针对约束动态多目标优化问题，提出了一种基于种群分类的动态约束多目标进化算法，其基本思想是：首先将每代种群分为不可行群体、非Pareto群体、非聚类Pareto群体和聚类Pareto群体，然后对这4类群体按一定规则赋以适应度，最后采用动态多目标进化算法中的进化算子进行进化操作，产生新种群。数值实验和性能指标统计数据表明，该算法不仅能较好地处理复杂约束，而且能产生分布性较佳的Pareto最优解。%Aiming at the dynamic multi‐objective optimization problem with complex constraints ,a dynamic constraint multi‐objective evolutionary algorithm based on population classification is proposed ,and the basic idea is that first ,every population is divided into four groups :infeasible group ,non‐Pareto group ,non‐clustering Pareto group and clustering Pareto groups ,then these four groups are assigned to fitness according to certain rules ,finally dynamic multi‐objective evolutionary operators are used to produce new population .The numerical experiments and statistical data of performance indicators show that the algorithm not only can deal with complex constraints ,but also can produce the Pareto optimal solutions with better distribution .
展开▼
