Moving Loads Identification on Bridge Structures with Quasi-Transmissibility Metric

This chapter presents a novel quasi-transmissibility metric designed for model-free identification of moving loads on bridge structures by strategically integrating bridge response measurements obtained from different moving vehicles. The proposed approach is situated within a unified Bayesian inference framework, wherein the quasi-transmissibility metric obviates the need for bridge-specific models, such as influence lines. At zero frequency, this metric reduces to the ratio of gross vehicle weights across two distinct scenarios, facilitating robust estimation of gross vehicle weights. Across broadband frequency ranges, the metric encapsulates the spectral relationship between axle load distributions under varying moving load conditions, enabling the concurrent identification of axle weights and spacings. Analytical derivations of the likelihood function incorporate uncertainties arising from measurement noise and modeling errors into a complex Gaussian ratio probabilistic formulation. Furthermore, closed-form expressions for the covariance matrix quantify posterior uncertainties associated with both load parameters and prediction errors. Experimental validations confirm the framework’s adaptability to diverse vehicle speeds, sensor configurations, and axle arrangements, achieving high accuracy without reliance on prior bridge models. This chapter thus introduces an innovative methodology for practical bridge weigh-in-motion systems.