Unknown input observer design for LPV OSL-QIB nonlinear systems

Abstract

This paper presents the design of an unknown input observer (UIO) for linear parameter-varying (LPV) one-sided Lipschitz quadratically inner-bounded (OSL-QIB) systems. This family of systems extends conventional LPV systems by adding nonlinear terms to improve observer performance and reduce errors due to the transformation of a nonlinear system to its LPV version (either because the nonlinearities are difficult to wrap as it is in the polytopic case, or because the Jacobian does not yield a good approximation). The UIO is initially designed in the absence of noise, allowing to establish LMI conditions to guarantee the convergence of the error dynamics, and then sensor noise is explicitly considered in the design so that its effiect can be mitigated. Finally, a case study based on the Corning channel benchmark is used to test the performance of the UIO in the presence of unknown inputs (e.g., leakage, evaporation or rain).

This work has been co-financed by the Spanish State Research Agency (AEI) and the European Regional Development Fund (ERFD) through the projects SaCoAV (ref. PID2020-114244RB-I00) and L-BEST (PID2020-115905RB-C21).

Peer Reviewed

Postprint (published version)