摘要
针对投影寻踪方法对多属性决策问题建模时,无法兼顾决策者经验及偏好、权重系数可能违背实际的问题,提出了一种基于层次分析和模糊专家评判的投影寻踪决策方法。借助层次分析法的思想构建指标的层次结构,然后专家根据经验进行模糊评判,得到准则的重要程度序关系,将其以约束的形式融入投影寻踪模型中。同时,针对差分进化算法的不足,提出了自适应聚类差分进化算法,并用于投影寻踪模型中的指标函数优化,得到最佳权重系数。该方法在客观赋权的基础上,融合了主观信息,弥补了两种赋权方法的不足,实际的算例验证了所提出的决策方法与优化算法的可行性与有效性。
A projection pursuit decision-making method based on analytic hierarchy process and fuzzy expert evaluation was proposed to recover the problem in modeling multiple attribute decision making problems with projection pursuit method which discarded the experience and preference of decision-makers and sometimes gave impractical weight coefficient. A hierarchy of indicators was built with the thought from Analytic Hierarchy Process and the importance order of the criterions was obtained according to the fuzzy evaluation of some experts. Serving as a constraint, the importance order was blended in the projection pursuit model The optimal weight coefficients were achieved by optimizing the projection index function with the new-proposed adaptive clustering differential evolution algorithm which was proposed to avoid the disadvantages of differential evolution algorithm. This method combined the subjective information with objective information and overcame the disadvantages of two weight assigning methods. Finally, a numerical example illustrates the feasibility and effectiveness of the decision method and optimization algorithm.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2014年第3期567-573,620,共8页
Journal of System Simulation
基金
国家自然科学基金资助项目(60674021)
关键词
多属性决策
属性权重
模糊专家评判
层次分析
投影寻踪
自适应聚类差分进化算法
multiple attribute decision-making
combination weight
fuzzy expert evaluation
analytic hierarchy process
projecting pursuit
adaptive clustering differential evolution algorithm
