An improved deflection model for FRP RC beams using an artificial intelligence-based approach

Abstract

This study uses an AI technique called gene expression programming (GEP) to generate a deflection model for predicting the deflection of reinforced concrete (RC) beams using fibre reinforced polymer (FRP) bars as the main reinforcements through the effective moment of inertia. Taking into account the advantages of both theoretical and empirical models, the study trained GEP using a database created by calculating the effective moment of inertia (Ie) of 108 designed beams using 10 equations collected from the literature. The results with the affected parameters were input into GEP. The GEP then provided an expression for the prediction of Ie based on the training database. After that, the mid-span deflection (δ) of the beams was determined through the predicted Ie, the beam span length (L), the maximum moment in a member at the stage at which deflection is computed (Ma), and the elastic modulus of concrete (E). Experiments were conducted to verify the predicted results. Further analysis of the effect of tension stiffening was also conducted. The proposed model provided acceptable predictions.