基于三维运动数据分析的典型室内行人流自组织现象仿真研究

Pedestrians' self-organizing phenomenon directly reflects the aggregation state of pedestrian flow, and is the macroscopic expression of the relationship between regional traffic bearing capacity and pedestrian flow. Rational utilization of characteristics of pedestrians' self-organization can optimize pedestrian traffic organization. In view of the problems existing in current research such as lack of analysis of the relationship between macro-phenomena and micro-behaviors at the mechanism analysis level and the insufficient consideration of a pedestrian's individual decision-making mechanism at the simulation modeling level, this topic intends to carry out these tasks: (1) A method of extracting and modeling pedestrians' three-dimensional motion data based on computer vision is proposed, and the macro and micro data of self-organizing characteristics is analyzed; (2) Based on the characteristic data, the decision-making mechanism of individual behavior at the key phase transition nodes of macro-phenomena is analyzed, so as to find out the correlation between macro-phenomena and micro-behaviors, and to analyze the mechanism of self-organization phenomena in essence; (3) Considering the decision-making mechanism of individual behavior, the mechanical characterization methods of micro-behaviors just as walking sideways, turning and following is studied, the social force model is improved, the simulation model of pedestrians' self-organization behavior is established and the simulation experimental platform is built; (4) The evaluation index system of pedestrians' self-organizing behavior which includes pattern and mechanical index is established, and the hierarchical method of pedestrians' self-organizing horizontal state is proposed, and finally an actual case is selected for simulation evaluation. The research results can provide technical support for the design of pedestrian facilities and the formulation of pedestrian flow control strategies.

行人流自组织现象直观反映了人群的集聚状态，是步行设施通行能力与人流关系的宏观表现，合理利用自组织特性可以优化行人交通组织。现有研究在机理解析层面缺乏对宏观现象与微观行为之间关联性的分析，在仿真建模层面未能充分考虑行人个体行为决策机制。针对以上问题，本课题：(1)提出基于计算机视觉的行人三维运动数据提取与建模方法，精细提取宏微观自组织特性数据；(2)基于宏微观特性数据，剖析宏观现象演变关键节点处行人个体行为决策机制，实现宏观现象与微观行为的关联分析，从本质上解析自组织行为机理；(3)充分考虑个体行为习惯与外部环境因素，研究侧身、转向、跟随等微观行为的力学表征方法，完善社会力模型，构建行人流自组织行为仿真模型，搭建仿真实验平台；(4)建立包含宏观斑图指标、力学指标的行人流自组织行为评估指标体系，最终选取实例进行仿真评估。研究成果可为步行设施的设计方案以及行人流管控策略的制定提供技术支撑。

行人流自组织现象直观反映了人群的集聚状态，是交通区域通行能力与人流关系的宏观表现，合理利用自组织特性可以优化行人交通组织。现有研究在机理解析层面缺乏对宏观现象与微观行为之间关联性的分析，在仿真建模层面未能充分考虑行人个体行为决策机制。针对以上问题，本课题：(1)提出基于计算机视觉的行人三维运动数据提取与建模方法，对宏微观自组织特性数据进行统计分析，开展了多场行人可控实验，并以三维视角对可控实验进行了分析；(2)基于特性数据，剖析宏观现象关键相变节点处个体行为决策机制，从而分析宏观现象与微观行为之间的关联性，从本质上解析自组织现象机理，探索了行人流自组织现象的成因机理解析方法，以及影响自组织现象的关键微观参数；(3)充分考虑个体行为决策机制，研究侧身、转向、跟随等微观行为的力学表征方法，对社会力模型进行改进，构建T型通道、双向通道内的行人流自组织行为仿真模型，基于C++语言以及OPENgl图像编程技术自研了城市大型公共场所自组织疏散行为仿真分析与评估平台；(4)建立行人流自组织行为评估指标体系，提出行人流自组织水平状态分级方法，在针对各类型行人自组织行为的指标体系构建之后，依据实测数据、仿真实验数据和相关国家标准对行人流自组织水平进行分级，最终选取实例进行仿真评估。研究成果可为步行设施的设计方案以及行人流管控策略的制定提供技术支持。