基于繁殖策略的求解昂贵约束单目标进化算法  

Apropagation Strategy based Evolutionary Optimization Algorithm for Expensive Constrained Single-objective Problem

作　　者：谭瑛[1] 张何萧 王浩[2] 李晓波 TAN Ying;ZHANG He-xiao;WANG Hao;LI Xiao-bo(College of computer Science and Technology,Taiyuan University of Science and Technology,Taiyuan 030024,China;School of electronic information engineering,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 [2]太原科技大学电子信息工程学院,太原030024

出　　处：《太原科技大学学报》2024年第2期119-124,共6页Journal of Taiyuan University of Science and Technology

基　　金：国家自然科学基金(61876123);山西省自然科学基金(201901D111262)。

摘　　要：实际工程优化中存在大量约束优化问题,且有一些优化问题目标函数和约束函数的评价非常耗时,导致该类问题无法直接使用传统优化算法求解。为此,为了在评价次数有限的情况下获得较好的可行解,针对昂贵单目标约束优化问题,为评价费时的目标函数和约束函数建立径向基函数(Radial Basis Function,RBF)预测模型,以及根据估值自适应选择个体的繁殖策略,以期能产生较好的可行解。在7个标准测试函数及3个工业测试函数上的测试结果表明,相比于其它现有针对昂贵约束问题的优化方法,本方法无需确保初始种群中必须有可行解,且能在优化目标和约束函数评价次数有限的情况下找到更好的解。In the real-world engineering optimization,there are a number of constrained problems,and some of them have time-consuming objective and constrained functions,resulting the evolutionary optimization algorithms are not able to be applied into this kind of problems.Thus,in order to achieve a feasible solution with better objective value in a limited computational budget,the radial basis function surrogate models are trained for objective and constrained functions,respectively.The propagation strategy of each solution will be adaptively determined according to its approximated values on objective and constrained functions,which is expected to improve a good feasible solution.The experimental results on seven benchmark problems and three engineering test problems show that compared to some state-of-the-art algorithms,the proposed method does not need to ensure that there is at least one feasible solution in the initial population,and can find better solutions in a limited computational cost.

关 键 词：约束优化 进化算法 径向基函数 昂贵单目标 

分 类 号：TP[自动化与计算机技术]