Set-membership state estimators based on cooperative output injections

Abstract

To estimate validated bounds on the actual state vector of uncertain non-linear systems, cooperative output injections methods are proposed in this contribution. The aim of the output injections is to design set-membership state estimators that ensure the order-preserving property between the lower, actual and upper state trajectories. Based on a special sensors placement, continuous-time and event-triggered output injections are proposed to cope with the conservatism of the classical bounding system methods. Furthermore, based on some properties of monotone dynamical systems, the convergence of the proposed set-membership state estimators is shown. It is worth pointing out that the proposed set-membership state estimation method allows one: on one hand, to avoid the conservatism related to the use of similarity transformations usually required in the framework of interval observer design approaches, and on the other hand, to circumvent the pessimism accumulation related to the wrapping effect of set-valued iterative numerical schemes.