A hybrid fuzzyeural system used to extract heuristic knowledgefrom a fault detection problem

摘要

Neural net have proven to be capable of solving the motormonitoring and fault detection problem using an inexpensive, reliableand noninvasive procedure. The neural net, unfortunately, cannot provideheuristic knowledge about the motor or the fault detection process. Thispaper introduces a novel hybrid fuzzyeural fault detector that usesthe learning capabilities of the neural net to detect if a motor has anincipient fault. Once the fuzzyeural fault detector is trained,heuristic knowledge about the motor and the fault detection process canalso be extracted. With better understanding of the heuristics throughthe use of fuzzy rules and fuzzy membership functions, we can have abetter understanding of the fault detection process of the system; thuswe can design better motor protection systems. The electric motors inindustry are exposed to a wide variety of environments and conditions.These factors, coupled with the natural aging process of any machine,make the motor subject to incipient faults. These incipient faults, leftundetected, contribute to the degradation and eventual failure of themotors. With proper monitoring and fault detection schemes, theincipient faults can be detected; thus maintenance and down-timeexpenses can be reduced while also improving safety. In this paper,motor bearing faults in single-phase induction motors are used toillustrate this novel system. This illustration demonstrates thesuccessful training of a hybrid fuzzyeural system that can provideaccurate fault detection, and gives the heuristic reasoning for thefault detection procedure