The Safety And Reliability Analysis Methods For Train Operation Based On Hybrid Automata

With the development of train control technology,train control system has made great progress and development.In the design and development process of the safety-critical system,the introduction of formal technology can not only ensure the quality of each development stage to the maximum extent,but also improve the development efficiency and control the development schedule effectively.However,it is still difficult to fully apply formal techniques in terms of ensuring the safety of trains and operations.1)in the preliminary design stage,the established formal model will inevitably contain some unknown parameters since the requirement specification is not very specific.Therefore,how to determine the constraint set of such uncertain parameters on the premise of ensuring the train operation safety is very important during the initial design stage.2)as the train control system is composed of multiple modules,the wireless communication system module plays an increasingly important role in modern train control system.In the moving block control system,the uncertainty of communication performance of wireless communication system may affect the control efficiency and train operation safety.Thus,it is of great significance to evaluate the influence of random events on train operational safety.3)as a probabilistic learning algorithm,intelligent learning algorithm does not have safety control mechanism.It is necessary to introduce a safety mechanism to ensure the safety of train operation when realizing the intelligent driving of multiple trains.4)the current reliability research is not closely related to the system safety analysis process.In addition,the research and analysis models for reliability and safety analysis are often independent of each other,which seriously affects the reliability of the reliability analysis results.In addition,system safety is an important prerequisite for reliability analysis.Aiming at ensuring the safety of train operation,the operational status and dynamic performance of critical equipment are monitored by using formal modeling and verification method,machine learning algorithms and reliability analysis method.The main research results are summarized as follows:1.In order to verify the hybrid automata model with uncertain parameters effectively,the preliminary design train control model is realized by solving the constraint set of uncertain control parameters for the corresponding safety requirements.In addition,a time-bounded online safety verification algorithm is designed to solve the state space explosion problem.By calculating all the executable paths of the hybrid automata control model in a certain period of time in the future,the proposed verification algorithm can reduce the difficulty of applying the traditional formal verification method in engineering practice,so as to monitor the train safety online.2.Based on probabilistic hybrid automata model,a safety monitoring method is proposed to monitor the train operation state in real time.To begin with,the feasible domains of uncertain control parameters and probabilistic event parameters for the corresponding quantitative performance indexes are obtained through a large number of off-line verification process.Furthermore,the dynamic performance index of the system can be obtained by online performance evaluation process.Finally,according to the specific system performance index,the constraints of train parameters under the corresponding operating state are used to evaluate and determine the safety level of the system.3.In order to improve the autonomy and intelligence of the train control process,an intelligent safe driving strategy for multi-train is proposed based on the intelligent driving algorithm of single train.Firstly,the driving data used by intelligent driving algorithm includes the driving data of excellent drivers and the simulated driving data of ATO algorithm.Then,based on the speed hierarchical braking modes,a hybrid automata controller is generated to ensure the tracking safety of multi-trains.In addition,the iterative pruning error minimization algorithm is designed to sparse driving data set to reduce the complexity and improve the generalization ability of intelligent driving model.Finally,the ensemble classification regression tree algorithm is used to discover the potential driving rules from the field driving data.4.To ensure the consistency between the system safety verification and system reliability evaluation,System Modeling Language is introduced to describe the structure and establish the system DFT model.Then,the continuous-time Markov chains(CTMC)model is transformed from the established DFT model to evaluate the system dynamic performance.The hierarchical iteration evaluation method is proposed to improve the calculation efficiency of reliability computation.Finally,the analysis method of Markov model is studied by considering the imperfect failure coverage.