Detection of broken bars in induction motor based on multiple coupled circuit model with optimized parameters

This paper presents an approach to the detection of broken bars in an induction motor based on a multiple coupled circuit model, whose stator and rotor can be represented by a number of coupled electrical circuits. All self-inductances and mutual inductances of the model are calculated based on the winding function theory. By applying a genetic algorithm (GA), optimized parameters are obtained by minimizing the errors between experimental results and simulation results. The comparison result of speed and current illustrates a good agreement between the model and the experimental data in a healthy condition. Based on the optimized model, simulation results of the machine with one broken bar and two adjacent broken bars are provided for fault diagnosis. By analyzing the power spectrum density of the stator current, the broken bars can be detected by clear and strong side frequency components.