Fault Recognition Method for Rolling Bearing Integrating VMD Denoising and FCM Clustering

摘要

A novel method for fault recognition of rolling bearing is proposed. The method integrates Variational Mode Decomposition (VMD) denoising and Fuzzy Center Means (FCM) clustering. Data preprocessing is as follows: firstly the preset scale K is obtained according to the phenomenon that center frequencies of adjacent Intrinsic Mode Functions (IMFs) coincide, when over-segmentation. Secondly vibration signal is decomposed by VMD to obtain IMFs. And according to the principle that the correlation coefficient is greater than 0.95, the useful IMFs are extracted and reconstructed. At last, the reconstructed data source retaining most of useful information is obtained. Fault recognition is as follows: firstly the sample entropy and the root mean square are calculated and used as feature vector. Then feature vectors set of training samples is received. Secondly it is put into FCM clustering, the two-dimensional clustering map and the cluster centers are acquired. Lastly, the fault type is recognized depending on the principle that the Euclidean distance between feature vector of test sample with cluster centers of training samples is the minimum. Measured data analysis shows that the proposed method can accurately recognize the fault type of rolling bearing, and the correct rate of fault recognition reaches 97.5%.