顾及有色噪声的改进EEMD站坐标时间序列降噪方法  

An Improved EEMD Station Coordinate Time Series Noise Reduction Method Considering Colored Noise

作　　者：贺小星 王海城[3] 聂启祥[2] HE Xiaoxing;WANG Haicheng;NIE Qixiang(College of Architecture and Surveying Engineering,Jiangxi University of Science and Technology,Ganzhou 341000,China;School of Civil Engineer and Architecture,East China Jiaotong University,Nanchang 330013,China;Hebei Water Conservancy and Hydropower Survey,Design and Research Institute Group Co.,Ltd.,Tianjin 300250,China)

机构地区：[1]江西理工大学土木与测绘工程学院,江西赣州341000 [2]华东交通大学土木建筑学院,江西南昌330013 [3]河北省水利水电勘测设计研究院集团有限公司,天津300250

出　　处：《测绘地理信息》2023年第6期40-45,共6页Journal of Geomatics

基　　金：国家自然科学基金(42104023);江西理工大学高层次人才科研启动(205200100564,205200100588);江西省教育科学规划课题(21YB072);江西省教育厅科学技术研究项目(GJJ200639);江西省自然科学基金(20202BABL213033);江西省数字国土重点实验室开放基金(DLLJ201905)。

摘　　要：针对全球导航卫星系统(global navigation satellite system,GNSS)站坐标时间序列信噪不易分离的问题,提出一种基于整体经验模态分解方法(ensemble empirical mode decomposition,EEMD)方法与相关系数准则的降噪新方法。该方法通过EEMD分解获取本征模态分量(intrinsic mode function,IMF)分量,再以相关系数准则进行高频噪声与低频信号的分离,可有效抑制传统EMD方法中的模态混叠问题,获得更精确的结果。以8个不同噪声、信号模型的仿真信号及5个实测站坐标时间序列进行实验分析,通过多种指标进行精度评估,结果表明:与传统EMD方法相比,基于EEMD与相关系数准则的方法能够获得更精确的坐标时间序列,且不受有色噪声模型影响,稳定性高。Aiming at the problem of difficult separation of sig⁃nal and noise for GNSS station coordinate time series,a new noise reduction method based on EEMD method and correla⁃tion coefficient criterion is proposed.This method obtains the IMF component through EEMD decomposition,and then sep⁃arates the high-frequency noise from the low-frequency signal according to the correlation coefficient criterion,which can ef⁃fectively reduce the mode mixing problem in the traditional EMD method and obtain more accurate results.The simula⁃tion signals of 8 different noise and signal models and the coor⁃dinate time series of 5 measured stations are used for experi⁃mental analysis,and the accuracy is evaluated through multi⁃ple indicators.The results show that compared with the tradi⁃tional EMD method,the method based on the EEMD and cor⁃relation coefficient criterion can obtain a more accurate coordi⁃nate time series,is not affected by the colored noise model,and has high stability.

关 键 词：全球卫星导航系统 坐标时间序列 整体经验模态分解 降噪分析 相关系数 

分 类 号：P228[天文地球—大地测量学与测量工程]