Adaptive diagnosis by pattern recognition: Application on an induction machine

摘要: In this paper, a pattern recognition method is used to provide the tracking and diagnosis of system. To illustrate it, we as application, an asynchronous motor 5.5 kW with squirrel-cage, in particular for detection broken bars, under any level load. From measurements carried out on system, parameters are calculated. These build up vector which considered system signature. determine vector, two methods applied. One, well-known, sequential backward selection (SBS) other, developed, based genetic approach, advantage optimal dimension representation space give better results (value criterion) than SBS. The determination decision using automatic classification called clustering. phase ldquok-nearest neighborsrdquo rule, associated evolution trajectory allowing not only states defined training set, but also intermediate states. appearance new operating mode taken into account order enrich initial knowledge base thus improve diagnosis.