Bayesian posterior distribution uncertainty of model parameters considering varying input seismic characteristics
A combination of global sensitivity analysis(GSA)and uncertainty quantification(UQ)is often used for updating uncertain structural parameters.GSA can extract significant features of model outputs before UQ is applied.The validity of resulting posterior distributions from UQ strongly depends on how sensitive each parameter is on the outputs.However,sensitivity degree significantly varies under different seismic loadings,and therefore has unneglectable impacts on the validity of the distributions.This study aims to construct an optimal updating procedure of the uncertain model parameters using monitoring data under various excitation levels.The test-bed structure is a standard two-degree-of-freedom(DOF)isolated bridge used in Japan design specifications for highway bridges.Firstly,a numerical model of bridge piers and isolators with the nonlinear response characteristics were constructed using a Takeda model and a bilinear model respectively.Monitoring data was then created by adding white noise to the model outputs under an assumed deterioration condition of the structure.Significant feature extraction was then implemented by GSA using Kriging metamodels.The procedure chooses optimal updating parameters corresponding to different seismic loadings for UQ implementation afterward.Finally,a sequential posterior distribution updating procedure of the structural parameters was constructed using available monitoring data under various seismic excitations.
Uncertainty quantification global sensitivity analysis seismic response isolated bridge pier
T.Tran M.Nishio
Department of Civil Engineering,Yokohama National University
国际会议
The 7th World Conference on Structural Control and Monitoring(7WCSCM)(第七届结构控制与监测世界大会)
青岛
英文
1913-1922
2018-07-22（万方平台首次上网日期，不代表论文的发表时间）