一种CORS站高程序列非线性趋势项分析方法  

An analysis method of non-linear trend items in the high sequence of CORS stations

作　　者：王志勇 王崇倡[1] 刘宇晨 金泽林 焦亚沁 WANG Zhiyong;WANG Chongchang;LIU Yuchen;JIN Zelin;JIAO Yaqin(School of Geomatics,Liaoning Technical University,Fuxin,Liaoning 123000,China;Nanjing Institute of Geography and Limnology,Chinese Academy of Sciences,Nanjing 210000,China;School of Marine Technology and Geomatics,Jiangsu Ocean University,Lianyungang,Jiangsu 222000,China)

机构地区：[1]辽宁工程技术大学测绘与地理科学学院,辽宁阜新123000 [2]中国科学院南京地理与湖泊研究所,南京210000 [3]江苏海洋大学海洋技术与测绘学院,江苏连云港222000

出　　处：《测绘科学》2022年第4期44-51,共8页Science of Surveying and Mapping

基　　金：国家自然科学基金项目(41471351)。

摘　　要：针对传统线性模型通过拟合连续运行参考站(CORS)高程时间序列趋势项,使CORS站运动趋势单一化导致无法体现测站各自运动特性的问题,该文提出一种基于CORS高程时间序列本身的非线性趋势项定量识别与估计方法。首先,用K均值聚类整体经验模态分解方法对CORS站高程时间序列进行分解;其次,基于排列熵理论进行非线性趋势项的定量识别与合成;最后,采用正弦函数模型、线性多项式模型、傅里叶模型对非线性趋势项进行建模拟合并讨论拟合精度。实验结果表明:基于排列熵的非线性趋势项定量识别方法避免了人为判别误差,准确地反映了序列自身的变化趋势;3种模型中,傅里叶模型拟合效果最佳。According to the traditional linear model,by fitting the trend item of continuously operating reference stations(CORS)station elevation time series,the simplification of the movement trend of the CORS station leads to the problem that the respective movement characteristics of the station cannot be reflected,a method for quantitative identification and estimation of nonlinear trend items based on CORS elevation time series itself was proposed in this paper.Firstly,the K-means clustering overall empirical mode decomposition method was used to decompose the CORS station elevation time series.Secondly,the nonlinear trend terms were identified and composed quantitatively based on Permutation entropy theory.Finally,the sinusoidal function model,linear polynomial model,and Fourier model were used to simulate the nonlinear trend term and discuss the fitting accuracy.Experimental results showed that the quantitative identification method of nonlinear trend items based on Permutation entropy could avoid human-made discrimination errors and accurately reflect the changing trend of the sequence itself;among the three models,the Fourier model had best fitting effect.

关 键 词：CORS 非线性趋势项 K均值 排列熵 线性多项式模型 傅里叶模型 正弦函数模型 

分 类 号：P228[天文地球—大地测量学与测量工程]