基于EMD的改进型3δ粗差探测方法

Improved 3δgross error detection method based on EMD

针对全球卫星导航系统(GNSS)时间序列粗差多、分布随机的问题,提出一种基于经验模态分解(EMD)的改进型3δ粗差探测方法。新方法基本思想是:先利用3δ准则初步探测全球导航卫星系统(GNSS)时间序列粗差并插补,再将GNSS时间序列分解为若干时间尺度的固有模态函数(IMF)分量,再利用相关系数法信噪分离,最后利用3δ准则对残差序列进行粗差探测。利用模拟数据和实际工程数据对比EMD-3δ方法相对于常规LS(Least Squares)-3δ法的粗差探测效果。模拟实验数据表明,EMD-3δ法能够探测92.7%的粗差,LS-3δ方法仅能探测48.3%的粗差。来自斯克里普斯轨道和永久阵列中心(SOPAC)的GNSS时间序列数据进一步表明EMD-3δ能较传统方法更有效地探测粗差。

Aiming at the numerous gross errors in the deformation monitoring time series of the global navigation satellite system(GNSS)and their random distribution,an improved three-standard-deviation(3δ)gross error detection method based on empirical mode decomposition(EMD)is proposed.The basic idea of the EMD-3δmethod is as follows:firstly,the gross error of GNSS time series is detected based on the 3δcriterion and interpolated;secondly,the GNSS time series are decomposed into intrinsic mode function(IMF)components of several time scales;thirdly,the signal and noise are separated via a correlation coefficient;finally,the gross error of residual series is detected using the 3δcriterion.The simulation data and actual engineering data are used to compare the gross error detection effect of the proposed EMD-3δmethod with that of the least squares(LS)-3δmethod.The simulation experiment data show that the improved EMD-3δmethod can detect 92.7%of the gross error,while the LS-3δmethod can only detect 48.3%of the gross error.GNSS time series data from the Scripps Orbit and Permanent Array Center(SOPAC)further show that the EMD-3δmethod can detect gross errors more effectively than traditional methods.