BAYESIAN UPDATING BY VARIATIONAL AUTOENCODER FOR SEISMIC RESPONSE ANALYSIS MODEL

Abstract

This paper proposes a nonparametric Bayesian updating method for nonlinear seismic response analysis models considering uncertainties using Variational Auto-Encoder (VAE). The proposed method is used for calculating the likelihood of the parameters for the response analysis model using VAE without any assumptions about the probability distribution. The proposed method is confirmed to successfully update the probability distribution of the parameters of a bilinear single-degree-of-freedom system. To discuss the characteristics of the proposed method, the relationship between the uncertainty due to the lack of information on the nonlinear response and the width of the posterior distribution of the parameters is investigated.