一种改进的自适应步长的人工萤火虫算法

PDF下载HTML

Title:

An improved adaptive step glowworm swarm optimization algorithm

作者:

唐少虎^1,2, 刘小明^1,2

1. 北方工业大学 电气与控制工程学院, 北京 100144;
2.北方工业大学 城市道路交通智能控制技术北京市重点实验室, 北京 100144

Author(s):

TANG Shaohu^1,2, LIU Xiaoming^1,2

1. College of Electrical and Control Engineering, North China University of Technology, Beijing 100144, China;
2. Beijing Key Lab of Urban Intelligent Traffic Control Technology, North China University of Technology, Beijing 100144, China

关键词:

人工萤火虫算法; 自适应步长; 觅食行为; 全局收敛性

Keywords:

glowworm swarm optimization; adaptive step; foraging behavior; global convergence

分类号:

TP183

DOI:

10.3969/j.issn.1673-4785.201403025

文献标志码:

A

摘要:

在基本的人工萤火虫算法中, 萤火虫的固定移动步长导致算法容易陷入局部最优并可能出现函数适应值的震荡现象.在一些自适应步长的人工萤火虫算法(A-GSO)中, 算法迭代过程中会出现一些萤火虫的邻域集合为空集的现象, 这将导致算法收敛速度降低并陷入局部最优值.为此, 设计了改进的自适应步长的人工萤火虫算法(FA-GSO), 改进的算法针对邻域无同伴的萤火虫引入觅食行为寻找优化方向并自适应调整移动步长, 进一步提高求解精度和稳定性, 并给出了算法的收敛性分析, 结合GSO、A-GSO 2种算法对多个标准测试函数进行寻优并提取相关指标.通过指标对照, 验证了FA-GSO算法的有效性, 表明算法可以改善函数寻优的精度并提高迭代速度.

Abstract:

In the basic glowworm swarm optimization (GSO), it is easy to fall into local optimum and the oscillation phenomenon of function adaptive values may occur because of the fixed step length. In some adaptive-step glowworm swarm optimization (A-GSO) algorithms, neighborhood sets of some fireflies may be empty in the iterative process of the algorithm, which leads to lower convergence speed and falls into local optimal value. Therefore, an improved foraging-behavior adaptive-step GSO (FA-GSO) algorithm was designed. The foraging behavior of the fireflies without neighborhood peer and adaptive step is introduced in order to find the optimization direction in the improved algorithm. The precision, stability, and global convergence analysis of FA-GSO is presented. After extracting and comparing the relevant optimization indicators of GSO, A-GSO and FA-GSO by several standard test functions, the effectiveness of the FA-GSO algorithm was verified, which indicates that the improved algorithm can improve the accuracy of function optimization and the iteration speed.