Scholarlink

<p>Minimax optimization is crucial for robust design, seeking optimal decisions under worst-case scenarios. However, existing algorithms face limitations in handling diverse problem types and suffer from computational inefficiency when dealing with expensive black-box functions. To address these challenges, this paper proposes a novel framework that integrates two key innovations: (1) an asymmetrical evolutionary strategy enabling reliable optimization for both symmetric and asymmetric minimax problems and (2) a Bayesian optimization component tailored for expensive simulation-based objectives without closed-form expressions. The proposed method advances robustness and computational efficiency compared to conventional approaches. Comprehensive experiments on eight benchmarks and five state-of-the-art methods demonstrate superior performance. Practical validation in real-world engineering applications further underscores its effectiveness for robust product design under uncertainty.</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

Asymmetrical evolutionary strategy with adaptive Bayesian optimization for robust design of complex electromechanical product

  • 期刊论文

  • 发表日期：

    2026年1月13日

查看全文
  • Hao Chen,
  • Fabin Mei,
  • Xuerong Ye,
  • Guofu Zhai

摘要

Minimax optimization is crucial for robust design, seeking optimal decisions under worst-case scenarios. However, existing algorithms face limitations in handling diverse problem types and suffer from computational inefficiency when dealing with expensive black-box functions. To address these challenges, this paper proposes a novel framework that integrates two key innovations: (1) an asymmetrical evolutionary strategy enabling reliable optimization for both symmetric and asymmetric minimax problems and (2) a Bayesian optimization component tailored for expensive simulation-based objectives without closed-form expressions. The proposed method advances robustness and computational efficiency compared to conventional approaches. Comprehensive experiments on eight benchmarks and five state-of-the-art methods demonstrate superior performance. Practical validation in real-world engineering applications further underscores its effectiveness for robust product design under uncertainty.