Determination of moment, shear and ductility capacities of spiral columns using an artificial neural network

Abstract

Intelligent systems are frequently used to solve difficult and complex problems in today's world. Especially in the discipline of civil engineering, much research has been conducted with the help of intelligent systems. The objective of this study is to identify the moment and shear force capacities of reinforced concrete spiral columns and their displacement ductility values by using an algorithm based on an Artificial Neural Network (ANN). In the study, 86 different spiral column experiments tested in the literature were compiled, and the moment and shear force capacities obtained through the experiments were arranged at a certain level. In addition, the ductility values of the tested columns were calculated using graphical experimental data. The output layer consists of four neural network models, which are ANN1 (with three output neurons), ANN2, ANN3, and ANN4 (with one output neuron), and moment, shear force capacity, and displacement ductility parameters were obtained, respectively. According to results of this study, it was observed that moment and shear force capacities of spiral columns could be particularly well estimated using ANN. However, it was found that the estimation success of ANN for displacement ductility was not sufficient.