Influence of earthquake-induced hydrodynamic pressure on train-bridge interactions based on back-propagation neural network

Forecasting the behavior of the train-bridge system under strong waves is crucial in designing the cross-sea bridge. However, the influence of earthquake-induced hydrodynamic pressure on the dynamic system may be significant and need to be adequately understood. This study investigates the influence of earthquake-induced hydrodynamic force on the stochastic responses of the train-bridge interaction using the Newmark-13 method. A surrogate model named back-propagation neural network is implemented by correlating wave samples with the stochastic responses of the train-bridge system. Such a model improves computation efficiency and avoids further time-step integration. The Pintan's bridge, located in China's Eastern Region, is selected as a case study. The results show that the earthquake-induced hydrodynamic force involves significant responses of the train-bridge system. Moreover, the maximum dynamic amplification factor of the deck and the pier are 6% and 12.5%, respectively. Finally, a significant value of the peak period minimizes the effect of the earthquake-induced hydrodynamic responses on the train-bridge system.

Automated summary

Forecasting the behavior of the train-bridge system under strong waves is crucial for designing cross-sea bridges. This study investigates the influence of earthquake-induced hydrodynamic force on the stochastic responses of the train-bridge interaction using a surrogate model called back-propagation neural network. The results show that the earthquake-induced hydrodynamic force significantly affects the train-bridge system, and the choice of peak period minimizes this effect.