Induction motor faults diagnosis using support vector machine to the motor current signature

Bearing defects were difficult to be identified effectively by Motor Current Signature Analysis (MCSA) due to the obscure features on the Fourier spectrum. An automatic induction motor diagnosis system based on motor current signature was proposed in this research. Wavelet packet decomposition was applied to analyze the Fourier spectrum of the stator current. The generated wavelet coefficients were found to have specific time-frequency features which reflect multiple failure modes of the induction motor. Then support vector machine was used to build a motor fault prediction model. The experimental result showed that the motor fault identification accuracy could reach more than 90%, including the diagnosis of bearing defects.