Rapid and reliable prediction of the dynamic response of long-span bridges is a critical approach for assessing the safety, serviceability, and resilience of key infrastructure. However, existing refined dynamic interaction models struggle to efficiently handle complex time-varying load scenarios due to their high computational costs. To address this, this study proposes a fast prediction framework for the dynamic response of long-span bridges based on neural operators and an equivalent dynamic wheel load method. By integrating physics-driven principles with deep learning, the proposed method significantly improves the computational efficiency of dynamic interaction models, and its effectiveness is validated through numerical experiments on a long-span bridge prototype under random loading conditions.
Liu et al. (Wed,) studied this question.