Raised Cosine Filter Embedded in Stationary Wavelet Packet Transform for Induction Motor Fault Detection

This paper proposes an original combination of Raised Cosine (RC) wavelet filter with the Stationary Wavelet Packet Transform (SWPT) to detect the Broken Rotor Bar (BRB) fault in Induction Motors (IM) through Motor Stator Current Analysis. The combination contributes to the establishment of an efficient and a straightforward approach for extracting the BRB fault feature using wavelet techniques even at low slip. A wise tune of the RC wavelet filter enabled the implementation of a simple SWPT, without the need to its recursivity for fundamental component (i.e. 50Hz) removal. The experimental results show that the proposed method is able to detect faulty conditions at incipient stage with high accuracy even at low slip. A comparative study between the Meyer wavelet based RSWPT and the proposed method shows the superior filtering features of the RC wavelet. The tunable-at-will RC wavelet filter and the reduced sampling frequency and samples size are considered as appreciated features for a hardware implementation in embedded systems.