Virtual reality research on vibration characteristics of long-span bridges with considering vehicle and wind loads based on neural networks and finite element method

A finite element of the bridge was established to realize the virtual reality. Vehicle and wind loads were applied on it, and modal computation was carried out. Results show that serious vibration mainly took place on the bridge deck. Stresses and strains of the bridge were further extracted, where the maximum stress appeared around the positions with load application, while the maximum strain appeared in the middle of the bridge deck. Vibration response of the bridge under different excitations including vehicle loads and wind loads was computed by using the finite element model. Finally, neural network was also used to compute the vibration characteristic of the bridge, and the computational result was compared with that of the finite element method. The comparison result showed that they were consistent with each other, and the prediction model of neural networks was reliable. Using neural networks can improve the computational efficiency.