Fighter aircrafts are the systems that operate in highly nonlinear aerodynamic regime, performing complex maneuvers. This situation requires the use of nonlinear control design techniques rather than conventional linear control approaches. Nonlinear dynamic inversion (NDI) has been proved to be a suitable technique for fighter aircraft. However, it has some drawbacks, first is related to the modeling error that make inversion not to be perfect; second is the limited robustness provided by the linear controller; and finally, the structure of NDI requires the knowledge of the full dynamics of the system which leads to a complex design. The incremental form of NDI (INDI), a Lyapunov based NDI (L-NDI), and an Adaptive NDI using neural networks, are solutions to overcome these limitations.
In this thesis, a control augmentation system (CAS) of a fighter aircraft is designed using L-NDI technique and its incremental form (INDI) for F-18 HARV aircraft, whose performances are tested by a high-g maneuver. The control system presented is developed using the time-scale separation approach structure that provide longitudinal and lateral-directional control of the aircraft. The results are compared and discussed. Then, an Adaptive NDI is designed using an on-line neural network structure, which provides robustness to the system. Finally the three systems L-NDI, Adaptive NDI and INDI are tested under some cases of uncertainties and modeling errors, to check their robustness, compare the results and point it out the advantages and disadvantages.
The results show that the three methodologies covered in this thesis are robust to modeling errors, however Adaptive NDI and INDI present better tracking performance. Additionally INDI simplifies the design of the controller.