Broken rotor bar diagnosis in induction machines through Stationary Wavelet Packet Transform under lower sampling rate

In induction motors (IM), broken rotor bars (BRB) create abnormal amplitude of the sidebands around the fundamental supply frequency. Several monitoring techniques developed for non stationary signal are essentially based on Wavelet Transform (WT). In the WT, the sampling rate determines the frequency resolution and it is an important factor in the embedded diagnosis system design. This paper proposes a new methodology for broken-rotor-bar diagnosis under lower sampling rate based on Stationary Wavelet Packet Transform (SWPT). The SWPT coefficients affected by the rotor fault are computed then the relative energy value is proposed as the diagnosis parameter. The statistical results based on boxplot show that SWPT with a lower sampling rate has a good performance to detect BRB under different load conditions and different fault severity.