CN105717543B - A kind of noise drawing method and system - Google Patents

A kind of noise drawing method and system Download PDF

Info

Publication number
CN105717543B
CN105717543B CN201610048523.2A CN201610048523A CN105717543B CN 105717543 B CN105717543 B CN 105717543B CN 201610048523 A CN201610048523 A CN 201610048523A CN 105717543 B CN105717543 B CN 105717543B
Authority
CN
China
Prior art keywords
signals
vector
signal
relational expression
observed
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201610048523.2A
Other languages
Chinese (zh)
Other versions
CN105717543A (en
Inventor
高亚力
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Inspur Beijing Electronic Information Industry Co Ltd
Original Assignee
Inspur Beijing Electronic Information Industry Co Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Inspur Beijing Electronic Information Industry Co Ltd filed Critical Inspur Beijing Electronic Information Industry Co Ltd
Priority to CN201610048523.2A priority Critical patent/CN105717543B/en
Publication of CN105717543A publication Critical patent/CN105717543A/en
Application granted granted Critical
Publication of CN105717543B publication Critical patent/CN105717543B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
    • G01V1/00—Seismology; Seismic or acoustic prospecting or detecting
    • G01V1/28—Processing seismic data, e.g. for interpretation or for event detection
    • G01V1/36—Effecting static or dynamic corrections on records, e.g. correcting spread; Correlating seismic signals; Eliminating effects of unwanted energy

Landscapes

  • Engineering & Computer Science (AREA)
  • Remote Sensing (AREA)
  • Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Acoustics & Sound (AREA)
  • Environmental & Geological Engineering (AREA)
  • Geology (AREA)
  • General Life Sciences & Earth Sciences (AREA)
  • General Physics & Mathematics (AREA)
  • Geophysics (AREA)
  • Indication And Recording Devices For Special Purposes And Tariff Metering Devices (AREA)
  • Radar Systems Or Details Thereof (AREA)

Abstract

本发明公开了一种噪声压制方法，包括获取M个检波器接收到的M个观测信号；其中，每个观测信号由相互独立的N个源信号组成；M和N均为正整数；对M个观测信号进行加性白噪声去除处理，得到去除加性白噪声之后的M个观测信号；对去除加性白噪声之后的M个观测信号进行独立分量分析处理，得到相互分离的N个源信号；从相互分离的N个源信号中去除噪声源信号，得到有效信号。该方法不仅能够去除噪声源信号，还能够去除观测信号中的加性白噪声，去除噪声的效率高；且能够避免加性白噪声对独立分量分析处理造成影响，准确性高。本发明还公开了一种噪声压制系统。

The invention discloses a method for suppressing noise, which includes acquiring M observation signals received by M detectors; wherein, each observation signal is composed of mutually independent N source signals; both M and N are positive integers; for M Additive white noise removal processing is performed on the observation signals to obtain M observation signals after the additive white noise removal; the independent component analysis is performed on the M observation signals after the additive white noise removal, and N source signals separated from each other are obtained. ; Remove the noise source signal from the separated N source signals to obtain an effective signal. The method can not only remove the noise source signal, but also remove the additive white noise in the observation signal, and the noise removal efficiency is high; and it can avoid the influence of the additive white noise on the independent component analysis processing, and the accuracy is high. The invention also discloses a noise suppression system.

Description

Noise suppression method and system
Technical Field
The invention relates to the technical field of seismic exploration, in particular to a noise suppression method and a noise suppression system.
Background
As the target of seismic exploration gradually shifts to the deep underground region and the region with complex earth surface, higher requirements are put forward on the seismic data processing technology. In seismic exploration, due to the influence of a large number of high-voltage transmission lines around the acquisition environment, observation signals received by a ground detector not only comprise effective waves, but also comprise a plurality of alternating current interference waves and various noises, so that the accuracy of the seismic exploration is seriously influenced, and therefore the alternating current interference waves and the noise signals need to be filtered.
Since the effective wave and the ac interference wave in the observation signal received by the ground detector are mutually independent source signals, an ICA (independent component analysis) algorithm is usually adopted to separate the observation signal, so as to remove the ac interference wave. However, the existing ICA algorithm can not remove additive white noise in an observed signal, and the efficiency of removing the noise is low; the ICA algorithm is based on the premise that "noise (additive white noise, etc.) is negligible", and actually acquired observation signals have additive white noise, which affects the ICA algorithm, so that the result is inaccurate.
Therefore, how to provide a noise suppression method and system with high noise removal efficiency and high accuracy is a problem that needs to be solved by those skilled in the art.
Disclosure of Invention
The invention aims to provide a noise suppression method and a system, which can remove noise source signals, can also remove additive white noise in observation signals and have high noise removal efficiency; and the influence of additive white noise on the analysis and the processing of the independent component can be avoided, and the accuracy is high.
In order to solve the above technical problem, the present invention provides a noise suppression method, including:
acquiring M observation signals received by M detectors; wherein each of the observation signals is composed of N source signals independent of each other; m and N are positive integers;
performing additive white noise removal processing on the M observation signals to obtain M observation signals after the additive white noise is removed;
performing independent component analysis processing on the M observation signals from which the additive white noise is removed to obtain N source signals which are separated from each other;
and removing the noise source signals from the N source signals which are separated from each other to obtain effective signals.
Preferably, the performing the additive white noise removal processing on the M observation signals specifically includes:
and performing additive white noise removal processing on the M observation signals by adopting a two-step eigenvalue decomposition method.
Preferably, the process of performing additive white noise removal processing on the M observation signals by using the two-step eigenvalue decomposition method specifically includes:
step s 201: obtaining an observation signal vector X (k) from M observation signals, wherein X (k) is [ X ]1,x2,……,xM]T，x1、x2、……、xMRespectively carrying out sampling processing on the M observation signals to obtain corresponding vectors;
step s 202: centering the observation signal vector X (k) to make the average value of the observation signal vector X (k) be 0;
step s 203: based on the observed signal vector X (k) and the correlation matrix relationObtaining a correlation matrixWherein the correlation matrix relation is:
step s 204: according to the correlation matrixAnd obtaining the characteristic value Lambda by the first decomposition relationX；
Wherein the first decomposition relation is:
wherein, ΛX＝diag{λ1，λ2，……，λN}；
Step s 205: pre-whitening the observation signal vector X (k) according to a pre-whitening relational expression to obtain a pre-whitened vector Z (k);
wherein the pre-whitening relational expression is:
wherein,VSis ΛSA corresponding feature vector;
step s 206: obtaining a diagonal matrix according to a preset time lag value p and a second decomposition relation
ΣX(ii) a Wherein the second decomposition relation is:
wherein, UXFor the feature matrix, p is not equal to 0;
step s 207: judging the diagonal matrix ΣXIf all singular values in the data are different, if yes, the step s208 is carried out; otherwise, returning to the step s206, and presetting different time lag values p;
step s 208: according to the pre-whitening vector Z (k) and the feature matrix UXAnd obtaining the observation signal vector X after the additive white noise is removed by the pre-whitening relational expression1(k)；
Wherein the observation signal vector X1(k) The method comprises M signal vectors, wherein the M signal vectors respectively correspond to the M observed signals after the additive white noise is removed.
Preferably, the time lag value p takes 1.
Preferably, the process of performing independent component analysis processing on the M observed signals from which the additive white noise is removed to obtain N source signals separated from each other specifically includes:
step s 401: setting the initial value of the iteration times q as 1;
step s 402: randomly generating an initial weight vector Wq；
Step s 403: vector X of the observed signal1(k) And the initial weight vector WqCarrying in a target relational expression, wherein the obtained result is a target weight vector W;
wherein the target relation is:
wherein E is the mean value operation, g is the nonlinear function;
step s 404: judging whether the target weight vector W is converged, if so, entering the step s405, otherwise, returning to the step s 403;
step s 405: adding 1 to the iteration number q, judging whether the iteration number q is greater than a preset iteration number threshold value, and if so, entering a step s 406; otherwise, return to step s 402;
step s 406: combining the target weight vector W and the observed signal vector X1(k) And substituting a separation relation to obtain a separation vector Y (k), wherein the separation relation is as follows:
Y(k)＝WX1(k)
the separation vector Y (k) includes N source signal vectors separated from each other, and the N source signal vectors correspond to the N source signals, respectively.
Preferably, the non-linear function g [. cndot. ] may be:
g (y) tanh (ay) or g (y) yexp (-y)2Y is/2) or g (y)3Wherein a is more than or equal to 1 and less than or equal to 2.
In order to solve the above technical problem, the present invention further provides a noise suppression system, including:
the acquisition unit is used for acquiring M observation signals received by the M detectors; wherein each of the observation signals is composed of N source signals independent of each other; m and N are positive integers;
the noise removing unit is used for performing additive white noise removing processing on the M observation signals to obtain M observation signals after the additive white noise is removed;
an independent component analysis processing unit, configured to perform independent component analysis processing on the M observation signals from which the additive white noise is removed, to obtain N source signals that are separated from each other;
and an effective signal obtaining unit configured to remove a noise source signal from the N source signals separated from each other to obtain an effective signal.
The invention provides a noise suppression method and a system, which improve the traditional ICA algorithm. The method comprises the steps of firstly carrying out additive white noise removal processing on M observation signals received by M detectors to remove additive white noise in the observation signals, then carrying out independent component analysis processing on the M observation signals to separate and remove noise source signals in the observation signals, and thus obtaining effective signals. Namely, the invention not only can remove the noise source signal, but also can remove the additive white noise in the observation signal, and the efficiency of removing the noise is high; and the independent component analysis processing is carried out after the additive white noise is removed, so that the premise of 'noise neglect is not counted' in the independent component analysis processing is met, the influence of the additive white noise on the independent component analysis processing is avoided, and the accuracy of the result is improved.
Drawings
In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings needed in the prior art and the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.
FIG. 1 is a flow chart of a process of a noise suppression method provided by the present invention;
FIG. 2 is a schematic diagram of a source signal composition in a noise suppression method according to the present invention;
FIG. 3 is a schematic diagram of an observed signal composition in a noise suppression method according to the present invention;
fig. 4 is a schematic diagram of a separation signal corresponding to a separation vector Y (k) in a noise suppression method according to the present invention;
FIG. 5 is a schematic frequency spectrum diagram of a source signal in a noise suppression method according to the present invention;
FIG. 6 is a schematic frequency spectrum diagram of a separated signal in a noise suppression method according to the present invention;
fig. 7 is a schematic structural diagram of a noise suppression system according to the present invention.
Detailed Description
The core of the invention is to provide a noise suppression method and a system, which can remove noise source signals, can also remove additive white noise in observation signals, and has high noise removal efficiency; and the influence of additive white noise on the analysis and the processing of the independent component can be avoided, and the accuracy is high.
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
The invention provides a noise suppression method, and as shown in fig. 1, fig. 1 is a flow chart of a process of the noise suppression method provided by the invention; the method comprises the following steps:
step s 101: acquiring M observation signals received by M detectors; wherein each observation signal is composed of N source signals which are independent of each other; m and N are positive integers;
step s 102: performing additive white noise removal processing on the M observation signals to obtain M observation signals after the additive white noise is removed;
step s 103: carrying out independent component analysis processing on the M observation signals from which the additive white noise is removed to obtain N source signals which are separated from each other;
step s 104: and removing the noise source signals from the N source signals which are separated from each other to obtain effective signals.
For ease of understanding, the noise suppression method of the present invention is described below with respect to a specific relationship:
the M observation signals received by the M detectors are xiI ═ 1,2, … …, M; each observation signal is composed of N mutually independent source signals SjJ is 1,2, … …, N (including effective wave, alternating current interference wave, single frequency noise, etc.) is obtained by linear mixing; the relationship between the observed signal and the additive white noise and the source signal can be represented by the following relationship:
X＝A×S+N
wherein X is ═ X1,x2,……,xM]TIs an observed signal vector; s ═ S1,s2,……,sN]TIs an unknown independent source signal vector; n is an additive white noise vector; a is an m n unknown mixing matrix. The invention can remove the additive white noise vector N through the additive white noise removal processing, thereby meeting the assumption that the noise can be ignored in the ICA algorithm, and at the moment, the relational expression can be simplified as follows:
X＝A×S
and then separating N source signals in X through independent component analysis processing, and removing noise source signals to obtain effective signals.
The noise source signal herein mainly refers to a single-frequency noise source signal, such as an alternating-current interference wave with a frequency, a phase and an amplitude that are substantially unchanged from a shallow layer to a deep layer of the ground. Of course, the present invention is not particularly limited to this.
It can be understood that the noise suppression method in the present invention is an improved ICA algorithm, the conventional ICA algorithm directly processes the observed signal, and the assumption of the ICA algorithm is that "noise is negligible", so that additive white noise present in the observed signal will affect the ICA algorithm, so that the result obtained after the independent component analysis processing is inaccurate. The invention adds additive white noise removal treatment on the basis of the original ICA algorithm, thereby improving the accuracy of the ICA algorithm.
The additive white noise herein generally refers to thermal noise, shot noise, and the like, and their relationship to the signal is additive, that is, the additive white noise exists regardless of whether the detector receives the observed signal.
The sources of additive white noise in a channel are generally classified into three types:
1. artifacts, which originate from unrelated other signal sources, such as: external station signals, switch contact noise, industrial ignition emissions, etc.
2. Natural noise, which refers to various electromagnetic wave sources existing in nature, such as: lightning, lightning strikes, electrical storms in the atmosphere, various cosmic noises, and the like.
3. Internal noise, which is various kinds of noise generated by the system device itself, such as: thermal movement of free electrons in the resistor and fluctuation of carriers in the semiconductor, etc.
Preferably, M and N are both positive integers not less than 2.
The method for removing the additive white noise of the M observation signals specifically comprises the following steps:
and (3) performing additive white noise removal processing on the M observation signals by adopting a two-step eigenvalue decomposition method.
It can be further known that the process of performing additive white noise removal processing on the M observation signals by using the two-step eigenvalue decomposition method specifically includes:
step s 201: obtaining an observation signal vector X (k) from the M observation signals, wherein X (k) ═ X1,x2,……,xM]T，x1、x2、……、xMRespectively carrying out sampling processing on the M observation signals to obtain corresponding vectors;
step s 202: centering the observation signal vector X (k) to make the average value of the observation signal vector X (k) be 0;
step s 203: obtaining a correlation matrix according to the observation signal vector X (k) and a correlation matrix relational expressionWherein, the relational expression of the correlation matrix is:
step s 204: according to a correlation matrixAnd obtaining the characteristic value Lambda by the first decomposition relationX；
Wherein the first decomposition relation is:
wherein, ΛX＝diag{λ1，λ2，……，λN}；
Step s 205: pre-whitening processing is carried out on the observation signal vector X (k) according to the pre-whitening relational expression to obtain a pre-whitened vector Z (k);
wherein, the prewhitening relational expression is:
wherein,VSis ΛSA corresponding feature vector;
step s 206: obtaining a diagonal matrix according to a preset time lag value p and a second decomposition relation
ΣX(ii) a Wherein the second decomposition relation is:
wherein, UXFor the feature matrix, p is not equal to 0;
step s 207: determining diagonal matrix ΣXIf all singular values in the data are different, if yes, the step s208 is carried out; otherwise, returning to the step s206, and presetting different time lag values p;
here, the singular value means the diagonal matrix ΣXThe value on the middle diagonal;
step s 208: according to the pre-whitening vector Z (k) and the feature matrix UXAnd obtaining the observation signal vector X after the additive white noise is removed by the pre-whitening relational expression1(k)；
Wherein, observing a signal vector X1(k) The M signal vectors are respectively corresponding to the M observed signals after the additive white noise is removed.
Here, the time lag value p takes 1. Of course, the specific value of the time lag value is not limited, and the staff can set the value according to the actual situation.
In addition, the process of performing independent component analysis processing on the M observation signals from which the additive white noise is removed to obtain N source signals separated from each other specifically includes:
step s 401: setting the initial value of the iteration times q as 1;
step s 402: randomly generating an initial weight vector Wq；
Step s 403: will observe signal vector X1(k) And an initial weight vector WqCarrying in a target relational expression, wherein the obtained result is a target weight vector W;
wherein, the target relational expression is:
wherein E is the mean value operation, g is the nonlinear function; the target relation is a target relation based on negative entropy maximization;
step s 404: judging whether the target weight vector W is converged, if so, entering the step s405, otherwise, returning to the step s 403;
step s 405: adding 1 to the iteration number q, judging whether the iteration number q is greater than a preset iteration number threshold value, and if so, entering a step s 406; otherwise, return to step s 402;
step s 406: combining the target weight vector W and the observed signal vector X1(k) And substituting a separation relation to obtain a separation vector Y (k), wherein the separation relation is as follows:
Y(k)＝WX1(k)
the separation vector Y (k) includes N source signal vectors separated from each other, and the N source signal vectors correspond to the N source signals, respectively.
The preset iteration threshold can be determined through multiple tests, and the preset iteration threshold is set to enable separation signals corresponding to the finally obtained separation vector Y (k) to be respectively equal to the N source signals as far as possible through multiple iteration processes. Of course, the present invention does not limit the specific value of the preset iteration threshold.
Preferably, the non-linear function g [. cndot. ] may be:
g (y) tanh (ay) or g (y) yexp (-y)2Y is/2) or g (y)3Wherein a is more than or equal to 1 and less than or equal to 2.
Of course, the present invention is not particularly limited as to the type of nonlinear function g [. cndot. ] specifically employed.
In order to facilitate understanding of the invention, a simulation test can be carried out on the process of independent component analysis processing in the invention by adopting a mode of simulating seismic signals, so that the effectiveness and the accuracy of the improved ICA algorithm can be verified.
First, a zero-phase signal S with a frequency of 60Hz is selected1As an active wave, a square wave signal S with a frequency of 100Hz2And a random signal S3As alternating current interference waves (i.e. noise source signals) constituting the source signal S(i.e. each observed signal is composed of three source signals S independent of each other1、S2、S3Linear mixing), as shown in fig. 2, fig. 2 is a schematic diagram of a source signal composition in a noise suppression method provided by the present invention, where a time sampling interval of the source signal is 1ms, and a signal length is 0.5 s; the abscissa represents the number of sampling points and the ordinate represents the amplitude.
The source signal passes through a random mixing matrix A5×3Then, 5 observation signals X are obtained1～X5Referring to fig. 3, fig. 3 is a schematic diagram of an observed signal composition in a noise suppression method according to the present invention;
at this time, assume that the source signal S and the random mixing matrix A5×3Are unknown, and separate signals Y corresponding to the separate vectors Y (k) are obtained by independent component analysis processing only according to the observed signals X1～Y3Referring to fig. 4, fig. 4 is a schematic diagram of a separation signal corresponding to a separation vector Y (k) in a noise suppression method according to the present invention.
Wherein the independent component analysis process has a convergence error accuracy of 10-7In this case, the iteration time is 0.0780 s. The correlation coefficients of the isolated signal and the corresponding source signal are calculated to be-0.9980, -1.0000/0.9763, respectively.
Referring to fig. 5 and fig. 6, fig. 5 is a schematic frequency spectrum diagram of a source signal in a noise suppression method according to the present invention; FIG. 6 is a schematic frequency spectrum diagram of a separated signal in a noise suppression method according to the present invention; separating two independent components Y in a signal1And Y2Respectively 100Hz and 60Hz, respectively corresponding to the source signal S2And S1In contrast, the frequency is unchanged. That is, when the observed signal does not contain additive white noise, the independent component analysis processing can accurately separate the observed signal to obtain separate signals respectively corresponding to the source signals, so as to separate the noise source signals and obtain effective signals.
The simulation results of the simulation experiments show that the improved ICA algorithm has high convergence speed and high calculation precision, does not need to perform filtering processing in a frequency domain, can effectively overcome the influence of additive white noise on the conventional ICA algorithm, and better separates out the noise source signals in the pre-stack seismic data, thereby achieving the purpose of suppressing the noise source signals in the pre-stack seismic data.
The invention provides a noise suppression method which improves the traditional ICA algorithm. The method comprises the steps of firstly carrying out additive white noise removal processing on M observation signals received by M detectors to remove additive white noise in the observation signals, then carrying out independent component analysis processing on the M observation signals to separate and remove noise source signals in the observation signals, and thus obtaining effective signals. Namely, the invention not only can remove the noise source signal, but also can remove the additive white noise in the observation signal, and the efficiency of removing the noise is high; and the independent component analysis processing is carried out after the additive white noise is removed, so that the premise of 'noise neglect is not counted' in the independent component analysis processing is met, the influence of the additive white noise on the independent component analysis processing is avoided, and the accuracy of the result is improved.
The invention further provides a noise suppression system, and fig. 7 is a schematic structural diagram of the noise suppression system provided by the invention. The system comprises:
an obtaining unit 11, configured to obtain M observation signals received by M detectors; wherein each observation signal is composed of N source signals which are independent of each other; m and N are positive integers;
the noise removing unit 12 is configured to perform additive white noise removal processing on the M observation signals to obtain M observation signals from which the additive white noise is removed;
an independent component analysis processing unit 13, configured to perform independent component analysis processing on the M observation signals from which the additive white noise is removed, so as to obtain N source signals that are separated from each other;
the effective signal acquisition unit 14 removes the noise source signal from the N source signals separated from each other, and obtains an effective signal.
The invention provides a noise suppression system which improves the traditional ICA algorithm. The method comprises the steps of firstly carrying out additive white noise removal processing on M observation signals received by M detectors to remove additive white noise in the observation signals, then carrying out independent component analysis processing on the M observation signals to separate and remove noise source signals in the observation signals, and thus obtaining effective signals. Namely, the invention not only can remove the noise source signal, but also can remove the additive white noise in the observation signal, and the efficiency of removing the noise is high; and the independent component analysis processing is carried out after the additive white noise is removed, so that the premise of 'noise neglect is not counted' in the independent component analysis processing is met, the influence of the additive white noise on the independent component analysis processing is avoided, and the accuracy of the result is improved.
It should be noted that, in the present specification, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising an … …" does not exclude the presence of other identical elements in a process, method, article, or apparatus that comprises the element.
The previous description of the disclosed embodiments is provided to enable any person skilled in the art to make or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments without departing from the spirit or scope of the invention. Thus, the present invention is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims (5)

1.一种噪声压制方法，其特征在于，包括：1. A noise suppressing method, characterized in that, comprising: 获取M个检波器接收到的M个观测信号；其中，每个所述观测信号由相互独立的N个源信号组成；M和N均为正整数；Obtaining M observation signals received by M detectors; wherein, each of the observation signals is composed of mutually independent N source signals; M and N are both positive integers; 根据M个所述观测信号得到观测信号矢量X(k)，其中，X(k)＝[x1,x2,……,xM]T，x1、x2、……、xM分别为M个所述观测信号进行采样处理后对应得到的矢量；k为X(k)的采样参量；Obtain the observed signal vector X(k) according to the M observed signals, wherein, X(k)=[x 1 ,x 2 ,...,x M ] T , x 1 , x 2 ,...,x M respectively A vector obtained after sampling processing for the M observed signals; k is a sampling parameter of X(k); 对所述观测信号矢量X(k)进行中心化处理，使其均值为0；Centralize the observed signal vector X(k) so that its mean value is 0; 根据所述观测信号矢量X(k)以及相关矩阵关系式得到相关矩阵其中，所述相关矩阵关系式为：Obtain the correlation matrix according to the observed signal vector X(k) and the correlation matrix relational expression Wherein, the relational expression of the correlation matrix is: 根据所述相关矩阵以及第一分解关系式得到特征值ΛX；According to the correlation matrix And the first decomposition relation obtains eigenvalue Λ X ; 其中，所述第一分解关系式为：Wherein, the first decomposition relational expression is: 其中，ΛX＝diag{λ1，λ2，……，λN}； Among them, Λ X =diag{λ 1 ,λ 2 ,...,λ N }; 根据预白化关系式对所述观测信号矢量X(k)进行预白化处理，得到预白化矢量Z(k)；Perform pre-whitening processing on the observed signal vector X(k) according to the pre-whitening relational expression to obtain a pre-whitening vector Z(k); 其中，所述预白化关系式为：Wherein, the pre-whitening relational expression is: 其中，VS为ΛS对应的特征向量；in, V S is the eigenvector corresponding to Λ S ; 根据预设的时间滞后值p以及第二分解关系式得到对角矩阵ΣX；其中，所述第二分解关系式为：Obtain the diagonal matrix Σ X according to the preset time lag value p and the second decomposition relational expression; wherein, the second decomposition relational expression is: 其中，UX为特征矩阵，p≠0； Among them, U X is the characteristic matrix, p≠0; 判断所述对角矩阵ΣX中的全部奇异值是否均不相同；如果是，根据所述预白化矢量Z(k)、所述特征矩阵UX以及所述预白化关系式得到去除加性白噪声后的观测信号矢量X1(k)；其中，所述观测信号矢量X1(k)包括M个信号矢量，M个所述信号矢量分别对应于去除所述加性白噪声之后的M个所述观测信号；否则，返回根据预设的时间滞后值p以及第二分解关系式得到对角矩阵ΣX的步骤，并预设不同的时间滞后值p；得到去除加性白噪声之后的M个所述观测信号；Judging whether all the singular values in the diagonal matrix Σ X are different; if so, obtain the removal of additive whitening according to the pre-whitening vector Z(k), the feature matrix U X and the pre-whitening relational expression The observed signal vector X 1 (k) after noise; Wherein, the observed signal vector X 1 (k) includes M signal vectors, and the M signal vectors respectively correspond to the M observed signals after the additive white noise is removed; otherwise, return according to the preset The step of obtaining the diagonal matrix Σ X by the time lag value p and the second decomposition relational expression, and preset different time lag values p; obtain the M observed signals after removing the additive white noise; 对去除所述加性白噪声之后的M个所述观测信号进行独立分量分析处理，得到相互分离的N个所述源信号；performing independent component analysis on the M observed signals after removing the additive white noise, to obtain N mutually separated source signals; 从所述相互分离的N个所述源信号中去除噪声源信号，得到有效信号。The noise source signal is removed from the N mutually separated source signals to obtain an effective signal. 2.根据权利要求1所述的方法，其特征在于，所述时间滞后值p取1。2 . The method according to claim 1 , wherein the time lag value p is 1. 3 . 3.根据权利要求1所述的方法，其特征在于，所述对去除所述加性白噪声之后的M个所述观测信号进行独立分量分析处理，得到相互分离的N个所述源信号的过程具体为：3. The method according to claim 1, wherein the independent component analysis is carried out to the M observed signals after removing the additive white noise, to obtain the N source signals separated from each other. The process is specifically: 步骤s401：令迭代次数q的初始值为1；Step s401: Let the initial value of the number of iterations q be 1; 步骤s402：随机生成初始权矢量Wq；Step s402: Randomly generate the initial weight vector W q ; 步骤s403：将所述观测信号矢量X1(k)以及所述初始权矢量Wq带入目标关系式，得到的结果即为目标权矢量W；Step s403: Bring the observed signal vector X 1 (k) and the initial weight vector W q into the target relational expression, and the obtained result is the target weight vector W; 其中，所述目标关系式为：Wherein, the target relational expression is: 其中，E[.]为均值运算，g[.]为非线性函数；Among them, E[.] is the mean value operation, and g[.] is a nonlinear function; 步骤s404：判断所述目标权矢量W是否收敛，如果是，进入步骤s405，否则，返回步骤s403；Step s404: judging whether the target weight vector W is convergent, if yes, proceed to step s405, otherwise, return to step s403; 步骤s405：令所述迭代次数q加1，判断所述迭代次数q是否大于预设迭代次数阈值，如果是，进入步骤s406；否则，返回步骤s402；Step s405: increase the number of iterations q by 1, and judge whether the number of iterations q is greater than the preset threshold of iterations, if yes, go to step s406; otherwise, return to step s402; 步骤s406：将所述目标权矢量W以及所述观测信号矢量X1(k)带入分离关系式，得到分离矢量Y(k)，其中，所述分离关系式为：Step s406: Bring the target weight vector W and the observed signal vector X 1 (k) into a separation relation to obtain a separation vector Y(k), where the separation relation is: Y(k)＝WX1(k)Y(k)=WX 1 (k) 其中，所述分离矢量Y(k)中包括相互分离的N个源信号矢量，N个所述源信号矢量分别对应于N个所述源信号。Wherein, the separation vector Y(k) includes N source signal vectors separated from each other, and the N source signal vectors respectively correspond to the N source signals. 4.根据权利要求3所述的方法，其特征在于，所述非线性函数g[.]可以为：4. method according to claim 3, is characterized in that, described nonlinear function g[.] can be: g(y)＝tanh(ay)或g(y)＝yexp(-y2/2)或g(y)＝y3，其中，1≤a≤2。g(y)=tanh(ay) or g(y)=yexp(-y 2 /2) or g(y)=y 3 , where 1≦a≦2. 5.一种噪声压制系统，其特征在于，包括：5. A noise suppression system, characterized in that, comprising: 获取单元，用于获取M个检波器接收到的M个观测信号；其中，每个所述观测信号由相互独立的N个源信号组成；M和N均为正整数；An acquisition unit, configured to acquire M observation signals received by M detectors; wherein each observation signal is composed of mutually independent N source signals; M and N are both positive integers; 观测信号获得单元，用于根据M个所述观测信号得到观测信号矢量X(k)，其中，X(k)＝[x1,x2,……,xM]T，x1、x2、……、xM分别为M个所述观测信号进行采样处理后对应得到的矢量；k为X(k)的采样参量；An observation signal obtaining unit, configured to obtain an observation signal vector X(k) according to the M observation signals, where X(k)=[x 1 ,x 2 ,...,x M ] T , x 1 , x 2 , ..., x M are the corresponding vectors obtained after sampling processing of the M observed signals respectively; k is the sampling parameter of X(k); 中心化单元，用于对所述观测信号矢量X(k)进行中心化处理，使其均值为0；A centralization unit, configured to centralize the observed signal vector X(k) so that its mean value is 0; 矩阵获得单元，用于根据所述观测信号矢量X(k)以及相关矩阵关系式得到相关矩阵其中，所述相关矩阵关系式为： A matrix obtaining unit for obtaining a correlation matrix according to the observed signal vector X(k) and the correlation matrix relation Wherein, the relational expression of the correlation matrix is: 第一分解单元，用于根据所述相关矩阵以及第一分解关系式得到特征值ΛX；其中，所述第一分解关系式为：其中，ΛX＝diag{λ1，λ2，……，λN}；The first decomposition unit is used for according to the correlation matrix And the first decomposition relational expression obtains eigenvalue Λ X ; Wherein, the first decomposition relational expression is: Among them, Λ X =diag{λ 1 ,λ 2 ,...,λ N }; 预白化单元，用于根据预白化关系式对所述观测信号矢量X(k)进行预白化处理，得到预白化矢量Z(k)；其中，所述预白化关系式为：其中，VS为ΛS对应的特征向量；A pre-whitening unit, configured to perform pre-whitening processing on the observed signal vector X(k) according to a pre-whitening relational expression to obtain a pre-whitening vector Z(k); wherein, the pre-whitening relational expression is: in, V S is the eigenvector corresponding to Λ S ; 第二分解单元，用于根据预设的时间滞后值p以及第二分解关系式得到对角矩阵ΣX；其中，所述第二分解关系式为：其中，UX为特征矩阵，p≠0；The second decomposition unit is used to obtain the diagonal matrix ΣX according to the preset time lag value p and the second decomposition relational expression; wherein, the second decomposition relational expression is: Among them, U X is the characteristic matrix, p≠0; 判断单元，用于判断所述对角矩阵ΣX中的全部奇异值是否均不相同；如果是，触发去除单元，否则，触发所述第二分解单元，并预设不同的时间滞后值p；A judging unit, configured to judge whether all the singular values in the diagonal matrix ΣX are different; if yes, trigger the removal unit, otherwise, trigger the second decomposition unit, and preset a different time lag value p; 所述去除单元，用于根据所述预白化矢量Z(k)、所述特征矩阵UX以及所述预白化关系式得到去除加性白噪声后的观测信号矢量X1(k)；其中，所述观测信号矢量X1(k)包括M个信号矢量，M个所述信号矢量分别对应于去除所述加性白噪声之后的M个所述观测信号；得到去除加性白噪声之后的M个所述观测信号；The removing unit is used to obtain the observed signal vector X 1 (k) after removing the additive white noise according to the pre-whitening vector Z(k), the feature matrix U X and the pre-whitening relational expression; Wherein, the observed signal vector X 1 (k) includes M signal vectors, and the M signal vectors respectively correspond to the M observed signals after the additive white noise is removed; after the additive white noise is removed, The M observed signals of ; 独立分量分析处理单元，用于对去除所述加性白噪声之后的M个所述观测信号进行独立分量分析处理，得到相互分离的N个所述源信号；An independent component analysis processing unit, configured to perform independent component analysis processing on the M observed signals after removing the additive white noise, to obtain N mutually separated source signals; 有效信号取得单元，从所述相互分离的N个所述源信号中去除噪声源信号，得到有效信号。The effective signal acquisition unit removes the noise source signal from the N mutually separated source signals to obtain an effective signal.
CN201610048523.2A 2016-01-25 2016-01-25 A kind of noise drawing method and system Active CN105717543B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201610048523.2A CN105717543B (en) 2016-01-25 2016-01-25 A kind of noise drawing method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201610048523.2A CN105717543B (en) 2016-01-25 2016-01-25 A kind of noise drawing method and system

Publications (2)

Publication Number Publication Date
CN105717543A CN105717543A (en) 2016-06-29
CN105717543B true CN105717543B (en) 2018-07-13

Family

ID=56154910

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201610048523.2A Active CN105717543B (en) 2016-01-25 2016-01-25 A kind of noise drawing method and system

Country Status (1)

Country Link
CN (1) CN105717543B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112925024B (en) * 2021-01-26 2023-10-20 中国石油化工股份有限公司 A square wave suppression method for earthquake recording

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1873443A (en) * 2005-06-03 2006-12-06 中国石油大学(北京) Method for analyzing noise elimination of earthquake based on independent components in Pearson system
CN101729157A (en) * 2009-12-11 2010-06-09 南京航空航天大学 Method for separating vibration signal blind sources under strong noise environment

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7869303B2 (en) * 2007-08-14 2011-01-11 Pgs Geophysical As Method for noise suppression in seismic signals using spatial transforms
US9103943B2 (en) * 2011-11-28 2015-08-11 Fugro-Geoteam As Acquisition and processing of multi-source broadband marine seismic data

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1873443A (en) * 2005-06-03 2006-12-06 中国石油大学(北京) Method for analyzing noise elimination of earthquake based on independent components in Pearson system
CN101729157A (en) * 2009-12-11 2010-06-09 南京航空航天大学 Method for separating vibration signal blind sources under strong noise environment

Also Published As

Publication number Publication date
CN105717543A (en) 2016-06-29

Similar Documents

Publication Publication Date Title
CN107451557B (en) Short circuit fault diagnosis method of transmission line based on empirical wavelet transform and local energy
CN106771905B (en) A pulse extraction method suitable for high-frequency current partial discharge detection
Chen et al. Ground-roll noise attenuation using a simple and effective approach based on local band-limited orthogonalization
CN104777442B (en) A kind of nuclear magnetic resonance depth measurement FID signal noise suppressing method
CN104614778B (en) Nuclear magnetic resonance underground water detectable signal noise cancellation method based on ICA
CN109557429A (en) Based on the GIS partial discharge fault detection method for improving wavelet threshold denoising
CN111999607B (en) Method and device for separating partial discharge narrow-band interference blind source under single-channel signal
CN106154330B (en) Electromagnetic type controlled source parabolic Radon transform suppresses parallel focus crosstalk noise method
CN109885903B (en) Model-based ground nuclear magnetic resonance signal spike noise removing method
Yusoff et al. Denoising technique for partial discharge signal: A comparison performance between artificial neural network, fast fourier transform and discrete wavelet transform
CN113640891A (en) A Noise Filtering Method for Transient Electromagnetic Detection Data Based on Singular Spectrum Analysis
Mattsson et al. Noise reduction and error analysis for a towed EM System
Lan et al. Partial discharge location of power cables based on an improved phase difference method
CN103323819A (en) SAR time-varying narrow-band interference suppression method based on time-frequency spectrogram decomposition
CN108802175A (en) A kind of Pulsed eddy current testing signal de-noising method based on EMD
CN107688135B (en) Single-phase earth fault line selection method of small-current grounding system based on FastICA
CN104459498A (en) Method for suppressing interference on partial discharge test of ultra-high voltage AC transformer
Chan et al. Hybrid method on signal de‐noising and representation for online partial discharge monitoring of power transformers at substations
CN105717543B (en) A kind of noise drawing method and system
CN109973251A (en) A Blind Source Separation Method for Noise Reduction of Solid Rocket Motor Vibration Signals
CN117250585A (en) Parameter estimation method of multi-component chirp signal based on incomplete spectrum reconstruction
CN108089100A (en) The detection method of small current neutral grounding system arc light resistance earth fault
CN110244360A (en) Seismic data separation method and system based on anti-aliasing in effective frequency wavenumber domain
Shuchong et al. Seismic signals wavelet packet de-noising method based on improved threshold function and adaptive threshold
CN113391354B (en) Method and device for suppressing aliasing noise

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant