Recursive identification of time-varying systems using minimum variance
TIAN ; WAHL ; VASSEUR
Type de document
COMMUNICATION AVEC ACTES INTERNATIONAL (ACTI)
Langue
anglais
Auteur
TIAN ; WAHL ; VASSEUR
Résumé / Abstract
This paper presents a new on-line identification method based on minimum variance using a sliding data window in order to improve robustness against noise and effective tracking ability. This method is based on a local linearization model between the sliding data window bounds and on an incremental procedure. The corresponding algorithm is applied to a non-linear fermentation process. The results illustrate the performances of this method in comparison with other existent techniques.
