Early Fault Diagnosis Research Under The Background Of Strong Noise For High-speed Railway Wheels

With the rapid development of railway transportation in China,the railways are more declined to be higher speed and heavier load,so that leading increasing stress operation in the wheel tread for high-speed rail vehicle wheels,like wheel rim,spokes and a web plate hole which easily occur near local stress concentration and fatigue crack initiation.Based on train severe accidents happened in history,it has been of apparent importance and necessity of early railway wheels fault diagnosis research.This thesis adopting metal magnetic memory(MMM)method to get the early fault information of stress concentration and fatigue crack initiation and diffusion to make decisions of the risk of early accurate diagnosis to decide the fault degree,that can effectively forecast and avoid dangerous accidents.Adopting the adaptive lifting wavelet threshold de-noising method,which combines with HHT transform on the process of the magnetic flux leakage signals of de-noising and feature value extraction,that makes the leakage magnetic signal de-noise and obtain more accurate signal characteristic value,and through the imitation of the Matlab2013 to verify the method to be effective.Based on the support vector machine(SVM)theory to put forward the SVM mixed algorithm to establish fault diagnosis model of train wheels early faults.Through the three parameters peak-to-peak value,wave crest value and wave valley value to establish fault signal model and then through the training sample and test sample test to verify the reliability and accuracy to recognize signal faults.Above all,by adopting the metal magnetic memory testing technology to acquire magnetic flux leakage signal,through adaptive lifting wavelet threshold de-noising and HHT transform method to deal with magnetic flux leakage signals to de-noise and get feature value extraction,and then establish the fault diagnosis model based on support vector machine(SVM)to identify the state and the model of fault signal.Finally,through above methods to effectively realize the high-speed train wheels early fault diagnosis of stress concentration and fatigue crack fault.