Reliability Analysis Of EMUs Pantograph System Based On Fuzzy Bayesian Network

With the gradual formation of the high-speed railway network and the proposal of "the Belt and Road" strategy,more and more high-speed Electric Multiple Unit(EMUs)are being put into use,Therefore,as the amount of EMUs trains and running mileage in China continually to increase,the reliability and safety of the operation of existing EMUs has become the primary concern of the people.The pantograph system as a key part of the high voltage traction system of the EMUs,directly obtains power from the contact network,and provides traction power for the operation of the high-speed EMUs.Therefore the pantograph system failure has a significant impact on the safety operation of EMUs,so it is crucial to analyze how to reduce the failure rate of the pantograph system and ensure electric safety and stability of the bow system work of the operation of EMUs,Therefore the reliability analysis of the pantograph is necessary.The EMUs has the characteristics of continuous running distance,fast speed,high running frequency.The EMUs is a complex system with complex working conditions,complex functional structure and diversity of failure modes.The EMU is affected by many factors,such as natural environment,maintenance and the effects of a variety of conditions.There are many intermediate states between the intact and completely failing state of the system and it is a typical multi state and multi failure mode system.Traditional reliability analysis methods cause complex problems such as event polymorphism,uncertainty of components and system failure mechanism and modeling difficulties.Bayesian network has been proved by theory and practice that it is suitable for reasoning and analysis of uncertain events in the process of reliability analysis,Therefore it has been widely applied in complex system reliability analysis.However,due to the heterogeneity and missing data of historical data,it is still difficult to model and analyze the reliability of the EMU based on Bayesian network.The main contents that the uncertainty of fault data in the process of reliability analysis of EMU,the polymorphism of components and events,the unclear mechanism of failure and the complexity of computation in the reliability analysis of EMUs complex system.The T-S fuzzy fault tree is used to construct a fuzzy system of Bayesian network reliability analysis model for complex systems,the weak linles which affects its reliability are found out by analyzing the pantograph system.This thesis carries out the following research work:1.According to the current research status of reliability analysis methods and reliability analysis of high-speed EMUs,this thesis combined with the actual analysis of the existing EMUs,the basic theories of fault tree and Bayesian networks and their advantages and disadvantages in reliability analysis are introduced in detail.And Based on the existing research and expand the two,proposes to use T-S fuzzy fault tree to build a fuzzy Bayesian network reliability analysis model for EMU reliability analysis.2.For the reliability analysis of complex systems and the failure mechanism is unclear,Therefore the variety of fault states is introduced.The T-S fuzzy model is introduced combined with the fuzzy theory,a T-S fuzzy fault tree reliability analysis model is constructed;multi-source heterogeneity for fault data with data missing and other issues,combined with the existing expert knowledge base,an expert survey method based on confidence index correction was proposed to process the basic data to improve the accuracy of reliability analysis.Combined with the T-S fuzzy fault tree,the reliability analysis of complex systems was carried out.3.Because the computation of T-S fault tree is complex,it cannot express the uncertain knowledge,and it cannot reverse reasoning.According to certain mapping rules,the T-S fuzzy fault tree is transformed into polymorphic fuzzy Bayesian network,and the evidence theory is introduced to enhance the analysis.During the process,uncertain knowledge is expressed,a fuzzy Bayesian network reliability analysis model is built,the model is analyzed and verified,and the example analysis of the pantograph system is used to prove the authenticity and effectiveness of the analysis method.which enhanced the promotion of fault tree and Bayesian networks in reliability analysis methods.