Application of artificial neural network to predict dynamic displacements from measured strains for a highway bridge under traffic loads

Abstract

This paper addresses the experimental verification of vertical displacement predictions for a highway bridge under dynamic vehicle loads. In the structural health monitoring of bridge structures, the measurements of vertical displacements are relatively difficult than the measurements of axial strains, and the vertical displacements can give an intuitive information for monitoring the structural conditions. Therefore, an artificial neural network (ANN) was introduced for the accurate predictions of the vertical displacements from the axial strains. In the experiments, both the strains and displacements at the different locations were measured during 48 h for obtaining the training (Set 1) and testing (Set 2) data, which were utilized to develop and validate the ANN-based prediction model. The environmental effects such as temperature loads were eliminated from the experimental data using a calibration approach, and the pure contributions of the external vehicle loadings to the 3D highway bridge were considered. The validation results showed that the ANN-based model can accurately predict the vertical displacements. It is expected that the proposed ANN-based model, as a fast and accurate framework, can be utilized for various civil infrastructures during the structural health monitoring.