Peak stress and peak strain evaluation of concrete columns confined with lateral ties under by Artificial Neural Networks

Abstract

Abstract The peak stress and peak strain of concrete columns confined with lateral stirrups were an important indicators for evaluating the load-bearing capacity and axial deformation of concrete columns under axial compression. However, it was hard to determine the peak stress and peak strain of concrete confined with lateral ties under axial compression due to complicated arching actions of transverse reinforcements and longitudinal reinforcements and the complex interaction between concrete and lateral ties. In this paper, two typical artificial neural networks (ANN) including (BP networks and Elamn networks) were applied to evaluate the peak stress and peak strain of concrete columns confined with lateral ties based on a reliable database consisting of 196 test data sets for peak stress and 166 test data sets for peak strain collected from previous studies. Both of the proposed ANN models had high prediction performance in the training and testing process. And By comparing with existing analytical models, the proposed BP networks had high reliability and high applicable in predicting the peak stress of confined concrete, while the Elman network had high reliability and high applicable in peak strain of concrete columns confined with lateral ties. Furthermore, based on the sensitive analysis, the concrete strength and the properties of lateral ties have obvious influence on the peak stress of confined concrete, while the volumetric ratio of lateral ties had significant effects on the peak strain of confined concrete.