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]

Chip fatigue inside an insulated gate bipolar transistor (IGBT) module is a kind of incipient defect. It can be considered as an indication of the impending failure, and is utmost important for the safe operation of IGBT modules. Therefore, monitoring the chip fatigue is one of crucial measures to enhance the operating reliability of IGBT modules. This paper presents a prognostic approach for the chip fatigue based on grey relational analysis (GRA), which uses dynamic change of the gate voltage as precursor parameter. This dynamic change is caused by aging of the intrinsic parasitic elements involved in gate drive circuit, which reflect the advent of chip fatigue. Grey relational grade is employed in this proposed prognostic approach to quantify the extent of those dynamic changes by little data, and find out potential chip fatigue. Then the operator would have a chance to schedule the maintenance and replace defective IGBT modules timely to avoid wear out. So it can be seen as a prefault diagnostic method. Finally, a confirmatory experiment is also carried out, and the correctness of the proposed approach is verified. [ABSTRACT FROM AUTHOR]