Determination of number of broken rotor bars in squirrel-cage induction motors using wavelet, PCA and neural networks

Abstract

For determination the number of broken rotor bars in squirrel-cage induction motors when these motors are working, this paper presents a new method based on an intelligent processing of the stator transient starting current. In light load condition, distinguishing between safe and faulty rotors is difficult, because the characteristic frequencies of rotor with broken bars are very close to the fundamental component and their amplitudes are small in comparison. To overcome this problem, an advanced technique based on the wavelet transform and artificial neural network is suggested for processing the starting current of induction motors. In order to increase the efficiency of the proposed method, the results of the wavelet analysis, before applying to the neural networks are processed by Principal Component Analysis (PCA). Then the outcome results are supposed as neural network's training and testing data set. The trained neural networks undertake of determining the number of broken rotor bars. The given statistical results, announce the proposed method's high ability to determine the number of broken rotor bars. The proposed method is independent from loading conditions of machine and it is useable even when the motor is unloaded. © 2009 Praise Worthy Prize S.r.l. - All rights reserved.