基于博弈论的多Agent协作研究

发布时间：2018-03-24 13:52

本文选题：博弈论　切入点：多Agent　出处：《河南师范大学》2015年硕士论文

【摘要】：随着Agent技术的不断发展,Agent的应用已逐渐深入到了各个领域。Agent所执行任务的难度以及应用环境的复杂度也在不断增加,单个Agent已经不能满足应用的要求,多Agent系统研究已成为Agent技术的一个重要研究方向。目前多Agent系统成为分布式人工智能领域的一个重要分支,同时也是当今国际上人工智能领域研究的热点问题之一,而多Agent协作问题是多Agent系统研究的核心问题之一。多Agent环境中,Agent的行为选择受到环境中其它Agent行为选择的影响,博弈论为这种相互影响的决策行为提供了很好的数学工具,本文利用博弈论的方法,来研究多Agent的协作问题,具体内容如下:(1)对国内外多Agent协作的研究现状进行综述,介绍了多Agent协作的主要研究内容、目前的热点问题,对已取得成果进行分析,并指出其中存在的不足。(2)介绍了Agent和多Agent系统的相关概念,博弈论的概念及分类,Nash均衡的定义及相关定理。(3)研究了多Agent协作团队的形成问题。针对复杂环境中协作团队容易被打破的问题,提出了一种基于博弈论和信誉值的多Agent团队形成算法,并用惩罚机制对试图退出协作团队的Agent成员进行惩罚,有效地保证了任务执行过程中协作团队的稳定性。(4)研究了多Agent协作任务执行的问题。针对复杂环境中的多Agent团队协作问题,提出了基于博弈论的多Agent协作算法,并利用虚拟行动法来预测团队中其它成员的行为选择策略,使得Agent能够选择更好的行为策略,快速到达平衡。(5)最后,对论文的主要工作进行总结,并阐述下一步的研究工作。
[Abstract]:With the continuous development of Agent technology, the application of agent has gradually penetrated into the difficulty of executing tasks and the complexity of application environment in various fields. A single Agent can no longer meet the requirements of the application. The research of multiple Agent systems has become an important research direction of Agent technology. At present, multiple Agent systems have become an important branch of the field of distributed artificial intelligence, and also one of the hot issues in the field of artificial intelligence in the world. The problem of multi-agent collaboration is one of the core problems in the study of multi-agent systems. The behavior selection of multi-agent in multi-agent environment is influenced by other Agent behavior choices in the environment. Game theory provides a good mathematical tool for this kind of interactive decision-making behavior. This paper makes use of the game theory method to study the cooperation problem of multiple Agent. The specific contents are as follows: (1) the research status of multi-#en1# collaboration at home and abroad is summarized, and the main research contents and current hot issues of multi-#en2# collaboration are introduced. This paper analyzes the results achieved, and points out the shortcomings. (2) it introduces the related concepts of Agent and multiple Agent systems. The concept and classification of game theory and the definition of Nash equilibrium and related theorems. 3) the formation of multiple Agent cooperative teams is studied. In view of the problem that cooperative teams are easily broken in complex environments, In this paper, a multi-#en0# team formation algorithm based on game theory and credit value is proposed, and the penalty mechanism is used to punish the Agent member who tries to quit the cooperative team. This paper effectively guarantees the stability of cooperative team in task execution. We study the problem of multi-#en0# cooperative task execution. Aiming at the problem of multi-#en1# team cooperation in complex environment, a multi-#en2# cooperation algorithm based on game theory is proposed. The virtual action method is used to predict the behavior choice strategies of other members of the team, so that Agent can choose better behavior strategies and reach balance quickly. Finally, the main work of this paper is summarized, and the next research work is described.
【学位授予单位】：河南师范大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：O225;TP18

【相似文献】