基于优序图加权的多维稀疏模糊推理方法  

Multidimensional Sparse Fuzzy Reasoning Method Based on Weight of Precedence Chart

作　　者：刘文远[1] 武丽霞[1] 王宝文[1] 

机构地区：[1]燕山大学信息科学与工程学院,秦皇岛066004

出　　处：《计算机工程》2009年第11期210-212,215,共4页Computer Engineering

基　　金：国家科技部高新技术计划基金资助项目(2005EJ000017);河北省科技研究与发展计划基金资助项目(02547015D)

摘　　要：在稀疏规则库条件下,多数的多维稀疏规则条件近似推理方法都难以保证推理结果的凸性和正规性,且没有考虑到多维变量对结论的影响权值。提出一种基于优序图加权的多维模糊推理方法,运用优序图确定权值,实验结果表明,该方法不仅减小推理结果的误差,而且能较好地保证推理结果的凸性和正规性。Most interpolative reasoning methods in multi-dimension cannot guarantee the convexity and normality of result, and cannot consider weights of each variable influencing conclusion. This paper proposes a fuzzy multidimensional reasoning method based on weight of Precedence Chart（PC）, which confirms the weights by Precedence Chart（PC）. Experimental results show that it not only reduces the errors, but also keeps the convexity and normality of the reasoning consequence.

关 键 词：多维稀疏模糊推理 优序图 权值 相似性 

分 类 号：TP311.13[自动化与计算机技术—计算机软件与理论]