AI-driven seismic optimization of outrigger systems in high-rise buildings: A machine learning framework for enhanced performance in earthquake-prone regions

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F61989100%3A27730%2F25%3A10258521" target="_blank" >RIV/61989100:27730/25:10258521 - isvavai.cz</a>

Výsledek na webu

<a href="https://www.sciencedirect.com/science/article/pii/S2352710225021011" target="_blank" >https://www.sciencedirect.com/science/article/pii/S2352710225021011</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1016/j.jobe.2025.113864" target="_blank" >10.1016/j.jobe.2025.113864</a>

Jazyk výsledku

angličtina

Název v původním jazyce

AI-driven seismic optimization of outrigger systems in high-rise buildings: A machine learning framework for enhanced performance in earthquake-prone regions

Popis výsledku v původním jazyce

This study introduces a novel machine learning-based framework to optimize outrigger systems for enhancing the seismic performance of tall buildings. Unlike conventional static or heuristic design methods, the proposed model integrates Artificial Neural Networks (ANN) and Support Vector Machines (SVM) for seismic response prediction, and applies Genetic Algorithms (GA) within a closed-loop optimization cycle. The framework uniquely combines supervised learning, reinforcement-based refinement, and structural simulations (pushover and time-history analysis) to automatically identify the optimal outrigger configuration position, stiffness, damping, and material properties under varying seismic loads. Empirical data and synthetic simulations are used to train the models, resulting in significant reductions in lateral displacement (46.67 %), inter-story drift (55.56 %), and improved energy dissipation (33.33 %). Benchmarking against standalone ANN and SVM models demonstrates superior prediction accuracy (RMSE = 0.089) and design efficiency. The proposed ANN-SVM-GA model is code-adaptable, resource-efficient, and scalable across seismic zones, achieving an 18 % RMSE reduction over ensemble baselines marking a significant advancement in intelligent structural design for earthquake-prone regions.

Název v anglickém jazyce

AI-driven seismic optimization of outrigger systems in high-rise buildings: A machine learning framework for enhanced performance in earthquake-prone regions

Popis výsledku anglicky

This study introduces a novel machine learning-based framework to optimize outrigger systems for enhancing the seismic performance of tall buildings. Unlike conventional static or heuristic design methods, the proposed model integrates Artificial Neural Networks (ANN) and Support Vector Machines (SVM) for seismic response prediction, and applies Genetic Algorithms (GA) within a closed-loop optimization cycle. The framework uniquely combines supervised learning, reinforcement-based refinement, and structural simulations (pushover and time-history analysis) to automatically identify the optimal outrigger configuration position, stiffness, damping, and material properties under varying seismic loads. Empirical data and synthetic simulations are used to train the models, resulting in significant reductions in lateral displacement (46.67 %), inter-story drift (55.56 %), and improved energy dissipation (33.33 %). Benchmarking against standalone ANN and SVM models demonstrates superior prediction accuracy (RMSE = 0.089) and design efficiency. The proposed ANN-SVM-GA model is code-adaptable, resource-efficient, and scalable across seismic zones, achieving an 18 % RMSE reduction over ensemble baselines marking a significant advancement in intelligent structural design for earthquake-prone regions.

Druh

J_imp - Článek v periodiku v databázi Web of Science

CEP obor

—

OECD FORD obor

20100 - Civil engineering

Projekt

<a href="/cs/project/TN02000025" target="_blank" >TN02000025: Národní centrum pro energetiku II</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2025

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název periodika

Journal of Building Engineering

ISSN

2352-7102

e-ISSN

2352-7102

Svazek periodika

112

Číslo periodika v rámci svazku

Volume 112

Stát vydavatele periodika

NL - Nizozemsko

Počet stran výsledku

32

Strana od-do

1-32

Kód UT WoS článku

001569209600008

EID výsledku v databázi Scopus

—