摘要
针对大坝变形监测GNSS时间序列数据中常见的噪声干扰问题,该文提出了一种基于改进的完备继承经验模态分解(ICEEMDAN)和色散熵的去噪方法。首先利用ICEEMDAN方法将形变数据分解为若干本征模态函数(IMF),然后通过计算各个IMF的色散熵值,识别并筛选出包含噪声的高频模态分量,最后应用小波变换对其中的噪声进行去除,并对经过去噪处理的模态与剩余的低频模态分量进行重构,生成去噪后的形变序列。采用模拟分析方法和实际GNSS大坝监测数据进行对比分析,并通过信噪比、相关系数和均方根误差等指标评估该方法与传统的小波变换及EMD去噪方法的性能差异。结果表明,该文提出的方法在去噪效果上明显优于现有方法,能够更有效地抑制噪声干扰,提升去噪效果。
Aiming at the common noise interference problems in the time series data of dam deformation monitoring,a denoising method based on improved complete inheritance empirical decomposition(ICEEMDAN)and dispersion entropy is proposed.First of all,the deformation data is decomposed into several intrinsic modal functions(IMF)using the ICEEMDAN method.Then,by calculating the dispersion entropy of each IMF,high-frequency modal components containing noise are identified and screened out.Finally,wavelet transform is applied to remove the noise,and the denoised modes and the remaining low-frequency modal components are reconstructed.to generate denoised deformation sequences.simulation data and actual GNSS dam monitoring data are used for comparative analysis,and the performance differences between this method and traditional wavelet transform and EMD denoising methods are evaluated through indicators such as signal-to-noise ratio,correlation coefficient and root mean square error.The results showed that the proposed method was obviously superior to the existing methods,which could suppress noise interference more effectively and improve the denoising effect.
作者
朱亦鹏
贾东振
何秀凤
ZHU Yipeng;JIA Dongzhen;HE Xiufeng(School of Earth Sciences and Engineering,Hohai University,Nanjing 21ll00,China)
出处
《测绘科学》
北大核心
2025年第6期64-70,共7页
Science of Surveying and Mapping
