Evaluation of Wavelet-Functions for Broken Rotor Bar Detection of Induction Machine Using Coefficient-Related Features

Early fault detection of the induction machine is necessary in order to guarantee its stable and high performance. To evaluate the motor's health and detect existence of any failure in it, any motor parameter is first measured using condition monitoring techniques. The raw signal acquired is then interpret applying signal processing and data analysis procedures. Wavelet analysis of the motor current has been considered as an effective fault detection method. However, there are different types of the wavelet function that can be used for signal decomposition. This paper intends to investigate the ability of different types of wavelet functions for early broken rotor bar detection. Different harmonic components introduced by this fault such as maximum wavelet coefficient, left and right gradients of the maximum coefficient, were extracted and used as a characteristic signature for fault detection. The results indicate that the reliability of the fault detection depends on the type of wavelet function applied for decomposition of the signal.

induction machine, fault, broken rotor bar, current signature, signal processing, wavelet analysis