Research On Bearing Fault Identification Model Based On VMD Multiscale Entropy And Improved DBN

Rolling bearing is one of the most commonly used parts in rotating machinery,its working state directly affects the performance of the entire equipment,and even affects the safety of the entire production line.Therefore,the research on intelligent diagnosis technology of rolling bearing faults has important theoretical value and practical significance for avoiding accidents.In this paper,rolling bearing is taken as the research object,aiming at the key technical problems such as feature extraction and pattern recognition model in rolling bearing fault diagnosis,a series of research works are carried out from the aspect of bearing vibration signal processing.In order to solve the timeconsuming problem of network layer structure debugging when deep confidence network(DBN)is used for bearing fault diagnosis,a bearing fault identification model based on Particle Swarm Optimization(PSO)is proposed.The model uses particle swarm optimization(PSO)algorithm to find the optimal solution of the node parameters of the hidden layer,and makes a functional comparison between DBN model and PSO-DBN model and draws a conclusion.By analyzing the experimental results and comparing the fault identification accuracy with the multi-scale entropy eigenvectors of three kinds of VMD decomposition by DBN and PSODBN models,it can be concluded that the PSO-DBN model has strong adaptive ability and does not need to rely on experience to select parameters and has a higher recognition rate.When the composite multi-scale dispersion entropy feature vector is used as the input,its feature vector classification effect is better and the recognition degree is higher.It is proved that the improved intelligent fault diagnosis model,VMD-CMDE-PSO-DBN,has a certain application value.