Research On Railway Switch Fault Diagnosis Based On VMD-EEMD-KFCM Model

China’s railway has made a historic leap in recent decades.The mileage and speed of the railway have increased greatly,in which the signal equipment plays an important role.The turnout,as the basic signal equipment of high-speed railway,is the premise of the operation of all signal systems to ensure its safe,stable and efficient operation.If there is a problem in the working state of the turnout system,it will greatly affect the operation safety and transportation efficiency of the train.Therefore,a series of methods that can quickly judge the operation state of the turnout system are needed,and the alarm prompt can be given in time in case of turnout failure.At present,from the actual operation of the high-speed railway on site,there are many problems in the fault of the turnout system,whether it depends on the maintenance personnel to judge directly according to experience on site or the maintenance personnel to judge according to experience from the current power curve of the switch machine of the turnout conversion equipment.First of all,these methods are affected by the on-site experience and knowledge level of the staff,which are prone to misjudgment and omission in fault diagnosis.In addition,the high-speed railway line is complex,the number of turnouts is large,and relying only on manpower obviously does not meet the needs of future railway development.In addition,the operation environment of the turnout system is complex and changeable,including special environments such as plateau,rain and snow,This also increases the difficulty of manual fault diagnosis.Under this background,by studying the structure and working principle of turnout switch system,consulting data and comparing the fault diagnosis methods in other fields,this paper puts forward an intelligent fault diagnosis system suitable for turnout switch machine.Its purpose is to eliminate the influence of human and environmental factors,and make the diagnosis process efficient and real-time.The main research work is as follows:First is the preparation stage.S700 K switch machine,which is most used in the transfer and speed-up turnout,is selected as the research object.Its internal structure and working principle are analyzed.It is found that the output tension and other parameters of the switch machine can directly reflect the operation state of the turnout system.Through the microcomputer monitoring system in the railway,the current and power curve of one action of S700 K switch machine can be directly obtained.According to the formula derivation,the conversion relationship between output power and output tension of switch machine can be obtained.Then,through the steps of on-site records,literature review and data collection,the parameter information and treatment methods of common turnout fault types are summarized,and the fault database is established.The second part is the data processing part.According to the obtained power curve of S700 K switch machine,it is proposed to decompose the power signal of switch machine with variational mode decomposition(VMD),select and reconstruct the effective component according to the correlation coefficient,calculate the sample entropy and root mean square value of the component as the characteristic parameters,and construct the characteristic vector.At the same time,in order to form a comparison,the ensemble empirical mode decomposition(EEMD)is selected to process under the same conditions.EEMD can adaptively decompose the power signal of the switch machine to obtain the intrinsic mode function(IMF)with frequency ranging from high to low.The first few items are selected to construct the eigenvector according to the same method.The above feature vector sets are clustered by kernel fuzzy C-means(KFCM)clustering algorithm to obtain the clustering center of each state type of turnout.The sample type can be determined according to the minimum principle by calculating the Euclidean space distance from the sample to the clustering center.According to the simulation results,the advantages and disadvantages of the two decomposition algorithms in turnout fault diagnosis are analyzed.After the comparison of simulation experiments,it is found that the combination of the two decomposition algorithms can improve the accuracy of fault diagnosis.Therefore,a turnout fault diagnosis system based on vmd-eemd-kfcm model is established in this paper,which includes the establishment of database,the design of system structure,the decomposition and clustering of power signals,the comparison and output of results and other functions.Finally,the power signal of S700 K switch machine in a certain electric service section is input as a test sample to verify the feasibility of the system fault diagnosis function and compare it with other methods.According to the analysis of the simulation results,it is proved that the system has higher accuracy in diagnosing the faults of the turnout system according to the power signal of the switch machine.