EMD-ICA滤波降噪法及其在GPS多路径效应中的应用

发布时间：2018-02-04 05:43

本文关键词： 经验模态分解(EMD) 独立分量分析(ICA) 滤波降噪 GPS多路径效应模态相关斜率绝对值　出处：《东华理工大学》2017年硕士论文　论文类型：学位论文

【摘要】：在GPS测量中,例如:大桥桥面监测、水库大坝监测以及高楼监测等,测量结果往往会有一定的偏差。上述GPS测量通常是短基线测量并且采用双频GPS接收机,因此电离层误差以及对流层误差可以采用差分方法消除,而GPS多路径误差及随机高频噪声误差却难以消除。由于GPS多路径误差不存在空间相关性以及高频随机噪声误差不具备规律性,因此削弱这些误差能够提高GPS短基线测量精度。本文针对高频噪声消除以及GPS多路径效应误差改正进行了相关分析,本文主要分析的内容及方法如下:1)基于独立分量分析(Independent Component Analysis,ICA)与经验模态分解(Empirical Mode Decomposition,EMD)算法,针对EMD与ICA滤波进行详细分析,提出将经验模态分解与独立分量分析联合进行滤波降噪,以下简称EMD—ICA滤波降噪法。2)通过振动台实验数据探讨天线发生不同位移时,动态GPS多路径效应表现特征。采用约1h的观测数据,并通过其频谱图分析动态GPS多路径效应与静态GPS多路径效应的关系。3)针对GPS多路径效应频带分布较宽以及快速ICA算法分离信息的能力取决于参考信息的准确性,本文引入模态相关准则,提出一种基于模态相关的EMD-ICA滤波降噪法。通过实验将其与基于模态相关的EMD法及基于固有模态函数(IMF)频谱分析的ICA法进行对比分析,实验表明该方法效果更好。4)针对信噪比过低,提出一种附加斜率绝对值法的模态相关EMD-ICA滤波降噪法,主要思想是通过能量大小进行判别。该方法与小波阀值联合EMD滤波去噪类似,均为高频分量中的有效信息。通过实验将该方法与基于模态相关EMD-ICA法及基于模态相关EMD法进行对比分析,实验表明该方法能够有效提取出高频分量中的低频信息并且效果优于其他两种方法。
[Abstract]:In GPS measurement, such as: bridge monitoring, monitoring and monitoring of dam building, the measurement results tend to have a certain deviation. The GPS is usually measured by short baseline measurement and dual frequency GPS receiver, so the ionospheric error and troposphere error can be used to eliminate the difference method, but it is difficult to eliminate GPS multipath error the high frequency noise and random error due to multipath error. GPS there is no spatial correlation and high frequency random noise error has no regularity, thus weakening these errors can improve the GPS short baseline measurement accuracy. According to the high frequency noise correction is analyzed and the elimination of GPS multipath error, this paper mainly analyzes the contents and methods are as follows: 1 independent component analysis (Independent) based on Component Analysis, ICA) and empirical mode decomposition (Empirical Mode Decomposition EMD) algorithm for E MD and ICA filter are analyzed in detail, put forward the empirical mode decomposition and independent component analysis combined filter, hereinafter referred to as EMD - ICA.2) to investigate the denoising methods through the shaking table experimental data of different antenna displacement, dynamic GPS multipath characteristics. Using the observation data about 1H, and through the spectrum figure.3 analysis the relationship between dynamic GPS multipath and static GPS multipath effects) for accuracy of GPS multipath frequency distribution is relatively wide and fast ICA algorithm to separate information depends on the ability of reference information, this paper introduces the mode correlation criterion, is proposed based on EMD-ICA modal correlation denoising methods. Through the experiment based on EMD method and the modal correlation and intrinsic mode function (IMF) based on ICA method for spectrum analysis comparative analysis, experimental results show that this method has better effect in low SNR of.4), is proposed An additional slope absolute value modal correlation EMD-ICA filtering method, the main idea is to determine the energy size. This method and wavelet threshold denoising is similar with EMD, were valid information in the high frequency component. Through experiments, the method with the EMD-ICA method based on modal correlation and comparative analysis of the modal EMD method based on the experimental results show that this method can effectively extract the high frequency component in the frequency information and the effect is better than the other two methods.

【学位授予单位】：东华理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：P228.4

【参考文献】