Application of deep learning in civil engineering: boosting algorithms for predicting strength of concrete.

This article focuses on using machine learning (ML) boosting ensemble algorithms to predict the compressive strength of high-performance concrete (HPC). Five boosting methods—XGBoost, AdaBoost, Gradient Boosting Decision Tree (GBDT), LightGBM, and CatBoost—were compared, with CatBoost achieving the highest accuracy (R² = 0.975 after hyperparameter optimization). The study incorporated a broad set of input variables, including mix design parameters such as sand ratio, water-cement ratio (W/C), and water-binder ratio (W/B), to enhance model reliability. The SHapley Additive exPlanations (SHAP) method, based on game theory, was applied to interpret the model's predictions, identifying AGE (curing time), W/B, and W/C as the most influential factors on compressive strength. These findings provide theoretical and practical guidance for optimizing HPC mix designs while reducing the need for extensive experimental testing.