改进的混沌理论与自适应惯性权重PSO算法提升优化性能

129 浏览量更新于2024-08-26 3 收藏 168KB PDF 举报

本文献主要探讨了一种融合混沌理论与自适应惯性权重的粒子群优化算法（Particle Swarm Optimization Algorithm Based on Chaotic Theory and Adaptive Inertia Weight, PSO-CTAIW）。传统粒子群优化算法(Particle Swarm Optimization, PSO)在搜索过程中存在固定惯性权重和早熟收敛的问题。惯性权重对算法性能至关重要，它影响着粒子的速度更新，过小可能导致搜索效率低下，过大则可能陷入局部最优。作者通过深入分析了惯性权重、种群大小、粒子适应度和搜索空间维度之间的关系，提出了一种动态自适应调整策略，旨在增强算法的全局和局部优化能力。为解决早熟收敛问题，该算法引入了混沌映射方法。混沌映射能够增加种群的多样性，避免粒子在搜索过程中过于集中在某个区域，从而减少陷入局部最优的可能性。同时，通过结合负梯度方向，算法能够有效地调整群体极值，提高全局最优解的探索效率。文章作者AN Peng详细验证了提出的算法改进策略的有效性和正确性，通过在多目标优化问题上的实验，展示了PSO-CTAIW在复杂问题求解中的优越性能。这一研究成果对于提升粒子群优化算法的性能，尤其是在处理高维优化问题时，具有重要的理论和实际意义，为优化算法的设计提供了新的思路和改进方向。这篇研究论文旨在解决PSO算法的传统局限，通过混沌理论和自适应机制，提高了算法的优化效果，有助于拓展粒子群优化技术在纳米电子学和光电子学等领域中的应用潜力。

Delivered by Ingenta to: Jiaxing Wen

IP: 127.0.0.1 On: Thu, 14 Sep 2017 02:48:33

ARTICLE

Printed in the United States of America

Journal of

Nanoelectronics and Optoelectronics

Vol. 12, pp. 404–408, 2017

www.aspbs.com/jno

Partic le Swarm Optimization Algorithm Based on

Chaotic Theory and Adaptive Inertia Weight

AN Peng

Fixed inertia weight and premature convergence are obvious ﬂaws of particle swarm algorithm. On the basis

of a detailed analysis of the relationship among the inertia weight, population size, particle ﬁtness and search

space dimension, a dynamic adaptive adjustment strategy for inertia weight is proposed, which effectively

enhances the global and local optimization ability of the algorithm. For premature problem, the chaotic map-

ping method is used to increase the diversity of the population, while taking advantage of the negative gradient

direction to adjust group extreme, greatly reducing the probability of fall into the local extreme. The correct-

ness and effectiveness of the proposed algorithm are demonstrated by comparison with other algorithms on

multiple test functions commonly used.

Keywords: Particle Swarm Optimization Algorithm, Chaotic, Iner tia Weight, Adaptive.

1. INTRODUCTION

Swarm intelligence (SI) algorithms have been a popu-

lar research area since 1990s.

1–3

These algorithms are

inspired by those social behaviors of animals, such as ant

colonies, bird ﬂocking, animal herding, bacterial g rowth,

and ﬁsh schooling. Among the published works, Parti-

cle Swarm Optimization (PSO) is a typical SI algorithm

which is widely used in multi-restriction optimization,

multi-purpose optimization, non-linear complex optimiza-

tion an d complicated network optimization, where some

good results have been achieved.

4–7

PSO is a typical algorithm based on group behavior

involvement, and it has a lot of advantages: a simple def-

inition, a small number of parameters, and it is easy to

implement in programming.

8–10

However, it also has some

obvious weaknesses, such as it is slow to converge, com-

mon premature convergence, and the solution precision is

low, etc. Many works have been proposed to tackle these

ﬂaws, and a lot of measures have been presented. Shi and

Eberhart discussed the parameter selection and inertia

parameter setting, and proposed an PSO algorithm based

on adaptive inertia weight.

Chatterjee and Siarry pro-

posed a non-linear adjustment algorithm for inertia weight,

which greatly improved the convergence speed.

In order

School of Electronics and Information Engineering, Ningbo University

of Technology, Ningbo 315016, China

Email: anp04@126.ccom

Recei ved: 5 March 2016

Accepted: 28 July 2016

to suppress premature convergence, a local PSO algorithm

was proposed which maintains a good variety of groups by

making a particle to study its best neighboring domain.

With the development of o ther swarm algorithms, more

works o n the combination of various algorithms have been

investigated.

14–18

Frans Van den Bergh et al. h ave studied

the utilization of SVM to improve the PSO and effectively

prevented local optimal solution, then the algorithm was

applied to the real time stability control of aerial eng ine

and the satisfactory results were achieved.

Liang et al.

took the gene algorithm into consideration and proposed

a new hybrid GA-PSO algorithm, which introduces cross

and mutation to classic PSO that can speed up convergence

and ﬁnd the global optimum solution.

In this paper, we try to improve the classic PSO algo-

rithm in two aspects: ﬁrstly, a new inertia weight updating

algorithm, which analyzes the relationship among the pop-

ulation size, particle ﬁtness and search space dimension

to improve the self-adapting ability of the algorithm and

secondly, the premature convergence in PSO. Herein, we

propose a fast convergence method based on chaotic the-

ory. The detailed experiments show that the proposed PSO

algorithm greatly improves the performance.

2. OVERVIEW OF PSO ALGORITHM

2.1. The Principle of Classic PSO Algorithm

PSO algorithm was ﬁrst proposed by Kennedy and

Eberhart.

The idea comes from artiﬁcial life and

evolution computatio n technique theor y and the initial

404 J. Nanoelectron. Optoelectron. 2017, Vol. 12, No. 4 1555-130X/2017/12/404/005 doi:10.1166/jno.2017.2033

下载后可阅读完整内容，剩余4页未读，立即下载

weixin_38611508

粉丝: 1
资源: 884

改进的混沌理论与自适应惯性权重PSO算法提升优化性能

基于局部搜索与混合多样性策略的多目标粒子群算法

基本粒子群算法MATLAB代码

局部搜索解决最优化问题matlab程序

自适应混沌变异粒子群算法

matlab中基于Cricle混沌映射自适应权重的麻雀算法优化BP神经网络

混沌自适应粒子群算法

混沌粒子群优化算法matlab

针对粒子群优化算法陷入局部最优解的问题有什么好的改进方法吗

MATLAB的混沌粒子群优化算法

粒子群算法求权重因子

最新资源