Bayesian System Identification of MDOF Nonlinear Systems using Highly Informative Training Data

摘要

The aim of this paper is to utilise the concept of ‘highly informative trainuding data’ such that, using Markovudchain Monte Carlo (MCMC) methods, one can apply Bayesian system idudentification to multi-degree-of-udfreedom nonlinear systems with relatively little computational cost.udSpecifically, the Shannon entropy isudused as a measure of information content such that, by analysing tudhe information content of the posteriorudparameter distribution, one is able to select and utilise a relatively smaudll but highly informative set of train-uding data (thus reducing the cost of running MCMC).