Reliability Analysis Of EMUs Braking System Based On Fuzzy Multi-state Bayesian Network

"A country with convenient transportation,railway first." With the vigorous development of human society,China’s high-speed railway has developed from the initial introduction of foreign technology to independent research and development that shocks the world.The railway network is becoming denser and denser,and there are more and more EMU trains.Run through all the lines in all directions.In this context,the safe and economical operation of EMUs has also become one of the focuses of attention.As one of the EMU subsystems,the brake system is related to the safe operation of the entire EMU,and its importance is self-evident.Therefore,the reliability of the EMU brake system is analyzed,and optimize reliability costs under the premise of reliability assurance.It is of great significance to consider the economy based on safety.The diverse working conditions,functional levels,and operating relationship of the internal structure of the brake system are extremely complicated.It is affected by a variety of factors,such as wear,vibration,fatigue,impact,and excessive maintenance.In the actual operation process,there are multiple failure states,multiple failure modes and different failure levels between normal operation and complete failure of components;Historical operation data presents multiple characteristics such as time and space multi-scale,small sample characteristics,and so on,resulting in poor data availability;Diversified working environment and its complexity make the system and component failures not obey specific laws,the logical mechanism is unclear,and the failed relationship is uncertain.In addition,how to achieve high reliability(low failure rate)while keeping resource costs low is one of the urgent problems to be solved.In response to existing problems,this thesis has done the following:(1)Based on combing the related research on the reliability of EMUs,research on the multi-state of the CRH5 EMU braking system.The expert survey method based on confidence index correction processes the basic data,and calculates the fuzzy failure rate of parts with a higher failure rate according to the expansion principle of fuzzy mathematics,and then applies the logical relationship of the T-S fuzzy fault tree to clear and handle the uncertainty of the event to build model and analyze,calculate the importance,and find the weak link.(2)As the calculation of the T-S fuzzy fault tree is complex and cannot be inferred backwards,a Bayesian network can intuitively express the uncertainty between the system and components,so according to the corresponding relationship,the bayesian fault tree is mapped by the T-S fuzzy fault tree,and evidence theory is introduced to expand the ability of a bayesian network to process uncertain information.Finally,the hyper-ellipsoidal Bayesian network model is constructed to analyze the braking system,the analysis results can provide theoretical guidance for the formulation of maintenance strategies and technological transformation.(3)Based on the Bayesian network analysis method and reliability cost function,with the lowest reliability cost as the objective function,the system failure rate is the constraint condition,combining the bat-ant colony algorithm to optimize the reliability of the system,to get the optimal reliability index.The optimization model based on a fuzzy Bayesian network reduces the difficulty of modelling system reliability optimization,and the optimization results can provide a theoretical reference for improving the economy of the EMU braking system.