This study aims to facilitate damage evaluation and detection in concrete bridge girders without the need for visual inspection while minimizing field measurements. Beams with different material and cracking parameters (cracks’ depth, width and location) are modeled using ABAQUS finite element analysis software in order to obtain stiffness values at specified nodes. The resulting database is then used to train two Artificial Neural Networks (ANNs). The first network (ANN1) solves the forward problem of providing a health index parameter based on predicted stiffness values. The second network (ANN2) solves the inverse problem of predicting the most probable cracking pattern, where a unique analytical solution is not attainable. Accordingly, simple span beams with 9 stiffness nodes and two cracks were modeled in this work. For the forward problem, ANN1 had the geometric, material and cracking parameters as inputs and stiffness values as outputs. This network provided excellent prediction accuracy measures (R2 99%). For the inverse problem, ANN2 had the geometric and material parameters as well as stiffness values as inputs and cracking parameters as outputs. This network provided less accurate predictions (R2 = 65%) compared to ANN1, however, ANN2 still provided reasonable results, considering the non-uniqueness of this problem’s solution.

Related Publications