Bending failure modes and moment capacity prediction of steel and FRP bars hybrid reinforced UHPC beams based on machine learning

This study aims to develop and identify an optimal machine learning model for predicting the bending performance of UHPC beams reinforced with steel bars, FRP bars, and their hybrid reinforcement. Five different machine learning models were employed for training and prediction. Unlike previous studies focusing primarily on conventional reinforced UHPC beams, this study establishes a dedicated database for steel-FRP hybrid reinforced UHPC beams and performs a comparative evaluation of multiple machine learning algorithms for both classification and regression tasks. For training and evaluating these predictive model a dataset of 120 UHPC beams with steel bars, FRP bars, and hybrid reinforcement was collected, containing their failure modes and bending moment capacity. The models were evaluated using standard metrics for classification and regression tasks. The results indicate that the XGBoost model achieves the best overall performance in both failure mode classification and moment capacity prediction, demonstrating superior accuracy and generalization ability. Furthermore, based on the best-performing model in this study, the input feature values were analyzed using Shapley Additive Explanations (SHAP) and Spearman's rank correlation coefficient. A graphical user interface (GUI) was designed to concurrently estimate failure modes and moment capacities for UHPC beams reinforced with steel, FRP, or hybrid systems. The results show that XGBoost achieves the best overall performance, demonstrating superior accuracy and generalization ability in both failure mode classification and moment capacity prediction. SHAP analysis indicates that section width, equivalent reinforcement ratio, effective depth, and FRP strength are the dominant factors influencing bending capacity. The developed predictive model and GUI provide a valuable reference for engineering researchers to more accurately predict the failure modes and load-carrying capacity of UHPC beams reinforced with steel bars, FRP bars, and hybrid reinforcement.