Artificial Neural Network Models for Predicting Required Cross-section Dimensions of Concrete Filled Steel Tubular Columns

Background Artificial Neural Networks (ANN) can be a useful tool to assist in the design of reinforced concrete structures. The aim of this paper is to develop artificial neural network models for predicting the required diameter of eccentrically compressed concrete-filled steel tubular (CFST) columns and the wall thickness of the steel pipe. Methods Within the framework of the set goal, three models of ANN were developed. The first model predicts the required cross-sectional diameter for the minimum pipe wall thickness. The second model solves the same problem for the maximum possible wall thickness. The input parameters of the first two models are the axial force, bending moment, strength characteristics of steel and concrete, column length, and a coefficient characterizing the share of constant and long-term loads in the total load. The third model predicts the required wall thickness based on the listed parameters, as well as the column diameter. Synthetic data, including more than 2 million samples, was generated to train the models. The ANN architecture is a feedforward neural network with 2 hidden layers containing 16 neurons each. Machine learning models are implemented in the MATLAB environment. Results The trained models showed high performance in terms of mean squared error and correlation coefficient between the target and predicted values ​​of the output parameter. The importance of features was also assessed using the variable fixation method. It has been established that the required value of the column diameter is most significantly influenced by the magnitude of the bending moment, axial force and column length. The required pipe wall thickness is most influenced by the magnitude of internal forces and the diameter of the column. Conclusion The developed models of artificial neural networks are an effective and reliable tool that can help a civil engineer in the design of buildings and structures that include CFST elements.