Bearing Fault Diagnosis Based On EEMD-SVD And FCM Clustering

Rolling bearing is the most commonly used in the machinery is also the most easily damaged parts, its working status directly affect the efficiency of the whole machine, so the working state of rolling bearing testing and fault diagnosis has important research significance.The process of machinery fault diagnosis includes the acquisition of information and extracting feature and recognizitions of which feature extraction and condition identification are the priority.When the rolling bearing occurs,the vibration signals often have the characteristics of nonlinear and non-stationary.In this paper Ensemble Empirical Mode Of Decomposition(EEMD),a self-adaptive method,using the White Gaussian Noise with a uniform frequency distribution statistics,the signal in different scales is continuous,avoid the phenomenon of mode mixing. Also, a feature extraction method based on EEMD combinate Singularity Value Decomposition(SVD)was put forward used to make the judgment on the basis of fault diagnosis.Then Fuzzy C-means Clustering(FCM) algorithm was appilied to idenitify the fault.Fristly, the basic princiciples of Ensemble Empirical Mode Of Decomposition was researched, and this method was used to analyse signals. And its anti mode mixing performance is analyzed, In view of the problem of false component and noise component in EEMD, the correlation coefficient analysis is used to extract the effective intrinsic mode function.Secondly, the recognition and extraction of fault feature often becomes difficult when the mechanical equipment is in the background of strong noise. For this kind of situation by the EEMD and singular value decomposition combination of feature extraction methods, is conducive to improve the rolling bearing fault signal recognition accuracy,enhanced fault signal feature extraction ability retained effective signal component and to eliminate noise components so as to improve the signal-to-noise ratio.Finally, with the fuzzy c-means clustering as a fault identification method, to the building of the bearing vibration signal fault feature vector clustering analysis, and realized the bearing fault diagnosis, at last, by comparing several different methods offault diagnosis, EEMD- SVD and FCM clustering can improve the accuracy of rolling bearing fault diagnosis.