摘要
基于快速非支配排序以及拥挤距离的NSGA-Ⅱ算法是一种较好的多目标进化算法(Multi-objective Optimization Evolutionary Algorithm,MOEA),降低了非劣排序遗传算法(elitist Non-dominated Sorting Genetic Algorithm,NSGA)的复杂性。为了进一步提高种群的多样性,避免算法过早地收敛于局部最优,提出了一种改进NSGA-Ⅱ算法(IPNSGA-Ⅱ),通过变异目标空间中的重合个体,以及在每一代增加若干个新个体的方法,提高种群的多样性。标准多目标数学测试问题、多目标背包问题和多目标钢铁一体化生产合同计划问题的仿真结果表明,IPNSGA-Ⅱ算法在求解整数规划的多目标组合优化问题时,可以获得很高的覆盖度和1较好的分布多样性。
NSGA-Ⅱbased on fast-non-dominated-sort and crowding-distance is one of the better multi-objective evolutionary algorithms.It could reduce the NSGA's complexity.For the purpose of increasing the diversity and avoiding the algorithm prematurely converging to local optimal,an improved NSGA-Ⅱalgorithm(IPNSGA-Ⅱ) is proposed, which mutates the superposition individuals in the objective space and adds some new individuals on every generation. Simulation results on three type of multi-objective problems,includes standard multi-objective mathematic test problem, multi-objective knapsack problem and multi-objective integrated steel production order planning,show that IPNSGA-Ⅱis able to find better spread of solutions and has higher coverage than NSGAII on multi-objective integer programming combination optimization problem.
出处
《控制工程》
CSCD
北大核心
2009年第S1期61-63,67,共4页
Control Engineering of China
基金
建设部基金资助项目(03-2-117)
辽宁省教育厅基金资助项目(20060699)