Adaptive nonlinear attitude control of spacecraft

Tang, Cheng
Major Professor
Ping Lu
Committee Member
Journal Title
Journal ISSN
Volume Title
Research Projects
Organizational Units
Aerospace Engineering
Organizational Unit
Journal Issue
Aerospace Engineering

Three adaptive nonlinear control approaches are proposed for attitude control of a Space Station. These control algorithms avoid the need for linearization, either by truncating a Taylor series or by a feedback linearization approach, which poses some difficulties in the presence of both uncertain mass properties and external disturbances. Global stability of the Space Station is guaranteed;A conventional linear-quadratic regulator synthesis technique is investigated. This linear control scheme provides a natural way to select control gains for the known spacecraft system parameters. Simulation results reveal that the uncertain mass property causes the nonlinear Space Station to be uncontrollable;On the basis of two nonlinear adaptive control approaches originally developed for robot manipulators, Two controllers are developed to overcome the problems introduced by uncertain inertias. The improvement on transient state responses is obtained by optimizing control parameters. A compensation term which adaptively adds periodic signals to the input is appended to both controllers to cancel the effect of the cyclic disturbances. The convergence properties of the adaptive algorithms are proved. The application to the Space Station attitude control of both nonlinear control schemes shows good performance in the presence of uncertain inertias and disturbances. Robustness with the unmodeled dynamics for both algorithms are also tested;Another new approach that utilizes the neural network with a radial basis function technique to compensate for the effect of the disturbances is also presented. The neural networks provide an estimate of the cyclic disturbances and act as an open-loop control so that the overall attitude oscillation caused by the disturbances is minimized. This neural network control law is applied to the Space Station in conjunction with the nonlinear controllers which are used to shape the system responses. The simulation results show that, compared to the case where only nonlinear controllers are used, the required control torques are significantly smaller. It has also been shown that the trained networks are capable of extrapolating;Finally, when the attitude control is achieved by control momentum gyros (CMGs), an approximate feedback linearization method is applied to simultaneous attitude control and momentum management with unknown constant disturbances. A direct adaptive controller is developed to stabilize the system with large disturbances. Simulations are carried out to show that the Space Station can be stabilized within a reasonable range, whereas ignoring the disturbances in the controller can lead to destruction of the Space Station and saturation of the CMGs.