Enhancing Concrete's Structural Integrity: A Novel Approach for Predicting the Mechanical Properties of Glass Wool Fiber Reinforced Concrete

ABSTRACT This research presents a novel hybrid technique to forecast the mechanical properties of glass wool fiber reinforced concrete (GWFRC) using progressive graph convolutional networks (PGCN) and sand cat swarm optimization (SCSO), termed as PGCN‐SCSO. The primary goal is to optimize the composition of GWFRC by accurately forecasting its compressive, flexural, and tensile strengths. The PGCN model learns complex relationships between input features, while the SCSO algorithm optimizes the hyperparameters of the PGCN for enhanced prediction accuracy. Experimental data from M20 and M30 grade concrete mixtures, incorporating glass wool fibers (0.5%–3%) and 30% ground granulated blast furnace slag (GGBS) replacement, were used to validate the proposed approach. The model's performance was assessed using the mean absolute error (MAE), coefficient of determination ( R 2 ), and root mean square error (RMSE), showing better outcomes than established methods such as artificial neural network (ANN), genetic algorithm‐extreme gradient boosting (GA‐XGBoost), and light gradient boosting machine (LightGBM). The PGCN‐SCSO approach achieved R 2 values of 0.94, 0.99, and 0.98 for flexural, compressive and tensile strength predictions, respectively, indicating its effectiveness in accurately predicting GWFRC properties and optimizing concrete formulations.