ESA GNC Conference Papers Repository

Scope of this work is to analyze the performance of different algorithms in the field of Sun vector estimation. This has been done using real flight data of the AGILE mission, a scientific mission for study of X-ray and gamma ray burst funded by the Italian Space Agency (ASI). Since the Sun estimation algorithm currently used on board for the mission in certain conditions shows some undesired effects, other algorithms have been implemented in order to perform a comparison. All of the examined algorithms receive as input the currents of the Coarse Sun Sensors, which are simple solar cells as the ones used for power generation. In order to evaluate algorithm accuracy, all the algorithms estimations have been compared using a common reference. This reference is the attitude estimated by the Star sensors, since they are by far more accurate than the Coarse Sun Sensors. As alternative algorithms, the followings have been considered: <br>- an improved version of the algorithm currently in use <br>- an algorithm based on the Two Cosine Detectors method, which using two couple of cells, estimates the direction cosines of the Sun respect to a reference axis <br>- the Grubin algorithm, which performs Sun vector estimation through a cones intersection technique. It has been used in the classical version and in an improved version <br>- Least Square algorithm, which receives as input the currents of different sensors and estimates the Sun direction by multiplication of the pseudoinverse of the unit vectors matrix with the currents data matrix. <br>Least Square algorithm and the improved Grubin algorithm are the methods that lead to the better results. In fact, these methods use a larger number of cells than the others, thus the errors are minimized. In particular, all the algorithms considered are influenced by the presence of the Albedo of the Earth, which acts as a disturbance, especially near the end of an eclipse period. Furthermore, the analysis of data taken in different periods highlighted the influence of the calibration state of the sensors on estimation performance. The results of this work have been used to choose the best algorithm to accomplish this task for future missions.