Attitude Control of Underpowered Spacecraft Assisted by Multimicrosatellites via Incomplete Information Nonzero-Sum Differential Game

In the integrated satellite–terrestrial Internet of Things, an underpowered target spacecraft would affect the Internet performance. In this article, a learning-based game control algorithm is developed for autonomous microsatellites that are implemented to assist an underpowered target spacecraft in controlling its attitude. The target spacecraft and microsatellites are modeled as rational players, each of which has an individual cost function for describing its control behavior. Note that, typically, neither the control strategy nor the cost function of the target spacecraft is available to the microsatellites. Accordingly, the attitude assistance control system is modeled as an incomplete information, nonzero-sum differential game (DG). To address this issue, the idea of learning is integrated into game control methods. First, for autonomous microsatellites, a concurrent learning (CL) law that leverages system state data to identify the target spacecraft’s feedback matrix is developed. The convergence of the estimate to its true value is guaranteed if the system state is finite-time exciting. Compared with the persistent excitation condition needed by conventional online adaptive parameter estimation techniques, the convergence condition in this article is more relaxed. Then, the game control policies of the autonomous microsatellites are calculated online in a distributed manner, which are also proven to converge to a Nash equilibrium with the feedback control law of the target spacecraft. Finally, a simulation example demonstrates the validity of the developed learning-based game control algorithm.