Detection of Induction Motor Incipient Short Circuit Faults Based on Continuous Wavelet Transform With Higher-Order Statistics

Abstract

The increased use of induction motors (IM) in industry requires efficient solutions to control and prevent from faults aiming to optimal production efficiency. Accordingly, several studies focused on identifying IM defects using motor current signature analysis (MCSA) and other techniques like acoustic noise and vibration analysis. In the context of MCSA strategies such as extended Park’s vector approach and negative sequence current, fault signatures measured from stator currents, provide rich information about faults. Conversely to conventional spectrum analysis approaches like Fast Fourier Transform (FFT) and Short Time Fourier Transform (STFT) of well-known limitations, this paper introduces a novel method for early detection of induction motors faults using stator current measurements. Consequently, the proposed approach utilizes the combination of Continuous Wavelet Transform (CWT) and the third-order statistics, specifically, the skewness to produce discriminative features. Then, the correlation coefficient is used to locate the exact onset of the fault. The validation of the method using various short-circuit fault, issued from experimental acquired signals confirms its effectiveness in detecting both severe and light faults when compared to reported conservational approaches.