PHM Framework for Discrete IGBTs in EV Applications: A PoF-Based Approach with a State-of-Damage Metric

Abstract

Power converters are increasingly required to deliver higher power density and longer service life, which increases the need for accurate remaining useful life (RuL) prediction of their switching devices. Although module-type switching devices dominate many applications, discrete insulated-gate bipolar transistors (IGBTs) remain commercially important and are deployed in traction inverters, making robust lifetime prediction models (LPMs) for discrete devices directly relevant to prognostics and health management (PHM).

This paper presents a Physics-of-Failure (PoF)-based LPM for discrete IGBTs that models die-attach solder degradation as the dominant wear-out mechanism. A novel State-of-Damage ( ) framework is further proposed to unify high- and low-cycle fatigue degradation within a single state variable, enabling damage accumulation under mixed-stress mission profiles. The proposed LPM and  metric are experimentally validated through power cycling tests.

The proposed approach provides a practical alternative to data-driven methods, which typically require extensive failure datasets and additional sensing or data acquisition hardware that are costly to obtain and often unavailable in industrial environments. By grounding RuL estimation in the dominant degradation mechanism, the framework remains computationally efficient and data-lean while enabling reliable health assessment and RuL prediction for discrete IGBT-based power electronic systems.