Studies Of Fault Diagnosis Algorithm Based On Fuzzy Clustering Method

With the development of science and technology,the tendency of industrial equipment is automation and intelligent,however,it brings lots of problems.Complex structure of industrial equipments makes that the existing mechanical equipment fault diagnosis technologies has been unable to meet the needs of equipment fault diagnosis.Therefore,the author studies the fault diagnosis algorithm based on fuzzy clustering method under the engineering background of fault diagnosis of metro vehicle auxiliary inverter.The article included six chapters.In the first chapter,the origin,engineering background and significance of the subject are introduced;what's more,the classification of fault diagnosis technology,domestic and international developments are also,introduced and the main content and the structure of article are listed.The second chapter includes the basic concept of fuzzy clustering method,fault signals analysis method and similary discrimination method.The 3rd,4th and 5th chapters are the main content of the article,also the core of article,which include as following.(1)The basic principle of fuzzy C-means clustering is introduced,its application to fault diagnosis is studied and its general steps used in fault diagnosis are concluded in the 3rd chapter.What's more,on the basis of FCM clustering algorithm,weighted fuzzy C-means clustering algorithm is proposed.The difference between FCM clustering algorithm and WFCM clustering algorithm is that attribute matrix is introduced to show the different effects of different components of the feature vector to clustering in WFCM.The results show that FCM and WFCM have the same convergence rate and diagnostic results of WFCM is slightly better than FCM.(2)The main content of the 4th chapter is application of AP clustering algorithm in the field of fault diagnosis.Experiment results show that a suitable clustering center can be selected among fault samples as initial clustering center by AP clustering algorithm,therefore,AP clustering algorithm can be used to solve fault diagnosis problems.(3)In the 5th chapter,the author studies fault diagnosis method based on PCM and KPCM clustering algorithm.Kernel function is referenced to map the data samples from low dimensional space to high dimensional space in order to classify them easily.Combined with PSO algorithm,KPCM clustering algorithm can be able to find a suitable clustering center with fewer iterations.Concluded from the experiment results,PSO-KPCM algorithm can be used in the field of fault diagnosis,and it even has better results and faster convergence rate than FCM,WFCM,AP clustering algorithm.The final chapter of the article includes the conclusions of the whole article and the problems which exist during the process of research and thesis writing.