Premium
State estimation of a solid‐state polymerization reactor for PET based on improved SR‐UKF
Author(s) -
Liu Ji,
Gu Xing Sheng
Publication year - 2010
Publication title -
asia‐pacific journal of chemical engineering
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.348
H-Index - 35
eISSN - 1932-2143
pISSN - 1932-2135
DOI - 10.1002/apj.306
Subject(s) - kalman filter , control theory (sociology) , extended kalman filter , collocation (remote sensing) , estimator , nonlinear system , discretization , convergence (economics) , orthogonal collocation , noise (video) , filter (signal processing) , computer science , algorithm , state (computer science) , mathematics , collocation method , ordinary differential equation , differential equation , physics , control (management) , statistics , artificial intelligence , mathematical analysis , quantum mechanics , machine learning , economics , image (mathematics) , computer vision , economic growth
A state estimator for the continuous solid‐state polymerization (SSP) reactor of polyethylene terephthalate (PET) is designed in this study. Because of its invalidity in the application to some of the practical examples such as SSP processes, the square‐root unscented Kalman filter (SR‐UKF) algorithm is improved for the state estimation of arbitrary nonlinear systems with linear measurements. Discussions are given on how to avoid the filter invalidation and accumulating additional error. Orthogonal collocation method has been used to spatially discretize the reactor model described by nonlinear partial differential equations. The reactant concentrations on chosen collocation points are reconstructed from the outlet measurements corrupted with a large noise. Furthermore, the error performance of the developed ISR‐UKF is investigated under the influence of various initial parameters, inaccurate measurement noise parameters and model mismatch. Simulation results show that this technique can produce fast convergence and good approximations for the state estimation of SSP reactor. Copyright © 2009 Curtin University of Technology and John Wiley & Sons, Ltd.