【作者】 王建丽；

【导师】 王希云； 夏桂梅；

【作者基本信息】 太原科技大学 ， 应用数学， 2010， 硕士

【摘要】 微粒群算法(particle swarm optimization,简称PSO)模拟的是鸟群寻找栖息地的行为。微粒群算法简单、易于实现、收敛速度快且需要调整的参数少,自该算法提出以来引起了诸多学者的广泛关注。论文选取如下两个方面的问题作为研究内容:(1)在前人工作的基础上将随机微粒群算法和协同进化结合在一起寻求一种更有效的算法。(2)把微粒群算法和传统优化算法(如信赖域算法、共轭梯度法、最速下降法等)相结合尝试寻找更有效的优化算法。论文的研究内容主要包括以下几个部分:第一章是绪论,主要介绍了微粒群算法的历史和现状,本文的创新和突破,以及本文的现实意义。第二章在多种群协同进化和随机微粒群算法基础上,提出了一种改进的多种群随机微粒群算法,将各个子种群独立地按照随机微粒群进化,周期性的更新共享信息。第三章在随机微粒群算法和函数梯度信息基础上,提出了基于梯度的随机微粒群算法。数值计算表明算法对于求解连续可微函数的全局优化问题是非常有效的。第四章提出一种基于协同进化的单纯形随机微粒群算法。该算法采用多个优化种群,分别在奇数种群和偶数种群并行运行随机微粒群法和单纯形法,周期性更新相邻种群最优信息。通过优化两个典型的测试函数验证了算法的有效性。更多还原

【Abstract】 Particle Swarm Optimization(PSO) is a random optimization algorithm .The basic idea of PSO comes from the research of the behavior of birds find habitat. Since the algorithm has been proposed, because of its easy implementation, fast convergence, and the need to adjust less parameters, attracted wide attention from many scholars. The following two problems have been chosen as the main goal in the dissertation.(1) Construct a few algorithms based on SPSO and cooperative evolutionary.(2) Develop a mixed search method by combining SPSO with other optimization algorithms.The content of this paper is organized into four chapters.In chapter I, we introduce the history and status about particle swarm optimization, the creativities and the practical importance of this paper.In chapter II, A modified cooperative stochastic particle swarm optimization(CSPSO), is presented based on the analysis of the SPSO and the cooperative evolutionary PSO with multi-populations. The whole group is divided into several sub-groups. Every subgroup evolved independently and updated sharing information periodically.In chapter III, The stochastic particle swarm optimization basied the gradient method, is presented based on the analysis of the SPSO and the gradient of the continuous-differential function.Numerical experiments have proved that this hybrid algorithm is very reasonable.In chapter IV, A cooperative Simplex Method-Stochastic Particle Swarm Optimization(SM-SPSO) is proposed, The conception of multipopulations is adopted in this method,where SPSO and SM run on odd populations and even populations,respectively. Experimental results on optimization two benchmark functions demonstrate its usefulness.更多还原