CubeSat Attitude System Calibration and Testing

This thesis concentrates on the development of Aalto-2 CubeSat attitude system calibration and testing methods. The work covers the design and testing phase of the calibration algorithms to the analysis of experimental data in order to verify the performance of the attitude instruments. The instruments under test are two-axis digital Sun sensor, three-axis magnetometer, three-axis gyroscope, and three-axis magnetorquer. These devices are all commercial off-the-shelf components which are selected for their cost-to-performance efficiency. The Sun sensor and gyroscope were calibrated with linear batch least squares method and the results showed that only minor corrections were required for the Sun angle and angular velocity readings, while the brightness readings from the Sun sensor required more corrections. For magnetometer calibration, a specific particle swarm optimization algorithm was developed with novel approach to estimate the full calibration parameters, without having to simplify the sensor model. The calibration results were evaluated with simulation data with satisfying results, while the results from experimental data itself showed heading error improvement from \SIrange[range-phrase=--]{5.24}{13.24}{\degree} to \SIrange[range-phrase=--]{1.9}{7.3}{\degree} for unfiltered data. Besides the magnetometer calibration parameters estimation, the magnetic properties of the spacecraft were also analyzed using inverse multiple magnetic dipole modeling approach, where multiple magnetic dipoles positions and moments are estimated using particle swarm optimization from the magnetic field strength readings around the spacecraft. The estimated total residual magnetic moment of the spacecraft is \SI{58.5}{\milli\ampere\square\meter}, lower than the maximum magnetorquer moment which is \SI{0.2}{\ampere\square\meter} in each axis. The magnetorquer was tested for verifying the validity of magnetic moment generated by the magnetorquer. The result shows that the magnetorquer moment is nonlinear, in contrast to the linear theoretical model