Faulty bearing detection, classification and location in a three-phase induction motor based on Stockwell transform and support vector machine

This paper presents faulty bearing detection, classification and its location in a three-phase induction motor using Stockwell transform and Support vector machine. Stockwell transform is applied to stator current signals to extract a number of features in both time and frequency domain. A set of non-correlated and high ranking features are selected based on Fisher score ranking. These features are in turn used to classify the faults such as ball, cage and outer-race faults using Support vector machine. Subsequent to fault identification, features of Stockwell transform are used to locate the defective bearing, i.e, either at fan-side or load-side of the motor. This algorithm is successfully implemented on the experimental data of defective bearings collected from the industry.