Remaining Useful Life Prediction of IGBT Module Based on Particle Filter Combining with Particle Swarm Optimization

Abstract

A data-driven lifetime prediction is proposed and implemented on the power module in this paper. Insulated gate bipolar transistors (IGBTs) are widely used in various power electronic converter systems. The IGBT modules suffering failure may influence the reliability of the power systems enormously. Thus, it is significate to accurately predict the remaining useful life (RUL) of this critical component. Based on the wide-used particle filter (PF) prediction algorithm, the particle swarm optimization (PSO) is combined to optimize the step of sequential important resampling in PF and solve the particle impoverishment problem. In addition, a power cycling test is designed, which is conducted to obtain the degradation data under specified operating stress. The method in this paper can effectively process the experimental results under power cycling tests.