Estimating unbounded unknown inputs in nonlinear systems. (June 2019)

• Record Type: Journal Article
• Authors: Chakrabarty, Ankush; Corless, Martin
• Is Part Of: Automatica. Volume 104(2019)
• Page Start: 57
• Page End: 66
• Publication Date: 2019-06
• Subjects: Unknown input observers -- Convex programming -- Nonlinear disturbance model -- Internal model -- Incremental quadratic constraints -- Fault detection and isolation; Automatic control -- Periodicals; Automation -- Periodicals; 629.805
• DOI: 10.1016/j.automatica.2019.02.050
• Languages: English
• ISSNs: 0005-1098

Abstract

Abstract: Unknown inputs such as attacks on safety-critical systems is a major concern in the design of modern engineering systems. If the underlying system to be protected exhibits nonlinear dynamics, few observers/estimators in the current literature can handle unknown input estimation, particularly when the unknown inputs are unbounded. This paper proposes an extended state observer methodology for designing observers capable of reconstructing the state along with unknown inputs for a class of nonlinear systems that can be characterized via incremental quadratic constraints. Two classes of unknown inputs are considered: completely unknown inputs that have bounded derivatives, or (the generalization of the first) unknown inputs that are generated by nonlinear internal state-space models, where the model states are driven by completely unknown inputs; for instance, completely unknown inputs with bounded higher-order derivatives. The potential of the proposed design is demonstrated via numerical simulations.