Induction motor mechanical fault online diagnosis with the application of artificial neural network

An online fault diagnostic algorithm for induction motor mechanical faults is presented based on the application of artificial neural networks. Two mechanical faults, the rotor bar breakage and air gap eccentricity, are considered. New feature coefficients obtained by wavelet packet decomposition of the stator current are used together with the slip speed as the input of a multi-layer neural network. The proposed algorithm is proved to be able to distinguish healthy and faulty conditions with high accuracy.