Research On Bayesian Networks In The Application Of Rail Switch Control Circuit Fault Diagnosis

The development of the high speed railway forces people to change the traditional rail maintenance models, and the post-repair for rail switch control circuit fault would take more operation time. Especially, the complexity of the swith environment, failure causes are complex and changeable, and there are obvious random uncertainty and fuzziness, these leads to not repair the failures in time, thus, implementation of the intelligent switch control circuit fault diagnosis has been a key resource to protect the safety of switch equipment.Bayesian networks is one of the most efficient theoretically models on uncertain things expression and inference field. By using bayesian networks’advantages of learning ability, solid theoretical foundation, high efficiency reasoning algorithm, it will be introduced into the yard to solve problems about fault diagnosis.Mainly completes below works:First of all, S700K switch control circuit is choosed as the research object. The composition and principle of switch control circuit are analyzed, and the switch control circuit fault types and characteristics are summarized, this thesis gives some failure analysis and treatment measures finally.Then, application of Bayesian networks in fault diagnosis is researched. Through the three kinds of fault knowledge acquisition modes, a large fault knowledges are acquired, then combining with the characteristics of the bayesian theory, the thesis puts forward to building the three layer diagnosis network model which is based on "failure cause-failure mode-failure effect " causal relations. Using clique tree propagation to calculate, this thesis shows the part of fault diagnosis results of no evidence input and there is evidence input, the model in fault diagnosis field is effective and practical.Finally, this thesis designs the fault diagnosis system. Switch control circuit fault diagnosis system based on Bayesian networks is built, and background database is established by Access, the system is carried out with Visual C++6.0, including network modeling, diagnostic reasoning, repair decision making, etc. Through the fault data, using Bayesian networks fault system of diagnosis, diagnosis results can be confirmed by this method is practical and feasible, this system can rapidly deal with the different aspects of fault information in time, greatly improving the efficiency of fault diagnosis.