Bayesian Network-Based Reliability Analysis Of Metro Train Brake System

Urban rail transit has become one of the most sustainable transportation in urban transit because of its high reliability,safety and large volume.As the carrier of urban rail transit,the safety of metro train plays a very important role in ensuring the safe operation of urban rail transit.Brake system is an important part of metro train,and the normal operation of the brake system is an important guarantee for the safe operation of the metro train.The safety and reliability of brake system affects not only the operational quality and service quality,but also the safety of the passengers and the normal operation of the urban transit.Therefore,the analysis of the brake system in order to increase the reliability of the system is of vital importance.However.because of the complex structure of the brake system,the different location,and varied failure modes,the fault data of the components is difficult to be recorded systematically and the accurate failure rate of each component is difficult to obtain.The traditional reliability analysis method is difficult to adopt.Thus,a reliability analysis method based on Bayesian network is proposed in this paper,and the analysis and research are mainly from the following aspects.(1)Firstly,according to the analysis of FMEA of the next generation metro train brake system and the analysis of the function and failure mode,the fault tree model of the next generation metro train brake system is established.(2)The transformation modeling method of fault tree model to Bayesian network(BN)is put forward,and the transformation mode of different system structure is studied and improved.Based on the fault tree model of the next generation metro train brake system,the BN model of brake system is built.The parameters of the BN are studied by EM algorithm,and the joint tree algorithm is used in the network inference.The reasons and sorts of the brake system failure and the weakness of the next generation metro train brake system are obtained.The weakness is gateway valve,smart valve,compressor set,air dryer unit and so on.The results are helpful to find out the fault components in the maintenance stage of the brake system,which is of great guiding significance for the daily maintenance of the metro train.(3)The dynamic Bayesian network of brake system considering maintenance factors is established.The effect of time factors and maintenance factors on the various variables of the system is not considered into the traditional static Bayesian network.Therefore,this paper establishes a dynamic Bayesian network along the time axis,which combines the maintenance factors with the conditional probability table between the time slices of the dynamic Bayesian network,and simulated by the GeNIE software.Based on the analysis and inference of dynamic Bayesian networks,the reliability allocation method based on inference results and the formulation of preventive maintenance cycle based on the key components under given reliability are proposed.And the method will assist the maintenance personnel in optimizing the maintenance strategy.Meanwhile the maintenance cost will be reduced and the reliability of the system will be improved.