基于用户偏好模型的交互式遗传算法  

Interactive Genetic Algorithm Based on User Preference Model

作　　者：蔡美菊[1] 梁昌勇[2] 

机构地区：[1]合肥工业大学管理学院,合肥230009 [2]合肥工业大学过程优化与智能决策教育部重点实验室,合肥230009

出　　处：《小型微型计算机系统》2016年第4期758-762,共5页Journal of Chinese Computer Systems

基　　金：国家自然科学基金重点项目(71131002)资助;国家自然科学基金面上项目(71271072)资助;高等学校博士学科点专项科研基金项目(20110111110006)资助

摘　　要：针对交互式遗传算法的核心问题——用户的疲劳问题,提出一种基于用户偏好模型的交互式遗传算法.在用户较偏爱个体的历史评价信息基础上,构建了基于部分的用户偏好模型,并据此结合考虑各构成部分权值及构成部分间关联关系对进化个体适应度值的影响,给出了一种更能符合用户情感偏好的适应值估算方法,以便在用户疲劳时由计算机有效地代替人工自动对新产生的进化个体进行评分.给出所提算法的关键技术与实现步骤,并将其扩展应用在汽车造型设计这一实践领域.与其他算法的实验结果对比验证了本文提出的算法具有较高运行性能,并能有效降低用户疲劳.To tackle the user fatigue problem——the core problem of interactive genetic algorithm,an improved interactive genetic algorithm on basis of user preference model was proposed.On the foundation of the history information of preferred evolutionary individuals,user preference model based on parts was established.Considering the weights of parts and the connections between parts,an estimation method of the evolutionary individual fitness was presented in order to implement automatic computation the fitness of newevolutionary individuals by computers when users are fatigued.The key technologies and implementation steps of the above algorithm were given,and then the algorithm was extended to the practice field of automobile modeling design.By contrast with the experimental results of other algorithms,it shows that the proposed algorithm has high performance,and can effectively reduce user fatigue.

关 键 词：用户偏好模型 交互式遗传算法 用户疲劳 偏好组合 汽车造型设计 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]