Application of Fault Diagnosis Methodologies to a General Aviation Aircraft

This paper addresses the problem of the detection and isolation of input sensor faults on a general aviation aircraft, characterised by a nonlinear model, in the presence of wind gust disturbance and measurement errors. In particular, this work studies and compares different residual generator designs in order to realise complete diagnosis schemes, when additive faults are present. These different methods comprise linear and nonlinear filters, neural networks and unknown input Kalman filters, that can achieve disturbance signal de-coupling and robustness properties with respect to modelling error and sensor measurement noise. The results obtained in the simulation of the faulty behaviour of a PIPER PA30 longitudinal aircraft model are finally reported.