Zastosowanie algorytmu klasteringu rozmytego w optymalizacji neuronowo-rozmytych detektorów uszkodzeń wirnika silnika indukcyjnego

In this article neuro-fuzzy detectors of faults in induction machine supplied by power converter are presented. Proposed models are implemented for analysis of symptoms of damages in squirrel cage motor. Selection of the center location of each membership function at the initial stage of processing is a problematic task in the design process of presented detectors. For optimization of the structures of the neuro-fuzzy detectors fuzzy clustering algorithm is implemented. Presented results of researches show high precision of detection of asymmetry of the rotor cage for changeable supply frequency and load of the drive. In addition the number of broken bars is also very precisely determined.