基于势能网格的人群流动模型及其在渠化效应中的应用

Pedestrian Flow Simulation Model Based on Potential Grid and Its Application to Lane Formation Phenomenon

将宏观流体力学模型中的一些特性和微观社会力模型的部分特征融合起来,利用Dijkstra算法建立覆盖全场的势函数分布,进而建立了一种新的人群流动模型,利用光滑粒子流体动力学的思想指导运动过程,使得模拟的过程兼具宏观和微观特性。该模型在模拟某音乐厅散场算例中得到了验证,也成功重现了渠化效应现象。针对渠化效应这一典型的人群流动自组织现象提出了渠化效率这一概念,分析了它的变化特性与影响因素,为今后进一步的定量理论研究提供了工具。

By combining macro characteristics in fluid mechanics model and micro characteristics in social force model together, and using Dijkstra algorithm to establish a potential function covering the whole field of simulation, with the guidance of Smoothed Particle Hydrodynamics, a new model was created in order to keep both macroscopic and microscopic characteristics during the process of simulation. This new model was verified by evacuation simulation in a music hall. Using this model, lane formation phenomena was reproduced. The definition of lane formation ratio makes it possible to analyze variations and causes of this typical self-organization phenomena, providing us with new tools for quantitative pedestrian simulation studies.