摘要
证据推理是一种不确定推理的人工智能方法。首先研究几种组合规则的改进方法和造成其缺陷的机理 ,然后从分析冲突的来源入手 ,利用冲突信息提出一种新的组合规则——吸收法。仿真结果表明 ,新的组合规则在处理冲突方面的性能得到了明显改进 ,并且具有较强的鲁棒性。
Evidence reasoning has good performance in dealing with uncertain information in decision making system. Some important improved combination rules are firstly reviewed. The reasons of drawbacks of Dempster′s rule are analyzed. According to the resource of conflict, a new combination rule, named “absorptive” method, is proposed by employing the information of conflict. As simulation results, this new method has better performance in dealing with conflict than Dempster′s rule and other methods.
出处
《控制与决策》
EI
CSCD
北大核心
2000年第5期540-544,共5页
Control and Decision
基金
国家自然科学基金项目!( 69772 0 3 1)
国家重点基础研究发展规划项目!( G19980 3 0 417)
国防预研科学基金项目!( 94J1A0 .4.10 0 )
关键词
证据推理
组合规则
智能决策
人工智能
evidence reasoning, Dempster′s rule, intelligence decision
