共查询到19条相似文献，搜索用时 156 毫秒
1.
基于经验模态分解的地震波特征提取的研究 总被引：2，自引：0，他引：2
本文原始数据为35个天然地震和27个人工爆破事件的离震中最近的5个台站的垂直分量波形数据.从原始波形数据分别提取最大振幅对应的周期、倒谱的方差、自相关函数的最大值3个特征.依据希尔伯特黄变换原理,用经验模态分解方法把原始波形信号分解为10个左右的本征模态函数分量后,再从每个分量中分别提取这3个特征.接着对所获取的特征样本集合采用随机划分法分为学习样本集与检验样本集,然后再通过支持向量机进行分类识别,如此反复进行多次样本划分和分类识别.结果表明经验模态分解后的分量信号提取的这3个特征具有更高的识别率,说明了经验模态分解有利于识别天然地震和人工爆破事件,值得进一步深入研究. 相似文献
2.
3.
完备总体经验模态分解(CEEMD)克服了经验模态分解(EMD)的模态混叠问题,依据信号自身的特点,将待分析的复杂信号分解为一系列不同尺度的固有模态函数(IMF)的子信号,且各IMF分量的频率由高到低依次排列,是一种适用于分析处理非线性非平稳信号的强大的信号分析技术.地震资料中的随机噪声一般属于高频率的信号,在CEEMD中往往分布在前几个高频IMF分量,本文针对基于CEEMD的分频去噪和基于CEEMD的小波阈值去噪等方法的不足,在前人基于EMD阈值去噪的基础上设计了自相关函数统计特性与CEEMD全局阈值联合去噪方法.该方法先对CEEMD分解的若干个模态分量进行自相关,寻找到噪声主导模态和信号主导模态,然后利用设计的全局阈值对噪声主导模态进行去噪,最后将处理后和未处理的固有模态函数进行重构,得到最终的去噪结果.模型试算和实际地震资料处理都验证了此方法在提高信噪比,保留原信号高频有效成分和弱信号信息上的有效性. 相似文献
4.
基于鹤岗地震台阵勘选数据,分析各子台记录的矿震和天然地震的频谱特征,从各子台频谱一致性角度对观测场地进行评估;采用希尔伯特-黄变换（Hilbert-Huang,简称HHT）方法,分析矿震和天然地震的时频特征。结果显示：勘选记录的地震频谱特征与已有研究一致,各子台特征也较一致,该勘选场地适合建设台阵;由经验模态分解（EMD）后的本征模态函数（IMF）分量可以明显识别出矿震面波,矿震和天然地震的HHT谱的时频特征有明显区别。 相似文献
5.
针对微震信号具有高噪声、突变快、随机性强等特点,基于经验模态分解(EMD)及独立成分分析(ICA)提出一种微震信号降噪方法.首先,对含噪信号进行EMD分解,获得一系列按频率从高到低的内蕴模态函数(IMF),利用原信号与各IMF之间的互相关系数辨识出噪声与信号的分界,将分界之上的高频噪声滤除;其次,为有效去除分界IMF中的模态混叠噪声,基于ICA算法对分界IMF进行盲源分离,提取其中的微震有效信号,并将其与剩余的IMF累加重构,从而得到降噪后的微震信号;最后,利用快速傅里叶变换(FFT)时频谱对比分析降噪前后的信号特征,定性说明本文方法的有效性;引入信噪比和降噪后信号占原信号的能量百分比两个参数,定量说明本文方法能充分保留微震信号的瞬态非平稳特征,降噪效果明显. 相似文献
6.
地震信号是典型的非平稳随机信号,容易受到场地条件和周边环境的影响,产生噪声干扰。采用自适应噪声完备集合经验模态分解（CEEMDAN）方法,对地震数据进行分析预处理,将地震信号分解成一系列具有不同特征时间尺度的固有模态函数（IMF）;对各IMF分量进行自相关计算,筛选含噪IMF分量,使用软阈值小波包方法进行去噪处理,与无噪分量进行重构,从而降低噪声干扰,提高数据质量。 相似文献
7.
8.
完备集合经验模态分解(CEEMD)通过添加正负成对辅助噪声可较好的解决集合经验模态分解(EEMD)中信号被噪声污染的问题,但CEEMD方法分解后的单个本征模态函数(IMF)分量中仍存在部分随机噪声信息.通过转变辅助噪声形式和分解流程提出自适应噪声完备集合经验模态分解(CEEMDAN)方法,该方法在较少集总次数和筛选迭代次数的情况下,即可实现优良的信噪分离功能,大大缩减处理耗时,具备分解精度高、具有完备性的特征.同时,针对传统经验模态分解(EMD)类方法去噪时直接舍弃第1～2阶高频IMF分量,导致其内高波数有效能量损失的问题,通过计算相邻IMF分量互信息熵获取高频噪声和低频有效信号的最优能量分界,对分界点前的各阶IMF分量进行同步压缩小波变换(SWT)处理,分离有效高频信息,最后与低频IMF分量重构达到噪声压制的目的.合成及实际地震资料处理结果表明,本文联合多步骤地震随机噪声压制策略具有较好的去噪效果和能量保持能力,在运算耗时指标上优于传统的EMD噪声辅助类方法. 相似文献
9.
在地震观测中,地震数据中普遍包含有噪声信号。由于噪声信号的干扰,地震分析的效率会受到不同程度的影响。传统的去噪方法通常需要噪声的先验知识,并且滤波时会造成部分有效信号丢失。针对这一问题,本文提出一种将自适应噪声完全集合经验模态分解（CEEMDAN）算法与Hurst指数相结合的地震数据去噪方法。首先通过CEEMDAN方法将信号分解为一系列本征模函数（IMF）,然后利用Hurst指数对滤波后的IMF分量进行识别,最后对地震数据IMF分量进行重构,从而实现数据去噪。与传统方法的去噪效果对比表明,本文方法可将低信噪比波形的去噪效果提高32%,将高信噪比波形的去噪效果提高6倍。同时对地磁数据的去噪结果表明,本文方法能够较完整地将地铁噪声从地磁信号波形中滤除。 相似文献
10.
希尔伯特—黄变换(Hilbert-Huang transform,HHT)是一种新的适合非平稳和非线性信号的分析方法,由于地震信号一般呈现出非平稳与非线性特性,因此HHT非常适合地震信号的分析。本文首先介绍了HHT中关于经验模态分解(Empirical Mode Decomposition,EMD)的实现过程,在此基础上分析了几种基于EMD获得本征模态函数(Intrinsic Mode Functions,IMF)来计算瞬时频率的算法,其中利用了两个采样间隔瞬时频率的平均来计算瞬时频率,较好地反映了地震信号频率成分随时间变化的特征。将该方法应用于四川东北部某地区海相碳酸盐岩地层三维地震叠后偏移数据处理,提取"三瞬"地震属性,与传统的希尔伯特变换提取的"三瞬"地震属性进行对比,结果表明基于HHT的"三瞬"地震属性结果具有更高的分辨率,IMF2的瞬时相位能够较好地刻画台地边缘生物礁相,IMF2的瞬时频率亦具有较好的分带性。将IMF2的"三瞬"地震属性与钻井等资料结合分析,能够更好地识别沉积相的分布。 相似文献
11.
Due to strong heterogeneity of marine carbonate reservoir, seismic signals become more complex, thus, it is very difficult for hydrocarbon detection. In hydrocarbon reservoir, there usually exist some changes in seismic wave energy and frequency. In their instantaneous spectrums there often exist such phenomena that show the characteristics of attenuation of high frequency energy and enhancement of low-frequency energy. The three EMD-based time-frequency analysis methods' instantaneous spectra all have certain oil and gas detection capability. In this paper, we introduced the Normalized Hilbert Transform (NHT) and a new method named the HU method for hydrocarbon detection. The model results in the Jingbian Gas Field which is located in the eastern Ordos Basin, China, show that NHT and HU methods can be adopted. They also detect the gas-bearing reservoir efficiently as the HHT method does. The three EMD-based methods, that is, the Hilbert–Huang transformation (HHT) and NHT and HU methods, were respectively applied to analyze the seismic data from the Jingbian Gas Field. Firstly, the seismic signals were decomposed into a finite number of intrinsic mode functions (IMFs) by empirical mode decomposition (EMD) method. The second IMF signal (IMF2) of the original seismic section better indicates the distribution of the reservoir. Information on hydrocarbon-bearing reservoir is mainly in IMF2. Secondly, the HHT, NHT and HU methods were respectively used to obtain different frequency division sections from IMF2. Hydrocarbon detection was realized from the energy distribution of the different frequency division sections with these three EMD-based methods. The practical application results show that the three EMD-based methods can all be employed to hydrocarbon detection. Frequency division section of IMF2 using NHT method was better for the seismic data from the Jingbian Gas Field than when using the HHT method and HU method. 相似文献
12.
经验模态分解(Empirical Mode Decomposition, EMD)是一种具有较大应用潜力的去噪算法.目前,该算法存在的一个较大问题是过渡内蕴模态函数(Intrinsic Mode Function, IMF)中混叠噪声不能有效处理.过渡内蕴模态函数中混叠噪声不易剔除,限制了该算法的应用.本文针对此问题,通过研究过渡IMF的特点,首次提出一种有效去除过渡IMF中混叠噪声的方法.该方法首先对原信号进行一次EMD处理,得到包含过渡IMF的初步去噪结果,并将其与合适的余弦信号结合,改变其包络分布,然后对其结果再次进行EMD处理,仿真实验表明该方法在保留有效信号的同时,可以有效的去除过渡IMF中混叠的噪声,并将该方法用于实际地震资料随机噪声压制,处理效果令人满意. 相似文献
13.
基于EMD的信号瞬时特征的小波分析方法 总被引：8，自引：1，他引：7
提出了一种基于EMD(Empirical Mode Decomposition)的信号瞬时特征的小波分析方法。用这种方法提取非平稳信号的瞬时频率和瞬时幅值分三个基本步骤：首先,用EMD把信号分解成IMF(Intrinsic Mode Function)分量;接着,对IMF分量进行小波分析,从小波系数的幅角函数中提取小波脊线;最后,从小波脊线中提取瞬时频率和瞬时幅值。通过对仿真信号的分析,验证了该方法能有效地分析非平稳信号。 相似文献
14.
This paper proposes a new multi-step prediction method of EMD-ELM (empirical mode decomposition-extreme learning machine) to achieve the short-term prediction of strong earthquake ground motions. Firstly, the acceleration time histories of near-fault ground motions with nonstationary property are decomposed into several components of intrinsic mode functions (IMFs) with different characteristic scales by the technique of EMD. Subsequently, the ELM method is utilized to predict the IMF components. Moreover, the predicted values of each IMF component are superimposed, and the short-term prediction of ground motions is attained with low error. The predicted results of near-fault acceleration records demonstrate that the EMD-ELM method can realize multi-step prediction of acceleration records with relatively high accuracy. Finally, the elastic and inelastic acceleration, velocity and displacement responses of single degree of freedom (SDOF) systems are also predicted with satisfactory accuracy by EMD-ELM method. 相似文献
15.
基于EMD与关联维的储层识别(英文) 总被引：1，自引：1，他引：0
在储层发育带,通常地震波波形会变复杂,所以关联维会变小。但由于地震波是有一定带宽的信号,波形受这个带宽内所有频率成分的影响,如果直接进行关联维的计算,其结果往往不能很好地反映储层。鉴于这种情况,提出了经验模态分解（EMD）与关联维相结合的方法,利用EMD对地震波进行分解,对分解后的每一个平稳的固有模函数（Intrinsic ModeFunction,IMF）分量进行关联维计算,将计算结果与有效的IMF分量进行储层预测。利用该方法对xx区三叠系中油组进行了试算,取得了较好的效果。 相似文献
16.
选取甘肃省测震台网测定的甘肃平凉地区塌陷地震、甘东南地区天然地震各50个,同地区的30个地震事件为待测事件。采用近年来计算机领域里较先进的图像识别方法——卷积神经网络识别两种地震事件类型,设定波形通道总数的80%为地震事件分类阈值,超过分类阈值的设定为“0”或“1”,即可判定该待测事件为对应的事件类型。将地震事件作为原始图像,提取图像中最具代表性特征点;接着提取图像特征进入池化层,池化层会对该图像特征点进行归类压缩,提取最具代表性的图像特征,最后输出识别图像。结果表明:塌陷地震最终分类准确识别率为86.7%,天然地震准确识别率为93.3%,总识别率为90%,为今后平凉地区塌陷地震事件类别识别工作提供了可靠的参考价值。 相似文献
17.
针对结构损伤检测中损伤的识别、定位以及程度的标定这三个独立并按一定先后顺序进行的检测过程,提出了一种能将以上三者同时进行的联合检测方法。该方法首先利用经验模态分解(EMD)方法将三层钢筋混凝土剪切型结构在各种损伤工况下的顶层地震作用加速度响应分解为若干固有模态函数(IMF)分量,然后以此IMF分量和未经EMD分解的原始加速度响应数据来构造损伤标识量,作为特征参数依次输入到径向基函数神经网络(RBFNN)中进行损伤检测。给出了应用此方法的具体步骤,通过仿真实验证明了利用该方法进行结构损伤一次检测的可行性和有效性,结果表明,由加速度响应经EMD分解而得到的IMF分量输入到RBFNN中能够更为精确地一次检测出结构所有损伤信息,并且RBFNN在结构损伤损度大时具有更好的检测效果。 相似文献
18.
In this work, we compare Fourier transform, wavelet transform, and empirical mode decomposition (EMD), and point out that EMD method decomposes complex signal into a series of component functions through curves of local mean value. Each of Intrinsic Mode Functions (IMFs — component functions) contains all the information on the original signal. Therefore, it is more suitable for the interface identification of logging sequence strata.Well logging data reflect rich geological information and belong to non-linear and non-stationary signals and EMD method can deal with non-stationary and non-linear signals very well. By selecting sensitive parameters combination that reflects the regional geological structure and lithology, the combined parameter can be decomposed through EMD method to study the correlation and the physical meaning of each intrinsic mode function. Meanwhile, it identifies the stratigraphy and cycle sequence perfectly and provides an effective signal treatment method for sequence interface. 相似文献
19.
DISCUSSION ON ATTENUATION CHARACTERISTICS,SITE RESPONSE AND MAGNITUDE DETERMINATION IN SICHUAN 
下载免费PDF全文
下载免费PDF全文 The attenuation characteristics and site response are calculated respectively for each individual tectonic unit in Sichuan (Sichuan Basin,west Sichuan plateau and Panzhihua-Xichang area),using digital waveform data recorded by regional seismic networks and relevant seismic phase data collected from China Seismograph Network.The frequency dependent Q(f) is obtained by the iterative grid-search technique described by Atkinson and Mereu based on trilinear geometrical spreading model.The source spectra are determined by the model of Brune and the site responses of seismic stations are derived by Moya's method using genetic algorithms.Comparison to conventional ML estimates shows that the network local magnitude bias is quite significant at low and intermediate magnitudes.The bias at the jth station for the ith event is defined as ΔMij=Mij-Mi, where ΔMij is the station magnitude and Mi the network-average value.For comparison,we mapped the spatial distribution of biases by digital seismograms recorded from 10535 earthquakes of magnitude 2.5≤ML≤4.9 that occurred in Sichuan from January 1,2009 to June 30,2015.Based on the above data,the attenuation characteristics,site response and their effects on magnitude determination in Sichuan are analyzed.Our results demonstrate that the associated model for regional quality factor for frequencies can be expressed as Q1(f)=450.6f0.513 4 for Sichuan Basin,Q2(f)=136.6f0.581 3 for west Sichuan Plateau and Q3(f)=101.9f0.666 3 for Panzhihua-Xichang area.Site response results indicate that different stations show different amplifications.Maps of biases appear to be different,but with similar dominant spatial distribution.For stations in Sichuan Basin,their greater magnitudes are functions of low attenuation in structure and amplification effects of both seismic stations and basin effects.For stations in west Sichuan Plateau,the possible causes of these lower magnitudes are severe dependence upon source region due to extreme lateral variations in either structure or path effect attenuation.For stations in Panzhihua-Xichang area,broken medium caused by strong tectonic activity or large earthquakes and heat flow up-welling along active faults may be the main reasons of low magnitude values when earthquakes occur in western Sichuan and eastern Tibetan region.And the greater magnitudes for earthquakes along the Longmen Mountains appear to be well correlated with edge effect of sedimentary basin on strong ground motion.In our study,stations magnitude biases appear to be extremely correlated with tectonic structures and different regions when seismic rays passing through,magnitudes are affected significantly by lateral variations in attenuation characters rather than site responses. 相似文献