分阶段二次变异的多目标混沌差分进化算法

Multi-objective chaotic differential evolution algorithm with grading second mutation

提出一种结合分阶段二次变异和混沌理论的改进差分进化(DE)算法,以解决多目标约束优化问题.其核心思想是,在DE进化前期采用基于非支配解的随机二次变异来提高算法的全局寻优能力,进化后期采用基于非支配解的混沌二次变异来提高DE的局部寻优能力.通过对典型测试问题的仿真实验验证了所提出的算法能在全局搜索性能与局部搜索性能之间维持较好平衡,而且保持了DE算法的简洁性能,其收敛性、分布度和均衡性均优于标准DE.

To solve the multi-objective constraint optimization problem, this paper proposes an advanced differential evolution（DE）. In the proposed algorithm, grading second mutation and chaotic theory are combined into standard DE. At early evolution process of DE, random second mutation based on non-dominance Pareto solution is adopted in order to improve global exploring ability. And in the later evolution process, the chaotic second mutation based on non-dominance Pareto solution is added into DE evolution operation in order to enhance local searching ability of algorithm. By testing benchmarks functions, simulation results show that, this algorithm has better convergence and distribution property, and is superior to standard DE in keeping balance between diversity and convergence.