企画セッション（数値解析の検証と妥当性確認（V&V)、不確かさ評価）①

[12013-20-03] Uncertainty quantification for dependent parameters using hierarchical Bayesian model updating and model class selection (Proceedings of Symposium on Applied Mechanics)

*Masaru Kitahara¹, Takeshi Kitahara², Michael Beer³ (1. The University of Tokyo, 2. Kanto Gakuin University, 3. Leibniz University Hannover)

Keywords:Stochastic model updating, Bayesian model updating, Bayesian model class selection, Copula, Staircase density function

In stochastic model updating, a probabilistic model is assumed for the model parameters and its distribution parameters are calibrated. We have recently developed a hierarchical Bayesian updating framework, where the staircase density function (SDF) is used to arbitrarily approximate a wide range of probability distributions. This study aims to extend the approach to the calibration of dependent parameters. The dependent structure is represented by copulas and the marginal distributions are modeled using the SDFs. Hence, the copula parameter and the SDF parameters are calibrated through a Bayesian fashion. Furthermore, the most appropriate copula is determined through Bayesian model class selection.