突发事件下人车混合疏散行为及应急疏导策略研究

China is one of the countries in which unexpected emergencies are happening with increasing frequency. Current emergency evacuation research mostly considers single crowd evacuation or vehicle evacuation, and has poor practical applicability because of lack of evacuation behaviors and evolving regulations of pedestrian-vehicle mixed traffic. Therefore, aimed at the actual demand of pedestrian-vehicle mixed evacuation of China, the project will explore the evacuation behavior and guidance strategy of pedestrian-vehicle mixed traffic under emergencies. Firstly, with the aid of psychological questionnaire design, behavior surveys, aspiration surveys, scenario experiments, the sophisticated social psychology and behavioral characteristics of pedestrian-vehicle mixed evacuation will be analyzed, and the real-time movement trails and conflict behaviors will be extracted by using video recognition and positioning techniques. Subsequently, by integrating multi-agent system with social force model, the microscopic behaviors and macroscopic characteristics of pedestrian-vehicle mixed complicated traffic flow will be simulated and imitated. Furthermore, by combining swarm intelligence with multi-objective optimization theory, the evacuation route optimization algorithm will be designed and improved, which could satisfy the macroscopic efficiencies of system and microscopic demands of individual at the same time. Finally, based on above research, the control and guidance strategies of pedestrian-vehicle mixed evacuation under emergencies will be investigated and discussed. The research findings in the project can make up for a lack of existed studies on conflict behaviors of pedestrian-vehicle mixed evacuation, contribute to complexity modeling theory and method study of emergency evolving regulations, and provide scientific theoretical basis and technical support for emergency decision and guidance under pedestrian-vehicle mixed evacuation scenarios.

我国是突发事件频发的国家之一，现有研究大多只考虑单一的人群疏散或车辆疏散，缺少人车混合疏散行为及演化规律研究，因而缺乏现实适用性。本研究针对我国人车混合疏散的实际需求，开展面向突发事件的人车混合疏散行为及应急疏导策略研究。 借助心理学问卷设计、行为调查、意愿调查、仿真情景实验等方法分析人车混合疏散的复杂社会心理与行为特征；通过视频图像识别与物联网定位相结合提取人车混合疏散运动的实时轨迹和冲突行为；集成多智能体系统与社会力模型实现突发事件下人车混合复杂交通流的微观行为仿真与宏观特征模拟；结合群智能算法与多目标优化理论设计兼顾系统宏观效率与个体微观需求的应急疏散路径优化算法；在此基础上探讨突发事件下人车混合疏散的控制与疏导策略。研究成果可弥补现有研究对人车混合冲突行为刻画的不足，有助于突发事件演化规律的复杂性建模理论与方法研究，为人车混合疏散情景下的应急决策与疏导提供科学的理论基础和技术支撑。

随着经济社会快速发展和世界环境变化的加剧，全球发生着一系列的重大突发事件，特别是近年来我国突发事件频发，给国家稳定和人民群众生命财产造成了巨大的威胁和损失。现有应急疏散研究大多只考虑单一的人群疏散或车辆疏散，与真实突发事件场景下人车混合通行的基本特征并不相符，因而缺乏现实适用性。. 本研究针对我国人车混合疏散的实际需求，开展了面向突发事件的人车混合疏散行为与策略研究。通过视频图像识别技术对地震突发事件下的教室区域、不可视环境下教室区域以及楼梯区域的行人疏散开展了大量真实实验研究；借助无人机等物联网技术提取了复杂交叉路口人车混合的运动轨迹和冲突行为；采用改进社会力模型实现了突发事件下人车混合复杂交通流的微观行为仿真、不可视环境下教室区域和楼梯区域的应急疏散模拟；在此基础上探讨了突发事件下人车混合疏散的控制与疏导策略。. 取得的重要进展包括：(1) 借助差分演化算法对社会力模型的参数进行最优化标定,使仿真模型的行人位置、疏散速度与芦山地震教室真实疏散视频基本一致，从而验证了社会力模型在真实地震疏散情景下的正确性和科学性。(2) 对出口处障碍物的形状尺寸和位置进行了最优化设计，并开展了真实的障碍物疏散实验，实验结果表明在出口处放置障碍物后能有效提高人群疏散速度，其根本原因在于通过合理的障碍物放置，人群的高密度拥堵区域大大减少，从而提升了整个人群的疏散速度。(3)开展了一系列行人、自行车、汽车的环形运动与直线运动实验，通过提取三者的实时轨迹数据，发现由于三者一致的内在心理行为驱动机制，其所表现出的动态时空图和静态直方图也是极为相似的；基于这一发现，本研究提出了一个统一的行人、自行车、汽车交通模型，仿真结果验证了该统一模型的正确性。