基于动－静态风险因素耦合机制的公路几何线形安全评价

Safety evaluation of highway geometric alignment is an effective method to guarantee safe design of highway as well as to prevent road crashes. However, the lack of evolution mechanism of road crash with the coupling effect of dynamic and static risk factors and corresponding crash prediction theory are critical bottlenecks of highway geometric design evaluation. The vehicle instability index and dynamic safety boundary reflecting track-holding, steering and roll stability are to be explored based on vehicle system dynamics at first, and then the risk evaluation theory of single vehicle crash can be improved. The interaction mechanism among highway alignment, dynamic characteristics of traffic flow and weather condition in multi-vehicle crashes is to be revealed with the assistance of statistics and data mining, after which crash probability prediction model is to be developed. Finally, taking ambiguity, discreteness of dynamic risk factor variables and correlation among variables into consideration, the method of highway geometric alignment safety evaluation with the coupling effect of dynamic and static risk factors is to be proposed based on cloud model theory. In this project, theoretical exploration will be undertaken by employing vehicle system dynamics, safety systems engineering, uncertainty theory and data mining, and then will be integrated with empirical research. The new geometric alignment safety evaluation method which characterizes various types of accident risks will be put forward, which is significant for enhancing the science, accuracy and targeting effect of highway geometric alignment safety evaluation, from both theoretical and engineering perspectives.

公路几何线形安全评价是提升公路安全设计水平、预防交通事故的有效手段，但动、静态风险因素耦合作用下公路交通事故演化机理及事故预测理论的匮乏成为制约几何线形安全评价理论发展的重要瓶颈。本项目拟基于汽车系统动力学，探索反映车道保持能力、转向能力、侧倾稳定性的汽车侧向失稳评价指标及动态安全边界，完善单车行车风险评价理论；综合运用统计学、数据挖掘技术，揭示多车事故中几何线形、交通流动态特征、天气特征的交互作用机理，构建多车事故风险预测模型；考虑各变量间的相关性及动态风险因素变量的模糊性、离散性，依据云模型理论提出动-静态因素耦合的公路几何线形安全评价方法。项目将运用汽车系统动力学、安全系统工程学、不确定性理论、数据挖掘等工具开展理论探索，并与实证研究相结合，提出准确表征各类型事故风险的公路几何线形安全评价新方法，对于提升评价的科学性、准确性和靶向性具有重要的理论和工程实践意义。

公路几何线形安全评价是提升公路安全设计水平、预防交通事故的有效手段，但动、静态风险因素耦合作用下公路交通事故演化机理及事故预测理论的匮乏成为制约几何线形安全评价理论发展的重要瓶颈。基于此，本项目依据汽车系统动力学、数值分析理论，提出了人-车-路-天气耦合作用下的单车事故的形成机理和风险预测方法；运用数据挖掘技术，深入解析了交通事故中几何线形、交通流动态特征、天气状况的耦合作用机制；考虑事故前天气、交通流状态和事故数据中的异质性，建立了动-静态风险因素耦合下的事故频次分析模型，探讨了各风险因素及其交互作用对事故频次的影响；针对动态风险因素的随机性、相关性和时变性，构建了基于实时事故风险的公路交通事故频次预测方法，并将其应用于公路几何线形安全评价。本项目重点突破了汽车失稳模态与单车事故概率的关联机制、公路事故中动-静态风险因素的耦合机理、考虑事故前动态风险因素时变特征的公路几何线形安全评价体系等关键科学问题，在动-静态风险因素交互机理、动-静态风险因素耦合下的事故预测方面实现了创新。研究成果可促进交通工程、安全工程、道路工程、车辆工程的学科交叉，为科学、准确地开展公路几何线形安全评价提供了可靠的理论依据和技术支持。