一种基于模糊学习子群的多目标粒子群算法

摘要

为提高多目标粒子群算法的局部搜索能力,提出了一种模糊学习子群多目标粒子群算法(FLSMOP-SO).在搜索过程中,每个粒子模糊自适应学习生成不确定的p个粒子形成一个子群而不是只产生一个新粒子,然后在其中选择模糊满意解作为其下一代新粒子.对四个典型测试函数的实验结果表明,新算法比NSGAⅡ和MOPSO两种经典多目标优化算法有显著的优越性.%To improve the local search ability of the MOPSO,this paper put forward a new fuzzy learning sub-swarm multi-ob-jective particle swarm optimization ( FLSMOPSO). In the searching process, each particle in the swarm could have linear re-gressive p particles by self-adaptive learning to form a sub-warm rather than a single particle. Then selected a fuzzy satisfied so-lution particle as the new position of the particle. Comparative analysis to the two typical algorithm shows that the new algorithm have prominent advantages.