Vibration Fault Diagnosis of Turbo-generator Set Based on Hybrid Fuzzy Clustering Analysis

To overcome the disadvantages of fuzzy C-means(FCM) method in the turbo-generator vibration fault diagnosis,a novel weighted chaos option fuzzy C-means(WCOFCM) algorithm is proposed.The WCOFCM algorithm combines chaos optimization strategy with the conventional FCM method,and optimizes fuzzy clustering objective function via chaos variable searching and gradient operator to achieve effective convergences.The similarity in samples is considered in assigning weights for different characteristics to obtain an adaptive classification.Moreover,the WCOFCM algorithm can automatically get the number of clusters according to the validity function.The algorithm is successfully applied in analyzing an international standard testing data set,as well as diagnosing vibration faults of a turbo-generator set in a power plant.The results show that it reduces the misclassification rate and therefore effectively diagnoses the turbo-generator vibration fault.