Analysis Of Machinery Vibrating Fault Based On Wavelet Theory

With the development of scientific technology and automation,machinery equipment is developing to be more and more complex,which makes the application field become widened.If it can be monitored and diagnosed in advance,the risk of machinery fault can be remarkably reduced and the profit of enterprise can be increased availably.Wavelet is a method of time-frequency analysis,which is adapted to probing the instantaneous abnormal phenomena of signals exactly.It can separate the real signal from the noise signal effectively and attains the purpose of de-noise and it is not affected by signal state at the same time.According to this characteristic,application problem of wavelet theory are investigated,such as signal threshold de-noising,fault analysis and diagnosis.An improved threshold algorithm for signal detection and de-noising was developed based on wavelet transform.Two different thresholds were established according to the Signal-to-Noise Ratio(SNR) of the signal.The measurements of the threshold vary with the wavelet scale,due to the different transformation characteristics of signals and noises at different scales.Through the decomposition of the signals,each high frequency coefficient was shrunk by a threshold,and then restricted the signal.This thesis chose the threshold algorithm from Matlab wavelet box and the improved threshold mentioned above to simulate the signals under the circumstance of different SNR.The experiments indicate that this method gives better SNR performance than other wavelet de-noising methods.Through multi-analysis of vibration theory of the machinery fault analysis,an improved wavelet packet energy algorithm was developed,which combined the wavelet point with the time parameter aider the wavelet packet analysis.This algorithm considered the distribution characteristic of wavelet packet energy in each band,which shown the distribution of the energy with the time variation effectively.The enveloping analysis had been introduced at the same time,which could show the spike pulse effectively and improve the precision of the signal analysis. Machinery vibration signal fault diagnosis system was developed based on Matlab7.0 situation.This system contains six contact surfaces,such as time-domain analysis, frequency-domain analysis,wavelet de-noising analysis,wavelet packet analysis,fault diagnosis and rotor experiment platform system.Signals were from 750W chemical fiber electrode and WS-ZHT1 rotor experiment platform.