基于源域选择的动态多目标优化算法  

Dynamic Multi-objective Optimization Algorithm Based on Source Domain Selection

作　　者：上官晨曦 时振涛 SHANGGUAN Chen-xi;SHI Zhen-tao(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：《太原科技大学学报》2024年第4期336-341,共6页Journal of Taiyuan University of Science and Technology

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

摘　　要：迁移学习是一种解决动态多目标优化问题的有效方法,但当源域和目标域差异性较大时,会产生负迁移,大大降低求解优化问题的效率。针对这种现象该算法提出一种基于源域选择策略的迁移学习方法。该方法首先根据历史环境的最优解集与新环境目标域的差异性对历史环境数据进行排序,选择一个差异性最小的历史环境数据作为迁移解;同时,对t时刻环境的最优解集进行交叉变异生成多样性解,将其与迁移解合进行非支配排序得到源域数据;然后将源域数据映射到嵌入空间,求出最优解作为新环境下的初始种群进行下一时刻迭代运算。这种方法考虑了多个历史环境知识的重用,可以加强种群全局搜索能力,有效抑制负迁移的产生,从而提高算法效率。通过实验,结果证明本文提出的算法能显著提高动态多目标优化方法的性能。Transfer learning is an effective method to solve dynamic multi-objective optimization problems,but when the difference between the source domain and the target domain is large,negative transfer will occur,which greatly reduces the efficiency of solving optimization problems.Aiming at this phenomenon,this algorithm proposes a transfer learning method based on the source domain selection strategy.The method first sorts the historical environmental data according to the difference between the optimal solution set of the historical environment and the target domain of the new environment,and selects a historical environmental data with the smallest difference as the migration solution;at the same time,the optimal solution set of the environment at time Performs crossover mutation to generate diversity solutions,and perform non-dominated sorting with migration solutions to obtain source domain data;then map the source domain data to the embedding space,and obtain the optimal solution as the initial population in the new environment for the next iteration.operation.This method considers the reuse of multiple historical environmental knowledge,which can strengthen the global search ability of the population and effectively suppress the generation of negative migration,thereby improving the efficiency of the algorithm.Through experiments,the results show that the algorithm proposed in this paper can significantly improve the performance of dynamic multi-objective optimization methods.

关 键 词：动态多目标优化 迁移学习 源域选择策略 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]