.png){kind=icon}
.png){kind=icon}
Return
Machine learning-based predictive models for maximum lateral strength CFST column under axial loading and lateral cyclic loading
.png){kind=icon}
PRE
AI
DOI：10.12989/cac.2026.37.3.507.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
In this study, a lateral strength predictive model of CFST columns under lateral cyclic loading was developed using machine learning (ML) algorithms. A total of 82 experimental datasets on the CFST columns under axial and lateral loading collected from the literature served as the training and testing data to build the predictive model. An extreme Gradient Boosted Tree (ExGBT) was utilized to interpret the trained ML models. To demonstrate the efficiency of such a model, the ExGBT performance was compared with other ML techniques, such as Multiple Linear Regression (MLR), Artificial Neural Networks (ANN), and Decision Trees (DT). The overall correlation coefficient was 0.951, 0.989, 0.994, and 0.995 for MLR, ANN, DT, and ExGBT, respectively, indicating that the lateral strength of the CFST column can be well-predicted by all considered methods. However, ExGBT was the optimum algorithm for lateral strength prediction of the CFST column with the lowest MAPE of 15.32%.
Keywords:
artificial neural network (ANN)
concrete-filled steel tube
cyclic loading
decision tree (DT)
extreme gradient boosted tree (ExGBT)
maximum lateral strength
multiple linear regression (MLR)