A Stochastic Model Updating Method With Unknown Parameter Distribution

摘要

For stochastic model updating (SMU) problem with unknown parameter probability density function (PDF), an approach is proposed to estimate the statistical properties of the structural parameters. The random variable is represented with polynomial chaos expansion (PCE) and therefore the SMU problem is converted to the evaluation of the PCE coefficients. The Bayesian method is applied to estimate the PCE representation, while the likelihood function in Bayesian calculation is constructed using kernel density estimation (KDE). In order to save computation time, Kriging approach is utilized to establish the surrogate model between structural input and output. A numerical example is given to verify the proposed method.