37

CHAPTER 3

DEVELOPMENT OF AN EXPERT SYSTEM TO IDENTIFY

AND ESTIMATE HARMONICS PRODUCED BY POWER

ELECTRONIC CONVERTERS

3.1 INTRODUCTION

The existence of harmonics is one of the significant means of

deteriorating the quality of an electric power supply. In order to fully

understand the problems caused by harmonics, sources of electric power

harmonics needs to be identified and this normally requires power quality

monitoring and analysis. However, monitoring usually generates vast amount

of data which makes manual inspection of the disturbance waveforms a

tedious and time-consuming task. Therefore, it is desirable that the data

collection process is automated such that monitoring equipment not only

captures harmonic waveforms but also performs the task of recognizing the

various types of nonlinear loads present in a system. With such an automated

system, faster and more efficient analysis of the databases can be achieved.

The issue of automatic analysis of various power quality disturbances has

been addressed in the form of the development of expert systems for power

quality event identification or the classification of disturbances such as

transients and voltage sags (Kazibwel and Sendaula 1992). The expert system

for analyzing power system harmonics focuses on symptoms associated with

harmonics related problems and providing possible solutions (Schlabbach

1994, Shipp et al 1995).

38

In this chapter an improved model is proposed to identify the

harmonic distortion produced by the different types of power electronic

converters. The rules are derived to identify harmonics produced by single

phase, three phase fully controlled AC-DC converters, and single phase

voltage source inverters. The proposed model provides automatic and

intelligent identification of harmonic sources originating from power

electronic converters loads by classifying the different types of nonlinear

loads that contribute harmonics to the supply system and then quantifying the

harmonic distortion levels of each type of load. This expert system provides

all the necessary input side and output side parameters that are not available

in the scheme proposed by Azah Mohamed et al (2006). In the development

of the expert system, an important consideration is given to the application of

advance signal processing techniques such as the fractal and fast Fourier

transform (FFT) analyses for characterizing the harmonic signatures of

various types of nonlinear loads. Unique features extracted from the analyses

are in the form of fractal numbers, harmonic distortion levels including

individual harmonic components and total harmonic distortion, crest factor

and power factor. The accuracy of the expert system is illustrated by

comparison with hardware.

3.2 TOOLS FOR ANALYSING HARMONICS

For most conventional analyses, the power system is essentially

modeled as a linear system with passive elements excited by constant

magnitude and constant frequency sinusoidal voltage sources. However, with

the widespread proliferation of power electronics loads nowadays, significant

amounts of harmonic currents are being injected into power systems.

Harmonic currents not only disturb loads that are sensitive to waveform

distortion, but also cause many undesirable effects on power system elements.

As a result, harmonic studies are becoming a growing concern. Harmonics are

39

usually defined as periodic steady state distortions of voltage and/or current

waveforms in power systems. In the harmonic polluted environment, the

theory regarding harmonic quantities needs to be defined to distinguish from

those quantities defined for the fundamental frequency.

3.2.1 Fourier Series and Analysis

The French physicist and mathematician, Joseph Fourier, in his

article ‘Analytic Theory of Heat’ which was published in 1882, first

introduced the theory of the Fourier series. The theory involves expansions of

arbitrary functions in certain types of trigonometric series. It proves that any

periodic function in an interval of time could be represented by the sum of a

fundamental and a series of higher orders of harmonic components at

frequencies, which are integral multiples of the fundamental component. The

series establishes a relationship between the function in time and frequency

domains. A distorted periodic wave can be decomposed into a fundamental

wave and a set of harmonics. This decomposition process is called Fourier

analysis. With this technique the effects of nonlinear elements in power

systems can be analyzed systematically (Chang 1998). A periodic function

can be defined as any function for which

( ) ( )f t f t T (3.1)

for all t. The smallest constant T that satisfies the equation (3.1) is called the

period of the function. By iteration of equation (3.1),

( ) ( ), 0, 1, 2...,f t f t nT n (3.2)

40

Let a function f(t) be periodic with period T, then this function can

be represented by the trigonometric series

) sin cos (

2 1 ) (

1 0 t n b t n a a t f n n n

(3.3)

where )/2( T . A series such as (3.3) is called trigonometric Fourier

series. It can be rewritten as

01

( ) sin( )n nn

f t c c n t

(3.4)

where 2

00

ac , 22nnn bac , and

n

nn b

a1tan

From the equation (3.4), it can be observed that the Fourier series

expression of a periodic function represents a periodic function as a sum of

sinusoidal components with different frequencies. The component of no is

called the nth harmonic of the periodic function. C0 is the magnitude of the dc

component. The component with n=1 is called the fundamental component.

Cn and n are known as the nth order harmonic magnitude and phase angle,

respectively. The magnitude and phase angle of each harmonic determine the

resulting wave shape of f(t). Equation (3.3) also can be represented by its

complex form as

0( ) jn tn

n

f t c e

(3.5)

where n=0, ±1, ±2, ...

0

2

2

1 ( )T

jn tn

T

c f t e dtT

(3.6)

41

3.2.2 Fourier Transform (FT)

The Fourier transform of a function f(t) is defined as

( ) ( ) j tF f t e dt

(3.7)

and f(t) is called the inverse Fourier transform of F(), which is defined as

1( ) ( )2

j tf t F e d

(3.8)

Equations (3.7) and (3.8) are often called the Fourier transform

pair, and they are used to map any function in the interval of (-∞, ∞) in time

or frequency domain into a continuous function in the inverse domain. The

key property of the Fourier transform is its ability to examine a function or

waveform from the perspective of both the time and frequency domains.

A given function can have two equivalent modes of representations: one is in

the time domain and is called f(t), and the other is in the frequency domain

and is called F(). Equation (3.7) transforms the time function into a

frequency spectrum, and (3.8) synthesizes the frequency spectrum to regain

the time function.

3.2.3 Fast Fourier Transform (FFT) Analysis of Harmonic

Waveform

To characterize the waveforms generated by nonlinear loads, signal

processing of the harmonic load current waveforms need to be performed.

The FFT is one of the traditional techniques widely applied in harmonic

analysis because of its effectiveness in computing harmonic distortion

components of a current waveform (Azah Mohamed et al 2006). The total

42

harmonic distortion (THD), crest factor, distortion or form factor and power

factor which can be computed using FFT analysis are useful factors for

quantifying harmonic distortions. Mathematically, the total harmonic

distortion (THD) in terms of current is given by,

12 2

1

( )nITHD

I (3.9)

where, I1 - Fundamental component of the current

In - Current harmonic component.

The THD normally varies from a few percent to 100% for current

and less than 5% for voltage. In an electrical distribution system, a voltage

THD of 5% or less is considered acceptable. Two other measures of distortion

are the crest and the form factors. The crest factor is the ratio of the peak of a

waveform to its RMS value:

_ ( ) peak

rms

ICrest Factor CF

I (3.10)

where Ipeak - Peak value of the harmonic waveform

Irms - RMS value of the harmonic waveform

The form or distortion factor on the other hand, is the ratio of the

RMS value of a waveform to the RMS of the waveform’s fundamental value

which is given by:

1

_ ( ) rmsIForm Factor FFI

(3.11)

where rmsI = RMS value of a harmonic waveform,

1I = RMS of the waveform’s fundamental value

43

3.3 FRACTAL ANALYSIS OF HARMONIC WAVEFORM

In recent years, fractal techniques have attracted increased attention

as a tool for signal processing. The method has been suggested in many

applications as an alternative for analyzing time-varying signals where other

techniques have not achieved the desired results (Crownover 1995). Fractal

theory is an extension of the classical geometry that can be used to make

precise models of physical structures (Barnsley 1993). Fractal analysis can be

useful for representing the nonlinear load harmonic current waveforms in

terms of fractal features through the computation of fractal numbers. Different

concepts of fractal geometry are available for quantization of chaotic behavior

of nonlinear systems. The fractal technique provides both time and spectral

information of the nonlinear load harmonic patterns. The analysis results

shows that the various harmonic current waveforms can be easily identified

from the characteristics of the fractal features. This investigation proves that

the fractal technique is a useful tool for identifying harmonic current

waveforms and forms a basis towards the development of the harmonic load

recognition system. Apart from increased speed, accuracy or efficiency, the

features of the new method as compared to the other existing techniques are:

1. The approach is able to present both fractals and time

information simultaneously.

2. It is very effective in monitoring time-varying signal dynamics

making it easier to locate the area of interest for proper

investigation.

3. It can be widely applied to localize various disturbance

waveforms for better visualization of the signal characteristics

4. It is a suitable candidate for pattern recognition of the various

load harmonic waveforms.

44

The algorithm has been well tested on data sets ranging from a few

hundreds to tens of thousands of data points. Fractals are often studied with

the aid of fractal dimension. Fractal dimension is a measure of the

“convoluted ness" or “degree of meandering" of the fractal set. The following

method presents a fast and simple way of estimating the fractal dimension of

a waveform (in the plane) (Pe 2003). The term waveform refers to the shape

of a wave, usually drawn as instantaneous values of a periodic quantity versus

time. In practice, a waveform is represented by a finite sample of N points,

,.....1:),( NiyxW ii . To estimate the fractal dimension D of the wave-

form ,.....1:),( NiyxW ii , W is first normalized using the equations

(3.12) and (3.13).

*

max

ii

xxx

(3.12)

* max

max min

ii

y yyy y

(3.13)

Here maxx , maxy denote the maximum values of the numbers xi,

iy respectively, and miny denotes the minimum of the numbers iy . The fractal

dimension D can be estimated using the formula,

*

ln( )1ln(2 )

LDN

(3.14)

where L is the length of the normalized wave-form W, and N* = N - 1. L is

easily calculated by repeated application of the distance formula on ),( **ii yx

and ),( *1

*1 ii yx , 1N,.....2,1ifor , that is

*

* * 2 * * 21 1

1( ) ( )

N

i i i ii

L x x y y

(3.15)

45

The process is quantified in terms of fractal numbers which is

derived using the equation (3.16).

2

1

2

1

N l

j k jj

N l

j l jj

y yF

y y

(3.16)

where, F is the fractal number of thm data set mS containing data points, k and

l are small integer sampling steps size which specifies the time interval

between close data points, such that l is greater than k. For the effectiveness

of the equation, different norms can be selected, however for this particular

work, the Euclidean norm is used due to its simplicity. The fractal number

computation turned out to be more efficient in this case if the ratio l to k is not

an integer. The data points have to be properly classified such that data points

belonging to k-step sampled subsets do not fall into the l-step subsets (Umeh

et al 2004).

3.4 DESIGN OF THE EXPERT SYSTEM FOR

IDENTIFICATION OF HARMONICS

For harmonic measurements of the nonlinear loads, harmonic

waveforms were obtained from single phase full converter (AC-DC), three

phase fully controlled converter (AC-DC), and single phase inverter

(DC-AC). The converters mentioned above were simulated using SIMULINK

in MATLAB. Harmonic measurements were performed to obtain data for

identification of harmonic loads. Hardware circuits are fabricated to verify the

performance of the algorithm developed. The proposed expert system is

designed in such a manner that the user can recognize typical harmonic

46

features of nonlinear loads and its structure is as shown in Figure 3.1. The

system consists of attribute extraction module and expert system development

module. The procedure involved in the development of the expert system is:

1. to determine the type of nonlinear loads.

2. to extract the current signatures or empirical features of the

nonlinear loads using FFT and fractal analyses.

3. to transfer the features in terms of knowledge base into a set of

computer programs.

4. to validate the developed expert model.

Figure 3.1 Structure of the expert system for identification of

harmonics in nonlinear loads

In the implementation of the expert system, considerations are

given to the preparation of data used in the formation of rules and in the

development of the graphical user interface (GUI). The attribute extraction

47

module is meant for pre-processing of the harmonic current waveforms using

FFT and Fractal analysis is shown in Figure 3.2. In the pre-processing stage,

unique characteristic features of harmonic current waveforms are extracted by

means of harmonic decomposition and fractal number computations. The

parameters namely, RMS current, total harmonic distortion, crest factor, form

or distortion factor, percentage of individual harmonic distortions, minimum,

maximum and the average value of the fractal number computations are

computed. The extracted features are then used as inputs into the expert

system.

Figure 3.2 Attribute extraction module

The knowledge base is poised of a set of rules in the form of

expertise knowledge from an elaborate analysis of the extracted features. The

rules are in the form of IF-THEN-ELSE description as suggested by Ibrahim

et al (1999). The ‘IF’ statement developed expert system comprises the

characteristic feature of various single phase nonlinear loads. The THEN part

Fractal Number

Computation

Individual Harmonic distortion calculation

Total Harmonic Distortion

Calculation

Crest factor Calculation

Form factor Calculation

Input to Expert System

Fractal Analysis

Fourier analysis (FFT)

Power Electronic Converter

Harmonic waveform generation

48

consists of a particular load or load composition matching the IF statement,

while the ELSE part acts as activation to the next rule. To formulate rules,

strict threshold values are set for each converter type or each combination of

converters. This is achieved by a comprehensive analysis and comparison of

derived features from each type of nonlinear load. The experimental data for

training and evaluation of the system are acquired by harmonic currents

generated by using the single phase and three phase fully controlled

converters at various firing angle, and single phase voltage source inverter.

Examples of the derived rules for single phase fully controlled

converter are described as follows:

Rule: 1 Firing Angle = 36o

IF

3rd Harm > 5th Harm &

7th Harm < 9th Harm &

13th Harm > 15th Harm &

19th Harm < 21st Harm &

22nd Harm < 20th Harm &

THD >10.95 &

THD < 16.5 &

Crest Factor > 1.43 &

Crest Factor < 1.45 &

Form Factor > 0.7 &

Form Factor <.72 &

Fractal Number Mean > 50 &

Fractal Number Mean < 55

THEN

Single phase full converter RL load at α=36o

49

Rule: 2 α = 60o IF

3rd Harm > 5th Harm &

7th Harm < 9th Harm &

13th Harm > 15th Harm &

19th Harm < 21st Harm &

22nd Harm < 20th Harm &

THD > 31.5 &

THD < 31.7 &

Crest Factor > 1.6 &

Crest Factor < 1.62 &

Form Factor > 0.73 &

Form Factor < 0.75 &

Fractal Number Mean > 45 &

Fractal Number Mean < 50 &

THEN

Single phase full converter RL load at α=60o

3.5 CREATING GRAPHICAL USER INTERFACE

In the MATLAB, Graphical User Interface Development

Environment (GUIDE), provides a set of tools for creating Graphical User

Interfaces (GUIs). These tools greatly simplify the process of designing and

building of GUIs. GUI is developed using the GUIDE layout editor, easily by

clicking and dragging the components such as panels, buttons, text fields,

sliders, menus, and so on into the layout area. To program the GUI, GUIDE

automatically generates an M-file that controls the operation of GUI

(www.mathworks.com). The M-file initializes the GUI and contains a

framework for all the GUI callback the commands that are executed when a

user clicks a GUI component. Using the M-file editor, one can add code to the

callbacks to perform the functions want to do.

50

3.5.1 Graphical User Interface Development Environment (GUIDE)

To start GUIDE, enter guide at the MATLAB command prompt.

This displays the GUIDE Quick Start dialog, as shown in Figure 3.3. From

the Quick Start dialog, one can,

1. Create a new GUI from one of the GUIDE templates - Prebuilt

GUIs that can be modifying for own purposes.

2. Open an existing GUI. It can be selected one of these options,

clicking OK opens the GUI in the Layout Editor

Figure 3.3 Graphical User Interface Development Environments

(GUIDE) quick start

3.5.2 Layout Editor

When open a GUI in GUIDE, it is displayed in the Layout Editor,

which is the control panel for all of the GUIDE tools. The following figure

shows the Layout Editor with a blank GUI template. Layout the GUI by

dragging components, such as push buttons, pop-up menus and axes from the

51

component palette at the left side of the Layout Editor into the layout area.

For example, drag a push button into the layout area; it appears in the layout

area Figure 3.4 shows the layout editor -control panel.

Figure 3.4 Layout editor-control panel

3.5.3 Running a Graphical User Interface (GUI)

To run a GUI, select Run from the tools menu, or click the run

button on the toolbar. This displays the functioning GUI outside the Layout

Editor ( ). GUIDE stores a GUI in two files, which are generated the first

time saved or run the GUI: A figure file, with extension.fig, which contains a

complete description of the GUI layout and the components of the GUI: push

buttons, menus, axes, and so on. The M-file with an extension.m, which

contains the code and callbacks that are controls the Graphical User Interface

(GUI). When the GUI is laid out in the Layout Editor, work is stored in the

FIG-file. When the GUI is programmed, work is stored in the M-file.

52

After laying out GUI, program the GUI M-file using the M-file

editor. GUIDE automatically generates this file from the first time save or run

the GUI. The GUI M-file

1. Initializes the GUI.

2. Contains code to perform tasks before the GUI appears on the

screen, such as creating data or graphics.

3. Contains the callback functions that are executed each time a

user clicks a GUI component.

Initially, each callback contains just a function definition line. Then

use the M-file editor to add code that makes the component function the way

want it to. To open the M-file, click the M-file Editor icon on the Layout

Editor Toolbar.

3.5.4 Development of Graphical User Interface

A graphical user interface (GUI) program has been developed in

MATLAB to allow communication between a user and the expert system.

Initially, an input current waveform or voltage waveform is generated using

Simulink model. The developed Graphical User Interface model is shown in

Figure 3.5. The interface illustrates the entire identification process

dynamically in which at a click of the “RUN” button, the load composition

exhibits the characteristic features of the input current waveform together

with a display of harmonic attributes of waveform. The displayed attributes

include the individual harmonic contents from third harmonic up to the

twenty-third harmonic, the current THD, crest factor, form factor, power

factor, RMS values of source voltage and output voltage and RMS values of

source current and output current. By selecting the “FFT plot” button, a FFT

analysis of the current waveform is performed; crest factor, form factor,

53

power factor and FFT plot are displayed. Likewise, by selecting the “Fractal”

button fractal plot and fractal numbers are displayed.

Figure 3.5 The developed expert system output model

3.6 TESTING OF THE EXPERT SYSTEM, RESULTS AND

DISCUSSION

The developed expert system model identifies and estimates the

harmonic levels for different types of nonlinear loads from its distorted supply

current and load voltage. The expert system is tested for single phase, three

phase fully controlled converters and single phase voltage source inverter.

The simulation results are compared with the developed hardware

experimental results for the following cases.

3.6.1 Single Phase Fully Controlled Converter Feeding RL Load

The single phase fully controlled converter feeding RL load with

the following specifications is considered for testing: Supply Voltage (Vs)

=230V, Firing Angle =36o, Load Resistance(R) =153 Ω, Load Inductance

54

(L)=0.328H. The simulink model for the single phase fully controlled

converter is shown in Figure 3.6. The results of the developed Graphical User

Interface are shown in Figure 3.7. The hardware test setup is shown in

Figure 3.8. The supply current waveform and harmonic spectrum of the

hardware is shown in Figure 3.9 and Figure 3.10 respectively. Comparative

results of simulation and hardware are shown in Table 3.1. Table 3.2

represents the comparison of important parameters for different firing angles.

Figure 3.6 Simulink model for the single phase fully controlled converter

Figure 3.7 Results of the developed graphical user interface for single

phase fully controlled converter

55

Figure 3.8 Hardware test setup of single phase fully controlled converter

Figure 3.9 Supply current waveform of single phase fully controlled converter

Figure 3.10 Harmonic spectrum of supply current of single phase fully controlled converter

56

Table 3.1 Comparative results of harmonic levels in single phase fully

controlled converter for a firing angle α=36o

Harmonic Order Power GUI

in MATLAB Real time

PQA* Expert System

(GUI)# THD % 26.00 23.4 24.61

Fundamental 100.00 100.00 100.00 3 25.71 21.20 16.58 5 3.30 8.00 2.75 7 1.21 4.20 1.35 9 0.69 2.80 0.52

11 0.52 1.40 0.68 13 0.38 1.00 0.15 15 0.52 0.70 0.30 17 0.14 0.30 0.17 19 0.13 0.70 0.18 21 0.15 0.80 0.15 23 0.13 0.80 0.21

*Power Quality Analyzer Readings

#-Develop Graphical User Interface Results

Table 3.2 Comparison of performance parameters in single phase

fully controlled converter

Sl. No.

Performance Parameter

Firing Angle (in degrees) 18o 36o 54o 72o

Expert System

Hardware Results

Expert System

Hardware Results

Expert System

Hardware Results

Expert System

Hardware Results

1. THD 11.95 10.80 24.61 23.40 32.63 31.10 40.23 39.10 2. Crest Factor 1.26 1.64 1.26 1.64 1.38 1.78 1.62 2.01 3. Power Factor 0.95 0.94 0.93 0.94 0.86 0.84 0.78 0.76

The total harmonic distortion of the single phase fully controlled

converter for the expert system is 24.61% and for the developed hardware is

23.4% according to Table 3.1. The important performance parameters of the

57

converter are compared in Table 3.2 and the readings are very close to the

hardware results. Hence the results obtained in the simulation and hardware is

found to be satisfactory.

3.6.2 Three Phase Fully Controlled Converter Feeding Resistive

Load

The three phase fully controlled converter feeding resistive load

with the following specifications is considered for testing: SupplyVoltage

(Vs) =400 V, Firing Angle =30o, Load Resistance(R) = 1000 Ω. The simulink

model for the three phase fully controlled converter is shown in Figure 3.11.

The results of the developed Graphical User Interface are shown in

Figure 3.12. The hardware test setup is shown in Figure 3.13. The supply

current waveform of the hardware is shown in Figure 3.14. Comparative

results of simulation and hardware are shown in Table 3.3. Table 3.4

represents the comparison of important parameters for different firing angles.

Figure 3.11 Simulink model for the three phase fully controlled converter

58

Figure 3.12 Results of the developed graphical user interface for three

phase fully controlled converter

Figure 3.13 Hardware test setup of three phase fully controlled converter

59

Figure 3.14 Supply current waveform of the three fully controlled

converter (for firing angle α=30o)

Table 3.3 Comparative results of harmonic levels in three phase fully controlled converter for a firing angle α=30o

Harmonic Order Power GUI in Matlab

Real time PQA*

Expert System (GUI)#

THD % 37.82 34.11 33.92 Fundamental 100.00 100.00 100.00

3 0.01 0.14 0.01 5 28.28 26.32 27.67 7 14.70 12.92 12.30 9 0.01 0.14 0.01

11 12.67 10.96 10.45 13 9.29 8.21 7.89 15 0.01 0.14 0.01 17 8.65 6.92 7.30 19 6.94 6.14 5.86 21 0.01 0.14 0.01 23 6.64 5.53 5.25

* - Power Quality Analyzer Readings

#-Develop Graphical User Interface Results

60

Table 3.4 Comparison of performance parameters of three phase fully

controlled converter

Sl. No.

Performance Parameter

Firing Angle (in degrees) 0o 30o 45o 60o

Expert System

Hardware Results

Expert System

Hardware Results

Expert System

Hardware Results

Expert System

Hardware Results

1. THD 11.95 10.80 24.61 23.40 32.63 31.10 40.23 39.10 2. Crest Factor 1.26 1.64 1.26 1.64 1.38 1.78 1.62 2.01 3. Power Factor 0.95 0.94 0.93 0.94 0.86 0.84 0.78 0.76

The total harmonic distortion of the three phase fully controlled

converter for the expert system is 33.92% and for the developed hardware is

34.11% according to Table 3.3. The important performance parameters of the

converter are compared in Table 3.4 and the readings are very close to the

hardware results. Hence the results obtained in the simulation and hardware is

found to be satisfactory.

3.6.3 Single Phase Voltage Source Inverter Feeding Resistive Load

The single phase voltage source inverter feeding resistive load with

the following specifications is considered for testing: Supply Voltage (Vs)

=230V DC, Load Resistance(R) = 400 Ω. The simulink model for the single

phase is voltage source inverter shown in Figure 3.15. The hardware test setup

is shown in Figure 3.16. The results of the developed Graphical User Interface

are shown in Figure 3.17. The output voltage wave form and harmonic

spectrum of the hardware is shown in Figures 3.18 and 3.19 respectively.

Comparative results of simulation and hardware are shown in Table 3.5.

61

Figure 3.15 Simulink model for the single phase voltage source inverter

Figure 3.16 Hardware test setup for the single phase voltage source

inverter

62

Figure 3.17 Results of the developed graphical user interface for single

phase voltage source inverter

Figure 3.18 Output voltage waveform of the single phase voltage source

inverter

63

Figure 3.19 Harmonic spectrum of the output voltage waveform of the single phase voltage source inverter

Table 3.5 Comparative results of individual harmonic levels of single

phase voltage source inverter

Harmonic Order Power GUI in Matlab

Real time PQA*

Expert System (GUI)#

THD % 46.25 47.00 46.89 Fundamental 100.00 100.00 100.00

3 33.34 33.33 33.28 5 20.00 20.00 20.02 7 14.29 14.30 14.26 9 11.12 11.10 11.13

11 9.10 9.10 9.13 13 7.71 7.70 7.69 15 6.68 6.70 6.72 17 5.90 5.90 5.90 19 5.28 5.20 5.21 21 4.78 4.70 4.73 23 4.37 4.30 4.32

* - Power Quality Analyzer Readings

#-Develop Graphical User Interface Results

64

The total harmonic distortion of the single phase voltage source

inverter for the expert systems is 46.89% and for the developed hardware is

47.00% according to Table 3.5. Hence the results obtained in the simulation

and hardware is found to be satisfactory.

3.7 CONCLUSION

In this chapter, an improved model is proposed to identify the

harmonic distortion produced by the different types of power electronic

converters. The rules are derived to identify harmonics produced by single

phase, three phase fully controlled AC-DC converters, and single phase

voltage source inverters. The proposed model provides automatic and

intelligent identification of harmonic sources originating from power

electronic converters loads by classifying the different types of nonlinear

loads that contribute harmonics to the supply system and then quantifying the

harmonic distortion levels of each type of load. FFT and Fractal analyses of

the current waveforms of the nonlinear loads have been carried out to obtain

the harmonic characteristic features of the loads. Each type of load is

represented by its total harmonic distortion, crest factor, form factor, power

factor, individual harmonic distortions with respect to fundamental and the

average value of the fractal number. The process of identifying the various

types of nonlinear loads is automated by developing a rule-based expert

system. The system is verified and found to be accurate in identifying the

respective nonlinear load type from the observed hardware current/voltage

waveform. The developed expert system also incorporates a graphical user

interface program, which can be used in a user-friendly and interactive

environment.