Dynamic Service Life Prediction for Large-Tonnage Bearing of Long-Span Railway Bridges under Evolving Traffic Scheduling

Large-tonnage bearing is a critical yet vulnerable component of a long-span railway bridge, which faces unexpected wear due to the evolving high-density traffic scheduling, and has a much shorter service life than the main structure. Thus, its life prediction is recognized as a vital issue for railway bridge maintenance and life extension. However, existing studies cannot address large-tonnage bearings with a double-sliding mechanism that separates translation and rotation. In addition, model-based methods suffer from high computational cost, while existing data-based methods are limited to monitoring scenarios, data quality, and cannot account for future traffic evolution. To address these limitations, this study presents an unsupervised and dynamic service life prediction method for large-tonnage bearings of long-span railway bridges, considering evolving traffic scheduling. This study first derives an analytical formula for the transmission among three types of sliding in large-tonnage bearings and main beam deformation. On this basis, a framework is developed to estimate the bearing behavior online, including spherical plate sliding (SPS) and plane plate sliding (PPS), across various monitoring scenarios. Then, the long-term train-induced sliding processes of the SPS and PPS are reconstructed by loading evolving railway traffic load on dynamically updated influence coefficient sets that can efficiently track the bearing behavior of a multilane long-span bridge. The bearing life is predicted by integrating the nonstationary wear process with time-dependent reliability analysis. Finally, the proposed method is successfully applied to a monitored in-service long-span railway bridge with specially designed large-tonnage bearings. The application results show that the method accurately reconstructs the long-term SPS and PPS under multispan layouts, even with intermittent data missing, while automatically predicting the life of multiple large-tonnage bearings under various evolving traffic conditions. The method provides reliable guidance for long-span railway bridge maintenance decision-making and life extension.