• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

计算机工程与科学

• 论文 • 上一篇    下一篇

嵌入共轭梯度法的混合蛙跳算法

庞凯立,梁昔明   

  1. (北京建筑大学理学院,北京 100044)
  • 收稿日期:2016-01-11 修回日期:2016-05-03 出版日期:2017-10-25 发布日期:2017-10-25
  • 基金资助:

    北京市自然科学基金(4122022);中央支持地方科研创新团队项目(PXM2013-014210-000173)

A hybrid SFLA algorithm based on conjugate gradient method
 

PANG Kai-li,LIANG Xi-ming   

  1. (School of Science,Beijing University of Civil Engineering and Architecture,Beijing 100044,China)
  • Received:2016-01-11 Revised:2016-05-03 Online:2017-10-25 Published:2017-10-25

摘要:

针对基本蛙跳算法在处理复杂函数优化问题时求解精度低且易陷入局部最优的缺点,提出了一种嵌入共轭梯度法的混合蛙跳
算法。该算法在基本蛙跳算法划分模因组的基础上引入共轭梯度法,由于基本蛙跳算法模因组的划分规则,使得排在最后的
青蛙子群个体位置较差,严重影响着整个群体的寻优速度,因而选取排列在后面的一部分模因组使用共轭梯度法进行求解,
这使得算法在进化中后期易跳出局部最优,提高了算法的收敛精度。所得混合蛙跳算法有效结合了基本蛙跳算法较强的全局
搜索能力和共轭梯度法快速精确的局部搜索能力。数值实验结果表明,所提出的改进蛙跳算法较基本蛙跳算法具有更高的收
敛精度,避免了陷入局部最优的缺点,且优化结果更加稳定。
 

关键词: 混合蛙跳算法, 共轭梯度法, 数值实验, 适应度函数

Abstract:

We propose a hybrid shuffled frog leaping algorithm (SFLA) based on the conjugate gradient method to solve
the problem of low solution precision and easy to fall into local optimal solutions of the traditional SFLA.
Since the rules of dividing meme groups make the fitness values of the last frog subgroup too bad, which
affects the searching speed of the SFLA in the whole group, we introduce the conjugate gradient method into
the SFLA. We select the last meme groups to search the solution by using the conjugate gradient method, which
prevents the hybrid SFLA from falling into local optimal solution and improves the convergence accuracy. The
proposed hybrid SFLA algorithm combines the strongly global searing ability of the SFLA with the fast local
searching ability of the conjugate gradient method. Numerical experiment results show that the hybrid SFLA
has a higher convergent precision, more stable optimal results, and jumps out of local optimal solution
easily in comparison with the basic SFLA.
 

Key words: shuffled frog leaping algorithm, conjugate gradient method, numerical experiment, fitness function