一种基于历史模型集成辅助的差分进化算法  

Ensemble-assisted Differential Evolution Algorithm Based on Historical Model

作　　者：谭瑛[1] 曹修 王浩 李晓波 TAN Ying;CAO Xiu;WANG Hao;LI Xiao-bo(School of Computer Science and Technology,Taiyuan University of Science and Technology,Taiyuan 030024,China)

机构地区：[1]太原科技大学计算机科学与技术学院,太原030024

出　　处：《小型微型计算机系统》2022年第6期1315-1321,共7页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61876123)资助;山西省自然科学基金项目(201901D111262,201901D111264)资助.

摘　　要：当前,基于代理模型辅助的进化算法广泛用于解决昂贵优化问题.其中,由于集成模型策略可以有效的集合多种模型的特点从而提高模型预测的准确度,所以被广泛应用.但是建立多个模型会增加优化过程的计算成本,因此本文提出一种基于历史模型集成辅助的差分进化算法.本文工作分为两部分:首先,提出由一部分历史模型和当前模型构成集成模型,该策略可以有效的降低计算成本.其次,提出一种新的基于决策空间欧式距离的不确定度评价标准,用于选择个体进行真实计算.为了验证本文提出算法的有效性,将本文方法与相关算法在CEC2005测试函数上测试,并且进行比较.实验结果证明本文提出的算法可以更有效的解决昂贵优化问题.At present,the surrogate-assisted evolutionary algorithms(SAEA)are widely used to solve expensive optimization problems.The ensemble model strategy is widely used in SAEA,because it ensembles the characteristics of various models to improve the accuracy of model prediction.However,the computational cost that is used to train models is increases dramatically.Therefore,in this paper,an ensemble-assisted differential evolution algorithm based on historical model(EHDE)is proposed.The main work of this paper is divided into two parts:first,an ensemble model strategy based on a part of historical models and the current model is proposed that can efficiently reduce the calculation costs.Secondly,a new degree of uncertainty evaluation criterion based on euclidean distance in decision space is proposed in order to select the solution to calculate by expensive function.In order to verify the effectiveness of the proposed algorithm,the method and related algorithms are tested on CEC2005 test suit and compared.Experimental results show that the proposed algorithm is more effectively when is used for solving expensive optimization problem.

关 键 词：昂贵优化问题 代理模型辅助的进化算法 集成历史模型 不确定度 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]