基于信息融合的轿车碰撞两轮车事故车速分析  

An Analysis on Vehicle Speed in Car-bicycle Accidents Based on Information Fusion

作　　者：袁泉[1] 郭锐[1] 胡晓佳[1] 李一兵[1] 

机构地区：[1]清华大学汽车安全与节能国家重点实验室,北京100084

出　　处：《汽车工程》2009年第12期1169-1172,共4页Automotive Engineering

基　　金："十一五"国家科技支撑计划项目(2007BAK35B06)资助

摘　　要：在对北京市轿车与两轮车碰撞事故调研分析的基础上,深入分析车辆、路面遗留痕迹特征与汽车碰撞特点之间的联系,选择与轿车碰撞速度相关的人体-风窗玻璃接触、人体及车辆与路面接触的痕迹特征参数,运用人工神经网络方法建立了轿车与两轮车碰撞事故的车辆碰撞速度分类预测模型。该模型融合了人-车-路相互作用的痕迹特征信息,采用实际事故案例提取的可靠样本对其进行训练并应用于车速估算。Based on the investigation into car-bicycle accidents in Beijing, the relations between vehicle impact characteristics and trace features remained on vehicle and road surface are analyzed. By using ANN method, the car impact speed related trace feature parameters of contacts between human and windscreen, human and road surface as well as vehicle and road surface are selected to build a car impact speed classification/prediction model for car-bicycle accidents. The model, merging together the information of trace features mentioned above, is trained by the reliable samples extracted from real accident cases and applied to vehicle-speed estimation.

关 键 词：轿车两轮车碰撞事故 车辆碰撞速度 信息融合 人工神经网络 

分 类 号：U491.31[交通运输工程—交通运输规划与管理]