基于维度扰动的快速非支配排序遗传算法II-《郑州大学学报(工学版)》

文章信息/Info

Title:: Non-dominated Sorting Genetic Algorithm II with Dimensionality Perturbation

作者:: 张茂清; 李东洋; 胡博; 汪镭; 崔志华; 郭为安; 1. 同济大学电子与信息工程学院;2. 太原科技大学计算机科学与技术学院;3. 同济大学中德工程学院

Author(s):: Zhang Maoqing 1; Li Dongyang 1; Huber 1Wang Lei 1; Cui Zhihua 2; Guo Wei’an 3; 1. School of Electronic and Information Engineering, Tongji University; 2. School of Computer Science and Technology, Taiyuan University of Science and Technology; 3. Sino-German School of Engineering, Tongji University

Keywords:: NSGA-ⅡＮ; multi-objective optimization algorithmtournament selectiondimensionality perturbation

摘要:: 快速非支配排序遗传算法II(Non-dominated Sorting Genetic Algorithm II, NSGA-II)是经典多目标优化算法然而，其采用的锦标赛选择策略在选择交叉父代时会产生大量重复个体，并进一步导致减少种群个体多样性，降低算法性能。为解决此类问题，本文提出了基于维度扰动的NSGA-II，即通过在待交叉的父代个体每个维度上引入扰动参数改变其值，然后将扰动父代做正常交叉操作产生新后代，以此避免了后代重复个体的产生。为验证算法有效性，本文采用ZDT测试集作为测试函数。与现有算法相比，所提策略可有效地改善算法性能,证明了所提策略的有效性。

Abstract:: Non-dominated Sorting Genetic Algorithm II （NSGA-II） is a classical multi-objective optimization algorithm. The strategy of tournament selection employed in NSGA-II may produce a large amount of repeated individuals, and thus decreases the diversity of population and the overall performance. To tackle this problem, this paper proposed NSGA-II based on Dimensionality Perturbation. Firstly, perturbation parameter is introduced, and then it is further used to modify each dimensionality of the parent individuals to do crossover operator. After that, the modified parent individuals do the crossover operator as usual to avoid generⅡating the repeated offspring individuals. To verify the proposed algorithm, ZDT test suit is employed as benchmark problems. Compared to the state-of-art algorithms, the proposed algorithm is capable of effectively improving the performance of NSGA-II, thus demonstrating the effectiveness of the proposed strategy.