3-phase Induction Motor Bearing Fault Detection and Isolation using MCSA Technique based on neural network Algorithm

Abstract

This paper shows a system that has the ability to diagnose bearing fault in three phase induction motor by using Motor Current Signature Analysis (MCSA) technique associated with artificial neural network (ANN) algorithm. Mathematical models for healthy and faulty conditions built to demonstrate theoretically the behavior of 3-phase induction motor in both cases. The effects of such a fault on motor currents waveforms at different loads studied experimentally using practical data acquisition and Fast Fourier Transform (FFT) analysis. The harmonic content for this fault current, through the loading range, is studied, and fed to neural network algorithm. A numerical optimization technique using Levenberg-Marquardt algorithm has been done for ANN training and testing.This system prepared to be used in industrial applications to diagnose and isolate the faulty motors immediately at their incipient stage, and to avoid any damage occur for the motors, or for their supply system