摘要
从国家车辆事故深度调查体系中筛选255起二轮车事故案例进行分析,在事故录像中提取参与方运动信息,利用帧间差法估算速度,提取案例中的时段、路段、天气等环境信息,利用有序Logistic回归法选择相关特征变量,并对其进行系统聚类;根据聚类结果分析车辆速度信息,建立了8种乘用车与二轮车事故危险场景,为自动紧急制动系统仿真测试提供参考依据。
255 cases of two-wheeled vehicle accidents were selected from the National Vehicle Accident In-depth Investigation System for analysis,the motion information of the participants was extracted from the accident video,the speed was estimated by the inter-frame difference method,the time,road section,weather and other information in the cases were extracted,and the relevant feature variables were selected by the ordinal Logistic regression method and clustered systematically.According to the clustering results,the vehicle speed information was analyzed,and 8 kinds of accident risk scenarios of passenger cars and two-wheelers were established according to the actual situation,which provides a reference for Automatic Emergency Braking System simulation test.
作者
李梦迪
李方媛
LI Mengdi;LI Fangyuan(School of Automotive Engineering,Shandong Jiaotong University,Jinan 250357,Shandong,China)
出处
《农业装备与车辆工程》
2024年第1期81-84,共4页
Agricultural Equipment & Vehicle Engineering
基金
山东省自然科学基金面上项目“面向车联网的道路交通事故链阻断关键技术”(ZR2020MG022)。
关键词
交通安全
乘用车
二轮车
事故危险场景
聚类分析
traffic safety
passenger car
two-wheelers
accident hazard scenarios
cluster analysis
