面向全局搜索的自适应领导者樽海鞘群算法
作者:
作者单位:

1.河南大学智能网络系统研究所;2.河南大学软件学院;3.河南省智能网络理论与关键技术国际联合实验室,河南大学

作者简介:

通讯作者:

中图分类号:

TP301.6

基金项目:

河南省重点研发与推广专项;河南大学研究生教育创新与质量提升项目(SYL18060145,SYL18020105)


Global Search-oriented Adaptive Leader Salp Swarm Algorithm
Author:
Affiliation:

1.Institute of Intelligent Network System, Henan University;2.College of Software, Henan University;3.Henan International Joint Laboratory of Theories and Key Technologies on Intelligence Networks, Henan University, Kaifeng, China

Fund Project:

Special R&D and promotion programs in Henan; Graduate Education Innovation and Quality Improvement Project of Henan University (Grant No. SYL18060145 and SYL18020105)

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    为了进一步改善基本樽海鞘群算法容易陷入局部最优,寻优精度有时不高,求解结果不太稳定的不足,提出了一种面向全局搜索的自适应领导者樽海鞘群算法。在领导者位置更新公式中引入上一代樽海鞘群位置,增强了全局搜索的充分性,有效避免算法陷入局部极值。然后在领导者位置更新公式中加入惯性权重,并在全局和局部搜索的选择上引入领导者-跟随者数量自适应调整策略,使算法在迭代前期领导者数目较多且受全局最优解影响较大,能以较大的全局搜索步幅快速收敛到全局最优区域;而在迭代后期领导者步幅较小且跟随者数量较多,可以在最优解附近深度挖掘,提高算法的收敛精度。随后给出了算法流程并对时间复杂度进行了理论分析。最后通过5种代表性对比算法在12个不同特征基准测试函数多个维度上的函数优化仿真实验,测试结果表明改进算法的寻优精度和稳定性均有明显提升。

    Abstract:

    In order to further improve the shortcomings that basic salp swarm algorithm is easy to fall into the local optimum, the optimization accuracy is sometimes not high, and the solution results are not stable, a global search-oriented adaptive leader salp swarm algorithm is proposed. The location of the last generation salp swarm group was introduced into the leader position update formula, which enhanced the sufficiency of global search and effectively avoided the algorithm falling into local extremum.Then add the inertia weight to the leader position update formula, and introduce the leader-follower adaptive adjustment strategy in the choice of global and local search, so that the algorithm has a large number of leaders in the early iteration and is greatly influenced by the global optimal solution, it can quickly converge to the global optimal region with a larger global search step; At the end of the iteration, the leader"s stride is small and the number of followers is large, so the algorithm can be mined deeply near the optimal solution to improve the convergence accuracy. Then the algorithm flow is given and the time complexity is analyzed theoretically. Finally, through the simulation experiment of function optimization of 5 representative comparison algorithms on multiple dimensions of 10 different feature benchmark functions, the test results show that the optimization accuracy and stability of the improved algorithm are significantly improved.

    参考文献
    相似文献
    引证文献
引用本文
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:2020-01-19
  • 最后修改日期:2020-03-22
  • 录用日期:2020-04-03
  • 在线发布日期:
  • 出版日期: