摘要
局部均值分解方法降噪过于粗糙,将认定为噪声的乘积函数(PF)分量直接剔除,导致有用信息丢失。为了有效提取GNSS站坐标时间序列的有用信息,该文提出一种局部均值分解和小波阈值相结合的降噪方法。通过局部均值分解将坐标时间序列分解为一系列PF分量和余项,依据消除趋势波动分析方法计算各PF分量的Hurst指数,利用小波阈值提取H≤1的PF分量中的有用信息,将提取出的信息与剩余PF分量叠加重构获得最终降噪的坐标时间序列。通过对5个测站的坐标时间序列进行实验,结果表明局部均值分解和小波阈值相结合的方法能够有效提取噪声分量中的有用信息,信噪比提高了27.8%,从而验证了该方法的有效性。
The local mean decomposition method is too rough to reduce noise, and the product function(PF) components identified as noise are directly removed, resulting in the loss of useful information. In order to effectively extract the useful information of GNSS station coordinate time series, a method of noise reduction combining local mean decomposition with wavelet threshold was proposed. The coordinate time series was decomposed into a series of PF components and residual by local mean decomposition, Hurst index of each PF component was calculated according to detrended fluctuation analysis. The useful information in the H≤1 PF components extracted by the wavelet threshold, the extracted information was superimposed and reconstructed with the remaining PF components to obtain the final noise-reduced coordinate time series. Through experiments on the coordinate time series of 5 stations,the results showed that combined local mean decomposition with wavelet threshold could effectively extract useful information from the noise component.The validity of the method was verified,and signal-to-noise ratio was increased by27.8%.
作者
邱小梦
陶国强
王奉伟
周世健
QIU Xiaomeng;TAO Guoqiang;WANG Fengwei;ZHOU Shijian(Yangtze River College,East China University of Technology,Fuzhou,Jiangxi 334000,China;Faculty of Geomatics,East China University of Technology,Nanchang 330013,China;College of Survcying and Gco-informatics,Tongji University,Shanghai 200092,China;Nanchang Hangkong University,Nanchang 330063.China)
出处
《测绘科学》
CSCD
北大核心
2021年第8期28-32,48,共6页
Science of Surveying and Mapping
基金
江西省教育厅科技项目(181523)
东华理工大学长江学院院长基金项目。
