Flight control design using a blend of modern nonlinear adaptive and robust techniques.

摘要

In this dissertation, the modern control techniques of feedback linearization, μ synthesis, and neural network based adaptation are used to design novel control laws for two specific applications: F/A-18 flight control and reusable launch vehicle (an X-33 derivative) entry guidance. For both applications, the performance of the controllers is assessed.; As a part of a NASA Dryden program to develop and flight test experimental controllers for an F/A-18 aircraft, a novel method of combining μ synthesis and feedback linearization is developed to design longitudinal and lateral-directional controllers. First of all, the open-loop and closed-loop dynamics of F/A-18 are investigated. The production F/A-18 controller as well as the control distribution mechanism are studied. The open-loop and closed-loop handling qualities of the F/A-18 are evaluated using low order transfer functions. Based on this information, a blend of robust μ synthesis and feedback linearization is used to design controllers for a low dynamic pressure envelope of flight conditions. For both the longitudinal and the lateral-directional axes, a robust linear controller is designed for a trim point in the center of the envelope. Then by including terms to cancel kinematic nonlinearities and variations in the aerodynamic forces and moments over the flight envelope, a complete nonlinear controller is developed. In addition, to compensate for the model uncertainty, linearization error and variations between operating points, neural network based adaptation is added to the designed longitudinal controller. The nonlinear simulations, robustness and handling qualities analysis indicate that the performance is similar to or better than that for the production F/A-18 controllers. When the dynamic pressure is very low, the performance of both the experimental and the production flight controllers is degraded, but Level I handling qualities are still achieved.; A new generation of Reusable Launch Vehicles (RLVs) demands improvements to the current entry guidance capability. The acceleration entry guidance method has proven effective for the current Space Shuttle Orbiter. The basic concept is to track a reference drag acceleration generated by a planner, replanning the drag profile periodically if necessary. Tracking the drag profile ensures that the vehicle will fly the correct distance to reach the target without violating the vehicle constraints. (Abstract shortened by UMI.)