针对不确定多属性决策问题,就区间数综合属性值排序过程中现有2种可能度的定义(即文中的定义3和定义4)以及可能度矩阵排序算法进行了分析比较并给出其适用范围.算例的计算结果和数理分析表明,基于定义4的排序算法具有更强的比较能力,可以解决基于定义3导致的区间差别较小而无法排序的问题,适用于区间数的精确比较.而定义3在参与比较的区间数长度接近且重叠比例接近1/2时误差较大,不宜使用;但该方法简单、计算速度较快,适于在大规模计算中的应用.在实际应用中可按照上述原则使用这2种方法或结合使用.
Focusing on multi-attribute decision, two definitions of possibility (definitions 3 and 4 in this paper) in the ranking process and the ranking method of possibility matrix were analyzed, and then the scope of application was given. The result of case study and mathematical analysis show that the ranking method corresponding to definition 4 has stronger ability, and can deal with ranking problems resulting from slight difference of interval values based on definition 3. So, it is suitable for an accurate comparison of interval numbers. Definition 3 will create a larger error with close length of two interval values and overlapping ratio close to 1/2, so it is unsuitable for the case. However, this ranking method is simple to calculate and suitable for large-scale computation. In practice, each of the two definitions or their combination can be used according to these principles.