A Deep Learning-Based Approach for the Identification of a Multi-Parameter BWBN Model

A restoring-force model is a versatile mathematical that can describe the relationship between restoring force and deformation obtained from large number of experiments. Over past few decades, body work on development models has been reported in literature. Under high intensity cyclic loadings or seismic excitations, reinforced concrete (RC) structures undergo wide range hysteretic deteriorations such as strength, stiffness pinching degradations. These characteristic behaviors be described by multi-parameter Bouc-Wen-Baber-Noori (BWBN) model, which offers applicability. This applied for response prediction modeling behavior structural mechanical engineering systems, adjusting distribution this model’s parameters. However, major difficulty utilizing BWBN parameters’ identification. In paper, deep neural network used to estimate hysteresis parameters model. one most widely general RC columns. The experimental data columns paper are collected database Pacific Earthquake Engineering Research Center (PEER). Firstly, loop physical experiment identified via genetic optimization algorithm. Then established backpropagation (BP) algorithm associating with column. Finally, regression analysis carried out obtain characteristics trained directly identify based columns, effective computationally efficient proposed approach overcomes difficult problem identifying provides promising wider application