Proceedings of Engineering & Technology (PET)

pp. 389-393
Copyright IPCO-2016

Identification of the fault leg in a three phase voltage

source inverter
Fatima zahra Bensouda#1, Ali Haddi#2, Tandjaoui Mohammed Nasser*3
#
Laboratory of Innovative technologies, National School of Applied Science
Tangier, Morocco
fatimazahra.bensouda@gmail.com
haddi.ensat@gmail.com
*
Department of Technology, Bechar University
Algeria
tanjaoui_08@yahoo.fr

Abstract Nowadays, inverters are power electronics devices A much methods based on the analysis of the current vector
used in the renewable energy conversion systems and also to fed have been suggested and introduced by Fuchs and Peuget
induction motors, therefore the continuity of service of power which proved its effectiveness [4-5].
converter has been a major concern. This paper presents a new Peuget technique requires only one period to detect and
approach of open circuit fault detection in a pulse width
isolate the fault after its occurrence, in the other hand, Fuchs
modulation three phase voltage source inverter. The fault
occurrence detection was based on the standard deviation of the detection is achieved in one fourth of a cycle. The current
measured current vector and the localization of the fault leg vector approach is used also for two level voltage inverter [6]
where the open circuit fault occurred was achieved analyzing the and three parallel power converters [7-8].
current vector in d-q frame system. The reliability of this Zidani used only two current sensors for simplicity and
technique was confirmed in a real time simulation. cost-effectiveness based on the Concordia current pattern with
a fuzzy diagnosis [9] while Diallo suggested a pattern
Keywords inverter, IGBT, power converter, induction motor, recognition approach based also on Concordia stator mean
current vector current vector [10]. Further, Caseiro published on the fault
diagnosis using the average current vector [11].
I. INTRODUCTION One of the biggest challenges in open circuit fault diagnosis
In many industrial applications, inverters are power is to minimise the fault detection and localization time, most
electronics device used to control the speed of induction techniques require one period to detect the fault occurrence,
motors and to convert energy from natural renewable sources the present paper propose a fault diagnosis technique for
to an electrical network. Generally, induction motors are fed single IGBT open circuit fault based on the analysis of the
from pulse width modulation (PWM) voltage source inverter. current vector, the fault is immediately detected and the fault
Much research in recent years has focused on the reliability of leg where the open circuit fault occurred is defined in a
inverters. In fact, switching devices of power converters such maximum of one current period and half without any
as IGBT insulated gate bipolar transistor are the weakest additional sensor or complex training algorithm.
components. Switching device faults in power converters are
short circuit and open circuit fault, short circuit generates
abnormal over current and the system is immediately halted II. SYSTEM DESCRIPTION
for safety. The voltage source inverter is a power electronic device
For open circuit faults in power converter unbalanced which converter energy from a direct current (DC) to
currents results and healthy switches are overstressed which alternating current (AC), used in many applications such in
can damage the system if the fault is not immediately renewable energy systems and in the industrial field to fed an
compensated. Rui and Huang cover the causes of IGBT induction motor. Therefore the reliability of power converter
insulated-gate bipolar transistor open circuit [1]. is extremely important. A fault diagnosis technique must be
Different fault detection methods have been proposed. developed to detect in a short time the faulty leg and isolate it
Ribiero proposed four techniques using extra voltage sensors to avoid possible damages in the system. The configuration of
at specific points to minimize time of fault diagnosis [2]. A the VSI considered in this study has been developed in
simple hardware have been proposed by Trabelsi to detect Matlab/Simulink as shown in fig.1, the pulse width
single and multiple open circuit faults in a maximum of one modulation (PWM) was used to control the output voltage of
switching period analysing the PWM switching signals and inverter.
line to line voltage levels [3].

ISSN: 2356-5608
 Fig. 3: values of x before and after fault occurrence

Fig.1: Matlab-Simulink model of a power inverter B. Fault leg identification

To identify the fault leg , the current in the d frame id was
A. Open circuit fault detection analysed, when a fault occurs after one period and half the
The proposed fault diagnosis detects the open circuit fault maximum and the minimum of the current id are calculated,
immediately after its occurrence; this technique requires the three cases are to discuss, if the maximum of id is zero or the
measurement of the output inverter currents then the currents minimum of id is zero the faulty switch is in the third leg
are expressed in the d-q frame on which the horizontal-axis is otherwise the sign of the current in q frame is obtained for the
- maximum of the current in d frame id , if the sign of iq is
The current in the d-q frame are expressed as follow: positive the open circuit fault occurred in the second leg , in
the contrarily if the sign of the current iq is negative the first
id =square root(1/6)*(2*ia-ib-ic) (1) leg must be isolated. The flow chart diagnosis below resume
the fault localization approach proposed. Therefore, a
iq =square root(1/2)*(ib-ic) (2) regression tree was built.

A new variable x is introduced, this variable in healthy Start

mode changes in the same interval but once the fault occurs
this variable diverges and changes in a larger interval which ia, ib, ic
indicates the fault occurrence at this specific time. Therefore x
is used to detect an open circuit fault in the inverter. d-q frame transformation
id, iq
x= square root (id²+id²) (3)
Fault detection: Tracking the
Fig.2 shows the patterns of the current vector in the healthy standard deviation & data storage
mode and in the faulty mode and fig.3 indicates the the fault
occurrence simulated at t=0.028 sec is visible. No
Fault ?
Healthy mode Faulty mode
Yes

Calculate max (id) and min (id)

Max (id) =0 or min (id)

Third leg faulty sign (iq) for max (id)

positif negatif
Fig. 2: Current patterns in healthy and faulty mode
Second leg faulty First leg faulty

Fig. 4: Flow chart fault diagnosis system

 Rigorously, the patterns of one current period were d-q frame can be found in fig.5. The fault occurrence is
analysed for every open circuit fault and were obtained after obvious which show a fairly consistent detection.
one period and half of the fault occurrence. The measured
current data were stored in the data storage step. For each leg
the patterns of the two open circuit fault possible are
represented in the following table:

TABLE I

Specific faulty leg

First leg Second leg Third leg

Fig.4: Phase currents in faulty mode

Fig.5: Currents in d-q system in faulty mode

Other researchers used the analysis of the current vector in Special patterns are also found in other researches using the
d-q frame; almost the same patterns were obtained in this analysis of the current vector by Peuget, Trabelsi and Fuchs
study. To detect the open circuit fault is to detect the faulty leg which confirm and support the results obtained.
which was the purpose of this research. Although, the Most techniques require one period cycle to detect the fault
technique proposed identify the fault leg only for one open occurrence and the open circuit fault is identified within a
circuit fault. maximum of two periods, this technique detects immediately
the fault occurrence and the localization of the fault leg is
III. RESULTS AND DISCUSSION achieved in a short time, one period and half. Ideal shapes of
As mentioned earlier, the aim of this study is to introduce a the vector currents were obtained after one period and half of
new approach based on the analysis of the current vector in d- the fault occurrence and the currents in d-q frame for one
q frame; to detect the fault leg where the open circuit fault period were analysed. This technique was achieved without
occurred. In this study the fault occurrence was immediately any additional sensor or complex training algorithm. However,
detected and the identification was achieved by a regression detection of multiple open circuit fault were not explored in
tree; to detect the fault occurrence the variable x was this study, nevertheless it can be introduced in a future work.
introduced which is the standard deviation of the currents in d- Only simulation results using simulink were obtained,
q frame , if the parameter x changes in a larger interval than experimental results were not achieved, although simulink is a
usual an open circuit fault occurred . In the other hand the real time simulation and the results were repeatedly and
localization was based on the maximum and the minimum of accurately obtained.
the current id, if both are equal to zero than the fault occurred
in the third leg otherwise, the sign of the current iq is IV. CONCLUSION
determined for the maximum of the current iq , if it is positive A new approach for open circuit fault detection was
the second leg is faulty else ways the first leg is faulty. introduced; this method was based on the analysis of the
Simulation results were obtained using Simulink/ Matlab current circuit in d-q system, which detected the fault leg
real time simulation, for each open circuit fault a particular where the open circuit fault occurred in a short time, one
shape was obtained as shown in table 1. The current vector in period and half of the current vector. The technique was
d-q coordinate system in healthy mode and faulty mode were verified through simulation. A multiple open circuit fault
represented in fig.2, the fault occurrence is visible in fig.3. detection and a fault tolerant and compensation system will be
The T4 switch experiences an open circuit at t=0.028sec, the presented in a future work.
faulty phase currents are illustrated in fig.4 and the currents in
 REFERENCES
[1] W. Rui, B. Frede, W. Huai, L. Marco and I. Francesco , Catastrophic
Failure and Fault-Tolerant Design of IGBT Power Electronic
Converters -An Overview, Proceedings of the 39th Annual
Conference of the IEEE Industrial Electronics Society, IECON 2013.
[2] R. L. de Araujo Ribeiro, C.B. Jacobina, E.R.C. da Silva, and A.M.N.
-switch damage in voltage-fed PWM
motor drive system, ,
Mar. 2003.
[3] M. Trabelsi, M. Boussaka, M. Gossa , -Switching pattern-based
diagnosis scheme for single and multiple open-switch damages in VSI
ISA Transactions 2012.
[4] F.W. Fuch Some diagnosis methods for voltage source inverter in
variable speed drives with induction machines, IEEE IECON 03,
November 2003, vol. 2
[5]
on a PWM inverter by knowledge-based mod
Applications, Vol. 34, No. 6 Nov./Dec. 1998.
[6] T. Orlowska-Kowalska and P. Sobanski Simple diagnostic technique
of a single IGBT open-circuit faults for a SVM-VSI vector controlled
induction motor drive Bulletin of the polish academy of sciences
technical sciences, vol.63; No. 1, 2015.
[7] Y. Ko and K. Lee , Fault Diagnosis of a Voltage-Fed PWM Inverter
for a Three-parallel Power Conversion System in a Wind Turbine,
Journal of Power Electronics, Vol. 10, No. 6, November 2010.
[8] Y.J. Ko, K.B. Lee1, D.-C. Lee and J.M. Kim, Fault diagnosis of three-
parallel voltage-source converter for a high-power wind turbine, IET
Power Electron., 2012, Vol. 5, Iss. 7.
[9] F. Zidani, D. Diallo, M. Benbouzid and R. Naît-Saîd, A Fuzzy-Based
Approach for the Diagnosis of Fault Modes in a Voltage-Fed PWM
Inverter Induction Motor Drive, IEEE Transactions on Industrial
Electronics, Institute of Electrical and Electronics Engineers (IEEE),
2008, 55 (2).
[10] D. Diallo, M. El Hachemi Benbouzid, D. Hamad and X. Pierre, Fault
Detection and Diagnosis in an Induction Machine Drive: A Pattern
Recognition Approach Based on Concordia Stator Mean Current
Vector, IEEE Transactions On Energy Conversion, Vol. 20, No. 3,
sep. 2005.
[11] J. A. A. Caseiro, A. M. S. Mendes and A. J. Marques Cardoso, Fault
Diagnosis on a PWM Rectifier AC Drive System with Fault Tolerance
, in Proc. Eur.
Conf. Power Electron. Appl., Barcelona, Spain, Sep. 8 10, 2009.