A hybrid machine learning approach for predicting the flexural strength of concrete reinforced with waste aluminium fibres.

The reutilization of industrial waste in concrete offers an effective pathway toward sustainable construction. In this study, waste aluminium fibres are investigated as an alternative reinforcement material, and a novel hybrid stacking ensemble model is developed to predict the flexural strength of aluminium fibre reinforced concrete. An experimental program evaluating aluminium fibre contents ranging from 1% to 5% identified an optimal dosage of 3% for flexural performance. A dataset comprising 195 samples, collected from experimental results and published studies, was used to train and test machine learning models. Random Forest (RF), Extreme Gradient Boosting (XGBoost), and Artificial Neural Network (ANN) models were optimized and combined within a stacking ensemble framework. The proposed hybrid model achieved superior predictive accuracy, with coefficients of determination (R2) of 0.9913 for training and 0.9627 for testing, outperforming the individual models. Explainability analysis using SHAP revealed that specimen age and aggregate content exert the greatest influence on flexural strength. These findings demonstrate that hybrid ensemble learning, coupled with explainable artificial intelligence, provides a reliable and interpretable approach for predicting the flexural behaviour of sustainable concrete incorporating waste aluminium fibres.