Induction motor rotor fault diagnosis using wavelet analysis of one-cycle average power

Abstract

Conventional techniques of detection of broken rotor bars, bearing faults and air-gap eccentricity were based on frequency domain analysis of voltage, current, and instantaneous input power. The accuracy of these techniques depends on the loading conditions of the machine, the signal-to-noise ratio of the spectral components being examined, and the ability to maintain a constant speed. This paper presents a method for induction motor rotor fault diagnosis using wavelet analysis with higher signal to noise ratio under varying load conditions. It also includes an interactive technique to detect broken rotor bar in varying load conditions. The fault severity is derived by wavelet analysis of single-phase active one-cycle average power. Wavelet allows analyzing non-stationary waveform and one-cycle average power allows detecting fault characteristic frequency component under low load conditions without removing the fundamental component. Finally, the proposed method is verified from the experimental results of two induction motors with different configurations.