基于K均值聚类的高速公路汽车碰撞事故分析

Statistical Analysis of Car Crash on Freeways Based on K-means Clustering

为优化我国汽车碰撞试验指标和高速公路安全管理,搜集重庆市所辖高速公路在近5 a发生的194起车-车碰撞事故数据,对其进行深度分析。统计出该地区所发生车-车碰撞事故的形态特征,利用K均值聚类方法建立乘员损伤程度同重叠率、质量比、碰撞方位、侵入量之间的相关模型。结果表明,重乘员损伤程度与其他的事故形态特征有显著的相关性,其中,重叠率在1/3～1/2、质量比为0.1～0.5或2～10以及12点碰撞方位、侵入量为1/2～2/3、2/3～1这样的车辆碰撞事故的乘员损伤最严重。

Car collision accidents on freeways happen frequently, of which the damage is very huge. However, there is no in-depth study about the characteristics of ear collision accidents on freeways. In order to optimize the car collision test indicators and safety management of freeways, data on car collision accidents was researched deeply, which was collected on Chongqing freeways in recent five years. Characteristics of the car collision accidents were got through a descriptive statistical way, such as overlap ratio, mass ratio, occupant injury severity and so on. The K-means clustering method was used to analyze the accidents. At last,a correlation model between occupant injury severity and other characteristics was built. The results show that there is a significant correlation between occupant injury severity and other characteristics, and that the accidents of which overlap ratio is 1/3 - 1/2, and mass ratio is 0. 1 - 0. 5 or 2 - 10 ; collision direction is 12,and crush intrusion is 1/2 -1/3 or 2/3 -1 are the most serious.