卫星钟差异常值探测的Bayesian方法

Bayesian Method of Satellite Clock Bias Outliers Detection

顾及钟差的物理特性,提出了一种新的卫星钟差时间序列异常值探测方法.利用二次多项式模型将卫星钟差分解为钟差、钟速、钟漂三个物理意义明确的分量,然后对每个分量通过ARIMA模型异常值探测的Bayesian方法进行异常值探测与估计.最后,采用IGS钟差数据进行实验,验证了该方法的有效性.

Bayesian method of satellite clock bias outliers detection is presented taking into account the physical characteristics of satellite clock bias.Then the satellite clock bias is divided into three independent components contained physical meanings,which are clock deviation,clock speed and clock frequency drift.And the outliers is detected and estimated by the Bayesian method of ARIMA model outliers detection.At last,the IGS examples illustrate that the new algorithm is effective for handling satellite clock bias outliers.