The Study On Pedestrian Traffic Flow Simulation And Self-organization Based On Improved Social Force Model

Although human beings can enjoy the convenience of various advancedtransportation tools with the scientific and technological development, there is nomethod replacing walk. The congestion caused by high density pedestrian flow will notonly extend travel time, but also may lead to stampede accident in urgent evacuation,which endangers personal and property safety. In the recent years more and morescholars pay close attention to the problem about pedestrian traffic flow, and theyexplore the problem both from microscopic pedestrian behaviors and macroscopic rules.The modeling of human behavior is a significant approach to reproduce realisticpedestrian flow phenomena, and we utilized social force model to study theself-organized processes in this paper. We improved the basic social force modelrespectively according to different actual scenes, including small groups in the effect ofsurroundings, human crowds under impacts of target point and wall, and queuesinfluenced by disturbance. In the simulation of small groups, the line between membersof2-person group could not be vertical to the direction of the street; one member in3-person group was ignored;4-person group degenerated into two2-person group. Therepulsive force caused by wall was relative to both the distance between pedestrians andwall and the distance between them and target point in specific direction. Compared totraditional semicircle-like formation, a novel sector-like pattern emerged under theextended model. Furthermore, while the target moving, the following pedestriansformed a comet-like configuration, but the pattern’s size and stability depended ontarget’s moving velocity. The disturbance from the target point of queue’s leader couldresult in fluctuation of the whole queue, and initial distance between pedestriansdetermined if the fluctuation was convergent or not. When the initial distance was largerthan critical value, the fluctuation of queue would be convergent. The pedestrianbehavior rules revealed from this paper contribute to enrich pedestrian traffic flowtheory, which will be of great value in future gathered infrastructures planning andoperations.