Reliability Analysis Of EMU Traction Drive System Based On Dynamic Bayesian Network

In order to meet the needs of China's social and economic development,the running mileage of high-speed railway is increasing,and the scope of operation is expanding.As one of the important parts of high-speed railway EMUs,the traction drive system traction will cause the EMUs to delay or abnormal stop when it breaks down,which will usually cause great economic loss and safety problems.In the daily maintenance process of EMUs,it is urgent to grasp the real-time change trend of EMUs' real-time operation status,and then make flexible status maintenance plan.The dynamic reliability analysis of EMUs in the operation process is one of the powerful means to grasp the real-time reliability level change of EMUs,which can play an important role in the realization process of condition-based maintenance.Taking the traction drive system of CRH3 EMUs as an example,based on the actual fault data of EMU,this paper proposes a dynamic reliability analysis method of EMUs' traction drive system based on dynamic variable structure Bayesian network method.In this paper,the static reliability analysis of the traction drive system is realized based on the traditional Bayesian N1 etwork,and then the dynamic reliability analysis of the system is realized by using the Structure-Variable Dynamic Bayesian Network(SVDBN).Firstly,according to the fault data of EMUs,this paper analyzes the reliability of the components of traction drive system,obtains the static and dynamic reliability parameters of the components of the system,and provides the data basis for the reliability modeling of the system.Then,according to the reliability logic relationship between the components and the system of the traction drive system,the reliability block diagram of the traction drive system is constructed hierarchically.According to the transformation relationship between the Reliability Block Diagram,Fault Tree and Bayesian Network,the method of directly obtaining the Bayesian Network from the reliability block diagram is obtained,and then transforms reliability block diagram of the traction drive system to Bayesian Network.After that,we extend static Bayesian Network in time sequence,a SVDBN model of traction drive system is constructed by combining the dynamic reliability parameters of components and the extended static Bayesian Network,and the relevant reasoning program is written to realize the reasoning of variable structure dynamic Bayesian network.Finally,according to the inference results of the static Bayesian network model and the dynamic Bayesian network model of the traction drive system,the reliability analysis of the traction drive system is carried out.And then the static and dynamic reliability indexes of the traction drive system are obtained,such as: the average failure rate and reliability of the system;the thin and weak links of the system;the reliability change curve;the fault distribution function of the system;the dynamic change trend of the weak link,Based on this,the maintenance system can be dynamically optimized to make the basic maintenance system more close to the actual service state change rule in the vehicle operation process,and finally transition to the state maintenance.There are 82 figures,48 tables and 73 references in this paper.