In order to improve the accuracy of rolling bearing fault diagnosis,a new method of rolling bearing fault diagnosis based on EEMD-BP and kurtosis criterion is proposed.The vibration signal is mixed with different levels of noise,which results in low diagnostic accuracy.Firstly,a number of IMF are obtained by EEMD decomposition to eliminate the modal aliasing and endpoint effects;secondly,the IMF component is selected by the method of cross-correlation coefficient and Kurtosis criterion to reconstruct the signal,and then the vibration signal is de-noised;then,the BP network is used to diagnose the bearing fault;finally,the data based on cross-correlation coefficient method,kurtosis criterion,EEMD denoising and original data are input into BP network to test and compare,to verify the validity of rolling bearing di-agnosis.The results show that the diagnostic accuracy of EEMD-BP model is 97.33％,which is 23.16％higher than that of BP model.The noise reduction method based on the fusion of EEMD,cross-correlation coefficient method and kurtosis criterion is suitable for fault diagnosis of rolling bearings.
万方数据