基于多场景建模的动态鲁棒多目标进化优化算法
Dynamic robust multi-objective evolutionary optimization algorithm based on multi-scenario modeling
徐标 1吕修豪 2李文姬 2范衠 3巩敦卫 4贺杰5
作者信息
- 1. 汕头大学工学院,广东汕头 515063;梧州学院广西机器视觉与智能控制重点实验室,广西梧州 543002
- 2. 汕头大学工学院,广东汕头 515063
- 3. 电子科技大学(深圳)高等研究院,广东 深圳 518110
- 4. 青岛科技大学自动化与电子工程学院,山东青岛 266100
- 5. 梧州学院广西机器视觉与智能控制重点实验室,广西梧州 543002
- 折叠
摘要
为了解决实际生产中的动态多目标优化问题,提出一种基于多场景建模的动态鲁棒多目标进化优化算法.首先,所提出算法将不同环境下的问题视为不同场景,并通过相似度计算和场景聚类建立多个场景;然后,利用改进的多场景多目标进化优化算法求解各场景的折中解,当环境发生变化时,根据新问题所属的场景类,直接应用该场景类的折中解作为新问题的最优解,从而加快算法的响应速度;最后,通过对场景类中问题的约减,保留最具代表性的问题,逐步提高算法的鲁棒性,并降低解的切换成本.实验结果表明,所提出算法能够快速响应环境变化,并提高解的鲁棒性.
Abstract
This paper proposes a dynamic robust multi-objective evolutionary optimization algorithm based on multi-scenario modeling,aiming to address dynamic multi-objective optimization problems in practical production.The algorithm treats problems in different environments as different scenarios and establishes multiple scenarios through similarity calculation and scenario clustering.Subsequently,it utilizes an improved multi-scenario multi-objective evolutionary optimization algorithm to find compromise solutions for each scenario.When the environment changes,the algorithm directly applies the compromise solution of the corresponding scenario class as the optimal solution for the new problem,thus speeding up the algorithm's response rate.Through reducing the number of problems in scenario classes and retaining the most representative ones,the algorithm gradually improves its robustness and reduces solution switching costs.Experimental results demonstrate that the proposed algorithm can rapidly respond to environmental changes and enhance solution robustness.
关键词
动态优化/多场景/多目标/鲁棒优化/进化算法Key words
dynamic optimization/multiple scenarios/multi-objective/robust optimization/evolutionary algorithm引用本文复制引用
出版年
2024
NETL
NSTL
万方数据