Fault Diagnosis Techniques for Aircraft Simulated Model Sensors

This paper addresses the study and the application of a comprehensive set of methodologies for fault detection and isolation of aircraft model input--output sensors. In particular, a novel nonlinear approach and an efficient linear method are developed and analysed with extensive simulation results obtained from a commercial aircraft model. The developed nonlinear method exploits adaptive filters for fault identification with disturbance de--coupling achieved via a nonlinear geometric approach. On the other hand, the linear filter design for FDI is based on a polynomial method that allows also to achieve good disturbance de--coupling properties. The FDI strategies are applied to the aircraft simulator data in a flight condition characterised by tight--coupled longitudinal and lateral dynamics. Finally, the design reliability of the two FDI schemes is assessed by means of extensive simulations in the presence of turbulence, measurement noise and modelling errors.