A Novel Multi-Stage Ensemble Method for Online Junction Tempera-ture Estimation of SiC MOSFETs

SiC MOSFETs have been identified in numerous studies as having a significant risk of thermal failure. Consequently, temperature monitoring has become a crucial component of online condition monitoring for SiC MOSFETs. Currently, temperature-sensitive electrical parameters (TSEP) are commonly used for real-time monitoring of the junction temperature of single SiC chips. However, existing TSEPs face two primary challenges: low resolution, and difficulties on sensor acquiring posed by the high switching speed of SiC MOSFETs. Additionally, the high switching speed also induces electromagnetic interference (EMI), which adversely affects the accuracy of TSEP measurements. This study proposed a novel temperature prediction method, using the period between the rising edge of both the gate PWM and drain current. Experimental results show high sensitivity and strong interference immunity. The approach also features a simple, cost-effective circuit design and broad compatibility, allowing seamless integration with commercial gate drivers, thereby enhancing the intelligence of gate driver systems.