文中在传统粒子群优化(Particle Swarm Optimization,PSO)算法的基础上,提出了智能单粒子优化算法(Intelligent Single Particle Opti mizer,ISPO).与传统的PSO算法不同,该算法采用了一个粒子在解空间中搜索,粒子的位置矢量被分成一定数量的子矢量,并基于子矢量对粒子进行更新.在子矢量更新过程中,通过分析之前的速度更新情况,引入一种新的学习策略,使粒子在搜索空间中能够动态地调整速度和位置,从而向全局最优靠近.实验表明,此算法对大部分标准复合测试函数都具有很强的全局搜索能力,其寻优能力超过了国际上最近提出的基于PSO的改进算法.
Intelligent single particle optimizer(ISPO)is proposed based on conventional particle swarm optimization(PSO).ISPO applies a particle,which is different from conventional PSO,to search in the problem space.The whole position vector of particle is split into a certain number of subvectors,and the particle is updated based on these subvectors.During the process of updating each subvector,a novel learning strategy is introduced based on the analysis of previous velocity subvectors,and the particle adjusts its velocity and position subvector dynamically.Experimental results demonstrate that ISPO has an outstanding ability to find the global optimum.ISPO performs much better than most recently proposed PSO-based algorithms on the optimization of most complicated composition test functions.