分布协同交互式遗传算法及其在群体决策中的应用  

A Distributed Co-Interactive Genetic Algorithm and Its Applications to Group Decision-making

作　　者：孙晓燕[1] 王煦法[2] 巩敦卫[1] 

机构地区：[1]中国矿业大学信息与电气工程学院,江苏徐州221008 [2]中国科学技术大学信息科学技术学院,安徽合肥230027

出　　处：《信息与控制》2007年第5期557-561,共5页Information and Control

基　　金：国家自然科学基金资助项目(60304016);中国矿业大学青年科研基金资助项目

摘　　要：针对交互式遗传算法单机实现模式存在的局限性,本文提出一种分布协同交互式遗传算法,并介绍了算法实现的关键技术.基于群体决策满意度和用户评价偏好,给出了共享个体数量确定方法和共享个体的选择方法.还提出了合理的评价平台和有效的决策评价指标以及隐含的信息交互方式等.为减轻用户疲劳,基于个体相似度,提出一种类适应度近似策略.最后基于服装的色彩设计问题,给出实验结果,以验证该算法的可行性.　A distributed co-interactive genetic algorithm is proposed in this paper to overcome the disadvantages of interactive genetic algorithms implemented in the single PC environment,and the key technologies to implement the algorithm are given.Based on the satisfaction degree of group decision and the users＇ preference,methods are pre-sented to determine the number and selection method of sharing individuals.An appropriate evaluating platform,the effective decision-making evaluation indices and the implicit way to interactive information are also presented.Based on individual similarities,an approximation strategy of similar fitness is put forward to reduce user fatigue.At last,considering the color design of dress,the experiment results are given to validate the feasibility of the proposed algorithm.

关 键 词：交互式遗传算法 分布式 协同进化 群体决策 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]