Title :
Bayesian data rectification of nonlinear systems with chains in cell space
Author :
Ungarala, Sridhar ; Chen, Zhongzhou
Author_Institution :
Dept. of Chem. Eng., Cleveland State Univ., OH, USA
Abstract :
Data rectification extracts state variables from data corrupted with errors. The Bayesian formulation for nonlinear dynamic systems with non-Gaussian state pdfs does not possess an analytical solution. It is typically simplified by linearization and Gaussian assumptions. This paper presents a novel approach to solve the true nonlinear, non-Gaussian problem. The evolution of state pdf is modeled with finite state Markov chains in discretized state space using Monte Carlo simulations. The approach is compared with grid filters and particle filters.
Keywords :
Bayes methods; Markov processes; Monte Carlo methods; linearisation techniques; nonlinear systems; state estimation; Bayesian data rectification; Bayesian formula; Gaussian assumption; Markov chain; Monte Carlo simulation; cell space; grid based filter; linearization; nonGaussian state pdfs; nonlinear dynamic systems; particle filter; state space; state variable; Bayesian methods; Chemical engineering; Current measurement; Filters; Hidden Markov models; Least squares approximation; Monte Carlo methods; Nonlinear systems; Shape; State-space methods;
Conference_Titel :
American Control Conference, 2003. Proceedings of the 2003
Print_ISBN :
0-7803-7896-2
DOI :
10.1109/ACC.2003.1242492
