High Performance Robust Control Of Flexible Space Structures

Many spacecraft systems have ambitious objectives that place stringent requirements in terms of pointing and vibration control systems. Because of the difficulty of obtaining accurate models for flexible space structures, limitations must be placed on performance in order to ensure stability. In some cases, to achieve sufficiently high performance to accomplish mission objectives will require the ability to refine the control design model based on closed-loop test data and tune the controller based on the refined model. The objective of this thesis is to develop a procedure for control system design which maximizes achievable performance with robust stability guarantees. Two facts of this dissertation are control design and system identification. A control design and system identification. A control design procedure is developed based on mixed H2/H[Infinite symbol] [subscript] optimization which is used to design a set of controllers which explicitly trade between nominal performance and robust stability. A homotopy algorithm is presented which generates a trajectory of gains along the boundary which trades between robustness and performance. This set of controllers may be implemented to determine the maximum achievable performance for a given model error bound. Examples are given which show that a better balance between robustness and performance is obtained using the mixed H2/H[Infinite symbol][subscript] design method than either H2 or ư-synthesis control design. A second contribution of this dissertation is a new procedure for closed-loop system identification. Using closed-loop response data, the parameters of a control design model in a canonical realization are refined. Examples are provided to demonstrate the convergence of the parameter esimation and improved performance realized by using the refined model for controller redesign. These developments result in an effective mechanism for achieving high performance control of flexible space structures.