A Bayesian approach to the detection of bridge scour damage using ‘drive-by’ measurements

‘Drive-by’ health monitoring uses sensors installed in passing vehicles to detect damage in bridges. Previous studies often assume vehicle properties since they are not readily available. Some of these properties are difficult to measure, making ‘drive-by’ monitoring more challenging. This paper proposes to use Sequential Monte Carlo (SMC) sampling method to probabilistically estimate vehicle properties via Bayes’ theorem. Updated vehicle parameters are then adopted to detect scour deflection of a bridge support. First, measurements including body car acceleration, rotational velocity and axle acceleration are generated by adding 5% random noise to simulated signals. A half-car model simulates axle acceleration using an Inverse Newmark-beta algorithm with a set of vehicle properties randomly sampled from prior distributions based on one’s belief. The Bayesian updating aims to minimise the residual between Power Spectral Density (PSD) of axle acceleration measurements and model outputs, resulting in the best fit vehicle properties. Having found vehicle properties, ‘apparent profile’, a combination of the true surface profile and elements of bridge deflection, can be inferred. Any step change of the apparent profile at the support indicates the deflection of bridge support due to scour. Updating the vehicle parameters using SMC following the Bayes’ rule shows high accuracy considering mean values of the posterior distributions, with less than 0.5% error for all properties. Using the updated vehicle properties, the deflection of the support caused by scour is detected at 1.6mm with 20% error compared to a ‘true’ value, 2mm.

TRAN Thanh T.X.;  O'BRIEN Eugene;  BONO Flavio;  OZER Ekin; 

2025-01-08

Civil Engineering Research Association of Ireland

JRC138956

978-1-3999-9618-1 (online),   

https://researchrepository.universityofgalway.ie/entities/publication/a20c47af-9b4f-4b30-b534-9d639565d920,    https://publications.jrc.ec.europa.eu/repository/handle/JRC138956,   

10.13025/29113 (online),