Novel models based on support vector regression to predict the compressive strength of concrete with recycled aggregate

Reusing recycled concrete aggregate (RCA) from destroyed structures has attracted the attention of academics due to its potential to mitigate environmental deterioration caused by the large amount of construction materials created globally in recent times. Recycled aggregates exhibit significant variations in composition and properties compared to natural aggregates, making it challenging to predict the performance of recycled aggregate concrete (RAC) and determine the appropriate mix proportions. This study aims to evaluate the effect of replacing natural fine and coarse aggregates with recycled materials on the 28-day compressive strength of concrete, utilizing machine learning (ML) techniques for assessment. Models based on Support Vector Regression (SVR) are created, trained using input parameters, and evaluated. To optimize the training process in the developed model, 2 meta-heuristic algorithms were employed, including one inspired by the Ebola virus disease (EOS) propagation mechanism and the other inspired by Jellyfish behavior in nature (JSO). The results confirm the high accuracy of the proposed predictive models, with the SVEO (SVR model optimized with EOS) in this study being the most precise with R² = 0.992 and RMSE = 1.345, while the accuracy of SVJS (SVR model optimized with JSO) is noticeably lower with R² = 0.984 and RMSE = 2.433.