TLS方法解算时间序列模型及其在卫星钟差预报中应用.pdfVIP

Times Series Model Solved by TLS Method and Applications in Prediction of Satellite Clock Error Zhao Jun 1, Gui Qing-ming 2, Chen Chen 3 1. Institute of Surveying and Mapping, Information Engineering University, Zhengzhou China 2. Institute of Science, Information Engineering University, Zhengzhou China 3 ．Institute of Navigation and Aerospace Engineering, Information Engineering University, Zhengzhou China 1. zhaojun4368@163.com, 2.guiqingming@126.com, 3. Gzxbcc@163.com Abstract: The satellite clock error is an important error factor to affect precision of GNSS navigation and positioning, which has become a focused research topic in GNSS navigation. As a common method for predicting, time series analyses have been applied to prediction of satellite clock error successfully. If least-squares (LS) method is used to obtain the estimator of unknown parameter for predicted model, it can be considered as errors-in-variable (EIV) model because the elements of coefficient matrix are consists of observations. If the errors of observations in designed matrix are not taken into consideration, unknown parameter estimator may be affected. Furthermore, the precision of predicting is probably destroyed. As the parameter estimation method of EIV model, total least-squares (TLS) method is more proper to adopt this problem. Given that the calculating model of time series is a structured TLS problem, the partial EIV will be employed and high efficient algorithm is developed. At last, combining with precise ephemeris provided by IGS, the clock data is considered as the time series. Then the TLS method is employed to obtain the unknown parameter