Toward Developing A Machine Learning Model For The Prediction Of The Bearing Capacity Of The Square Concrete-Filled Steel Columns

Abstract

In this thesis project, based on studies analyzing the research, the thesis aims to develop a machine learning model to predict the bearing capacity of concrete-filled steel (CFST) columns by utilizing influential extreme gradient boosting (XGBoost) for training with a dataset containing 253 data points. In the dataset used for the machine learning model, there are a total of eleven that cover various geometric and material properties including the shape of the CFST (shape), the width of square steel pipe (B), depth of square steel pipe (H), where the diameter of round steel pipe is exchanged equivalent to the width and depth of square steel pipe, column length ( L), the wall thickness of steel pipe (t), concrete thickness (d), cylindrical compressive strength of concrete (f'c), compressive strength of concrete prism (fck), yield strength of steel (fy), and load eccentricity (e), the bearing capacity of the concrete-filled steel columns (Nu). Features are divided into nine input variables, one category instance variable, and one output variable. Machine learning model performance in estimating the bearing capacity of concrete-filled steel columns using MAE, RMSE, and R2 values. Furthermore, when compared to other models, the model utilizing XGBoost performs better than the performance derived from the stated formulae, highlighting the usefulness of forecasting bearing capacity values.