用CEEMDAN-ICA去除GNSS坐标时间序列噪声  

Remove GNSS Coordinate Time Series Noise with CEEMDAN-ICA

作　　者：范小猛 胡川 张重阳 李成洪 赵立都 FAN Xiaomeng;HU Chuan;ZHANG Chongyang;LI Chenghong;ZHAO Lidu(Tianjin Surteying and Design Instinte for Water Transport Engineering Co.Ltd,Tianjin 300456,China;School of Smart City,Chongqing Jiaotong Unitersity,Chongqing 400074,China)

机构地区：[1]天津水运工程勘察设计院有限公司,天津300456 [2]重庆交通大学智慧城市学院,重庆400074

出　　处：《测绘科学技术学报》2025年第1期15-20,共6页Journal of Geomatics Science and Technology

基　　金：国家重点研发计划项目(2021YFB2600600,2021YFB2600603);重庆市基础科学与前沿技术研究(一般)项目(cstc2017jcyjAX0102);重庆交通大学研究生科研创新项目(CYS21341)。

摘　　要：为了改善相关系数准则识别噪声不准确的问题,提出一种联合自适应噪声完备经验模态分解CEEMDAN(Complete Ensemble Empirical Mode Decomposition with Adaptive Noise)和独立成分分析ICA(Independ Component Analysis)的坐标时序降噪方法。首先对坐标时间序列进行CEEMDAN分解,根据相关系数准则得到高频分量;然后对其执行ICA分解,并根据排列熵剔除含噪声的独立分量重构坐标时间序列;最后通过模拟数据实验和实测数据实验,验证所提方法的有效性。实验结果表明,CEEMDAN-ICA可以很好地分离模拟数据中添加的噪声,降噪后数据的均方根误差相较于相关系数准则和EEMD(Ensemble Empirical Mode Decomposition)-ICA分别减少28.4%和18.8%,信噪比分别提高18.5%和8%;对于坐标时间序列,CEEMDAN-ICA降噪结果在N、E和U方向上均方根误差均最小,平均达到1.44、1.27和2.92 mm;信噪比最大,平均达到11.13、12.54和15.78 dB。In order to improve the problem of inaccurate noise identification by the correlation coefficient criterion,a coordinate time series noise reduction method combining complete ensemble empirical mode decomposition with adaptive noise(CEEMDAN)and independent component analysis(ICA)is proposed.Firstly,the coordinate time series is decomposed by CEEMDAN,and the high frequency components are obtained according to the correlation coefficient criterion.Then the ICA decomposition is performed and the independent components containing noise are removed according to the permutation entropy.Finally,the coordinate time series is reconstructed by summing the remaining components.The validity of the proposed method is verified by simulation data and actual data experiment.The experimental results show that the noise added in the simulated data can be well separated by CEEMDAN-ICA.Compared with the results of correlation coefficient criterion and EEMD(ensemble empirical mode decomposition)-ICA method,the root-mean-square error of the denoised data are reduced by 28.4%and 18.8%respectively,and the signal-to-noise ratio are increased by 18.5%and 8%respectively.For coordinate time series,the root-mean-square error of CEEMDAN-ICA denoising results in N,E and U directions are the smallest,with an average of 1.44,1.27 and 2.92 mm.The signal-to-noise ratio are the largest,with an average of 11.13,12.54 and 15.78 dB.

关 键 词：GNSS坐标时间序列 相关系数准则 自适应噪声完备经验模态分解 独立成分分析 排列熵 

分 类 号：P228[天文地球—大地测量学与测量工程]