Data validation and diagnosis using interval analysis

José Ragot (1) , Didier Maquin (1)

1 CRAN - Centre de Recherche en Automatique de Nancy

Parameter or state estimation plays an important role in numerous engineering fields such as function estimation, system identification, controler design, data validation, diagnosis. Classical estimation methods compute the parameters so that the mean-squared error between the model output and the observed response is minimized. A robust approach to the estimation problem may be used when the error is bounded. This approach, called parameter set estimation, aims to find the feasible set of parameters consistent with all observed data and error bounds. State and parameter estimation problems are usually solved by statistical approaches, which are relevant when an explicit characterization of the perturbances is available. Unfortunately, it is often difficult to get this information and, in some cases, it is more natural to assume that all perturbations (measurement and modelling errors, model uncertainty), belong to a known set. In this case, guaranteed estimation, also known as bounded-error estimation, allows the characterization of the whole set of state or parameter vectors that are compatible with the measured data, the model of the process and the error bounds. This paper deals with model based data validation and fault detection-isolation considering the effect of model uncertainty, these uncertainties being represented by interval values.

Domaines: Automatique / Robotique

José Ragot, Didier Maquin. Data validation and diagnosis using interval analysis. Workshop on Advanced Control and Diagnosis, Nov 2005, Mulhouse, France. pp.CDROM. ⟨hal-00151263⟩