基于多种群的高维多目标混合进化算法

Many-objective hybrid evolutionary algorithm based on multi-population

目的基于多种群的高维多目标混合进化算法求解高维多目标优化问题。方法使用K-means聚类将初始种群划分为若干个子种群,引入粒子群优化算法加快种群的收敛速度;引入遗传算法提高解的质量;引入差分进化算法维护种群的多样性。此外,提出基于角度选择的存档机制进行子种群间的信息交流,进一步增加了种群的多样性。结果与结论在DTLZ标准测试集函数上进行仿真实验,数值结果表明MaOEA_MP在大多数测试实例上具有较好的收敛性与多样性。

Purposes—To solve the many-objective optimization problems by using many-objective hybrid evolutionary algorithm based on multi-population.Methods—In this algorithm,the K-means clustering is used to divide initial population into several subpopulations,and particle swarm optimization algorithm is used to increase the convergence speed of the population,genetic algorithm is introduced to improve the quality of the solution,and differential evolution algorithm is adopted to maintain the diversity of the population.In addition,an archive mechanism based on angle selection is proposed to exchange search information among the different subpopulations,which increases the diversity of the population.Result and conclusion—The simulation results on DTLZ standard test function set show that MaOEA_MP has good convergence and diversity on most test functions.