A combined soft computing-mechanics approach to inversely predict damage in bridges

Abstract

This study aims to facilitate damage detection in concrete bridge girders without the need for visual inspection while minimizing field measurements. Beams with different material parameters and cracking patterns are modeled using mechanics-based ABAQUS finite element analysis software program in order to obtain stiffness values at specified nodes. The resulting database is then used to train an Artificial Neural Network (ANN) model to inversely predict the most probable cracking pattern. The aim is to use ANN approach to solve an inverse problem where a unique analytical solution is not attainable. Accordingly, simple span beams with 3, 5, 7 and 9 stiffness nodes and a single crack were modeled in this work. To confirm that the ANN approach can characterize the logic within the databases, networks with geometric material and cracking parameters as inputs and stiffness values as outputs were created. These networks provided excellent prediction accuracy measures (R2 values > 99%). For the inverse problem, the noted trend shows that better prediction accuracy measures are achieved when more stiffness nodes are utilized in the ANN modeling process. It was observed that decreasing the number of required outputs immensely improves the quality of predictions provided by the ANN.