Estimation of axial load-carrying capacity of concrete-filled steel tubes using surrogate models

The main objective of the present work is to estimate the load-carrying capacity of concrete-filled steel tubes (CFST) under axial compression using hybrid artificial intelligence (AI) algorithms. In particular, the adaptive neuro-fuzzy inference system (ANFIS) with various metaheuristic optimization methods, such as the biogeography-based optimization (ANFIS-BBO), the particle swarm optimization (ANFIS-PSO), and the genetic algorithm (ANFIS-GA), have been employed taking into account the variability of input parameters. Commonly used statistical criteria, such as the coefficient of determination ( R 2 ), the a20-index, and the root mean squared error (RMSE), were utilized to evaluate and compare the effectiveness of the proposed AI models. The Monte Carlo approach was used to propagate the variability in the input space to the predicted output. The results showed that the ANFIS system, optimized by PSO, was the most effective and robust model with respect to three considered criteria (a20-index = 0.881, R 2 = 0.942 and RMSE = 185.631). Sensitivity analysis was performed, indicating that the minor axis length and thickness of the steel tube exhibited the highest contribution to the axial compression load-carrying capacity of the CFST.