Nonlinear Geometric Approach to Fault Detection and Isolation in an Aircraft Nonlinear Longitudinal Model

In this paper, aircraft actuator fault detection and isolation (FDI) is investigated and designed using a nonlinear geometric FDI approach based on a nonlinear longitudinal aircraft model. Two detection filters are designed for the throttle position and the elevator angle, respectively, which are the two main actuation signals in the longitudinal aircraft model. In nonlinear geometric FDI approach the objective is to find state and output transformations, if such transformations exist, that lead us to a new set of observable states which are unaffected by all faults but one. Numerous simulation results show the excellent performance of the designed nonlinear diagnosis filters in detecting and isolating certain types of faults such as float and loss of effectiveness in the input channels. Comparative simulation results are conducted to demonstrate the superiority of the proposed nonlinear filters to their linear counterparts. It is shown that the linear geometric approach may fail to detect and isolate the above faults mainly due to the model inaccuracies that are inherent to the linearization of the nonlinear aircraft model.