Study On Reliability Of Power Converter For Switched Reluctance Motor System

Switched reluctance motor system can show excellent performance in some harsh operating conditions.With the increasing problem of system stability,its reliability evaluation has gradually become an important part of system-level reliability research.In this paper,the focus is placed on reliability analysis of power converter that is the weakest part and can greatly affect the overall reliability of system.Firstly,the power loss of converter is accurately calculated with the help of Simulink and LTspice,especially for switching power loss of MOSFET and reverse recovery loss of diode.Afterwards,the thermal coupling effects and dissipating boundary conditions are analyzed in 3-D FEM model.According to the step power response extraction,a 3-D compact RC network model can be built for junction temperature estimation with the coupling impedances and sub-heatsink impedances in series.Based on the estimated junction temperature,the failure rate of power device can be obtained by Stress Analysis method,which lays foundation for quantitative evaluation of reliability.Finally,in order to determine the criterion of system failure,the fault mode analysis is carried out to bulid Fault Tree model,and thereby,the Bayesian Network model of switched reluctance motor system can be transformed form Fault Tree model.According to forward inference of Bayesian Network,the probability of system failure can be calculated by prior probability,and in contrast,the cause of system failure can be diagnosed in reverse inference.Consequently,the evaluation results of Bayesian Network for switched reluctance motor system is similar to Markov model method,which reveal that Bayesian Network evaluation method is an accurate reliability evalution method with low time-consuming and simple inference process.