Sensitivity of Piecewise control laws with respect to perturbation of the state-space partition - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2017

Sensitivity of Piecewise control laws with respect to perturbation of the state-space partition

Rajesh Koduri
Sorin Olaru

Résumé

The control design techniques for linear or hybrid systems (e.g. anti-wind up, MPC) lead often to off-line state-space partitions with non-overlapping convex polyhedral regions. This corresponds to a piecewise state feedback control laws associated to polyhedral partition of the state-space. In this work, we consider the perturbation in the representation of the vertices of the polyhedral regions. For such control laws, this problem is of particular interest with respect to the reduced precision representation of the state-space partition. The perturbed state-space partitions might lose one of the important property of the explicit controllers: the " non-overlapping " characterization. We derive a set called vertex sensitivity region to determine the admissible perturbation independently for each vertex of the polyhedral partition. A perturbation is deemed admissible if the non-overlapping property of the polyhedral regions is preserved. In the present work, the analysis of the sensitivity of each vertex is done under the assumption that the rest of the partition remains on the nominal configuration.

Domaines

Automatique
Fichier principal
Vignette du fichier
main_new.pdf (1.28 Mo) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-01648113 , version 1 (24-11-2017)

Identifiants

Citer

Rajesh Koduri, Sorin Olaru, Pedro Rodriguez-Ayerbe. Sensitivity of Piecewise control laws with respect to perturbation of the state-space partition. The 21st International Conference on Control Systems and Computer Science (CSCS21), May 2017, Bucarest, Romania. ⟨10.1109/CSCS.2017.9⟩. ⟨hal-01648113⟩
174 Consultations
184 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More