You are on page 1of 7

Inverter Fault-Identification for

VSI Motor Drives


A. Gaeta, G. Scarcella, G. Scelba, S. De Caro, A. Testa

Φ
Abstract – The paper presents an identification procedure frequency nature of the injected signals [7].
able to detect open circuit failures in voltage source inverters. Failure detection is generally accomplished according to
The procedure is suitable to equip fault tolerant electric motor measurement of the inverter phase currents or voltages,
drives and in principle only requires the measurement of the through additional sensors. However, extra sensors increase
DC link current as it accomplishes a reconstruction of the
inverter output phase currents from the DC link current the cost and complexity of any motor drive for industrial
measurement, thus eliminating the need of conventional applications. A possible way to obtain information about the
current sensors. However, if they are present at the cost of phase currents through a reduced number of current sensors
minimal modifications the PWM technique, response times as is represented by a reconstruction of such currents from the
short as one switching period can be obtained. Compared to DC link current, based on the sequence of the inverter states
other technical solutions found on past literature, the proposed [8]. Only a single current sensor is therefore needed to
approach considerably reduces the required computational
effort. Simulations and experimental tests are presented to reconstruct the three phase currents, moreover it can be
confirm the consistence of the proposed approach. implemented through a simple and inexpensive circuitry.
Unfortunately, this approach fails in some working
Index Terms— Current Measurement, Failure Diagnosis, conditions. In fact, at low modulation indexes and when
Reliability, Space Vector Pulse Width Modulation, Variable two-phase voltages feature the same duty cycle an incorrect
Speed Drives. reconstruction of the phase currents is obtained. As this
drawback largely prevents any practical application, more
I. INTRODUCTION sophisticated techniques have been proposed using either

T HREE poles, six switches, Voltage Source Inverters


(VSI) are the normal equipment for three phase
electrical motor drives. However, more complex inverter
special filters, modifications of PWM patterns, or state
observers [9]÷[12]. However, although they are suitable to
reconstruct the phase currents for normal control purposes,
structures featuring extra power switches, or additional they hardly can be exploited to detect motor or power switch
poles, have been developed in the past to obtain fault failures.
tolerant electric motor drives [1]. In these systems, a This paper proposes a failure-identification procedure able
suitable procedure is implemented, able to detect the to detect open circuit power device failures, in voltage
occurrence of a failure, to identify the damaged power source inverters, by using a reduced set of current sensors. In
switch and the nature of the malfunctioning, as well as to particular, with a single current sensor on the DC link, a
reconfigure the structure of the inverter [2]÷[6]. The procedure able to detect open circuit failures is carried out,
response time of the failure detection procedure is an suitable to equip conventional or sensored, vector controlled
important aspect to ensure the required service continuity of industrial motor drives. Failure detection times are lower
a reconfigurable electric motor drive and is crucial in HF than two switching periods, thus suitable to permit a fast
injection-based sensorless motor drives. In these last, a inverter structure reconfiguration and to avoid any system
power switch failure has effects not only on the motor power shutdown, especially in case of HF injection-based
supply, but also on estimation of the rotor position the sensorless motor drives. The proposed procedure does not
injection of the HF signal also influenced by the failure. require modifications of the switching patterns during the
On standard sensored drives and open circuit failure healthy operation of the drive and can be also exploited to
detection and identification can be accomplished in several reduce the number of current sensors in conventional
tens of milliseconds, without affecting drive operations. industrial motor drives. If compared to previously developed
On the contrary, on sensorless drives, an open circuit failure approaches, the proposed technique considerably reduces the
may remarkably alter the signal injection and so the rotor computational effort, leading to easy and inexpensive
position estimation in few milliseconds, due to the high practical implementation.

Φ
This work was partly supported by the Project: “Optimization and II. OPEN CIRCUIT FAILURE DETECTION AND
Control of Electrical Drives for Fault Tolerant and Ride-Through IDENTIFICATION
Operation”, sponsored by the Italian Ministry of Education and Research,
PRIN 2008.
The actual state of a three-phase inverter can be described
A. Gaeta, G. Scarcella, G. Scelba are with University of Catania, Viale by a sequence of binary digits to indicate if each leg output
Andrea Doria 6, 95125 Italy (e-mail: alberto.gaeta@dieei.unict.it) is connected to the higher (1) or to the lower (0) DC rail.
A. Testa S. De Caro are with University of Messina, Viale Ferdinando Considering all possible combinations, a three-phase inverter
Stagno D’Alcontres 31, 98166 Sant’Agata (ME) (e-mail:
testa@ingegneria.unime.it). can generate only six active voltage vectors and two zero

978-1-4244-9303-6/11/$26.00 ©2011 IEEE

413
vectors as depicted in Fig. 1. When an inverter power device Similar considerations can be applied to the case of a
Sx x=1,2…6 (Fig. 2), is affected by an open-circuit failure, negative leg current. In fact, as soon as the leg current is
the real sequence of voltage vectors applied to the load can negative, only the lower device and the upper diode of the
differ from the commanded one, depending on the direction leg take part to the current conduction, and an open-circuit
of the current in the failed leg. failure to the upper device will not alter normal operation.
On the other hand, a failure of the lower device will connect
the output of the affected leg to the higher dc rail,
compromising the capability of the leg to generate a
modulated voltage and causing the negative leg current to
suddenly increase.

A) Failure Detection
For a wye connected system, like that shown in Fig. 2, in
normal operations, the DC link current is zero during
inverter zero states (S1-S3-S5 or S2-S4-S6), because all the
phases are short-circuited each other. However, under open
circuit failure conditions, the DC-link current is not zero
during a commanded zero state, because the actual state of
3
3
V DC the inverter is rather an active state. As shown in Fig. 3, in
case of a failure of an upper device, the state of the failed leg
Fig. 1. Voltage space vector representation for three leg VSI.
is (0) while is (1) if the failed device is the lower one.

Healthy System

Fig. 3. Failure on S3: CH1) VaN [100V/div.], CH2) VbN [100V/div.], CH3)
VcN [100V/div.], CH4) iDC [0.5A/div.], Time [20µs/div.].

Fig. 2. Fault Tolerant three phase VSI. In particular, a unique output current combination
corresponds to each state of the inverter represented in Fig.
As an example, if the leg current is positive (flowing from 1. These combinations can be detected through a measure of
the inverter to the load), only the upper power device of the the DC link current. During a commanded zero state, the
leg (S1, S2 or S3) will take part to current conduction: when faulty DC-link current is equal to minus the absolute value
the upper device is active, it will carry the leg output current, of the current in the affected leg. In this way, by measuring
while when inactive the current will be conducted by the the DC link current and using a threshold comparator, it is
lower diode of the same leg (D2, D4 or D6); therefore, under possible to detect an open-circuit failure of the upper devices
the hypothesis of positive leg current, the leg output voltage (S1-S3-S5) during a commanded (111) state or of the lower
is imposed by the state of the upper device of the leg. devices (S2-S4-S6) during a (000) state. The threshold should
Consequently, an open-circuit failure to the lower device (S2, be suitably selected in order to avoid false detections, due to
S4 or S6) will not affect the inverter operation as long as the the presence of an unavoidable minimal DC-link currents,
leg output current will remain positive. common mode currents, sensor noises, etc [3]. In addition,
Obviously, the occurrence of a failure of the upper power the DC-link measurement can be used to detect inverter earth
device, when the leg output current is positive, will cause the faults according to [2]. The DC-link current measured during
current to be carried by the lower diode of the leg, a commanded zero state will be always positive in case of
independently of the gate signals. The failed leg output earth leakage currents, while it will be negative in case of an
voltage will be tied to the lower dc bus rail, causing the open-circuit fault.
positive leg current to suddenly decrease up to zero if no In the following, it is assumed that in the inverter a power
action is rapidly taken [12]. device is affected by an-open-circuit failure only if it is

414
conducting current, or in the previous PWM period Ts it has (010), (001) in case of failure of one of the upper devices, or
been conducting current. Moreover, the analysis neglects the the sequence (011), (101), (110) in case of failure of a lower
case of multiple contemporary faults and diode failures as device as described in TABLE II.
well. The failure detection time may vary depending on the For example, if S3 is the failed device, the DC-link
particular modulation strategy used. In particular, if a current, measured during commanded (010) state, should be
symmetric PWM switching pattern is used, the maximum below the threshold because the actual inverter state is (000)
delay time is equal to two PWM sampling periods. Such a and the state of the affected leg is (0), independently of its
value is obtained as the worst case occurring when an open gate signals. However, such a procedure is incomplete
circuit failure happens just after the end of a zero state DC- because the absolute value of the measured DC link current
link current measurement. This delay can be reduced by during an active test state, can fall below the threshold even
using an asymmetric PWM switching pattern, as the adjacent if the power device under test is not damaged. It must be
zero states modulation shown in TABLE I. This strategy emphasized that, after that a failure has occurred, sudden
allows to reduce the detection delay to one sampling period, variations will appear also on the output currents of the
at expenses of increased switching losses, due to the healthy legs, as represented in Fig. 4. Such variations will
additional commutations required. increase/decrease the amplitude of the healthy currents,
which have equal/opposite sign respect to the faulty current.
Table I TABLE I: Switching pattern of the adopted SVM technique Therefore, during the unavoidable time delay required by
fault identification, healthy currents with the same sign with
Sector Switching Pattern respect to the faulty current will become larger while healthy
V1 V2 V7 V0
I currents with opposite sign will become smaller, making the
(100) (110) (111) (000)
identification procedure more difficult. A critical situation
V3 V2 V7 V0
II
(010) (110) (111) (000)
arises when two measured DC-link current are below the
V3 V4 V7 V0
identification threshold, and cannot be solved just applying
III small duration active state tests. A possible solution could be
(010) (011) (111) (000)
V5 V4 V7 V0 to prolong the duration of each active test state until a
IV possible fault is identified. The time duration enlargement
(001) (011) (111) (000)
V5 V6 V7 V0 should allow current amplitude to increase up to exceed the
V detection threshold if the device is not failed. In this way a
(001) (101) (111) (000)
V1 V6 V7 V0 second DC- link current measurement, performed at the end
VI
(100) (101) (111) (000) of the longer active test state, will eliminate any uncertainty.
The main disadvantage of this approach is that each an
If a continuous failure monitoring is required, a third limit enlarged active test state should last more than:
is that both zero states must be present in each PWM period
and with a time duration sufficiently high to allow a correct tenlarged >= 2*Threshold*(3*L/2)/VDC
DC-link current measurement. In fact, each zero state should
last at least: tmin = td + tacq where tacq is the time required by where VDC is the DC-bus voltage and L represents the
current sampling and td is the time delay required to allow inductive component of the load.
the DC-link current settling [1], [2]. This limits the An alternative solution is to exploit the presence of
maximum fundamental amplitude of the motor voltages to redundant hardware, used for inverter reconfiguration after a
(1-2*tmin/Ts)Vsmax. Moreover, if an open-circuit fault happens fault, in order to retrieve additional information useful for
when the current direction in the leg makes the inverter fault identification. As an example, assuming the availability
operating normally, the fault detection is not effective. of a fourth leg connected to each of the inverter output by
However, if the inverter is properly designed and assembled, means of three couples of anti-paralleled GTO, as
the possibility that a power device fails while it does not represented in Fig. 2, after a possible fault identification, the
carry any current, is very low. These considerations apply to active state is extended using the following sequence:
the power section (gate drivers, power devices) of the - both gate signals of the leg under test are set to the low
inverter. In addition, unforeseen open-circuit state of power logic level, thus disabling the leg;
devices can also be caused by failures of digital components - the proper GTO couple is activated in order to connect
of the inverter. These occurrences are not necessarily related the fourth leg in parallel with the leg under test;
to current conduction across the affected device. In this case, - the same active test state is applied using the fourth leg
the fault detection can be accomplished by feeding back to instead of the leg under test.
the control unit the actual devices driving voltages, in order If the measured DC-link current exceeds the identification
to allow an immediate detection of failures of digital threshold, the power device under test is the failed one,
components. otherwise, the device is healthy and the current in the leg
under test is physiologically below the threshold. If a fast
B) Failure Identification fault detection and identification is mandatory, long active
After that a failure has been detected a failure test states may be avoided by properly exploiting the
identification procedure must be performed, to identify the information regarding the inverter output currents. If the
failed device. A straightforward method to identify the failed amplitude and sign of the inverter output currents are known,
device is to apply the additional active states sequence (100), it is possible to predict which fault can be detected.

415
Table II
Additional
Fault Phase currents Detectable Faults Fault Detection Fault identification
states
(ia>imin,ib<imin,ic<imin) S1 After (111) state
Contemporary with fault
(S1, S3, S5) (ia<imin,ib>imin,ic<imin) S3 current Not required
detection
(ia<imin,ib<imin,ic>imin) S5 measurement
(ia>imin,ib>imin,ic<imin) S1,S3 After (111) state (100),(010) S1 after (100) curr. meas.
(S1, S3, S5) (ia>imin,ib<imin,ic>imin) S1,S5 current (100),(001) S3 after (010) curr. meas.
(ia<imin,ib>imin,ic>imin) S3,S5 measurement (010),(001) S5 after (001) curr. meas.
(ia<-imin,ib>-imin,ic>-imin) S2 After (000) state
Contemporary with fault
(S2, S4, S6) (ia>-imin,ib<-imin,ic>-imin) S4 current Not required
detection
(ia>-imin,ib>-imin,ic<-imin) S6 measurement
(ia<-imin,ib<-imin,ic>-imin) S2,S4 After (000) state (011),(101) S2 after (011) curr. meas.
(S2, S4, S6) (ia<-imin,ib>-imin,ic<-imin) S2,S6 current (011),(110) S4 after (101) curr. meas.
(ia>-imin,ib<-imin,ic<-imin) S4,S6 measurement (101),(110) S6 after (110) curr. meas.

Fig. 4. Fault detection and identification at the same time. Fig. 6. Fault detection and identification using two additional test states.

As a consequence, it is possible to eliminate the If the failed device is identified when the first active test
uncertainties of the fault identification procedure and to state is applied, the identification procedure is stopped (Fig.
reduce the number of the required active test states from 5). If only one leg current meets the above criteria, the
three to one, as shown in Fig. 5, to two, (Fig. 6), or even to identification procedure is not necessary, because only one
zero (Fig. 4). For example, if a fault to an upper power leg may have caused the fault detection and the fault can be
device has been detected, only legs with sufficiently large immediately identified when detected (Fig. 4). By
positive currents may be affected by the fault. Therefore legs considering that, in the case of wye connected load, it is not
with small or negative currents, which are responsible for possible that all the three currents have contemporary the
indeterminations of the fault identification procedure, can be same sign, it is clear that applying three active test states for
immediately excluded before applying the active states test. fault identification is a waste of time, if the inverter output
By supposing that a fault to an upper power device is current information is available.
detected, if the two leg currents are positive and sufficiently
large, only two active states are required to properly identify III. PHASE CURRENTS RECONSTRUCTION FROM
the failed device (Fig. 6). DC-LINK SENSOR
Information about the inverter output currents can be
obtained by using additional current sensors or can be
reconstructed using the same DC-link sensor required for
fault detection and identification. In the latter case, in order
to assure a continuous output currents reconstruction, each
active state within a PWM period should last enough time to
allow proper active state current measurement. Two phase
currents can be measured in each Ts, and the third can be
calculated by using the constraint imposed by the wye load
connection. Such measurement are used in the control only
at the end of the next zero state (000), in order to discard any
possible wrong measurement due to the fault, as shown in
Fig. 7. Time duration of the active states become small when
Fig. 5. Fault detection and identification using one additional test state. the reference output voltage vector is located near the origin

416
of the state space plane of Fig 1, or when it is near to the the start-up of the machine. The results indicated in Figs. 9
boundary between two adjacent sectors. In the first case both and 10 show the behavior of the drive when an open circuit
the active vectors that are used to synthesize the reference fault occurs in S1, and when ias>0, ibs<0,and ics>0. As it is
vector, are too small to allow a correct DC-link current possible to note in Fig. 9, the fault situation occurs during
measurement, while in the second case only one of the active the inverter state (1,1,1), after the currents measurement.
vectors is too small. The system is not aware of the fault till the (1,1,1) inverter
state is again applied. In fact, a DC-link current higher than
the detection threshold is detected only in the latter state.
TABLE III: VSI Motor Drive Data
Rated Power 20HP Lm 0.076
Rated Voltage 460V J 0.1kgm2
Rated Frequency 60Hz F 0.02Nms
Rated Speed 1760 rpm Pole Pairs 2
Rs 0.2761Ÿ DC bus Voltage 796 V
Lls 0.0022 H Switching Frequency 10kHz
Llr’ 0.0022 H Dead Time 1µs

Since two power devices in the upper side of the inverter


are conducting contemporary, additional inverter states are
Fig. 7. Phase current reconstruction time sequence. necessary to identify which of these devices is failed. The
In this paper, in order to allow a continuous phase current first additional state (1,0,0) provides the solution because
reconstruction from the DC-link current measurement, active during this state the DC link current is zero. When the failed
vectors that are not enough long are suitably extended to device is identified, the inverter is reconfigured in order to
allow a proper current measurement. The distortion of the disable the broken leg and restoring the functionality of the
reference voltage, caused by such a modification of the drive.
switching pattern, is compensated by applying the
complementary active state. This is, referring to Fig.1, the
active state which has same direction but different sign, as
shown in Fig. 8. From the inverter output voltage point of
view, the increment of the duration of the active state and the
introduction of the complementary state is equivalent to the
application of a zero state. Consequently, the full ability in
reconstructing the phase current using the DC-link sensor is
obtained at the expenses of a reduced maximum inverter
output voltage.

Fig. 9. Current measurements and phase currents update.

The effects of this transient on the motor variables are


displayed in Fig. 10. Since the identification procedure lasts
less than Ts, a limited influence on the machine operation is
Fig. 8. Fault detection and identification using two additional test states. observed. As shown in Fig. 11, a similar analysis has been
conducted in case of an open circuit fault occurring in S5.
IV. SIMULATION RESULTS Differently from previous case, here the fault identification
The digital behavior of the fault identification procedure time is increased due to the use of a second additional test
has been tested by means of numerical simulations. In state to identify the failed device. Being S5 the failed device,
particular, the proposed method was applied to a three-phase the first additional test state only confirm that S1 is not
fault tolerant VSI motor drive using the same topology of affected by a fault. Even in this condition, the power system
Fig. 2, main technical characteristics are indicated in Tab.III. is rapidly reconfigured after fault identification, and the
The effectiveness of the method has been evaluated during effects on the machine are displayed in Fig. 12.

417
after a fault, correctly detected by the developed procedure,
is performed without any change in the motor current and
assuring the continuity of drive operations.

Fig. 10. Current measurements and phase currents update.

Fig. 13 Operation before fault: CH1) ia at motor terminal[1 A/div.], CH2) ia


at inverter terminal [1 A/div.], CH3) id [1 A/div.], Time [10ms/div.].

Fig. 11. Current measurements and phase currents update.

Fig. 14 Operation after a fault to A-leg, (A-leg gate signals disabled and
freewheeling diodes healthy): CH1) ia at motor terminal[1 A/div.], CH2) ia
at inverter terminal [1 A/div.], CH3) id [1 A/div.], Time [10ms/div.].

Fig. 12. Current measurements and phase currents update.

V. EXPERIMENTAL RESULTS
The proposed procedure has been experimentally tested,
using a three-phase VSI motor drive topology with the
characteristics of the system used for the simulations. Figs.
Fig. 15 Operation after a fault to A-leg, (A-leg gate signals disabled and
13-16 show the behavior of the system before and after a freewheeling diodes healthy): CH1) ia at motor terminal[1 A/div.], CH2) ia
switch fault. As it is possible to see the voltage modulation at inverter terminal [1 A/div.], CH3) id [1 A/div.], Time [10ms/div.].

418
[10]Kim Hongrae, T.M. Jahns, "Phase Current Reconstruction for AC
Motor Drives Using a DC Link Single Current Sensor and
Measurement Voltage Vectors",IEEE Trans. Power Electronics,
vol. 21, pp.:1413-1419,Sept. 2006.
[11]Jung-Ik Ha, "Current Prediction in Vector-Controlled PWM Inverters
Using Single DC-Link Current Sensor", IEEE Trans. Industrial
Electronics, vol. 57, pp.:716-726,Febr. 2010.
[12]D.P. Marc etiü, E.M. Adz iü, "Improved Three-Phase Current
Reconstruction for Induction Motor Drives With DC-Link Shunt",
IEEE Trans. Industrial Electronics, vol. 57, pp.:2454-2462,July 2010.
[13]G.L. Skibinski, R.J. Kerkman, D. Schlegel, "EMI emissions of modern
PWM AC drives", IEEE Industry Applications Magazine, vol. 5,
pp.:47-80, Nov.-Dec. 1999.
[14]S. Chakrabarti, T.M. Jahns, R.D. Lorenz, "Reduction of parameter
sensitivity in an induction motor current regulator using integrated pilot
sensors in the low-side switches", IEEE Trans. Industry Applications,
vol. 41, pp.:1656-1666, Nov.-Dec. 2005.
[15] S. Karimi, P. Poure, S. Saadate, "A Fault Tolerant Three-leg Shunt
Fig. 16 Fault transient on leg A, (A-leg gate signals disabled and free- Active Filter Using FPGA for Fast Switch Failure Detection", PESC
wheeling diodes healthy): CH1) ia at motor terminal[1 A/div.], CH2) ia at 2008, pp.:3342-3347, June 2008.
inverter terminal [1 A/div.], CH3) id [1 A/div.], CH4) TRIAC1 signal
[5V/div.], Time [10ms/div.].
VIII. BIOGRAPHIES
VI. CONCLUSIONS Alberto Gaeta was born in Lentini (SR), Italy, in 1980. He received the
M.S. degree in electronic engineering from the University of Catania, Italy,
With the aim to improve the fault tolerant capability of in 2008. He is working on his Ph.D. in Electrical Engineering through
electrical motor drives, this paper has presented a new collaboration with CePTIT of Misterbianco (CT) and with Dept. of
Electrical Electronics and Systems Engineering of University of Catania.
technique able to detect and identify open circuit faults of His fields of interest are electric machinery and electric drives with
the power devices in VSIs. According to the proposed particular attention to the research and optimization of sensorless control
method, standard current sensors can be eliminated and techniques.
response time of failures identification is greatly reduced. Giuseppe Scarcella received the M.S. degree in 1995 and the Ph.D. degree
Furthermore, compared to other failure-identification in 1998 both in electrical engineering from the University of Catania.
solutions, the proposed approach reduces the computational During 1998, he was Visiting Researcher at the University of Wisconsin,
Madison. Since 1999 he is at the University of Catania where, currently, he
efforts. Simulations and experimental tests are presented to is an Associate Professor in the areas of electrical machines, power
confirm the consistence of the proposed approach. electronics and drives. He has authored or co-authored over 100 technical
papers and three patents. He has achieved two IEEE awards, respectively
obtained in the year 2000 for the best paper published in the IEEE
VII. REFERENCES Transactions on Power Electronics, and in 1998 as the third prize paper
presented at the IEEE-IAS Annual Meeting. Both papers were on sensorless
[1] B. A. Welcko, T. A. Lipo, T. M.Jahns, S. E. Schulz "Fault Tolerant control of ac motor drives. His research interests include sensorless control
Three-Phase AC Motor Drives Topologies: A Comparison of Features, of electrical drives, digital modulation techniques, and electromagnetic
Costs and Limitations”, IEEE Trans. Power Electronics, vol. 19, compatibility.
pp.:1108-1116,July 2004.
[2] Bin Lu, S.K. Sharma, "A Literature Review of IGBT Fault Diagnostic Giacomo Scelba was born in Caltagirone (CT), on January 1, 1976. He
and Protection Methods for Power Inverters", IEEE Trans. Industry received the M.S. and Ph.D. degrees in electrical engineering from the
Applications, vol. 45, pp.:1770-1777, Sept.-Oct. 2009 University of Catania, Catania, Italy, in 2002 and 2005, respectively. In
2004, he was a Visiting Student at Rockwell Automation Standard Drives
[3] M.B. de Rossiter Correa, C.B. Jacobina, E.R.C. da Silva, A.M.N. Lima, Development, Mayfield Heights, OH. He is currently a Researcher in the
"An Induction Motor Drive System with Improved Fault Tolerance", Department of Electric, Electronic Engineering and Computer Science,
Trans. Industrial Electronics, vol. 37, pp.:873-879, May/June 2001. University of Catania. His current research interests include sensorless
[4] S.M. Jung, J.S. Park, H.S. Kim, H.W. Kim, M.J. Youn "Simple Switch control, digital signal processing, and ac drive control technologies.
Open Fault Detection Method of Voltage Source Inverter", ECCE 2009,
pp.:3175-3181, November 2009 Salvatore De Caro was born in Frankfurt am Main, Germany. He received
[5] O.S Yu, N.J. Park, D.S. Hyun "A Novel Fault Detection Scheme for the M.S. degree in electronic engineering from the University of Messina,
Voltage Fed PWM Inverter” IECON 2006, pp.:2654-2659, November Messina, Italy, in 2000, and the Ph.D. degree in electrical engineering from
2006 the University of Catania, Catania, Italy, in 2004. He is currently a
researcher in the area of power electronics at the University of Messina. His
[6] S. Bolognani, M. Zordan, M. Zigliotto, "Experimental fault-tolerant major research interests include power converters, power devices, and EMI.
control of a PMSM drive", Trans. Industrial Electronics, vol. 47,
pp.:1134-1141, October 2000. Antonio Testa was born in Catania, Italy, in 1962. He received the Degree
[7] A. Consoli, A. Gaeta, G. Scarcella, G. Scelba, A. Testa, "HF injection- in electrical engineering from the University of Catania, Catania, Italy, in
based sensorless technique for fault-tolerant IPMSM drives", ECCE 1988. During 1989–1990, he was the recipient of an SGS-Thomson Grant.
2010, pp.:3131 – 3138, October 2010. From 1990 to 1998, he was an Assistant Professor of Electrical Drives and
[8] F. Blaabjerg, J.K. Pedersen, "A new low-cost, fully fault-protected Power Electronics in the Department of Electrical, Electronic, and Systems
PWM-VSI inverter with true phase-current information", IEEE Trans. Engineering, University of Catania. He is currently a Professor of Electric
Power Electronics, vol. 12, pp.: 187-197, Jan. 1997. Machines and Power Electronics at the University of Messina, Messina,
Italy. During 1991, he was a Visiting Researcher at the University of
[9] F. Blaabjerg, J.K. Pedersen, U. Jaeger, P. Thoegersen, "Single current Wisconsin, Madison. His major research interests are in the fields of
sensor technique in the DC link of three-phase PWM-VS inverters: a sensorless control of electrical drives, advanced digital control systems,
review and a novel solution", IEEE Trans. Industry Applications, vol. power converters, power devices, and electromagnetic compatibility.
33, pp.:1241-1253, Sept.-Oct. 1997.

419

You might also like