Experimental Investigation of Machine Learning Based Fault Diagnosis for Induction Motors

In this paper, experimental data are used to develop a practical machine learning based fault diagnosis method for induction motors. Both motor stator currents and vibration signals are measured simultaneously in experiments and used in fault diagnosis. Various single- and multi- electrical and mechanical faults are applied to two identical induction motors in experiments. Two signal processing techniques, Matching Pursuit (MP) and Discrete Wavelet Transform (DWT), for feature extraction purpose are chosen. It is found that the proposed method can accurately detect electrical and mechanical faults using several machine learning algorithms.