Scholarlink

<p>Metaheuristic optimization is a research area that allows automating the seismic design of structures while prioritizing the sustainable use of resources. Unfortunately, these AI techniques have limited applicability due to their long execution times.&#xa0;To address this limitation, this paper explores the use of artificial neural networks (ANNs) as surrogate models to accelerate the optimization process of seismic retrofitting of mid-rise frames equipped with buckling-restrained braces (BRBs). To identify the benefits of this strategy, the performance of simulated annealing and genetic algorithms, both with and without&#xa0;a surrogate model, is compared. From the results of this study, it is shown that: (a) ANNs can reduce the metaheuristic optimization time by up to 51%; (b) surrogate models can return infeasible designs if they are used to drive the entire optimization process, and (c) for the considered problem, the characteristics of the BRBs that provided the most efficient designs were identified.</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

Seismic optimization of buckling-restrained brace mid-rise frames by metaheuristics and machine learning surrogative models

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年3月1日

查看全文
  • Luis Velasco,
  • Hector Guerrero,
  • Antonio Hospitaler

摘要

Metaheuristic optimization is a research area that allows automating the seismic design of structures while prioritizing the sustainable use of resources. Unfortunately, these AI techniques have limited applicability due to their long execution times. To address this limitation, this paper explores the use of artificial neural networks (ANNs) as surrogate models to accelerate the optimization process of seismic retrofitting of mid-rise frames equipped with buckling-restrained braces (BRBs). To identify the benefits of this strategy, the performance of simulated annealing and genetic algorithms, both with and without a surrogate model, is compared. From the results of this study, it is shown that: (a) ANNs can reduce the metaheuristic optimization time by up to 51%; (b) surrogate models can return infeasible designs if they are used to drive the entire optimization process, and (c) for the considered problem, the characteristics of the BRBs that provided the most efficient designs were identified.