ABSTRACT
The VBISI (Vehicle-Bridge Interaction System Identification) method is a drive-by monitoring method that estimates the vehicle’s and bridge’s parameters and road unevenness simultaneously only from vehicle vibration and position data. The estimation process randomly assumes the mechanical parameters of vehicle and bridge, and solves two processes, namely the IEP (Input Estimation Problem) of the vehicle and the DRS (Dynamic Response Simulation) of the bridge to estimate the road unevenness. It is realized by updating the mechanical parameters by minimizing the residual of the estimated road unevenness. This paper introduces and compares the potential schemes: PSO (Particle Swarm Optimization), NM (Nelder-Mead) method and MCMC (Monte Carlo Markov Chain). The differences between these schemes are also examined numerically in this study. PSO is very efficient but needs large calculation resources, NM is very efficient and shows high accuracy, MCMC is very costly but gives the reliable.
