Soft computing approach for fault Detection in electrical machines

Abstract

In this thesis the broken rotor bar end ring fault detection schemes newlinefor three phase induction motor are proposed The harmonic components of newlinethe stator current are extracted for fault diagnosis The neural network and newlinefuzzy logic are used to diagnose the faults To improve diagnostic accuracy of newlinefault diagnosis system and to obtain the optimal values of fuzzy membership newlinefunctions genetic algorithm is used The performance of fuzzy fault diagnosis newlinesystem with conventional semi circular and optimized membership functions newlineis compared The bearing fault detection schemes for three phase induction newlinemotor are implemented using neural network and fuzzy logic The technique newlineis based on monitoring the stator current spectrum and speed The harmonic newlinecomponents of stator current spectrum and speed are taken as the inputs for newlineneural network and fuzzy logic fault detectors The performance of neural and newlinefuzzy fault detection schemes is compared in terms of percentage of error newlineThe fuzzy logic based fault detection scheme is implemented in real time The newlineexperimental results are presented newline