双车道公路事故预测与防治

关键技术研究

Study  on Key Technologies of Accident Prediction

and  Prevention for Two-lane Highways

研究生：唐琤琤

指导教师：郭忠印教授

二○○九年三月

摘要

针对我国双车道公路交通安全现状“公路安全保障工程”实施工作中的需求，论文依托西部交通建设科技项目《公路交通安全手册研究》（合同号 2004 318 223 33－01）和《西部地区公路安全性评价》（合同号 2003 318 223 23），进行了双车道公路事故预测及防治关键技术的研究。

论文首先根据我国双车道公路交通和安全特性，把研究对象分为山区和平原区的普通路段、村庄路段和交叉口，应用交通统计学知识，建立了事故预测基础模型。对用于建模的数据采集区域之外的地区应用事故预测模型提供了标定系数的方法；对历史事故数据采取EB过程以使预测值更接近真实水平；建立了平原区和山区双车道公路部分要素的事故修正因子（AMF）：接入口密度、路基宽度指标、货车比例、路侧危险度等。

然后提出路网归一化处理方法，以及依据预测事故、历史事故进行路网事故多发点段判别的方法和标准。

针对事故多发点段的防治，论文提出安全诊断程序，分析事故成因及特征并提出对策。根据双车道公路事故预测模型及国内外研究成果，评估对策的效果。

论文建立了关于安全改进点和改进方案选择的0－1整数规划模型，采用基于线性规划的分枝定界法求解模型，解决了资金约束下路网安全效益最大化为目标的对策方案排序问题。

最后，依托山东省德州市的区域公路网，将双车道公路事故预防及防治的研究成果应用于实践，并对进一步工作的方向进行了简要的讨论。

关键词：双车道公路，交通事故预测，事故修正因子，公路网，事故防治，公路安全保障工程

ABSTRACT

To  aim at the current status of two-lane highway traffic safety in China as well  as the demands for implementing “Highway Safety Enhancement Projects”, this  dissertation relies on the science and technology projects titled with  “Research on Highway Safety Manual” (contract number: 2004 318 223 33-01) and  “Safety Assessment for Highways in Western China” (contract number: 2003 318  223 23)  and is devoted to the study on key technologies of accident  prediction and prevention for two-lane highways.

Firstly,  taking into account the characteristics of infrastructure and traffic safety,  the two-lane highways are divided into ordinary sections, village sections  and intersections in mountainous area and flat area. The basic accident  prediction models are developed with the application of mathematical  statistics. The method for calibrating the coefficients is provided when the  accident prediction models are applying in the areas which are unrelated to  the collected data for developing models. The EB process is adopted with the  combination of historical accident data so that the predicted value is more  close to the actual one. The accident modification factors (AMF) for some  factors such as access density, subgrade width and percentage of trucks,  roadside hazardous ratings for two-lane highways in flat and mountainous  areas are established.

Then  the dissertation proposed that the different kinds of highways in a network  are disposed as same criterion and also provided the methods and standards  for identifying accident-prone sections based on accident predicton and  history data.

The  dissertation provided the safety diagnose procedure to analysis the  characteristics and causes of the accidents and recommended the  countermeasures to precent the accidents. Base on the accident pridection  models and the research results, the effect of the countermeasures can be  evaluated.

Furthermore,  the 0-1 integer programming model for selecting improvement sites and  improvement measures is developed, and branch and bound method based on  linear programming is adopted for solving the ranking and selection of safety  improvement sites and improvement measures with the objective to maximize the  safety benefits of highway networks subjected to the limited budget.

Lastly,  the research findings about accident analysis and safety countermeasures are  put into practice supported by the regional highway networks in Dezhou of Shandong  Province. Moreover, the further research work on this topic is briefly  discussed.

Key Words:  Two-Lane Highways, Traffic Accidents Prediction, Accident Modification  Factors, Highway Networks, Accident Prevention, Highway Safety Enhancement  Projects