摘要
研究在多目标优化进化算法中引入强选择压力机制,以促使搜索群体在有效保证多样性的前提下向Pareto最优前沿迅速收敛,并引入空间超体积测度.针对当前空间超体积测度计算代价高的问题,提出了一种基于空间切片的快速空间超体积贡献计算方法FH.基于该方法,发展出一种基于快速计算空间超体积贡献机制的多目标进化算法(FH-MOEA),并应用于解决复杂的多目标优化问题.用一组测试问题对算法性能进行检验,实验结果表明,该算法在收敛性和分布性两方面均比著名的NSGA-Ⅱ算法有显著提高.
The method for incorporating strong selection pressure was introduced into multi-objective evolutionary optimization algorithms （MOEAs） to force the evolution population approaches rapidly towards the Pareto optimal front with a spread as uniform as possible over the Pareto front. An effective measure called ＂hyper-volume contribution＂ was adopted to provide the strong selection pressure. Based on the fast method for calculating hyper-volume contribution proposed, a new multi-objective optimization evolutionary algorithm multi-objective evolutionary algorithm based on fast hyper-volume contribution （FH-MOEA） was proposed for the complex multi objective optimization problem （MOP） tasks. Via a suite of designed experiments, it is distinctly indicated that FH-MOEA has a great advantage over the famous MOEA ＂NSGA- Ⅱ ＂ in terms of both convergence and diversity.
基金
National Natural Science Foundation of China(60401015,60572012)
Natural Science Foundation of Anhuiprovince(050420201).
关键词
多目标优化
超体积贡献
强选择压力机制
遗传算法
智能计算
multi-objectives optimization
hyper-volume contribution
strong selection pressure mechanism
genetic algorithm
intelligence computation
作者简介
Biography：WANG Yu, PhD candidate. Research field： evolutionary computation, multi-objective optimization, link biologic concept to evolutionary computation and application of evolutionary computation in power system domain. E-mail： wyustc@mail. ustc. edu. cn
Corresponding author： LI Bin, PhD/associate professor. E-mail： binli@ ustc. edu. cn
