Research On Rolling Bearing Fault Diagnosis Method Based On Variational Mode Decomposition And Entropy Theory

Rolling bearing is an important intermediary component in rotating machinery system.Online monitoring and fault diagnosis of rolling bearing is of great significance to ensure production safety and reduce economic losses.Due to the influence of many nonlinear factors in the environment,the vibration signals of rolling bearings show strong non-stationary characteristics,which bring difficulties for fault diagnosis.Based on this,this paper takes the rolling bearing as the research object,around the three key problems in fault diagnosis: first,the extraction of fault feature frequency;second,the identification of bearing damage parts;third,the quantification of bearing damage degree,the fault diagnosis method of rolling bearing is studied,in order to improve the accuracy of fault diagnosis.The main contents of this paper are as follows:(1)Aiming at the problem that the original signal of rolling bearing is interfered by environmental noise,which makes it difficult to extract fault information,a fault feature extraction method based on the parameters optimized maximum correlation kurtosis deconvolution(MCKD)and the variational mode decomposition(VMD)combined with Teager energy operator(TEO)spectrum is proposed.Firstly,the Particle swarm optimization(PSO)is used to optimize the relevant parameters of MCKD and VMD,which improves the performance of the algorithm.Secondly,based on the optimized algorithm,the vibration signal is denoised jointly,and the natural mode components are screened to achieve effective suppression of background noise.Finally,the feature frequency of sensitive IMF components is enhanced and extracted by TEO.The experimental results show that the method can effectively extract the fault characteristic frequency of rolling bearing vibration signal,and has high frequency domain resolution.(2)In order to accurately judge the damage location of rolling bearing,a identification method for damaged parts of rolling bearing based on parameter optimization VMD singular value entropy was proposed.Firstly,the optimized VMD is used to decompose the vibration signal,and IMF components are composed of initialization matrix to realize the effective separation of different modulation components in the signal.Next,the singular value decomposition is used to decompose the matrix,and the noise reduction order in the singular value sequence is reasonably determined according to the singular value difference curve,so as to realize the high concentration of characteristic information.Finally,the singular value entropy of the denoised singular value sequence is calculated and the entropy distribution range of the same type of signal can distinguish the damage position of rolling bearing.The experimental analysis of the measured bearing signals shows that this method can effectively distinguish different bearing damage locations,and has good feature mapping ability and algorithm robustness.(3)In order to realize the accurate identification of rolling bearing damage degree,a rolling bearing damage degree identification method based on VMD composite multiscale permutation entropy(CMPE)and fuzzy neural network(FNN)is proposed.Firstly,the influence of CMPE parameters on algorithm performance is studied,and the parameters of CMPE suitable for non-stationary signal feature extraction are set reasonably.After that,the vibration signal is decomposed by VMD,and the CMPE of each IMF component is calculated to form the entropy sequence,so as to realize the effective extraction of feature information.Finally,the entropy sequence is composed of entropy feature matrix and input into FNN.Based on FNN’s excellent fuzzy logic reasoning ability,the accurate identification of the damage degree of rolling bearing is presented.Based on the measured data of rolling bearing published by Case Western Reserve University,the experimental results show that the proposed method has high accuracy in judging the damage degree of rolling bearing.