An Investigation Into the Axial Capacity of Hot‐Rolled I‐Section Steel Columns Using Machine Learning

ABSTRACT Hot‐rolled steel (HRS) is a fundamental material in structural engineering, notable for its mechanical properties and diverse applications. This study focused on I‐sections (ISMB) of varying dimensions, conducting both experimental and numerical investigations. A nonlinear finite element model (FEM) was developed using ABAQUS to validate the experimental results regarding axial capacity and failure modes. The FEM results of the parametric study were subsequently compared against the IS 800:2007 and BS 5950‐1:2000 code specifications. Additionally, this study investigated the application of machine learning methods to predict the axial capacity of HRS sections. Specifically, soft computing models such as Artificial Neural Networks (ANN), Gradient Tree Boosting (GTB), and Multivariate Adaptive Regression Splines (MARS) were developed to predict the axial capacity of HRS sections based on the findings of finite element analysis. Comparison of the predicted results with experimental observations demonstrated the reliability and robustness of these machine learning models in approximating the axial capacity of hot‐rolled steel columns. This suggests that these soft computing models can be effective tools for predicting the ultimate strength or axial capacity of ISMB sections.