为提高4目标以上高维多目标优化问题的求解性能,提出一种基于改进??支配排序的高维多目标进化算法(KS-MODE)。该算法针对??支配的支配关系和排序方法进行改进,避免循环支配并增强选择压力;设计新的全局密度估计方法提高局部密度估计精确性;设计新的精英选择策略和适应度值评价函数;采用CAO局部搜索算子加速收敛。在4~30个目标标准测试函数上的实验结果表明, KS-MODE能够在保证解集分布性的同时大幅提升收敛性和稳定性,能够有效求解高维多目标优化问题。
To improve the convergence performance in dealing with multi-objective optimization problems(MOPs), a multi-objective evolutionary algorithm(MOEA) based on improved K-dominated sorting(KS-MODE) is proposed. KS-MODE improves K-dominance to avoid the circular dominance and to enhance the selection pressure. The new global density estimation method, the elitist selection strategy and fitness evaluation function are also designed. In addition, the CAO operator is adopted to accelerate convergence. Simulation results on MOPs with 4-30 objectives show that, KS-MODE significantly outperforms several state-of-the-art MOEAs in terms of convergence, distribution and stability.