Unknown Input Observer-Based Fault Diagnosis in Induction Machine: Theory and Experiment

摘要

This paper presents an observer-based approach to diagnosis and detect inter-turn short-circuit fault in induction machine based on residual generation. Residual signals are generated by taking measured three-phase stator currents and those reconstructed by unknown input observer. By comparing these residuals to a predefined threshold, such fault can be detected and fault signatures are obtained. These signatures are used to develop a faulty phase isolation algorithm. The ability to detect the fault and isolate the faulty phase is tested by simulation and validated using experimental test bench with real-time interface based on the dSpace environment.