提出了一种适应于多目标进化算法的变异越界处理策略,成功地将这些变异算子应用到多目标进化优化问题中,从多目标优化收敛性的角度比较了这些变异算子的性能。通过一组实验表明这种越界处理方法是非常有效的,单目标优化中的这些变异算子具有与多项式变异算子相当的分布性,同时取得了更好的收敛性能。
This paper proposes a mutation over-flow dealing method to fit for the environment of MOEAs,and applies these operators successfully to multi-objective optimization problems.Then it compares these operators' performance through its convergence quality,and it demonstrates that this over-flow dealing method is effective through a group of experiments,the mutation operators used in single-objective optimization have the respectable distribution to polynomial mutation and achieve better convergence qua]ity.