A STUDY ON ASSESSING THE DAMAGE OF A BEARING AND A PIER WITH NEURAL NETWORK USING THE ACCELERATION RESPONSES OF A BRIDGE

Abstract

 After an earthquake, quickly identifying damage on bridges is crucial to early recoveries. However, current method of structural health monitoring on bridges rely on cameras, cable networks, and manpower. These methods require high costs and also lead to slower recovery effort. In order to address these issues, a health monitoring model of a bridge bearing and a pier using neural networks and accelerometers were proposed. Such a model ensures real time evaluation of bearing and pier damage while using only the acceleration responses of a bridge. In order to evaluate the bearing damage, a neural network was proposed with acceleration responses of girder and pier top being the input while the output being bearing displacement. In order to evaluate the pier damage, first a neural network was proposed with acceleration responses of footing, girder and pier top being the input while the output being footing displacement. Then a neural network was proposed with displacement responses of footing, pier top and acceleration responses of girder being the input while the output being pier curvature. Several different combinations of seismic motions were considered as the learning data set. The results demonstrated that the neural network estimated the bearing displacement and the pier curvature with high precision.