Bearing Fault Detection in Induction Motors with Hilbert-Huang Transformation

Abstract

Currently, induction motors are widely used in many industrial applications owing to their simple construction and high reliability. There has been a gradual increase in the demands for on-line health monitoring and fault detection techniques to avoid unscheduled maintenance and economic losses caused by sudden failures. In this study, we focused on the detections of bearing faults, which are the most frequently occurring faults in electric motors, and we adopted the Hilbert-Huang transformation, which is a recent signal processing technique, to analyze non-linear or non-stationary signals. We prepared several bearing samples with varying degrees of anomalies by heat treatments and built them into the motors. Based on the measurements of the stator current waveform to the motors, we determined the factors related to the degree of anomalies of the bearings, by comparing their loudnesses. Consequently, we confirmed that the components of higher-order (more than 3rd order) intrinsic mode functions have a positive correlation with the degree of anomalies of the bearings.