Search

Potential defects inside an insulated gate bipolar transistor (IGBT) module can be seen as precursors of being about to wear out, and are important for health monitoring of the IGBT module which is used to enhance reliability. A method for detecting defects in the IGBT module by identifying dynamic changes of the gate current is presented in this paper. It is based on the fact that parasitic elements inside the IGBT module are affected by local damage induced by aging over time, and subsequently, these influences are easily distinguished by dynamic changes of the gate current. The proposed prognostic approach is based on relevance vector machine framework, which can give accurate predictions and allow operators to observe unhealthy IGBT modules inside a power converter, and then take appropriate measures timely to avoid breakdown. The parasitic elements involved in the gate circuit are extracted, and their contributions to dynamic changes of the gate current are also discussed. Finally, a confirmatory experiment is carried out, and the correctness of the proposed method is verified. [ABSTRACT FROM AUTHOR]