Scholarlink

<p>Traditional evolutionary algorithms (EAs) often require a large number of function evaluations (FEs) to solve expensive constraint optimization problems (ECOPs), leading to high computational costs. To address this challenge, this paper proposes a fast generalized surrogate−assisted evolution algorithm based on a multi-strategy hybrid sparrow search algorithm (FGSAEA-MSHSSA). The proposed algorithm introduces an innovative model management approach that integrates both global and local surrogate models. The proposed algorithm constructs the global surrogate model using an ensemble surrogate method and introduces a variance and distance (VD) criterion to select appropriate individuals for exact evaluation, effectively guiding the evolutionary search process. It also proposes the concept of a population trust region to build local surrogate models, enhancing the exploration capability of FGSAEA-MSHSSA. Additionally, the top feasible mean rule (TFMR) is incorporated to maintain a balance between the feasible and infeasible regions during the search, improving the quality of feasible solutions and significantly accelerating convergence. Experimental results demonstrate that the proposed algorithm outperforms other advanced algorithms on benchmark functions with varying characteristics and spring design problems. It not only significantly reduces computational costs but also greatly enhances optimization performance, with an average improvement of approximately 48.4% compared to other 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

A fast generalized surrogate−assisted evolutionary algorithm for solving expensive constrained optimization problems

  • 期刊论文

  • 发表日期：

    2025年7月24日

查看全文
  • Genggeng Liu,
  • Xuehui Su,
  • Ruping Zhou,
  • Nengxian Liu,
  • Liyuan Zhang,
  • Ye Tian

摘要

Traditional evolutionary algorithms (EAs) often require a large number of function evaluations (FEs) to solve expensive constraint optimization problems (ECOPs), leading to high computational costs. To address this challenge, this paper proposes a fast generalized surrogate−assisted evolution algorithm based on a multi-strategy hybrid sparrow search algorithm (FGSAEA-MSHSSA). The proposed algorithm introduces an innovative model management approach that integrates both global and local surrogate models. The proposed algorithm constructs the global surrogate model using an ensemble surrogate method and introduces a variance and distance (VD) criterion to select appropriate individuals for exact evaluation, effectively guiding the evolutionary search process. It also proposes the concept of a population trust region to build local surrogate models, enhancing the exploration capability of FGSAEA-MSHSSA. Additionally, the top feasible mean rule (TFMR) is incorporated to maintain a balance between the feasible and infeasible regions during the search, improving the quality of feasible solutions and significantly accelerating convergence. Experimental results demonstrate that the proposed algorithm outperforms other advanced algorithms on benchmark functions with varying characteristics and spring design problems. It not only significantly reduces computational costs but also greatly enhances optimization performance, with an average improvement of approximately 48.4% compared to other algorithms.