多元多维偏好组合集结的群体决策与多目标优化方法

项目名称： 多元多维偏好组合集结的群体决策与多目标优化方法

项目编号： No.61473118

项目类型： 面上项目

立项/批准年度： 2015

项目学科： 自动化技术、计算机技术

项目作者： 李武

作者单位： 湖南理工学院

项目金额： 80万元

中文摘要： 同一多属性决策问题不同方法求解结果不一致是决策分析研究和实践中经常面临的一大难题，将不同方法得出的结果进行适当组合是一种有效办法。但已有研究未考虑多元偏好，不能兼顾评价值差异和序关系，有效性评价方法不具一般性。本项目拟采用距离测度、社会选择、多目标优化及计算机仿真方法，提出描述多元偏好相似性和差异性的距离测度函数；建立组合方法集构造和决策结果权重计算方法；构建多元结果组合的多目标优化模型及求解算法；从社会选择角度提出其有效性评价原则及测度指标并进行理论论证，通过仿真将结果与已有方法进行比较；最后开发网络化混合型组合多属性决策支持系统原型。通过研究，可望在理论上建立多元偏好共存时同一多属性决策问题不同方法的结果组合集结（多元多维偏好组合集结）及其有效性评价方法，为解决不同方法求解结果非一致性难题提供新的思路；在应用上为求解各类多属性决策问题提供有效的方法和工具支持，提高决策的科学性和可信度。

中文关键词： 决策分析；组合多属性决策；多元偏好；多目标优化；群决策

英文摘要： It is a problem need to be solved urgently for decision analysis research and parctice that different methods may yield different results for same multi-attribute decision-making. Integrating properly these results from these methods is a effective methodology for this problem. However, these existing methods do not take multiple preferences into account, can not give consideration to both cardinal evaluation and ordinal preference, and their effectiveness evaluation methodology is not universal. Thus, the project will research the results inconsistency problem by distance measurement, social selection, multiobjective optimization and computer simulation. The research content includes defining similarity and difference of multiple preferences, presenting the methods to build combined decision method set and determine the weights of different decision methods, developing the multiobjective optimization model for combined integration of multiple decision results and its resolution method, presenting and proving the effectiveness evaluation criteria from social selection aspect, comparing the results obtained by our methodology and some existing ones by simulation and finally developing a networked hybrid combined multiattribute decision support system prototype. This research achievement can build a methodology of combined integration and its effectiveness evaluation of results from different methods for same multi-attribute decision making with multiple preferences, and provide new ideas for the results inconsistency problem of different methods for same multi-attribute decision-making. In addition, our reseaches can provide effective methods and tools for all kinds of practical multiattribute decision problems and thus make the decisions more scientific and convincing.

英文关键词： decision analysis;combined multiattribute decision-making;multiple preferences;multiobjective optimization;group decision-making