粒子群算法改进及其应用(硕士论文).pdf

摘要
粒子群算法改进及应用

摘要

粒子群优化算法最早是由 Eberhart 和 Kennedy 模拟自然界的生物群体觅
食提出的一种群智能化方法。后来 Shi 等人引入惯性权重来更好的控制收敛和
探索,形成了当前的标准 PSO 算法。由于该算法实现简单,需要调整的参数
少,已被广泛地应用于函数优化、通信系统设计、电子系统设计以及经济管
理等领域。
粒子群算法已经被国内外学者认为是一种有效的优化方法,但是自身也
存在着一些缺点,比如在搜索后期易陷入局部最优和出现早熟现象。如何加
快粒子群算法的收敛速度和避免出现早熟收敛,一直是研究者关注的重点。
本文在基本粒子群算法的基础上,进行了一些改进。引入云理论把粒子群分
为三个种群,用云方法修改粒子群算法中惯性权重,同时修改速度更新公式
中的“认知部分”和“社会部分”,加入“均值”的概念,提出一种基于均值
的云自适应粒子群算法;考虑惯性权重对算法的影响,较大的权值有利于提
高算法的全局搜索能力,而较小的权值会增强算法的局部搜索能力。提出了
一种基于位置多样性和种群多样性来修改惯性权值的粒子群优化算法。让惯
性权值随着位置移动的长短和适应度的大小来改变。最后把改进的方法应用
在求解工程约束优化问题中。数值实验结果表明,改进的算法对于高维非线
性的无约束优化问题表现出了良好的性能,对工程实例的约束优化问题也显
示了其优越性。
将人工萤火虫算法与粒子群算法结合提出一种基于萤火虫算法感知范围
的粒子群算法。并应用到求解工程实例约束优化问题中,实验结果也表明了
改进算法的有效性和正确性。

关键词:粒子群自适应云理论均值约束优化萤火虫算法








I
ABSTRACT

IMPROVED AND APPLICATION BASED
ON PARTICLE SWARM OPTIMIZATION



ABSTRACT


The particle swarm optimization was a kind of modern optimization method
that was proposed by Eberhart and Kennedy through mimic natural biological
community grazing. Later, Shi, who was the introduction of inertia weight to better
control the convergence and, thus, the current standard PSO algorithm. Because the
algorithm is simple, needs to adjust the few parameters, has been widely applied to
function optimization, communication system design, electronic system design and
economic management, etc.
Particle swarm optimization is thinking an efficient optimization method by
the domestic and overseas scholars, but oneself also exist some ings, such
as easily trapped into local optimal in the later and premature phenomenon. How to
speed up the particle swarm algorithm convergence speed and avoid premature
convergence is always the most researchers’ focus of concern. In this paper, based
on the standard particle swarm algorithm, some improvements were made.
Introducing cloud theory, the particle swarm is divided into three populations. It is
modified inertia weight using cloud method,