CHAPTER 1

INTRODUCTION

1.0 Background

Most of electrical drive systems in industrial process are exposed to hash

environment which reduces lifespan of the inverter because of natural aging process.

Therefore, to maintain the performance of the inverter, reliability and preserving the health

is extremely crucial. This issue has attracted many researchers to improve faults detection

and diagnosis techniques. Statistical data shows that 63% of the inverter experience faults

within the first year of operation and 70% of the faults occurred are related to power switch

which categories into short circuit faults, open circuit faults and gate-misfiring faults (Ui-

Min et al., 2015).

Generally, fault diagnosis can be categorized into component based and system

based. For component based techniques, standard features of industrial power converter

such as gate signal monitoring and overcurrent scheme are used to protect solid state

switches. To ensure the successful of remedial action, these schemes which are integrated

with auxiliary analog circuitry for switch abnormalities detection must operate in less than

10sec to prevent the silicon chip from damages (Ubale et al., 2013). System based

1
techniques on the other hand focusing on waveform analysis while algorithmic techniques

to identify type of faults and its location (Gaeta et al., 2011).

1.2 Problem Statements

Three-phase voltage source inverters (VSI) play a key roles in motor drives,

flexible power transmissions, and recently grid-tied renewable energy generation units.

Therefore, the availability and reliability of inverters have become, increasingly, important.

Early stage of fault detection in inverters is needed for remedial actions prolonged normal

operation of inverters and, in some cases, prevents the system from unexpected shutdowns.

Normally, when a voltage source inverter (VSI) fault occurs, the associate system has to be

stopped for an un-programed maintenance schedule. This sudden interruption can be costly

and therefore the development of fault detection system is a must for improving reliability

and lowering the maintenance costs (Salehifar et al., 2015).

A number of techniques for detection and identification for the switches faults have

been proposed by various researcher (Yong et al., 2014). For example, Parks Vector used

neural network for diagnosis (Mendes et al., 2014). But this technique requires a very

complex pattern recognition algorithm fault detection by using voltage sensors has also

been presented. This method is fast but it requires additional sensors to operate (Ghorpade

and Jagtap, 2014). Besides, many techniques have been presented by various researchers

for identifying switches faults using signal processing (Yong et al., 2014). One of the

widely used tools is fast Fourier transform (FFT) analysis for detecting an open transistor.

However, the technique only presents spectral of the signal and does not provide temporal

information (Smirnov and Tushev, 2014). To solve the problem, short-time Fourier

transform (STFT) was proposed that it can provide temporal and spectral information of

2
the signal by representing it jointly in time-frequency representation (TFR) which are

very appropriate to analyse non-stationary signal. However, this technique cannot track

the non-stationary signal dynamics properly because of the limitation of fixed window

width. (Alavi et al., 2014).

Wavelet transform (WT) provides the time-scale analysis of the non-stationary

signal. It decomposes the signal to time scale representation rather than time frequency

representation, but it produces time-scale plots that are unsuitable for intuitive visual

analysis. Also the absolute referenced phase information cannot be deduced from the

wavelet transform (Bandyopadhyay et al., 2014). A superior method for time-frequency

analysis known as S-transform has been explored, extensively, in various researches as an

alternative to the STFT. These techniques allow high time resolution for high

frequency component and high frequency resolution for low frequency component

(Huda et al., 2013).

Several topology of fault tolerant strategy has been introduced as a remedial action

to assure a continuous and safety operation of power converter such as switch-redundant,

neutral leg, additional phase leg and cascaded inverter. But the technique required major

modification of the three-phase VSI (Nguyen et al., 2015). On the other hand, switches

faults classification and identification system for open and short circuit has not yet realized

until now (Mirafzal, 2014). Therefore, it is important to develop a comprehensive system

which offer detection, classifying type of fault and identifying the location of switches

fault.

1.3 Objective of the Research

The objectives of this research are:

3
 i. To analyze three-phase voltage source inverter (VSI) switches faults signals using

fast Fourier transform (FFT) and time-frequency distributions (TFDs) which are

STFT and S-transform. The best TFD for the switches faults analysis is selected in

terms of the accuracy, computation complexity and memory size used of time-

frequency representation (TFR).

ii. To detect, classify and identify the VSI switches faults using the best TFD.

Parameters of the switches faults are estimated from the TFR and then are used for

switches faults detection, classification and identification process.

iii. To develop three-phase voltage source inverter switches faults detection,

classification and identification system using the best TFD. The system also

exhibits user-friendly graphical user interface (GUI), displays the waveform and

power spectrum of the faults signal as well as can provide and log the parameters

obtained for the analysis.

1.4 Scope of Work

This research analyses two types of there-phase switches faults in VSI which are

open circuit and short circuit switches fault. The main focus of this research is to

concentrate on the signal analysis using linear time-frequency distribution (TFD) for

detection, classification and identification system. The TFDs selected are STFT and S-

transform are analysed and the best TFD is chosen in terms of accuracy, computation

complexity and memory size of the TFR. Based on the signal faults parameters obtained

from the analysis, the switches faults are detected, classified and identified. The analysis

are done by using MATLAB software whereas the system is developed by using visual

basic 2010.

According to the characteristics of the switches faults, maximum signal frequency

will be measured is 5000 Hz. Therefore, in order to fulfil the requirement of the Nyquist

theorem that states the sampling frequency to be at least double the size of the measured
4
frequency. The sampling frequency of this system is set at 10000 Hz. The three-phase VSI

is fed by 50V power supply and is connected to the load that consist of resistant and

inductance which are 4800 and 15.3H, respectively.

Moreover, determination of the type of classifier to be used is not the main focus for

this work. Therefore, rule-based classifier which is a subset of tree classifier is used due to

its simplicity. The research will not look into the root causes of these switches faults

problems, instead propose a solution for existing problem and predict future IGBT failure

in VSI. Furthermore, the research does not scrutinise any associated costs to industries.

1.5 Thesis Contributions

In this research, three-phase voltage source inverter switches faults analysis using

FFT and linear TFDs which are STFT and S-transform are presented. The switches faults

are modeled, generated and then represented in, jointly, TFR using the TFDs. Based on the

analysis of TFR, the selection of the best TFD is demonstrated in terms of accuracy,

computation complexity and memory size used. This criteria is necessity in order to

increase the reliability and performance for the system implemented on hardware. Many

researches have been done in this area but no research has presented switches faults

analysis using S-transform and STFT.

Besides, from the analysis, information of the switches faults signal are also

provided. Parameters of the faults signal that are estimated from TFR can present

signal information needed in time. The signal faults parameters are instantaneous of

RMS current, RMS fundamental current, average current, total waveform distortion,

total harmonic distortion and total inter-harmonic distortion. Then, from the signal

parameters, characteristics of the signal can be calculated. The signal characteristics are

such as fault duration, mean of RMS current, average current, total harmonic distortion,

5
total nonharmonic distortion and total waveform distortion. This information is useful to

give more understanding and diagnosis of the signal.

An automated three-phase VSI switches faults detection, classification and

identification system is also demonstrated. The system analyzes the faults signals using the

best TFD to detect, classify and identify the switches faults. In addition, user-friendly

graphical user interface (GUI) is designed to show the waveform and power spectrum of

the faults signal as well as able to display and log the parameters obtained from the

analysis. Therefore, the system can be used to monitor switches faults signal and help

utilities, regulators and customers to have a clear understanding of what is happening in the

VSI switches faults.

1.6 Thesis Outline

This thesis is organized into five chapters. The first chapter contains an overview

of this research. Some explanations about the literature and the recent development of VSI

switches faults analysis are covered in Chapter 2. Chapter 3 describes the theory of time-

frequency analysis techniques for identifying signal information and classifier technique

for signals classification. This is followed by the discussion of the performance analysis

and results for the TFD and the classification system using the best TFD are demonstrated

in Chapter 4. This thesis ends with the conclusion and suggestions for further research as

presented in Chapter 5.

6
 REFERENCES

ALAVI, M., DANWEI, W. & MING, L. 2014. Short-Circuit Fault Diagnosis for Three-
Phase Inverters Based on Voltage-Space Patterns. Industrial Electronics,
IEEE Transactions on, 61, 5558-5569.
BANDYOPADHYAY, I., DAS, S., PURKAIT, P., DAS, P. P. & KOLEY, C. Application of
wavelet transform to identify faulty IGBTs in 3-phase induction motor
drives. Control, Instrumentation, Energy and Communication (CIEC),
2014 International Conference on, Jan. 31 2014-Feb. 2 2014 2014. 296-
300.
GAETA, A., SCARCELLA, G., SCELBA, G., DE CARO, S. & TESTA, A. Inverter fault-
identification for VSI motor drives. Diagnostics for Electric Machines,
Power Electronics & Drives (SDEMPED), 2011 IEEE International
Symposium on, 2011. IEEE, 413-419.
GHORPADE, S. S. & JAGTAP, S. R. Detection and identification of open circuit
faults in VSI-fed induction motor drive. Advanced Communication
Control and Computing Technologies (ICACCCT), 2014 International
Conference on, 8-10 May 2014 2014. 306-308.
HUDA, N. H. T., ABDULLAH, A. R. & JOPRI, M. H. Power quality signals detection
using S-transform. Power Engineering and Optimization Conference
(PEOCO), 2013 IEEE 7th International, 3-4 June 2013 2013. 552-557.
MENDES, A. M. S., ABADI, M. B. & CRUZ, S. M. A. 2014. Fault diagnostic
algorithm for three-level neutral point clamped AC motor drives, based
on the average current Park's vector. Power Electronics, IET, 7, 1127-
1137.
MIRAFZAL, B. 2014. Survey of Fault-Tolerance Techniques for Three-Phase
Voltage Source Inverters. Industrial Electronics, IEEE Transactions on,
61, 5192-5202.
NGUYEN, N., MEINGUET, F., SEMAIL, E. & KESTELYNKESTELYN, X. 2015. Fault-
Tolerant Operation of an Open-End Winding Five-Phase PMSM Drive with
Short-Circuit Inverter Fault. Industrial Electronics, IEEE Transactions on,
PP, 1-1.
SALEHIFAR, M., SALEHI ARASHLOO, R., MORENO-EGUILAZ, M., SALA, V. &
ROMERAL, L. 2015. Observer-based open transistor fault diagnosis and
fault-tolerant control of five-phase permanent magnet motor drive for
application in electric vehicles. Power Electronics, IET, 8, 76-87.
SMIRNOV, N. & TUSHEV, S. Two Methods of Fault Detection in the PMSM Electric
Drive IGBT-Based Inverter. Manufacturing Engineering, Automatic
Control and Robotics: Proceedings of the 14th International Conference
on Robotics, Control and Manufacturing Technology (ROCOM'14), 2014.
1.
UBALE, M. R., DHUMALE, R. & LOKHANDE, S. 2013. OPEN SWITCH FAULT
DIAGNOSIS IN THREE PHASE INVERTER USING DIAGNOSTIC VARIABLE
METHOD.
UI-MIN, C., BLAABJERG, F. & KYO-BEUM, L. 2015. Study and Handling Methods
of Power IGBT Module Failures in Power Electronic Converter Systems.
Power Electronics, IEEE Transactions on, 30, 2517-2533.
YONG, Y., JINYANG, H., ZIYUAN, W. & DIANGUO, X. IGBT open circuit fault
diagnosis in VSI fed induction motor drives based on modified average
current method. Industrial Electronics and Applications (ICIEA), 2014
IEEE 9th Conference on, 9-11 June 2014 2014. 1334-1338.