Shear Strength Prediction in Reinforced Concrete Deep Beams Using Nature-Inspired Metaheuristic Support Vector Regression

Abstract

The shear strength of reinforced concrete (RC) deep beams is a dynamic phenomenon that varies with many mechanical and geometrical factors. Accurately estimating shear strength in RC deep beams is a vital issue in engineering design and management. However, prediction accuracy is still poor. This study presents a nature-inspired metaheuristic regression method for accurately predicting shear strength in RC deep beams that combines a novel smart artificial firefly colony algorithm (SFA) and least squares support vector regression (LS-SVR). The SFA integrates the firefly algorithm (FA), chaotic map (CM), adaptive inertia weight (AIW), and Lévy flight (LF). First, an adaptive approach and randomization methods (i.e., CM, AIW, and LF) were incorporated in FA to construct an effective metaheuristic algorithm for global optimization. The SFA was then used to optimize the hyperparameters of the LS-SVR model. The proposed model was constructed using a data set for RC deep beams which was derived from the literature. Model performance was evaluated by comparing results with those of a baseline SVR model and with previous methods via a cross validation algorithm. Analytical results show that the novel optimized prediction model is superior to others in predicting the shear strength of RC deep beams and that it can assist civil engineers in designing RC deep beam structures.