基于弱关联的自适应高维多目标进化算法

对现有的分解方法进行改进,提出一种基于弱关联的自适应高维多目标进化算法(WAEA).首先,提出一种基于夹角子空间的关联策略,使得一个解能与多个参考向量相关联;其次,提出弱关联概念,并基于此概念设计双模态标量函数,使算法能够更好地处理复杂PF问题,此外,算法通过检测参考向量子空间内解的数量,自适应调整惩罚参数大小,使其能...

A weak association-based adaptive evolutionary algorithm for many- objective optimization

This paper proposes a weak association-based adaptive evolutionary algorithm(WAEA) on many-objective optimization by improving the previous decomposition approaches. Firstly, an association strategy has been presented based on the angle subspace, which can make a solution associated with multiple reference vectors. Then, the idea of weak association has been employed to design a bimodal scalar function which improves the capability of dealing with the complex PF problem. Moreover, through the detection of the number of solutions in the reference vector subspace, the proposed algorithm is capable of doing self-adaption to adjust the size of penalty parameters to efficiently deal with multi-type issue on many-objective optimization. Finally, the proposed WAEA is compared with eight representative many-objective based algorithms, respectively. The results show that the WAEA has the capability of gaining better balance of the Pareto optimum in convergence and diversity while dealing with high-dimensional many-objective problems.