宏观交通安全建模研究与安全影响因素分析

Macro Level Safety Modeling and Impact Factor Analysis

摘要:

由于对规划层面的交通安全研究不足，道路网络特征、区域交通特征及区域其他与交通安全的关系尚不明确，导致在交通规划阶段缺少可用的方法和工具来评价不同规划方案的安全性.依托美国佛罗里达州Orange县的数据对宏观交通安全建模进行了系统研究，基于拓扑学理论计算交通分析小区（traffic analysis zone, TAZ）路网形态结构，提取TAZ层面交通特征和其他影响因素.TAZ在空间上有远近之分，相邻的TAZ在社会经济发展、道路交通特征方面有某种程度的趋同性，距离较远的TAZ之间有某种程度的相异性.贝叶斯条件自回归模型可以分析空间相关数据，利用先进的贝叶斯空间统计模型分析TAZ层面交通安全与影响因素之间的关系.基于事故发生机理考虑，提出将主干路事故和支路事故分别建模的新思路.分析发现新的建模思路能够很好区分不同道路类型的安全影响因素；路网形态结构和安全之间有显著的关系.

Abstract:

Due to the lack of systematic research for traffic safety at the planning stage, the relationship between macro level impact factors such as road network patterns, traffic features and other variables with transportation safety remains unclear, which leads to the shortage of available methods and tools to evaluate the safety of various transportation plans. Based on the data from the Orange County, Florida, this paper describes the system research on macro transportation safety modeling at traffic analysis zone (TAZ) level, abstracting effective indices of road network patterns, traffic features and other macro level impact factors. Bayes Space Static Model was adopted to investigate the association between transportation safety and impact factors. Adjacent TAZs are always similar in economic development, road network structures, and traffic features; while TAZs located separately tend to be more diversity. Bayesian Conditional Autoregressive Regression (CAR) Models were used to analyze these spatial correlated data. Considering the different crash mechanisms of major roads and minor roads, new modeling strategy was proposed by modeling crashes on arterials and on local roads separately. The paper finds out there is a significant relationship between the road network patterns and safety, and the new modeling strategy is able to differentiate the safety effect for different types of crashes.