Load-carrying capacity of locally corroded steel plate girder ends using artificial neural network

In aged steel bridges, an area of local damage may be created in girders nearby the bearing region due to corrosion. The existence of local corrosion damage in the plate girder end can reduce the load-carrying capacity of bridge. A three-layer Back-Propagation neural network (BPNN) has been developed to predict the residual buckling strength of such damaged members. In this paper, train, test and validation sets of the neural network were obtained by using the finite element software ABAQUS. The accuracy of the nonlinear finite element method (FEM) to evaluate the residual bearing capacities of damaged beams is discussed. Buckling and post-buckling behavior of plate girders ends were quantitatively evaluated from nonlinear finite element analyses (FEA) model varying the corrosion scenario. A parametric study is achieved based on FE and an empirical equation is proposed based on BPNN to estimate the residual bearing capacity of deteriorated steel plate girder by local corrosion damage. The obtained results show that the prediction of the residual bearing capacity of the locally corroded steel plate girder ends is accurate and effective. (C) 2015 Elsevier Ltd. All rights reserved.