Employing the hybrid RBF prediction algorithm to predict the failure energy properties of concrete

Recently, fracture energy has received considerable attention owing to its importance in crack propagation in concrete structures. However, measuring fracture energy of concrete require experimental tests, causes cost and time. The alternative ways could be effective such as machine learning methods. So, this study concentrated on the developing prediction models to estimate the preliminary (G(f)) and total (G(F)) fracture energy of concrete. This research integrates the radial foundation function nervous network (RBF) with the equilibrium optimizer (EO) and the salp swarm optimization algorithm (SSOA), abbreviated EORB and SSRB, respectively, to get a more in-depth understanding of G(f) and G(F). The 264 research recordings were used to build and examine theories from previous research. Optimization procedures were used to determine the best propagation value and concealed substrate neuron count. Estimates demonstrate that both the improved EORB and SSRB were able to operate very well throughout the whole estimating procedure of G(f), with coefficient of determination (R-2) values of 0.9578 and 0.9907 and 0.8851, and 0.9617 for the training and assessment data subsets, respectively. Regarding G(F), R-2 values of 0.9119 and 0.8763, and 0.8586 and 0.8097 for the training and assessment data subsets were obtained. The comparison with the literature depicted a significant improvement in the efficacy of the EORB simulation, with R-2 going up from 0.8281 to 0.9119 related to G(f) and from 0.9025 to 0.9907 related to G(F). Given the logic and ease of model processing, the EORB analysis seems to be highly reliable for computing G(f) and G(F), even though the SSRB technique has distinctive features for simulating.