See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/273635142

Open and Short Circuit Switches Fault Detection of Voltage Source Inverter
Using Spectrogram

Article · June 2014

DOI: 10.11142/jicems.2014.3.2.190

CITATIONS READS

4 513

3 authors:

Nur Sumayyah Ahmad Abdul Rahim Abdullah

16 PUBLICATIONS   120 CITATIONS   
Technical University of Malaysia Malacca
155 PUBLICATIONS   555 CITATIONS   
SEE PROFILE
SEE PROFILE

Norhazilina Bahari
Technical University of Malaysia Malacca
18 PUBLICATIONS   65 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

PROTON EXCHANGE MEMBRANE FUEL CELL (PEMFC) PARAMETERS ESTIMATION USING TIME-FREQUENCY ANALYSIS TECHNIQUE View project

Switch Faults Classification using Signal Processing View project

All content following this page was uploaded by Abdul Rahim Abdullah on 17 March 2015.

The user has requested enhancement of the downloaded file.

 http://dx.doi.org/10.11142/jicems.2014.3.2.190
190 Journal of International Conference on Electrical Machines and Systems Vol. 3, No. 2, pp. 190~199, 2014

Open and Short Circuit Switches Fault Detection of

Voltage Source Inverter Using Spectrogram

N. S. Ahmad *, A. R. Abdullah** and N. Bahari ***

Abstract –In the last years, fault problem in power electronics has been more and more
investigated both from theoretical and practical point of view. The fault problem can cause
equipment failure, data and economical losses. And the analyze system require to ensure fault
problem and also rectify failures. The current errors on these faults are applied for identified type
of faults. This paper presents technique to detection and identification faults in three-phase voltage
source inverter (VSI) by using time-frequency distribution (TFD). TFD capable represent time
frequency representation (TFR) in temporal and spectral information. Based on TFR, signal
parameters are calculated such as instantaneous average current, instantaneous root mean square
current, instantaneous fundamental root mean square current and, instantaneous total current
waveform distortion. From on results, the detection of VSI faults could be determined based on
characteristic of parameter estimation. And also concluded that the fault detection is capable of
identifying the type of inverter fault and can reduce cost maintenance.

Keywords: Voltage Source Inverter (VSI), Spectrogram, Time-Frequency Distribution (TFD),

Fast Fourier Transform (FFT), Open-circuit fault, Short-circuit fault.

1. Introduction devices in the drive module. Then a survey was done by

Thorsen and Dalva on the reliability of VSI for industrial
In general of industrial drives and its applications, drives. According to results of their survey, 50 percent of all
voltage source inverter (VSI) is commonly used in wide failures are in the control circuit, 7.7 percent in the cooling
variety, including a number of critical application in which fans, and 37.9 percent are in the power circuits. So from
very high levels of reliability are required because of loss at recent research and survey, faults in power electronics are
VSI operation can have serious consequences. Cause of large percentage of the VSI faults [4]. Due VSI faults is
VSI faults, the effect of motor drive will stop drive running higher issues, analyze of faults is important in industrial
or at least it affects the drive performances. For example, in drives, thus fault can identified early.
conveyor applications the drives motor has collapsed due to
VSI failure. As mentioned [1] , the VSI faults have several Different methods and analytical techniques were used
fault can appear as open-circuit faults, short-circuit faults, to detection and identification the VSI faults. As an
and intermittent gate-misfiring faults. However, a VSI fault example, B. Raison et al [2] used Fast Fourier Transform
has certain limitations such as difficulty in detecting type of (FFT) to analyzed current spectrum for detecting the
faults. characteristics of an open transistor [2, 5]. L. Zhang et al.
[6] used time domain for analyze fault modes with
The recent research shows that the fault occurred in the characterized the patterns of the current waveform While F.
power electronic are about 82.5 percent of faults occur in Charfi et al.[7] examined information about the fault
the VSI motor drives by Ricardo et al. [2, 3]. Also has been signatures for detection and classification fault using
estimated by Diqi Zhu et al. and Shenlin et al. [2, 3] wavelet transform. But each technique has limitations and
about 38 percent the fault in inverter driven industrial disadvantages for analyze the performance. For the FFT
induction motors are due to malfunctioning of power technique is does not appropriate for non-stationary signal
and not provide temporal information [8].
* Dept. of Electrical Engineering, Uni.Tek.Malaysia Melaka,Malaysia
([email protected]) Due to VSI faults signal in non-stationary pattern, this
** Dept. of Electrical Engineering, Uni.Tek.Malaysia Melaka,Malaysia
paper present time frequency distribution (TFD) which is
([email protected])
*** Dept. of Electrical Engineering, Uni.Tek.Malaysia Melaka, Malaysia.
spectrogram for analyze VSI faults. Spectrogram technique
([email protected]) capable to represent parameter estimated in time frequency
Received 21 April 2014; Accepted 23 April 2014
 N. S. Ahmad , A. R. Abdullah and N. Bahari 191

representation (TFR). Based on TFR, parameters of VSI Commonly, intersective method is simplest way to
current signal are estimated and used to identify open and generate a PWM signal, which requires a triangle waveform
short circuit switches fault based on current pattern then and a comparator. When the value of the references signal
characteristic of VSI are calculated. is more than the modulation waveform, the PWM signal is
in the high state, otherwise it is in low state. Besides that,
2. Voltage Source Inverter System the module of fault generation that used for give pulse at
switching gate signal means give disturbance at gate signal
Voltage source inverter (VSI) is an electronic device
which concern direct current (DC) to Alternating current The switches fault VSI for open circuit fault condition,
(AC). VSI has advantages like higher efficiencies, IGBT falls in the off state and remains in this situation
minimizing installation timing, elimination interconnect regardless of the gate voltage value. Open circuit fault
power cabling costs and reducing building floor space. generally do not cause system shutdown, but degrade its
Fig.1 (a) shows the diagram of general system for this performance. Fig. 2 show the model of circuit when VSI
research and Fig.1 (b) shows the modeling of three-phase switches open circuit fault.
VSI with the fault generation. In the model S11 to S32 are
Insulated Gate Bipolar Transistors (IGBTs) which can be on
or off pulse width modulation (PWM) is a common IM
technique for controlling the switches.

Inverter Motor
Load

Collect
Data Fig. 2. The model of Voltage Source Inverter when
Switches Open Circuit Fault

The short-circuit fault where this faults leads to

catastrophic failure of the inverter if the other transistor of
the same inverter leg is turned-on, this resulting in a direct
short-circuit as show Fig.3
(a)

IM IM

Fault Gate Fig. 3. The model of Voltage Source Inverter when

Generation Signal Switches Short Circuit Fault

Based on modeling of three-phase VSI with the fault

Pulse Width Modulation
generation, the transistors are S11 – S32 and the output stage
is considered as a general R load. To control this VSI it was
(b) used the pulse width modulation (PWM) as switching to
control the transistor. A switching function of the VSI is
Fig. 1. (a) Diagram general of system
defined as Fig. 4, where j represents the phase, and variable
(b) The model of Voltage Source Inverter (VSI)
p and n, respectively, represent the upper and lower
 192 Open and Short Circuit Switches Fault Detection of Voltage Source Inverter Using Spectrogram

component of the phase leg. The switching status represent Fourier Transform (STFT) and can be expressed as below.
in term of “1” when the switch is closed and “0’ when open. Where h(t ) is the input signal while w(t) is the window
The cause of fault in VSI for this paper depends on pulse function. This technique depends on window size since it
width modulation (PWM). Such the fault is open circuit results better time resolution and lower frequency
fault upper, open circuit fault lower, short circuit fault upper resolution by using smaller window size and vice versa[12,
and short circuit fault lower [9-11]. 13, 15, 16].
 2

 h( )w(  t )e
 j 2 f 
Sjp
1
S x (t , f )  d (1)


1 4. Parameter Estimation
Sjn
Parameters of the faults signal are estimated from the
Fig. 4. Gating signals of the inverter in an ideal case TFR to identify the characteristics of VSI faults.

The gate switching fault is one of cause open circuit and 4.1 Instantaneous Average Current
short circuit VSI. Based on Fig.5 and 6, gate pulse
switching is one of causes fault VSI. Open circuit fault 1
T

T 0
causes by gate pulse switching “0” then short circuit fault I  I rms (t )dt (2)
rms, ave
causes by gate pulse switching “1”.

Sjp
4.2 Instantaneous Root Mean Square [17]
The equation below from, Sx(t,f) is the distribution and
1
Sjn fmax is the maximum frequency of interest.

f max
Fig. 5. Gating signals of the inverter in an open-circuit
fault Irms  t    S x (t , f )df (3)
0
1
Sjp
0
4.3 Instantaneous RMS Fundamental Current
1
Sjn The instantaneous RMS fundamental current is as the RMS
0
current at power system frequency [12, 15, 18] can be
Fig. 6. Gating signals of the inverter in short-circuit
calculated as
fault

f hi
3. Result Time-Frequency Distribution
I1rms (t )  2  S x (t , f )df
flo (4)
Spectrogram is one of the time-frequency
representations (TFR) that represents a three-dimensional f f
f hi  f1  , f lo  f1 
of the signal energy with respect to time and frequency. 2 2
Time-frequency representations are used to analyze or
characterize signals whose energy distribution varies in
time and frequency. They map the one-dimensional time- 4.4 Instantaneous Total Waveform Distortion
domain signal into a two-dimensional function of time and
frequency [12-14]. A time-frequency representation The total waveform distortion consists of harmonic and
presents the variation of spectral energy over time [8-10]. In non-harmonic distortion. The TWD can be defined
this paper, time-frequency analysis technique which is
spectrogram is used to analyze the faults signal. Vrms (t )2  V1rms (t )2
Spectrogram is the squared magnitude of the Short Time
TWD(t)  (5)
V1rms (t )
 N. S. Ahmad , A. R. Abdullah and N. Bahari 193

4.4 Instantaneous Total Harmonic Distortion

THD is commonly used to measure harmonic 5.1 Open Circuit Switches Fault in VSI (Upper Fault)
distortion in signal. Harmonics have frequency that
The fault analysis done by using spectrogram technique,
are integer multiples of the waveform’s fundamental we can see observe from Fig.7 (a) for the fault in phase A
frequency [18]Such as, given a fundamental there is discontinuity in the signal and the severity of the
frequency 60 Hz, the harmonics component will be fault can be clearly understood by spectrogram analysis.
120 Hz, 180 Hz and 240 Hz. According on signal, fault current occur at 0.2 – 0.3 s where
current at phase A lower compare phase B and C. From the
TFR that display time, frequency and amplitude on the

H same graph and as can be seen two frequency detected. Fig.
h2
Vh,rms (t)2
THD(t)  (6) 7(b) the higher power of signal in the contour plot is
V1rms (t) represented in red color, while the lowest in blue color.
Based on result, TFR shows the 60 Hz for fundamental
frequency and frequency of VSI fault at DC component is 0
4.4 Instantaneous Total Non-Harmonic Hz
Total non-harmonic represents total non-harmonic Open-Circuit Faults of the Upper Switch Signal
3
distortion. Non-harmonic refer are signal not integers

Phase A
2
1
multiples of the fundamental frequency [1212, 14, 19]. For 0
Current, A

-1
example, the fundamental frequency given 60 Hz, the -2

Phase B
3
-3
interharmonic will be 70 Hz, 85 Hz. 20 0.1 0.2 0.3 0.4 0.5 0.6
1
0
-1
-2
-3
3
0 0.1 0.2 0.3 0.4 0.5 0.6

Phase C
2

V (t )rms   h 0 Vh,rms (t )
2 H 2 1
0

TnHD(t)  (7) -1
-2

V1rms (t ) -3
0 0.1 0.2 0.3 0.4 0.5 0.6

Time (s)
(a)

5. Results
Frequency, %

The results of analysis the VSI switches open and short

fault are represented from Fig. 7 to Fig. 14. That signal 60Hz
0 Hz
consists of three-phase which is phase A, phase B and phase
C. The red color represent phase A, green color represent
phase B and blue color represent phase C.
Time (s)
Based on the shown Table 1, that considers dc supply, (b)
carrier frequency (Hz), IGBT and fundamental frequency Fig.7. (a) Signal of Open-Circuit Fault Upper (b)
(Hz). From simulation, the result of open-circuit and short- Spectrogram open-circuit fault upper
circuit faults for upper and lower switches has been analyze
and calculate all characteristic of parameter From the TFR, the parameters are estimated for all
phase in per unit (pu). For Fig. 8(a), the instantaneous
average current phase A is greater while phase B and C,
Table 1. Voltage source inverter specifications
respectively is lower than zero between 0.2 and 0.3 s. From
Parameter, unit Value
the observed on Fig. 8(b), the RMS value of magnitude of
DC supply, V 50
the phase A momentary decrease at 0.2 – 0.3 s compare
Carrier frequency, Hz 5kHz
with phase B and C. The RMS fundamental value for phase
Snubber resistance 1 x 105
A is lower but phase B and C is higher show as Fig. 8(c).
Internal resistance 1 x 106 Lastly, the value of THD is 1 percent for phase A but phase
Fundamental frequency 60Hz
B and C is below than 0.5 percent shown in Fig.8(d)
respectively.
 194 Open and Short Circuit Switches Fault Detection of Voltage Source Inverter Using Spectrogram

Instantaneous Average Current, Iave(t) 5.2 Open Circuit Switches Fault in VSI (Lower Fault)
1
Fig. 9(a) shows the three-phase VSI for fault at lower
0.5 switches. We can see, observation from the open-circuit
Current, A
Amplitude

Phase B & C
fault lower signal where current phase A higher than phase
0
B and C. The current phase A increase starting from 0.2 –
-0.5 Phase A 0.3 s but phase B and C respectively.
Open-Circuit Faults of the Lower Switch Signal
-1 3

Phase A
0 0.1 0.2 0.3 0.4 0.5 0.6 2

Time (sec)
Time (s)
1
0
-1
(a) -2
-3
30 0.1 0.2 0.3 0.4 0.5 0.6

Current, A

Phase B
2
Instantaneous RMS Current, Irms(t) 1

1.1 0
-1
-2
1.05 Phase C -3
3

Phase C
Current, A

0 0.1 0.2 0.3 0.4 0.5 0.6

2
1
1 0
-1

Phase B -2

0.95 -3
0 0.1 0.2 0.3 0.4 0.5 0.6

Phase A Time (s)

0.9
(a)
0.85
0 0.1 0.2 0.3 0.4 0.5 0.6 Time Frequency Representation
Time (s)
(b)
Frequency, Hz

Instantaneous RMSFundamental Current, Irms(t)

1.1
Phase C
Current, A

1 0 Hz 60 Hz

0.9
Phase B

0.8 Phase A

0.7
0 0.1 0.2 0.3 0.4 0.5 0.6
Time (s)
Time (s) (b)
(c)
Instantaneous Total Waveform Distortion, TWD(t) Fig. 9. (a) Signal of Open-Circuit Fault Lower
1.5 (b) Spectrogram Open-Circuit Fault Lower
Percentage, %

Based on result, the contour of TFR detect 60 Hz for

1
Phase A fundamental frequency and fault current at DC component
(0 Hz) at phase A. Based on TFR, the instantaneous average
0.5
Phase B & C current, instantaneous RMS current, instantaneous RMS
fundamental are estimated as shown in Fig. 10(a) – (d). Fig.
0
0 0.1 0.2 0.3 0.4 0.5 0.6 10(a) show the average current for phase A is higher while
Time (s) phase B and C is lower. The current is higher at phase A
(d)
when RMS value result compares than phase B and C.
Fig.8. Parameter Estimation for open-circuit fault upper
(a) Instantaneous Average Current, Iave(t) Besides that, the RMS fundamental value is lower compare
(b) Instantaneous RMS Current, Irms(t) at phase A and higher current at phase B and C, respectively.
(c) Instantaneous RMS Fundamental Current, From all parameter, the percentage of THD for the open
Irms(t) circuit at lower fault is higher than 1 percent while both of
(d) Instantaneous Total Waveform Distortion, phase B and C is lower 1 percent
TWD(t)
 N. S. Ahmad , A. R. Abdullah and N. Bahari 195

Instantaneous Average Current, Iave(t) 5.3 Short Circuit Switches Fault in VSI (Upper Fault)
1
Fig. 11(a) shows signal short circuit switches fault
Phase A where three-phase current indicate the fault also occur at
Current, A

0.5
Amplitude

phase A where current signal is increase at 0.2 – 0.3 s. At

current signal at phase B and C is decrease when fault occur.
0 Based on Fig. 11(b), the TFR detect two frequency at faults
Phase B & C signal, where fundamental and fault frequency. From the
contour, the spectrogram is show three-axis where the
-0.5
0 0.1 0.2 0.3 0.4 0.5 0.6 detect fault at DC component is 0 Hz and 60 Hz for
Time (sec)
Time (s) fundamental frequency.
(a)
Instantaneous RMS Current, Irms(t)
1.15 Short-circuit Faults of the Upper Switch Signal
3

1.1 2
Current, A

Phase A
0
-1
1.05 Phase C -2
3
-3
0 0.1 0.2 0.3 0.4 0.5 0.6
Current, A
2
1 1

Phase B
0
0.95 Phase A Phase B -1
-2

0.9 -3
3
0 0.1 0.2 0.3 0.4 0.5 0.6
0 0.1 0.2 0.3 0.4 0.5 0.6 2
1

Phase C
Time (s) 0
-1
(b) -2
-3
Instantaneous RMS Fundamental Current, Irms(t) 0 0.1 0.2 0.3 0.4 0.5 0.6
1.1
Time (s)
Current, A

Phase C (a)
1
Time Frequency Representation
0.9 Phase B

0.8 Phase A
Frequency, Hz

0.7
0 0.1 0.2 0.3 0.4 0.5 0.6
Time (s)
0 Hz 60 Hz
(c)
Instantaneous Total Waveform Distortion, TWD(t)
3
Percentage, %

Phase A Time (s)

1 (b)

Phase B & C Fig.11. (a) Signal of Short-Circuit Fault Upper

0 (b) Spectrogram Short-Circuit Fault Upper
0 0.1 0.2 0.3 0.4 0.5 0.6
Time (s)
(d) Fig. 12 (a) shows the average current, phase A is higher
Fig.10. Parameter Estimation for open-circuit fault compare phase B and C at 0.2 – 0.3 s. During the RMS
upper current, the phase A is phase faults occur where the value of
(a) Instantaneous Average Current, Iave(t) current is higher compare with phase B and C as shows Fig.
(b) Instantaneous RMS Current, Irms(t) 12(b). According Fig.12 (c), the RMS fundamental for
(c) Instantaneous RMS Fundamental Current, phase A is momentary decrease but other phase is higher
Irms(t)
than phase fault occurring. While, value of TWD gives 1.4
(d) Instantaneous Total Waveform Distortion,
TWD(t) percent and phase B and C is below than 1 percent as show
Fig. 12 (d).
 196 Open and Short Circuit Switches Fault Detection of Voltage Source Inverter Using Spectrogram

Instantaneous Average Current, Iave(t) 5.4 Short Circuit Switches Fault in VSI (Lower Fault)
1
Fig. 13(a) shows the signal short-circuit faults of the
Phase A lower switch where switch at phase A fault at 0.2-0.3s. The
Current, A

0.5
Current(A)

current momentary decrease and for phase B and C have

0 certain distortion. According TFR shows harmonics and
Phase B Phase C interharmonics that occurred as represented Fig. 13(b). As
-0.5 previous results, faults occurred at dc component and at 60
0 0.1 0.2 0.3 0.4 0.5 0.6
Hz.
Time (s)
(a)
3
Short-circuit Faults of the Lower Switch Signal
Instantaneous RMS Current, Irms(t)

Phase A
2
1
1.6 0
-1

Current, A
-2
Amplitude(pu)
Current, A

1.4 Phase A 3
-3
0 0.1 0.2 0.3 0.4 0.5 0.6

Phase B
2
1
0
1.2 Phase B -1

Phase C -2
3
-3
0 0.1 0.2 0.3 0.4 0.5 0.6
2
1

Phase C
1
0
-1
-2
0.8 -3
0 0.1 0.2 0.3 0.4 0.5 0.6 0 0.1 0.2 0.3 0.4 0.5 0.6

Time (s) Time (s)

(b) (a)

Instantaneous RMS Fundamental Current, Irms(t) Time Frequency Representation

1.2

1.1
Current, A

Frequency, Hz

1
Phase B & C
0.9 Phase A
0 Hz 60 Hz
0.8
0 0.1 0.2 0.3 0.4 0.5 0.6
Time (s)
(c)
Instantaneous Total Waveform Distortion, TWD(t)
1.5
Time (s)
Percentages, %

(b)
Phase A
1
Fig.13. (a) Signal of Short-Circuit Fault Lower (b)
Phase B Phase C Spectrogram Short-Circuit Fault Lower
0.5
The parameter that estimated from TFR with the
0 instantaneous average current for phase A is lower than
0 0.1 0.2 0.3 0.4 0.5 0.6
Time (s) phase B and C as shows Fig. 14(a). But for the RMS
(d) current value, phase A is higher while phase B and C is
Fig.12. Parameter Estimation for short-circuit fault lower, respectively as Fig. 14(b). For the instantaneous
Upper RMS fundamental current on Fig. 14(c), phase A is lower
(a) Instantaneous Average Current, Iave(t) than phase B and C. Besides that, instantaneous waveform
(b) Instantaneous RMS Current, Irms(t)
distortion as shows in Fig. 10(d), where faults signal occur
(c) Instantaneous RMS Fundamental Current,
Irms(t) at DC component, there TWD gives 1.42 percent but phase
(d) Instantaneous Total Waveform Distortion, B and C is below then 1 percent.
TWD(t)
 N. S. Ahmad , A. R. Abdullah and N. Bahari 197

Instantaneous Average Current, Iave(t)

According all results that shows four patterns which is
1
open circuit fault upper, open circuit fault lower, short
Phase B
0.5 circuit fault upper and short circuit fault lower, the
Current, A

Phase C
Current(A)

parameter that estimate gives information for detect type of

0
fault. Fig. 15 shows the bar graph indicates that error of
-0.5 Phase A current detection for VSI faults can be known from the
-1 parameter estimation by applying TFD. The condition of
each fault can be identifying according to the result as a bar
0 0.1 0.2 0.3 0.4 0.5 0.6
Time (S) graph below. The details of each fault are shows in table 1
Time (s)
(a) and 2
Instantaneous RMS Current, Irms (t)
1.6 In order to find out the relation between parameter to
detect condition of the VSI faults have been conducted.
Current, A

1.4
Phase A Analysis results were made based on current parameter
1.2 estimation from spectrogram. Table 1 shows general of
Phase C
Phase B current error during fault detected. Next, Table 2 shows
1 overall observation of current error from parameter
estimation. From the observation of average current, that
0.8
0 0.1 0.2 0.3 0.4 0.5 0.6 found the value of IB is equal IC but less than 0.5 and other
Time (s) of current is IB equal with IC. Both of current is greater than
(b)
Instantaneous RMS Fundamental Current, Irms(t) 0.5 and less than zero. Next, the open-circuit lower can be
1.2 detect with IA below than 1, respectively is greater than
zero. For this condition IA is greater than 1 but IA less than
1.1 zero. higher than zero for open circuit fault upper. Besides
Current, A

that, the current error at phase A is greater than 1 when

1 short circuit fault upper detected. But other IA is less than 1.
Phase B & C When short-circuit lower, IA less than -1 and other IA is
0.9 Phase A greater than -1. On the other hand, the error of rms current
Phase B & C when open circuit fault either upper and lower, value IA is
0.8 equal IC but for short-circuit is IA different IC. During short
0 0.1 0.2 0.3 0.4 0.5 0.6
Time(s) circuit fault upper detected, value of IB is greater than IC
(c) and other fault IB is greater than IC. Then, open circuit fault
Instantaneous Total Waveform Distortion, can be detected during RMS fundamental current IA is
1.5 TWD(t) greater than IB and IC, where both of current are equal. But,
short circuit fault can be detected IA is greater than IC and
Percentage, %

e (s) greater then IB. Lastly, table 2 give summaries of

1 Phase A
characteristic for each fault cased on error current where
average current is important used to detect type of faults.
Phase B Phase C
0.5 The rules based on error current detection during fault are
used to detection the faults event instanteously are
summaries in Table 1.
0
0 0.1 0.2 0.3 0.4 0.5 0.6
Time (s) Table 1. General of current error during fault
(d)
Switch Type of fault Current Observation
Fig. 14. Parameter Estimation for short-circuit fault Upper Open Average IA < (IB =IC)
lower RMS (I
(IAB ==ICI)C>0
)> IB
(a) Instantaneous Average Current, Iave(t) Short Average IA> (IB =IC)
(b) Instantaneous RMS Current, Irms(t) RMS (I
(IAB ==ICI)C<0
)> IB
(c) Instantaneous RMS Fundamental Current,Irms(t) Lower Open Average IA> (IB =IC)
(d) Instantaneous Total Waveform Distortion RMS I(IAB>=IICB)><0
IC
Short Average IA < (IB =IC)
RMS (I
IAB>=IICC)>>0IB
 198 Open and Short Circuit Switches Fault Detection of Voltage Source Inverter Using Spectrogram

Open-circuit Upper Open-circuit Lower Short-circuit Upper Short-circuit Lower

Fig. 15. The bar graph for error of current detection

Table 2. Characteristic for each fault detection estimation. So, the detection faults can be early and
based on parameter estimation fault in VSI will be decrease.
Parameter Type of fault Observation
Average Open Upper ( IB = IC) < 0.5
Current Others ( IB = IC) > 0.5
( IB = IC) < 0 Acknowledgment
Open Lower IA< 1 but IA >0
Others IA> 1 & IA< 0
Short Upper IA >1
Faculty of Electrical Engineering, Universiti Teknikal
Others IA< 1 Malaysia Melaka
Short Lower IA < -1
Others IA > -1
RMS Open ( IA = IC)
References
Current Other ( IA = IC)
Short Upper ( IB > IC) [1] Trabelsi, M., et al. An improved diagnosis technique
Other ( IB = IC) for IGBTs open-circuit fault in PWM-VSI-fed
Short Lower ( IC >IB)
Other ( IC =IB) induction motor drive. in Industrial Electronics (ISIE),
RMS Open IA > ( IB = IC) 2011 IEEE International Symposium on. 2011.
Current
Short IA > IC > IB [2] Bin, L. and S. Sharma. A Literature Review of IGBT
Fault Diagnostic and Protection Methods for Power
Inverters. in Industry Applications Society Annual
Meeting, 2008. IAS '08. IEEE. 2008.
[3] Fuchs, F.W. Some diagnosis methods for voltage
6. Conclusion source inverters in variable speed drives with induction
machines - a survey. in Industrial Electronics Society,
The analysis and detection system of type VSI faults is 2003. IECON '03. The 29th Annual Conference of the
presented using TFD. From the parameter calculation, IEEE. 2003.
the TFR is estimated to identify fault signal parameter [4] Bojoi, R., M.G. Neacsu, and A. Tenconi. Analysis and
and then, from the parameters, fault signal survey of multi-phase power electronic converter
characteristic can be calculated. Next, the characteristic topologies for the more electric aircraft applications. in
for each fault detection can be identifying based on Power Electronics, Electrical Drives, Automation and
current error. The performance of the faults system is Motion (SPEEDAM), 2012 International Symposium
verified by the value current error which is average on. 2012.
[5] B. Raison, G.R.a.J.P.R., Towards a Global Monitoring
current and fundamental current. It shows the VSI
Scheme for Induction Motor Drives. Proc. Int. Power
faults can be identified depends on value parameter
Electron. Conf.2000, 2000: p. pp. 1183-1188.
 N. S. Ahmad , A. R. Abdullah and N. Bahari 199

[6] L. Zhang, I.B.A.a.L.N.H., A Knowledge-Based Nur Sumayyah Bt Ahmad was born in

System forOn-line Fault Diagnosis of Power Inverter Terengganu, Malaysia, on November 6 on
Circuits for AC machines. 1995. 1988. She received B. Eng. degrees with
honours in electrical engineering from the
[7] F. Charfi, F.S.a.K.A.-H., Fault diagnosis in power
Universiti Teknikal Malaysia Melaka
system using wavelet transforms and neural networks. (UTeM) on 2012. Currently, she is a
2006. master degree student researcher in
[8] Rioul, O. and M. Vetterli, Wavelets and signal electrical engineering at Universiti Teknikal Malaysia
processing. Signal Processing Magazine, IEEE, 1991. Melaka (UTeM).
8(4): p. 14-38.
[9] Jardan, K.R., S.B. Dewan, and G.R. Slemon, General
Dr. Abdul Rahim Abdullah was born in
Analysis of Three-Phase Inverters. Industryand General
Kedah, Malaysia on 1979. He received his
Applications, IEEE Transactions on,1969. IGA-5(6): p. B. Eng., Master, PhD Degree from University
672-679. of Technology Malaysia in 2001, 2004 and
[10] Sibasish Pandaa, A.K.P.a.H.N.P., Fault Analysis on 2011 in Electrical Engineering and Digital
Grid Connected MPPT based Photovoltaic System. Signal Processing. He is currently a Senior
International Journal of Current Engineering and Lecturer and Coordinator of Center of Excellent Robotic and
Technology, 2013. Industrial Automation (CeRiA) in Universiti Teknikal Malaysia
[11] Abdullah, A.R., et al. Short-circuit switches Melaka (UTeM)
faultanalysis of voltage source inverter using
spectrogram. in Electrical Machines and Systems
Norhazilina Bt Bahari was born in
(ICEMS), 2013 International Conference on. 2013.
Melaka, Malaysia. She received her
[12] Abdullah, A.R., et al. Open switch faults analysis in
Master Engineering in Electrical
voltage source inverter using spectrogram. in Power
Engineering (Power),UTM and completed
Engineering and Optimization Conference (PEOCO),
her Bachelor Engineering degree in
2013 IEEE 7th International. 2013.
Electrical Engineering at UTeM.
[13] Abidin, N.Q.Z., et al. Leakage current analysis on
polymeric surface condition using time-frequency
distribution. in Power Engineering and Optimization
Conference (PEDCO) Melaka, Malaysia, 2012 Ieee
International. 2012.
[14] Bin Abdullah, A.R., A.Z. Bin Sha'ameri, and A.B.
Jidin. Classification of power quality signals using
smooth-windowed Wigner-Ville distribution. in
Electrical Machines and Systems (ICEMS), 2010
International Conference on. 2010.
[15] Abdullah, A.R., A.Z. Sha'ameri, and N.M.
Saad.Power quality analysis using spectrogram and
gabor transformation. in Applied Electromagnetics,
2007. APACE 2007. Asia-Pacific Conference on.2007.
[16] Abdullah, A.R. and A.Z. Sha'ameri.
Powerqualityanalysis using linear time-frequency
distribution. in Power and Energy Conference, 2008.
PECon 2008. IEEE 2nd International. 2008.
[17] R.C.Dugan, S.S., M. f. McGranaghan and H.W.beaty,
Electrical Power System Quality. McGraw Hill 2002.
[18] M.T.Thompson, A.K.a., Power Quality in Electrical
Systems. McGraw Hill, 2007.
[19] Gu, H.J.B.a.Y.H., Signal Processing of Power Quality.
Wiley-Interscience , 2006.

View publication stats