Model selection via worst-case criterion for nonlinear bounded-error estimation

S. Brahim-Belhouari (1) , Michel Kieffer (2, 3, 4) , G. Fleury (1) , Luc Jaulin (5, 6, 1) , Eric Walter (2)

1 LISA - Laboratoire d'Ingéniérie des Systèmes Automatisés
2 L2S - Laboratoire des signaux et systèmes
3 LTCI - Laboratoire Traitement et Communication de l'Information
4 IUF - Institut universitaire de France
5 STIC - Département STIC [Brest]
6 Lab-STICC - Laboratoire des sciences et techniques de l'information, de la communication et de la connaissance

In this paper the problem of model selection for measurement purpose is studied. A new selection procedure in a deterministic framework is proposed. The problem of nonlinear bounded-error estimation is viewed as a set inversion procedure. As each candidate model structure leads to a specific set of admissible values of the measurement vector, the worts-case criterion is used to select the optimal model. The selection procedure is applied to a real measurement problem, grooves dimensioning using Remote Field Eddy Current (RFEC) inspection.

Domaines: Sciences de l'ingénieur [physics]

S. Brahim-Belhouari, Michel Kieffer, G. Fleury, Luc Jaulin, Eric Walter. Model selection via worst-case criterion for nonlinear bounded-error estimation. IEEE Instrumentation and Measurement Magazine, 2000, 49 (3), pp.653-658. ⟨hal-00844852⟩