摘要
在被评价对象的指标值与理想值之间的广义距离和充分小的情况下,追求不同赋权方法权重组合系数的信息分配最合理。随着广义距离和不断变小,得到一组不同方法赋权后的组合权重,进而得到了评价结果。本文的特色与创新一是本文得到的权重兼顾了信息分配最合理与指标数据距离理想值的广义距离和最小两个目标。二是提出一个单目标模型求解多目标问题Pareto解集的方法,并根据解集对评价对象进行排序,增加了排序的可靠性,也为多目标模型求解提供了一种新思路。三是改变了组合赋权系数为近似平均的结果。四是解决了多目标线性加权求解时多个目标组合系数不确定问题。
Considering the minimum sum of generalized distance between the actual values and ideal values of the alternatives,this paper presents a model for distributing the information of the combination coefficients of different weighting methods more reasonably.With the sum of generalized distance becoming smaller gradually,the combination weights are obtained by different weighting methods.The valuations of alternatives are calculated and the rank of alternatives is given.The contribution of the paper is as follows.Firstly,the combination weights can reflect the most reliable information distribution and the minimum generalized distance sum.Secondly,a new method is presented to calculate the Pareto solution set of multi-objectives programming by a single objective optimization model.The rank of alternatives is given according to the solution set.Therefore,the result of the valuation is more reliable.At the same time,a new idea is provided for solving multi-objectives model.Thirdly,the combination coefficients are not equal with the decreasing sum,which are more reasonable.Fourthly,the uncertainty of the combination coefficients between the objectives is eliminated.
出处
《运筹与管理》
CSCD
北大核心
2011年第1期53-59,共7页
Operations Research and Management Science
基金
国家社会科学基金重大资助项目(06&ZD039)
大连理工大学人文社会科学研究基金重大资助项目(DUTHS2007101)
辽宁省社科规划基金资助项目(L07DJY064)
大连市科技局项目(2009C11CY185)
关键词
线性组合赋权
最大熵原理
对偶规划
组合系数计算
linear combination weighting
maximum entropy principle
dual programming
combination coefficients calculation
