A machine learning and game theory-based approach for predicting creep behavior of recycled aggregate concrete

Abstract

In this paper, the SHapley Additive exPlanation (SHAP) is utilized in conjunction with the ensemble machine learning (EML) model to study the creep behaviors of recycled aggregate concrete (RAC) for the first time. Five typical EML models, such as Random Forest (RF), Adaptive Boost Machine (AdaBoost), Gradient Boosting Decision Tree (GBDT), Extreme Gradient Boosting Machine (XGBoost), and Light Gradient Boosting Machine (LGBM) are considered. The proposed method can sort the contributions of input features for creep behaviors and interpret the prediction results of the best EML model. The findings show that the existing empirical models fib Model Code 2010 and JTG 3362-2018 cannot satisfy the requirements for RAC creep behavior prediction because the impact of RAC aggregate ratio and other factors are ignored. Moreover, the water-cement ratio and loading age are the two most significant factors. Therefore, this study has the potential to provide insight into the performance of RAC structures and help engineers adjust RAC mechanical behaviors.