基于PET极值统计的高速公路车道变换行为安全性研究  

Study on Safety of Lane Changing Behaviours on Expressway Based on PET Extreme Statistics

作　　者：郑来[1] 邓晓庆[1] 孟祥海[1] 

机构地区：[1]哈尔滨工业大学交通科学与工程学院,黑龙江哈尔滨150090

出　　处：《公路交通科技》2016年第8期120-126,共7页Journal of Highway and Transportation Research and Development

基　　金：中央高校基本科研业务费专项项目(HIT.NSRIF.201660);广东省交通运输厅科技项目(2012-2013)

摘　　要：以高速公路基本路段上车道变换过程中后侵入时间(PET)的界定、测量和分析为基础,通过极值统计方法建立了车道变换PET的超阈值模型,进而结合模型参数估计构建了百万次车辆变道事故重现水平指标,并将该指标与交通事故数据次数的关系进行了分析。结果表明:百万次变道事故重现水平与交通事故次数之间的皮尔逊相关性系数高达0.77,决定性系数也达到了0.60,二者都明显优于PET=3.0 s所界定的车道变换冲突次数与交通事故次数之间的相关性。这说明通过对PET信息的进一步挖掘和利用,本文所得到的完全不依赖于历史事故数据的安全分析指标具有和交通事故指标近似等效的评价效果,也从侧面进一步证明了交通冲突技术的有效性。Based on defining, measuring and analyzing post encroachment time （PET） of lane changing maneuvers on expressway basic sections, the over threshold models for PETs of lane changing are established based on the extreme value statistic approach. By the parameter estimation of the over threshold models, the accident recurrence level of one million vehicle lane changing maneuvers is built, and the relationship among accident recurrence level and traffic accidents data is analysed. The result shows that （1） the Pearson＇s correlation coefficient between the accident recurrence level of one million vehicle lane changing maneuvers and the number of accidents is as high as 0. 77, and the coefficient of determination is 0. 60, both of the coefficients are better than the correlation between the number of lane changing conflicts determined by PET = 3.0 s and the number of accidents. This indicates that through further analyzing and utilizing the PET information, the proposed safety analysis index obtained without using the historical accident data performs almost as well as the traffic accident indexes, and this also proves the effectiveness of the traffic conflict technique.

关 键 词：交通工程 超阈值模型 极值统计 车道变换行为 高速公路 后侵入时间 

分 类 号：U491.31[交通运输工程—交通运输规划与管理]