Online rotor bar breakage detection of three phase induction motors by wavelet packet decomposition and artificial neural network

Online detection algorithm for induction motor rotor bar breakage is presented using a multi-layer perception network (MLP) and wavelet packet decomposition (WPD). New features of rotor bar faults are obtained by wavelet packet decomposition of the stator current. These features are of multiple frequency resolutions and obviously differentiate the healthy and faulty conditions. Features with different frequency resolutions are used together with the speed slip as the input sets of a 4-layer perceptron network. The algorithm is evaluated on a small three-phase induction motor with experiments. The laboratory results show that the proposed method is able to detect the faulty conditions with high accuracy. This algorithm is also applicable to the detection of other electrical faults of induction motors.