摘要
总结了社会力模型的研究现状、存在问题和基本原理.将分子动力学的Gear预测校正法应用到社会力仿真模型中,并根据柯朗条件确定了行人仿真显式方法的最小时间步长,引入了链接列表元胞(linked-listcell method)加速算法.最后设计了编程实现框架,采用VC++实现了上述方法,并进行了仿真实验,再现了自动渠化等行人群自组织现象,仿真结果显示减速避让机制的加入可以得到更符合实际的效果,而链接列表元胞法可大幅缩减仿真时间.
The principle, current research and existing problems of social force model for pedestrian simulation are summarized in this paper. Gear＇s predictor-corrector method in molecular dynamic is applied in social force method. According to Courant-Friedriehs-Lewy condition, the minimum time step for explicit method pedestrian simulation is determined, and the linked-list cell method is proposed to accelerate computing. Then the framework of programming for the method is designed and the simulation method is realized. The simulation test recreated some pedestrian self-organization phenomena including lane formation. Simulation result demonstrates that deceleration and collision avoiding mechanism results in more realistic outcome, and the link cell method can reduce simulation time significantly.
出处
《系统工程理论与实践》
EI
CSSCI
CSCD
北大核心
2014年第4期1003-1010,共8页
Systems Engineering-Theory & Practice
基金
国家自然科学基金(51278029)
北京市教育委员会共建项目建设计划(C12H100010)
关键词
社会力模型
分子动力学
仿真
算法
social force model
molecular dynamics
simulation
algorithm
