基于证据理论的不完全信息多属性群决策方法研究

Study on Multi-attribute Group Decision Making Problem with Incomplete Information Based on the D-S Theory

针对多属性群决策中信息不完全、评价指标存在相关性以及专家权重无法确定等问题,本文提出了一种基于证据理论的决策方法。该方法利用分级排序表处理指标间的相关性,并依据评价信息定义不同属性下各焦元的基本信度分配函数,用DS合成方法对信度分配值进行合成,得到专家的个体评价结果;然后综合专家意见对方案进行排序,在专家权重无法确定时,提出基于焦元距离最小的定权方法;最后通过案例分析与DS-AHP方法进行比较,说明方法的有效性和使用范围。

A new method based on the evidence theory is proposed for multi-attribute group decision making problem with incomplete information, index relativity and experts＇ weights unknown. The approach first processes the index relativity through rating scale analysis and defines the basic probability assignment function according to the assessment information, and then the BPA values are integrated into a comprehensive one through D-S rule of combination. To solve the problem of experts weights unknown, a method which minimizes the total deviationlof each expert from the integrated assessment of the expert group is put forward to synthesize the experts＇ opinions to rank the alternatives, Finially, an example is examined by comparing with the result of DS-AHP to demonstrate the validity and limitation of the method.