Abstract
This article demonstrates the performance of an improved particle swarm optimization (PSO) algorithm with scalar checking and rotation axis fitting objectives using in-orbit data, which is obtained from two CubeSats missions, Aalto-1 and ESTCube-1, as well as simulation as reference. The improved algorithm uses sequential objectives refinement process to combine the two optimization objectives. This improvement addresses some challenges of magnetometer calibration when using in-orbit data. First, the change in the magnetic field vector direction at different points in orbit which is uncorrelated to the rotation of the spacecraft itself. Second, the uncertainty of the rotation axis information used as the reference, e.g., from gyroscope noise. Third, the available data set is heavily affected by the rotation mode of the spacecraft, which imposes some limitation in the rotation axis information needed by the algorithm. The improved PSO algorithm is applied on simulated data in order to analyze the calibration performance under different spacecraft tumbling rates and noise levels. In ideal condition (varying rotation axis during measurements and sufficient sampling rate relative to the spin rate), the rotation axis fitting objective can reach ∼0.1° of correction accuracy.
Highlights
A TTITUDE Determination and Control System (ADCS) performance of a spacecraft is a combination of its sensors and actuators technical performance and their calibration, which ensures their reliability in acceptable range of error
This relaxed knowledge requirement implies that the usable information for calibration process is limited, resulting in an under-determined estimation problem. This appears mathematically as a constraint in the calibration parameters, in the calibration matrix: the matrix is assumed as a diagonal or triangular matrix (3–6 independent elements instead of 12 elements in a full 3 × 3 matrix) [2]–[9]. This means that the distortion which can be corrected is limited to non-orthogonality and non-uniform scaling of the axes, with the assumption that at least one axis is perfectly aligned with reference, i. e. the magnetic field scalar based calibration cannot resolve rotational error
The rotation axis fitting algorithm uses the estimated rotation axis from magnetometer data to fit into the reference rotation axis; in this paper, the reference rotation axis is gathered from gyroscope measurement
Summary
A TTITUDE Determination and Control System (ADCS) performance of a spacecraft is a combination of its sensors and actuators technical performance and their calibration, which ensures their reliability in acceptable range of error. Tikka, Kestila, et al [13] have proposed a method to resolve this rotational error component using rotation axis information derived from the magnetometer reading and fit it into a reference rotation axis information, which, in spacecraft settings, can be read directly from gyroscopes or determined from other sensors This rotation axis fitting method was used complementary with the scalar checking method implemented inside a particle swarm optimization (PSO) algorithm. The algorithm is applied on simulated data that models the attitude dynamics in orbit, as well as using real inorbit flight data from Aalto-1 [14], [15] and ESTCube-1 [16], [17] satellites This means that the algorithm robustness is tested under the changing magnetic field direction experienced by the spacecraft in orbit due to Earth’s magnetic field vector variation with relation to the position in orbit, which will appear as if the magnetometer is rotating.
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call