Adaptive limit margin detection and limit avoidance

This thesis concerns the development of methods, algorithms, and control laws for the development of an adaptive flight envelope protection system to be used for both manned and unmanned aircraft. The proposed method lifts the requirement for detailed a priori information of aircraft dynamics by enabling adaptation to system uncertainty. The system can be used for limits that can be either measured or related to selected measurable quantities. Specifically, an adaptive technique for predicting limit margins and calculating the corresponding allowable control or controller command margins of an aircraft is described in an effort to enable true carefree maneuvering. This new approach utilizes adaptive neural network based loops for the approximation of required aircraft dynamics. For limits that reach their maximum value in steady state, a constructed estimator model is used to predict the maneuvering quasi-steady response behavior—the so called dynamic trim—of the limit parameters and the corresponding control or command margins. Linearly Parameterized Neural Networks as well as Single Hidden Layer Neural Networks are used for on-line adaptation. The approach does not require any off-line training of the neural networks, instead all learning is achieved during flight. Lyapunov based weight update laws are derived. The method is extended for multi-channelled control limiting for aircraft subject to multiple limits, and for automatic control and command limiting for UAV's. Simulation evaluations of the method using a linear helicopter model and a nonlinear Generalized Tiltrotor Simulation (GTRSIM) model are presented. Limit avoidance methods are integrated and tested through the implementation of an artificial pilot model and an active-stick controller model for tactile cueing in the tiltrotor simulation, GTRSIM. Load factor, angle-of-attack, and torque limits are considered as examples. Similarly, the method is applied to the Georgia Tech's Yamaha R-Max (GTMax) unmanned helicopter test bed for automatic limit detection and avoidance. Angular rate limits and rotor stall limits are considered. Software-in-the-loop simulation and flight test results of the GTMax are included.