Fault Diagnosis And Status Assessment Based On Bayesian Network For Running System Of Urban Rail Vehicle

The operating conditions of large dynamic transportation are more complex,the dynamic and static loads affect the reliability of the system.When the service time of the transportation vehicle is gradually accumulated,the performance of each part gradually degrades,and some of them will reach their safety limit,so the whole system will face the safety problem of continuous application.In view of this phenomenon,industry and academia have paid attention to the reliability,availability and maintainability of components of large equipment for a long time,and conducted comprehensive research from the perspective of equipment in-service condition detection and maintenance countermeasures.City rail vehicle is integrated into a complex dynamic transport system with the multi-system and multi-disciplinary.Its running system is composed of framework device,wheel on the axle box device,suspension device,driving device,braking device,coupling device,accessory device and so on,and involve many parts.The failure mode is occurred in a variety of forms,affect the safety of the operation of vehicle system.All subway companies are actively dealing with the risks and safety of vehicle system operation,and gradually establishing a health management platform for system operation and a service status information detection network,so as to establish a reliability-centered health management system with predictable,maintainable and guaranteed failures.This thesis takes the urban rail vehicle running system as the research object,discusses the connection relationship between parts of the running system,forms the structure hierarchy of the running system and the corresponding cascade fault synthesis,and analyzes the potential fault forms.According to the level of different parts of system structure,the occurrence of fault can be divided into system-level fault,functional components,parts and components failure,based on the design of fault cascade relationship,structure of the multilayer fault tree model,and based on multi-level fault tree model is established in this paper,Bayesian diagnostic network possible faults of running system to predict or to maintain fault has occurred.Taking wheelset axle box components of the running system as an example,the FullBNT toolbox of the Matlab software system was used to realize Bayesian network simulation model,and combined with the connection tree engine,to carry out the causal reasoning and diagnostic reasoning analysis of wheelset axle box faults.Due to the complex structure of modern tram walking system,the correlation between components and the situation that it is difficult to assess the health status of the system caused by operating in different environments,this thesis adopts the method of combining conventional AHP and fuzzy comprehensive evaluation(FCE)to evaluate the health status of the running system.Hierarchical analysis method is used to classify the equipment structure,determine the evaluation index of each subsystem and calculate the corresponding weight vector.The health status membership vector of system components is obtained by using fuzzy comprehensive evaluation method,and a membership function evaluation system is constructed by using the health value to evaluate the health status of mechanical equipment.