Construction And Verification Of Rolling Bearing Fault Diagnosis Model Based On Vibration Signal Analysis

As one of the most critical components in rotating machinery,rolling bearings are very prone to faults and even failures due to their complex and precise structure and the harsh and diverse working environment.In order to guarantee the production safety,the research on the rolling bearing fault diagnosis method is necessary.Rolling bearing fault diagnosis is actually a process of pattern recognition,which mainly consists of fault feature extraction and fault identification.Since the vibration signal,which is rich in fault characteristics,has different degrees of non-smoothness and non-linearity,it makes the extracted fault features insignificant,which leads to a low fault identification accuracy.Therefore,it is important to deeply study the rolling bearing fault mechanism and fault diagnosis model,select systematic and intelligent signal processing and feature extraction methods,and accurately and efficiently complete the fault identification.In view of this,a new rolling bearing fault diagnosis model based on Hierarchical Refined Composite Multiscale Fluctuation-based Dispersion Entropy(HRCMFDE)and Particle Swarm Optimization-based Extreme Learning Machine(PSO-ELM)is constructed in this paper.In the feature extraction part,HRCMFDE is used to extract the fault features embedded in the vibration signals of rolling bearings.By introducing the hierarchical theory algorithm into the vibration signal decomposition process,the problem of missing high-frequency signals in the coarse granulation process is solved.The Fluctuating Discrete Entropy(FDE)maps the adjacent elements in the vibration signal to different classes,which has the characteristics of insensitivity to noise interference and high computational efficiency,making the feature vector more effective in highlighting the fault features.In the fault identification part,the PSO algorithm is used to optimize the input weights and hidden layer neuron thresholds of the ELM network to improve the fault identification capability of the ELM classifier.In order to illustrate the effectiveness and generalization of the constructed rolling bearing fault diagnosis model,two data sets are utilized for experimental validation in this paper.Firstly,using the CWRU public data set,the data samples are divided into 12 categories according to the fault type and fault severity,and the fault identification accuracy can reach up to 100% and has a good load migration effect.Secondly,the bearing fault simulation experiment platform was constructed,and the vibration signals of four kinds of experimental bearings under six working conditions are collected,and the accuracy of bearing fault identification for different working conditions is obtained by the constructed fault diagnosis model,which can reach up to 98%.In summary,the proposed rolling bearing fault diagnosis model has excellent performance and practical application prospects.