Estimating high-performance concrete compressive strength with support vector regression in hybrid method

Save time and energy by accurately predicting the mechanical parameters of concrete by employing artificial intelligence (AI) methods. Existing nonlinear relationships between concrete members have introduced uncertainties in estimating compressive strength (CS), among the most essential variables in concrete design. The use of conventional methods for individual use of AI models in predicting dependent components has been adopted in many investigations. The objective of this paper was to create estimation models for the CS of high-performance concrete (HPC) using hybrid approaches. Specifically, the study combined support vector regression (SVR) models with optimization algorithms such as Multi-objective Artificial Vultures Optimization (MAVO), African vulture’s optimization algorithm (AVOA), and Tunicate Swarm Algorithm (TSA). Linking the three predictive models and tuning their internal settings via optimization algorithms can result in different sorts of models. After evaluating the results of the proposed models utilizing multiple metrics, the performance of the SVMiA is computed to be greater than that of the other hybrid models, with R2 = 0.9672 and RMSE = 2.1884, respectively. In general, employing advanced kinds of individual models, such as hybrids, showed improved performance with a lower cost modeling process.