International Journal of Metrology and Quality Engineering (Jan 2017)
Singular value decomposition based satellite attitude determination using different sensor configurations
Abstract
Satellite's attitude can be determined by using several different ways. In order to use a recursive filter to determine the attitude especially after crucial intervals for the satellite including detumbling, eclipse, etc., single-frame methods can be used, and singular value decomposition (SVD) method has been selected because of its robustness for this paper. They can be placed for the initial values, or continuous input to the filter in addition to covariance matrix at each step. Also, sensor configurations depending on the position of the Sun are considered and used in the SVD method. Wahba's loss function is minimized with separating the basic matrix including the reference and model vectors into singular values because many researches show the SVD as the most robust algorithm after error analyses between several single-frame methods. Sun sensor, magnetometer, and horizon sensor are the selected common attitude sensors for small satellites in order to compare their attitude determination performances. For this purpose, absolute errors for different sensor configurations in the simulation environment in addition to the RMS errors are presented. The purpose of the paper is to present an attitude determination concept for a nanosatellite. Also, different sensor configurations are considered in the single-frame method with switching logic depending on the vector data. Horizon sensor and sun sensor are the optimum pair for the light side of the Earth on the other hand, during the eclipse; horizon sensor and magnetometer give the most accurate results. Study shows that the proposed algorithm can be performed for initial values to the Kalman filter, and in the transient phase before the filter tuned in practical satellite usage. In the eclipse period, sun-independent sensors are used and changing configurations provide better attitude estimation results to filtering techniques.
Keywords
