Axial Load Capacity Prediction of Concrete-Filled Steel Tubes Using Machine Learning: A Comparative Study

Concrete-filled steel tubes (CFSTs) combine the advantages of both steel and concrete and have been widely applied in recent years due to their excellent mechanical performance and economic efficiency. This paper focuses on the axial load capacity of CFST columns. First, 227 sets of experimental data were collected to develop axial load capacity prediction models using six machine learning algorithms. Second, the existing design code-recommended calculation models were evaluated. Subsequently, parameter importance and sensitivity analyses were conducted using the optimal machine learning model, and a graphical user interface was developed for predicting the axial load capacity of circular CFST columns. The results show that the eXtreme Gradient Boosting model is the most suitable for predicting the axial load capacity of CFST columns. In contrast, the stability and accuracy of the code-recommended models need improvement. The steel tube diameter and thickness have a significantly greater influence on axial load capacity compared to other parameters. Within specific ranges, increasing all parameters except column length enhances the axial load capacity.