Rockfall impacts simple supported girder bridge piers: deep learning model

As the mountainous economy continues to grow rapidly, simply supported girder bridges are becoming more commonly used in highway projects. However, the challenging geological environments and frequent rockfall hazards present considerable risks to their structural safety. In this paper, we set against the backdrop of typical mountainous terrain, employs ROCKFALL software to simulate and analyze the trajectory of rockfalls. It integrates this with an LS-DYNA finite element model to systematically investigate the damage distribution characteristics and dynamic response patterns of bridge piers subjected to rock impacts. The study evaluates the dynamic response of the bridge pier, including impact force, displacement, and direct damage, to assess the potential hazards posed by rockfalls. It also examines how various impact parameters influence the damage distribution and displacement response of bridge piers. To enhance prediction accuracy, a deep learning model for rockfall impact response is utilized, and an intelligent prediction model that integrates Long Short-Term Memory (LSTM), Bidirectional Long and Short Term Memory (BiLSTM), and Adaptive Boosting (Adaboost) is proposed.