Shadow-Aware Formation Flying Using Solar Radiation Pressure with Neural Network

Propellant-free formation flying using environmental forces offers significant benefits in terms of system simplification and mission life extension. In high orbits around the Earth, solar radiation pressure (SRP) is available. Formation is controlled by appropriately changing the attitude and shape of the satellite to generate the desired SRP acceleration. For this purpose, it is essential to model SRP corresponding to these satellite states. Conventional formation control methods using SRP rely on models that ignore self-shadowing caused by, for example, solar array paddles. The gap between modeled and actual SRP can degrade control performance. To solve this problem, we propose a novel SRP model that takes into account self-shadowing for nonconvex satellites and a formation control law based on the model. The self-shadowing effect is estimated at low computational cost by a neural network that imitates ray-tracing analysis. Numerical simulation results for formation establishment and maintenance show the effectiveness of the proposed method and the importance of considering the self-shadowing.