Development of Bayesian Estimation Method for Fault Cause Based on 1DCAE Model

Abstract

A fault cause estimation method applying Markov Chain Monte Carlo (MCMC) methods was developed. The method can mechanistically estimate fault causes based on a 1DCAE model. Artificial malfunction data generated from a 1DCAE model with white noise is applied to demonstrate the fault cause estimation method. 1DCAE model simulates depressurization process of water tank with vapor supply. A surrogate model by machine learning using the results of 1DCAE model is employed in the fault cause estimate method. As results, the fault causes are verified correctly.