第５部門②

[2E06-10-04] Application of replica exchange MCMC method to Bayesian structure model update by dual sampling method

*Kiyoyuki Kaito², Kodai Matsuoka¹ (1. Railway Technical Research Institute, 2. Osaka University)

Keywords:Bayesian Estimation, Uncertainty, Dual Sampling

This study incorporates the replica exchange MCMC method which is one of the typical wide-area spatial estimation methods into the dual sampling method which is the tail estimation method for simultaneous posterior distribution in Bayesian structural model update. The proposed method is applied to estimate the tail of simultaneous posterior distribution from the data obtained on past study, and the obtained simultaneous posterior distribution is compared with the results of the existing method (GA) to confirm the estimation accuracy and computational efficiently.