Simulation of the ultimate conditions of fibre-reinforced polymer confined concrete using hybrid intelligence models

Abstract Fibre Reinforced Polymer (FRP) composites can provide efficient enhancements in terms of strength and deformability for concrete structures, in which accurate predictions of FRP confined concrete ultimate conditions is highly essential to maintain safety levels, further structural analysis and members design. In this paper, three novel hybrid intelligence models were proposed based on the hybridization of Support Vector Regression (SVR) model with three bio-inspired optimization algorithms as genetic algorithm (GA), particle swarm optimization (PSO) and Whale optimization algorithm (WOA) for predicting the ultimate conditions of FRP-confined concrete. Moreover, 15 existing empirical relations for the prediction of the ultimate strength and strain of FRP-confined concrete have been comprehensively reviewed. The performances of the empirical models and the proposed hybrid models as SVR-GA, SVR-PSO and SVR-WOA are evaluated and compared based on a large database, including 780 circular FRP-confined concrete specimens, which are collected from the open-source published experiments. By comparing the predicted results based on several statistical indicators, the proposed hybrid SVR-models are generally outperforming the existing empirical relations in terms of accuracy and agreement with the experimental database. SVR-WOA provides superior performances than SVR-PSO, SVR-GA and all existed empirical models. The root mean square error is improved using SVR-WOA by 0.9%, 14.9 % and 37% for the ultimate strain capacity, and 2.7%, 4.6% and 17.3% for the ultimate strength compared to SVR-PSO, SVR-GA and the best empirical relation, respectively.