Modeling of uncertainty in bridge-vehicle system and the interaction force identification.

摘要

Since the randomness in the bridge structure has not been introduced in the bridge-vehicle interaction problem and the perturbation method adopted in the existing research works tends to loss accuracy with the increasing in variation of uncertainty. The research work in this Thesis aims to provide theoretical studies on the stochastic analysis of the bridge-vehicle interaction problem as well as on the identification of vehicle axle loads from samples of bridge response with uncertainties in both system parameters and road surface roughness.;The bridge is modeled as a simply supported planar Euler-Bernoulli beam with a vehicle either modeled as multiple forces or a mass-spring system moving on top. The finite element method is adopted to build the bridge-vehicle interaction model in which the system parameters as well as the road surface roughness are assumed as random processes. Firstly, only the randomness in road surface roughness is included and to be assumed as Gaussian random process represented by the Karhunen-Loeve Expansion. Based on the formulated model, both the dynamic analysis and the moving force identification are conducted. Secondly, the uncertainty in the material properties of the bridge structure which is assumed to be small and have Gaussian property is further included. Based on the model, a general stochastic moving force identification algorithm is proposed to identify the statistics of the vehicle axle loads from samples of bridge response. Finally, larger variation of uncertainty in the system parameters are investigated with the Spectral Stochastic Finite Element Method adopted in system modeling. The system parameters are assumed as Gaussian random processes and are further extended to non-Gaussian case. Dynamic analysis on the bridge-vehicle interaction problem with large variation of uncertainty in both system parameters and excitation forces is conducted. All the methods proposed are verified with numerical examples in which the Monte Carlo Simulation is adopted to obtain the reference solutions. Results show that the proposed methods on the dynamic analysis of the bridge-vehicle interaction problem and on the identification of statistics of moving vehicle axle loads with uncertainties are effective.