Reliability Analysis Of ### Traction Drive System Based On Bayesian Networks

With the rapid development of high-speed railway, more high-speed EMU would be put into use, and the high density run of EMU would bring tremendous workload to the maintenance staffs. To make a scientific maintenance plan, studying the reliability of high-speed EMU is critical. Reducing the maintenance frequency and improving the maintenance efficiency. Traction drive system is the core power train technology and equipment system, which directly affects the reliability and stability of train operation. Therefore, to ensure the safe and stable run of the train and safeguard the normal order of rail transport, studying the reliability of traction system has engineering significance.Firstly,### EMUs are the research object in this paper, which run on the*** line. all the fault data is analyzed which were collected from the beginning operation in &&to the end of &&. And in order to find a suitable distribution, the fault data is fitted by using different distribution functions. The reliability characteristic function of the main components existing can be gotten based on the failure distribution function which were gotten by using parameter estimation. Moreover, the reliability characteristic function curve can be plotted and the law of fault can be found.Secondly, the method of Bayesian network is used, and the process of transformation from fault tree to Bayesian networks is given. The Bayesian network analysis models of traction converters and high voltage electrical subsystem are established based on their fault tree model. And the probability of failure of the system was calculated by using exact inference algorithm bucket elimination method. On the basis of the advantages of Bayesian network diagnostic reasoning, the traction converters and high voltage electrical vulnerabilities were confirmed.Finally, the Bayesian network model of traction drive system is established based on its corresponding fault tree model, and the probability of failure of the entire system was gotten by forward reasoning, then the weak links in the system were identified but based on the diagnostic reasoning. The important element is the key indicator which was used to assess the effects of reliability of components on the reliability of system. The structure importance, probability importance and critical importance of were calculated respectively in this paper, and the effects of each traction drive system unit are analyzed from different aspects, therefore, the units were recognized which should be focused on in daily maintenance. In addition, based on the results of Bayesian network diagnostic, the reliabilities of traction drive system were allocated in this paper, and the results of allocation meet the design requirements, therefore. the validity of this method was testified.