metaheuristic Algorithms. Luniver Press

(2008).

[19] Payne RB, Sorenson MD, Klitz K, (2005) The Cuckoos, Oxford
University Press.

101
[20] Yang XS, Deb S, (2009) Cuckoo search via Lévy flights, Proceeings of
World Congress on Nature & Biologically Inspired Computing, India,
pp 210-214.

[21] Yang XS, Deb S, (2010) Engineering Optimisation by Cuckoo Search,

Int. J. Mathematical Modelling and Numerical Optimisation. Vol. 1,
No. 4, pp 330–343.

[22] Brown C, Liebovitch LS, Glendon R, (2007) Lévy flights in Dobe

Ju/hoansi foraging patterns, Human Ecol. Vol. 35, pp 129-138.

[23] H. Shayeghi, H. Shayanfar, and A. Jalili, "Load frequency control

strategies: A state-of-the-art survey for the researcher," Energy
Conversion and management, vol. 50, pp. 344-353, 2009.

[24] X.-S Yang Nature-Inspired Optimization Algorithms (first ed.), Elsevier,

London (2014).

102
 APPENDIX A

Simulink Model of a Single Area Reheat Thermal System

103
 APPENDIX B

Firefly Algorithm Code

%Firefly Algorithm
no_fireflies = 30;
MaxGeneration = 100;
D=5;
ub=2;
lb=0.2;
Range = ub-lb;

gamma=1.0;
delta=0.97;
alpha=0.8;
betamin=0.2;

% ------------------------------------------------
%Initialization
firefly = rand(no_fireflies,D).* Range + lb;
acc = 0.005;
value = firefly;
firefly = round(value/acc)*acc;

ObjVal = tracklsqabc6(firefly,no_fireflies);
ObjVal = ObjVal';
[firefly ObjVal];

for k=1 :MaxGeneration %% start iterations

% ------------------------------------------------
delta=1-(10^(-4)/0.9)^(1^MaxGeneration);
alpha=(1-delta)*alpha;
% ------------------------------------------------
% Evaluate new solutions (for all n fireflies)
for i=1:no_fireflies
ObjVal(i) = tracklsqabc6(firefly(i,:),1);
Lightn(i)=ObjVal(i);

104
 end
% Ranking fireflies based on their light intensity/objectives
[Lightn,Index]=sort(ObjVal);
ns_tmp=firefly;
for i=1:no_fireflies
firefly(i,:)=ns_tmp(Index(i),:);
end
% Find the current best
fireflyo=firefly;
Lighto=Lightn;
Fireflybest=firefly(1,:);
Lightbest=Lightn(1);

% For output only

fbest(k,:)=Lightbest;

% Move all fireflies to the better locations

% [ns]=ffa_move(n,d,ns,Lightn,nso,Lighto,nbest,...
% Lightbest,alpha,betamin,gamma,Lb,Ub);
%
% Scaling of the system
scale=abs(ub-lb);

% Updating fireflies
for i=1:no_fireflies

% The attractiveness parameter beta=exp(-gamma*r)

for j=1:no_fireflies
r=sqrt(sum((firefly(i,:)-firefly(j,:)).^2));
% Update moves
if Lightn(i)>Lighto(j), % Brighter and more attractive
beta0=1;
beta=(beta0-betamin)*exp(-gamma*r.^2)+betamin;
tmpf=alpha.*(rand(1,D)-0.5).*scale;
firefly(i,:)=firefly(i,:).*(1-
beta)+fireflyo(j,:).*beta+tmpf;
end
end % end for j
end % end for i

105
 %%Check Limit of Kp Ki Kd

for i2=1:no_fireflies
for i3 = 1:D
if (firefly(i2,i3)<lb)
firefly(i2,i3) = lb;
end
if(firefly(i2,i3)>ub)
firefly(i2,i3) = ub;
end
end
end

% for i2=1:no_fireflies
% if firefly(i2,:)<lb , firefly(i2,:) = lb; end
% if firefly(i2,:)>ub , firefly(i2,:) = ub; end
%
% for i3=1
% if firefly(i3,:)<lb , firefly(i3,:) = lb; end
% if firefly(i3,:)>ub , firefly(i3,:) = ub; end
% end
% end

pp(k,:) = Lightbest;
k
Lightbest
Fireflybest
end
Fireflybest = [Fireflybest Lightbest];
Kp = Fireflybest(:,1)
Ki = Fireflybest(:,2)
Kd = Fireflybest(:,3)
mu = Fireflybest(:,4)
lambda = Fireflybest(:,5)

%set_param('sim12ex51','SimulationCommand','start');

% ScopeData = evalin('base', simout12);

106
 APPENDIX C

Cuckoo Search Algorithm Code

clear all
clc

% Change this if you want to get better results

N_IterTotal=100;

% Discovery rate of alien eggs/solutions

pa=0.25;

% Simple bounds of the search domain

n = 20;
nd= 5;
% Lower bounds
Lb=0.2*ones(1,nd);
% Upper bounds
Ub=2*ones(1,nd);

% Random initial solutions

for i=1:n,
nest(i,:)=Lb+(Ub-Lb).*rand(size(Lb));
end

% Find the current best nest

fitness=10^10*ones(n,1);
[fmin,bestnest,nest,fitness]=get_best_nest(nest,nest,fitness,n);

N_iter=0;
%% Starting iterations
for iter=1:N_IterTotal,
% Generate new solutions (but keep the current best)
new_nest=get_cuckoos(nest,bestnest,Lb,Ub);

107
[fnew,best,nest,fitness]=get_best_nest(nest,new_nest,fitness,n);
% Update the counter
N_iter=N_iter+n;
% Discovery and randomization
new_nest=empty_nests(nest,Lb,Ub,pa) ;

% Evaluate this set of solutions

[fnew,best,nest,fitness]=get_best_nest(nest,new_nest,fitness,n);
% Update the counter again
N_iter=N_iter+n;
% Find the best objective so far
if fnew<fmin,
fmin=fnew;
bestnest=best;
end

pp(iter,:) = fmin;

end %% End of iterations

%%

Kp = bestnest(:,1)
Ki = bestnest(:,2)
Kd = bestnest(:,3)
mu = bestnest(:,4)
lambda = bestnest(:,5)

%set_param('sim12ex51','SimulationCommand','start');

108