基于加权证据理论的受电弓事故树分析

Fault tree pantograph security analysis based on weighted evidence theory

受电弓一旦发生故障,就很可能造成严重交通事故。在传统事故树的基础上采用加权证据理论算法优化事故树定量分析,融合不同专家意见,处理了概率评估中的不确定信息,结合机务段的技术人员和专家的意见,对受电弓进行风险评估。基于定性定量分析的结果总结出受电弓安全运行的薄弱环节,提出了具体的安全控制措施。

The present paper intends to propose a new approach to the pantograph security analysis by using the evidence theory in hoping to heighten the quantitative analysis efficiency of the fault tree application to the pantograph failure. As is known, there has always been existing a problem in presenting evaluation results of the basic events in traditional fault tree analysis. What is more, the probability evaluation given by the expertise in different fields may not turn out to be the same. As a matter of fact, such problems often exist in the pantograph failure analysis, which are likely to be so serous as tO influence the proper solution to the practical problems, on the condition of merely resort to the traditional fault tree methods. To change the situation, we have proposed a method that intends to join the fault tree analysis with the evidence theory in this paper. The proposed method of ours intends to divide the pantograph failure analysis into the following steps. First of all, we have made a survey of the traffic accidents that took place in recent years caused by the pantograph failure. Secondly, we have worked out a fault tree model based on our investigative survey under the recommendations offered by the corresponding experts. And, thirdly, we have invited two experts in the field and ask them to provide the probability assessments of the basic events according to their own judgment and experience in terms of such items such as the fault, not fault, and/or uncertainty. And, the next, fourthly, the primary probability assessment results were invited to be given by the third-party, or by the on-looking experts to test the degrees to be trusted over the results of the evaluations given by the experts preciously mentioned in the corresponding forms （such as given by expert 1, expert 2, and the others unknown）. And, fifthly, some other corresponding componential parts of uncertain information prefer to be assigned to the focal elements with Bayesian approximation. And, sixthly, the final probabilities of the basic events can be achieved based on the weighted calculation and evidence fusion. And, last of all, it would be natural to figure out the degrees of the structure and probability importance based on the previous work that has been done. It is worthwhile noticing that the data which have been achieved in the significant probability degree analysis are likely to contain some kinds of improbable information assigned by Bayesian approximation. Thus, it can be concluded that the method proposed in this paper is in a position to make the final safety decision more scientific, reasonable and reliable for its advantages of conforming the pantograph existing security risks in time and higher probability.