Receding horizon output feedback control for linear systems with input saturation

摘要

A constrained receding horizon output feedback control method that is based on a state observer is suggested. The proposed method adopts the receding horizon dual-mode paradigm, which consists of an 'admissible invariant set' and 'free control moves'. Polyhedral admissible invariant sets of estimated state are derived, along with guaranteed bounds on state estimation errors. The guaranteed bounds on the state estimation errors are developed by considering invariant sets of state estimation errors that include possible initial estimation errors. Predictions of future states are made based on estimated current state and bounds on current estimation error. The free control moves are determined so that the predicted future state belongs to the polyhedral admissible invariant set, despite input constraints and measurement noise. This methodology can yield a stabilisable set of initial states that is infinite (in certain directions), even in the case of open-loop, unstable systems.