Research On Fault Diagnosis Of Double Slip Switches Based On Bayesian Network

The double slip switches,which is a large amount of laying in the arrival field,the classification yard and some important connecting throat areas,is the most complicated type of turnout with outdoor structure.As an important device to realize the change of driving vehicle lines in railway transportation,once the double slip switches system fails during the work process,it would cause the train to be delayed and even the train would derail and other major accidents,which seriously endanger passengers' personal safety.Its fault diagnosis and daily maintenance quality will have a direct impact on railway transportation efficiency in my country.When the double slip switches fail,the indoor faults of the switch are often checked faster,but the outdoor faults are mostly based on the work experience of field staff,sothe daily maintenance strategy for double slip switches lacks pertinence.In this thesis,obtained the common fault types of the double slip switches system by analyzing the fault text data of the double slip switches;on this basis,a fault tree model of the double slip switches system is constructed;The model was transformed into a bayesian network,using the existing data for parameter learning,and using the bayesian network to perform reliability analysis and fault diagnosis on the double slip switches,which provided a reference for the daily operation and maintenance of the double slip switches.The main work and content of the thesis are as follows:(1)In this thesis,the fault data of the double slip switches recorded in the “Registration Book of Driving Equipment(Construction&Inspection)”(Yuntong-46)for a railway group for three years from 2016 to 2018 are used for research.Compared with the detection data of the switch system,the text data has the advantages of high authenticity,low acquisition cost and no noise interference.Through the text mining process,sort out the fault type of the double slip switches,and at the same time generate the experimental data set that can be recognized by the computer.(2)In view of the lack of targeted problems in the daily maintenance of double slip switches,based on the results of text data mining,combined with the fault tree analysis method,the failure mechanism of the double slip switches was sorted out and the reliability analysis was performed to find the double slip switches.The weak link in the system lays the foundation for the subsequent design of the bayesian network structure of the double slip switches.(3)The bayesian network with strong learning ability is used to conduct fault diagnosis research on compound intersections.First,the structure design of bayesian network is completed based on the model of the double slip switches fault tree.bayesian network parameter learning,comparative analysis of its results to verify the accuracy of bayesiannetwork parameter learning results.Combined with the reliability analysis results of double slip switches based on fault tree,it provides reasonable suggestions for the daily maintenance strategy of double slip switches.(4)Using the established bayesian network model for fault diagnosis and reasoning,the reasoning results are analyzed and compared to provide guidance for the double slip switches fault repair and improve the efficiency of double slip switches repair.