基于关联规则的高速公路纵面线形事故风险概率研究

Study on Risk Probability of Vertical Alignment Accident of Expressway Based on Association Rule

随着汽车保有量持续增加,交通事故问题仍然严重,虽然基于事故的道路线形模型研究比较成熟,但由于事故数据缺失,很多模型是否可行很难判断。以事故统计数据为基础,通过交通事故风险概率的分析研究高速公路纵断面线形指标的合理性。通过对不同分类竖曲线段和纵坡度、坡长不同组合段事故风险概率的分类统计,利用改进的管理规则对不良线形组合和事故风险概率变化趋势进行分析。以支持度和置信度作为关联规则生成的筛选条件,提升度作为关联规则关联性的评判标准,由提升度大小将关联规则的关联性分为4个不同的区间,更加直观地表达道路纵面线形与交通事故之间的关系,并基于行车安全提出了推荐指标范围。结果表明:同向竖曲线上的事故黑点明显多于反向竖曲线;大半径竖曲线接短坡或者长坡接小半径平曲线时事故黑点较多;凸曲线间的直坡段长度在100~500 m之间时事故风险概率较高;凹曲线间的直坡段长度在100~400 m之间时事故风险概率较高;凸曲线直坡长度应与竖曲线半径大小相配合,凹曲线半径应尽量取到60 000 m以上;事故风险概率随着纵坡度的增大呈现先减小后增大的趋势,当纵坡度大于4%之后事故风险概率增大较明显,当纵坡度小于0.5时,事故风险概率较高;当坡段长度小于400 m时,事故风险概率较高,尤其是坡长小于250 m时。

As the number of cars continues to increase, the traffic accidents are still serious. Although the study of accident-based road alignment models is relatively mature, it is difficult to judge whether many models are feasible due to the lack of accident data. Based on accident statistics, the reasonableness of expressway longitudinal section alignment indicators is studied through the analysis of traffic accident risk probability. By classification statistics of the accident risk probability of different classified vertical curve sections and the sections with different combinations of longitudinal slope and slope length, the trends of poor alignment combinations and accident risk probability are analyzed by using the improved management rule. Taking the support degree and confidence degree as the screening conditions for generating the association rule, and the elevation degree as the judging criterion for the association rule relevance, the correlation of the association rule is divided into 4 intervals by the magnitude of the elevation degree, which expresses the relationship between road vertical alignment and traffic accidents more intuitively, and the recommended indicator range is put forward based on traffic safety. The result shows that(1) there are significantly more accident black spots on the same direction vertical curve than on the reverse vertical curve;(2) there are more accident black spots when a large radius vertical curve connects with short slope or long slope connects with small radius flat curve;(3) the accident risk probability is higher when the length of the straight slope section between convex curves is between 100 and 500 m;(4) the accident risk probability is higher when the length of the straight slope section between concave curves is between 100 and 400 m;(5) the length of the straight slope of the convex curve should match the radius of the vertical curve, and the radius should be over 60 000 m as far as possible;(6) the accident risk probability shows the trend of decreasing and then increasing with the increase of longitudinal slope, the accident risk probability increases more obviously when the longitudinal slope is more than 4%, and the accident risk probability is higher when the longitudinal slope is less than 0.5;(7) the accident risk probability is higher when the length of slope section is less than 400 m, especially when the slope length is less than 250 m.