基于Logistic模型的双车道公路事故易发路段预测研究

事故易发路段的判别是安全管理的重要内容之一。传统的事故易发路段判别方法多应用于运营阶段的公路,其判别精度很大程度决定于历史事故记录的周期和质量,且受"回归到平均"等现象的影响。对于实际的安全管理工作,预先对某一对象成为事故易发路段的可能性进行预测,进而预先采取安全改进措施,能够使得安全完善工作由被动变主动,还可增加事故易发路段判别工作的应用阶段。通过对大量双车道公路事故易发路段样本的整理和分析,发现接入口密度、道路线形、交通组成等要素对事故易发路段的产生有很大影响。为此,借助Log istic模型在概率预测上的优势对上述要素对事故易发路段形成的影响情况进行了研究,建立了事故易发路段预测模型,并进行了实际检验。

A Study on Prediction Model of Accident Prone Location of Two-Lane Highway Based on Logistic Model

Accident prone location identification is an important part in safety management.Tradition methods are often used in the running roads.The validity of those methods is based on the quality of the accident data,and is influenced by the phenomenon named "Regression to Mean".It is important to predict the probability of being an accident prone location in actual safety management works.Many safety improvement works can be done based on the analysis results.This will be more active to safety management works and the methods can be used widely.It is found that accident prone locations are influenced by the driveway density,alignments,traffic composition,and other elements through the analysis of many accident prone locations.So some study works are done based on the logistic model.The prediction model of accident prone location model is advanced.Some test works are done too.