Identification of errors-in-variables ARX model with time varying time delay

Abstract

An identification method is proposed for errors-in-variables (EIV) ARX model with input time-varying time-delays. A Markov chain is used to model varying time delays whose parameters are also estimated. The EIV system accounts for noises in both input and output. To estimate noise-free input, a linear state space model is used to describe input generation process and a Kalman smoother is adopted for its estimation. An expectation maximization algorithm is used to estimate ARX model parameters. A spinning process of polyester fiber and a continuous stirred tank reactor process are used to verify the effectiveness of the proposed approach.