基于经验贝叶斯的事故多发信控交叉口判别.docVIP

  基于经验贝叶斯的事故多发信控交叉口判 别# 王雪松，李佳* 5 10 15 20 25 30 35 40 （同济大学 道路与交通工程教育部重点实验室,上海 201804） 摘要： 信控交叉口是交通事故的多发点，判别事故多发信控交叉口是安全改善的基础。由 于交通事故是小概率事件，事故绝对数具有随机波动的特性，同时传统的事故绝对数法判别 出的往往是几何尺寸、交通流量较大的交叉口，安全改善空间有限。经验贝叶斯方法结合事 故观测数和类似地点的事故期望估计事故发生数，消除了事故绝对数的随机波动性。在此基 础上，采用安全可提高空间判别方法，计算经验贝叶斯方法估计事故数与类似地点事故期望 数的差值，对事故多发信控多发交叉口进行判别；差值越大，表明可以改善的潜能越大。本 文基于上海市信控交叉口，采用经验贝叶斯方法，基于安全可提高空间对事故多发信控交叉 口进行判别。 关键词：交通工程；信控交叉口；事故多发点判别；经验贝叶斯；安全可提高空间 中图分类号：U491 Signalized Intersection Hotspot Identification Based on Empirical Bayes WANG Xuesong, LI Jia (Key Laboratory of Road and Traffic Engineering of the Ministry of Education, Tongji University, Shanghai 201804) Abstract: Signalized intersections are traffic crash hotspots in road network. Hotspots identification is the basis for safety improvement. Traffic crash is small probability event which will lead to the Crash Frequency’s random fluctuation during hotspot identification and intersections which have large scale and volume are often identified. While the Empirical Bayes method combines clues from both the crash history of a specific site and expected crash of similar sites to estimate the crash number which can overcome this problem. Potential for Safety Improvement method identifies signalized intersection hotspots with the difference between the Empirical Bayes estimated crash number and what is normal for similar sites obtained from regression model. The larger the difference is, the more Potential for Safety Improvement is. Based on signalized intersections’ crash data in Shanghai and Empirical Bayes, the Potential for Safety Improvement is used to identify signalized intersection hotspot. Key words: traffic engineering; signalized intersection; hotspot identification; Empirical Bayes; Potential for Safety Improvement 0 引言 信控交叉口是道路网络中的事故多发设施。判别事故多发信控交叉口，是进 行路网安全改善的基础。传统事故绝对数法根据事故数的多少进行排序忽略了事 故发生的随机波动性[1]，并且筛选出的往往是几何尺寸、交通流量较大的交叉口， 而这些交叉口的安全改善空间有限。 基金项目：高等学校博士学科点专项科研基金（20090072120016） 作者简介：王雪松(1977-),男,教授,交通安全研究,交通运输规划与管理. wangxs@tongj