改进的蚁群聚类算法及在多属性大群体决策中的应用  

Improved ants-clustering algorithm and its application in multi-attribute large group decision making

作　　者：徐选华[1] 范永峰[1] 

机构地区：[1]中南大学商学院,湖南长沙410083

出　　处：《系统工程与电子技术》2011年第2期346-349,共4页Systems Engineering and Electronics

基　　金：国家自然科学基金(70871121);国家创新研究群体科学基金(70921001);湖南省软科学基金(2008ZK3038)资助课题

摘　　要：多属性复杂大群体决策中,对决策人员的决策结果进行有效地聚类,是分析以及完成群体决策的基础。针对蚁群聚类算法参数选取复杂、自适应性差以及随机性等缺点,提出了一种改进的蚁群聚类算法,该算法将决策群体成员对决策问题的若干个评价准则值转化成偏好矢量,以偏好矢量相聚度作为邻域相似度的计算公式,形成一个启发式聚类算法。通过一个算例计算说明该算法具有聚类质量高、自组织和鲁棒性的特点,适用于解决多属性复杂大群体聚类与决策问题。In multi-attribute complex large group-decision,effectively clustering the decision results of decision makers is the base to analyze and complete group decision.Aiming at the disadvantages for parameters selecting＇s complexity,self-adaptive shortage and their randomicity in ants-clustering algorithm,an improved ants-clustering algorithm is proposed.Several evaluating criterion values of a decision problem from decision makers are converted into preference vectors in the algorithm.The preference vectors＇ clustered degree is taken as the calculating formula of the neighborhood similarity degree to form a heuristic clustering algorithm.A calculation example is used to show that the algorithm holds the characteristic of high clustering quality,self organization and robustness,which is applicable to solve the problem for multi-attribute complex large group clustering and decision making.

关 键 词：蚁群聚类算法 群体聚类 大群体 群决策 

分 类 号：TP311.52[自动化与计算机技术—计算机软件与理论]