首页 | 本学科首页   官方微博 | 高级检索  
     

半Markov决策过程折扣模型与平均模型之间的关系
引用本文:殷保群,李衍杰,唐昊,代桂平,奚宏生.半Markov决策过程折扣模型与平均模型之间的关系[J].控制理论与应用,2006,23(1):65-68.
作者姓名:殷保群  李衍杰  唐昊  代桂平  奚宏生
作者单位:1. 中国科学技术大学,自动化系,安徽,合肥,230026
2. 合肥工业大学,计算机系,安徽,合肥,230009
基金项目:国家自然科学基金资助项目(60274012,60574065); 安徽省自然科学基金资助项目(050420301)
摘    要:首先分别在折扣代价与平均代价性能准则下,讨论了一类半M arkov决策问题.基于性能势方法,导出了由最优平稳策略所满足的最优性方程.然后讨论了两种模型之间的关系,表明了平均模型的有关结论,可以通过对折扣模型相应结论取折扣因子趋于零时的极限来得到.

关 键 词:半Markov决策过程  折扣模型  平均模型  最优性方程  最优平稳策略
文章编号:1000-8152(2006)01-0065-04
收稿时间:2004-04-29
修稿时间:2004-04-292005-04-30

Relations between discounted models and average models for semi-Markov decision processes
YIN Bao-qun,LI Yan-jie,TANG Hao,DAI Gui-ping,XI Hong-sheng.Relations between discounted models and average models for semi-Markov decision processes[J].Control Theory & Applications,2006,23(1):65-68.
Authors:YIN Bao-qun  LI Yan-jie  TANG Hao  DAI Gui-ping  XI Hong-sheng
Affiliation:Department of Automation,University of Science and Technology of China,Hefei Anhui 230026,China;Department of Computer,Hefei University of Technology,Hefei Anhui 230009,China
Abstract:The semi-Markov decision problems are discussed for discounted-cost and average-cost performance criteria,respectively.Based on a potential approach,the optimality equations satisfied by the optimal stationary policies are derived.Then the relation between the discounted model and average model is studied.It shows that the related conclusions for the average model can be obtained by taking the limits of results about the discounted model as the discounted factor tends to zero.
Keywords:semi-Markov decision processes  discounted model  average model  optimality equation  optimal stationary policy
本文献已被 CNKI 维普 万方数据 等数据库收录!
点击此处可从《控制理论与应用》浏览原始摘要信息
点击此处可从《控制理论与应用》下载全文
设为首页 | 免责声明 | 关于勤云 | 加入收藏

Copyright©北京勤云科技发展有限公司  京ICP备09084417号