Prediction of compressive strength of concrete for high-performance concrete using two combined models, SVR-AVOA and SVR-SSA

Abstract

High-performance concrete (HPC) is a crucial material for constructing critical structures, such as dams, bridges, and high-rise buildings, as its exceptional compressive strength (CS) is vital for ensuring structural integrity. To improve this strength, additives, such as fly ash (FA) and micro-silica (MS), can be added to the mixture, often by reducing the water-to-cement ratio, alongside other factors. However, accurate modeling is imperative to estimate the CS of HPC. In this paper, support vector regression (SVR) is a newly developed regression model demonstrating superior performance in HPC compressive strength. Furthermore, two novel optimization algorithms, the African Vulture Optimization Algorithm (AVOA), and Salp Swarm Algorithm (SSA), are utilized to improve the performance of the SVR model. Each SVR-AVOA and SVR-SSA hybrid model was evaluated by 168 experimental samples, of which 70% of the sample belonged to training and 30$ to testing sections. Results indicate that the coupled model, SVR-AVOA, with suitable values, containing ${R}^{2}=0.973$, $RMSE=2.79$, $MSE=7.82$, $NRMSE=0.0443$, $NMSE=7.5$, $MAPE=3.13$, and $WAPE=0.0318$, is the most suitable for accurately estimating the CS of HPC. These findings highlight the efficacy of this hybrid model in HPC modeling and offer the potential for further research in the field.