电子学报 ›› 2017, Vol. 45 ›› Issue (12): 2997-3004.DOI: 10.3969/j.issn.0372-2112.2017.12.023

• 学术论文 • 上一篇    下一篇

基于熵极大准则的非用户赋适应值交互式遗传算法

郭广颂1, 陈良骥2   

  1. 1. 郑州航空工业管理学院机电工程学院, 河南郑州 450046;
    2. 天津工业大学机械工程学院, 天津 300387
  • 收稿日期:2016-04-19 修回日期:2016-06-07 出版日期:2017-12-25
    • 通讯作者:
    • 陈良骥
    • 作者简介:
    • 郭广颂,男,1978年9月出生,吉林省集安市人,副教授.2001年和2007年分别在郑州大学和中国矿业大学获工学学士、硕士学位.主要从事智能控制与进化优化方面研究.E-mail:guogs78@126.com
    • 基金资助:
    • 国家自然科学基金 (No.51275485); 河南省科技攻关项目 (No.172102210513)

An Interactive Genetic Algorithms Based on Maximum Entropy Principle with Individuals' Fitness not Assigned by User

GUO Guang-song1, CHEN Liang-ji2   

  1. 1. School of Mechatronics Engineering, Zhengzhou University of Aeronautics, Zhengzhou, Henan 450046, China;
    2. School of Mechanical Engineering, Tianjin Polytechnic University, Tianjin 300387, China
  • Received:2016-04-19 Revised:2016-06-07 Online:2017-12-25 Published:2017-12-25

摘要: 针对交互式遗传算法适应值人工赋值极易疲劳导致的算法进化代数不足、优化效率低下这一难题,提出了适应值非用户赋值方法.首先,用户对个体采用二元评价机制评价个体,将个体划分为满意集合和不满意集合;然后,根据个体评价时间与偏好的内在联系,通过个体评价时间确定评价满意度;最后,基于熵极大准则求解满意度最大条件下的个体适应值.为了确保优势基因遗传,加快算法收敛,采取种群精英基因构建优势个体保留策略.将该方法应用于装饰性墙壁纸选型系统中,并与其他代表性算法比较.结果表明,该方法能有效降低疲劳,提高算法优化效率.

关键词: 交互式遗传算法, 熵极大准则, 适应值, 非用户赋值

Abstract: The fitness assigned by user can easily make fatigue which causes insufficient evolution algebra and low optimization efficiency for interactive genetic algorithms.In this study,a method of interactive genetic algorithms with individuals' fitness not assigned by user is presented.First,the individuals could be divided into satisfied sets and not satisfied sets; then,the individuals satisfaction is determined through evaluation time; finally,the fitness is calculated based on the maximum entropy principle under the biggest satisfaction.In order to ensure protogene inheritance,the reserved elite individual is built by population elite genes.This method is applied to selection system of decorative wallpaper,and the results show that it can effectively reduce fatigue and improve the optimization efficiency.

Key words: interactive genetic algorithms, maximum entropy principle, fitness, not assigned by user

中图分类号: