交互式蚂蚁算法  

Interactive ant system

作　　者：黄永青[1] 张祥德[2] 李旭东[1] 

机构地区：[1]铜陵学院信息技术与工程管理研究所,安徽铜陵244000 [2]东北大学理学院,沈阳110819

出　　处：《控制与决策》2012年第4期609-612,617,共5页Control and Decision

基　　金：国家自然科学基金项目(70771037);教育部人文社会科学研究青年基金项目(11YJC630074;11YJC630283);安徽省自然科学基金项目(090416247);安徽省高等学校优秀青年人才基金项目(2009SQRZ175)

摘　　要：传统蚁群算法需利用显式评估函数引导算法搜索,因而不适用于那些优化性能指标不能或者难以数量化的系统.为此,提出一种将人对问题解的数量评价值作为目标函数值的交互式蚂蚁算法.从人机交互的特点出发,设计了算法模型的结构、信息素的放置方式与更新策略和用户的评价方式.最后利用模拟算法环境的函数优化实验和汽车造型草图设计实验进行了测试,测试结果表明所提出的算法具有较高的运行效率,并能较好地克服用户疲劳问题.Conventional ant colony optimization needs explicit evaluation function to boost searching process, so it cannot effectively solve the systems whose optimization indices are unable or difficult to be quantified. Therefore, an interactive ant system（IAS） that the objective function is determined by subjective human evaluation of the potential solutions in quantitative way is proposed. According to the characteristics of human-computer interaction, the structure of IAS model, placement method and update strategy of pheromone and user assessment approach are presented. Finally, IAS is tested in function optimizations experiment that is designed as simulation environment for IAS and car styling draft design. The experimental results show that the proposed method has significant operation efficiency and overcomes user fatigue as well.

关 键 词：蚂蚁算法 人机交互 汽车造型 用户疲劳 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]