Reliability Assessment Methodology of Power Devices Based on Physics of Failure and Multivariate Copulas

The reliability assessment of power devices based on Physics of Failure (PoF) involves factors such as performance, structural materials, and operating conditions. However, PoF methods are mainly used for the specific failure mechanisms (FMs) of the device. Now a fusion multiple FMs reliability assessment is needed. This paper proposes a reliability assessment methodology considering the failure correlation of different structures of power devices based on multivariate copulas. We illustrate the assessment process by taking a FS discrete device as a case. Firstly, the FMs and PoF models of power devices are studied. The parameter values in PoF models are determined; The unknown parameters are obtained by finite element analysis (FEA). Secondly, the multivariate copulas are used to derive the reliability function with multi-FMs, and the unknown parameters in copulas are estimated. Finally, comparing the Monte Carlo-Competition (MCC) method, the results show that the proposed method can consider the cross¬influence factors between PoF models, and the correlation of different structural lifetime distributions is not negligible. Moreover, the computational efficiency is also improved.