Fault detection and identification for a dual-redundant brushless DC motor system using wavelet transform and hierarchical clustering algorithm

In order to implement the high reliability and wide-ranged speed control of an electro-mechanical actuator (EMA), a novel scheme was proposed to detect and identify faults of a dual-redundant brushless DC motor (BLDCM) system, which combines wavelet transform (WT) technique with hierarchical clustering algorithm (HCA) by testing the existing bus-current signals in each channel and using the features of the specific system structure and commutation process. Experimental results reveal that the motor open-phase faults and all kinds of inverter-transistor open-circuit faults of this BLDCM system can be correctly detected and identified with high robustness for the impact of wide-ranged motor speed, operating load, and even unexpected noise. This method is so sensitive to the sudden changes of the bus currents that it is very powerful for detecting those faults which lead to abnormal commutations. It needs no more extra devices, and is of low complexity, well fault-distinguish ability, high reliability and practical simplicity.