Reaserch On Bayesian Network Based Fault Diagnosis Of ZPW-2000K Jointless Track Circuit

In recent years, with the rapid development of high-speed railway, The speed of train has been improved, and the safety of the railway transportation has become more and more important. ZPW-2000K jointless track circuit system is an important infrastucture of high-speed railway signal system, which plays a very important role in ensuring train safety and the efficiency of railway transportation. Therefore, the normal operation of the ZPW-2000K jointless track circuit is an important method which can ensure the safety of railway transportation.Currently, in the fault diagnosis applications of the ZPW-2000K track circuit, the railway electricity maintenance personnel mainly adopts artificial fault diagnosis, the main problem is that the the intelligent level of diagnosis method is still low, which causes to depend on artificial experience seriously. This is not consistent with our status of high-speed railway.Therefore, it is important to research and improve the fault diagnosis technology of the ZPW-2000K jointless track circuit.Based on the above objectives, this paper focuses on this research, the main work is as follows:First of all, thesis analysed the system structure and principle of the ZPW-2000K jointless track circuit, and introduced fundamental knowledge of bayesian network. Considering the uncertainty of the track circuit and the advantage of bayesian network which solves the uncertain probelm, the fault diagnosis system which thesis researchde took bayesin network as an key technology and appliance.Thesis built fault date table by using historic information table which track circuit generated and fault information which the railway electricity maintenance personnel recorded. Besides, thesis analysed how to build the best bayesian network by using fault data and expert experience, and thesis completed parameter learning and reasoning by using the best bayesian network.The track circuit fault diagnosis system is designed for the user interaction layer, data processing layer and data storage layer based on the system principal analysis. On the basis of the analysis of each function module and fault process of the fault diagnosis system, thesis developed the ZPW-2000K jointless track circuit fault diagnosis system by using MATLAB and VS2010 development tools. When fault diagnosis system diagnoses fault, the system will gives the results of fault diagnosis and the corresponding maintenance recommendations which help the maintenance people deal with failure to provide technical support and improve the efficiency of fault diagnosis.