高维多目标进化算法研究综述

Survey on large-dimensional multi-objective evolutionary algorithms

传统的多目标进化算法能够有效地解决2个或3个目标的优化问题,但当优化目标超过4维即具有高维目标时,其优化效果将大大下降,因此高维多目标进化算法的研究得到了较多的关注.鉴于此,对高维多目标进化算法的研究进展进行系统地分类综述,分析了高维目标对优化算法造成的困难以及改进的可视化技术;总结了各类算法的特点与缺陷,并给出进一步可能的研究方向.

The conventional multi-objective evolutionary algorithms （MOEAs） can solve two-objective optimization problems successfully,but their search ability and performance will deteriorate badly when the number of objectives exceeds four. So,large-dimensional multi-objective evolutionary algorithms are attracting more attention. The large-dimensional multi-objective evolutionary algorithms are surveyed systematically by categories. The influences of large-dimensional objectives bringing on optimization problems are analyzed,and the visualization techniques are introduced. Finally,the proposed algorithms are evaluated and topics for future research are suggested.