Full and Semi-Probabilistic Analysis of Ultra-High-Performance Concrete Beams in Bending via Machine Learning

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21110%2F25%3A00385805" target="_blank" >RIV/68407700:21110/25:00385805 - isvavai.cz</a>

Nalezeny alternativní kódy

RIV/68407700:21610/25:00385805

Výsledek na webu

<a href="https://doi.org/10.1080/10168664.2025.2500458" target="_blank" >https://doi.org/10.1080/10168664.2025.2500458</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1080/10168664.2025.2500458" target="_blank" >10.1080/10168664.2025.2500458</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Full and Semi-Probabilistic Analysis of Ultra-High-Performance Concrete Beams in Bending via Machine Learning

Popis výsledku v původním jazyce

Ultra-High-Performance Concrete (UHPC) offers exceptional mechanical properties, yielding it an attractive choice for advanced structural engineering applications. However, predicting the bending capacity of UHPC beams may remain a challenge due to its complex physical behavior and inherent variability. This study integrates a machine learning (ML) model based on Extreme Gradient Boosting (XGBoost) algorithm, with probabilistic analysis to enhance predictive accuracy and reliability. A database of 187 UHPC beams is collected and used to develop the XGBoost model, achieving superior performance metrics compared to traditional analytical methods like the Federal Highway Administration (FHWA) and the Swiss Standard (SIA 2052). Model uncertainty is evaluated, and a global resistance factor ($gamma _R, = , 1.35$gamma R=1.35) is derived from semi-probabilistic design. Fully probabilistic reliability analyses using Monte Carlo simulations demonstrate the efficacy of XGBoost in achieving target reliability indices for diverse beam configurations. Reliability indices for UHPC beams designed with XGBoost range from 3.0 to 4.3 depending on the steel reinforcement ratio, with lightly reinforced beams averaging the optimal value of 3.8, reflecting a balanced cost-safety design. In contrast, heavily steel-reinforced beams consistently exceed this optimal value, indicating potential overdesign under the semi-probabilistic approach. Additionally, SHAP (Shapley Additive Explanations) analyses provide insights into the influence of key variables when predicting with XGBoost, emphasizing the importance of reinforcement ratios and effective depths. The proposed framework facilitates robust and efficient design of UHPC beams with reduced model uncertainties and corresponding global resistance factors preferred for structural engineering practices.

Název v anglickém jazyce

Full and Semi-Probabilistic Analysis of Ultra-High-Performance Concrete Beams in Bending via Machine Learning

Popis výsledku anglicky

Ultra-High-Performance Concrete (UHPC) offers exceptional mechanical properties, yielding it an attractive choice for advanced structural engineering applications. However, predicting the bending capacity of UHPC beams may remain a challenge due to its complex physical behavior and inherent variability. This study integrates a machine learning (ML) model based on Extreme Gradient Boosting (XGBoost) algorithm, with probabilistic analysis to enhance predictive accuracy and reliability. A database of 187 UHPC beams is collected and used to develop the XGBoost model, achieving superior performance metrics compared to traditional analytical methods like the Federal Highway Administration (FHWA) and the Swiss Standard (SIA 2052). Model uncertainty is evaluated, and a global resistance factor ($gamma _R, = , 1.35$gamma R=1.35) is derived from semi-probabilistic design. Fully probabilistic reliability analyses using Monte Carlo simulations demonstrate the efficacy of XGBoost in achieving target reliability indices for diverse beam configurations. Reliability indices for UHPC beams designed with XGBoost range from 3.0 to 4.3 depending on the steel reinforcement ratio, with lightly reinforced beams averaging the optimal value of 3.8, reflecting a balanced cost-safety design. In contrast, heavily steel-reinforced beams consistently exceed this optimal value, indicating potential overdesign under the semi-probabilistic approach. Additionally, SHAP (Shapley Additive Explanations) analyses provide insights into the influence of key variables when predicting with XGBoost, emphasizing the importance of reinforcement ratios and effective depths. The proposed framework facilitates robust and efficient design of UHPC beams with reduced model uncertainties and corresponding global resistance factors preferred for structural engineering practices.

Druh

J_imp - Článek v periodiku v databázi Web of Science

CEP obor

—

OECD FORD obor

20102 - Construction engineering, Municipal and structural engineering

Projekt

<a href="/cs/project/GA24-10892S" target="_blank" >GA24-10892S: Strojové učení pro víceúrovňové modelování prostorové variability a trhlin pro zajištění udržitelnosti betonových konstrukcí</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

Structural Engineering International

ISSN

1016-8664

e-ISSN

1683-0350

Svazek periodika

35

Číslo periodika v rámci svazku

4

Stát vydavatele periodika

CH - Švýcarská konfederace

Počet stran výsledku

15

Strana od-do

634-648

Kód UT WoS článku

001510634800001

EID výsledku v databázi Scopus

2-s2.0-105008354012