基于可能度矩阵的区间型多属性决策方法.docVIP

基于可能度矩阵的区间型多属性决策方法 摘 要: 研究了几组可能度公式之间的关系，提出一种基于可能度矩阵的区间型多属性决策(madm)方法。对决策矩阵中各指标下的属性区间值两两比较并建立各指标的可能度矩阵，通过各个可能度矩阵的排序向量把属性值为区间数的决策矩阵转化为以精确数为测度的矩阵，把求解区间型多属性决策中指标权重的不确定性问题转化为确定性问题处理，随后利用区间数排序的可能度法获得最优方案。实验结果表明了所提方法的可行性和有效性。最后对多属性决策问题中由不确定性转化为确定性的求解策略及其可能产生的问题作了必要讨论。 关键词: 多属性决策；指标权重；区间数；可能度 中图分类号: c934；tp18 文献标志码:a abstract: the authors studied the relation between several possibility degree formulas, and proposed a possibility degree matrices-based method that aimed to objectively determine the weights of criteria in multiple attribute decision-making (madm) with intervals. each pair of interval values belonging to the same attributes in a decision matrix was compared to construct corresponding possibility degree matrices, whose priority vectors were subsequently utilized to convert the decision matrix expressed as intervals into a matrix with precise numbers as a measure. in this way, an uncertainty of determining weights of criteria in madm with intervals could be converted into a certainty which was easier to handle, and with the attribute weights obtained, the possibility degree method for ranking interval numbers was still used to get the priorities of alternatives. two numerical examples were given to illustrate the proposed method and examine its feasibility and validity. finally, a necessary discussion was made on the conversion from uncertainty to certainty in madm with intervals, and some potential problems coming from it. key words: multiple attribute decision-making (madm); weight of criteria; interval number; possibility degree 0 引言 区间型多属性决策（multiple attribute decision-making， madm）［1］是不确定信息决策科学的一种重要表现形式，具有广泛的应用背景，目前的研究多集中在基于uowa(uncertain ordered weighted averaging)算子的区间数排序法［2-3］，基于指标加权的区间数排序法［4-6］，基于区间数距离的topsis(technique for order preference by similarity to an ideal solution)法［7-10］，基于c-owa(continuous ordered weighted averaging)请补充uowa、topsis和c-owa的英文全称。等算子理论的属性值集成法［11-14］，基于理想属性区间数偏离度的灰色关联系数法［15］以及区间型理想点投影法［16］等几种方法上。这其中，指标（属性）权重信息的合理确定对决策结果有着重要