偏好关系粗近似的一种新方法  

A Novel Method of Rough Approximation of Preference Relation

作　　者：花文健[1] 刘作良[1] 

机构地区：[1]空军工程大学电讯工程学院,陕西西安710077

出　　处：《计算机应用研究》2004年第12期23-25,28,共4页Application Research of Computers

基　　金：国家"863"计划资助项目(863 306 04 02 1)

摘　　要：提出一种多准则方案对偏好程度粗度量的一般方法。构造了一种从参考方案集中获取偏好规则并用于全体方案排序和优选的算法。将参考方案对的偏好关系定量分级表示,应用粗集理论的近似原理对综合偏好关系进行粗近似以获取基于偏好信息的决策规则。计算实例表明新方法能充分表示决策者的偏好,有效避免了规则不一致性。A new method for rough measurement of the preference degree between action pairs in MCDM is presented based on the Rough Sets theory. The preference degree of each reference action pair is quantificationally described by the set of intervals. The comprehensive preference relations of the reference action pairs are approximated respectively by the upper set and the lower set to derive the preference information potential in reference action set in terms of decision rules with no (inconsistence) found in LRES method. A computational study shows that new method based on the graded preference relation achieves significant improvement over the method based on the indiscernibility relation.

关 键 词：分级优先关系 粗近似 决策规则 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]