摘要： 因参考点选择不恰当及折扣方式不合理，DS/AHP群决策方法存在决策信息损失的问题。为此，基于由决策主体推理判断出的互斥方案组和七级标度相对偏好信息，构建能够对所有决策主体在特定属性上进行偏好集成的主体信息融合模型。在此基础上参照传统方法中由知识矩阵向BPA函数转换的思想，以及利用Dempster组合规则进行信息融合的思想，提出能够综合集成所有属性上证据信息的方法步骤，并通过案例模拟验证该方法的科学有效性和应用可行性，其Pignistic概率与标准结果之间的总差异程度较小。

关键词: DS/AHP方法, 群决策方法, 信息无损视角, 主体信息融合模型, Dempster组合规则, 知识矩阵

Abstract: There is a problem of loss decision information resulting from unsuitable reference points and unreasonable discount belief functions in the traditional DS/AHP method for group decision making. In order to solve above problem, a subjective information fusion model is constructed to integrate preferences of decision subjects on a specific attribute, in terms of mutually exclusive alternative groups induced and judged by decision subjects and relative preference information with seven-scale. After that, with reference to the transformation from knowledge matrix to Basic Probability Assignment(BPA) as well as information fusion thoughts by utilizing Dempster combination rule in traditional method, the proposed decision procedures are introduced to fuse evidence information on all of attributes. A simulation case shows the proposed method is scientific, applicable and feasible. The total degree of difference between its Pignistic probability and standard result is small.

Key words: DS/AHP method, group decision making method, information lossless perspective, subjective information fusion model, Dempster combination rule, knowledge matrix

中图分类号: 

• TP391