Scholarlink

Purpose <p>This study presents a comprehensive investigation into the optimization of NZ-bearing parameters for base-isolated buildings using metaheuristic algorithms (MHAs). Eight MHAs: artificial bee colony (ABC), cuckoo search (CS), differential evolution (DE), human evolutionary optimization algorithm (HEOA), teaching–learning-based optimization (TLBO), jaya algorithm (JA), gravitational search algorithm (GSA), and big bang-big crunch (BBBC), collectively representing all major categories of MHAs, namely swarm intelligence-based, evolution-based, human behavior-based, and physics-based algorithms, are employed, and their comparative performance is assessed.</p> Methodology <p>A regression-based multi-objective optimization framework employing a weighted linear combination approach, in conjunction with MHAs, is developed to determine optimal NZ-bearing design parameters. The framework simultaneously minimizes multiple structural responses, including isolator displacement, roof displacement, roof acceleration, and base shear of a six-story base-isolated building subjected to multiple earthquake excitations. The performance of the MHA classes is rigorously compared in terms of consistency, convergence behavior, computational efficiency, and robustness.</p> Results and Conclusions <p>All eight tested MHAs successfully identified optimal design parameters with negligible variation. However, DE, along with swarm intelligence-based and human behavior-based MHAs, consistently delivered superior performance in terms of stability, reliability (SD and CV values in the order of 10<sup>–17</sup> to 10<sup>–14</sup> across 100 independent runs), success rate (100%), and convergence speed. DE emerges as the most reliable and efficient algorithm overall. In contrast, physics-based MHAs and HEOA showed higher variability and lower reliability, although HEOA exhibited faster execution time (average 2.17&#xa0;s). Furthermore, the proposed weighted mean approach offers a viable option for identifying optimal parameters without earthquake-specific tuning.</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

Performance Assessment of Multi-Class Metaheuristic Algorithms for Optimal Design of NZ-Bearings in Base-Isolated Buildings

  • 期刊论文

  • 发表日期：

    2026年5月26日

查看全文
  • Sukamal Kanta Ghosh,
  • Diptesh Das

摘要

Purpose

This study presents a comprehensive investigation into the optimization of NZ-bearing parameters for base-isolated buildings using metaheuristic algorithms (MHAs). Eight MHAs: artificial bee colony (ABC), cuckoo search (CS), differential evolution (DE), human evolutionary optimization algorithm (HEOA), teaching–learning-based optimization (TLBO), jaya algorithm (JA), gravitational search algorithm (GSA), and big bang-big crunch (BBBC), collectively representing all major categories of MHAs, namely swarm intelligence-based, evolution-based, human behavior-based, and physics-based algorithms, are employed, and their comparative performance is assessed.

Methodology

A regression-based multi-objective optimization framework employing a weighted linear combination approach, in conjunction with MHAs, is developed to determine optimal NZ-bearing design parameters. The framework simultaneously minimizes multiple structural responses, including isolator displacement, roof displacement, roof acceleration, and base shear of a six-story base-isolated building subjected to multiple earthquake excitations. The performance of the MHA classes is rigorously compared in terms of consistency, convergence behavior, computational efficiency, and robustness.

Results and Conclusions

All eight tested MHAs successfully identified optimal design parameters with negligible variation. However, DE, along with swarm intelligence-based and human behavior-based MHAs, consistently delivered superior performance in terms of stability, reliability (SD and CV values in the order of 10–17 to 10–14 across 100 independent runs), success rate (100%), and convergence speed. DE emerges as the most reliable and efficient algorithm overall. In contrast, physics-based MHAs and HEOA showed higher variability and lower reliability, although HEOA exhibited faster execution time (average 2.17 s). Furthermore, the proposed weighted mean approach offers a viable option for identifying optimal parameters without earthquake-specific tuning.