Diagnosis of Bearing And Interturn Faults In Induction Motor

Abstract

ABSTRACT newlineInduction motor is the most widely used rotating machinery in industries. The newlineprimary objective of the present work is to detect the most widely occurring faults, for newlineexample faults on rolling element bearings and inter turn faults in the stator windings newlineof an induction motor, at an early stage to take corrective action prior to the newlinecatastrophic failure. In this context, it is important to be able to discriminate between newlinehealthy and faulty conditions. A number of conventional condition monitoring newlinetechniques exist by which faults in induction motors may be detected. Some newlinetechniques use current measurements for the detection of electrical faults like inter newlineturn faults, some use vibration signals to detect mechanical faults, like the bearing newlinefaults. However, under circumstances involving multiple fault conditions newlineconventional condition monitoring techniques may fail. Therefore, this work deals newlinewith diagnosis of both electrical (stator- inter turn) and mechanical (bearing) faults in newlineInduction motor using vibration analysis only. An experimental set up was designed newlineand fabricated to simulate bearing and stator inter turn faults. Different fault newlineconditions were introduced in the set up and vibration data were acquired for faults in newlinebearing inner race, outer race; inter turn faults and multiple faults. An Artificial newlineNeural Network (ANN) was developed to use this data for diagnosis purpose. By newlineusing ANN, it is demonstrated that both electrical and mechanical faults could be newlineidentified by using vibration data alone and there is no need to use current signal for newlinedetecting inter turn faults. In addition to use of ANN, methods of waveform, spectrum newlineand wavelet analysis are also used for fault diagnosis. Thus, it is demonstrated that newlineANN with statistical parameters of time domain vibration data or with wavelet newlinecoefficients data, can successfully detect electrical and mechanical faults and also the newlinecoexistence of both the types of faults in induction motor. newline