A data-driven approach for IGBT aging degree evaluation based on multi-observation sequence particle filtering and support vector regression

Insulated gate bipolar transistors (IGBTs) are capable of efficiently and stably converting and regulating electrical power. Precise evaluation of the aging degree of IGBTs is particularly important. This study proposes a data-driven approach for IGBT aging degree evaluation based on multi-observation sequence particle filtering and support vector regression. This method effectively integrates the aging data from different devices, significantly reducing data uncertainty, and constructs an IGBT aging degree evaluation model based on a small amount of data. A series of experiments has verified the effectiveness of this method, demonstrating its high accuracy and reliability.