Research On Rolling Bearing Fault Diagnosis Based On Wavelet Analysis And Support Vector Machine

Rolling bearing is one of the important components in rotating machinery.Its function is to change the sliding friction between the running shaft and the shaft seat into rolling friction.Bearing failure is one of the main causes of faults in rotating machinery.Therefore,the fault diagnosis of rolling bearings has been a research topic in recent years.The research on fault diagnosis of rolling bearings also has important theoretical significance and practical application value,research on fault diagnosis management of mechanical equipment also has important theoretical significance and practical application value.With the development of computer technology and signal processing technology,more and more new methods have been introduced into the field of rolling bearing fault diagnosis.The research method used in this paper is based on wavelet analysis and support vector machine.The two methods are combined to realize the rolling bearing.Comprehensive diagnosis of fault type and fault level.Firstly,the wavelet signal analysis in the signal processing technology is used to denoise the original vibration signal data collected by the data acquisition system and extract the effective fault feature vector,and then use the support vector machine(SVM)in the machine learning algorithm.The theory has obvious advantages in small sample size space classification.Finally,a fault diagnosis method based on wavelet analysis and support vector machine is proposed.The main work done in this paper is as follows:1.By introducing the vibration mechanism,failure mode and related detection technology of rolling bearing,the advantages and disadvantages of common detection methods of rolling bearing are analyzed and compared.The vibration signal data of rolling bearing is used to diagnose the fault of bearing,and then the vibration mechanism and characteristics of rolling bearing are studied.frequency;2.On the basis of studying various analysis methods of wavelet theory,the application of wavelet transform in signal denoising and feature extraction is studied,and it is verified by simulation analysis.By comparing the signal-to-noise ratios of wavelet signals with different wavelet bases and decomposition layers,it is further determined that the db6 wavelet base is used in this experiment,and the number of decomposition layers is 4.The vector composed of the energy components of the signal after wavelet decomposition is selected as the eigenvector,and the component ratio map of the energy in each frequency band is drawn.Through multiple comparisons,it is found that when the inner ring of the bearing fails,the energy is mainly concentrated in the first The frequency band and other frequency bands also have a certain distribution;when the bearing outer ring fails,the energy is mainly concentrated in the first and second frequency bands,and other frequency bands also have a certain distribution;when the bearing rolling element fails,the energy is concentrated only in the first frequency band.There are few distributions in other frequency bands;the distribution of bearing energy in the normal state is more uniform in each frequency band,indicating that the selection of feature vectors is effective for the identification of fault types;3.This study proposes a two-layer regression support vector machine for rolling bearing fault diagnosis.Compared to the traditional classification methods of many other studies,this study not only aims to identify different bearing failure modes,but also assesses the size of their defects.This scheme is more meaningful in practice,because the bearing life curve is relatively smooth and smooth in the early stage,and only a sharp increase occurs at the end.According to the above description,the proposed solution involves extracting representative statistical parameters from the vibration signal,identifying bearing conditions in the first layer,and predicting the degree of failure in the second layer once the failure mode is identified in the first layer.Experiments show that the overall accuracy of the method in fault type identification is above 90%,and it has a good effect on the determination of various types of defects in rolling bearings.In summary,the support vector machine based rolling bearing fault diagnosis method is feasible.The analysis and diagnosis results are in good agreement with the actual conditions.It can meet the requirements of rolling bearing fault diagnosis and has a certain guiding role in the fault diagnosis of rolling bearings.