Résumé:
This thesis deals with a method for detection and isolation sensors faults on
a dynamic linear time-varying system. This method consists in estimating the faults
magnitude thanks to a least square module coupled to the Kalman filter estimating the
system state. The proposed approach allows considering additive and simultaneous
faults.
Keywords: detection, localization, faults, drifts, sensors, estimation, Kalman filter, least
squares, dynamical systems.