Prediction of the Shear Strength of Steel-Fiber-Reinforced Concrete Using a Swarm Intelligence Based Extreme-Learning Machine

Steel fibers are added to concrete to enhance the shear strength of reinforced concrete beams and improve the post-cracking tensile strength of the material. However, creating an accurate simulation model to estimate the shear strength of steel-fiber-reinforced concrete (SFRC) beams is challenging. Several factors affect the ultimate shear strength of fiber-reinforced concrete beams, that include the compressive strength of the concrete, the adequate depth, the ratio of shear span to depth, longitudinal reinforcement, the quantity of fibers, and the fiber length. This paper evaluates the effectiveness of using machine-learning models, including artificial neural networks (ANN), support vector machines (SVM), gradient tree boosting regressor (GTB), CatBoost regressor (CATB), extreme-learning machine (ELM), and extreme-learning machine optimized with particle swarm optimization (ELM-PSO). The models were built using experimental data gathered from the literature. To assess the performance of each model, several statistical metrics were employed, such as the Wilmott Index (WI), Nash–Sutcliffe Efficiency (NSE), Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Kling-Gupta Efficiency (KGE). Based on the results, the ELM-PSO model proved to be a viable method for predicting the shear strength of SFRC beams. During the test phase, the ELM-PSO model achieved a coefficient of determination (R²) of 0.95, making it the most accurate estimator of shear strength in SFRC beams when compared to other models.