风险环境下步行设施合流区行人流通行机理与主动干预方法研究

Crowd accident such as stampede often occurs at pedestrian merging facilities due to the complex multi-directional movements of pedestrian flow. Empirical data on the pedestrian flows at merging facilities are insufficient and few researches have been devoted to the analysis and prevention of crowd accidents. For the safe and efficient operation of pedestrian merging facilities, the proposed project aims to explore the crowd motion characteristics, contributing factors of accident risk and safety design and guidance methods for pedestrian merging facilities. Empirical data on pedestrian flow and gait characteristics under safety critical situations were collected in laboratory and Virtual Reality (VR) environment. The experiment scenarios included limited boundary and vision field conditions, limited viewing and hearing conditions, simulated panic situation. Then, the contributing factors and precursory features for pedestrian merging facilities were exploited, and the quantitative relations between the macro-micro features and the risk of crowd accident were established. Based on the associating rules among infrastructures, environment, pedestrian flow and crowd safety, a simulation model for evaluating the level-of-safety service of pedestrian merging facilities was established. Using the proposed model, the safety of pedestrian merging facilities with different designs were evaluated and the design parameters were optimized. At last, guidance strategies for merging zones were designed based on Augmented Reality (AR) equipment. Expected outcomes of the proposed project can provide theoretical and empirical foundations to the safe and efficient operations of pedestrian merging facilities.

步行设施合流区行人流具有复杂的运动特性，且亦是人群踩踏事故的多发地。针对合流区行人流实测数据匮乏、人群事故风险主动干预研究不够深入的现状，本项目拟针对合流区行人流通行特性、人群事故风险影响机理、安全优化设计与诱导展开研究。基于实验室与虚拟现实环境相结合的控制性实验方法，采集多重风险因素影响下行人流实测数据，研究边界视野受限、视觉听觉受限、模拟恐慌等环境下合流区人群运动机理和行人微观步态特性；深入挖掘分析合流区人群事故诱因与前兆特征，探究行人流宏观、微观特性与人群事故风险之间的表征关系，建立基础设施-环境-行人流-人群事故风险关联机制，构建合流区安全服务水平模型；在此基础上，标定面向安全分析需求的微观行人流仿真模型，用以评估不同设计参数下步行设施合流区的安全性，通过优化设计参数，并应用增强现实设备进行合流区的行人诱导，为保障步行设施合流区安全、高效运行提供科学支撑。

无论是步行通道还是房间出口，多股行人流交汇的合流区是人群事故多发地。在诸如视觉受限、群体恐慌等风险环境下，合流区通道与出口的不合理的设计与不协调的运行会显著降低步行设施的服务水平与质量，甚至引发灾难性重大安全事故。然而，由于感知技术瓶颈、伦理道德限制等因素，风险环境下行人流实测与实验数据较为稀缺，无法满足微观行人交通仿真日益精细化建模及公共场所自动化人群管理需求。因此，探索间接、非侵入式地恐慌数据替代性采集方法势在必行。.本项目重点关注风险环境下步行设施行人流通行机理与主动干预方法，组织视野受限与模拟恐慌环境下行人流疏散实验，解析风险环境下行人流通行机理，构建考虑真实行为机理的行人流模型，开展行人流仿真模型在紧急与非紧急状态下的适用性验证研究，提出行人流事故风险建模与主动干预方法，并在应用层面研发多人在线虚拟现实线上实验平台，建立多源数据融合的步行设施状态监测系统，研制面向步行设施行人流主动干预的自动巡检机器人。.研究发现：1）视野受限下与正常视野下行人轨迹显著不同：视野受限环境下行人的寻路策略倾向于优先找到房间边界，并顺墙找到出口；此外，改进后的社会力模型可以较好地仿真顺墙行为。2）金钱激励对行人流的疏散效率、路径决策与选择均有显著影响：总体上，金钱奖励额度越高，人群总体疏散时间越短、平均速度越大。3）针对行人双向合流运动，原版场域元胞自动机（FFCA）模型模拟结果会发生真人实验几乎不会发生死锁现象，改进后的FFCA模型可以降低死锁概率。4）Viswalk中经过实验数据标定的社会力模型在紧急状态下模拟仿真效果总体要弱于非紧急状态。.本项目研究丰富了风险环境下行人流动力学经验数据库，并成功建立支持多人在线交互的虚拟现实行人仿真系统，为步行设施内行人诱导等主动干预策略提供循证设计依据与实验平台，促进自主式灾害预防理论发展，为保障步行设施安全、高效运行提供科学支撑。