摘要
多 agent 的合作决策可克服单个 agent 知识的不完备性,提高决策结果的可靠性.本文在基于证据理论的多 agent 合作决策机制的基础上,对合作决策过程的 agent 可靠性进行研究.基于证据理论,提出一种 agent 可靠性的表示方法.不仅描述单个 agent 的可靠性,同时也刻划 agent 群体的可靠性.agent 适应度的引入,反映决策过程中agent 可靠性的动态变化,实现合作决策时 agent 的优化选择,降低合成各 agent 决策结果时的计算复杂度.
Multi-agent cooperative decision-making is capable of overcoming incomplete knowledge of single agent and improving reliability of the results. In this paper, the reliability of agent is studied on the basis of the evidence theory based multi-agent cooperative decision-making mechanism. A definition of agent reliability is proposed to characterize the reliability of single agent and agent group. Agent fitness is introduced to describe the dynamic reliability of agent during decision-making process, and the agents are optimally chosen for cooperative decision-making. The experimental results show the proposed algorithm effectively decreases computational complexity of combining multi-agent conclusions.
出处
《模式识别与人工智能》
EI
CSCD
北大核心
2008年第5期621-626,共6页
Pattern Recognition and Artificial Intelligence
基金
国家自然科学基金项目(No.60775014)
安徽省"十一五"科技攻关重点项目(No.07010200093)
安徽省教育厅重点自然科学基金项目(No.2006KJ028A)资助
关键词
合作决策
可靠性
适应度
证据理论
Cooperative Decision-Making, Reliability, Fitness, Evidence Theory