Self-compacting concrete strength evaluation using fire hawk optimization-based simulations

Single and integrated forms of machine learning analysis were developed, named multi-layered perceptron neural network (MLP), support vector regression (SVR), radial basis function neural network (RBF), and Random forests (RF) were applied to determine the self-compacting concrete’s (SCC) compressive strength (CS). The fire hawk optimization algorithm (FHO) determined the optimal values of the main parameters in each model (abbreviated as FMLP, FSVR, FRBF, and FRF). A comprehensive dataset was created by creating experimental specimens, including fly ash, granite powders, silica fume, granulated blast furnace slag, superplasticizer, steel slag powder, and viscosity-modifying admixture, in addition to the standard concrete ingredients. The calculations and analysis of metrics demonstrated that the FSVR, FMLP, FRBF, and FRF algorithms could significantly achieve favored efficiency. For example, the FSVR models recognized as the best framework depicted almost 45% improvement considering error metrics compared to the FMLP. This study is important for its potential to greatly improve the efficiency and precision of \(SCC\) $\mathit{SCC}$ mix design. By offering a reliable predictive tool, it presents a way to cut both the cost and time required for concrete production. This, in turn, can promote more sustainable construction practices, more efficient material use, and improved quality control in producing high-performance concrete.