计算机工程与应用 ›› 2019, Vol. 55 ›› Issue (9): 124-129.DOI: 10.3778/j.issn.1002-8331.1801-0457

• 模式识别与人工智能 • 上一篇    下一篇

基于近邻套索算子的磷虾群算法

谈  庆,黄樟灿   

  1. 武汉理工大学 理学院,武汉 430070
  • 出版日期:2019-05-01 发布日期:2019-04-28

Krill Herd with Nearest Neighbor Lasso Operator

TAN Qing, HUANG Zhangcan   

  1. School of Science, Wuhan University of Technology, Wuhan 430070, China
  • Online:2019-05-01 Published:2019-04-28

摘要: 针对标准磷虾群算法(KH)在求解复杂函数优化问题时局部搜索能力差,开采能力不足的问题,提出了一种基于近邻套索算子的磷虾群算法(NLKH)。该算法将一种新的近邻套索算子加入了标准磷虾群算法,使得处理复杂函数优化问题更加有效。近邻套索算子通过比较磷虾个体之间的欧式距离来选取目标磷虾对,然后通过在优质个体附近加速操作产生新磷虾个体和剔除劣质磷虾个体的方式,提高了磷虾个体局部搜索的能力。通过比较PSO算法、KH算法、KHLD算法、NLKH算法在10个测试函数上的结果表明,NLKH算法相较于PSO算法、KH算法和KHLD算法有着更强全局搜索能力,寻优精度更高,收敛速度更快,稳定性更好。并且NLKH算法相较于KH算法和KHLD算法有着更强的局部勘测能力,开采能力更强。

关键词: 磷虾群算法, 近邻套索算子, 函数优化, 开采能力

Abstract: Aiming at the problem that basic Krill Herd(KH) algorithm has poor local search ability and insufficient exploitation capacity when solving complex function optimization problems, an NLKH algorithm based on nearest neighbor lasso operator is proposed in this paper. In this algorithm, a new nearest neighbor lasso operator is added to the basic krill herd algorithm, which makes it more effective to deal with the optimization problem of complex functions. The nearest neighbor lasso operator selects the target krill pair by comparing the Euclidean distance between krill individuals. Then the local search ability of the krill individual is improved by accelerating the operation of high quality krill individuals and eliminating poorly performing krill individuals. By comparing PSO algorithm, KH algorithm, KHLD algorithm and NLKH algorithm on 10 test functions, the results show that NLKH algorithm has stronger global search ability, higher accuracy, faster convergence speed and better stability. And the NLKH algorithm has stronger local search ability than the KH algorithm and KHLD algorithm, and exploitation ability is also stronger.

Key words: krill herd algorithm, nearest neighbor lasso operator, function optimization, exploitation ability