The precise forecasting of the compressive strength of Steel Fibre Reinforced Concrete (SFRC) under elevated temperatures is crucial for the secure design of structures exposed to fire and thermal stresses. This study establishes a robust machine learning (ML) framework to predict the compressive strength of SFRC utilising an experimental database of 307 data points. Six machine learning techniques, namely Random Forest (RF), Gradient Boosting Regressor (GBR), Bagging Regressor (BR), Support Vector Machine (SVM), K-Nearest Neighbours (KNN), and Artificial Neural Networks (ANN), were meticulously assessed and compared. The input parameters comprised fibre volume fraction, specimen geometry, heating rate, temperature, and cementitious composition. Of all models, RF exhibited the greatest predictive accuracy, attaining a R² of 0.96 and an RMSE of 6.27 on the test data. The outcomes were juxtaposed with recognised empirical formulations from the literature, validating the superiority of ensemble machine learning models in elucidating nonlinear interactions among SFRC properties. This study highlights the efficacy of data-driven methodologies as dependable substitutes for conventional models, enhancing the accuracy of predictions regarding SFRC behaviour under elevated temperature conditions. The predictive performance of six machine learning algorithms was compared with conventional empirical models, and the results demonstrated that ensemble-based methods such as Random Forest provide superior accuracy in estimating compressive strength loss of SFRC under elevated temperatures.
Khan et al. (Mon,) studied this question.