Refined Instrumental Variable methods for closed-loop system identification

Marion Gilson (1) , Hugues Garnier (1) , Peter J.W. Young (2) , Paul van den Hof (3)

1 CRAN - Centre de Recherche en Automatique de Nancy
2 CRES - Centre for Research on Environmental Systems and Statistics
3 DCSC - Delft Center for Systems and Control [Delft]

This paper describes optimal instrumental variable methods for identifying discrete-time transfer function models when the system operates in closed-loop. Several noise models required for the design of optimal prefilters and instruments are analyzed and different approaches are developed according to whether the controller is known or not. Moreover, a new optimal refined instrumental variable technique is developed to handle the identification of a linear (ARX) predictor combined with an ARMA noise model in a closed-loop framework. The proposed refined instrumental variable algorithm achieves minimum variance estimation of the process model parameters. The performance of the proposed approaches is evaluated by Monte-Carlo analysis in comparison with other alternative closed-loop estimation methods.

Mots clés: optimal instrumental variable; System identification; closed-loop identification; optimal instrumental variable.

Domaines: Automatique / Robotique

Marion Gilson, Hugues Garnier, Peter J.W. Young, Paul van den Hof. Refined Instrumental Variable methods for closed-loop system identification. 15th IFAC Symposium on System Identification, SYSID 2009, Jul 2009, Saint-Malo, France. pp.CDROM. ⟨hal-00411754⟩