The purpose of this research is to create and test a machine learning-based model to predict the structural response of moment- and shear-loaded CFRP-confined concrete elements. The numerical simulation of the dataset was carried out in SAP2000, with the number of stories, seismic scaling factor and analysis type as the varying input parameters (the concrete compressive strength being carried as an additional feature to represent CFRP confinement in the validation stage) and moment and shear capacities as the target structural responses. The numerical results were then used to train and test the predictive model, and its performance was evaluated using the coefficient of determination (R²), root mean square error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE). The results showed good predictive capability for moment capacity, with R² of 0.846, RMSE of 7.8, MAE of 6.7, and MAPE of 9.4%, indicating good generalisation to unseen cases and identification of the most dominant patterns for flexural behaviour. The novelty of this study is the combination of machine learning and finite element-based structural simulation, enabling a more cost-efficient predictive approach that reduces the need for repeated numerical analysis for preliminary assessment, structural design optimisation, and data-driven decision-making in practical CFRP strengthening applications.
Copyrights © 2026