Rolling Bearing Fault Diagnosis Based On TQWT Sparse Basis Tracking Denoising And Adaptive Feature Extraction

As an important component of rotating machinery,when the rolling bearing fails,it will seriously affect the performance of the machinery,and at the same time cause a series of shocks,and the pulses generated by these shocks will stimulate the periodic vibration of the system.Due to the complex working environment of bearings,the bearing vibration signals collected by the experiment contain serious noise interference,resulting in the weak fault characteristics of the bearing in the early stage often drowned out by noise.If the noise is not treated,it will cause difficulties in bearing health monitoring and fault diagnosis,so noise reduction treatment of bearing vibration signal is an important prerequisite.This paper takes the early weak fault detection of rolling bearings as the research object,uses Tunable Q-Factor Wavelet Transform(TQWT)as the signal analysis method,and uses the Support Vector Machine(SVM)as the classification recognition method,and the main research content includes signal noise reduction,eigenvalue extraction and bearing operating state recognition.The details are as follows:(1)Aiming at the problem of TQWT parameter selection and decomposition,this paper proposes a KSR index that can measure the signal impact intensity and period smoothness at the same time on the basis of steepness and signal smoothness,and constructs an adaptive TQWT parameter search method based on KSR,which adaptively searches for the best parameters according to the signal characteristics to decompose the signal.On the basis of the decomposition of TQWT optimal parameters,the components with rich impact characteristics are selected according to the KSR component selection rules,and the components are reconstructed,and the fault characteristics in the signal are further extracted to realize the weak fault diagnosis of bearings.Through simulation signal simulation experiments,the effectiveness and advantages of this method are verified.(2)Aiming at the problem of strong noise interference in the vibration signal of rolling bearings,a signal noise reduction method combining TQWT optimal subband component selection and sparse base tracking is proposed.For the optimal subband components screened by KSR,the TQWT sparse basis tracking noise reduction algorithm is introduced to reduce the noise of the reconstructed signals,and the characteristic frequency of bearing faults is analyzed by using Hilbert envelope spectrum.The experimental analysis using the bearing engineering data of Case Western Reserve University shows that the proposed method can effectively eliminate the noise in the vibration signal of rolling bearings and effectively highlight the fault frequency characteristics.(3)In order to accurately extract the fault characteristics of rolling bearings,a rolling bearing fault identification method based on TQWT multi-scale arrangement entropy and kernel principal component analysis(KPCA)is proposed.For the vibration signal after TQWT noise reduction,the initial features are extracted according to the multi-scale arrangement entropy value,and the relationship between the features is analyzed.In order to reduce the dimension of the feature set and the redundant information,KPCA is used to process the initial feature set.The low-dimensional feature set is fed into the Support Vector Machine(SVM)to identify bearing faults.The method proposed in this paper and the method of single SVM are applied to the bearing engineering data of Case Western Reserve University,and the experimental results show that the proposed method can accurately identify the bearing operating state.