基于奇异谱分析的GNSS坐标时间序列粗差探测与噪声估计

Gross Error Detection and Noise Estimation of GNSS Coordinate Time Series Based on Singular Spectrum Analysis

考虑到传统谐波模型难以精确描述GNSS坐标时间序列的非线性变化,导致信号和噪声不能很好地分离,进一步影响粗差探测和噪声估计,本文提出一种基于奇异谱分析的粗差探测与噪声估计算法。首先采用奇异谱分析方法分离出GNSS坐标时间序列中的信号与噪声,然后基于IQR准则探测噪声中的粗差,最后采用最小二乘方差分量估计(LS_VCE)方法定量估计各噪声分量。算例表明,相比于传统基于谐波模型的算法,该算法的粗差探测准确率更高,且估计的噪声分量与真值更接近。

Considering that the traditional harmonic model has difficulty accurately describing the nonlinear variation of GNSS coordinate time series,the signal and noise cannot be separated well,which further affects the gross error detection and noise estimation.This paper proposes an algorithm for gross error detection and noise component estimation based on singular spectrum analysis(SSA).The basic idea of the proposed algorithm is to separate the signal and noise with the SSA firstly,and then detect gross error in noise based on the inter-quartile range(IQR)criterion.Finally,we employ the least squares variance component estimation(LS_VCE)to quantitatively estimate each noise component.The analysis results show that the success rate of gross error detection of the new algorithm is higher than that of the traditional algorithm and the noise component estimation derived by the new algorithm is closer to the true value compared with the traditional algorithm.