Comparison of different wavelet families for broken bar detection in induction motors

In previous research works dealing with induction motor rotor assessment it has been shown that, when broken rotor bars exist, the analysis of the stator start-up current using the Discrete Wavelet Transform (DWT) leads to clearly recognizable patterns associated with the failure. This is valid for different operating conditions (loaded and unloaded machine, periodical fluctuation in the supply voltage, pulsating load torques...) and for distinct start-up modalities (direct start-up, star-delta start-up, soft-starter-operated motors). However, this fault-related pattern can significantly vary or even may not appear depending on the wavelet family used for the analysis. This paper discusses about some fundamental mathematical properties of wavelet families and how these properties influence the appearance of the characteristic pattern caused by failure. For this purpose, Daubechies, Meyer and Symlet families are considered in the work, since they have provided the most satisfactory results regarding the pattern detection.