Discussion on: 'A Stacked Model Structure for Off-line Parameter Variation Estimation in Multi-equilibria Nonlinear Systems'

摘要

Ref. [1] presents a methodology for estimating constant parameters of a nonlinear system, with a specific focus on systems that operate around a finite set of operating points. A continuous stirred tank reactor (CSTR) with nonlinear dynamics is used to demonstrate the approach. The method uses a stacked Kalman Filter (KF) or stacked Extended Kalman Filter (EKF) along with data about multiple equili-briumpoints to determine a set of constant parameters. The authors show how a KF does not work well for data about single equilibrium point data because of observability problems; similar issues occur with the EKF. The stacked filter is then used to select the appropriate equilibrium points and to estimate the constant parameters using multiple datasets. This discussion note focuses on three important points related to parameter estimation in general, and the approaches in Ref. [1] specifically: (a) the KF for parameter estimation, (b) the EKF for parameter estimation, and (c) the Kalman smoother.