Innovative modeling framework of chloride resistance of recycled aggregate concrete using ensemble-machine-learning methods

This study investigates the feasibility of introducing machine learning algorithms to predict diffusion resistance chloride penetration recycled aggregate concrete (RAC). A total 226 samples collated from published literature were used train and test developed framework, which integrated four standalone models two ensemble models. The hyperparameters involved fine-tuned by grid search 10-fold cross-validation. Results showed that all had good performance in predicting RAC among them, gradient boosting model outperformed others. water content was identified as most critical factor affecting ion permeability based on standardized regression coefficient analysis. model’s interpretability greatly improved through a two-way partial dependence Finally, proposed models, performance-based mixture design method service life prediction approach for developed, thereby offering novel robust tools achieving more durable resilient development goals procuring sustainable concrete.