Method of Rolling Bearing Fault Identification Based on Cloud Theory and GG Clustering
In order to explore the contribution of different features to distinguish the fault states of rolling bear-ings,the goal is to extract sensitive features,which are associated with the qualitative concept.Cloud theory is introduced into the feature screening,and the proposed method is applied to the fault identification in combina-tion with Gath-Geva(GG)clustering.Firstly,Extracting the high-dimensional feature set from the noise reduc-tion vibration signal,which establishes cloud distribution model under different operating conditions.Then,the forward cloud generator is used to find out the determinacy of each feature on the bearing state under different samples respectively,a threshold is set to screen the features in the original feature set that contribute signifi-cantly to the bearing operating state,the probability of their occurrence is calculated and used as weights to pro-pose a cloud theory-based weighted feature selection method to screen the sensitive feature set.Finally,princi-pal component analysis(PCA)is used to reduce the feature set and input to GG clustering to complete the fault identification.Experimental results show that the proposed algorithm has obvious advantages over the traditional feature selection method in terms of clustering evaluation index and fault identification rate.
rolling bearingcloud theoryfeature selectionfeature weightingcertaintyGG clustering
NETL
NSTL
万方数据
维普
