面向高速公路的广义交通事故预测模型构建方法研究

项目名称： 面向高速公路的广义交通事故预测模型构建方法研究

项目编号： No.51308114

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 建筑科学

项目作者： 叶智锐

作者单位： 东南大学

项目金额： 25万元

中文摘要： 掌握交通事故发生规律、制定交通安全对策是预防交通事故的有效途径。交通事故预测模型是研究交通事故发生规律最有效的手段。然而，传统的交通事故预测模型受到假设条件的限制，只适用于特定离散程度的交通事故数据，使得交通事故预测需要通过模型的对比和优选，不但过程复杂，而且易造成模型的不当使用，影响了交通事故预测的准确性和可靠性。针对这一问题，本项目拟通过分析交通事故的影响机理和分布规律，结合广义预测理论和方法，构建适用于各种离散程度的交通事故预测模型，研究预测模型的修正方法，并结合仿真模拟和实例验证广义预测模型的适用性和实用性，提出广义交通事故预测模型，为交通事故预测提供科学指导。项目成果将为我国的交通安全管理提供理论依据和预测方法，充实和发展交通事故预测理论和方法，有助于提高交通事故预测的准确性和可靠性、合理制定交通安全管理措施，缓解交通安全所引起的社会问题。

中文关键词： 交通安全管理；广义事故预测模型；离散性；概率分布；

英文摘要： Traffic crash modeling is the most effective way to develop safety performance function， which is further used for the prediction of traffic crashes and development of traffic safety improvement strategies. However, traditional crash models can be only applied to traffic crash data that display certain degree of dispersion (e.g., underdispersion, equidispersion, overdispersion) and thus severely limit their usefulness. More importantly, this limitation makes the analysis of traffic crash data more complicated and can result in inappropriate use of crash models. In light of this, the goal of this project is to investigate and develop general crash models that can handle different degrees of dispersion in traffic crash data. To achieve this goal, this project will first investigate the factors that affect traffic safety, explore the characteristics of traffic crash data, and analyzing probability distributions of crash data. Then, this project will investigate general count models that can be applied to data with unknown degree of dispersion. The selected models will be further used to develop general crash models. Based on the proposed models, both Monte Carlo simulation and real world examples will be used to comprehensively examine model performance. It is anticipated that the proposed model(s) will ease the an

英文关键词： Traffic Safety Management；General Crash Prediction Model；Dispersion；Probability Distribution；