融入关系分组的人群运动仿真  

Relationship-integrated Crowds Simulation

作　　者：柳广鹏 刘弘[1,2] 吕蕾[1,2] 李焱[1,2] 

机构地区：[1]山东师范大学信息科学与工程学院,济南250014 [2]山东师范大学山东省分布式计算机软件新技术重点实验室,济南250358

出　　处：《小型微型计算机系统》2016年第8期1735-1740,共6页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61272094;61472232;61572299;61402269;61402270)资助;山东省自然科学基金项目(ZR2014FQ009)资助;山东省高等学校科技计划项目(J13LN13)资助

摘　　要：在人群行为的运动仿真研究中,人群的分组行为是不能忽略的因素.家庭成员、同事、朋友等关系较密切的人会依据亲密度形成分组,这种现象在整个疏散过程中一直存在,并且个体之间的关系越密切,他们在组内的聚集度越高.目前在人群运动仿真方法中,要么没有考虑分组,要么只根据距离进行分组,而不考虑个体间的关系.本文针对上述不足,在运动过程中同时考虑个体间距离和个体间关系这两个因素对运动的影响,提出了一种改进的K-Medoids算法,该算法加权考虑两类不同的特征值.为验证本文方法的有效性,设计了多组实验.实验结果表明本文方法能够较好地提高疏散效率,并且由于分组过程中考虑了人与人之间的距离以及亲密程度等因素,能够使得虚拟环境中的人群疏散效果更加贴近真实环境下的人群疏散行为.In the research of crowd behavior simulation, the divide group behavior of crowd is an important factor. They will group based on their relationships, such as family members, colleagues or friends. The group will exist in the process of evacuation and the more closely the relationships between individuals are, the higher the degree of gathering is. At present, the distance between the individual is seen as the only influencing factor in crowd simulation or never considered group behavior,ignored relationships between individuals. Considering shortcomings above, this paper considered two kinds of characteristic values in crowd simulation which have in- fluences on group behavior. This paper improved a K-Medoids algorithm using two kinds of characteristic values. To verify the effectiveness of this method, a multi group experiment is designed. The results of experiments illustrated that this method can realize the process of evacuation more efficient and reality, as it takes the distance and intimacy between individuals into consideration, the crowd evacuation in virtual environment is more close to the real environment.

关 键 词：K.Medoids K-MEANS 人群仿真 聚类 关系 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]