计算机科学与探索 ›› 2024, Vol. 18 ›› Issue (8): 2065-2079.DOI: 10.3778/j.issn.1673-9418.2305101

• 理论·算法 • 上一篇    下一篇

动态拓扑结构混合粒子群算法及其应用

王浩丞,李凌   

  1. 1. 湘潭大学 物理与光电工程学院,湖南 湘潭 411105
    2. 苏州中科华影健康科技有限公司,江苏 苏州 215123
  • 出版日期:2024-08-01 发布日期:2024-07-29

Hybrid Particle Swarm Algorithm with Dynamic Topology and Its Applications

WANG Haocheng, LI Ling   

  1. 1. College of Physics and Optoelectronics, Xiangtan University, Xiangtan, Hunan 411105, China
    2. Suzhou Ultimage Health Technology Co., Ltd., Suzhou, Jiangsu 215123, China
  • Online:2024-08-01 Published:2024-07-29

摘要: 针对传统粒子群算法面对较高维度参数整定问题时所表现出的寻优速度慢、易陷入局部最优解的问题,提出一种动态环形拓扑结构混合粒子群算法(DynRing-hfpso)。该算法以粒子群算法为基底,融合萤火虫算法的优点,通过定义选择逻辑使粒子在迭代过程中各自独立地交替进行全局搜索与局部探索,并以自适应的粒子速度与位置约束方法提高算法在迭代过程中信息的利用率。对粒子群拓扑结构进行改进,以动态多邻域环形拓扑结构提高搜索空间覆盖能力,均衡收敛速度。设置动态性能分析以及消融实验,验证算法中粒子分布质量与改进措施的有效性。采用分数阶比例-积分-微分([FOPIλDμ])控制下的速度伺服系统作为应用场景,将该算法与其他四种算法进行对比。结果表明,DynRing-hfpso算法较已有的元启发式优化算法有更快的收敛速度与更优的收敛精度,且在多次实验中展现出更强的鲁棒性。

关键词: 环形拓扑结构, 粒子群算法, 萤火虫算法, 分数阶PID, 自整定

Abstract: The traditional particle swarm optimization algorithm is slow to find the optimal solution and easy to fall into the local optimal solution when facing the parameter tuning problem of higher dimensions. This paper proposes a dynamic ring topolopy hybrid firefly and particle swarm optimization (DynRing-hfpso) algorithm. This algorithm takes the particle swarm algorithm as the base, incorporates advantages of firefly algorithm, enables the particles to alternate between global search and local exploration independently in the iterative process by defining the selection logic, and improves the utilization of information in the iterative process of the algorithm with an adaptive particle velocity and position constraint method. In addition, the particle swarm topology is improved to enhance the search space coverage and equalize the convergence speed with dynamic multi-neighborhood ring topology. Dynamic performance analysis and ablation experiments are set up to verify the effectiveness of the particle distribution quality and improvement measures in the algorithm. Then, a velocity servo system under fractional order proportion integration differentiation [(FOPIλDμ)] control is used as an application scenario to compare the performance of this algorithm with the other four algorithms. The results show that the DynRing-hfpso algorithm has faster convergence speed and better convergence accuracy than the existing metaheuristic optimization algorithms, and shows stronger robustness in several experiments.

Key words: ring topology, particle swarm optimization, firefly algorithm, fractional order PID (proportion integration differentiation), self-tuning