计算机工程与应用 ›› 2018, Vol. 54 ›› Issue (18): 34-39.DOI: 10.3778/j.issn.1002-8331.1707-0154

• 理论与研发 • 上一篇    下一篇

具备反向学习和局部学习能力的磷虾群算法

肖素琼1,2,罗  可1,2   

  1. 1.长沙理工大学 计算机与通信工程学院,长沙 410114
    2.长沙理工大学 综合交通运输大数据智能处理省重点实验室,长沙 410114
  • 出版日期:2018-09-15 发布日期:2018-10-16

Krill Herd optimization algorithm with reverse-learning and local-learning behavior

XIAO Suqiong1,2, LUO Ke1,2   

  1. 1.School of Computer and Communication Engineering, Changsha University of Science and Technology, Changsha 410114, China
    2.Hunan Provincial Key Laboratory of Intelligent Processing of Big Data on Transportation, Changsha University of Science and Technology, Changsha 410114, China
  • Online:2018-09-15 Published:2018-10-16

摘要: 针对磷虾群算法易陷入局部最优、收敛速度慢等缺点,提出了具备反向学习和局部学习能力的磷虾群算法。利用混沌映射和反向学习的思想初始化种群,根据算法迭代次数自适应调整学习维度,对精英个体进行反向学习,能有效保持种群的多样性,选取精英群体,通过自适应的Lévy飞行分布和改进的差分变异算子,提高种群的局部学习能力。这种新颖的元启发方式能加速收敛速度的同时可以保证磷虾群算法的鲁棒性。通过对8个基准函数进行仿真测试,实验结果表明:与最近的KH优化算法相比,该算法在收敛速度、收敛精度等方面得到明显改进。

关键词: 磷虾群优化算法, 种群初始化, 精英反向学习, 差分变异算子, 局部学习

Abstract: As Krill Herd(KH) optimization algorithm is easy to fall into local optimum and has slow convergence velocity, so the paper presents an effective approach, called Krill Herd optimization algorithm with reverse-learning and local-learning behavior. First of all, the population is initialized by the chaos mapping and reverse learning method. Then, the elite individual generate their opposite solutions by elite opposition-based learning of adaptive learning dimension space according to the iterations, which can effectively maintain the population diversity. Finally, this paper selects the elite group and uses adaptive Lévy flight distribution and improved differential mutation operator to enhance the local learning ability. This novel meta-heuristic method can accelerate the global convergence speed while preserving the robustness of the basic KH algorithm. The experimental results on 8 benchmark functions show that the proposed algorithm achieves significant improvements in the convergence speed and convergence accuracy.

Key words: Krill Herd(KH) algorithm, population initialization, elite opposition-based learning, differential mutation operator, local-learning