Moving load identification in the presence of uncertainties in the computational model

Most standard methods for moving load identification make use of a deterministic computational or analytical models under strong modelling assumptions concerning the geometry of the supporting structure on which the moving loads are applied and the speed of these moving loads. Nevertheless, these quantities are in general not perfectly known (thus uncertain) and the identified moving loads, as well as the dynamic response of the structure, can be very sensitive to their values. It is therefore crucial to examine the effects of these uncertainties on the predictability of computational models and support decision-makers. This paper introduces a stochastic framework for modelling a vehicle-bridge interaction (VBI) system, accounting for uncertainties in the road profile and the speed of the moving loads yielding uncertainties in the geometry of the supporting structure and the position of the moving loads. The explicit form of the Newmark- β method is utilised to establish an inverse model and identify the moving loads, and the Monte Carlo simulation method is used to build the statistics of the identified moving loads. • The speed variability of each moving load is modelled as a stochastic process. • The roughness of the road profile is modelled as a random field. • The forward problem is solved with an explicit formulation of Newmark-beta scheme. • Statistics on the identified moving loads are estimated using MC method.