Sparse-Sensor Bayesian Updating of Railway Bridge Digital Twins under Train Passage

Abstract

The continuous and safe operation of railway networks relies heavily on the structural integrity of railway bridges, which are subjected to dynamic loads from increasingly frequent and high-speed train passages. Structural Health Monitoring has conventionally depended on dense sensor arrays to capture the nuanced dynamic responses of these structures. However, installing, maintaining, and processing data from dense sensor networks pose significant economic and computational burdens. This paper proposes a novel framework for the Bayesian updating of railway bridge Digital Twins utilizing a sparse sensor configuration under operational train passage events. By integrating vehicle-bridge interaction dynamics with advanced Bayesian inference algorithms, the proposed methodology successfully updates the physical parameters of the bridge digital twin, capturing both structural degradation and environmental uncertainties using minimal instrumentation. The research explores optimal sensor placement strategies based on information entropy to maximize structural observability. A comprehensive numerical investigation of a representative high-speed railway bridge demonstrates that the sparse-sensor Bayesian framework achieves parameter estimation accuracies comparable to traditional dense networks while reducing computational overhead by an order of magnitude. The findings validate the feasibility of deploying resource-efficient digital twins for predictive maintenance in critical transportation infrastructure, paving the way for scalable and economically viable structural health management systems.