LAUSR

In this article, a threshold-based induction motor fault diagnosis method is proposed using the measured stator current signal. A 0.25-HP three-phase squirrel-cage induction motor fed directly online is tested in the laboratory with various single- and multielectrical faults under six different loading conditions. The discrete wavelet transform (DWT) is chosen as the signal processing technique for the measured stator currents. The threshold and energy values at each decomposition level of the DWT processing results are evaluated. Threshold values appear to be more consistent than energy values at different measured data windows, and thus, the threshold at the decomposition level d8 is chosen as a fault indicator. Curve fitting equations are developed to calculate threshold values for the motor loadings that were not tested in experiments. The suitability using threshold values for induction motor fault diagnosis is further validated using two probabilistic methods, the correlation analysis and the confidence interval estimation.