Guaranteed nonlinear parameter estimation from error bounded data from NaI detectors
Résumé
Interval Analysis is a new tool, well known in Automatic Control, very powerfull for solving estimation problem. This appraoch is especially suited for models whose outputs is nonlinear in their parameters, a situation which is the rule for models based on prior physical knowledge but where most available methods fail to provide any guarantee as the global validity of the results obtained. It may be useful for NaI detector measurement system commonly used by the EURATOM safeguards system. One of the problems encountered during the in-field measurements of this type is the strong influence of the thickness and type of the material of container on the measurement results. We explore some possible solutions and validate these solutions on the real JRC Ispra PERLA dataset.
Domaines
Applications [stat.AP]
Origine : Fichiers produits par l'(les) auteur(s)