Neural Network Classifier for Three Phase Induction Motor Stator Fault Classification

Abstract

Understanding electric motor dynamics and electro-mechanical interaction is facilitated by computer simulations of their functioning. A proper model makes it possible to simulate motor defects and forecast changes in the associated parameters without the need for practical investigation. This suggested strategy provides an examination of asymmetric stator winding and rotor flaws in induction machines from both a theoretical and experimental standpoint. A three-phase induction motor was simulated and put through its paces while under normal, healthy operation, while experiencing short circuit winding faults, phase-to-phase winding faults, phase-to-ground winding faults, and voltage imbalances between supply phases. The findings show that the outcomes of simulations and experiments generally accord. First, the traditional Fast Fourier transform approach is used to analyze the fault situation and is tested under various winding fault scenarios. For the analysis of stator winding defects, a fuzzy logic controller based on fuzzy rule basis architecture is then created. Both scenarios make it evident that the FFT
analysis only calibrates the total harmonic distortion (THD) of the three-phase induction motor's input side voltage and current signals that are faulty (stator side). Artificial Neural Network, on the other hand, immediately examined the defect kind on the induction motor stator winding. Finally, the outcomes of the artificial neural network are compared to those of the fuzzy
logic controller classifier, wavelet transform, and fast Fourier transform analysis methods for signals. MATLAB 2015 Simulink software is used to develop a motor model and fault analysis system. This programmed is used to examine failure instances and motor parameters.