Bài báo khoa học

The cold-formed Steel Semi-oval Hollow Section (SSOHS) column is a new cross-section column and has been used a lot in construction projects. However, the design standards for steel structures in the world have not covered the cross-section classifications for the SSOHS columns in the design process. Therefore, the Axial Load Capacity (ALC) of the SSOHS column has been different between the design standards and experiments. This paper develops predictive tools (formula and graphical user interface) for calculating the ALC of the SSOHS columns based on an Artificial Neural Network (ANN) model. The ANN model has been developed with 219 datasets. The input parameters of the ANN model include the overall depth (D), the overall width (B), thickness (t) of the sections, and the length of the pin-ended columns (L). Meanwhile, the ALC of the SSOHS column is the output parameter of the ANN model. The predictive formula based on an ANN model is compared with three regression models and two existing formulas. The comparison results reveal that the performance of the ANN model outperforms three regression models and two existing formulas through indicators: R-squared, RMSE, and a20-index. The sensitivity analyses of the input parameters to the ALC of the SSOHS column are also performed. Finally, a mathematical formula and graphical user interface program are developed to practically calculate the ALC of the SSOHS column.