Machine Learning Approach to Analyze Converter Fault in Switched Reluctance Motor Drive Using Finite Element Method

Switched Reluctance Motor (SRM) has recently been in the limelight as its features are more suitable for battery-operated electric vehicles. This paper analyzes the switch short circuit and open circuit fault in the asymmetric bridge converter of the SRM drive and its impact on the motor parameters using the finite element method. With the help of the ANSYS Maxwell FEA tool, simulation is carried out on a 550 W, 220 V, Four Phase, 8/6 Pole, 940 rpm SRM, and its parameters like phase current, airgap flux density, flux lines, and torque are analyzed under healthy and faulty condition. From the finite element model, information about motor phase current is extracted, and machine learning algorithms in MATLAB, such as KNN, decision tree, and random forest, are used to investigate the fault. Based on the phase current feature, KNN provides 93.97%, decision trees 97.02%, and random forest 96.88% accuracy. An experimental model is also implemented to analyze the switch short & switch open circuit fault.