基于驾驶行为建模和安全-效率综合评价的交叉口信控方案优化方法

With growing attention paid to the traffic safety problem, the design method and level of service evaluation for the signalized intersection gradually change from mainly considering the traffic efficiency from involving the traffic safety. This project is based on the driving behavior modeling and studies the intersection signal control scheme optimization with the comprehensive evaluation of safety and efficiency. Firstly, from the viewpoint of driving behaviors (car following, stop and go decisions and gap acceptance behavior), the project studies drivers’ preference rules to the traffic potential risks at the signalized intersection utilizing the traffic conflicts technique, and explores the relationship between the signal control scheme and the safety level. Then, the heterogeneous driving behavior models and traffic simulation platform for the signalized intersection are constructed, based on which the critical behavioral factors to the traffic safety and efficiency and the influence mechanism are uncovered. After that, the established dual-objective optimization model regarding the traffic safety and efficiency and the traffic simulation platform constitute the simulation optimization system for the signalized intersection, by which the corresponding Pareto optimal state is resolved, and the influence modes and differences of key driving behavioral factors and signal control parameters on the traffic safety and efficiency are compared. These results will provide the theoretical basis and technical foundation for the driving behavior governance and the signal control scheme optimization so as to reach the tradeoff between the safety level and traffic efficiency for the signalized intersection.

随着交通安全问题日益获得重视，信号交叉口设计方法和服务水平评价从主要考虑通行效率逐渐转变到兼顾行车安全水平。本课题基于驾驶行为建模，研究融合安全和效率综合评价的交叉口信控优化方法。首先从驾驶行为的角度（跟车、停行决策和可接受间隙行为）采用交通冲突分析技术研究信号交叉口驾驶员的风险偏好规律，探讨信号控制方案与安全水平之间的关联性，构建异质驾驶行为模型及交通仿真平台，揭示影响行车安全和通行效率的关键行为因素及影响机制。然后，建立信号交叉口关于行车安全和通行效率的双目标优化模型，并与交通仿真平台构成仿真优化系统，获取关于双目标的帕累托最优状态，比较关键行为因素和信号控制参数对安全和效率相关指标的影响方式与差异，为权衡信号交叉口行车安全和通行效率而进行驾驶行为治理和信号控制方案优化提供理论依据和技术基础。

随着交通安全问题日益获得重视，交叉口信控方案优化和服务水平评价从只考虑通行效率逐渐转变到兼顾行车安全水平。本项目基于驾驶行为建模与仿真，研究了融合安全和效率综合评价的交叉口信控仿真优化方法。项目主要在三个方面取得了重要进展：1）通过COM接口与二次开发技术将Matlab控制平台与宏、微观交通仿真软件进行集成，搭建了支持不同评价精度的交通仿真优化平台，提出了综合考虑安全和效率的路网信控优化模型，并设计了相应的双目标随机仿真优化算法。该研究成果具有很好的理论与实际工程意义，并拥有良好的扩展性、移植性和推广性。2）车-车通信环境下，将驾驶员和车辆类型进行分类，研究车-车通信信息对驾驶行为的作用机制，提出了考虑安全行驶规则的微观交通流模型。同时，提出了适应于不同交通数据的灵活宏观交通流模型和满足各向异质性的宏观交通流模型。该研究成果有助于更好地分析和理解信号交叉口附近的驾驶行为特征，为仿真平台中交通流模型的二次开发提供理论支撑。3）将城市区域离散网格化，通过浮动车数据重构了快速路，可以有效辨别交通状态。同时，设计了离线-在线交通预测框架，并搭建了多任务深度学习预测算法框架，有助于实现信号交叉口实时在线优化控制。