Scholarlink

<p>RIME, a physics-based heuristic algorithm, simulates the natural phenomenon of rime generation and possesses a robust capacity for global exploration, enabling it to escape local optima. However, testing revealed that RIME suffers from slow convergence during the later stages of evaluation, weak individual exploitation capabilities, and subpar population quality when addressing numerical function optimization problems. This paper proposes a fast-convergence soft-rime search strategy to address these issues by enhancing the soft-rime coefficient, a critical parameter, to mitigate slow convergence in RIME's later evaluation stages. Additionally, the concepts of collaboration and competition, inherent in swarm intelligence-based algorithms, are introduced through the hard-rime puncture strategy, aimed at improving individual exploitation and population quality in RIME. An improved version, termed CGWRIME, is developed by integrating the proposed strategy with a comprehensive learning approach. Subsequently, qualitative analyses and ablation experiments validate the algorithm's search characteristics and the proposed strategy's effectiveness. Comparative experiments with well-known heuristic algorithms and high-performing metaheuristic algorithms confirm CGWRIME's advantages in unconstrained mathematical optimization. Finally, it is applied to five engineering design optimization cases, demonstrating that CGWRIME excels in managing unconstrained numerical functions and provides significant benefits in solving practical engineering optimization problems with constraints.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

CGWRIME: collaboration and competition-boosted RIME optimizer for engineering optimization problems

  • 期刊论文

  • 发表日期：

    2025年1月6日

查看全文
  • Zhen Wang,
  • Dong Zhao,
  • Ali Asghar Heidari,
  • Huiling Chen,
  • Guoxi Liang

摘要

RIME, a physics-based heuristic algorithm, simulates the natural phenomenon of rime generation and possesses a robust capacity for global exploration, enabling it to escape local optima. However, testing revealed that RIME suffers from slow convergence during the later stages of evaluation, weak individual exploitation capabilities, and subpar population quality when addressing numerical function optimization problems. This paper proposes a fast-convergence soft-rime search strategy to address these issues by enhancing the soft-rime coefficient, a critical parameter, to mitigate slow convergence in RIME's later evaluation stages. Additionally, the concepts of collaboration and competition, inherent in swarm intelligence-based algorithms, are introduced through the hard-rime puncture strategy, aimed at improving individual exploitation and population quality in RIME. An improved version, termed CGWRIME, is developed by integrating the proposed strategy with a comprehensive learning approach. Subsequently, qualitative analyses and ablation experiments validate the algorithm's search characteristics and the proposed strategy's effectiveness. Comparative experiments with well-known heuristic algorithms and high-performing metaheuristic algorithms confirm CGWRIME's advantages in unconstrained mathematical optimization. Finally, it is applied to five engineering design optimization cases, demonstrating that CGWRIME excels in managing unconstrained numerical functions and provides significant benefits in solving practical engineering optimization problems with constraints.