Time-Varying Model Predictive Attitude Control for Magnetically Actuated Dual-Spin Satellites

Abstract

Attitude control hardware for small satellites is often limited in power and space availability given the importance of the science instruments they exist to transport. To mitigate this, a dual-spin stabilized satellite actuated via magnetic torque rods reduces the space and power required of the attitude control system, but may require advanced control policies. This paper explores the attitude control of a magnetically actuated dual-spin stabilized CubeSat with model predictive control using time-varying prediction dynamics. An inertial pointing objective is used as a representative mission, where the satellite is able to deviate from its nominal orientation within some allowable amount. Three time-varying MPC policies are developed and compared to ensure the system does not violate constraints while minimizing control effort. These policies include a prediction model that accounts for the orbital position of the satellite and two iterative approaches that incorporate control inputs through either propagation of the linear dynamics, or nonlinear propagation with successive linearization. Results demonstrate that the nonlinear prediction policy outperforms other prediction methods with regards to not only minimal actuation, but to constraint satisfaction as well while imparting minimal computational burden.