基于支持向量机的导航星筛选方法

Selecting Method of Guide Stars Based on SVM

导航星表是天文导航中星敏感器用来实现星图识别和姿态确定的唯一依据,其容量、结构、存储等直接影响着识别的快速性和准确度。总结现有导航星表制定方法的优缺点,定义了表征天区恒星散布情况的特征量——稠密度,提出了以恒星视星等和稠密度为模式特征、基于支持向量机的导航星筛选方法。利用史密松天文台星表进行了仿真,制定了包含3 051颗恒星的导航星库,星库平均星等为5.09 Mv,98.10%的恒星星等在设定的星敏感器探测阈值以内。经蒙特卡洛枚举法验证,98.23%的随机模拟视场(10°×10°)内导航星达到3颗或3颗以上,满足局部天区星图识别条件。仿真算例表明,该方法能够兼顾恒星视星等和散布情况,其产生的导航星表满足星敏感器星图识别和姿态确定需要。

A guide star catalogue is the unique basis for star sensors to identify the stars and determine their attitudes in celestial navigation,and its capacity,structure and storage mode are vital to the rapidity and accuracy of identification. The advantages and disadvantages of the existing methods are analyzed. A new method based on support vector machine(SVM),which uses star magnitude and density as its characteristics,by defining star density to represent star scatter circumstance. The simulation on Smithsonian Astrophysical Observatory(SAO) star catalogue generates a guide star catalogue including 3 051 stars,in which mean star magnitude reaches 5. 09 Mv,and 98. 10% of the stars can be detected theoretically.Monte Carlo experiment demonstrates that 98. 23% of random simulation fields of views(10 × 10°) contain 3 guide stars or more,which can be used to identify the stars in part of field of view. The simulation shows that the proposed method which taking the star magnitude and scatter circumstance into consideration meets the needs for star sensors to identify the stars and determine their attitudes.