Scholarlink

<p>The dung beetle optimizer (DBO) demonstrates strong performance in optimization but suffers from limited global exploration and susceptibility to local optima. To address these issues, we propose an improved algorithm, called Weibull Difference Adaptive t-Gaussian Dung Beetle Optimization (WDTDBO), which integrates Bernoulli chaotic mapping and Lens opposition-based learning for population initialization to increase population diversity, employs a Weibull dynamic flight Strategy to enhance global search, and applies Mean difference perturbation with Adaptive t-Gaussian perturbation to escape local optima. Experimental results show that on the CEC2017 benchmark, WDTDBO improves stability and convergence accuracy by 75.72% and 47.20% in 30 dimensions, and by 60.84% and 60.98% in 100 dimensions. On the 20-dimensional CEC2022 benchmark, improvements reach 77.65% and 27.70%. The average rankings of 1.17, 1.21, and 1.50 on 30D/100D CEC2017 and 20D CEC2022 respectively indicate superior accuracy, convergence efficiency, and robustness compared with DBO and other algorithms. Moreover, experiments on challenging engineering problems further confirm the practical advantages of WDTDBO over competing algorithms.</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

Multi-strategy improved dung beetle optimization algorithm: a swarm intelligence-based metaheuristic algorithm for solving engineering design problems

  • 期刊论文

  • 发表日期：

    2026年7月24日

查看全文
  • Dongfu Xu,
  • Lanyu Zhang,
  • Lei Du,
  • Pu Liu,
  • Chunhe Wang

摘要

The dung beetle optimizer (DBO) demonstrates strong performance in optimization but suffers from limited global exploration and susceptibility to local optima. To address these issues, we propose an improved algorithm, called Weibull Difference Adaptive t-Gaussian Dung Beetle Optimization (WDTDBO), which integrates Bernoulli chaotic mapping and Lens opposition-based learning for population initialization to increase population diversity, employs a Weibull dynamic flight Strategy to enhance global search, and applies Mean difference perturbation with Adaptive t-Gaussian perturbation to escape local optima. Experimental results show that on the CEC2017 benchmark, WDTDBO improves stability and convergence accuracy by 75.72% and 47.20% in 30 dimensions, and by 60.84% and 60.98% in 100 dimensions. On the 20-dimensional CEC2022 benchmark, improvements reach 77.65% and 27.70%. The average rankings of 1.17, 1.21, and 1.50 on 30D/100D CEC2017 and 20D CEC2022 respectively indicate superior accuracy, convergence efficiency, and robustness compared with DBO and other algorithms. Moreover, experiments on challenging engineering problems further confirm the practical advantages of WDTDBO over competing algorithms.