基于空间聚类的城市路网事故多发区域共性特征分析应用  

Analysis of Common Features of Accidents-prone Spots in Urban Roads Network and Corresponding Application Based on Spatial Clustering

作　　者：樊鹏程 董宪元 FAN Pengcheng;DONG Xianyuan(College of Transportation Engineering, Tongji University, Shanghai 201804, China;Tongji University, The Key Laboratory of Road and Traffic Engineering, Ministry of Education, Shanghai 201804, China;Tongji University, Ministry of Education & Engineering Research Center of Road Traffic Safety and Environment, Ministry of Education, Shanghai 201804, China;Shanghai Key Laboratory of Rail Infrastructure Durability and System Safety, Shanghai 201804, China)

机构地区：[1]同济大学交通运输工程学院,上海201804 [2]同济大学道路与交通工程教育部重点实验室,上海201804 [3]同济大学道路交通安全与环境教育部工程研究中心,上海201804 [4]上海市轨道交通结构耐久与系统安全重点实验室,上海201804

出　　处：《交通科技》2019年第4期85-89,共5页Transportation Science & Technology

摘　　要：为了针对性地识别与治理城市道路事故黑点,需要深入挖掘事故特征。文中采用国内某市内城市道路交通事故数据,分别提取年事故数、每公里事故数为指标,利用空间聚类算法对事故点进行聚类,经过数据可视化得到城市道路交通事故风险分布的聚类结果图并进行特征分析。结果表明,该方法可以有效关联事故多发位置鉴别与特征分析,能在路网层级准确识别事故多发区域,为城市道路交通安全管理提供决策指导。In order to identify and ameliorate the accident black spots targetedly, it is imperative to extract the accident risks distribution apart from its level in the scopes of the urban networks. Based on the data sets of accidents happened in urban roads from a domestic city, accidents-prone spots clustering was completed by means of spatial clustering method with adopting the index of the number of annual accidents and the number of annual accidents per kilometer respectively. The clustering result diagrams depicting the accident risks distribution in the whole urban networks were also obtained to analyze features. The results show that the proposed method can correlate the identification to the features analysis of accident-prone locations, and accurately identify accident-prone locations at the level of road networks. Eventually it will contribute to providing decision guidance for urban road traffic safety management.

关 键 词：交通事故 空间聚类 事故挖掘 特征分析 可视化 

分 类 号：U491[交通运输工程—交通运输规划与管理]