Output feedback sliding mode control of Lorentz-augmented spacecraft hovering using neural networks

Abstract

Traditional thrusters using chemical fuels could hardly afford long-term hovering in low Earth orbits due to the considerable fuel consumptions, which, however, could be feasible with the geomagnetic Lorentz force acting on a charged Lorentz spacecraft as auxiliary propulsion. And this kind of hovering using hybrid control inputs consisting of the net charge and thruster-generated control force of Lorentz spacecraft is termed as Lorentz-augmented spacecraft hovering. Based on the dynamical model of Lorentz-augmented spacecraft relative motion, fuel-optimal open-loop control laws are first derived by a Lagrange method. To achieve closed-loop hovering in the presence of external perturbations and system uncertainties and in the absence of velocity measurements, a neural network–based observer and an adaptive output feedback sliding mode controller are designed. Neural networks are used to approximate the unknown nonlinear dynamics, and sliding mode control method is used to ensure the system robustness. The adaptive tuning law of the neural networks is derived by a Lyapunov approach to guarantee the uniform ultimate boundedness of the overall closed-loop system. The validity and feasibility of the proposed control schemes are substantiated by numerical simulations.