Enhancing shear strength prediction in steel fiber reinforced concrete: a hybrid machine learning approach

It is complicated to comprehend the shear strength behavior of steel fiber-reinforced concrete (FRC) because of various parameters that are involved. The goal of this paper is to develop a framework for precise estimation of the shear strength of steel FRC beams using a novel hybrid machine learning technique. The adaptive neuro fuzzy inference system (ANFIS) was chosen as a predictive model to address this problem, and a new metaheuristic algorithm based on the equilibrium optimizer (EO) was used, for optimal generalization and accurate prediction. In addition, the performance of the newly created predictive model was evaluated by using other popular metaheuristic algorithms, such as genetic algorithm (GA) and particle swarm optimization (PSO). To create the EO-ANFIS model, 241 experimental data sets from the literature were utilized. The input variables consisted of the steel fiber ratio, the concrete design strength, the effective depth, the reinforcement ratio, and the span-to-depth ratio, while the output variable was the shear strength capacity. The accuracy of the ANFIS models based on metaheuristic algorithms was evaluated against four equations from other research and against modern ensemble learners using ten-fold cross-validation. The results indicated that the ANFIS models based on the proposed algorithms outperformed the previous equations, and the EO-ANFIS model had a better performance than the GA-ANFIS and PSO-ANFIS models while remaining competitive with the ensemble methods and offering an interpretable rule structure. The sensitivity analysis revealed that the most influential factors on shear strength are reinforcement ratio and steel fiber ratio.