一个基于自组织多Agent系统的智能控制与决策模型  

An Intelligent Control and Decision Making Model Based on Self-Organizing Multi-Agent System

作　　者：杨斯博[1] 李敏强[1] 

机构地区：[1]天津大学管理与经济学部,天津300072

出　　处：《天津大学学报》2012年第10期903-911,共9页Journal of Tianjin University(Science and Technology)

基　　金：教育部高等学校博士学科点专项科研基金资助项目(20090032110065)

摘　　要：提出了一个基于自组织多Agent系统的智能控制与决策模型,它是一种基于行为主义的智能控制与决策模型,由环境和自组织多Agent系统两大部件构成,分别通过环境的定义、环境的识别、多Agent控制与决策过程以及多Agent控制与决策输出4个基本步骤来完成建模工作.该模型可以灵活地选择多种算法进行具体的实现工作,文中给出的算法是一种基于作用力机制(物理激励)的多Agent控制与决策算法,该算法是通过模仿经典物理学理论中的万有引力定律,将多Agent系统中的自组织交互过程转化为相互的作用力,并通过交互作用力的大小和方向体现多Agent之间的交互机制,充分利用了多Agent系统的群体决策优势.测试实验结果表明,该模型具有较好的应用效果并且其系统能量在Agent数目为300、迭代次数超过80次时具有稳定性.An intelligent control and decision making model based on the self-organizing multi-Agent system was proposed in this paper. This model is a behavior-based model and has two major components--environment and self- organizing multi-agent system. Its modeling work is completed by the following four steps： the definition of envi- ronment, the identification of environment, the process and the output of multi-Agent control and decision making. The model has the flexibility to choose a variety of algorithms for the implementation of specific work. The algorithm given in the paper is a multi-Agent control and decision making algorithm based on interaction force mechanism （physical inspiration）. Through the imitation of the law of gravity in classical physics, the algorithm changes self- organizing interaction processes in the multi-Agent system into mutual interaction forces and realizes the interaction force mechanism with sizes and directions of the interaction forces in the multi-Agent system. It takes full use of the advantage of group decision making of the multi-Agent system. Test results indicate that this model has good applica- tion results and its system energy is stable when the number of Agent is 300 and the number of iterations is more than 80 times.

关 键 词：自组织 多AGENT系统 智能控制与决策 环境 作用力机制 群体决策 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]