EM算法在广播星历计算卫星位置中的应用  

Application of EM algorithm to the calculation of the satellite position based on broadcast ephemeris

作　　者：肖琴琴[1] 宋迎春[1] 杜琨[1] 

机构地区：[1]中南大学地球科学与信息物理学院,湖南长沙410083

出　　处：《测绘工程》2013年第6期73-76,85,共5页Engineering of Surveying and Mapping

基　　金：国家自然科学基金资助项目(40874005)

摘　　要：在利用广播星历计算卫星坐标时,为了提高效率,经常会采用多项式来进行拟合。由于多方面的原因,可能会导致拟合点数据的精度不够或者拟合点的个数不够,继续用此数据进行拟合则会大大降低拟合结果的精度和可靠性。文中使用EM(expectation maximization)算法添加有益于多项式拟合的"潜在数据",即缺失数据的条件期望,有效提高缺失数据条件下卫星坐标的拟合精度和可靠性。A polynomial fitting method is often used in broadcast ephemeris to calculate the satellite coordinates in order to improve the efficiency.Due to various reasons,data precision of fitting point is insufficient or the number of fitting points is not enough.If the original data are used to fit the satellite positions,it will greatly reduce the fitting precision and reliability.Thus EM （expectation maximization） algorithm is used to add ＂potential data＂ benefitting the coordinate fitting,the conditional expectation of the missing data,which can effectively improve the fitting precision and reliability of satellite coordinates under the condition of the data missing.

关 键 词：广播星历 拟合法 切比雪夫 缺失数据 EM算法 

分 类 号：P228[天文地球—大地测量学与测量工程]