为了解决多目标优化过程中各个解之间存在的资源争夺、冲突，算法由于趋同性而带来的早熟无法收敛等缺点，文中提出了一种多子种群协同优化粒子群算法。算法分别采用不同的种群优化不同的目标，并且在算法中引入外部档案和精英学习策略，使得算法能够得到更多的外部档案的解供选择，精英学习策略是为了使算法的分布性和收敛性更好。最后将算法应用到多目标测试函数中，通过实验验证了改进后的算法的收敛性和分布性都比经典多目标算法NSGA-II要好。%To solve the problem that resource contention and conflict between the various solutions in multi-objective optimization pro-cessing,and can't be convergence duo to the precocious brought by convergence,introduce a multi-sub-population co-evolution mecha-nism to overcome these shortcomings. The algorithm has adopted different populations to optimize different targets. Meanwhile,it intro-duces an external archive and elite learning strategies,in this way it can obtain more solutions of external archive to choose. Elite learning strategies makes the algorithm has a better distribution and convergence. Finally,the algorithm is applied into the multi-objective test function,the experimental results show that the improved algorithm has a better convergence and distribution than NSGA II.
展开▼
