Inter-Turn Fault Detection of PMSM Based on Fuzzy Logic and Discrete Wavelet Transform Using Unsupervised Clustering

Abstract

<em>The idea of this paper is designing an automatic fault detection system based on fuzzy logic, therefore two signals of PMSM in fault condition are analyzed for inter turn fault detection: current and torque. In this fault type there is some distortion in these signals, but it is not good enough to detecting with fuzzy logic solely, so with combination of wavelet transform and FCM a new method for fault detection is introduced. In this method one detail signal of wavelet transform is chosen and then with FCM it is divided into 6 clusters, these clusters describe the situation of signal truly. Using FCM has two advantages: first in some clusters there had been fault therefore fault was detected, and second it is used for fuzzy logic system to deciding amount and intensity of fault of PMSM. By applying combination of wavelet transform and FCM, designing of fuzzy logic has been more effective, the MFs are directly come from output of FCM, therefore fuzzy logic system have more accurate answer. The output of fuzzy logic that is showed in surface view is based on tree situation that is defined in output MF, and describes whole conditions of PMSM and shows the amount of inter turn fault. </em>