A high-performance hybrid paradigm of ELM and augmented GWO for estimating tensile strength retention of GFRP rebars in alkaline environments

The durability of fibre-reinforced polymer (FRP) rebars in seawater and sea sand concrete (SWSSC) is crucial for the design of marine and offshore structures. The retention of tensile strength in rebars within alkaline environments is a complex, nonlinear function dependent on several parameters. This study compares traditional and high-performance machine learning algorithms (MLAs) for estimating the tensile strength retention of glass FRP (GFRP) rebars aged in laboratory-simulated alkaline concrete conditions. Four widely used MLAs, namely artificial neural network, adaptive neuro-fuzzy inference system, extreme learning machine (ELM), and least squares support vector machine, were used. Multiple hybrid intelligence paradigms of these MLAs were constructed using swarm intelligence algorithms. Furthermore, an augmented grey wolf optimizer (AGWO), was utilized to develop high-performance soft-computing paradigms. Experimental results exhibit that the constructed ELM-AGWO model achieved 90.18% accuracy during the testing phase. Overall results indicate that the developed ELM-AGWO model outperforms all other models developed in this study. Moreover, the proposed ELM-AGWO model was employed to evaluate a simulated dataset representing a low alkaline concrete environment, demonstrating a significant reduction in the degradation of GFRP rebars with a decrease in the pH of the surrounding environment.