VOLTAGE SOURCE INVERTER FAULTDIAGNOSIS

IN VARIABLE SPEED AC
DRIVES,BY THEAVERAGECURRENT
PARK'SVECTORAPPROACH

A. M. S. Mendes and A. J. Marques Cardoso

Universidade de Coimbra, Departamento de Engenharia ElectrotCcnica
P610 I1 - Pinhal de Marrocos, P-3030 Coimbra, Portugal
Tel: + +
351 39 796200, Fax: 351 39 796247, E-mail: [andresm], [acardoso]@dee.uc.pt

Abstract: This paper introduces a new approach, based on U. AVERAGE

CURRENT PARK'S VECTOR APPROACH
the average motor supply current Park's Vector monitoring,
for diagnosing voltage source inverter faults in variable speed As a function of the average motor supply currents Iuav,I,,"
AC drives. Both simulation and laboratory tests results dem- and IWav,the corresponding Park's Vector can be expressed
onstrate the effectiveness of the proposed on-line diagnostic by:
technique.

I. INTRODUCTION

The voltage source inverter (VSI) is one of the most used where
circuit configurations for speed control of three-phase induc-
tion motors. This type of converter creates a relatively well-
defined switched voltage waveform at the induction motor
terminals. If for any reason, one of its power switches fail, (3)
this voltage waveform will change and critical damages can
occur either in the inverter or in the motor connected to it.
Considering that unintentional stoppages of large and criti- The operating philosophy of the Average Current Park's
cal adjustable speed AC drives are not desirable, it is ex- Vector Approach is thus based on relating the fault type with
tremely useful the availability of fault diagnostic techniques the average motor supply current Park's Vector modulus and
that would aid in detecting, identifying and locating the faults the faulty element with the corresponding phase value [7].
and, subsequently, in isolating the faulty element(s). A non-
invasive technique for diagnosing VSI failures in variable HI. LABORATORY
TESTS
speed AC drives, based on the identification of unique sig-
nature patterns corresponding to the motor supply current A. Details of the Test Rig
Park'!; Vector, was proposed in [l]. The same approach was
also proposed for variable speed DC drives fault diagnosis An open-loop volt/Hz speed control strategy of an induc-
[ 2 ] ,[3]. A method for the on-line detection of the intermittent tion motor fed by a PWM voltage source inverter, based on
misfising of the switching devices in a voltage-fed PWM IGBT technology, was used in this study. Fig. 1 shows the
inverter induction-motor drive was also reported in [4]. The main power circuit of the VSI fed AC motor drive used for
method is based on the time-domain response analysis of the the laboratory tests. The DC supply voltage is obtained from
induction motor current space vector which the inverter sup- a three-phase diode rectifier bridge connected to a variable
plies, followed by the application of a suitable pattern recog- AC supply.
nition algorithm.
This paper introduces a different diagnostic approach,
which can be readily integrated into the drive control system
without any additional pattern recognition algorithms, there-
fore reducing considerably the required computational effort.
The integration of diagnostic routines into the drive control
VDC
i. "L V
b 4 : )
W *1.

system allow for the introduction of remedial control strate-

gies soon after failure occurrence, thereby improving reli- II 't-I
I I
ability and fault tolerant capability of the drive system [ 5 ] ,
[61. Fig. 1 - Power circuit of the VSI fed AC motor drive.

0-7803-5293-9/99$10.00 0 1999 IEEE 704

 The motor used for the laboratory tests is a '+-pole, 3801660 Notice that in the real occurrence of an IGBT short-circuit a
V, 1.610.92 A, 0.52 kW, cos $ = 0.75, AEG induction motor larger average motor supply current Park's Vector modulus
with delta stator connection. A mechanical load was provided will be obtained. Similarly to the open-circuit faults, the
by a DC generator feeding a variable resistor. identification of the short-circuited controlled power switch
can be reached by determining in which interval the average
B. Experimental Results motor supply current Park's Vector phase (esav) falls in (Ta-
ble III).
Under normal operating mode of the drive system the mo- The 180' displacement of the e,," intervals, for open-circuit
tor line currents waveforms present a zero (nearly) DC com- faults and short-circuit faults in the VSI controlled power
ponent (Fig. 2). In this situation the average motor supply switches (Table 111), is due to the opposite polarity of the DC
current Park's Vector modulus becomes also zero (nearly). components of the motor line currents (Fig. 3 and Fig. 4).
For a power switch open-circuit fault a non-zero DC com-
ponent is imposed in the machine stator windings, which can
be observed in Fig. 3, for motor line currents waveforms U
and W, concerning the case of an open-circuit fault in IGBT F"""' " " " " "" " " " ' "." '7 C h l Mean
4 2 74mV
T,. Therefore, the occurrence of an open-circuit fault in the
m
IGBT TI, for example, is characterised by the following aver- Ch2 Mean
-25 13mV
age motor line currents: IUav= -2.082 A, Iv, = 1.008 A (ex-
trapolated) and Iwav= 1.074 A (Fig. 3), from which an aver-
1 J
age motor supply current Park's Vector with a modulus of
2.550 A and phase 181.04' is obtained. Further results, con- 2'
1 D lRMS
Ch2 SV

cerning open-circuit faults in other power switches, can be

found in Table I.
For a power switch short-circuit fault a non-zero DC com- 2 5V M lOms C h l J'' "'ti
ponent can be also observed in the motor line currents, as can Fig. 2 -Experimental motor line currents (Chl: Line U. Ch2: Line
be seen in Fig. 4, for motor line currents waveforms U and W) for the normal behaviour of the VSI.
W, concerning the case of a short-circuit fault in IGBT TI.
Therefore, the occurrence of a short-circuit fault in the IGBT
TI, for example, is characterised by the following average
motor line currents: I,,"= 3.293 A, I",,= -1.533 (extrapo- C h l Mean
-2.082 V
lated) and IWav= -1.760 A (Fig. 4), from which an average
motor supply current Park's Vector with a modulus of 4.036 Ch2 Mean
1.074 V
A and phase 2.27" is obtained. Further results, concerning
short-circuit faults in other power switches, can be found in
Table 11.
t i h l RMS
C2.722 V

Notice that for the short-circuit tests a 17R auxiliary re- Ch2
2 . 7 1RMS
29

sistor was used, connected in parallel with the power switch

to be tested, in order to limit the short-circuit current.
5V Ch2 5V m C H l I 0

C,Fault Diagnostics Fig. 3 -Experimental motor line currents (Chl: Line U, Ch2: Line
W) for an open-circuit fault in IGBT T,.
Comparing the average motor supply current Park's Vector
values expressed in Table I with the situation described for
the normal behaviour of drive system, it can be shown that
the occurrence of controlled power switches open-circuit C h l Mean
3 293V
faults is easily detected by observing that the corresponding I
average motor supply current Park's Vector modulus is al-
ways different from zero. The identification of the faulty I I
power switch can be obtained by determining in which inter-
val the average motor supply current Park's Vector phase
(esa,)falls in (Table 111). 2+ Nul.
The occurrence of controlled power switches short-circuit
faults is characterised by average motor supply current Park's
ci'i'~"'''b
Vectors moduli, different from zero and greater than the ones 5v' %" ' ' s v ' " m m s
corresponding to the controlled power switches open-circuit Fig. 4 -Experimental motor line currents (ChI: Line U, Ch2: Line
faults, thereby allowing for a reliable fault type detection. W) for a short-circuit fault in IGBT T,.

705
 TABLE I 3.0

EXPERIMENTAL
RESULTS FOR AN OPEN-CIRCUIT FAULT
IN THE POWER SWITCHES (a)

-3.0
0.0 100.0 [mr]
1; -1.094 -1.095 2.189 2.680 240.01
7; 1.082 -2.285 1.203 2.799 298.25
Z4 2.287 -1.129 -1.158 2.801 0.41
r; 1.023 0.961 -1.984 2.430 58.90 ......... t....... 1 1...... ...1 ....
.. .-, ... +_ ...... ..., .......
7' -0.788 1.889 -1.101 2.324 114.54 -3.0
0.0 ioo.0 [mr]
(1)

TABLE I1 3.0
L .... ............
EXPERIMENTAL
RESULTS FOR A SHORT-CIRCUITFAULT
IN THE POWER SWITCHES

IGBT
- I,, (A) (A) I,, (A) 1 ,1 (A) e,o,r) ....... ....
....
..
..............
..

;rt 3.293 -1.533 -1.760 4.036 2.27 -3.0

00 1000 [mr]
;r, 1.645 1.647 -3.292 4.031 60.02
3.0
ir3 -2.004 3.729 -1.689 4.570 123.12
jr4 -3.729 1.539 2.190 4.590 185.75
(b)
T5 -1.483 -1.705 3.188 3.900 242.30
T' 1.987 -3.829 1.829 4.690 301.30
~~ -3.0
0.0 100.0 [msl
(2)
TABLE In Fig. 5 -Simulated motor line currents ((a) Line U; (b) Line W) for:
INTERVALS OF e,, FOR POWER SWITCHESFAULT IDENTIFICATION (1) Open-circuitfault in IGBT T,; (2) Short-circuitfault in IGBT T,.
IGBT Oven-circuif Faul Short-circuifFault
VI. ACKNOWLEDGEMENT

The authors wish to acknowledge the financial support of

the Portuguese Foundation for Science and Technology
(FCT), under Project Number PBIC/C/CEG/2441/95.

vn. REFERENCES
A. M. S . Mendes, A. J. Marques Cardoso and E. S . Saraiva, "Voltage Source
Iv. SIMULATION Inverter Fault Diagnosis in Variable Speed AC Drives, by Park's Vector Ap-
proach, IEE 7th International Conference on Power Electronics ond Vuriuble
Speed Drives. IEE Conference Publication No. 456, pp. 538-543, London. UK.
Fig. 5 shows simulated results, concerning the motor line September 21-23, 1998.
currents waveforms U and W, for open-circuit and short-
A. I. Marques Cardoso and A. M. S. Mendes, "Semi-Converter Fault Diagnosis
circuit faults in IGBT T,. in DC Motor Drives, by Park's Vector Approach, IEE 6th International Con-
ference on Power Electronics and Variable Speed Drives, IEE Conference Pub-
lication No. 429, pp. 93-98, Nottingham, UK, September 23-25.1996.
V. CONCLUSIONS
A. J. Marques Cardoso and A. M. S . Mendes. "Converter Fault Diagnosis in
Variable Speed DC Drives, by Park's Vector Approach, IEEE Internotional
This paper introduces a new approach, based on the aver- Symposium on Industrial Electronics, vol. 2, pp. 497-500, Guimarses. Portugal,
age motor supply current Park's Vector monitoring, for diag- July7-11, 1997.
nosing voltage source inverter faults in variable speed AC K. Scot Smith, Li Ran and Jim Penman, "Real-Time Detection of Intermittent
drives. Experimental and simulated results show that VSI Misfiring in Voltage-Fed PWM Inverter Induction-Motor Drive", IEEE Truns-
actions on Industrial Electronics. vol. 44, No. 4, pp. 468-476, August 1997.
controlled power switches open-circuit faults and short-
drives can be effectively diagnosed by using this new ap- K. Debaprasad and B. K. Bose, "Fault Mode Single-phase Operation of a Vari-
able Frequency Induction Motor Drive and Improvement of Pulsating Torque
proach, whose operating philosophy relies on the behaviour Characteristics". IEEE Transactions on Industrial Elecrronics. vol. 41. No. 4,
of the average motor supply current Park's Vector modulus pp. 426-433, August 1994.
and phase. K. Debaprasad and B. K. Bose, "Investigation of Fault Modes of Voltage-Fed
Further work concerning the application of the Average Inverter System for Induction Motor Drive", IEEE Trunsuctions on Industry
Applications, vol. 30. No. 4, pp. 1028-1038, JulyIAugust 1994.
Current Park's Vector Approach is currently in progress to-
A. M. S . Mendes, "Fault Diagnosis in a Rectifier-Inverter System usedfor
wards the development of fault tolerant adjustable speed AC Speed Control of Three-Phase Induction Motors" (in Portuguese), MSc. thesis,
drive systems. University of Coimbra. Portugal, 1998.

706