摘要
针对整体经验模态分解(EEMD)方法对连续运行参考站(CORS)高程时间序列分解结果整体平均时,噪声未能完全消除的问题,该文引入一种噪声自适应整体经验模态分解(EEMDAN)方法,在经验模态分解(EMD)处理时添加白噪声分解的模态分量,改进CORS站高程时间序列信号分解的模态混叠,根据平均周期和能量密度乘积指标区分噪声和周期信号,并基于正交性指数及均方根误差指标衡量分解的精度。两个CORS站20年高程时间序列实验结果表明,以正交指数衡量的分解精度EEMDAN比EEMD提高近35%,均方根误差指标提高近22%,实现了对CORS站高程时间序列的精确分解,并改善EEMD整体平均后噪声未完全消除对信号重构的影响。
According to the fact that the noise can not be completely eliminated when the ensemble empirical mode decomposition(EEMD)method decomposes the result of time series,in this paper,a noise adaptive(EEMD of adaptive noise)method was introduced,the modal component of white noise decomposition was added to EEMD processing to improve the modal aliasing of continuously operating reference stations(CORS)elevation time series signal decomposition,the noise and periodic signals were distinguished according to the product of average cycle and energy density,and the accuracy of decomposition was measured based on the index of orthogonality and the root mean square error index.The results of the twenty year elevation time series experiment of two CORS stations showed that the decomposition precision of the index of orthogonality was nearly 35%higher than that of the EEMD,and the mean square root error index was increased by nearly 22%,and the effect of the noise on the signal reconstruction was not eliminated after the overall mean of EEMD was improved.
作者
张恒璟
龙安森
文汉江
ZHANG Hengjing;LONG Ansen;WEN Hanjiang(School of Geomatics,Liaoning Technical University,Fuxin,Liaoning 123000,China;Chinese Academy of Surveying and Mapping,Beijing 100036,China)
出处
《测绘科学》
CSCD
北大核心
2020年第2期29-34,共6页
Science of Surveying and Mapping
基金
高分遥感测绘应用示范系统(一期)
辽宁省教育厅辽宁省高等学校基本科研项目(LJ2017QL008)
辽宁工程技术大学博士启动基金项目(13-1120)
中国测绘科学研究院基本科研业务费项目(AR1918).
作者简介
张恒璟(1982—),男,河南邓州人,副教授,博士,主要研究方向为空间大地测量数据处理与GNSS高程非线性运动。E-mail:sun_winter2009@163.com
