首页|数据驱动的公路典型风险场景聚类与特征分析

数据驱动的公路典型风险场景聚类与特征分析

扫码查看
为了明晰公路交通中典型风险场景,提高公路测试场景构建和公路安全分析的针对性和指导性.收集不同区域代表性省份5年约60 000条公路交通事故数据作为数据源,筛选确定关键分类变量,分别针对7类典型公路路段和路口开展典型交通冲突形式聚类分析,共获取16类典型风险场景,然后构建场景风险特征表征参数,针对典型风险场景的风险特征进行对比分析,进而深入分析道路和环境因素对场景事故数量和发生事故严重程度的影响特征.结果表明:路表情况、防护设施类型、交通信号方式、照明条件、天气和能见度等因素都对部分场景的风险度有较大影响,路面状况因素对场景风险度影响不大.
Clustering and Characteristic Analysis of Typical Highway Risk Scenarios Driven by Data
In order to clarify typical risk scenarios in highway traffic,improve the pertinence and guidance of highway test scenario construction and highway safety analysis.The data of about 60 000 road traffic accidents in five years was collected in representative provinces of different regions of China as the data source.The key categorical variables were screened and determined,and the typical traffic conflict forms of 7 typical highway sections and intersections were clustered.A total of 16 typical risk scenarios were obtained.Then the characterization parameters of scene risk characteristics were constructed.The risk characteristics of typical risk scenarios were compared and analyzed,and then the influence characteristics of road and environmental factors on the number and severity of scene accidents were analyzed.The results show that road surface conditions,types of protective facilities,traffic signal modes,lighting conditions,weather and visibility factors have a great impact on the risk degree of some scenes,while the road integrity or not has little impact on the risk degree of scene.

traffic safetyrisk characteristicsK-means clusteringtypical highway risk scenarios

胡伟超、陈艳艳、于鹏程、于士杰、牛世峰

展开 >

北京工业大学城市交通学院,北京 100124

公安部道路交通安全研究中心,北京 100062

长安大学汽车学院,西安 710064

交通安全 风险特征 K均值聚类 公路典型风险场景

国家重点研发计划中央级公益性科研院所基本科研业务费专项

2019YFB1600500111041000000180001221101

2024

科学技术与工程
中国技术经济学会

科学技术与工程

CSTPCD北大核心
影响因子:0.338
ISSN:1671-1815
年,卷(期):2024.24(8)
  • 12