Deciding Stability of a Switched System Without Identifying It

Balkan, Ayca;Jungers, Raphaël;Kenanian, Joris;Tabuada, Paulo
(2018) 2018 IEEE Conference on Decision and Control (CDC) — Location: Miami Beach, FL (17.December.2018)

Files

08619702.pdf
  • Open Access
  • Adobe PDF
  • 324.37 KB
Download

Details

Authors
  • Balkan, AycaECE Dept., University of California, Los Angeles (UCLA)
    Author
  • Jungers, RaphaëlUCLouvain
    Author
  • Kenanian, JorisECE Dept., University of California, Los Angeles (UCLA)
    Author
  • Tabuada, PauloECE Dept., University of California, Los Angeles (UCLA)
    Author
Abstract
We address the problem of deciding stability of a “black-box” dynamical system (i.e., a system whose model is not known) from a set of observations. The only assumption we make on the black-box system is that it can be described by a switched linear system. We show that, for a given (randomly generated) set of observations, one can give a stability guarantee, for some level of confidence, with a trade-off between the quality of the guarantee and the level of confidence. We provide an explicit way of computing the best stability guarantee, as a function of both the number of observations and the required level of confidence. Our results rely on geometrical analysis and combine chance-constrained optimization theory with stability analysis techniques for switched systems 1 .
Affiliations

Citations

  • APA
  • Chicago
  • FWB

Balkan, A., Jungers, R., Kenanian, J., & Tabuada, P. (2018). Deciding Stability of a Switched System Without Identifying It. 2018 IEEE Conference on Decision and Control (CDC). Published. 2018 IEEE Conference on Decision and Control (CDC), Miami Beach, FL. https://doi.org/10.1109/cdc.2018.8619702