CONTROL OF FLEXIBLE SPACECRAFT BY OPTIMAL MODEL FOLLOWING.

摘要

Spacecraft flexibility resulting from increased size and light-weight design techniques significantly increases the interaction between the structural bending modes and the vehicle control system. The subject of this thesis is the use of the unique properties of Optimal Model Following (OMF) to simplify the controller design process for flexible spacecraft. Model following controllers force the plant states to follow the dynamics of a model which incorporates the system performance specifications. OMF, model following using a linear quadratic cost function, trades off the degree of dynamics matching with the control effort utilized.; As part of this thesis, an OMF controller design program was developed. OMF controllers were designed for three examples: attitude control of a flexible spacecraft, pointing of a camera package on a flexible appendage, and control of a spacecraft based on the Galileo vehicle. The performance characteristics of the OMF designs were compared to those of optimal controllers.; OMF is a hybrid design technique which combines many of the advantages of classical and modern control design. System time and frequency performance specifications can be incorporated directly into the design model. OMF decouples the design states from the rest of the system and matches the transfer functions of the design states to those of the model states. In this manner, the model dynamics will predominate in the closed loop system time response. Decoupling isolates the design states from the effects of disturbances and unmodeled system dynamics. OMF facilitates the use of reduced order models in controller design. If initial designs are unsuccessful, the design program provides the information necessary to eliminate conflicting design requirements or to modify or supplement the existing controls in order to achieve the desired performance.; These advantages of OMF design are not achieved without accompanying disadvantages. The design state decoupling is generally achieved at the expense of higher feedback gains and an attendant increase in system sensitivity to sensor noise. The decoupling process can act as an "information barrier" as well as a "disturbance barrier", making the process of state measurement or estimation more difficult, and requiring additional sensors.