基于证据理论的多Agent重复囚徒困境博弈研究.pdfVIP

第24 卷第6 期 Vol. 24 No.6 系统工程学报 2009 年 12 月 JOURNAL OF SYSTEMS ENGINEERING Dec. 21α9 doi: 10. 3969/j. issn. 1∞0 明 578 1. 2∞9.06.003 慕于证据理论的多 Agent 重复囚徒因境博弈研究① 陈波，朱卫东，张洪涛 (合肥工业大学管理学院，安徽合肥 23ω09) 摘要:在多 Agent 系统中，利用证据缓论与爱某囚徒因珑博界对其中的合作行为进行丁研究.为加强 Agent 之间的合作行为，构建了:1.于传皮函数的综合信任模型.该综合信任皮的建立来于两个方面是由以前交 互历史中观察的直接证据而得到的直接信任度，二是由来自于第三方的椎荐证据而得到的椎荐信任皮.­ 于综合信任模型，给出了多 Agent 重复囚徒困境博弈算法，并对所研究的问题进行了仿真，仿真给采表明， 基于综合信任度的博弈策略能够有效地加强系统中的合作行为，提高系统的收益. 关键词:博弈;囊复囚徒困境;证据理论; Dempster 合成;综合信任 中图分类号: F224. 32 ;C934 文献标i只码:A 文章编号:1{削 -5781 (2仪9)俑 -0653 -fJ7 Iterated prisoner s dilemma game with multi-Agent based on evidence theory CHEN Bo , ZHU Wei-dong , ZHANG Hong-tao ( School of Managernent , Hefei University of Technology , Hefei 23以附9 ， China) Abstract: Cooperation behavior in rnulti嗣Agent systerns is studied by using evidence theory and iter酬 ated prisoner s dilernrna garnes. I To strengthen the cooperation between Agents ，由is paper constmc- ted a cornbined tmst rnodel using the belief structure. The cornbined 仿ust w制 cornposed of the direct tmst based on the direct evidence and the recornmended trust based on other Agents recommended evidence. ßased on the constmcted model , a combined tmst strategy and a game algorithrn in itera- ted prisoner s dilemma garne were presented. At last , an application case was simulated in a multi翩 Agent systern. The simulation results demonstrate that the proposed strategy can effectively strength唰 阳出e cooperation behavior and increase the revenues of the sy日tem. Key words: ga