Fleet monitoring using a Bayesian approach to detect bridge bearing damage

This paper proposes the use of a Bayesian approach to detect bridge bearing damage using simulated acceleration measurements collected from a fleet of notional instrumented vehicles. On-bridge vehicle accelerations are used to update bridge and vehicle properties. Four different bridge damage states are considered with different bearing damage conditions. The results show that bridge properties, i.e., second moment of area and bearing stiffnesses, can be inferred accurately from the acceleration measurements using a Bayesian updating approach. The fleet monitoring concept with the Bayesian approach is computationally efficient when a large quantity of measurement is involved.