摘要
人的疲劳问题是交互式遗传算法的核心问题,它制约了交互式遗传算法在复杂优化问题中的应用.为了解决该问题,本文提出基于多近似模型的交互式遗传算法.该算法首先将搜索空间划分,然后利用传统交互式遗传算法得到的数据,在不同子空间生成不同的近似模型,最后采用该模型近似人对进化个体的评价,从而减少人评价的数量,有效解决人的疲劳问题.算法性能分析及在服装进化设计系统中的应用验证了其有效性.
Human fatigue is a key problem which restricts the application of interactive genetic algorithms to complicated optimization problems. An interactive genetic algorithm with multiple approximate models is proposed to deal with this problem. In this algorithm, the search space is divided into several subspaces, in which different approximate models are generated with data from traditional interactive genetic algorithms. The approximate model is applied to approximate human subjective evaluations on individuals, thus reducing the number of human evaluations and effectively resolving the human fatigue problem. The efficacy of the algorithm is validated through the performance analysis and the application to fashion evolutionary design systems.
出处
《控制理论与应用》
EI
CAS
CSCD
北大核心
2008年第3期434-438,共5页
Control Theory & Applications
基金
国家自然科学基金(60304016
60575046
60775044)
教育部"新世纪优秀人才支持计划"资助项目.
关键词
遗传算法
交互
多近似模型
服装设计
genetic algorithms
interaction
multiple approximate models
fashion design
