Internal Short Circuit Fault Diagnosis for PM Machines Using FE-Based Phase Variable Model and Wavelets Analysis

This paper presents a procedure using the finite-element (FE)-based phase variable model combined with wavelet analysis to facilitate the fault diagnostic study for permanent magnet machines with internal short circuit faults. Our efforts are dedicated to the aspects of fault modeling and fault extraction. The FE-based phase variable model is developed to describe the PM machine with internal short circuit faults. This model is built with the parameters [inductances and back Electromotive Force (EMF)] obtained from FE computations of the machine with the same type of fault. The developed model has two features. It includes the detailed information of the fault including the location of the shorted turns and the number of turns involved. It keeps the accuracy of the FE model by taking only a fraction of time needed by FE operation. This is particularly desired for diagnosing faults in machines connected to a control circuit. The wavelet transform is used to perform machine current/voltage signature analysis. Excellent results were obtained providing information that would not be otherwise available except by measurement