Fault detection using differential flatness in flight guidance systems

Abstract : In this paper, flight guidance dynamics are shown to be implicit differentially flat with respect to the inertial position of an aircraft. This proves the existence of a set of relations between these flat outputs and the state variables representative of flight guidance dynamics and between these flat outputs and the basic inputs to flight guidance dynamics. A neural network is introduced to obtain, from the actual trajectory, nominal flight parameters which can be compared with actual values to detect abnormal behaviour.