Scholarlink

<p>The Grey Wolf Optimizer (GWO) faces premature convergence and limited global exploration in complex optimization. To overcome these issues, this paper introduces the Adaptive Step Size and Differential Evolution in Hybrid Grey Wolf Optimizer (ADHGWO), which couples an adaptive step size update for precise search, a differential evolution module to sustain diversity, and a nonlinear parameter control scheme for dynamic exploration-exploitation balance. Extensive experiments show statistically significant gains (Wilcoxon, <i>p</i> &lt; 0.05) over strong baselines on 23 classical functions and the IEEE CEC 2014 suite at 30–100 dimensions, while additional tests on the CEC 2022 suite at 10 and 20 dimensions confirm rank-1 performance with lower means and variances. ADHGWO achieves 25–35% faster convergence with higher final accuracy; a running-time study on CEC 2022 (10D/20D) indicates only modest overhead with the same asymptotic complexity O(T×N×D). An ablation study verifies positive contributions from each component and the largest margin from their combination. Practical effectiveness is further demonstrated on five constrained engineering designs (lower costs) and unmanned aerial vehicle path planning in threat-rich environments (18.7% trajectory-cost reduction), establishing ADHGWO as a robust and versatile optimizer across theoretical and real-world tasks.</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

Dynamic hybrid optimization for complex engineering and UAV path planning in threat-rich environments

  • 期刊论文

  • 发表日期：

    2026年2月2日

查看全文
  • Zhiwu Chen,
  • Hui Chang,
  • Dongliang Han,
  • Mengyao Chen,
  • Hao Cui,
  • Shengwei Zhang,
  • Kaihang Zhang,
  • Xinran Hao,
  • Shuyu Zhang,
  • Dengpan Zhang

摘要

The Grey Wolf Optimizer (GWO) faces premature convergence and limited global exploration in complex optimization. To overcome these issues, this paper introduces the Adaptive Step Size and Differential Evolution in Hybrid Grey Wolf Optimizer (ADHGWO), which couples an adaptive step size update for precise search, a differential evolution module to sustain diversity, and a nonlinear parameter control scheme for dynamic exploration-exploitation balance. Extensive experiments show statistically significant gains (Wilcoxon, p < 0.05) over strong baselines on 23 classical functions and the IEEE CEC 2014 suite at 30–100 dimensions, while additional tests on the CEC 2022 suite at 10 and 20 dimensions confirm rank-1 performance with lower means and variances. ADHGWO achieves 25–35% faster convergence with higher final accuracy; a running-time study on CEC 2022 (10D/20D) indicates only modest overhead with the same asymptotic complexity O(T×N×D). An ablation study verifies positive contributions from each component and the largest margin from their combination. Practical effectiveness is further demonstrated on five constrained engineering designs (lower costs) and unmanned aerial vehicle path planning in threat-rich environments (18.7% trajectory-cost reduction), establishing ADHGWO as a robust and versatile optimizer across theoretical and real-world tasks.