Distributed Model Predictive Control of Leader-Follower Systems Using an Interior Point Method with Efficient Computations

I. Necoara 1 D. N. Clipici 1 Sorin Olaru 2
1 Department of Automatic Control and Systems Engineering [Bucharest]
2 DISCO - Dynamical Interconnected Systems in COmplex Environments
L2S - Laboratoire des signaux et systèmes, Inria Saclay - Ile de France, SUPELEC, CNRS - Centre National de la Recherche Scientifique : UMR8506
Abstract : Standard model predictive control strategies implythe online computation of control inputs at each samplinginstance, which traditionally limits this type of control scheme tosystems with slow dynamics. This paper focuses on distributedmodel predictive control for large-scale systems comprised ofinteracting linear subsystems, where the online computationsrequired for the control input can be distributed amongstthem. A model predictive controller based on a distributedinterior point method is derived, for which every subsystemin the network can compute stabilizing control inputs usingdistributed computations. We introduce local terminal sets andcost functions, which together satisfy distributed invarianceconditions for the whole system, that guarantees stability of theclosed-loop interconnected system. We show that the synthesisof both terminal sets and terminal cost functions can be donein a distributed framework.
Document type :
Conference papers
Complete list of metadatas
https://hal-supelec.archives-ouvertes.fr/hal-00827199
Contributor : Josiane Dartron <>
Submitted on : Wednesday, May 29, 2013 - 8:09:27 AM
Last modification on : Thursday, April 26, 2018 - 2:18:53 PM

Identifiers

  • HAL Id : hal-00827199, version 1

Collections

SUPELEC | INRIA | SUP_LSS | TDS-MACS | L2S

Citation

I. Necoara, D. N. Clipici, Sorin Olaru. Distributed Model Predictive Control of Leader-Follower Systems Using an Interior Point Method with Efficient Computations. 2013 IEEE American Control Conference ACC, Jun 2013, Washington, United States. pp.1-8. ⟨hal-00827199⟩

Export

BibTeX TEI DC DCterms EndNote

Share

Metrics

Record views

433

Contact

support.ccsd.cnrs.fr
hal.support@ccsd.cnrs.fr
Logo CCSD