Artificial Neural Network Prediction for Seismic Response of Bridge Structure

Abstract

Based on identification and prediction ability of neural networks for nonlinear systems, an improved BP network was adopted to predict the seismic responses of the bridge structures. First, a multi-player BP networks based on Levenberg-Marquardt algorithm was formed. Then, the improved neural network was trained by the imitated seismic responses of the first 4 seconds which were obtained from artificial earthquake waves by finite element method. Thirdly, the seismic responses of 1st to 8th seconds for the same bridge structure were predicted use the neural network which has been trained, and the predict responses were compared with the calculation data. The error curves between the prediction and the calculation results show that the BP network combined with is Levenberg-Marquardt algorithm has very good convergence rate, and the artificial neural network can predict the dynamic response of bridge structures well enough.