Scholarlink

<p>The Pelican Optimization Algorithm (POA), a widely studied swarm intelligence method, has been applied to complex real-world problems but suffers from declining population diversity, limited global search, slow convergence, and a tendency to fall into local optima. To address these shortcomings, the Circle Nonlinear Dynamic Lévy Pelican Optimization Algorithm (CNDLPOA) is proposed. CNDLPOA employs Circle chaotic mapping for population initialization to enhance diversity, introduces a nonlinear inertia weight strategy during the exploration phase to strengthen global search and accelerate convergence, and develops a dynamically scaled Lévy flight strategy based on the traditional Lévy flight to improve local search and balance exploration with exploitation. Performance validation on the CEC2005 test suite, with comparisons against several popular swarm intelligence algorithms and statistical verification using the Wilcoxon rank-sum test, demonstrates superior results on 18 of 23 functions, comparable performance on 5, and no significant decline. Application to gear train design optimization further confirms effectiveness and feasibility.</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

CNDLPOA: a pelican optimization algorithm enhanced with multiple strategies

  • 期刊论文

  • 发表日期：

    2026年6月17日

查看全文
  • Mengran Zhou,
  • Yue Wen,
  • Feng Hu,
  • Yue Chen,
  • Zhengwei Chen,
  • Zhicheng Fang

摘要

The Pelican Optimization Algorithm (POA), a widely studied swarm intelligence method, has been applied to complex real-world problems but suffers from declining population diversity, limited global search, slow convergence, and a tendency to fall into local optima. To address these shortcomings, the Circle Nonlinear Dynamic Lévy Pelican Optimization Algorithm (CNDLPOA) is proposed. CNDLPOA employs Circle chaotic mapping for population initialization to enhance diversity, introduces a nonlinear inertia weight strategy during the exploration phase to strengthen global search and accelerate convergence, and develops a dynamically scaled Lévy flight strategy based on the traditional Lévy flight to improve local search and balance exploration with exploitation. Performance validation on the CEC2005 test suite, with comparisons against several popular swarm intelligence algorithms and statistical verification using the Wilcoxon rank-sum test, demonstrates superior results on 18 of 23 functions, comparable performance on 5, and no significant decline. Application to gear train design optimization further confirms effectiveness and feasibility.