A hybrid neuro-swarm model for shear strength of steel fiber reinforced concrete deep beams

ABSTRACT: The prediction of the shear strength capacity of Steel Fiber-Reinforced Concrete Deep Beams has been a challenge to civil engineering applications. Despite being widely used for its exceptional strength and durability, it remains to be byzantine due to the impacts of its beam dimensions, material strength, and reinforcements. In this research, a hybrid Neuro-Swarm model that combines Artificial Neural Networks (ANN) and Particle Swarm Organization (PSO) techniques was proposed to address this challenge. The model was trained on 96 datasets, and its performance was evaluated on 20 datasets. The study found that the proposed model had a high R-value of 0.99704 for the training dataset and 0.97737 for the testing dataset, indicating superior performance. An ANOVA test showed that there was no significant difference between the predicted and experimental shear strength values. Furthermore, the researchers found that the beam’s effective depth and yield strength of reinforcement bars were the most significant factors affecting the shear strength capacity of SFRC deep beams. The proposed model was also compared with other recent shear strength prediction models, and the Neuro-Swarm model was observed to be the best and the most robust model among them. It had the highest R2 value of 0.955243, the smallest RMSE of 53.33075, and the lowest percentage difference of 6.02607% between the predicted and actual values. The findings of this study demonstrate the potential of the hybrid Neuro-Swarm model for predicting the shear strength of SFRC deep beams providing useful insights for designers and engineers in the construction industry.

Filipiniana