The Fault Diagnosis Of The Motor Bearing Based On The Wavelet And The Neural Network

As an important electromechanical product, asynchronous motors are widely used in various fields. With the development of science and technology motors make even more great functions than before. However, motors fault will terribly damage motors and the normal work of the system, endanger operators ' lives and make heavy economic losses and wide-ranging influence in the society. Therefore, it is of great importance and practical engineering value to make researches on motors fault diagnosing.This thesis focuses on motor bearing, that is, to make a analysis of vibratory single in the three modes of normalbearin??-* damaged inner bal1 tracks, bearing balls pitting and pick up oignal characteristics as neural i:etvv"ork inpu: vector. Author examined the tendency of the pitting fault of ball by using Levenberg - Marquardt BP study algorithm and identifies the web that's of training and models under teachers' controls. The first part of this thesis makes a statement of the history and development of asynchronous motors and bearing fault diagnosis technology. According to bearings' structures and inner vibration source, the cause of bearings' fault and charactaristic frequency are analyzed. After analyzing the wavelet-neural networks and discussing methods application of bearing fault diagnosis, fault diagnosis system that is with the help of wavelet packet to single denoising and with the assistance of Levenberg-Marquardt BP algorithm to be the mode of fault or the way to identify the fault situation is built up. Then, motor bearing experiments are carried on the faults diagnose bench in the laboratory to make a research on motor bearings fault diagnosing. It is proved that this method candivide these three areas of fault modes perfectly and diagnose the development tendency of bearing pitting faults. Finally, the conclusion is drawn that it can examine and diagnose bearing faults as expected.In the process of data analysis, author makes use of Matlab that is of great ability of visual data processing of digital signals of big amount and realizes direct and valid data processing.