Research On Fault Diagnosis Methods For Rolling Bearing Based On EEMD And Least Squares-support Vector Machine

Rolling bearing is the most commonly used machine parts in modern industrial equipment, and it is also the most prone to failure of machine parts. The normal operation of rolling bearing has great influence on the reliability, life and precision of rotating machinery, so the analysis of the fault diagnosis of rolling bearing research to ensure smooth and safe operation of mechanical equipment has the vital significance.In this paper, the improved similar extreme extension is used to suppress the end effect of EMD, the problem is avoided by increasing the judgment of whether the endpoint as the extreme point. By analyzing the simulation signal and the vibration signal of bearing, the experiment result shows that the improved similar extreme extension can effectively inhibit end effect of EMD. The EEMD method is used to eliminate the influence of the modal aliasing. By analyzing the simulation signal, the experiment results show that the EEMD can effectively suppress the occurrence of the modal aliasing, and improve the accuracy of the EMD. An approach applied to the fault feature extraction of rolling bearing is proposed in this paper based on the combination of EEMD and spectral kurtosis. The parameters of EEMD are selected by adding the amplitude standard deviation criterion of white noise. And spectral kurtosis and correlation coefficient method is used to extract effectively the IMF component for signal reconstruction. Finally, the results of envelope spectrum analysis to diagnose bearing fault. The typical fault signal of rolling bearing analysis was performed to verify, and compared with the general EEMD, the experimental results show that this method can improve the accuracy of fault feature extraction and can be effectively applied to the fault diagnosis of rolling bearing. The paper proposed an approach based on EEMD energy entropy and DE-LSSVM applied to the fault pattern recognition of rolling bearing. In order to improve the diagnostic accuracy of model, the differential evolution algorithm is used to optimize the parameters of LS-SVM, the trained DE-LSSVM model is applied to the fault pattern recognition of rolling bearings. By analyzing the different degree of damage of the bearing inner ring signal, Experimental results show that the DE-LSSVM algorithm has the advantages of shorter training time and higher fault recognition rate than LS-SVM and PSO-LSSVM algorithm, this method can accurately diagnose and classify the faults of rolling bearings.