Flight Test Comparison of Different Adaptive Augmentations of Fault Tolerant Control Laws for a Modified F-15 Aircraft

Abstract

This report describes the improvements and enhancements to a neural network-based approach for directly adapting to aerodynamic changes resulting from damage or failures. This research is a follow-on effort to flight tests performed on the NASA F-15 aircraft as part of the Intelligent Flight Control System research effort. Previous flight test results demonstrated the potential for performance improvement under destabilizing damage conditions. Little or no improvement was provided under simulated control surface failures, however, and the adaptive system was prone to pilot-induced oscillations. An improved controller was designed to reduce the occurrence of pilot-induced oscillations and increase robustness to failures in general. This report presents an analysis of the neural networks used in the previous flight test, the improved adaptive controller, and the baseline case with no adaptation. Flight test results demonstrate significant improvement in performance by using the new adaptive controller compared with the previous adaptive system and the baseline system for control surface failures. Nomenclature

Neural network-based flight control systems: Present and future

“…The evaluation was based on performance measurements and pilot ratings. As reported in [26,27], for structural damages, NN augmentation was generally found to provide significant improvements in overall pitch performance. However, control surface faults led to mixed results from slight improvements in pitch rate response to a propensity for roll pilot-induced oscillation.…”

“…The utilized modifications included the use of alternate NN inputs in the de-signed framework which can satisfactorily tackle high-correlation and high-gain problems in the basic design, the adoption of a weight decay term (in updating the NN weights) to avoid the overfitting problem, and using scalar dead-zones in adaptation laws for simplicity. The results obtained by the modified control scheme in 2009 indicated a significant improvement over the basic design [27].…”

“…using the control command u c defined by ( 26)- (27), and employing the same procedure as given in Section 2.2 (by substituting e by ē), it can be concluded that both the tracking error ē and the estimation error of the weight matrix W are bounded. In this regard, due to the substitution of e by ē in the updating rule of the NN weights, they can be satisfactorily updated even at the time of input saturation owing to the elimination of the effect of input constraints from ē using the introduced term ν h .…”

Handling Qualities Evaluations of Low Complexity Model Reference Adaptive Controllers for Reduced Pitch and Roll Damping Scenarios

“…Examples that do exist include the first flight tests of a model reference adaptive system on the F-94A aircraft (Lockheed Martin, Bethesda, Maryland), 1 an experimental adaptive flight control system evaluated on the F-94C aircraft (Lockheed Martin, Bethesda, Maryland), 2 the implementation of an adaptive flight control system on the X-15 aircraft (North American Aviation, Inc., Downy, California), 3,4 testing of an indirect-adaptive self designing controller (SDC) on the F-16 VISTA (General Dynamics, now Lockheed Martin, Bethesda, Maryland) 5 and the intelligent flight control system (IFCS) research on the highly-modified NASA F-15 aircraft (McDonnell Douglas, now The Boeing Company). 6,7 By the mid-2000s, experiments such as SDC and IFCS had demonstrated that adaptive flight controls can be an effective technology for improving aircraft safety in the event of failures or damage. However, the nonlinear, timevarying nature of adaptive systems continues to challenge traditional methods of verification and validation of safety-critical flight control systems.…”