中文摘要：

导航星表的性能对于星敏感器姿态测量的实时性及精度至关重要。为了克服星等过滤算法的缺点,将支持向量机应用于导航星表的构造算法中。将基本星表中的恒星视为待分类的数据点,选取抽样视场中最亮的k颗星作为导航星,而非导航星的数量由抽样视场中恒星的密度决定。为了获得具有最大推广能力的抽样数据,采用了一种球面螺旋形算法生成抽样视场视轴指向,使用抽样数据构建最优导航星分类器,应用最优导航星分类器对基本星表中每一颗恒星进行分类判决。仿真结果表明,在满足8°×8°视场中至少出现3颗导航星的条件下,该算法生成的导航星表导航星总数约为星等过滤算法的33%,比传统支持向量机算法减少了7.8%,其标准差仅为星等过滤算法的21%,这表明本算法在导航星表容量及导航星分布均匀性方面大大优于星等过滤算法和传统支持向量机算法。

英文摘要：

The performance of guide star catalog（GSC）is crucial to accuracy and real-time ability of star sensor＇s attitude measurement.The support vector machines（SVM）is applied in the construction of GSC to avoid the flaws of magnitude filtering method（MFM）.All stars of the original star catalog are considered as data points to be classified.The k brightest stars are selected as guide stars inside the field of view（FOV）.The number of non-guide stars is determined by the density inside the FOV.Furthermore,the sphere spiral algorithm is adopted to generate the sampling boresight directions by virtue of obtaining the sampling data holding maximal generalization capability.Next,the optimal guide star classifier is constructed and every star from the original star catalog is judged using it.Experiment demonstrates that the total number of guide stars generated by the proposed algorithm is approximately 33% and the standard deviation is 21% compared with those of MFM.Furthermore it is less than the traditional SVM method about 7.8%,under the condition that the number of guide star inside the FOV is not less than 3.Consequently,the proposed algorithm makes a great progress relative to MFM and the traditional SVM method in capacity and uniformity of GSC.