Guaranteed fault detection based on interval Constraint Satisfaction Problem

This paper deals with a fault detection method taking model uncertainties described by bounded variables into account. This problem is treated as an interval Constraint Satisfaction Problem (CSP) whose relations depend on both sought unknown variables and on uncertain model parameters belonging to interval supports. A set-membership inversion method is used for solving this interval CSP. An analysis on used existential and universal quantificators is provided and an example shows that the proposed approach is also suited to a large variety of problems handled in Automatic Control.