Shear strength prediction of steel fiber reinforced concrete beam using hybrid intelligence models: A new approach

Abstract

Despite modern advancements in structural engineering, the behavior and design of reinforced concrete beams in shear are still a major concern for structural engineers. In this research, a new Support Vector Regression algorithm coupled with Particle Swarm Optimization (SVR-PSO) is developed to predict the shear strength (Ss) of steel fiber-reinforced concrete beams (SFRC) using several input combinations denoting the dimensional and material properties. The experimental test data are collected from reliable literature sources. The main variables used to construct the predictive model are related to the dimensional and material properties of the beams. SVR-PSO, the objective predictive model, is validated against a classical neural network model tuned with the same metaheuristic optimizer algorithm. The findings of the modeling study provide a clear evidence of the superior capability of the SVR-PSO used to predict the SFRC shear strength relative to the benchmark model. In addition, the construction of the predictive models with a lesser number of input data attributes are attained, leading an acceptable prediction accuracy of the SVR-PSO compared to the ANN-PSO model. In summary, the proposed SVR-PSO methodology has demonstrates an effective engineering strategy that can be applied in problems of structural and construction engineering prospective, applied to predict shear strength of steel fiber reinforced concrete beam using advanced hybrid artificial intelligence models developed in this study.