Research On Rolling Element Bearing Signal Feature Extraction And Fault Diagnosis Method

As one of the key components of high-speed trains,rolling bearing plays a vital role during the operation of high-speed trains.Every year the fault of rolling bearings causes many rail accidents which give the world enormous casualties and property losses.For monitoring the state of rolling bearing,this paper uses a method of fault detection which based on wavelet de-noising and Hilbert-Huang transform to develop a rolling bearing fault diagnosis system and research on the processing algorithm of the sound signal of bearing.Besides,it studies the vibration signal processing methods of gearbox bearings,raises an idea of adopting the time-domain parameters as the fault diagnosis methods,and based on the data collected by the bearing running test platform,the effectiveness for collected signals of the proposed algorithm has been verified.All the researches have an important theoretical significance and application value in reducing property losses,preventing train accidents,and ensuring train operation safety.Firstly,this paper studies the main components of the rolling bearing and its vibration mechanism of the components,and summarizes the causes and the solutions for fault occurrence of the components,and concludes the formula for calculating the frequency of fault features of different types of bearings(ball bearings,cylindrical roller bearings and tapered roller bearings).At the same time,analyzes some common signal processing methods and the advantages and disadvantages of different signal processing methods,then uses the method of time-domain parameters to diagnose acoustic signals and the method of Fourier transform,wavelet de-noising and Hilbert-Huang transform to diagnose vibration signals.Secondly,according to the research of virtual instrument technology and the respective advantages of Lab VIEW and MATLAB,the joint programming mode was used to develop the rolling bearing fault diagnosis system which has the functions of bearing fault characteristic frequency calculation,time-domain parameters,wavelet de-noising and HHT.Finally,conducting the rolling bearing operation test.This paper contains three sets of bearing model data: ball bearing,cylindrical roller bearing and tapered roller bearing model.In the first step,the 6205 ball bearing data was used to verify the feasibility and correctness of wavelet de-noising and HHT.And then,conducting the running test of the 26-type cylindrical roller bearing on the wheel-to-bearing running-in test,and diagnosing the fault of the bearing based on wavelet de-noising and HHT processing of sound signals.At last,using the running test of the R215 tapered roller bearing on the CRRC gearbox bearing running test bench,and estimating the bearing failure status of different maintenance levels by the trend of the time-domain parameters.