Sensor fault estimation for nonlinear systems described by multiple models

This paper deals with the problem of sensor fault estimation for linear and nonlinear systems. Thanks to the introduction of an augmented generalized state vector, including the original state vector and a filtered output of the system, the sensor fault ap- pears as an unknown input. Therefore, an adaptive proportional integral observer is used to estimate simultaneously the state of the system and the unknown input. In order to provide some robustness properties, the disturbance effect on the state and fault estimation errors is minimized in an L2 framework. State and sensor fault estimation is firstly presented for linear systems and is next extended to nonlinear systems described by Takagi- Sugeno models.