基于地理空间认知与多智能体的人群疏散模型

Crowd evacuation simulation is an important supporting tool for unconventional emergencies decision. However, the defects of current crowd evacuation models in the geographical environment perception and behavioral decision-making have become an important bottleneck for the model application in complex and real geographical environment. Based on the methods of modeling man-land relationship in Geographic Information System (GIS), a new crowd evacuation model will be developed using geo-spatial cognition and Multi-agent System (MAS). We will pay more attention to the model construction and its application. By modeling crowd based on individuals, three critical aspects related to crowd evacuation model will be investigated: the spatial data model considering environment semantic information, the modeling of geo-spatial cognitive and its application, and the modeling of individual spatial-temporal behavior based on MAS. Tianjin Normal University Gymnasium will be taken as the study area of crowd evacuation model. Combined with data acquisition and processing from real environment, a simulation system for virtual evacuation facing to bioterrorism is built and used to analyze the behavioral characteristics of individuals. This research can be utilized in efficient security, scientific management of the cities, urgent response, and management of emergencies and so on. It can also promote the applicability and accuracy of crowd evacuation mode, improve the scientific decision-making ability of disaster prevention and management, strengthen the GIS modeling and analyzing capabilities to the individual, deepen the study of Virtual Geographic Environments (VGE) and geographical process simulation, and promote the research of people-based geographic information science.

人群疏散模拟是非常规突发事件辅助决策的重要参考手段，然而当前人群疏散模型在地理环境感知和行为决策中的缺陷，成为制约其面向复杂真实地理环境应用的重要瓶颈。本项目结合地理信息系统（GIS）人地关系建模方法，对基于地理空间认知与多智能体的人群疏散模型构建与模型应用进行深入研究。在机理上，以个体作为人群建模的基本单元，重点探讨考虑环境语义的空间数据组织与表达方法，地理空间认知的建构与调用方法和基于多智能体的个体时空行为模型。在应用上，以天津师范大学体育馆为研究区，结合现实环境中的空间数据采集与个体行为特征参数获取，建立人群疏散模拟实验环境，面向特定案例分析个体的时空行为特征。本项目研究可服务于城市的高效安全科学管理与突发事件的应急处置与管理，提高疏散模型的适用性和准确度，加强GIS对个体时空行为的建模与分析能力，深化虚拟地理环境（VGE）和地理过程模拟的研究，促进面向人的地理信息科学发展。

本项目设计并完成采用地理空间认知和多智能体的人群疏散模型的建模。对考虑环境语义的空间数据组织与表达、地理空间认知的建构与调用方法、基于多智能体的个体时空行为模型、面向特定案例的人群疏散模拟试验等内容进行了研究；并重点对基于网格的空间数据表达模型、基于拓扑图的地理空间认知构建方法、基于场模型的人群行为计算方法、基于智能体和多层拓扑网络的人群疏散路径计算方法等进行探讨。并结合特定案例应用，构建了面向三维空间环境的人群疏散模拟原型系统，开展了原型应用试验。本项目在面向人群疏散的空间数据数据模型、人群行为计算等方面进行了突破，为加强GIS与人群疏散模拟的结合，深化虚拟地理环境（VGE）和地理过程模拟的研究方面做出了贡献。本项目共发表论文13篇（含录用），其中SCI论文5篇，申请软件著作权登记3个。