Research On Bearing Fault Diagnosis Of High-speed Trains Based On Time-frequency Analysis And Screening Bandpass Filtering

High speed train wheel bearings are one of the important components of highspeed trains,playing a role in bearing the weight,load,and reducing friction of the train body.Their operating status is directly related to the safe and stable operation of the train.Conducting research on bearing fault diagnosis for high-speed trains is of great significance for safer,more efficient,and smoother operation.The fault signal of wheel bearing is a typical nonlinear and non-stationary signal.Short Time Fourier Transform(STFT),as a typical time-frequency analysis method,is flexible in application in the field of fault diagnosis and demonstrates good time-frequency fault feature extraction ability.However,there is strong background noise interference in the wheel bearing fault signal,which seriously affects the accurate extraction of fault time-frequency features in the signal.In view of the above problems,the thesis conducts in-depth research on how to accurately extract the time-frequency characteristics of faults under strong background noise,and develops a set of weak fault diagnosis system for wheel set bearings including software and hardware equipment.The main research work of the thesis is as follows:(1)This article first introduces the structure and failure forms of rolling bearings,analyzes the failure mechanism of bearings,and analyzes the waveform of fault signals.Then,it introduces the mathematical framework of STFT and uses simulation signals to verify its effectiveness and shortcomings under different signal-to-noise ratios.When the signal-to-noise ratio is low,STFT can extract fault features from the time-frequency map.When the signal-to-noise ratio is high,signal fault feature extraction is poor.Then,the process of the STFT algorithm was explained,and the effectiveness of the algorithm’s perfect reconstruction was verified using experimental signals.(2)This article proposes a denoising method for adaptive screening bandpass filtering.Firstly,the basic principle of screening bandpass filtering is described in detail,and on this basis,the screening bandpass filtering and STFT method are fused to effectively detect weak faults in wheelset bearings.The screening bandpass filtering algorithm uses adaptive inner and outer window changes to extract periodic pulse components from time-frequency images.As an algorithm for extracting periodic components from two-dimensional images,it can extract periodic fault feature components from time-frequency images after short-term Fourier transform,achieving the goal of eliminating noise interference components.The validation results show that the proposed method can effectively denoise fault signals,And can extract periodic fault components that exist in the signal.Compared with other methods,the results show that this algorithm can extract weak fault information components of bearings under complex background noise,and has more superiority and effectiveness in extracting periodic components.(3)On the basis of the aforementioned methods,the thesis has developed a wheel bearing fault diagnosis system that includes both software and hardware devices.The system utilizes NI-DAQ acquisition board,NI chassis,built-in power amplifier acceleration sensor(IEPE),data cables,and other hardware facilities to build.Design using Labview software,including up to eight channels of time-domain analysis,frequency-domain features,and Hilbert envelope spectrum analysis,and be able to achieve three save mode acquisition methods and analyze the processing results of the proposed algorithm.The system can achieve multi-channel acquisition of bearing signals and perform real-time filtering and processing of the collected signals.It successfully combines theory with practice,software with hardware,and forms a complete scheme from signal acquisition to signal processing.