Ein lernender Autopilot mit selbsteinstellender robuster Regelung

This report examines the control of vehicle movement in three-dimensional space. This involves the planning and creation of adapted target trajectories for the control variables and their regulation. The main point of emphasis in the model-assisted multivariable design is in improving the degree of guidance accuracy, taking due account of any uncertainties there may be in the model. The control system comprises two parts: a H_#infinity#robust/H_2-optimum multivariable controller for counteracting unknown sources of interference (feedback control) and a trainable pilot control system using a nonlinear inverse model for achieving the guidance behaviour required (feedforward control). The investigations are simulated taking a submarine as an example. The model of motion for submarines which is generally applied by hydromechanical engineers and which is parameterized in planar motion tests is outlined. Models for actuator and sensor technology are added as they have to be taken into account in the control design. Simplified motion equations also have to be developed which shall provide the basis for all of the other investigations. The investigation is concluded with the results of the simulation for the purposes of assessing the motion control. The investigations described here have their origin in the flight control technology. (orig./AKF)Available from TIB Hannover: H93B2199 / FIZ - Fachinformationszzentrum Karlsruhe / TIB - Technische InformationsbibliothekSIGLEDEGerman