Life prediction methods for SiC MOSFETS

As power electronics technology advances toward high frequency and power density, silicon carbide (SiC) power devices are emerging as a promising replacement for traditional silicon devices due to their superior performance characteristics. The widespread adoption of SiC devices in power electronic systems has created an urgent need for their health management. However, research on lifetime prediction methods for SiC devices remains in its nascent stage and demands substantial time and datasets for accurate forecasting. This paper establishes a series-connected SiC MOSFET power cycling test platform that enables simultaneous aging of multiple groups of devices under test, thereby reducing the overall aging time. By integrating the CNN-BiLSTM-Attention lifetime prediction model and the exponential fitting method, this study investigates lifetime prediction methodologies for SiC MOSFETs.