Case Temperature Monitoring-Based Online Condition Monitoring of SiC MOSFET Power Modules using a Radial Basis Function Network

Abstract

Silicon carbide (SiC) metal-oxide-semiconductor field-effect transistor (MOSFET) power modules are being used for high power applications because of their superior thermal characteristics and high blocking voltage capabilities over traditional silicon power modules. This paper explores monitoring the temperature distribution of the baseplate of an SiC MOSFET power module for online condition monitoring of the power module. A radial basis neural network (RBFN) is trained to follow the operational temperature data of a healthy power module. As a module deteriorates the temperature distribution changes as well. Comparing the trained RBFN output and an unhealthy module’s temperature output at the same point, the differences in temperature signify deterioration in the health of the module. The proposed method of online condition monitoring is applied to an SiC MOSFET power module and validated by computer simulations using finite element analysis models for the power module in both healthy and unhealthy conditions.