主管：中国科学院
主办：中国优选法统筹法与经济数学研究会
　　　中国科学院科技战略咨询研究院
中国管理科学

中国管理科学 ›› 2009, Vol. 17 ›› Issue (1): 131-137.

• 论文 • 上一篇    下一篇

面向不确定多属性决策问题的范例检索算法研究

梁昌勇, 顾东晓, 范昕, 陈文恩   

  1. 合肥工业大学管理学院, 安徽 合肥 230009
  • 收稿日期:2008-06-30 修回日期:2008-12-24 出版日期:2009-02-28 发布日期:2009-02-28
  • 作者简介:梁昌勇(1965- ),男(汉族),安徽肥西人,合肥工业大学管理学院教授,博士生导师,研究方向:过程优化与决策支持技术.
  • 基金资助:

    国家自然科学基金重点资助项目(70631003);国家自然科学基金资助项目(70741046);教育部博士点项目(20050359006)

Research of Case Retrieval Algorithm for Decision Making with Uncertain Multiple Attributes

LIANG Chang-yong, GU Dong-xiao, FAN Xin, CHEN Wen-en   

  1. School of Management, Hefei University of Technology, Hefei 230009, China
  • Received:2008-06-30 Revised:2008-12-24 Online:2009-02-28 Published:2009-02-28

PDF (PC)

1155

被引次数

21

摘要： 将不确定多属性决策问题转化为不确定性多属性范例的检索问题,利用范例推理方法获取最相似范例作对目标决策问题的近似解,通过范例修正缩短该近似解与目标决策问题解之间的差距直至完全一致.文章系统地提出了适合于不确定多属性问题检索要求的范例检索算法(RA-UMDM),重点研究了范例聚类、用基于梯形模糊集和改进的欧氏距离检索算法分别解决范例中模糊概念属性、区间特征属性的相似度计算问题.将该方法应用于两家医院临床辅助诊断决策过程求得了较为准确的解,并且基于范例库对RA-UMDM法的有效性、准确度、效率等进行了实验,结果表明该方法能够满足不确定多属性决策问题的检索要求,检索速度较快、质量高.

关键词: 多属性决策, 不确定性, 范例, 检索算法, 聚类

Abstract: This paper transformed uncertain multi-attribute decision-making into the uncertain multi-attribute case retrieval question, and used the case reasoning method to gain the most similar case which is the approximate solution of the goal decision-making questions.The difference was reduced between the approximate solution and the solution of the goal decision-making question through case revision until completely consistent.This paper systematically propo sed a case retrieval algorithm(RA-UMDM) which is fit for the retrieval request of the uncertain multi-attribute question, and emphatically studied the fuzzy set based on the trapezoid and improved retrieval algorithm based on Eulerian-Lagrangian distance which would be used to solve in separately the case fuzzy concept attribute, the similarity of sector characteristic attribute and the case weight adjustment.This paper applied this method in the dentistry assistance diagnosis decision-making process and obtained a more accurate solution, and then tested the validity, accuracy and efficiency of RA-UMDM algorithm.It was demonstrated that RA-UMDM could satisfy the retrieval requirements of uncertain multi-attribute decision-making with high retrieval speed and quality.

Key words: multi-attribute decision-making, uncertain, case, retrieval algorithm, clustering

中图分类号: