基于SEBI系统的多智能体通信机制研究

Collision Solution and Communication Mechanism for MultiAgent System Based on SEBI Model

首先介绍了 SEBI多智能体系统模型 ,并通过对模型中各智能体功能的分析 ,提出了利用智能体喜好地址表进行通信 ,通过相互间的接收系数来实现网状链式信息传输的思想 .每个智能体是网络中的一个节点 ,节点可直接与其相连的智能体进行通信 ,也可通过多个链接方式与其它智能体进行间接通信 ,从而可避免多智能体的 NEXP-完全及集中控制存在的信息阻塞等问题 ,一定程度地解决了多智能体通信与合作中的分布与集中这一矛盾 .本文还通过对智能体间通信机制的分析来探讨 SEBI系统中智能体的添加与删除机制及冲突预警机制等 .模拟通信实例结果表明 ,SEBI模型系统可较好地解决多智能体间的 NEXP-完全问题及具有冲突预警、层次分明以及安全稳定等优点 .

In this paper, SEBI MultiAgent System model is constructed based on the difference of social benefit. Through the analysis of functionality of agents in MAS SEBI model, this paper presents the movement mechanism for collisional agents and solution mechanism such as collision warning and security filtering. It also proposes a method for changing agents thinking strategy based on inter-connecting coefficients between agents. The analysis of a typical example shows that SEBI model could solve the problems of communication and collision solution successfully, and it also indicates that SEBI model is clearly hierarchical, open, flexible, secure and robust.