In Situ Parameter Identification Method for Thermal Network Models of Power Modules

Thermal network models provide an indirect and noninvasive method for online junction temperature estimation of power semiconductor devices. Currently, thermal network model parameters are mainly extracted from finite-element models or calculated from material property parameters and geometries. However, as the power module ages, the parameters of the thermal network model change, resulting in inaccurate junction temperatures estimated by the preset thermal network model. This article proposes a thermal network model for long-term junction temperature monitoring of power modules, taking into account cooling conditions and thermal coupling between chips. By measuring the junction temperature cooling curves in the standby state existing in the operating conditions and updating the parameters of the thermal network model using the identification results of the artificial intelligence algorithm, the synchronization between the thermal network model and the health state of the power module is achieved, and the aging degree and aging location of the power module are monitored. The algorithm, implementation, and result analysis of the proposed method are presented in this article. Experimental verification is given to prove the effectiveness, accuracy, and convenience of the proposed method.