In this paper we solve a non-smooth convex formulation for passivity enforcement of linear macromodels using robust localization based algorithms such as the ellipsoid and the cutting plane methods. Differently from existing perturbation based techniques, we solve the formulation based on the direct ℌ∞ norm minimization through perturbation of state-space model parameters. We provide a systematic way of defining an initial set which is guaranteed to contain the global optimum. We also provide a lower bound on the global minimum, that grows tighter at each iteration and hence guarantees δ - optimality of the computed solution. We demonstrate the robustness of our implementation by generating accurate passive models for challenging examples for which existing algorithms either failed or exhibited extremely slow convergence.
Robust localization methods for passivity enforcement of linear macromodels / Mahmood, Z.; Chinea, Alessandro; Calafiore, Giuseppe Carlo; GRIVET TALOCIA, Stefano; Daniel, L.. - STAMPA. - (2013), pp. 1-4. (Intervento presentato al convegno 17th IEEE Workshop on Signal and Power Integrity (SPI) tenutosi a Paris nel 12-15 May 2013) [10.1109/SaPIW.2013.6558312].
Robust localization methods for passivity enforcement of linear macromodels
CHINEA, ALESSANDRO;CALAFIORE, Giuseppe Carlo;GRIVET TALOCIA, STEFANO;
2013
Abstract
In this paper we solve a non-smooth convex formulation for passivity enforcement of linear macromodels using robust localization based algorithms such as the ellipsoid and the cutting plane methods. Differently from existing perturbation based techniques, we solve the formulation based on the direct ℌ∞ norm minimization through perturbation of state-space model parameters. We provide a systematic way of defining an initial set which is guaranteed to contain the global optimum. We also provide a lower bound on the global minimum, that grows tighter at each iteration and hence guarantees δ - optimality of the computed solution. We demonstrate the robustness of our implementation by generating accurate passive models for challenging examples for which existing algorithms either failed or exhibited extremely slow convergence.File | Dimensione | Formato | |
---|---|---|---|
cnf-2013-spi-passivity.pdf
accesso aperto
Tipologia:
2. Post-print / Author's Accepted Manuscript
Licenza:
PUBBLICO - Tutti i diritti riservati
Dimensione
200.57 kB
Formato
Adobe PDF
|
200.57 kB | Adobe PDF | Visualizza/Apri |
Pubblicazioni consigliate
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/11583/2510283
Attenzione
Attenzione! I dati visualizzati non sono stati sottoposti a validazione da parte dell'ateneo