Research On Fault Diagnosis Method Of Rolling Bearing Based On Feature Selection And Optimization LSSVM

As the core component of rotating machinery,rolling bearing often has a high failure rate due to harsh working environment.The research on rolling bearing fault diagnosis can maximize the uptime of components and minimize the maintenance cost,while reducing the frequency of failure.Among them,a good signal processing method can effectively extract the fault features from vibration signals.So as to realize efficient identification of bearing fault status.Therefore,this article has done the following work based on the signal processing method:(1)To solve the problem of incomplete features caused by single-domain feature extraction of bearing signals,this thesis proposes a multi-domain feature extraction method which decomposes sample entropy energy entropy from time domain,frequency domain and variational mode decomposition.Simulation experiments are carried out on the bearing fault dataset of Case Western Reserve University.The results show that the diagnostic accuracy of this method is improved by more than 4% compared with that of single-domain analysis.The diagnostic performance of the model is further improved.(2)In view of the poor diagnostic effect caused by the redundancy and noise in the high-dimensional feature vectors of the signal,this thesis adopts the Laplace score method as the feature selection method,and makes a comparative analysis between the experimental part and Fischer score method.The results show that the feature dimension is reduced to 14 dimensions after the feature selection by the Laplace score method,and the partial redundant features and noise are removed.which effectively improves the accuracy of fault state recognition.(3)Aiming at the blindness of Least Squares Support Vector Machine(LSSVM)parameter selection,this thesis proposes an improved particle swarm optimization method to optimize LSSVM model.By combining linear differential decrement inertial weight and asynchronous acceleration factor,the standard particle swarm optimization algorithm has a fixed parameter.Then,a bearing fault diagnosis method model based on Laplace score and improved particle swarm optimization LSSVM was constructed.The results of experimental simulation show that the overall model has a good diagnostic effect,and compared with other methods,the effectiveness and superiority of the model are confirmed.