A Fault Diagnosis Method with Residual Generator Selection for Permanent Magnet Synchronous Machines

Structural analysis for fault detection and isolation (FDI) is a powerful technique for designing structured residual generators in model-based diagnostic systems. However, in complex systems, the method often generates a large number of diagnostic test candidates. This paper presents a method for selecting residual generators from a potentially extensive set of candidate equations, focusing on optimizing the performance of the selected residual equations in terms of accuracy. The methodology is utilized for permanent magnet synchronous machine (PMSM) drive system fault diagnosis in the application of electric vehicles, demonstrating its effectiveness in automatically selecting the optimal residual generators and the associated residuals for diagnostic test implementation, even when starting with a large number of options generated by structural analysis algorithms.