Investigation on IGBT Failure Effects of EV/HEV Inverter Using Fault Insertion HiL Testing

Size: px

Start display at page:

Download "Investigation on IGBT Failure Effects of EV/HEV Inverter Using Fault Insertion HiL Testing"

Dorthy Holmes
6 years ago
Views:

1 EVS28 KINTEX, Kore, My 3-6, 25 Investigtion on IGBT Filure Effects of EV/EV Inverter Using Fult Insertion il Testing Chun-Ying Lin, Ping-Lun Li, Cheng-o Li, Wen-Shu Ching Industril Technology Reserch Institute, 95, Sec. 4, Chung sing Rd., Chutung, sinchu, 34, Tiwn, R.O.C., Abstrct This pper presents the investigtion of IGBT filure effects of electric vehicle/hybrid electric vehicle (EV/EV) inverter with rdwre-in-the-loop (IL) testing. The testing system model is creted with Automotive Simultion Models (ASM) toolbox nd the micro controller unit (MCU) is connected to dspace test bench. Bsed on Filure-Mode-nd-Effects-Anlysis (FMEA), the filure model of IGBT hs been creted nd injected to the system model. Through the rel-time simultion nd testing, the control strtegy nd defect of MCU hs been identified. Keywords: rdwre-in-the-loop (IL), Filure-Mode-nd-Effects-Anlysis (FMEA), fult insertion, Electric Vehicle Introduction ue to the vigorous development of electric vehicle in recent yers, the complexity of electronics nd controllers re incresing. The rdwre-in-the-loop testing is technique tht uses mthemticl models to represent the physicl dynmic of full system nd connect the models with the hrdwre component such s micro controller to test the function nd control strtegy. There re mny prts of cr cn be replced nd the demnd of the prototype vehicles cn be reduced with IL testing. Especilly in erly design stge, the required time, cost, nd effort cn be reduced sufficiently. Therefore, the demnd for test nd vlidtion with IL is widely incresing []-[2]. The electronics ply n importnt role in motor drives nd converters for electric nd hybrid vehicle. Therefore, the relibility of electronics is very importnt nd the issue how electronics fult could dmge or endnger the propulsion system to hzrd driver nd pssenger is the top priority to be considered in the design nd vlidtion stge. owever, the testing of electronics fult with rel components is dngerous nd costly [3]. This reserch ims to develop n electric propulsion system model for electric or hybrid vehicle system nd crete fult model to simulte IGBT filure. The models nd connected I/O port re equipped with interntionl vehiculr electronics tier one supplier level instrument. The vlidtion nd testing process includes Models In the Loop (MIL), Softwre In the Loop (SIL), nd IL to prove the stbility nd relibility of the controller. The dspace test bench ws used to crete electric vehicle propulsion system with rel-time models, such s electric motor, inverter, nd resolver, etc. The models of inverter nd electric motor nd the MCU re bsed on the reserch nd development result from Industril Technology Reserch Institute (ITRI) of Tiwn which is investigted for Commercil Plug-in Electric Vehicle (CPEV). The system model nd fult models re built with ASM [4] nd Simulink/SimPower Systems [5]. This pper presents the design nd implementtion for the IGBTs filure effect on EV/EV with IL testing pltform. This pper is orgnized s follows. Section II provides the theory to crete system model which cn be used to verify the control strtegy nd EVS28 Interntionl Electric Vehicle Symposium nd Exhibition

hrdwre/softwre integrtion for the MCU. Section III derives the fult model from the FMEA tble nd injects the fult model to system model. Section IV process the IL testing nd shows the result.

The IL test bench is shown s Fig. 2. MCU nd ASM hve been connected through S 522 I/O bord. The rel-time simultion between MCU nd models hs ccomplished by S 6 processor bord.

The dspace ASM is simultion tool suite tht provides model-bsed function development environment for vehicle dynmic simultion.

Model of MCU The MCU model cn be decomposed s speed closed-loop controller, q-xis current closed-loop controller, d-xis current closed-loop controller, d- xis voltge decoupler, nd q-xis voltge

2 hrdwre/softwre integrtion for the MCU. Section III derives the fult model from the FMEA tble nd injects the fult model to system model. Section IV process the IL testing nd shows the result. Figure : Overview of Motor Control System for IL gte driver, motor, protection circuit, nd the sensors re creted s virtul prt within the dsh squre line of Fig. nd relized with ASM on PC. The IL test bench is shown s Fig. 2. MCU nd ASM hve been connected through S 522 I/O bord. The rel-time simultion between MCU nd models hs ccomplished by S 6 processor bord. The I/O bord, processor bord, nd CAN interfce bord re integrted to dsapce control box. The dspace ASM is simultion tool suite tht provides model-bsed function development environment for vehicle dynmic simultion. The open Simulink models cn dd dditionl components esily to mke the entire system more flexible to implement fult injection Fig. 3 shows ASM model desk control pnel. 2. Model of MCU The MCU model cn be decomposed s speed closed-loop controller, q-xis current closed-loop controller, d-xis current closed-loop controller, d- xis voltge decoupler, nd q-xis voltge decoupler. The flux-wekening control model cn be dded utomticlly depends on the motor speed nd control strtegy. The MCU control model frme is shown s Fig. 4. Figure 2: igrm of rdwre with Test Bench Figure 4: MCU control model frme Figure 3: Model esk Control Pnel 2 Rel-Time Simultion models nd implementtion In this section, the models of motor control system for electric vehicle re presented. As Fig. shows, the micro control unit (MCU) bord is the only one rel hrdwre in the motor control system for IL test. The other components of motor control system such s bttery, inverter, Proportionl-Integrl (PI) controllers re used for relize input commnd to speed nd d/q-xis current control. The speed nd current commnd is sent to voltge decoupler to simplify the current control loop eqution. Then, the output commnd of voltge decoupler trnsfers to torque commnd nd chieve motor control strtegy. The closed-loop speed control cn be expressed: i q K p iq K p Ki Ts () z Where K p is the proportionl gin, K i is the integrl gin, Δω is the difference of speed commnd EVS28 Interntionl Electric Vehicle Symposium nd Exhibition 2

3 minus speed feedbck vlue, T s is step size of smpling, i d is d-xis feedbck current, nd i q is q-xis feedbck current. The function of d-xis current closed-loop controller cn be written s: do K p id Vd K p id Ki id Ts (2) z Where Δi d is the difference of commnd vlue minus feedbck vlue of d-xis current, V do is d- xis output voltge, nd V d is d-xis voltge. The function of q-xis current closed-loop controller cn be written s: qo K p iq Vq K p iq Ki iq Ts (3) z Where Δi q is the difference of commnd vlue minus feedbck vlue of q-xis current, V qo is q- xis output voltge, nd V q is q-xis voltge. The d-xis voltge decoupler cn be express s: V d i L V (4) d e q do Where ω e is electricl rottion speed of PMSM. The q-xis voltge decoupler cn be express s: V q i L (5) q e d qo 2.2 Model of PMSM The motor model consists of d/q-xis current, bck-emf, nd torque clcultion models. These models re co-simulted with the inductnce model which cn be switched between liner nd non-liner model. By input the inductnce nd flux linkge with different motor current, the look-up tble hs been completed for the nonliner inductnce model. With the non-liner inductnce tble, the motor model cn be more ccurte. The motor models re illustrted in Fig. 5. The specifiction of testing motor is shown s Tble. The prmeters re used to build the motor model. The bck-emf nd torque of PMSM cn be written s: Vqemf (6) r si L d Lq i diq iq si 3 Ppole (7) 2 Where V qemf is bck-emf, ω r is motor rottion speed, nd λ si is rotor flux linkge, P pole is rotor pole pir, L d is d-xis inductnce, L q is q-xis inductnce, nd T rq is output torque. Figure5: The clcultion of motor model frme Tble: Specifiction of testing motor Prmeter escription Vlue V C C voltge provided by bttery 44 V PWM Frequency of the sine wve 2 kz L q Inductnce of sttor (q).525 m L d Inductnce of sttor (d).387 m R s Resistnce of sttor.46 ohm P Pole pir of rotor 6 λ si Flux linkge of rotor.65 Wb The equivlent circuit of motor includes inductnce, resistnce, nd bck-emf. Prk trnsform hs been used for convert three phse current to d/q-xis form current. Two xis current is shown s below: riqlq d Rsid Ts id I d (8) L d q rid Ld r si Rsiq Ts iq I q (9) L Where R s is sttor resistnce, I d is d-xis output current, I q is q-xis output current nd T s is step size of smpling time. 2.3 Model of IGBT The six IGBTs nd freewheeling diodes in inverter model re shown s Figure 6. The phse current nd voltge estimtion of the model will be introduced. The model receives the feedbck signl from current sensor nd identifies s positive, negtive, nd zero-cross point current. As the current stte hs been confirmed, the output function of voltge, current, nd phse loss cn be expressed s eqution () to (2). q EVS28 Interntionl Electric Vehicle Symposium nd Exhibition 3

Figure6: Inverter model frme Tke phse A for exmple, the negtive current cn be written s: Vdiode Z diode i ROR VA i i A i ROR loss () Where is high side duty cycle, L is low side duty cycle, Z is equl

4 Figure6: Inverter model frme Tke phse A for exmple, the negtive current cn be written s: Vdiode Z diode i ROR VA i i A i ROR loss () Where is high side duty cycle, L is low side duty cycle, Z is equl to (- - L ), V is bttery voltge, V diode is diode drop voltge, R OR is stte of switch resistnce, i is the input phse current, i A is the consumption prt of phse current, V loss is loss of phse voltge nd V A is phse output voltge. The zero-cross point current cn be expressed s eqution (). In this eqution, if V is greter thn V + V diode, then V equls to V + V diode. If V is smller thn - V diode, then V equls to -V diode. If V is locte between V + V diode nd -V diode, then V remins its vlue. V i A V diode V diode V () loss The positive current cn be written s: V i V L i A OR diode V loss Z diode i R OR V i R V (2) 3 FMEA nd Fult injection unit FMEA is step-by-step pproch for collecting knowledge bout potentil filure in the design level of production. Then the methods to reduce or void the occurrence of filure will be proposed. The probbility of fult under wht kind of condition needs to be investigted from top to down in system level. A 3. Construction of FMEA The fult severity (SEV), occurrence (OCC) nd detection (EC) re used to determine the weight nd give number of risk priority number (RPN) to follow the order. The construction of filure scheme cn be clssified s: () Fult injection by hrdwre interfce; (2) Fult injection by MCU model or circuit; (3) Fult injection in rel-time model. For IGBT filure, the fult injection will be pplied with rel-time model. The initil condition of vehicle behviour hs been defined s before strting (speed: rpm, lod torque: Nm) nd during driving (speed: 3 rpm, lod torque: 9 Nm). The fult lso cn be clssified s temporry filure nd continul filure. Tble 2 shows the FMEA of IGBT filure. Tble 2: FMEA of IGBT filure Fult Period Sitution Test Tsk open circuit short circuit unblnce voltge Temp. Cont. Temp. Conti. Temp. Conti. Before strt: Vehicle strnded Before strt: Vehicle strnded Before strt: Vehicle strnded Phse -side Phse L-side Phse -side Phse -side switch open Phse -side Phse L-side Phse -side short Phse L-side short Phse short Line to line short Phse -side short Phse L-side short Phse short Line to line short Phse -side short Phse L-side short Phse short Line to line short Phse voltge unblnce Phse voltge unblnce Phse voltge unblnce 3.2 Models of Fult Injection In this section, constructing fult injection models re developed bse on FMEA tble. For IGBT EVS28 Interntionl Electric Vehicle Symposium nd Exhibition 4

components, fult injection in inverter could cuse open circuit, short circuit, nd unblnce voltge supply to phse voltge tht re considered s criticl dmge.

ssumption or possible relity. To verify the phenomenon, some electricity modelling softwre such s PLECS or SimPowerSystem is used to vlidte the fult model.

described. Fig. 7 shows control system nd fult in fult injection models.

8 shows some content inside the IGBT fult injection model. Figure 7: control system nd fult injection models controller is connected directly to physicl model of the system within Simulink digrm.

5 components, fult injection in inverter could cuse open circuit, short circuit, nd unblnce voltge supply to phse voltge tht re considered s criticl dmge. The most difficult prt in the progress of fult injection models developing is to identify fult of electronic components relted to open circuit or short circuit is the phenomenon occurred in simulted ssumption or possible relity. To verify the phenomenon, some electricity modelling softwre such s PLECS or SimPowerSystem is used to vlidte the fult model. In the reference [7] nd [8], the IGBT in open circuit, short circuit nd phse short circuit condition is expressed by using mthemticl lgorithm nd the fult phenomenon of electric component hs been described. Fig. 7 shows control system nd fult in fult injection models. The red block models re represent s fult injection switch models nd blue block models re used for connect to digitl nd nlog signl hrdwre pin with MCU. Fig. 8 shows some content inside the IGBT fult injection model. Figure 7: control system nd fult injection models controller is connected directly to physicl model of the system within Simulink digrm. The controller model cn be modified immeditely nd speed up development t this stge. SIL is kind of coding system testing. The model need to been coded into C or C++ nd inserted the coded model bck to the system model. After finish SIL testing, the IL test cn crry on. Fig. 9 shows the control desk window. The control desk cn switch the testing between SIL nd IL. Figure9: control desk window 4 Experimentl results of IL The IL test bench is shown in Fig.. In this section, the IGBT fult injection IL experimentl results re presented. Fig. shows the motor rottion speed chnge of phse A high-side IGBT continuously. Fig. 2 shows the motor torque vrition nd Fig. 3 is the three phse current distribution. The fult hs been injected t.5 sec. The dsh line in the figure mens the trigger stte of fult. The rted speed of test motor is 3 rpm. The result shows the motor is unstble fter fult inject. Controller tkes.28 sec to discover the fult nd shut down the switch. In Fig. 2 nd 3, the switches for ll phse re turned-off t.285 sec. After fult inject, only the negtive vlue of phse A current cn be seen in Fig.2 which mens high-side switch is opened circuit nd ll current turn to zero which results zero output torque fter.285 sec. Figure 8: IGBT fult injection model 3.3 MIL nd SIL testing Before integrte the hrdwre nd softwre component of MCU, the softwre cn be test nd verified with MIL nd SIL. For the MIL testing, we using model of the controller to work with model of the hrdwre. The model of the Figure : IL test bench EVS28 Interntionl Electric Vehicle Symposium nd Exhibition 5

6 I (A) (Nm) (Nm) been triggered, the controller cnnot fford stble torque to fit the lod torque Figure : Motor speed of phse A high-side switch open continuously Figure 4: Motor speed of phse A high-side switch open before strt Figure 2: Motor output torque of phse A high-side continuously I A -2 I B I C Figure 3: Three phse current of phse A high-side continuously Fig. 4 nd 5 show the motor speed nd torque of phse A before strt respectively. Before the fult inject, the negtive torque in Fig. 4 represents the lod torque. After fult hs Figure 5: Motor torque of phse A high-side switch open before strt Fig. 6 to 8 illustrte the phenomenon of temporry open of phse A low side switch. The temporry time lst for.7 ms. The minimum signl time resolution of S 6 processor bord is.7 ms. Consider the processor clcultion cpbility, temporry time period hs been chosen s.7 ms. The controller did not detect the temporry filure. Fig. 7 shows the torque drop when fult hppen. In the menwhile, the three phse current mplitude lso decresed in Fig. 8. ue to the closed-loop control, the controller tried to compenste the torque, so the current increse fter the end of fult injection. The most criticl fult of IGBT is the short circuit of switch component. It will let the vehicle lose the motive nd let driver nd pssengers been exposed to dnger, especilly in high speed driving environment. Fig. 9 to 22 illustrtes tht short circuit hppen between high side nd low side switch. Becuse of the inerti of motor, the speed EVS28 Interntionl Electric Vehicle Symposium nd Exhibition 6

7 I (A) I (A) (Nm) (Nm) of motor did not drop very fst in Fig. 9. owever, the output torque strts to oscillte quickly nd hs very huge torque ripple in Fig. 2. The controller detects the fult t.2 sec nd set the switch duty to zero to stop motor opertion Figure 6: Motor speed of phse A low-side switch temporry open Figure 9: Motor speed of phse A switch short continuously Figure 7: Motor torque of phse A low-side switch open temporrily Figure 2: Motor torque of phse A switch short continuously I A I B I C I A I B I C Figure 2: Three phse current of phse A switch short continuously Figure 8: Three phse current of phse A low-side temporrily EVS28 Interntionl Electric Vehicle Symposium nd Exhibition 7

8 (Nm) uty cycle (%) I (A) uty A ighside uty B ighside uty C ighside Figure 22: Three phse current switch duty of phse A switch short continuously To test the unblnced voltge for IGBT, the voltge supply for phse A hs been decresed for %. The results cn be seen in Fig. 23 to Fig. 25. Apprently, the motor speed did not chnge lot, but the torque ripple increse significntly. For the % single unblnce phse voltge supply, the motor torque cn remin stble Figure 23: Motor speed of provide unblnce voltge supply for phse A continuously Figure 24: Motor torque of provide unblnce voltge supply for phse A continuously Figure 25: Three phse current of provide unblnce voltge supply for phse A continuously 5 Conclusion This pper presented the design, implementtion, nd vlidtion for the IGBT filure effect on the EV/EV propulsion system with IL test bench. First, the propulsion system model hs been completed, nd then creted fult models bsed on the FMEA tble. Finlly, the fult injection to the system model with MCU control bord hs been implemented. The protection strtegy nd softwre/hrdwre integrtion hs been verified with rel-time monitored control interfce of IL test bench. The rel-time behviour cn be recorded nd used to finds the defect of MCU. It provides n efficient nd economicl testing to improve vehicle control system to mke it more stble nd relible. References I A I B I C [] S. Krimi, P. Poure, nd S. Sdte, An IL- Bsed Reconfigurble Pltform for esign, Implementtion, nd Verifiction of Electricl System igitl Controllers, IEEE Trns. on Industril Electronics, Vol. 57, No. 4, pp , April, 2. [2] J. J. Poon, M. A. Kinsy, N. A. Pllo, S. evds, nd I. L. Celnovic, rdwre-in-the-loop Testing for Electric Vehicle rive Applictions, Applied Power Electronics Conference nd Exposition (APEC), pp Feb. 2. [3] R. Wu, F. Blbjerg,. Wng, nd M. Liserre, Ctstrophic filure nd fult-tolernt design of IGBT electronic converters - n overview, Industril Electronics Society, IECON, pp Nov. 23. [4] ASM Vehicle ynmics, dspace Gmb. 22. EVS28 Interntionl Electric Vehicle Symposium nd Exhibition 8

9 [5] SimPowerSystems User's Guide R22b, The MthWorks, Inc. 22. [6] Controlesk Experiment Guide, dspace Gmb. 22. [7] M. Abul Msrur, Assumption or Fct? Line-to- Neutrl Voltge Expression in n Unblnced 3- Phse Circuit uring Inverter Switching, IEEE Trnsctions on Eduction, Vol. 52, No. 2, My 29. [8] S. Mllt, A Wvelet Tour off Signl Processing: The Sprse Wy, Acdemic Press, 3rd ed., 29. Mechnicl System reserch Lb in the Industril Technology Reserch Institute. is reserches re relevnt to chrging system for electric vehicle nd bttery swpping system for electric vehicle. Authors Chun-Ying Lin received his Ph. degree in the eprtment of Electricl nd Control Engineering, Ntionl Chio Tung University, sinchu, Tiwn, 22. e hs been resercher in the Industril Technology Reserch Institute since 22. is reserches re relevnt to virtul vehicle testing stge (IL & MIL) nd control strtegy for electric vehicle chrging process. Ping-Lun Li received his Ph. degree in the Ph.. Progrm of Electricl nd Communictions Engineering, Feng Chi University, Tichung, Tiwn, 23. e is currently resercher in the Industril Technology Reserch Institute. is min reserch domins re virtul vehicle testing stge (IL), electric mchine design, nd electromgnetic nlysis. Cheng-o Li received his Ph. degree in the eprtment of Power Mechnicl Engineering, Ntionl Tsing u University, sinchu, Tiwn, 28. e hs been resercher in the Industril Technology Reserch Institute since 28. is min reserch res re relevnt to electric vehicle motion control nd electric propulsion system IL test. Wen-Shu Ching received his Ph. degree in the eprtment of Mechnicl Engineering, Ntionl Chung Cheng University, Chiyi, Tiwn, 23. e is currently the division director of Intelligent Mobility Technology division of EVS28 Interntionl Electric Vehicle Symposium nd Exhibition 9

Design and Implementation of the Performance Test System for Battery-Powered Motorcycles

Design and Implementation of the Performance Test System for Battery-Powered Motorcycles Rhodes, Greece, August 0-, 008 Design nd Implementtion of the Performnce Test System for Bttery-Powered Motorcycles DING-TSAIR SU, YING-SHING SHIAO, JUI-LIANG YANG Deprtment of Electricl Engineering Ntionl