Sergey Nazin : February 1, 2007

Ellipsoidal and Interval Techniques for State Estimation in Linear Dynamical Systems under Model Uncertainty.

Abstract

The unknown-but-bounded description of system uncertainty is a main counterpart to its probabilistic characterization. This description forms the set-membership approach to the different kinds of problems in control and identification. My talk is devoted to the state estimation problem for linear discrete-time dynamical systems under unknown-but-bounded uncertainties. This problem is to characterize the set of all feasible system state vectors that are compatible with the model structure, measurement data and hypotheses on the perturbations. All points of this set are the candidate solutions of the estimation problem. Since the set obtained may become extremely complicated, its outer approximations by a simpler sets like ellipsoids or interval vectors (boxes) are then customary to use. Ellipsoidal and interval methods will be presented for state estimation in the case of model uncertainty, which is a natural extension of classical methods for deterministic models with unknown-but-bounded state perturbations and measurement noise.