Probabilistic Invariant Sets for Closed-Loop Re-Identification

ANDERSON, ALEJANDRO; FERRAMOSCA, ANTONIO; GONZÁLEZ, ALEJANDRO HERNÁN; KOFMAN, ERNESTO

Córdoba

Congreso; XVIº Reunión de Trabajo en Procesamiento de la Información y Control (RPIC 2015); 2015

Recently, a Model Predictive Control(MPC) suitable for closed-loop re-identificationwas proposed, which solves the potential conflict betweenthe persistent excitation of the system andthe stabilization of the closed-loop by extendingthe equilibrium-point-stability to the invariant-setstability.The proposed objective set, however, derivesin large regions that contain conservatively theexcited system evolution. In this work, based on theconcept of probabilistic invariant sets, the controllertarget sets are substantially reduced ensuring the invariancewith a sufficiently large probability (insteadof deterministically), giving the resulting MPC controllerthe necessary flexibility to be applied in a widerange of systems.