Open AccessParameter estimation of Markov‐switching Hammerstein systems using the variational Bayesian approach
Author(s) -
Ma Junxia,
Huang Biao,
Ding Feng
Publication year - 2019
Publication title -
iet control theory and applications
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.059
H-Index - 108
eISSN - 1751-8652
pISSN - 1751-8644
DOI - 10.1049/iet-cta.2018.6236
Subject(s) - control theory (sociology) , autoregressive model , estimation theory , markov chain , mathematics , markov process , computer science , markov model , bayesian probability , hidden markov model , variable order markov model , mathematical optimization , algorithm , artificial intelligence , statistics , control (management)
Parameter estimation of Markov‐switching Hammerstein systems is presented in this study. The random switching mode is described by a hidden Markov model. An input non‐linear controlled autoregressive Hammerstein model is applied to describe the dynamic process of each local model. By applying the decomposition technique to the non‐linear mapping of the input signal, each local model is written as a linear‐in‐parameter form. The variational Bayesian approach is applied to identify the switched Markov Hammerstein models. Both the discrete‐valued switching mode state and the unknown parameters in each local model are estimated. Moreover, instead of the point estimation, posterior distributions of unknown parameters for each local model are obtained. Numerical simulation examples and a hybrid tank experiment are presented to verify the effectiveness of the proposed approach.