基于贝叶斯网络的驼峰超速连挂事故分析  

Analysis on Over-speed Coupling Accidents on Hump Based on Bayesian Networks

作　　者：张春民[1] 李引珍[1,2] 陈志忠[1] 杨信丰[1] 何瑞春[1] 

机构地区：[1]兰州交通大学交通运输学院,甘肃兰州730070 [2]西北交通经济研究中心,甘肃兰州730070

出　　处：《铁道学报》2011年第10期7-14,共8页Journal of the China Railway Society

基　　金：国家自然科学基金(60870008);教育部新世纪优秀人才支持计划(NCET-09-0017)

摘　　要：针对驼峰超速连挂事故进行分析。分析对事故有重要影响的线路、调速设备和车辆等因素,考虑到贝叶斯网络能较好解决不确定性问题以及可以双向推理的特点,建立驼峰超速连挂事故树,并将其转化成贝叶斯网络,计算顶上事件的发生概率以及基本事件的后验概率。结果表明,超速连挂事故的主要影响因素为线路纵断面、调速设备制动能力以及新型重载车辆三方面。进一步考虑新型重载车辆的共因失效问题,建立贝叶斯共因失效混合网络,计算并分析新型重载车辆对超速连挂事故的影响程度以及对既有驼峰的适应性。Over-speed coupling accidents on humping process are analyzed.After analyzing the affects of tracks,speed control equipment and rolling stock,which are regarded as the main factors to cause accidents,and taking the characteristics of Bayesian network into account which it is good at solving uncertainty problem and having bidirectional inference,a fault tree of over-speed coupling accident in humping process is established and converted into a Bayesian network.The top event＇s probability and basic events＇ posterior probability is calculated.The results show that the over-speed coupling accidents are mainly caused by the unreasonable vertical section of the hump,the braking capacity of the hump equipment,and the newly-built heavy-haul cars.In addition,this paper considers the common cause failure problem on heavy-haul cars,and sets up a Bayesian mixture network of common cause failures to calculate and analyze the impact degree of the heavy-haul cars to the over-speed coupling accidents and the adaptability to the existing hump.

关 键 词：超速连挂 事故树 贝叶斯网络 共因失效 

分 类 号：U292.28[交通运输工程—交通运输规划与管理]