Stirrups fractured judgment in concrete columns confined with stirrups under axial compression by artificial neural networks

Abstract The stirrups may be fractured in stirrups confined concrete columns under axial compression. The fracture of stirrups had significant effects on the residual stress and deformability of concrete columns confined with stirrups. To explore the stirrups fracture in concrete confined with stirrups, two artificial neural networks, including a genetic algorithm (GA) optimization‐based support vector machine and a back propagation (BP) network based on GA optimization, were developed to judge whether stirrups fractured in concrete columns confined with stirrups. The main parameters investigated in this paper, including the compressive strength of concrete, the yield strength, diameter, spacing, and volumetric ratio of stirrups, as well as the yield strength and ratio of longitudinal reinforcement, were used as input parameters. The proposed models were trained and tested using 534 data sets from the test results conducted by the authors and previous studies. The proposed models had high reliability and predictability for judging whether stirrups fractured in concrete confined with stirrups. Furthermore, a BP network was developed to evaluate the residual stress of confined concrete, which can be utilized to evaluate the residual load‐bearing capacity of concrete columns after stirrups were fractured. Based on the proposed models, the strain corresponding to the residual stress of confined concrete should be introduced into the existing equation of the axial stress–strain curves of confined concrete as the ultimate strain of confined concrete. The ultimate strain of confined concrete should be the maximum value between the axial strain corresponding to the residual stress and the axial strain specified in the existing equations.