Variable sampling-time nonlinear model predictive control of satellites using magneto-torquers

Abstract

Satellite control using magneto-torquers represents a control challenge combined with strong nonlinearity, variable dynamics and partial controllability. An automatic differentiation-based nonlinear model predictive control (NMPC) algorithm is developed in this work to tackle these issues. Based on the previously developed formulation of NMPC, a novel variable sampling-time scheme is proposed to provide a better trade-off between transient control performance and closed-loop stability. More specifically, a small sampling time is adopted to improve the response speed when the satellite is far away from the desired position, and a large sampling time is employed for the closed-loop stability when the satellite is around its equilibrium position. This scheme also significantly reduces the online computational burden associated with fixed sampling-time NMPC where a large prediction horizon has to be adopted in order to the ensure closed-loop stability. The proposed approach is demonstrated through nonlinear simulation of a specific satellite case with satisfactory results obtained.