考虑评价偏差的交互式遗传算法  

An Interactive Genetic Algorithm with Deviation Evaluation

作　　者：刘冠宇[1] 葛方振[1] 洪留荣[1] 贾平平[1] 于雷 LIU Guanyu GE Fangzhen HONG Liurong JIA Pingping YU Lei(School of Computer Science and Technology, Huaibei Normal University, 235000, Huaibei, Anhui, Chin)

机构地区：[1]淮北师范大学计算机科学与技术学院,安徽淮北235000

出　　处：《淮北师范大学学报（自然科学版）》2018年第2期12-20,共9页Journal of Huaibei Normal University：Natural Sciences

基　　金：安徽高校自然科学研究重大项目(KJ2017ZD32);安徽省自然科学基金项目(1808085MF174);安徽省高校自然科学研究一般项目(KJ2016B019)

摘　　要：交互式遗传算法IGA求解隐式目标优化,存在用户评价不确定性和用户疲劳问题,为此提出考虑评价偏差的交互式遗传算法DE-IGA.算法DE-IGA根据用户认知规律,设计用户评价适应值修正模型,修正力度随不确定性由大到小;根据基因相似度,由上一代最优个体的基因特征选择下一代的用户评价个体;依据基因相似性将种群划分为粒,对粒内未评价个体的适应值估算并调整;将所提出的DE-IGA算法应用于不等面积布局问题UA-FLP.实验结果表明,算法能够有效提高用户评价准确性,提高收敛性,减少运行时间,降低用户疲劳.Interactive Genetic Algorithm （IGA） solves implicit objective optimization, which has theproblems of user evaluation of uncertainty and user fatigue, resulting in user deviation of evaluation.To tackle such problems, an interactive genetic algorithm with deviation evaluation, DE-IGA, is proposed.Firstly, ausereval-uation fitness correction model of DE-IGA is devised according to the users′cognitive law.The intensity cor-rection of the model is from big to small with uncertainty.Secondly, according to the gene similarity, user-evaluated individuals of the next generation are selected from the genetic characteristics of the best individu-al in the previous generation.Then, the population is divided into grains in terms of the gene similarity, and the fitness value of the non-graded individuals is estimated and adjusted.Finally, the proposed algorithm is applied to the unequal area facility layout problem, UA-FLP.The experimental results show that this algo-rithm can effectively improve the accuracy of user evaluation and the convergence, and reduce the running time and user fatigue.

关 键 词：交互式遗传算法 评价偏差 不等面积布局问题 

分 类 号：TP30[自动化与计算机技术—计算机系统结构]