In-Orbit Three-Axis Spacecraft Orbit Control Based on Neural Networks via Limited Thrust Budget

Abstract

The problem of spacecraft orbit control is divided into coplanar and non-coplanar control algorithms. The general case of an orbital manoeuvre for a low earth orbit spacecraft is represented by a mixed planar and non-coplanar manoeuvre. Several algorithms for developing non-coplanar orbital manoeuvre exist in the literature. These algorithms usually require intensive calculations, representing a severe problem for application on-board a spacecraft in-orbit. In this research article, a spacecraft orbit control algorithm is developed based on Neural Networks. The developed orbit control algorithm has numerous advantages over the ones found in the literature. Various sources of nonlinearities such as plant nonlinearity, disturbance nonlinearity, and thrust saturation level are taken into consideration during the controller design process. Random disturbances are also considered during the controller design process. Training data is provided by a PD control algorithm with proven stability. The Stability of the learning algorithm is also proven. The developed controller can mimic the performance of the PD control algorithm, which is used for training. The developed algorithm can completely control the spacecraft’s orbit during the manoeuvre and avoid collision with the surface of the earth while the PD control algorithm has failed. This has assured the superior performance of the developed control algorithm.
Copyright © 2021 Praise Worthy Prize - All rights reserved.