Analysis of Acoustic Emission Signals From SiC MOSFETs Using Robust Local Mean Decomposition and Generalized S-Transform

Power modules are the most vulnerable components in power electronic systems. It is essential to conduct fault diagnosis and implement protections on power modules to prevent further damage to the system. Existing methods primarily identify faults in power modules through electrical, magnetic, and junction temperature parameters. However, research on the application of acoustic emission (AE) method to monitor power modules is still limited. This article proposes a scheme that leverages the intrinsic relationship between AE signals and the switching states of SiC MOSFETs. The robust local mean decomposition (RLMD) and the generalized S-transform (GST) are introduced. The feasibility and superiority of these methods are initially validated through numerical simulations. Experimental investigations are conducted by a double pulse platform for SiC MOSFETs. The model between time-frequency characteristics of AE signals and external electrical parameters is established under normal conditions. By comparing the Rényi entropy of AE signals from SiC MOSFETs under normal and faulty conditions, which differ by approximately three times, it is demonstrated that Rényi entropy can be used to identify faults.