Physics-Integrated Machine Learning Framework for Predicting Torsional Strength and Response Behavior of Reinforced Concrete Beams

Fuente: Zenodo
Saved in:
Bibliographic Details
Main Authors: MEHTA, VIKAS, Kim, Jae Min, Rhee, Inkyu
Format: Recurso digital
Published: Zenodo 2026
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866902243117105152
author MEHTA, VIKAS
Kim, Jae Min
Rhee, Inkyu
author_facet MEHTA, VIKAS
Kim, Jae Min
Rhee, Inkyu
contents <p>This repository contains the source code, datasets, and supporting materials associated with the manuscript titled <em>“Physics-Integrated Machine Learning Framework for Predicting Torsional Strength and Response Behavior of Reinforced Concrete Beams,”</em> submitted to the journal <em>Engineering Applications of Artificial Intelligence</em>.</p> <p>The archived materials support the development, validation, and reproducibility of a hybrid artificial intelligence framework for torsional strength prediction in reinforced concrete beams. The framework integrates physics-guided neural networks (PGNN), graph neural networks (GNN), and a Bayesian-optimized stacking ensemble (BO-Stack) to provide accurate, interpretable, and physically consistent predictions of torsional behavior.</p> <p>Contents of this repository include:<br>• Experimental torsional test dataset<br>• Physics-informed synthetic dataset generated using variational autoencoder (VAE)-based augmentation<br>• PGNN implementation<br>• GNN implementation<br>• BO-Stack ensemble framework<br>• SHAP-based interpretability scripts<br>• Torsional boundary/failure domain generation scripts<br>• Cross-validation and benchmarking scripts<br>• Graphical user interface (GUI) for engineering application<br>• Reproducibility documentation and execution instructions</p> <p>The repository is intended to ensure transparency, reproducibility, and public accessibility of the computational framework and supporting data associated with this study.</p>
format Recurso digital
id zenodo_https___doi_org_10_5281_zenodo_20322665
institution Zenodo
language
publishDate 2026
publisher Zenodo
record_format zenodo
spellingShingle Physics-Integrated Machine Learning Framework for Predicting Torsional Strength and Response Behavior of Reinforced Concrete Beams
MEHTA, VIKAS
Kim, Jae Min
Rhee, Inkyu
<p>This repository contains the source code, datasets, and supporting materials associated with the manuscript titled <em>“Physics-Integrated Machine Learning Framework for Predicting Torsional Strength and Response Behavior of Reinforced Concrete Beams,”</em> submitted to the journal <em>Engineering Applications of Artificial Intelligence</em>.</p> <p>The archived materials support the development, validation, and reproducibility of a hybrid artificial intelligence framework for torsional strength prediction in reinforced concrete beams. The framework integrates physics-guided neural networks (PGNN), graph neural networks (GNN), and a Bayesian-optimized stacking ensemble (BO-Stack) to provide accurate, interpretable, and physically consistent predictions of torsional behavior.</p> <p>Contents of this repository include:<br>• Experimental torsional test dataset<br>• Physics-informed synthetic dataset generated using variational autoencoder (VAE)-based augmentation<br>• PGNN implementation<br>• GNN implementation<br>• BO-Stack ensemble framework<br>• SHAP-based interpretability scripts<br>• Torsional boundary/failure domain generation scripts<br>• Cross-validation and benchmarking scripts<br>• Graphical user interface (GUI) for engineering application<br>• Reproducibility documentation and execution instructions</p> <p>The repository is intended to ensure transparency, reproducibility, and public accessibility of the computational framework and supporting data associated with this study.</p>
title Physics-Integrated Machine Learning Framework for Predicting Torsional Strength and Response Behavior of Reinforced Concrete Beams
url https://doi.org/10.5281/zenodo.20322665