Predicting the Girder Rotation for Bridge Deck Construction Using Deep Neural Network

Bridge deck construction often uses overhang brackets to support the deck overhang formwork, the self-weight of the fresh concrete, and the loads from the heavy construction equipment. These loads can introduce significant torsional moments on the exterior girders and sometimes lead to the excessive rotation of the bridge girders, which may induce many safety and maintenance issues. To prevent these issues and limit the exterior girder rotation, detailed finite element analysis is usually recommended for the torsional analysis during the bridge deck overhang construction, which can be tedious and difficult in some cases. This study developed a deep neural network model to predict the maximum exterior girder rotation during deck construction. More than 10,000 finite element models were developed using SAP2000 to obtain sufficient data for the deep neural network model. The deep neural network model was validated using field bridge monitoring and finite element analysis data. The results indicate that the deep neural network is an effective and efficient method to predict exterior girder rotation during bridge deck construction.