基于改进EEMD的GPS多路径误差建模与削弱方法研究

引入一种改进的集合经验模式分解方法（improved ensemble empirical mode decomposition，IEEMD）进行多路径误差的建模。与EEMD方法相比较，IEEMD方法可以有效地克服模态分量数目不一致和分解不彻底等问题，并将原始序列分解为不同尺度的模态分量。同时，考虑到不同模态分量中高斯白噪声的能量密度与平均周期之积为常数，设计一种自动选择尺度与重构的方法，用于模态分量的选择与重构，进而构建GPS多路径误差模型。在此基础上，采用恒星日滤波技术，进行邻近周日GPS坐标序列中多路径误差的实时削弱。实验结果表明，采用同样的尺度选择方法和恒星日滤波技术，使用IEEMD方法可以得到比EEMD方法精度更高的GPS坐标序列。

Research on GPS Multipath Error Modeling and Mitigation in View of Improved EEMD

In this article, a new method called IEEMD (improved ensemble empirical mode decomposition) is introduced to build the correction of multipath error. Compared to EEMD (ensemble empirical mode decomposition), the new method can effectively solve the problem of model mixing and incomplete decomposition. It can also decompose the original coordinate sequence into different scales of modal components. The product of Gaussian white noise energy density and the average period is constant. Taking this condition into account, a new method is designed for automatic selection of scale and reconstruction to build the multipath error model. We can use it to correct the later coordinate series by the first day’s multipath model by the reason of the strong correlation between two successive days. The experiment results show that IEEMD can get more accurate coordinate sequence than EEMD, based on the same way of scale selection and stellar day filtering. To a certain extent, this study solidifies the theoretical basis of GPS high-precision real-time dynamic deformation monitoring.