Neuro Fuzzy Inference System based

Advanced STATCOM for Transient

Instability

Submitted by:
Muhammad Khizer 2014-EE-171
Muhammad Tayyab Hussain 2014-EE-179
Muhammad Rehan Majeed 2014-EE-180
Abdullah Rashid 2014-EE-182

Supervised by: Mr.Habib Wajid

Department of Electrical Engineering

University of Engineering and Technology Lahore
Neuro Fuzzy Inference System based
Advanced STATCOM for Transient
Instability

Submitted to the faculty of the Electrical Engineering Department

of the University of Engineering and Technology Lahore
in partial fulfillment of the requirements for the Degree of

Bachelor of Science
in
Electrical Engineering.

Internal Examiner External Examiner

Director
Undergraduate Studies

Department of Electrical Engineering

University of Engineering and Technology Lahore

i
Declaration
We declare that the work contained in this thesis is our own, except where explicitly
stated otherwise. In addition this work has not been submitted to obtain another degree
or professional qualification.

Signature:

Signature:

Signature:

Signature:

Date:

ii
Acknowledgments
Firstly we are grateful to Allah Almighty (the Most Gracious, the Most Merciful) for
showering his blessings upon us and to enable us to complete this project.

We are highly indebted to our Project Advisor Mr.Habib Wajid for his continuous guid-
ance and support throughout the project. His cooperation and technical expertise helped
us to overcome the difficulties faced during the project. His continuous moral support
motivated us to never lose hope and not to give up even at the eleventh hour. Without
his guidance we were not able to complete the project.

In the end we would thank our parents and friends for their continuous support and
encouragement.

iii
Dedicated to our parents because without their support we are
nothing. . .

iv
Contents

Acknowledgments iii

List of Figures viii

Abbreviations x

Abstract xi

1 Introduction 1
1.1 Motivation and Problem Statement . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Overview of the project . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4 Aims . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4

2 Theoretical Background 5
2.1 Static Synchronous Compensator . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.1 Need for Compensation . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.2 Series Compensation . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.1.3 Shunt Compensation . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.1.4 Flexible AC Transmission Systems . . . . . . . . . . . . . . . . . . 6
2.1.5 Static Synchronous Compensator . . . . . . . . . . . . . . . . . . . 6
2.1.5.1 Construction and Working Principle . . . . . . . . . . . . 7
2.1.5.2 DC Link Voltage . . . . . . . . . . . . . . . . . . . . . . . 7
2.1.5.3 STATCOM Control . . . . . . . . . . . . . . . . . . . . . 7
2.1.5.4 Phase Angle Control . . . . . . . . . . . . . . . . . . . . . 7
2.1.5.5 PWM Control . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2 Inverters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2.1 Types of Inverter . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.2.1.1 On the basis of type of Source . . . . . . . . . . . . . . . 8
2.2.1.2 On the basis of output waveform . . . . . . . . . . . . . . 8
2.2.1.3 On the basis of number of phases . . . . . . . . . . . . . 9
2.3 Pulse Width Modulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3.1 Sinusoidal Pulse Width Modulation . . . . . . . . . . . . . . . . . 12
2.3.2 Types of Sinusoidal Pulse Width Modulation . . . . . . . . . . . . 13
2.3.2.1 Bipolar Sinusoidal Pulse Width Modulation . . . . . . . . 13
2.3.2.2 Unipolar Sinusoidal Pulse Width Modulation . . . . . . . 13
2.4 Power System Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.4.1 Types of Stability . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

v
Contents vi

2.4.1.1 Steady State Stability . . . . . . . . . . . . . . . . . . . . 14

2.4.1.2 Dynamic Stability . . . . . . . . . . . . . . . . . . . . . . 14
2.4.1.3 Transient Stability . . . . . . . . . . . . . . . . . . . . . . 14
2.4.2 Synchronous Generator Stability . . . . . . . . . . . . . . . . . . . 14
2.4.2.1 Swing equation . . . . . . . . . . . . . . . . . . . . . . . . 15
2.4.2.2 Power Angle Curve . . . . . . . . . . . . . . . . . . . . . 15
2.4.2.3 Critical Angle . . . . . . . . . . . . . . . . . . . . . . . . 15
2.4.2.4 Equal Area Criterion . . . . . . . . . . . . . . . . . . . . 15
2.5 Artificial Neural Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.5.2 Supervised and Unsupervised Learning . . . . . . . . . . . . . . . . 16
2.5.2.1 Supervised Learning . . . . . . . . . . . . . . . . . . . . . 17
2.5.2.2 Unsupervised Learning . . . . . . . . . . . . . . . . . . . 17
2.5.3 Neural Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.5.3.1 Types of Neural Networks . . . . . . . . . . . . . . . . . . 19
2.5.3.2 Training Algorithms . . . . . . . . . . . . . . . . . . . . . 20
2.5.3.3 Stopping Criterion . . . . . . . . . . . . . . . . . . . . . . 20
2.6 Fuzzy Logic and Fuzzy Controller . . . . . . . . . . . . . . . . . . . . . . . 21
2.6.1 Fuzzy Logic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.6.2 Fuzzy Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.6.3 Types of Fuzzy Controller . . . . . . . . . . . . . . . . . . . . . . . 21
2.6.4 Mamdani Fuzzy Controller . . . . . . . . . . . . . . . . . . . . . . 21
2.6.4.1 Normalization . . . . . . . . . . . . . . . . . . . . . . . . 22
2.6.4.2 Fuzzification . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.6.4.3 Rule base . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.6.4.4 Defuzzification . . . . . . . . . . . . . . . . . . . . . . . . 24
2.6.4.5 Denormalization . . . . . . . . . . . . . . . . . . . . . . . 24
2.6.5 Neuro Fuzzy Inference System . . . . . . . . . . . . . . . . . . . . 24

3 Project Implementation and Methodology 26

3.1 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.2 Simulation Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.3 Software Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.4 Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.4.1 Inverter Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.4.1.1 Single Phase Square Wave Inverter . . . . . . . . . . . . . 27
3.4.1.2 Pulse Width Modulated Inverter . . . . . . . . . . . . . . 27
3.4.1.3 Trailing Edge Pulse Width Modulated Inverter . . . . . . 27
3.4.1.4 Leading Edge Pulse Width Modulated Inverter . . . . . . 28
3.4.1.5 Center Aligned Pulse Width Modulated Inverter . . . . . 28
3.4.1.6 Unipolar Pulse Width Modulated Inverter . . . . . . . . 29
3.4.1.7 Bipolar Pulse Width Modulated Inverter . . . . . . . . . 29
3.5 Open Loop Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.6 Closed Loop Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.7 Controller Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.7.1 Graphical User Interface . . . . . . . . . . . . . . . . . . . . . . . . 36
3.7.2 Selection of Parameters . . . . . . . . . . . . . . . . . . . . . . . . 37
Contents vii

3.7.3 Membership Functions . . . . . . . . . . . . . . . . . . . . . . . . . 37

4 Hardware Implementation and Results 38

4.1 Components . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.1.1 STM32F401 Microcontroller . . . . . . . . . . . . . . . . . . . . . . 38
4.1.2 Optocoupler tlp250 . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.1.3 Gate Driver IC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.1.4 IGBT IC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.1.5 LM7812 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.2 Isolated Supplies Implementation . . . . . . . . . . . . . . . . . . . . . . 41
4.3 Filter Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.4 Inverter Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.5 Square wave inverter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.6 SPWM inverter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.7 Final PCB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

5 Conclusion 45
5.1 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
5.2 Future Aspects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45

References 46
List of Figures

1.1 Flow diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

1.2 Schematic Diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2.1 Series Compensation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

2.2 Shunt compensation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
2.3 STATCOM block diagram . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.4 Half bridge inverter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.5 Full bridge inverter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.6 Output voltage waveform of Full Bridge Inverter . . . . . . . . . . . . . . 10
2.7 Three phase inverter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.8 Phase Voltage waveform for 120 degree conduction . . . . . . . . . . . . . 11
2.9 Line Voltage waveform for 180 degree conduction . . . . . . . . . . . . . . 11
2.10 Bipolar SPWM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.11 Unipolar SPWM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.12 Power Angle Curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.13 Equal Area Criteria in case of sudden increase in input . . . . . . . . . . . 16
2.14 Basic Structure of Neural Networks . . . . . . . . . . . . . . . . . . . . . . 18
2.15 Single Layer Perceptron . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.16 Basic Layout of Fuzzy Controller . . . . . . . . . . . . . . . . . . . . . . . 21
2.17 Triangular membership function . . . . . . . . . . . . . . . . . . . . . . . 23
2.18 Trapezoidal membership function . . . . . . . . . . . . . . . . . . . . . . . 23
2.19 Gauss membership function . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.20 Neuro Fuzzy Inference System structure . . . . . . . . . . . . . . . . . . . 25

3.1 Single phase square wave inverter simulation results . . . . . . . . . . . . 27

3.2 Trailing edge PWM inverter simulation result . . . . . . . . . . . . . . . . 28
3.3 Leading edge PWM inverter simulation result . . . . . . . . . . . . . . . . 28
3.4 Center aligned PWM inverter simulation result . . . . . . . . . . . . . . . 29
3.5 Unipolar Sinusoidal PWM inverter simulation result . . . . . . . . . . . . 29
3.6 Bipolar Sinusoidal PWM inverter simulation result . . . . . . . . . . . . . 30
3.7 Block Diagram of Single Phase Inverter . . . . . . . . . . . . . . . . . . . 30
3.8 Powerflow equations modeling in SIMULINK . . . . . . . . . . . . . . . . 31
3.9 Open loop SIMULINK diagram . . . . . . . . . . . . . . . . . . . . . . . . 31
3.10 Scope Window showing Active Power Flow . . . . . . . . . . . . . . . . . 32
3.11 Scope Window showing Reactive Power Flow . . . . . . . . . . . . . . . . 32
3.12 Sine block implementation in SIMULINK . . . . . . . . . . . . . . . . . . 33
3.13 PWM block implementation in SIMULINK . . . . . . . . . . . . . . . . . 33
3.14 Closed Loop implementation in SIMULINK . . . . . . . . . . . . . . . . . 34

viii
List of Figures ix

3.15 Scope Window showing Active Power Flow in closed loop . . . . . . . . . 35

3.16 Scope Window showing Reactive Power Flow in closed loop . . . . . . . . 35
3.17 Fuzzy Logic Controller in SIMULINK . . . . . . . . . . . . . . . . . . . . 36
3.18 Fuzzy Logic Controller GUI . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.19 GUI showing number of inputs and outputs . . . . . . . . . . . . . . . . . 37
3.20 Membership functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

4.1 STM32F4 Discovery Board . . . . . . . . . . . . . . . . . . . . . . . . . . 38

4.2 Optocoupler TLP250 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.3 IR2103 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
4.4 Igbt symbol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.5 Igbt IC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.6 LM7812 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.7 Isolated Supply block diagram . . . . . . . . . . . . . . . . . . . . . . . . 41
4.8 Isolated Supplies Implementation . . . . . . . . . . . . . . . . . . . . . . . 41
4.9 Filter implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.10 Filter Output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.11 Square wave inverter Output . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.12 SPWM pulses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.13 SPWM output . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.14 Final PCB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
Abbreviations

STATCOM Static Synchronous Compensator

SPWM Sinusoidal Pulse Width Modulation
CT Current Transformer
PT Potential Transformer
FACTS Flexible AC Transmission Systems
VSI Voltage Source Inverter
CSI Current Source Inverter
AC Alternating Current
ADC Analog toDigital Converter
DAC Digital to Analog Converter

x
Abstract
The purpose of our project is to enhance the stability of Power System against transients
with the application of Neuro Fuzzy Controller based Static Synchronous Compensator
(STATCOM). We know that the rotor of a generator rotates at a specific angle with
respect to the stator field and the angular position is proportional to the power generated.
This angle needs to be in its specific range for the synchronous and stable operation of
a generator and power system. In case of transient events like line outage and power
plant shut down the power stress on generator increases due to which rotor angle varies
and may cross its specified limit and generator may lost its synchronism with the rest
of the system which may cause other generators to lose their synchronism and this may
ultimately lead to blackout of whole system. Thus fast remedial measures are necessary
in case of transients so that the generator may not lose it’s synchronism. Our project
proposes the remedial solution by Static Synchronous Compensator (STATCOM)which
is controlled by Neuro Fuzzy Controller. The Statcom will provide or absorb the extra
power so that the rotor return to it’s normal value.The Neuro Fuzzy Control is fast
because of it’s non-linear nature so it can control STATCOM to provide compensation
in transient events.A three phase SPWM inverter with the DC battery as a DC link
source is developed to provide the compensated power. Power is continuously monitored
via PT and CT and processed in microcontroller and the controller will change the value
of angle and magnitude of STATCOM voltage in accordance with the power needed to
absorb or supply.
Chapter 1

Introduction

Stable and Synchronized operation of a Power System is crucial for the continuous and
regulated supply of power to consumers. Power System is exposed to several internal
and external disturbances which can be of steady state nature as well of transient nature.
There are different devices installed in a power system to compensate the effects caused
by these disturbances. Our project considers the transient disturbances and provides a
solution to compensate the effects caused by transient disturbances.This chapter gives
the Motivation and problem statement, aims and objectives of the project ,overview of
project, it’s schematic and flow diagram.

1.1 Motivation and Problem Statement

Power System is continuously exposed to different types of transient disturbances like
short circuit faults, line outages, increase in load and shut down of a plant. These distur-
bances directly affect the generator because these cause generator to undergo production
change. This production change directly affects the rotor’s position of the generator
which is at a certain angle with the stator. The rotor can make an angle with the stator
within a certain limit so that the generator does not lose it’s synchronism. If in case of
any transient interval the rotor crosses it’s upper and lower angle limit than the genera-
tor can become unstable and may lose it’s synchronism with the remaining system.And
we know that the generators in power system are in synchronism with each other so the
event may occur that the unstable generator may cause other generators to lose their
sysnchronism which may lead to blackout. So, it is necessary to deal with the effects
of transients. To deal with this problem a STATCOM is connected in parallel which
compensate for the extra power and to deal with the transients Neuro Fuzzy Controller
is proposed because it is fast and non-linear

1
Chapter 1. Introduction 2

1.2 Overview of the project

STATCOM is a shunt compensation device.It is a controlled voltage source inverter
(usually SPWM based) which can compensate both active and reactive power when
DC link supply is battery and only reactive power when DC link voltage is provided
by capacitor.The quantities which are controlled in STATCOM are the inverter output
voltage magnitude and phase. The quantities which are processed in controller are
current and voltage. The flow diagram of the project is shown in Figure 1.1.

Figure 1.1: Flow diagram

Chapter 1. Introduction 3

STATCOM is implemented as a three phase Sinusoidal PWM inverter with a DC battery

as a DC link source. The STATCOM is connected to the grid using coupling transformer
which also acts as a low pass filter. The voltages and currents are measured with
the CT’S and PT’s and fed to the controller which produces the required amount of
amplitude and phase angle to provide or absorb power.A Neuro Fuzzy Controller is
designed which adjust the PID gains dynamically thus responding to uncertainties and
making the response fast and robust. The schematic diagram of the project is shown in
Figure.1.2

Figure 1.2: Schematic Diagram

Chapter 1. Introduction 4

1.3 Objectives
The project objective is to develop a STATCOM. The second objective is to make the
response time of STATCOM fast by the application of Neuro Fuzzy Control because of
it’s non-linear and robust nature.

1.4 Aims
The project aims to enhance the Power System Stability by compensating the effects of
transient events.
Chapter 2

Theoretical Background

2.1 Static Synchronous Compensator

2.1.1 Need for Compensation
In this modern era, continuous increase in load demand has put tremendous stress
on power system. Due to increase in demand, reactive power which is necessary for
the flow of useful active power either increases or decreases beyond the certain limit
which disturbs the voltage regulation, increases power losses,decreases power factor and
decreases the stability margin of power system. Thus reactive power is need to be
compensated for the stable operation of power system. The reactive power is either
compensated by connecting a voltage source in series (series compensation) with the
bus or connecting current source in parallel (shunt compensation).

2.1.2 Series Compensation

In series compensation, a voltage source is connected in series with the bus. By adjusting
the value of voltage load voltage can be regulated to obtain the desired power factor.
Figure 2.1 is showing the line having voltage source connected in series and the phase
diagram depicting the improvement in power factor

5
Chapter 2. Theoretical Background 6

Figure 2.1: Series Compensation

2.1.3 Shunt Compensation

In shunt compensation, a current source is connected in parallel with the bus.The extra
current required by the inductive load is supplied by the current source and thus the
stress on line reduces and losses reduce.Figure 2.2 is showing line having current source
connected in parallel and the phasor diagram showing the process of series compensation

Figure 2.2: Shunt compensation

2.1.4 Flexible AC Transmission Systems

To overcome the problems which arise due to excess or shortage of reactive power and
for the better transmission of AC power several compensating devices have been devel-
oped which are called Flexible AC Transmission Systems. These include SVC’s, TCR’s,
TCSC’s etc.

2.1.5 Static Synchronous Compensator

STATCOM is a type of FACTS devices. It is a shunt connected compensation device.
It can compensate both active and reactive power.
Chapter 2. Theoretical Background 7

2.1.5.1 Construction and Working Principle

STATCOM is a three phase VSI (Voltage Source Inverter). The DC voltage is provided
to inverter either by capacitor or battery. If the energy storage element is a capacitor it
can exchange only reactive power. It can exchange both active and reactive power if the
DC source is a battery. The STATCOM is connected to the point of compensation by a
coupling transformer. The STATCOM exchange power on the basis of voltage and angle
difference.If the STATCOM is represented by a voltage source having magnitude and
phase V1 and δ and the grid is represented by a voltage source having magnitude V2 and
angle 0 and these two are connected by a coupling inductance X than the active power
flow is from leading source to the lagging source and reactive power from source having
voltage magnitude higher. The active and reactive power exchanged between STATCOM
and grid at the point of coupling takes place in accordance with the following powerflow
equations[9]
P = V 1V 2sinδ/X
Q = V 1(V 1 − V 2cosδ)/X
The basic block diagram of STATCOM is shown in Figure 2.3

Figure 2.3: STATCOM block diagram

2.1.5.2 DC Link Voltage

DC Link source dictates whether STATCOM will exchange only reactive power or both
active and reactive power. If the inverter is fed from Capacitor than the STATCOM
can only exchange reactive power and if the inverter is fed from the DC battery than
the STATCOM can exchange both active and reactive power.

2.1.5.3 STATCOM Control

In Statcom the controllable quantities are the output voltage magnitude and phase
angle.STATCOM is controlled by either PWM control or Phase Angle Control

2.1.5.4 Phase Angle Control

In Phase angle control, modulation index is kept constant and the controlled quantity
is phase angle. The output voltage is controlled by changing the DC link voltage.
Chapter 2. Theoretical Background 8

2.1.5.5 PWM Control

In PWM Control,the controlled quantity is amplitude of the reference signal and the
output voltage is controlled by changing the modulation index. The output voltage
of inverter is changed by varying the modulation index in accordance with reactive
power required to absorb or provide. In STATCOM, we use Sinusoidal Pulse Width
Modulation.

2.2 Inverters
Inverter is a power electronic circuit which converts DC into AC of desired frequency and
amplitude. Inverters produce AC by the switching of power electronics based switches.
Inverters have many applications like in UPS, variable AC frequency drives, induction
heating etc.

2.2.1 Types of Inverter

Inverter has many types on the basis of type of source, on the basis of output waveform
and on the number of phases.

2.2.1.1 On the basis of type of Source

On the basis of type of source inverters are classified into two types:
Voltage Source Inverters
In voltage source inverters (VSI’s), voltage is fed from DC voltage source and there is a
capacitor connected in parallel with the DC source.
Current Source Inverters
In Current source inverters (CSI’s), voltage is fed from DC current source and there is
an inductor connected in series with the DC source.

2.2.1.2 On the basis of output waveform

On the basis of output waveform, the inverters are of three types:

Square wave inverter
In this type of inverter, the output waveform generated by an inverter is of square shape.
The square wave as we know is a combination of many sinusoidals thus it has many har-
monics.

Modified Sine wave inverter

In modified sine wave inverter,the output waveform is square but the difference is that
the output becomes zero for a small interval of time before polarity its state from posi-
tive to negative and vice versa.
Chapter 2. Theoretical Background 9

Sine wave inverter

The output of this type of inverter is a pure sine wave.

2.2.1.3 On the basis of number of phases

Inverters are classified into two types on the basis of number of phases:

• S ingle phase inverter.

• T hree phase inverter.

Single phase inverter

Single phase inverters are of two types:

• H alf Bridge Single phase inverter.

• F ull Bridge Single phase inverter.

Half Bridge Single phase inverter

In half bridge inverter, there are two switches. The output of half bridge is one portion
of the square wave.This scheme has disadvantage that the source utilization is half.Half
bridge circuit taken from [10] is shown in figure 2.4

Figure 2.4: Half bridge inverter

Chapter 2. Theoretical Background 10

Full Bridge Single phase inverter

In full bridge inverter, there are four switches. Two of the four switches are ON at the
same time The output of half bridge is a square wave.This scheme has advantage that
the source utilization is increased. Full bridge inverter circuit taken from[10] is shown
in figure 2.5 The output of a full bridge inverter is a square wave. Output is shown in

Figure 2.5: Full bridge inverter

figure 2.6

Figure 2.6: Output voltage waveform of Full Bridge Inverter

Three phase inverter

In Three phase inverter, there are six switches. Three of the six switches are ON at the
same time. There are two modes of conduction for a three phase inverter.

• 1 20 degree Conduction.

• 1 80 degree Conduction.

Three phase inverters are mostly used in grid connected applications. These are used in
UPS and also in other applications.
Chapter 2. Theoretical Background 11

Three phase inverter circuit taken from[10] is shown in figure 2.7

Figure 2.7: Three phase inverter

120 degree conduction

In 120 degree conduction, two transistors remain ON at any instant of time and each
transistor remains in conduction mode for 120 degrees.Waveform of 120 degree conduc-
tion mode taken from[10] is shown in figure 2.8

Figure 2.8: Phase Voltage waveform for 120 degree conduction

180 degree conduction

In 180 degree conduction, three transistors remain ON at any instant of time and each
transistor remains in conduction mode for 180 degrees.Waveform of 180 degree conduc-
tion mode taken from[10] is shown in figure 2.9

Figure 2.9: Line Voltage waveform for 180 degree conduction

Chapter 2. Theoretical Background 12

2.3 Pulse Width Modulation

In pulse width modulation, we vary the duty cycle of pulses to change the voltage.Pulse
Width Modulation is most commonly used to control inverter output voltage. There are
different schemes available for Pulse Width Modulation which are listed below:

• S inusoidal Pulse Width Modulation.

• M odified Sinusoidal Pulse Width Modulation

• S pace Vector Pulse Width Modulation.

• T hird Harmonic Pulse Width Modulation

• 6 0 degree Pulse Width Modulation.

• S ingle Pulse Width Modulation

• M ultiple Pulse Width Modulation

2.3.1 Sinusoidal Pulse Width Modulation

In Sinusoidal Pulse Width Modulation,we vary the duty cycle sinusoidally that’s why
called as sinusoidal PWM.In sinusoidal PWM the reference signal is a sinusoid having
a specific frequency and the carrier signal is a triangular or sawtooth signal having high
frequency greater than the reference signals. These are compared with each other and
the resultant pulses are fed to the controller. The resultant output produced by an
inverter is in pulsated form which is than filtered through the filter to obtain a signal
having reference frequency and a sinusoidal wave is produced[8]. Reference Signal
Reference signal in sinusoidal pulse width modulation is a sinusoid having frequency
much smaller than the carrier frequency.We change the amplitude of reference signal to
vary the voltage.
Carrier Signal
A carrier signal is a high frequency signal (usually in kHz). It can be a triangular signal
or a sawtooth signal.
Modulation Index
We change the duty cycle of pulses by either varying the frequency (frequency mod-
ulation) or by amplitude of reference signal (amplitude modulation). The modulation
index is the ratio of the frequency of reference to the frequency of carrier or the ratio
of amplitude of reference and amplitude of carrier wave. For amplitude pulse width
modulation, modulation index is given by:

ma = V r/V c (2.1)

It is in the range between 0 and 1. With 0 being the smallest duty cycle and 1 being
the largest duty cycle.
Chapter 2. Theoretical Background 13

2.3.2 Types of Sinusoidal Pulse Width Modulation

Sinusoidal Pulse Width Modulation can be implemented in following two ways:

• U nipolar Sinusoidal Pulse Width Modulation.

• B ipolar Sinusoidal Pulse Width Modulation

2.3.2.1 Bipolar Sinusoidal Pulse Width Modulation

In Bipolar Sinusoidal Pulse Width Modulation, the carrier signal is triangular wave and
the reference signal is compared with it to produce bipolar signal. When we apply the
pulses which are produced by this scheme the mosfets in two legs are ON/OFF at the
same time[6]. This scheme can only be applicable to the single phase inverters because it
includes only one reference signal for the pulses.The figure 2.10 is showing the Unipolar
Sinusoidal Pulse Width Modulation

Figure 2.10: Bipolar SPWM

2.3.2.2 Unipolar Sinusoidal Pulse Width Modulation

In this SPWM scheme, there are two references signals one is with 0 degree phase while
one is 180 degree of other. This scheme is used for both single phase and three phase
inverters because it includes three different reference signals displaced by 120 degrees
from each other[7]. Figure 2.11 is showing Unipolar Sinusoidal Pulse Width Modulation

Figure 2.11: Unipolar SPWM

Chapter 2. Theoretical Background 14

2.4 Power System Stability

According to Power System Analysis by Hadi Saadat[12], the stability of a Power System
is defined as the ability to develop restoring forces against disturbances. Power System
is subjected to many small and big disturbances. Some exist for small interval of time
while others exist for longer interval of time.

2.4.1 Types of Stability

According to the type of disturbances, the stability of a power system is classified into
three categories:

2.4.1.1 Steady State Stability

By Steady State Stability, mean the ability of a system to return to it’s original state
after being subjected to slow gradual changes. These slow gradual changes are not large
and continuously occur in power system[12].

2.4.1.2 Dynamic Stability

Dynamic Stability derives it’s essence from Steady State Stability but the difference
between the two is that the dynamic stability involves the disturbances which last for
long intervals of time with the inclusion of control devices[12].

2.4.1.3 Transient Stability

By transient stability we mean the ability of a system to return to its equilibrium

state after subjected to transient disturbances. By transient disturbances we mean the
disturbances which last for very small interval of time and are sudden and large in nature.
Transient stability studies are crucial to make sure the stable operation of power system
after major disturbance or change. Transient disturbances may include the disturbances
arise from events like outage of a line due to any fault, power shutdown of a major power
station, short circuit faults etc [12].In transient stability studies we majorly deal with
the stability of synchronous generators.

2.4.2 Synchronous Generator Stability

The stability of a power system is ensured when all the generators are working well in
synchronism. We know that in power system all the generators are in synchronism with
each other. If one generator due to any reason go out of this synchronism chain than it
can cause other generators to lose their synchronism and the power system can become
unstable if multiple generators goes out of synchronism.We know that in synchronous
generators we have stator moving with synchronous speed and the rotor moves at the
relative speed at a certain angle with the stator which is called the power angle or torque
angle.This angle depends on the difference of mechanical and electrical power. If the
disturbance occurs than the rotor wither accelerate or deaccelerate. If the disturbance
involves any change in net power than the rotor changes its relative position i.e, power
angle and comes to a new position.
Chapter 2. Theoretical Background 15

2.4.2.1 Swing equation

The equation relating the relative motion between rotor and stator field due to change
in power is called a swing equation. The swing equation relating the electrical power
angle and change in power is given next:

p/2*dδ/dt=Pm-Pe

2.4.2.2 Power Angle Curve

The relation between electrical power and torque angle is graphically represented by the
curve known as the Power Angle Curve. The power angle curve taken from[12] is shown
in figure 2.12

Figure 2.12: Power Angle Curve

2.4.2.3 Critical Angle

For a stable operation of the generator,the rotor needs to make an angle with stator in a
certain range (between 0 to 180 degree). If the angle goes below the lower threshold or
beyond the upper threshold the generator will become unstable. The angle after which
the generator becomes unstable is called critical angle.

2.4.2.4 Equal Area Criterion

Equal Area Criterion is the graphical approach to depict the stability of generator. It
shows the energy stored in the rotor to show that whether generator will exhibit stable or
an unstable operation when exposed to different disturbances. It has many applications
like studying the behaviour of generator under different situations like three phase fault
in transmission line, sudden change in power input etc.It is very easy to infer about
the stability profile while looking at the equal area criterion graph because it gives
an observer a crucial information about the generator’s behaviour at a glance without
solving the complex non-linear swing equation. Equal Area Criteria applied to study the
generator stability in case of sudden increase in power input is discussed for elaborating
and highlighting the importance and application of Equal Area Criterion in stability
Chapter 2. Theoretical Background 16

studies.
Application to sudden increase in power input
This case has taken from Power System Analysis by Hadi Sadat[12].Suppose we have
a generator operating at the angle delta0 at steady state condition having mechanical
power (Pm) equal to electrical power (Pe). Suppose mechanical power is increased to
Pm2. Now due to increase in mechanical power rotor will accelerate and let at the angle
δ mechanical power becomes equal to the electrical power. The new operating point
becomes this point but due to momentum electrical power continues to increase and
due to excess of electrical power than mechanical power the rotor decelerate. Now the
rotor oscillates between the two points and ultimately settles to point corresponding to
δ. According to equal area criterion, for the generator to operate in the stable zone the
two areas should be equal, the one where rotor gains energy while accelerating and the
one where rotor loses energy while decelerating. This is shown in figure 2.13

Figure 2.13: Equal Area Criteria in case of sudden increase in input

2.5 Artificial Neural Networks

2.5.1 Introduction
Artificial Neural Networks is a very vast area of Machine Learning and lot of research
is being carried on them.It is an information processing algorithm. It is inspired from
the neural system of humans. Like in human nervous system, several neurons are in-
terconnected in a highly complex and interlinked system and transfer information by
coordinating with each other in a systematic way,the same is the case in artificial neural
networks.Artificial Neural Networks are nowadays very popular and have many applica-
tions like in robotics,medicine,automobile industry, security systems and nowadays re-
search is being conducted to inculcate neural networks in several devices used in power
systems.

2.5.2 Supervised and Unsupervised Learning

Machine Learning algorithms work by learning and training. Like a human brain learns
things and make decisions by pondering upon several times and by taking into consider-
ation several facts, figures, past experiences likewise machine learning algorithms learn
things by training. The learning can be supervised learning or unsupervised learning.
Chapter 2. Theoretical Background 17

2.5.2.1 Supervised Learning

In supervised learning, our algorithm is trained by using several examples. The output
class of these examples is known. The algorithm takes the input and than decides to
which output class does the input belong. This type of learning is used in gesture recog-
nition,classification problems etc.

2.5.2.2 Unsupervised Learning

In unsupervised learning, the output class is unknown and the output can be any function
of input data.

2.5.3 Neural Networks

Neural Networks consists of several units(neurons) connected in form of layers. These
neurons take input,process it and make a decision. There are three layers in a neural
network which are listed below

• I nput Layer

• H idden Layer

• Output Layer

The neurons in input layer takes input as vector made up of features.These features
can be of image e.g, colors, pixel values, numbers etc. The hidden layer processes this
information and neurons in output layer computes the output and take action. The
number of input and output layer is one while number of hidden layers can be more
than one depending on the complexity of problem. The number of inputs and outputs
can be more than one. The neurons in one layer are connected to the neurons in other
layer through the weighted connections.In the structure of neural networks one neuron
is connected to all other neurons in adjacent layer[3].The basic structure of Neural Net-
works is shown in figure 2.14

Figure 2.14: Basic Structure of Neural Networks

Chapter 2. Theoretical Background 18

Weights
The neurons are connected to each other through the connections. These connections
are weighted i.e, these have some strengths. These weights play very important role
in the decision making process. In training the network we actually update weights on
each iteration and we get the optimized weights after training which help in making
decision in unknown situation.These weights can be optimized by training the network
using different algorithms like least square, back propagation etc.

2.5.3.1 Types of Neural Networks

Neural Networks can be classified on the basis of propagation method or number of

layers
Types on basis of propagation method
On the basis of type of propagation, there are two types of neural networks.

• F eedforward Neural Networks

• F eedback Neural Networks

Feedforward Neural Networks

In feedforward neural networks there is not feedback or cycles or loops and there is one
way flow of information. The input layers accept input pass it to the hidden layer which
after processing it pass it to output layer which produces a output.
Feedback Neural Networks
In feedback neural networks there are cycles or loops and there is a two way flow of
information. The information can be passed to output and also from output to hidden
layer back using loops.
Types on basis number of layers
On the basis of number of layers,there are two types of neural networks.

• S ingle Layer Perceptron.

• M ultilayer Perceptron.

Single Layer Perceptron

In single layer perceptron, input directly goes into the output as a weighted sum. Output
neuron gives an output of 0 or 1 depending on the threshold value. These neurons are
also called threshold neurons[11]. Figure 2.15 shows a perceptron

MultiLayer Perceptron
In multilayer neural networks there is one input layer, one or more hidden layers and one
output layer. The information is passed through the input layer to the hidden layer and
than to output layer where output is calculated. Figure 2.14 is an example of multilayer
perceptron.
Chapter 2. Theoretical Background 19

Figure 2.15: Single Layer Perceptron

2.5.3.2 Training Algorithms

There are several algorithms available for the training of neural networks. Each one has
its own advantages and limitations. Following are the algorithms which are commonly
used to train neural networks

• M ethod of Least Square.

• Gradient Descent and Delta Rule.

• M ethod of Back Propagation.

Method of Least Square

In method of least square, we make a quadratic cost function. Cost function is the
function which we optimize in order to get the optimized weights. The cost function in
this method is
costf unction = (desiredo ut − predictedo ut)2 (2.2)

In this method we update the weights on each iteration based on the error value squared.
This method has less accuracy than the other methods
Gradient Descent Method
In this method we find the rate of change of weights with respect to error and we are
in search of error such that the rate of change (gradient) is minimum. That’s why it is
called gradient descent method.
Method of Back Propagation
This method is mostly used to train multilayer neural networks. In this method we back
propagate the errors. Feed forward output is calculated and the error is back propagated.
The error terms for the hidden units and output units are calculated. The weights are
updated on the basis of errors. This process continues until a specific termination criteria
is met.

2.5.3.3 Stopping Criterion

The stopping criterion to train the network can be of two types:Either it can be on the
basis of no of epochs (iterations) or error tolerance.
Chapter 2. Theoretical Background 20

2.6 Fuzzy Logic and Fuzzy Controller

2.6.1 Fuzzy Logic
Fuzzy Logic as it’s name implies deals with uncertainty. Unlike binary logic which gives
answer in binary variables YES(1) or NO(0) fuzzy logic considers the in between values
also. It has the ability to depict the situations where binary logic fails.It considers all
the possibilities thus can represent the scenerios where there is uncertainty, doubt and
fuzziness.

2.6.2 Fuzzy Controller

Fuzzy Controller is a controller which is based on Fuzzy Logic. It is employed in situ-
ations where there is an uncertainty.Nowadays, fuzzy controllers are employed in many
devices like washing machines, electric kiln etc.It is considered as an intelligent control
because it infers result based on human like reasoning[1]. It is fast and robust control
because it does not require any mathematical model and thus requires less computation
time. Nowadays, lot of research and study is being conducted to employ fuzzy controllers
in devices used for the control of equipments used in power system.

2.6.3 Types of Fuzzy Controller

Fuzzy Controllers are of two types:

• M amdani Type Fuzzy Controller

• S ugeno Type Fuzzy Controller

2.6.4 Mamdani Fuzzy Controller

This type of fuzzy controller is commonly used throughout the world.The block diagram
representing the basic layout of fuzzy controller is shown in Figure.2.16

Figure 2.16: Basic Layout of Fuzzy Controller

Chapter 2. Theoretical Background 21

From the above block diagram we can see that there are five blocks or processes in a
fuzzy controller.

• N ormalization

• F uzzification

• Rule Base

• Defuzzification

• Denormalization

2.6.4.1 Normalization

In normalization, we convert the crisp values into the normalized values which can be
included in universe of discourse.
Crisp values
Crisp values are the values which are in standard units that is real world values.
Universe of discourse
In fuzzy logic universe of discourse is the input data range in which the crisp value
should lie after normalized

2.6.4.2 Fuzzification

Fuzzification is a stage in which the normalized input is fuzzified i.e. it is assigned the
value in between 0 and 1. This fuzzy value is assigned by membership functions. Mem-
bership Function
The membership function is the function which decides how much membership value a
particular variable has in each function. There are several membership functions and
selection of membership function depends on the complexity of problem at hand.The
membership functions which are generally used are triangular, bell shaped, s-shaped,z-
shaped,trapezoidal etc.Some of them are described below.
Triangular membership function
Triangular membership function is a function of three parameters. Upper and lower
limits and midpoint. For example, for upper limit m and lower limit n and midpoint o
the triangular function is defined as:
f(x;m,o,n)= 0, x≤m
x−m
o−m , m≤x≤n
n−x
c−o , o≤x≤n
0,n≤x
Chapter 2. Theoretical Background 22

Figure taken from[6] is showing triangular function graphically

Figure 2.17: Triangular membership function

Trapezoidal membership function

Trapezoidal membership function is a function of four parameters. For example, for
points m,n,o and p trapezoidal function is defined as:
f(x;m,n,o,p)= 0, x≤m
x−m
n−m , m≤x≤n
1, n≤x≤o
p−x
p−o , o≤x≤n
0,p≤x
Figure taken from[5] is showing trapezoidal function graphically

Figure 2.18: Trapezoidal membership function

Gauss membership function

Gauss membership function is a function of σ of point. For example, for point c gauss
function is defined as:
f(x;σ)= e−(x−c)2sigma2
Chapter 2. Theoretical Background 23

Figure taken from[2] is showing gauss function graphically

Figure 2.19: Gauss membership function

2.6.4.3 Rule base

Fuzzy controller takes decision on basis of the rules. These rules are defined in the
form of rule base. These rules are made by the experienced man who has a sound
knowledge about the operation and behaviour of system. The fuzzy rules are in form of
if-else rules[4]. In these rules the first part is called antecedent while the second part is
called precedent.Each rule has a strength or value associated with it which is called rule
strength.

2.6.4.4 Defuzzification

In inference process different rules strength are found and decision is made on the basis
of these strengths and the crisp value is obtained.Different defuzzification techniques are
available which include centroid method,bisection method etc.

2.6.4.5 Denormalization

The crisp value obtained after defuzzification is than denormalized so that it comes back
in it’s standard range.

2.6.5 Neuro Fuzzy Inference System

Neuro Fuzzy Inference is a combination of Neural Networks and Fuzzy Logic.It is devel-
oped by merging the good qualities of both Neural Networks and Fuzzy Logic. Neural
Networks have a learning capability as these can learn the new situations by updating
the weights so with the passage of time these can update themselves according to the
conditions. Fuzzy Logic unlike Neural Networks has no learning capability i.e, it is not
adaptive but it has a inference capability. It has human like decision taking power and
can make decisions under uncertainties and vague conditions on the basis of rules. Also
less computation time is required for Fuzzy Logic because decision is made on the basis
of rules. In Neuro Fuzzy Inference System, we can have both the learning capability
and decision taking qualities of Neural Networks and Fuzzy Logic respectively. In Neuro
Fuzzy inference System,we have fuzzy membership functions which are tuned by neural
Chapter 2. Theoretical Background 24

networks. We have a neural networks like structure in which rules are implemented as
neurons and these rules connected through weighted connections. The basic structure of
neuro fuzzy inference system is shown in fig 2.20 In the above figure we can see that the

Figure 2.20: Neuro Fuzzy Inference System structure

first layer in this structure is fuzzification layer. Second and third layer is of antecedent
and consequent of rules and in final layer we have a aggregator.
Chapter 3

Project Implementation and

Methodology

3.1 Methodology
STATCOM has been implemented first by the single phase inverter and than by the
three phase inverter.Than STATCOM was be synchronized with grid. The voltage and
current are be measured using PT and CT. The power of generator is monitored. In
case of disturbance the extra power will be provided or absorbed by STATCOM. The
control is based on Neuro Fuzzy and Fuzzy PID.
In PID, we tune the gains of proportional, derivative and integral. But since these gains
provide same type of response in all situations and cannot handle the uncertainty so we
use Fuzzy PID instead of simple PID.In fuzzy PID,we decide the gains on the basis of
rules. Thus the uncertainty will be dealt and the gains will be distributed and scheduled.
This strategy makes the response fast as compared to the simple PID.The project will
be tested first on simulation and than on hardware. The project is implemented on two
levels:

• S imulation Implementation.

• P ractical Implementation.

3.2 Simulation Implementation

On simulation level,we implemented the project on MATLAB.We used SIMULINK to
make a STATCOM and Neural Networks and Fuzzy Logic Toolbox of MATLAB to
implement Neuro Fuzzy.

3.3 Software Introduction

MATLAB is an authorized software developed by Mathworks Corporation. It is mostly
used to solve Matrices as it’s name implies MATLAB. MATLAB is an efficient tool to

25
Chapter 3. Project Implementation 26

solve complex calculations. It’s library known as SIMULINK has toolboxes related to
many fields like Power Systems, Control Systems, aerospace engineering, Fluid mechan-
ics, Fuzzy Logic Toolbox etc.It is used extensively for educational purposes and also in
industries.

3.4 Simulations
3.4.1 Inverter Simulations
First we simulated single phase inverter based on different schemes and observed their
results. Than we synchronized the inverter (STATCOM) with the inverter observed the
response in open loop and than observed the response in closed loop.

3.4.1.1 Single Phase Square Wave Inverter

We simulated single phase square wave inverter. The inverter is implemented using
generic IGBT switches taken from powerlib library. The pulses are applied using pulse
generator. The load used is series RLC branch. Figure 3.1 is showing the pulses, output
voltage and current waveforms of inverter displayed in the scope window

Figure 3.1: Single phase square wave inverter simulation results

3.4.1.2 Pulse Width Modulated Inverter

Secondly we simulated pulse width modulated inverters based on different schemes.We

simulated center aligned, leading edge, trailing edge, unipolar and bipolar sinusoidal
pwm inverters.

3.4.1.3 Trailing Edge Pulse Width Modulated Inverter

We simulated trailing edge pulse width modulated inverter. In trailing edge pulse width
modulated inverter, the carrier wave is a sawtooth starting from 0.
Chapter 3. Project Implementation 27

Figure 3.2 is showing the pulses, output voltage and current waveforms of inverter dis-
played in the scope window

Figure 3.2: Trailing edge PWM inverter simulation result

3.4.1.4 Leading Edge Pulse Width Modulated Inverter

We simulated leading edge pulse width modulated inverter. In leading edge pulse width
modulated inverter, the carrier wave is a sawtooth signal starting from 1. Figure 3.3 is
showing the pulses, output voltage and current waveforms of inverter displayed in the
scope window

Figure 3.3: Leading edge PWM inverter simulation result

3.4.1.5 Center Aligned Pulse Width Modulated Inverter

We simulated center aligned pulse width modulated inverter. In center aligned pulse
width modulated inverter, the carrier wave can be a sawtooth or triangular wave signal
starting from 0.
Chapter 3. Project Implementation 28

Figure 3.4 is showing the pulses, output voltage and current waveforms of inverter dis-
played in the scope window

Figure 3.4: Center aligned PWM inverter simulation result

3.4.1.6 Unipolar Pulse Width Modulated Inverter

We simulated single phase pwm inverter based on unipolar scheme. The carrier frequency
was selected to be 10k HZ.The two refernce signals which are 180 degree out of phase
are provided by using two separate sinusoidal signal blocks.

Figure 3.5: Unipolar Sinusoidal PWM inverter simulation result

3.4.1.7 Bipolar Pulse Width Modulated Inverter

We simulated single phase pwm inverter based on bipolar scheme. The carrier frequency
was selected to be 10k HZ.The reference signal is provided by sinusoidal signal block.
Chapter 3. Project Implementation 29

Figure 3.6 is showing the results of Bipolar Pulse Width Modulated Inverter

Figure 3.6: Bipolar Sinusoidal PWM inverter simulation result

3.5 Open Loop Simulation

After simulating different schemes of inverter, we simulated the statcom in open loop.
We made unipolar single phase SPWM inverter. For LC filter,we calculated the values
of L and C using transfer function and plotting the gain of transfer function versus
frequency to find the values of L and C where the gain is 1.We found that at these
values the higher order components were approaching towards almost zero gain except
at 25kHz there was a harmonic component which was providing a gain of 10. We used
the following values of L and C:

• L=1mH.

• C =4e-6.

Figure 3.7 is showing the simulink block diagram of single phase SPWM inverter

Figure 3.7: Block Diagram of Single Phase Inverter

Chapter 3. Project Implementation 30

For connecting the STATCOM we connected it through coupling inductor having value
1e-3H. For grid we used the sinusoidal source and we connected the STATCOM with
grid through circuit breaker to which step input is applied. The circuit breaker will
remain open for the some duration of simulation running and at a time specified a step
input will be applied and circuit breaker becomes close and power exchange between
STATCOM and grid starts to take place. We run the simulation such that the specific
amount of active and reactive power should be exchanged between STATCOM and grid.
To find the required value of phase and amplitude of inverter which would produce this
much power we used the power flow equations. We made a subsystem which would
accept the specified values of P and Q and will give the required value of phase angle
and inverter output voltage amplitude.Figure 3.8 is showing the power flow equations
modeled in SIMULINK.

Figure 3.8: Powerflow equations modeling in SIMULINK

By finding required values of inverter voltage amplitude and phase angle we applied
the required DC and made us STATCOM lag or lead depending on requirement.The
SIMULINK diagram after grid connection is shown in figure 3.9

Figure 3.9: Open loop SIMULINK diagram

Chapter 3. Project Implementation 31

We ran the simulation for the random value of P and Q to see that whether the power
exchange takes place or not. We ran the simulation for P=100 W and Q=50 VAR.We
observed that there was a steady state error between the desired value and the power
exchange took place. We ran the simulation for 0.5s. We set the step time to be 0.2s. We
observed that until the breaker closes the power flow is almost zero but as soon as the
breaker closes power starts to flow between the two sources and power reaches almost
98 W and 48 VAR. Figure 3.10 is showing the P that flowed shown in scope window
Figure 3.11 is showing the Q that flowed shown in scope window

Figure 3.10: Scope Window showing Active Power Flow

Figure 3.11: Scope Window showing Reactive Power Flow

Chapter 3. Project Implementation 32

3.6 Closed Loop Simulation

After performing the simulation in open loop, we simulated the STATCOM in closed
loop by controlling it through the controller. We used the Fuzzy Logic Toolbox to
design the controller. Since in closed loop we have to continuously vary the amplitude of
inverter and phase angle of reference signal. Since in SIMULINK we cannot change the
internal parameter values of block on run-time so in order to deal with this problem we
made our own block of sinusoidal signal. This block accepts the amplitude and phase
and produces the sinusoidal according to these updated values. Figure 4.12 is showing
the SIMULINK block diagram of this sinusoidal signal. With this sinusoidal we modified

Figure 3.12: Sine block implementation in SIMULINK

our PWM block. Our PWM block now updates on run time. Figure 3.13 is showing
PWM block implemented in SIMULINK.

Figure 3.13: PWM block implementation in SIMULINK

Chapter 3. Project Implementation 33

In closed loop, we pass the error of power, rate of change of error and integration of
error into the controller and the controller than gives the values of gains which are than
multiplied with the error,rate of change of error and integration of error respectively
and new values of P and Q are than passed into the subsystem to find the new values
of modulation index and phase angle. The powerflow equation module gives us RMS
value of voltage required to produced by STATCOM and the phase. From this we have
to calculate the value of DC which would actually produce that much voltage. For this
we calculate the value of DC voltage by using KCL and considering the drops associated
with filter and coupling inductor. The calculated value of DC voltage is than divided
by the applied DC to find the modulation index. Figur 3.14 is showing the Simulink
diagram of closed loop implementation.

Figure 3.14: Closed Loop implementation in SIMULINK

We ran the simulation for the same values of P and Q to see that how the power exchange
takes place in closed loop. We ran the simulation for 1s. We set the step time to be
0.5s. We observed that until the breaker closes the power flow is almost zero but as
soon as the breaker closes power starts to flow between the two sources. Figure 3.15 is
showing the P that flowed shown in scope window Figure 3.16 is showing the Q that
flowed shown in scope window
Chapter 3. Project Implementation 34

Figure 3.15: Scope Window showing Active Power Flow in closed loop

Figure 3.16: Scope Window showing Reactive Power Flow in closed loop
Chapter 3. Project Implementation 35

3.7 Controller Implementation

We implemented Controller in Simulink using Fuzzy Logic Toolbox. It is equipped with
all the tools required for the development of Fuzzy Controller. It has a GUI where user
can choose membership functions,defuzzification method, and also has a rule base where
user can store rules. In SIMULINK there are two variants available for Fuzzy Logic
Toolbox. One is Fuzzy Logic Controller and other is Fuzzy Logic Controller with Rule
Viewer.The difference between the two is that in later there is a separate interface for the
observation of variations of rules and output.The fuzzy logic controller in SIMULINK
has following form:

Figure 3.17: Fuzzy Logic Controller in SIMULINK

3.7.1 Graphical User Interface

Fuzzy Logic Controller implemented in SIMULINK has a Graphical User Interface. It
provides developer to specify the number of inputs and outputs, specify type of mem-
bership functions and defuzzification method. It also has a built in rule base where user
can define and modify rules. Figure 4.18 gives the glimpse of this GUI

Figure 3.18: Fuzzy Logic Controller GUI

Chapter 3. Project Implementation 36

3.7.2 Selection of Parameters

We have implemented the Mamdani type fuzzy controller. We have a three input and
three output system. We used the min implication and aggregation method as sum and
defuzzification method as centroid method.Figure 3.19 is showing the GUI displaying
the inputs and outputs

Figure 3.19: GUI showing number of inputs and outputs

3.7.3 Membership Functions

We used trapezoidal and triangular membership functions. We defined the input and
output ranges. Figure 3.20 is showing input membership functions

Figure 3.20: Membership functions

Chapter 4

Hardware Implementation and

Results

In hardware implementation, we first tested the circuit on breadboard and than we

developed a printed circuit board (PCB). We first made a single phase full bridge inverter
and we synchronized it with grid and after that we made a three phase inverter.

4.1 Components
Following components are used in the implementation of project:

4.1.1 STM32F401 Microcontroller

We used STM32F401 Microcontroller. It is a 32 bit microcontroller. It is a part of 1st
series of STM32F4 Discovery Board Series. It has 7 General purpose Inputs/Outputs
ports, built-in timers,a watchdog timer, analog to digital converters (ADC’s), digital to
analog converters (DAC’s). It has builtin oscillator and a clock.Figure 4.1 is showing
STM32F4 discovery board

Figure 4.1: STM32F4 Discovery Board

37
Chapter 4. Hardware Implementation and Results 38

4.1.2 Optocoupler tlp250

Optocoupler is used to provide isolation between low current circuit and high current
circuit to protect sensitive electronic equipments like microcontroller. It consists of light
emitting diode and a transistor. When a voltage across LED is sufficient it conducts
and the light causes the transistor to go into active region and voltage applied across
transistor as Vcc appears as output.We used TLP250 optocoupler. It has 8 terminals.
One anode and cathode terminal and two terminals unconnected. Figure 4.2 is showing
tlp250

Figure 4.2: Optocoupler TLP250

4.1.3 Gate Driver IC

Gate driver circuit is used to drive MOSFETS’s and IGBT’s. We know that the power
MOSFET’s and IGBT’s require gate to source voltage in the range of 10-15 Volts in
order to work properly. Gate driver circuit serves this purpose. There are several gate
driver IC’s available like IR2110,IR2103 etc. We used IR2103 IC. It has two inputs
and two outputs to drive high side switches and low side switches respectively. It also
has bootstrap mechanism.Bootstrap mechanism is used to provide virtual grounding. It
also has built-in dead time to avoid short circuiting of the supply. Figure 4.3 is showing
IR2103 IC

Figure 4.3: IR2103

Chapter 4. Hardware Implementation and Results 39

4.1.4 IGBT IC
Choice of electronic switch used for the implementation of inverter depends on many
factors i.e frequency and voltage and current rating. For high frequency applications
MOSFETS’s are prefered. For low frequency and high voltage and high current appli-
cations IGBT’s are prefered.We used IGBT as a swicth. IGBT has three terminals like
MOSFET. Gate, Collector and Emitter IGBT’s have high voltage and current rating
and work at low frequency. We used IRG4BC30FD ic.Figure 4.4 is showing igbt symbol
and ic

Figure 4.4: Igbt symbol

Figure 4.5: Igbt IC

4.1.5 LM7812
LM7812 is a regulator IC. It is a part of LM78 regulator series.It regulates the voltage
to 12V. Figure 4.5 is showing LM7812 IC

Figure 4.6: LM7812

Chapter 4. Hardware Implementation and Results 40

4.2 Isolated Supplies Implementation

Gate driver IC IR2103 worked well for resistive load but when we connected transformer
as filter which is inductive load supply started to short circuited because dead time was
not enough. To overcome this problem we used isolated supplies. In isolated supplies
we used separate supply for high side switch and low side switch and we put the desired
dead time in programming. In this way ground is provided by the supply and there
is no need to use gate driver. Isolated supply is implemented as a regular DC power
supply. The block diagram showing the components in implementing an isolated supply
is shown in figure 4.6 A separate supply is needed for each high side switch while one

Figure 4.7: Isolated Supply block diagram

supply can be used for the low side switches because the ground of low side switches
is common.For three phase inverter, we made four isolated supplies,three supplies for
upper three high side switches and one for lower three low side switches.Figure 4.7 is
showing isolated supplies implemented

Figure 4.8: Isolated Supplies Implementation

Chapter 4. Hardware Implementation and Results 41

4.3 Filter Implementation

As we know that in STATCOM there are many harmonics produced in output voltage.
For filtering out the harmonics and to produce a sinusoidal we used a LC filter. For
inductor we used the transformer of ratio 1:1 which is also used as the coupling trans-
former. The transformer serves the filtering purpose because in SPWM the higher order
harmonics have magnitudes very small as compared to the fundamental so coupling
transformer served this purpose and there was no need to design the separate inductor
for filter. For capacitor we use a 4uF capacitor. Figure 4.8 is showing the filter imple-
mented

Figure 4.9: Filter implementation

The output produced by filter displayed on oscilloscope is shown in figure 4.9

Figure 4.10: Filter Output

Chapter 4. Hardware Implementation and Results 42

4.4 Inverter Implementation

4.5 Square wave inverter
We implemented a single phase square wave inverter. The pulses were generated from
microcontroller.The pulses generated are 180 degree out of phase. Figure 4.10 is showing
square wave output on oscilloscope

Figure 4.11: Square wave inverter Output

4.6 SPWM inverter

Than we implemented a single phase SPWM inverter. The pulses were generated using
microcontroller. The carrier frequency was taken to be 10k Hz.Figure 4.11 is showing
pulses displayed on oscilloscope

Figure 4.12: SPWM pulses

Chapter 4. Hardware Implementation and Results 43

The output of a SPWM inverter is shown in figure 4.12

Figure 4.13: SPWM output

4.7 Final PCB

The final PCB developed is shown in figure 4.13

Figure 4.14: Final PCB

Chapter 5

Conclusion

The project is about the improvement of power system stability by the application of
STATCOM.The control of STATCOM is by the use of Neuro Fuzzy Control. This control
is fast and non-linear control. Due to the fast response time this control can be used to
handle transients.A fuzzy PID control is implemented on simulation level on MATLAB
SIMULINK. A single phase and three phase SPWM inverter is made for this purpose.LC
low pass Filter is developed to filter out the high frequency components.The amplitude
and phase and frequency can be controlled. The P and Q can be controlled in open
loop.

5.1 Applications
The applications of a STATCOM includes:

• Reactive power compensation.

• F or voltage regulation.

• T o enhance Transient Stability

5.2 Future Aspects

The future aspects of a project includes:

• Response speed of control can be made better

• N euro Fuzzy Control can be implemented at hardware and results can be observed

• S TATCOM can be installed at power stations to enhance stability

44
References

[1] Artificial Intelligence-Fuzzy Logic Systems. www.tutorialspoint.com/

Artificial_intelligence_fuzzy_logic_systems.htm/.

[2] Gaussian curve membership function. www.mathworks.com/help/Fuzzy/gaussmf.

html/.

[3] Neural Networks and Deep Learning. neuralnetworksanddeeplearning.com/.

[4] Fuzzy rule-based systems. www.data-machine.com/fuzzy1.htm/.

[5] Trapezoidal-shaped membership function. www.mathworks.com/help/Fuzzy/

trapmf.html/.

[6] Triangular-shaped membership function. www.mathworks.com/help/Fuzzy/

trimf.html/.

[7] Unipolar and Bipolar SPWM Voltage Modulation Type inverter for Im-
proved Switching Frequencies. www.academia.edu/8951265/Unipolar_
and_Bipolar_SPWM_Voltage_Modulation_Type_inverter_for_Improved_
Switching_Frequencies/.

[8] Sine Pulse Width Modulation. www.efxkits.us/

Sine-pulse-width-modulation-spwm-working/, January 24, 2017.

[9] Power flow Study. www.wikipedia.org/Power-flow_study, March 30, 2018.

[10] M.H.Rahisd. Power Electronics Devices, Circuits and Applications. Pearson, 2011.

[11] Tom. M Mitchell. Machine Learning. The Mc-Graw Hill Companies.

[12] Hadi Saadat. Power System Analysis. The Mc-Graw Hill Companies, 1999.

45