Broken rotor bar detection via wavelet packet decomposition of motor current

Abstract

In this paper, discrete wavelet packet decomposition (DWPD) of induction motor current is proposed for detecting broken rotor bar conditions. Good frequency separation is essential for accurate detection of broken rotor bars since it is difficult to separate the rotor bar frequency components from the fundamental supply frequency component at rated speeds. The proposed method provides good frequency separation at a very low computational complexity. The motor current signal is first notch filtered to suppress the power system fundamental frequency. Then, the motor current is decomposed into equally spaced frequency bands by using all-pass implementation of elliptic IIR half-band filter. Next, baseline shifting of wavelet packet coefficients (WPCs) is applied to remove the filtering distortion caused by other harmonics. Finally, energy level of each frequency band is calculated by determining rms values from WPCs of associated frequency bands. The changes in the energy levels of frequency bands in which broken rotor bar related current frequencies lie are monitored to detect motor fault conditions. © 2009 Praise Worthy Prize S.r.l. © 2022 Elsevier B.V., All rights reserved.