An approach of attribute reduction based on FCM and its application in fault diagnosis on turbo-generator unit

Fuzzy c-means(FCM) clustering method is applied to discretizing continuous attributes of turbo-generator unit faults by the analysis of spatial distribution of pattern datasets.The attribute reduction algorithm for rough set(RS) is utilized to optimize original feature vectors of turbo-generator unit faults,eliminate redundant information and extract essential information of feature vectors.Fault diagnosis model based on support vector machine(SVM) is built up according to the reduced feature vectors.Experimental results demonstrate that the proposed method can improve the accuracy and real-time performance of fault diagnosis,decrease the number of test items and accordingly cut down the cost of fault diagnosis on turbo-generator unit.