Nonlinear deterministic observer for state estimation: Application to a continuous free radical polymerization reactor

Abstract

A nonlinear deterministic observer is developed for a class of multi-input multi-output systems. This observer is a Luenberger-like one that uses a constant gain which can be tuned by a high-gain technique to control the speed of convergence. The observer is used for estimating the state in a continuous free radical polymerization process. In open-loop simulations of this process the observer converges very well in much less than one residence time, provided the convergence parameter is well tuned. The observer shows good robustness with respect to measurement noise and to modelling errors. It must be pointed out that this observer works owing to the particular property of strong observability of the model, which in the general nonlinear case is not guaranteed.