计算机工程与应用 ›› 2020, Vol. 56 ›› Issue (8): 49-54.DOI: 10.3778/j.issn.1002-8331.1905-0175

• 理论与研发 • 上一篇    下一篇

基于博弈论的人员疏散演化研究

刘根旺,周颖,张磊,康增信   

  1. 河北工业大学 人工智能与数据科学学院,天津 300130
  • 出版日期:2020-04-15 发布日期:2020-04-14

Research on Pedestrian Evacuation Evolution Based on Game Theory

LIU Genwang, ZHOU Ying, ZHANG Lei, KANG Zengxin   

  1. School of Artificial Intelligence, Hebei University of Technology, Tianjin 300130, China
  • Online:2020-04-15 Published:2020-04-14

摘要:

在人员疏散中,撤离者之间的冲突行为复杂且对疏散的速率影响较大,为了模拟此过程,提出了一种将博弈论与元胞自动机相结合的模型。在该模型中设置了两种策略的人(叛逃者和合作者),并引入恐慌指数和惯性系数两个重要参数,来描述疏散过程中撤离者的恐慌和保持原策略的能力;每一次冲突即是一次博弈,发生冲突的撤离者之间通过收益矩阵来确定成功者;并且在博弈完成时采用费米函数进行策略更新。模型再现了“出口拱形”和“快即是慢”现象;同时发现在相同的恐慌下,叛逃者与合作者的比率总是倾向于一致状态。这些自组织现象支持了模型的正确性。该项工作也有助于理解惯性因素对于合作行为产生与存续的影响。

关键词: 冲突, 博弈论, 策略, 恐慌, 惯性, 人员疏散

Abstract:

In pedestrian evacuation, the conflict behavior between evacuees is complex and has a great influence on the rate of evacuation. In order to simulate this process, a model combining game theory with cellular automata is proposed. In this model, two strategies are set up(defectors and collaborators), and two important parameters of panic index and inertia coefficient are introduced to describe the panic of the evacuees in the evacuation process and the ability to maintain the original strategy. Every conflict is a game, the winners are identified by the income matrix between the evacuees, and the Fermi function is used to update the strategy when the game is completed. The model reproduces the phenomenon of “export arch” and “fast or slow”. At the same time, it is found that under the same panic, the ratio of defectors to collaborators tends to be consistent. These self-organizing phenomena support the correctness of the model. This work also helps to understand the impact of inertia on the generation and survival of cooperative behavior.

Key words: conflict, game theory, strategy, panic, inertia, evacuation