Research On Reliability Of Catenary Based On Dynamic Bayesian Network

With the increasing transportation volume of electrified railway in China,more and more requirements and challenges about the safety of train operation are put forward.As the key carrier for the transmission of required electric energy to trains,catenary is the core and weak link of electrified railway.Therefore,it is significant to carry out research on the reliability of catenary to improve the quality of train operation and ensure the traffic safety.On the basis of summarizing the current research status of the reliability of catenary system,aiming at the complex fault mechanism of catenary and the deficiency of traditional reliability research methods,this thesis proposes to apply dynamic Bayesian network theory to the reliability study of catenary system.The specific research work is as follows:Firstly,several key technologies of reliability modeling of dynamic Bayesian network are studied.In general,the technical route of reliability modeling and analysis based on dynamic Bayesian network is established.Then the mapping rules of fault tree model to traditional dynamic Bayesian network and variable structure dynamic Bayesian network are emphatically studied,which is the key to establish the topological structure and conditional probability distribution of the catenary dynamic Bayesian network reliability model.Furthemore,the reliability estimation and fault diagnosis formulas of dynamic Bayesian networks are derived.Secondly,the dynamic reliability of catenary system when the failure time follows exponential distribution is studied with the traditional dynamic Bayesian network.Through the analysis of the structure and fault propagation law of catenary system,its fault tree model is constructed.Based on this,its dynamic Bayesian network reliability analysis model is constructed with the mapping rules from fault tree model to traditional dynamic Bayesian network.And the bidirectional reasoning is carried out for the model.The dynamic change law of reliability and availability of catenary system with time is obtained by forward reasoning,the fault diagnosis of catenary system is completed by reverse reasoning,and the weak parts affecting the system reliability are found,which can provide guidance for the daily maintenance of catenary.Comparing the reliability estimation results of catenary system with those calculated by using Markov analysis method,which verifies the effectiveness of the traditional dynamic Bayesian network reliability modeling method proposed in this paper.Then,the dynamic reliability of catenary system under the condition of non exponential distribution of failure time is studied based on variable structure dynamic Bayesian network.The mapping rules from fault tree model to variable structure dynamic Bayesian network is used to dynamically estimate the reliability level of the catenary system and effectively judge the weak links.Comparing the reliability estimation results of catenary system with those calculated by Monte Carlo simulation,which verifies the effectiveness of the variable structure dynamic Bayesian network reliability modeling method proposed in this paper.Finally,the dynamic Bayesian network reliability analysis software of catenary is designed and developed.With this software,the reliability prediction and fault diagnosis of catenary system at any time can be realized quickly.Moreover,the software is presented in the form of GUI with simple and friendly interface,which is convenient for engineers to use.