基于目标空间划分的自适应多目标进化算法

目前，多目标进化算法在众多领域具有极高的应用价值，是优化领域的研究热点之一.分析已有多目标进化算法在保持种群多样性方面的不足并提出一种基于解空间划分的自适应多目标进化算法（space division basedadaptive multiobjective evolutionary algorithm，简称SDA-MOEA）来解决多目标优化问题.该方法首先将多目标优化问题的解空间划分为大量子空间，在算法进化过程中，每个子空间都保留一个非支配解集，以保证种群的多样性.另外，该方法根据每个子空间推进种群前进的距离，自适应地为每个子空间分配进化机会，以提高种群的进化速度.最后，利用3组共14个多目标优化问题检验SDA-MOEA的性能，并将SDA-MOEA与其他5个已有多目标进化算法进行对比分析.实验结果表明：在10个问题上，算法SDA-MOEA显著优于其他对比算法.

Objective Space Division Based Adaptive Multiobjective Optimization Algorithm

Currently, multiobjective evolutionary algorithm has been applied widely in various fields, and become one of the most attractive topics in the optimization area. This paper analyzes the deficiency of traditional multiobjective evolutionary algorithms in maintaining population diversity, and further proposes an objective space division based adaptive multiobjective evolutionary algorithm (SDA-MOEA) to solve multiobjective optimization problems. The proposed algorithm divides the objective space of a multiobjective optimization problem into a series of subspaces. During the evolution process, each subspace in SDA-MOEA can maintain a set of non-dominated solutions to guarantee the population diversity. Besides, SDA-MOEA self-adaptively distributes the evolutionary opportunities for each subspace according to its forward distance, which can promote the population convergence. Finally, 14 multiobjective problems of three groups are selected to measure the performance of SDA-MOEA. By comparing with five existing multiobjective evolutionary algorithms, the experimental results demonstrate that SDA-MOEA shows obvious superiority over these existing algorithms on 10 problems.