导航卫星原子钟Kalman滤波中噪声方差-协方差的确定  

Determination of Covariance Matrix of Kalman Filter Used for Time Prediction of Atomic Clocks of Navigation Satellites

作　　者：郭海荣[1,2] 杨元喜[3] 何海波 徐天河[3] 

机构地区：[1]北京市5136信箱,北京100094 [2]信息工程大学测绘学院,河南郑州450052 [3]西安测绘研究所,陕西西安710054

出　　处：《测绘学报》2010年第2期146-150,共5页Acta Geodaetica et Cartographica Sinica

基　　金：国家自然科学基金(40474001,40274002,40604003)

摘　　要：用Kalman滤波实时监测卫星钟运行状况,需要确定Kalman滤波协方差阵。首先介绍适用于铷钟的Kalman滤波方程,推导基于哈达玛总方差的Kalman滤波过程噪声参数和观测噪声估计方法,并在此基础上构造Kalman滤波状态噪声协方差阵和观测噪声协方差阵。利用GPSBlockIIR铷钟事后处理的采样间隔为5min的精密钟差数据进行了预报分析,结果表明,当预报时间为1h时,预报精度在1ns左右;当预报时间为6h时,预报精度在(8～9)ns之间。进一步验证Kalman滤波协方差阵估计方法的正确性。In order to manage the operational problems of satellite clocks using Kalman filter,its covariance matrixes need to be determined.A Kalman filter equation for Rubidium clocks was presented at first.A method for estimating the process noises and the observation noises of Kalman filter using total Hadamard variance was deduced.And then the prediction covariance represented by a function of the Kalman Filter process noises and the observation covariance was constructed.The precision of prediction was analyzed using the 5-min precise clock data of GPS Block IIR rubidium clocks.The results show that：① the prediction precision is 1 ns or so when prediction time is within one hour;② the prediction precision is（8~9） ns when prediction time is within 6 hours.Then the correctness is further verified.

关 键 词：卫星导航 原子钟 KALMAN滤波 哈达玛总方差 

分 类 号：P228[天文地球—大地测量学与测量工程]