论文部分内容阅读
A comprehensive Bayesian model updating approach is presented in this paper.The uncertainty of measurement,nonlinear distortions from the linearization of the model and modeling errors due to the limited predictability of the original model are considered.Tow strategies are implemented to make the Bayesian updating method feasible with general numerical models,the delaying rejection and adaptive metropolis(DRAM)samplers Markov Chain Monte Carlo mathematical algorithm(MCMC)is modified firstly to increase acceptance ratio in MCMC,and then support vector machine(SVM)is introduced as a surrogate mapping between the probability spaces of the prior random variables and the model modal parameters.The advantage of this proposed approach is computational efficiency and easy implementation for structural finite element model(FEM).Validation of the approach is demonstrated using a large-span continuous rigid frame bridge and a four-story steel lab-scale frame.Results show that parameters of FEM can be successfully updated.