Effective separation of vehicle, road and bridge information from drive-by acceleration data via the power spectral density resulting from crossings at various speeds

Abstract

This paper proposes a novel PSD-based algorithm to overcome limitations found in drive-by bridge health monitoring through the application of a transfer function to extract the contact point response from axle accelerations, the removal of the road irregularities by subtracting responses at different speeds in the spatial domain, and the estimation of damage by comparison of the PSD to a reference database. A strength of the algorithm lies in the use of a moving constant force to create the database, greatly simplifying computational and modelling requirements. Theoretical testing is conducted driving a quarter-car over a 15 m simply supported bridge in 1800 damage scenarios defined by a single crack randomly located across the span. The accuracy in locating and quantifying the crack remains unaffected overall when adding 5% Gaussian noise to the accelerations. The highly accurate quantification demonstrates the success in removing vehicular and road components from the axle accelerations.