We propose a new technique for the identication of discretetime hybrid systems in the Piece-Wise Ane (PWA) form. The identi- cation algorithm proposed in [10] is rst considered and then improved under various aspects. Measures of condence on the samples are introduced and exploited in order to improve the performance of both the clustering algorithm used for classifying the data and the nal linear regression procedure. Moreover, clustering is performed in a suitably de- ned space that allows also to reconstruct dierent submodels that share the same coecients but are dened on dierent regions.
A clustering technique for the identification of piecewise affine systems
M Muselli;D Liberati;
2001
Abstract
We propose a new technique for the identication of discretetime hybrid systems in the Piece-Wise Ane (PWA) form. The identi- cation algorithm proposed in [10] is rst considered and then improved under various aspects. Measures of condence on the samples are introduced and exploited in order to improve the performance of both the clustering algorithm used for classifying the data and the nal linear regression procedure. Moreover, clustering is performed in a suitably de- ned space that allows also to reconstruct dierent submodels that share the same coecients but are dened on dierent regions.File in questo prodotto:
Non ci sono file associati a questo prodotto.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
