Multi-agent-based Modeling For Passenger Emergency Evacuation From Metro Platforms

Urban rail transit can fully exploit urban space to alleviate the urban traffic pressure and optimize residents travel mode in an effective way at the time of rapidly expanding of urban transit system in China. However, urban rail transit station (especially subway station) is vulnerable to become the high-frequency point of traffic accidents because of its characteristics of narrow space, enclosed of architectural environment, the complex and liquid of evacuees’composed, high degree of social awareness.Evacuation behavior of passengers is affected by various factors and constraints, such as the impacts of psychological stress, evacuation guidance and evacuation personnel structure should be paid attention to when an emergency event occurs. The existed studies on evacuation behavior under impacts of these factors have raised extensive concerns. And detailed studies are not adequate and satisfied.Multi-agent-based modeling considering of psychological stress for passenger emergency evacuation from metro platforms focus on evacuation behavior characteristics is presented in this dissertation according to the existing studies. The contributions of this dissertation mainly summarized as following.(1) An improved social force model considering psychological stress of evacuees is presented. The concept of psychological pressure degree is introduced in the social forced model. And a mathematical model considering of residence time, crowd density and distance from the source of danger is proposed to calculate the value of psychological pressure and analysis the impacts of each factor on the number of pressure behavior.(2) Multi-agent-based modeling for pedestrian evacuation which is considered as a process of perception, decision-making and action. Each pedestrian is regarded as an agent according to the special properties and evacuation behavior characteristics of pedestrians. An agent can not only perceive the interaction with others and obstacles, but also collect information to judge whether the agent is stress or not in the proposed perception model, making decisions about the evacuation route and the desired speed in the proposed decision-making model, determining behavior rules according to influences of psychological stress, evacuation guide and the composed of evacuees in the proposed action model. (3) Case study. Different simulation scenarios based on the Multi-agent-based modeling for passenger emergency evacuation from the platform of the Xizhimen station along Metro Line2in Beijing are carried out to analysis the influence of three factors that stress behavior, guide behavior and group behavior on evacuation process.