摘要
为了提高变形监测数据的去噪精度及可靠性,基于变分模态分解(variational mode decomposition,VMD)构建一种新的变形监测数据去噪方法。首先,建立VMD高频噪声分量判定标准,引入T指标用于确定VMD去噪的最优K值。然后,将剔除高频噪声后的VMD分量进行叠加重构,建立VMD变形监测数据去噪方法。最后,通过仿真信号、桥梁、大坝变形监测数据去噪实例,对比分析VMD、小波及经验模态分解(empirical mode decomposition,EMD)去噪方法。实验结果表明,VMD对仿真信号去噪的相关系数、均方根误差、信噪比等指标均较大程度上优于小波及EMD去噪方法,理论上证实了VMD去噪方法的有效性及可靠性;VMD对桥梁、大坝变形监测数据去噪的结果比小波、EMD具有更好的精度及光滑性,同时较好地保留了局部变形特征信息。
In order to improve the denoising accuracy and reliability of deformation monitoring data, a new denoising algorithm for deformed data is constructed based on variational mode decomposition(VMD).Firstly, the criterion for judging the high frequency noise component of VMD is established, and T index is introduced to determine the optimal K value of VMD denoising. Then, VMD component after eliminating high frequency noise is reconstructed, and the denoising method of VMD deformation data is established.Finally, the denoising methods of VMD, wavelet and empirical mode decomposition(EMD) are compared and analyzed through the examples of simulation signal, bridge deformation data and dam deformation data.The experimental results show that the correlation coefficient, root mean square error and signal-to-noise ratio of VMD are better than those of wavelet and EMD. Therefore, the validity and reliability of VMD denoising method are proved theoretically. When denoising bridge deformation data and dam deformation data, VMD denoising results have better denoising accuracy and smoothness than wavelet and EMD, while retaining the local deformation feature information.
作者
罗亦泳
黄城
张静影
LUO Yiyong;HUANG Cheng;ZHANG Jingying(School of Geodesy and Geomatics Engineering,East China University of Technology,Nanchang 330013,China;School of Geodesy and Geomatics,Wuhan University,Wuhan 430079,China)
出处
《武汉大学学报（信息科学版）》
EI
CSCD
北大核心
2020年第5期784-790,共7页
Geomatics and Information Science of Wuhan University
基金
国家自然科学基金(41861058,41664001)
江西省数字国土重点实验室开放研究基金(DLLJ201612)。
关键词
变分模态分解
变形监测数据
去噪
精度分析
variational mode decomposition
deformation monitoring data
denoising
precision analysis
