Reliability Analysis And Evaluation Of Regional Computer Interlocking System

Regional computer interlocking is the key system to ensure the safe and efficient operation of regional traffic.With the development of railway to the direction of information,intelligence and network,regional computer interlocking system has become the main development direction of station interlocking system with its unique advantages(centralized placement of equipment,simplification of inter station connection,reduction of project investment,significant improvement of driving efficiency and section passing capacity,etc.).This is a typical safety-critical system.Once it breaks down,it will affect the driving efficiency in a small way and endanger the driving safety in a serious way.Therefore,the thesis takes the regional computer interlocking system as the research object,uses the dynamic Bayesian network(DBN)and the improved grey clustering model to analyze and evaluate its reliability,which can provide theoretical guidance for the railway reconstruction line and the new line,provide reference for the existing line evaluation,and help to improve the system design and performances,and promote the development of regional interlocking system in China.In view of the shortcomings of Markov model,fault tree analysis(FTA)and Bayesian network(BN)in failure correlation and maintainability,the thesis uses DBN to analyze the reliability of regional computer interlocking system.Firstly,according to the composition and failure mechanism of regional computer interlocking system,and considering the common cause fault(CCF)and repairable factor,the dynamic fault tree(DFT)models of regional two interlocking units and three interlocking units are established.Then,according to the rules of DFT to DBN conversion,the DBN models of regional two interlocking units and three interlocking units are established.The models are used to analyze the reliability of interlocking subsystem,and then select the appropriate number of redundant interlocking units.Based on it,the whole DBN model of regional computer interlocking system is established,so as to analyze the weak links of the whole system.The results show that the two interlocking units are the better way to achieve the redundancy of the interlocking subsystem,and at the same time,the weak link can be transferred to the unit with small influence on the system,which further improves the driving efficiency.According to the diagnosis and reasoning of DBN,CCF is the main cause of the system fault.In order to ensure the high reliability of the interlocking subsystem,it is necessary to strengthen the protection of the system's CCF.When analyzing the weak link of the overall system,the acquisition module,operation indication module and driving module are the weak links of the system,so we should strengthen their maintenance and inspection.In order to evaluate the reliability of the regional computer interlocking system correctly and objectively,the thesis adopts the improved grey clustering model and combines it with a comprehensive weighting method to evaluate the reliability of the system.Firstly,according to the composition of the regional computer interlocking system,the corresponding evaluation system is established,and the evaluation set of each basic unit of the system in the stable period is determined by combining the experience of experts.Secondly,the weight of each basic unit is determined by using the comprehensive weighting method.Then,the reliability of each basic unit in the stable period given by the expert based on the experience is passed through the whitenization weighting function by using the improved grey clustering evaluation model,and combining with comprehensive weight of each basic unit to determine the degree of each grey class to which the system belongs.Finally,it is compared with the evaluation result of the traditional gray clustering model.The result shows that the reliability level of the regional computer interlocking system in the stable period belongs to the fourth gray category,that is,the system can achieve the main functions,and the evaluation result conforms to the technical characteristics of the system and the experts' predictions.Meanwhile,the accuracy and reliability of the evaluation result of the improved model are verified by comparing with the result of the traditional gray clustering model,which can provide references for similar research.