基于MOEA/D的多类多目标优化算法数据集

本文定义了一个新的多目标优化问题，称为多类多目标优化问题（MMOP）。一个MMOP由几个多目标优化问题组成，具有不同的决策空间和相同的目标空间，其最优解是帕累托之间的非支配解所有单个 多目标优化问题的最优解。我们构建了一组具有不同特征的基准测试实例。我们提出一个基于分解的多目标进化算法求解MMOP (MOEA/D-MM)。测试结果表明MOEA/D-MM 比一些著名的传统方法再MMOP问题上更为有效。This paper defines a novel multi-objective optimization problem, termed Multi-class Multi-objective Optimization Problem (MMOP). A MMOP consists of multiple multi-objective optimization problems, which have distinct decision spaces but identical objective spaces, and its optimal solutions are the Pareto non-dominated solutions among the optimal solutions of all individual multi-objective optimization problems. We construct a set of benchmark test instances with diverse characteristics. We propose a decomposition-based multi-objective evolutionary algorithm tailored for solving MMOP, denoted as MOEA/D-MM. Experimental results demonstrate that MOEA/D-MM outperforms several well-established traditional methods when applied to MMOPs.

搜集汇总

背景与挑战

背景概述

该数据集提出了多类多目标优化问题（MMOP），并基于MOEA/D算法开发了解决方案MOEA/D-MM，包含60个文件，数据量4.91MB。测试结果表明该算法在MMOP问题上比传统方法更有效。