Robust fault detection using interval constraints satisfaction and set computations

Abstract

In this chapter, the robust fault detection problem for nonlinear systems considering both bounded parametric modeling errors and noises is addressed. Fault detection is formulated as a set-membership estimation problem being this one of the contributions of the chapter. A state estimator that describes the set of all the states consistent with modeling uncertainty, measured data, and noise bounds is presented. Two possible implementations of such state estimation based on constraint satisfaction and set computation techniques, of the state estimator are proposed and compared, being this a second contribution. Finally, the proposed approaches are applied to detect faults in limnimeters of a piece of Barcelona sewer network.