A Method Of Fault Diagnosis Based On FVMD Multi-scale Permutation Entropy And GK Fuzzy Clustering

The normal operation of mechanical equipment is an important guarantee for industrial production,so the fault diagnosis technology of mechanical equipment has been paid more and more attention.In this paper,the fault feature extraction and pattern recognition,which are taken as two key aspects,were studied based on the analysis of vibration signal,and a fault diagnosis method based on the fast variational mode decomposition(FVMD),the parameter optimized multi-scale permutation entropy and the feature weighted GK fuzzy clustering was proposed.Firstly,the idea of fast iteration was introduced on the basis of variational mode decomposition,and the fast variational mode decomposition method was proposed.In fast variational mode decomposition method,the Lagrange multiplication operator is updated twice,so that the convergence speed of the algorithm is accelerated.The running time and iteration number of fast variational mode decomposition are reduced while the iteration termination criterion is unchanged.The noise robustness of fast variational mode decomposition method and the ability to separate the multimode signal were analyzed and verified by the simulation signal,and the fast variational mode decomposition was also applied to the fault diagnosis of rolling bearing.Secondly,the parameters of multi-scale permutation entropy algorithm,which are length of time series N,embedding dimension m,delay time τ and scale factors s,were studied.Aiming at the problem of parameters determination of multi-scale permutation entropy and considering the interaction among parameters comprehensively,a method of parameters optimization based on multi force particle swarm optimization was proposed.Then,the multi-scale permutation entropy was used to quantify the fault features,and the parameter optimized multi-scale permutation entropy were combined with fast variational mode decomposition for fault feature extraction,in addition,the effectiveness was verified by vibration signal of rolling bearing.Thirdly,considering the contribution of each feature in the feature vector to the clustering process,a new method of GK fuzzy clustering based on ReliefF feature weighting was proposed.After that,the feature weighted GK fuzzy clustering was used to determine the standard clustering center,and the fault pattern recognition was realized by the principle of choosing the nearest.Then,the method was used to deal with ten kinds of vibration signals of rolling bearing,which contain different fault types and different degrees of damage.By comparative analysis,the superiority of the proposed method was verified.Finally,the fault data of rolling bearing and gear collected separately on the machinery fault simulator was taken as the research object.The fast variational mode decomposition method was used to reconstruct the original fault signal,and the fault feature was extracted by the parameter optimized multi-scale permutation entropy algorithm,then,the fault recognition was realized by combining the feature weighted GK fuzzy clustering based on ReliefF and the principle of choosing the nearest.The results show that compared with the improvement variational mode decomposition,multi-scale permutation entropy and traditional GK fuzzy clustering,the proposed method not only can extract the fault features effectively,but also can realize the classification and recognition of fault modes accurately,and the fault recognition rate is more higher.