Efficient and Robust Attitude Determination and Control System Design for the MIST CubeSat

Sammanfattning: This thesis investigated different configurations of the attitude determination and control system (ADCS) for the MIST satellite, to find a satisfying trade-off between computational demand and estimation/pointing accuracy. A model of the satellite dynamics was developed and used in a simulation. The designed ADCS consists of a discrete extended Kalman filter (EKF) and a model predictive control (MPC) controller tunable in different ways. The filter works with a linearization of the spacecraft dynamics model which is performed about the last attitude estimate and it is also capable of estimating the residual magnetic moment of the spacecraft without any initial guess. Three different models were used with the MPC and compared: a linear-like, state-dependent model, a model linearized around a fixed equilibrium point, and a model linearized around the last attitude estimate. The simulation, developed with Simulink, served as a testbed for the different tunings. From the simulation results, the filter proved to be capable of estimating the residual magnetic moment of the satellite with satisfying accuracy. Estimation and pointing requirements were met on average with a mean absolute estimation error of 0.8 deg and a mean absolute pointing error of 3.5 deg. This performance was achieved in face of measurement and model uncertainty.