基于几何线形和车速的高速公路事故预测及应用研究

发布时间：2018-03-03 08:00

本文选题：高速公路　切入点：线形指标　出处：《哈尔滨工业大学》2015年硕士论文　论文类型：学位论文

【摘要】：通过对调查数据的深入分析,分山岭区、重丘区和平原区三种地形条件,研究了高速公路几何线形、车速与事故率的关系,建立了基于负二项分布的事故预测模型,并结合敏感性分析给出了线形优化和车速控制的建议,具有重要的实际应用价值。从高速公路路政管理部门收集到共计23001起交通事故数据;依据“同质法”对上述高速公路进行了事故预测单元划分,共得最小单元15148个;从流量观测站得到了事故预测单元近几年以来的年平均日交通量数据。整合处理历年来的交通事故数据、几何线形数据、车速数据以及交通量数据,构建一一对应关系,建立了不同地形条件下高速公路事故率影响因素分析数据库。采用回归分析的方法,分山岭区、重丘区和平原区三种地形条件,分析单一线形指标与亿车公里事故率的关系,并在此基础上分析了组合线形指标(弯坡组合路段、平竖曲线组合路段、长大坡路段等)与事故率的关系,并据此确定了事故率最低时所对应的线形条件,即本文所指的理想线形条件。采用回归分析的方法分析了山岭区、重丘区和平原区高速公路平均速度、运行速度、大小型车速度差、车速离散度与事故率的关系。结果表明高速公路事故率随平均速度、运行速度、大小型车速度差、车速离散度的增加而显著增加,且山岭区高速公路事故数高于其他两种地形条件下的事故率。分析了Poisson系列模型的适用特点,针对交通事故分布较为离散的特性,选取了Vuong统计量、α的t统计量、AIC、BIC等参数作为模型选择的依据,最终构建了基于负二项分布的适用于不同地形条件的交通事故预测模型。事故预测结果表明对于具有一定长度的高速公路而言,事故预测模型具有比较高的预测精度。依据局部敏感性分析的原理,分析了高速公路事故预测模型中各变量的敏感系数大小。结合敏感性分析,用事故数增长系数的均值代表可以接受的安全水平上限。以山岭区和重丘区高速公路平纵组合路段为研究对象,得出基于行车安全的高速公路平纵组合路段关键线形指标的取值的推荐值。以山岭区和重丘区高速公路上下坡路段和平曲线路段上的运行速度为研究对象,得出基于行车安全的高速公路上下坡路段和平曲线路段车速控制的推荐值。
[Abstract]:Based on the deep analysis of the investigation data, the relationship among the geometry, speed and accident rate of expressway is studied, and the accident prediction model based on negative binomial distribution is established, which is divided into three kinds of terrain conditions: mountain area, heavy hill area and plain area. Based on the sensitivity analysis, the suggestions of linear optimization and speed control are given, which have important practical application value. A total of 23001 traffic accidents data were collected from highway administration department. According to the homogeneity method, the accident prediction units of the freeway mentioned above are divided into 15148 minimum units. The annual average daily traffic volume data of the accident prediction unit in recent years are obtained from the flow observation station. The traffic accident data, geometric line data, speed data and traffic volume data are integrated and processed over the years, and a one-to-one correspondence is constructed. In this paper, the analysis database of the factors influencing the accident rate of expressway under different terrain conditions is established. By means of regression analysis, the relationship between the single linear index and the accident rate of 100 million vehicles per kilometer is analyzed by means of regression analysis, which is divided into three topographic conditions: mountain area, heavy hill area and plain area. On this basis, the relationship between the combined linear index (bending and slope combination, horizontal and vertical curve combination, long slope section, etc.) and the accident rate is analyzed, and the corresponding linear conditions when the accident rate is lowest are determined. The method of regression analysis is used to analyze the average speed, running speed and speed difference of large and small vehicles in mountain area, heavy hill area and plain area. The result shows that the accident rate of expressway increases significantly with the increase of average speed, running speed, speed difference of large and small vehicles, and the increase of speed dispersion. The accident rate of expressway in mountain area is higher than that under the other two kinds of terrain. This paper analyzes the applicable characteristics of Poisson series models, aiming at the discrete characteristics of traffic accident distribution. The parameters such as Vuong statistic, 伪 -t statistic and AIC-BIC are selected as the basis of model selection. Finally, a traffic accident prediction model based on negative binomial distribution is constructed, which is suitable for different terrain conditions. The result of accident prediction shows that for freeway with certain length, According to the principle of local sensitivity analysis, the sensitivity coefficient of each variable in the accident prediction model of expressway is analyzed. The upper limit of acceptable safety level is represented by the mean value of the increase coefficient of accident number. The horizontal and vertical combination section of expressway in mountain area and heavy hill area is taken as the research object. The recommended value of the key linear index of the horizontal and vertical combination section of expressway based on driving safety is obtained. The running speed of the downhill section and the curve section on the downhill section of the expressway in the mountain area and the heavy hill area is taken as the research object. The recommended value of speed control on the downhill section and curve section of expressway based on driving safety is obtained.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：U491.31

【参考文献】