Switched iterative learning attitude and structural control for solar power satellites

A switched iterative learning controller (ILC) is proposed for a flexible solar power satellite (SPS) to periodically track the earth (transmitting antenna) and the sun (solar array). ILC is advantageous to deal with batch processes because the feedforward control signal learns from the inputs and errors of previous operations. However, the main difficulty applying ILC to attitude control of the SPS is that the periodic attitude motions are continuous processes instead of batch processes. The ILC would cause sudden change of control inputs and severe oscillation of control errors when the feedforward signals are suddenly added into the controller. To address this problem, a smooth switch is designed to avoid altering the control inputs abruptly. Furthermore, a filter is adopted based on Fourier series to eliminate the adverse influence of useless signals from the previous periods, such as sensor noise and structural vibrations. It is found from numerical simulations that the proposed switched ILC improves the attitude control accuracies by one to two orders of magnitude. Moreover, it reduces the interaction between attitude of the antenna and the structural vibrations of the truss.