State estimation for nonlinear systems under model unobservable uncertainties: application to continuous reactor

The issue of this paper is the classical problem of state estimation, considering partially unknown, nonlinear systems with noise measurements. Estimation of system's state variables with observable and unobservable unstructured uncertain terms is performed simultaneously. An alternative system representation is proposed to transform the measured disturbance onto system disturbance, which lead a more adequate observer structure. The proposed methodology contains an uncertainty estimator based on the predictive contribution to infer the unobservable uncertainties and corrective contribution to estimate the observable uncertainties; which provides robustness against noisy measurements and model uncertainties. Convergence analysis of the proposed estimation methodology is realized, analyzing the dynamic equation of the estimation error, where asymptotic convergence is shown. This estimation scheme is applied to a continuous stirred chemical reactor. Numerical experiments illustrate the good performance of the proposed observer.