CN112083508A - Artificial source electromagnetic exploration signal denoising method and system with noise reference channel - Google Patents
Artificial source electromagnetic exploration signal denoising method and system with noise reference channel Download PDFInfo
- Publication number
- CN112083508A CN112083508A CN202010713236.5A CN202010713236A CN112083508A CN 112083508 A CN112083508 A CN 112083508A CN 202010713236 A CN202010713236 A CN 202010713236A CN 112083508 A CN112083508 A CN 112083508A
- Authority
- CN
- China
- Prior art keywords
- noise
- signal
- observation
- denoising
- dictionary library
- 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.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01V—GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS
- G01V3/00—Electric or magnetic prospecting or detecting; Measuring magnetic field characteristics of the earth, e.g. declination, deviation
- G01V3/38—Processing data, e.g. for analysis, for interpretation, for correction
Landscapes
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Environmental & Geological Engineering (AREA)
- Geology (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- General Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Geophysics (AREA)
- Geophysics And Detection Of Objects (AREA)
Abstract
The invention discloses a method and a system for denoising artificial source electromagnetic exploration signals with noise reference channels, wherein a noise observation point is arranged in an observation area, and the noise observation point only collects magnetic field signals of noise; the denoising method specifically comprises the following steps: performing sparse decomposition on an observed noise signal based on orthogonal matching pursuit; and performing signal-noise separation on the observation signal according to the decomposition result of the noise signal. According to the invention, only pure noise signals are acquired by adding a pure noise observation device, and the denoising problem is converted into a noise solving problem by using the periodic characteristics of artificial source signals and based on the spectrum discrete characteristics, so that the denoising difficulty is reduced, and the denoising success rate is improved.
Description
Technical Field
The invention belongs to the technical field of artificial source electromagnetic exploration, and particularly relates to an artificial source electromagnetic exploration signal denoising method and system with a noise reference channel.
Background
The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art.
The artificial source electromagnetic exploration method has wide application in the fields of energy, resource exploration, geological disasters, urban geological investigation and the like due to the characteristics of controllable field source, known signal type, relatively high signal-to-noise ratio and the like, such as a controllable source audio frequency magnetotelluric method (CSAMT), a wide area electromagnetic method (WFEM) and the like.
However, in recent years, with the continuous development of human society, strong human noise interference generally exists, which has become one of the key factors influencing the artificial source electromagnetic exploration effect, and the existence of the strong human noise interference seriously influences the data acquisition quality and reduces the accuracy and precision of data interpretation. Aiming at noise interference in artificial source electromagnetic exploration, a large amount of research is carried out on the artificial source electromagnetic exploration noise, and the noise is suppressed by pertinently adopting different methods according to the influence characteristics of different noises on effective signals. The traditional denoising method comprises a statistical method, a filtering method, an empirical mode method and the like. The inventor finds that the statistical method adopts a superposition estimation method to realize signal-noise separation, but needs a large number of samples as a premise, and is difficult to realize effective denoising when the sample data is insufficient; the filtering method is mainly based on the fact that noise and signals are obviously different in a frequency spectrum or after the signals are projected to a specific space, so that signal-noise separation is achieved, if the frequency spectrum characteristics of the signals and the noise are complex and mutually overlapped, the signals and the noise are difficult to separate on a corresponding space, and if certain unsteady electromagnetic noise is faced, effective denoising processing is difficult to complete; the empirical mode method can process signal noise under various complex conditions, the signal noise separation effect is obvious, but when the frequency spectrum components of effective signals and noise signals are overlapped seriously, the effective signal noise separation is still difficult to realize, the method is mainly used for natural source electromagnetic exploration, the nature difference between artificial source electromagnetic signals and magnetotelluric signals is large, and the frequency spectrum discrete characteristic of the artificial source effective signals often influences the denoising effect.
Disclosure of Invention
In order to overcome the defects of the prior art, the invention designs a denoising method aiming at the artificial source electromagnetic exploration signal from the characteristics of the artificial source electromagnetic exploration signal and the spatial distribution characteristics of the electromagnetic field of the artificial source electromagnetic exploration signal, eliminates the noise in the exploration signal in an observation area, and improves the signal-to-noise ratio of the artificial source electromagnetic exploration data in the environment with strong human noise.
In order to achieve the above object, one or more embodiments of the present invention provide the following technical solutions:
a noise removing method for artificial source electromagnetic exploration signals with noise reference channels is characterized in that noise observation points are arranged in an observation area, and only noise magnetic field signals are collected by the noise observation points; the denoising method specifically comprises the following steps:
performing sparse decomposition on an observed noise signal based on orthogonal matching pursuit;
and performing signal-noise separation on the observation signal according to the decomposition result of the noise signal.
Further, the method for calculating the observation azimuth angle of the noise observation point comprises the following steps:
and calculating the azimuth angle of an observation reference point aiming at the magnetic field noise at any position in the remote area according to the approximate equation of the electromagnetic field of the controllable source remote area by utilizing the position and the direction of the electric source field source and the position of the observation point.
Further, the method for calculating the observation azimuth angle of the noise observation point comprises the following steps:
obtaining an azimuth angle beta when the electric field component of the artificial field source at the noise observation point is 0:
according to the orthogonal property of the electric field and the magnetic field in the far zone, the observation azimuth angle with the magnetic field component of 0 is obtained:
γ＝β+2/π
wherein the content of the first and second substances,
the included angle between the noise reference point position and the X axis and the field source connecting line is shown.
Further, based on the orthogonal matching pursuit, sparsely decomposing the observed noise signal includes:
constructing a redundant dictionary library, wherein the redundant dictionary library comprises a basic atom dictionary library and a dynamic dictionary library, the basic atom dictionary library comprises a series of narrow-band atoms covering the whole spectrum space of the noise signal, and the dynamic dictionary library comprises narrow-band atoms corresponding to time-frequency units with specific spectrum characteristics in the noise signal;
and according to the redundant dictionary library, performing sparse decomposition on the time-frequency unit of the noise signal by using an orthogonal matching pursuit algorithm to obtain a series of noise atoms with narrow-band characteristics to form a noise matching dictionary library of the noise signal.
Further, the narrow-band features in the basic dictionary library are obtained according to a wavelet dictionary library or a low-order Legendre polynomial with the narrow-band features of db10, sym10 and dct.
Further, the narrow-band features in the dynamic dictionary library have narrow-band features of Chirplet wavelet, Gabor wavelet, sin or cos wavelet.
Furthermore, when sparse decomposition is carried out on the time-frequency unit of the noise signal by utilizing an orthogonal matching pursuit algorithm, the iteration times are set.
Further, signal-to-noise separating the observed signal comprises:
and performing signal-noise separation on the observation signal based on a least square inversion denoising method according to the noise matching dictionary library.
Further, signal-to-noise separating the observed signal comprises:
picking up boundary distribution by using an edge detection algorithm for the frequency spectrum image of each time-frequency unit of the noise signal, counting the number of the boundaries, evaluating the image complexity based on the number of the boundaries, and reserving the time-frequency units with the complexity lower than a threshold value according to a given complexity threshold value;
and for the reserved time-frequency unit, performing signal-noise separation on the observation signal by frequency division based on a least square inversion denoising method.
One or more embodiments provide an artificial source electromagnetic survey system comprising a set of noise observation devices that collect only noisy magnetic field signals; the system denoises the artificial source electromagnetic exploration signal based on the method.
The above one or more technical solutions have the following beneficial effects:
according to the invention, a 'pure' noise observation device is added when a field source supplies power, only pure noise signals are collected, the denoising problem is converted into a noise solving problem based on the spectrum discrete characteristics by utilizing the periodic characteristics of artificial source signals, all noises do not need to be processed, only part of simple noises need to be processed or depicted, and the selected noise position to be processed can be fragmented, and does not need to be selected in a complete period, so that the denoising difficulty is reduced to a certain extent, and the denoising success rate is improved.
The collected noise signals are magnetic field signals, and the collecting device, namely the magnetic bar or the fluxgate, is small in size, small in occupied area, free of grounding treatment, less in measurement condition constraint and easy to arrange, and can be arranged horizontally or vertically.
In addition, the method is suitable for artificial source electromagnetic exploration in areas with strong interference, even if effective signals are weak, the noise signals can also realize characteristic analysis, and the orthogonal matching tracking algorithm is implemented to realize signal-noise separation.
The redundant dictionary library is established, narrow-band features are used as basic units, and the noise matching dictionary library obtained through orthogonal matching tracking can reflect the feature information of the noise signal more accurately.
When the signal-noise separation is carried out on the observation signals based on the least square inversion denoising method, the signal-noise separation is only carried out aiming at the frequency band with the complexity within the set range, and the processing efficiency can be greatly improved while the precision is ensured.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, are included to provide a further understanding of the invention, and are incorporated in and constitute a part of this specification, illustrate exemplary embodiments of the invention and together with the description serve to explain the invention and not to limit the invention.
FIG. 1 is a schematic view of a torsional measurement angle of an electric field according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of Hmin observation in the embodiment of the present invention.
Detailed Description
It is to be understood that the following detailed description is exemplary and is intended to provide further explanation of the invention as claimed. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.
It is noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of exemplary embodiments according to the invention. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, and it should be understood that when the terms "comprises" and/or "comprising" are used in this specification, they specify the presence of stated features, steps, operations, devices, components, and/or combinations thereof, unless the context clearly indicates otherwise.
The embodiments and features of the embodiments of the present invention may be combined with each other without conflict.
The invention designs a denoising method aiming at an artificial source electromagnetic exploration signal from the periodic characteristics of an artificial source signal and the spatial distribution characteristics of an electromagnetic field, which is mainly characterized in that a group of 'pure' noise observation devices are added in an observation area as noise reference points, and the denoising method specifically comprises the following steps: when a field source supplies power, the device does not or hardly collect electromagnetic signals from an artificial source by rotating a certain measuring angle, only pure noise signals are collected, time-frequency unit division is carried out on the pure noise signals to obtain noise time distribution characteristics, and an electric field and a magnetic field of actual exploration signals can be processed based on the characteristics; meanwhile, after a 'pure' noise signal of observation is obtained, a 'real-time' noise dictionary library can be established based on the noise data, orthogonal matching tracking is carried out on the noise data to obtain a noise matching dictionary library in an observation area, noise in actual exploration signals in the observation area is removed based on the matching dictionary library and a least square inversion denoising method is combined, and the signal-to-noise ratio of the artificial source electromagnetic exploration data in a strong human noise environment is improved.
Example one
The embodiment discloses a method for denoising artificial source electromagnetic exploration signals with noise reference channels. The noise reference channel is realized by adding a group of pure noise observation devices in an observation area when power is supplied for artificial source electromagnetic exploration, the noise observation devices are rotated by a certain measurement angle to enable the noise observation devices not to collect or hardly collect electromagnetic signals from artificial sources, only magnetic field signals of pure noise are collected, signal characteristics of the magnetic field signals are analyzed, a denoising problem is converted into a noise solving problem, effective signal-noise separation is further realized, and the signal-to-noise ratio of data is improved.
The denoising method comprises two stages: a noise reference point setting mode and a denoising method of artificial source electromagnetic exploration based on an orthogonal matching pursuit algorithm.
Stage one: noise reference point setting
Based on the known field source position, the remote area approximate formula of the remote area wide-area electromagnetic method and the controllable source audio frequency geoelectromagnetic method is utilized to calculate the corresponding angle deflection when the observed electric field at any position is zero, and the electric field value at the moment is called as EminUsing the property of minimum magnetic field orthogonal to minimum electric field, along EminAnd magnetic rods are arranged in the direction perpendicular to the observation direction to collect magnetic field signals. The magnetic rods are arranged at the observation azimuth angle, and are connected with an acquisition instrument through a connecting cable to perform acquisition, so that the magnetic field signals corresponding to the field source cannot be received or almost cannot be received, all or most of the received signals come from interference noise, and thus, only the noise signals are observed, the effective signals are not observed, and the magnetic field signals at the moment are called as 'pure' noise magnetic field signals Hmin. Meanwhile, because no effective periodic signal influence exists during acquisition, the noise type and the influence time of the signal are more accurately judged during analysis.
Specifically, the noise reference point setting includes the steps of:
step 1-1: calculating an observation reference point azimuth angle aiming at magnetic field noise at any position in a remote area according to an approximate equation of a controllable source remote area electromagnetic field by utilizing the position and the direction of the electric source field source and the position of an observation point;
step 1-2: and determining the azimuth angle of the embedded magnetic rod according to the azimuth angle of the noise observation reference point.
And finally, arranging the magnetic rods according to the observation azimuth angle.
The azimuth angle calculation method of the noise observation device comprises the following steps:
by means of HminAnd EminBy formula derivation, we can find EminObserving the azimuth angle of the noise reference point of (H)minThe observation azimuth angle of the noise reference point is indirectly solved, and the azimuth angle embedded by the magnetic rod is further obtained, wherein the calculation and derivation process of the observation azimuth angle is as follows:
according to the formula of the electric field component of the electric source in the uniform ground, the E of the horizontal plane is expressed by using the cylindrical coordinatesrAnd
component ErPerpendicular to
As shown in fig. 1, the electric field calculation formula is as follows:
wherein the content of the first and second substances,
in the artificial source frequency domain electromagnetic exploration method, the wave number is called as dielectric constant, mu is magnetic conductivity, and omega is angular frequency; rho is resistivity (omega. m), I is emission current (A), d is loop side length (m), and r is receiving and transmitting distance (m).
When the receiving-transmitting distance r is larger and enters a far zone in the electromagnetic exploration field, e-ikrBecome very small, e-ikr(1+ ikr) is negligible compared to the constants 1, 2, etc., and equation (1) and (2) are further simplified:
the embodiment aims to observe the electric field zero value band through the torsion measurement angle, so that the observed artificial field source electric field component is 0, and the magnetic field torsion measurement angle is solved by utilizing the orthogonal property of the electric field and the magnetic field. The torsion measurement angle of the electric field is schematically shown in fig. 1.
Wherein the content of the first and second substances,
the noise reference point position and the included angle between the field source connecting line and the X axis are measured, alpha is the rotating axis measuring direction and ErAnd the included angle beta is the included angle between the measurement direction and the Y axis after rotation. The angle component and the radial component of the electric field rotate anticlockwise by beta to reach the target position, the electric field value of 0 component can be obtained, and the electric field value at the moment is considered to be 0 or almost zero, namely EminWe satisfied the objective of obtaining an electric field signal without a significant signal, but with only a pure noise electric field component (where Emin may be found in patent application No. 2020100329921).
When the noise reference point measurement is carried out in the zero value band, the direction is called EηAnd the following equation can be established according to equations (3) and (4), by calculating the available rotation angle α:
in order to avoid the infinite condition when the arctangent is directly solved, the problem is converted into the problem of solving the arcsine, and the problem is obtained:
readily available by means of a complementary relationship between angles, EηAngle beta with Y axis, see formula
Knowing the electric field torsion angle, according to the orthogonal property of the electric field and the magnetic field in the far zone, the measurement angle gamma of the magnetic field required to be twisted when the magnetic field component is 0 can be obtained as follows:
γ＝β+2/π (9)
according to the azimuth angle of the magnetic field needing to be twisted, the magnetic bar buried underground is rotated by the angle, the measurement of the magnetic field noise reference point is carried out, and then a 'pure' noise magnetic field signal H is obtainedmin. Meanwhile, in order to assist in denoising, pure noise electric field signal acquisition can be implemented according to the measurement angle beta of the electric field needing to be twisted, and the acquired noise magnetic field signal H is comprehensively acquiredminAnd carrying out orthogonal matching analysis, and carrying out denoising processing on the acquired magnetic field signals.
For artificial source electromagnetic surveying, it is generally accepted that electromagnetic waves propagate in a plane wave manner in the "far zone". Assuming that a galvanic couple source is used for transmitting, under the condition of a uniform earth and a layered model, the angles of electric field 'zero-value bands' at different positions can be calculated in a far zone according to the relative position relation between a field source and a receiving point, the angle is measured by twisting a receiving electrode MN (namely a magnetic rod), so that the receiving electrode MN does not observe signals from an artificial source, namely, the signals are collected at the positions of the zero-value bands or the positions close to the zero-value bands, and the angles of the zero-value bands are different along with the position change. When the topographic relief is not large, the torsion angle is considered to be only related to the position of the topographic relief, and is almost unrelated to the resistivity and the frequency, and a strict zero-value zone region H does not necessarily exist in the complicated region with topographic relief and electrical inhomogeneities in the undergroundminThere is a high probability that the signal will contain an artificial source signal. In fact, H in this embodimentminIt does not have to strictly contain artificial source signals, but only needs to be as small as possible, and HminThe smaller the signal containing the artificial source is, the more accurate the noise depiction in the observation area is, and the better the denoising effect aiming at the exploration data is. At the same time, it is not necessary for all observation positions to be noise referencedThe observation device can make H according to the noise interference degree of the current measurement areaminAnd parameters such as observation density, observation polar distance and the like are obtained, a correlation formula is established, and the arrangement of noise reference points is implemented. As shown in FIG. 2, X and Y represent horizontal and vertical coordinates, respectively, A and B represent galvanic couples, the line between A and B represents the field source line, and the upper three points H in the figureyThree observation points are shown, the vertical line passing through each observation point represents the observation angle, and other horizontal or oblique lines represent HminThe observation angle of the signal at the corresponding position.
By adopting the method of the embodiment, when power is supplied, a group of noise observation devices are added in the area with larger noise influence in the observation area, and the observation devices are magnetic bars and can collect magnetic field signals (H is used as the magnetic bar)yDirection as an example). The magnetic bar is connected with an observation instrument through a connecting cable, and observation is carried out at a certain angle at a noise reference point; by observing the magnetic field component HyWhen the magnetic rod is detected, the magnetic rod is rotated by a measuring angle, the angle of the magnetic rod is calculated according to the formula (5), so that the magnetic rod hardly collects signals emitted from an artificial source, only noise magnetic field signals are measured, and the magnetic rod is called as H in the embodimentmin. When the magnetic bar is used for collecting, the horizontal magnetic field is measured only by rotating the angle calculated by the formula (9), the horizontal magnetic field is connected with a measuring instrument through a connecting cable, and the collecting instrument is started to collect the pure magnetic field noise. The occupied area is small when the magnetic rods are arranged, and the method is suitable for urban artificial source electromagnetic exploration with limited field space. Binding of the collected HminThe pure magnetic field noise signal is used for denoising the noise in Hy normally observed in an orthogonal matching mode, and the quality of the collected data of the artificial source electromagnetic exploration is improved.
The pure noise magnetic field signal H is carried out through the arrangement mode of the noise reference pointsminAnd (4) collecting, analyzing the noise signal, and further realizing denoising.
And a second stage: artificial source electromagnetic exploration based on orthogonal matching pursuit algorithm
Due to HminHas the characteristics of same position and simultaneous acquisition with the normal observation Hy, and the noise contained in the two has high homology and similarityMeanwhile, due to different measurement angles, the overall noise size, phase and the like of the measurement angles may be different, so that the noise cannot be directly removed through signal coherence or subtraction. This embodiment utilizes an Orthogonal Matching Pursuit algorithm (OMP) to rank H based on a given redundant dictionary libraryminSparse decomposition into a series of noise atoms (Atom), the acquisition constituent HminThe noise matching dictionary library, and the solution H is established based on the matching dictionaryxThe overdetermined equation of the effective frequency coefficient is used for carrying out sparse decomposition on an observed signal into noise basic atoms to obtain a real-time matching dictionary library corresponding to noise in view of Hy noise and HminAnd (3) the similarity between signals, and integrating artificial source electromagnetic exploration frequency information and the matching dictionary library (namely constructing an inversion matrix A in a least square inversion denoising method) to separate the signal and the noise of Hy. The method specifically comprises the following steps:
step 2-1: and constructing a redundant dictionary library, wherein the redundant dictionary library comprises a basic atom dictionary library and a dynamic dictionary library, the basic atom dictionary library comprises a series of narrow-band atoms covering the whole spectrum space of the noise signal, and the dynamic dictionary library comprises narrow-band atoms corresponding to time-frequency units with specific spectrum characteristics in the noise signal.
In areas with strong interference, the noise complexity is high. And aiming at the reconstruction unit with higher complexity, establishing a basic dictionary library and a dynamic dictionary library at the same time.
The step 2-1 specifically comprises:
step 2-1-1: for noise signal HminDividing time-frequency units;
specifically, first, for HminAnd step point monitoring is carried out on the time domain signal, a value point with large fluctuation in the signal is obtained, the signal is divided into different regions in time by taking the step point as a boundary, and the influence of frequency spectrum energy leakage caused by signal-to-noise mutation during post-denoising processing is avoided. Then, aiming at the different regions, according to the noise conditions of different frequencies, dividing the frequency domain corresponding to each region into several frequency bands to be processed, thereby forming a series of equidistant time-frequency units in the time-frequency spectrum, filling the corresponding frequency intervals, wherein the time length of the time-frequency unit and the time length of the server are the sameThe computing power is relevant. The continuous time frequency units can be combined into larger time frequency units, and the calculation complexity is reduced.
Step 2-1-2: respectively constructing a basic dictionary library and a dynamic dictionary library by combining frequency bands corresponding to all time-frequency units;
for the basic dictionary library, a db10 wavelet packet, a sym10 wavelet packet, a dct dictionary library and a basic dictionary library of low-order legendre polynomials (such as 0 order, 1 order and the like) with full time length can be initially established for different frequency bands in the actual observation process. The frequency spectrums of db10 and sym10 have good narrow-band characteristics and tight-support characteristics in a time domain, and the whole frequency spectrum space can be divided into a series of narrow-band units as dictionary base atoms.
And for the dynamic dictionary library, storing the frequency spectrum result of each time-frequency unit as an image, analyzing the image of each time-frequency unit, identifying the time-frequency unit with the specific frequency spectrum characteristic, sequentially reconstructing a signal with the specific frequency spectrum characteristic in the image as an updating atom, and establishing the dynamic dictionary library. The specific spectral features are spectral features with wavelet narrow-band features such as Chirplet, Gabor, sin, cos and the like.
The basic dictionary library is the same for all time-frequency units, and the dynamic dictionary library is dynamically updated according to different time-frequency unit characteristics, so that a redundant dictionary library consisting of basic and dynamic parts is built for each time-frequency unit.
Step 2-2: using orthogonal matching pursuit algorithm to pair H according to redundant dictionary libraryminSparse decomposition of time-frequency unit noise, and separation of HminSparse decomposition into a series of noise atoms (Atom) with narrow-band characteristics, resulting in the constituent HminThe noise matching dictionary base;
specifically, for each time-frequency unit of the noise signal, a narrow-band unit is matched from a redundant dictionary library based on an orthogonal matching tracking algorithm according to a time axis position, the narrow-band unit obtained through matching may be from a basic dictionary library or a dynamic dictionary library, and a noise matching dictionary library of the noise signal obtained through matching is used for depicting a pure noise part in the noise-containing signal.
The orthogonal matching pursuit algorithm is evolved from the matching pursuit algorithm. The basic idea of matching tracking is to find the atom most similar to the signal as the current matching atom by traversing a given redundant dictionary library according to an inner product maximization principle, subtract the similar component of the atom from the signal, continuously perform inner product on the rest part and other atoms in the dictionary library, find the corresponding atom with the largest inner product, subtract the similar part of the atom again, and stop iteration in sequence until a termination condition of a fitting error or iteration times is met. The problems of repeated atom selection, slow iterative convergence and the like are solved.
Due to Hy and HminThe measured angles are different, so that the noise is subjected to earth convolution before being collected, the earth filter coefficients of different frequency components are different, the orientations are different, and the same noise is in the H positionxAnd HminIn (b) is different from (b) isminThe method is different from the traditional orthogonal matching pursuit algorithm in that the method is decomposed into a series of narrow-band atoms, and in the embodiment, the redundant dictionary base has specific requirements, and the actual signals are not necessarily subjected to sparse decomposition, so that the method focuses more on fitting errors and fitting residual noise characteristics in the OMP operation process, and does not pursue that the corresponding time-frequency unit can be decomposed by using a few sparse coefficients. By estimating HminAnd designing an OMP convergence error threshold to control the final fitting precision. In this embodiment, the iteration number is set according to the actual observation requirement not to exceed a fixed proportion of a given signal length, for example, 10% of the total signal length, for example, the signal length is 1000, the iteration number of the orthogonal matching pursuit is not more than 100, and at most 100 sparse representation coefficients are obtained. After iteration is carried out for 100 times, if the fitting error does not meet the requirement, the time-frequency unit does not serve as an effective time-frequency unit to participate in the least square method denoising.
Orthogonal matching pursuit algorithm based on dictionary base and aiming at HminAnd Hy noise by the pair HminAnd a Hy corresponding noise matching dictionary library is constructed in the learning process, so that the uncertainty in the traditional dictionary learning process is greatly reduced.
Step 2-3: and performing signal-noise separation on the observation signal based on a least square inversion denoising method according to the noise matching dictionary library.
Step 2-3-1: and for the frequency spectrum image of each time-frequency unit of the noise signal, picking up the boundary distribution condition in the image by using an edge detection algorithm, counting the number of boundaries in each time-frequency unit, evaluating the image complexity according to the number of the boundaries, screening all the time-frequency units according to a given complexity threshold value, and reserving the time-frequency units with the complexity lower than the threshold value as reconstruction units during least square inversion.
The continuous time frequency units can be combined into larger time frequency units, and the calculation complexity is reduced. The positions of the screened time-frequency units are used as reconstruction units to participate in the least square method denoising, and the positions of the screened time-frequency units are not used as effective time-frequency units to participate in the least square method denoising when the complexity of the time-frequency units subjected to orthogonal matching pursuit decomposition exceeds a certain degree (for example, after 10% of the total iteration length mentioned in the step 2-2, the error fitting conditions are not met).
Step 2-3-2: and the sub-band carries out signal-noise separation on the observation signal based on a least square inversion denoising method.
On the basis of time unit division, carrying out denoising and result evaluation on frequency bands, dividing the frequency bands into a plurality of frequency bands, and respectively carrying out least square inversion denoising to obtain denoising results of the corresponding frequency bands; each band requires that the upper and lower boundaries need to include all bands to be processed, and therefore the upper and lower boundaries need to be extended to avoid the influence caused by spectrum leakage. In this embodiment, it is assumed that the frequency range of the frequency band to be processed is [ m, n ], and the upper and lower boundaries are [ m/2, n/2] after being expanded, and those skilled in the art can understand that the size of the expansion can be designed according to actual situations. In the actual processing process, different frequency bands can be superposed with each other, and when a signal is processed, the frequency spectrum coefficient of a high-frequency part outside the upper bound of the frequency can be set to zero and then reconstructed to a time domain, and down-sampling is carried out to reduce the calculation complexity; each frequency band is relatively independent in the denoising process and can run in parallel, so that the computing time is saved.
The least square inversion Denoising method (see patent ZL201610410616.5, or thesis, "Denoising controlled-source electronic data using least square inversion", Yang Y, Li D, Tong T, Geophysics,2018.83(4), E229-E244) has the following specific implementation modes: collecting a first time sequence signal of a current signal emitted by an emission source, and carrying out spectrum analysis on the first time sequence signal to obtain the emission frequency of the signal; when an effective signal is received, acquiring a second time series signal received by a receiving terminal Hy, and performing Fourier transform on the second time series signal to obtain a first coefficient of the second time series signal; performing spectrum analysis on the second time series signal, selecting a plurality of data points to perform inverse Fourier transform processing on the second time series signal, and establishing an over-determined equation set taking the transmitting frequency as an unknown number based on the plurality of data points:
Ax＝b (10)
acquiring a second coefficient of the non-periodic signal in the second time series signal according to the over-determined equation set; and obtaining an effective frequency coefficient in the Hy signal according to the first coefficient and the second coefficient, and finally obtaining a least square solution of a frequency domain coefficient corresponding to the exploration frequency, thereby realizing signal separation aiming at Hy.
The left side of equation (10) is augmented by a pass pair HminMatching dictionary obtained by orthogonal matching pursuit
(characterize noise details) and a low-order legendre polynomial (characterize large step undulations in the signal), an inversion matrix a is established, as shown in equation (11).
Wherein g [ m ]]For Gaussian noise of the aperiodic part at the reconstruction location, Fk]Is the coefficient of the non-periodic part of Ex at the non-significant frequency position (equal to the coefficient of the original signal at the non-significant frequency, which is a known coefficient), FNPC[l1]、FNPC[N-l1]、FNPC[l2]、FNPC[N-l2]Equal to the unknown coefficient of the Hy aperiodic part at the effective frequency position, where FNPC[lm]And FNPC[N-lm]Conjugation; a. b and c are matching dictionary libraries
Middle corresponding atom phi1、φ2And phi3D and e are respectively polynomial coefficients of 0 order and 1 order, and a, b, c, d and e are unknown coefficients.
More inclined to the horizontal direction E than the observation of noisy electric field signalsx-EminThe magnetic field signal can be carried out in all directions (H)x-Hmin、Hy-HminAnd Hz-Hmin) The observation direction is more flexible, the method is suitable for collecting the noise magnetic field signals in any direction, and the noise removal processing can be carried out on the magnetic field signals in any direction according to the noise removal method. In addition, the magnetic field is relatively easy to observe, only a magnetic rod or a fluxgate is needed, and too many measuring condition constraints are not needed, so that the device is more suitable for arranging the observation device for the artificial source electromagnetic exploration among cities.
Those skilled in the art will appreciate that the modules or steps of the present invention described above can be implemented using general purpose computer means, or alternatively, they can be implemented using program code that is executable by computing means, such that they are stored in memory means for execution by the computing means, or they are separately fabricated into individual integrated circuit modules, or multiple modules or steps of them are fabricated into a single integrated circuit module. The present invention is not limited to any specific combination of hardware and software.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Although the embodiments of the present invention have been described with reference to the accompanying drawings, it is not intended to limit the scope of the present invention, and it should be understood by those skilled in the art that various modifications and variations can be made without inventive efforts by those skilled in the art based on the technical solution of the present invention.
Claims (10)
1. A denoising method for an artificial source electromagnetic exploration signal with a noise reference channel is characterized in that a noise observation point is arranged in an observation area, and the noise observation point only collects a magnetic field signal of noise; the denoising method specifically comprises the following steps:
performing sparse decomposition on an observed noise signal based on orthogonal matching pursuit;
and performing signal-noise separation on the observation signal according to the decomposition result of the noise signal.
2. The method for denoising artificial source electromagnetic survey signals with noise reference channels according to claim 1, wherein the method for calculating the observation azimuth angle of the noise observation point comprises:
and calculating the azimuth angle of an observation reference point aiming at the magnetic field noise at any position in the remote area according to the approximate equation of the electromagnetic field of the controllable source remote area by utilizing the position and the direction of the electric source field source and the position of the observation point.
3. The method for denoising artificial source electromagnetic survey signals with noise reference channels according to claim 1, wherein the method for calculating the observation azimuth angle of the noise observation point comprises:
obtaining an azimuth angle beta when the electric field component of the artificial field source at the noise observation point is 0:
according to the orthogonal property of the electric field and the magnetic field in the far zone, the observation azimuth angle with the magnetic field component of 0 is obtained:
γ＝β+2/π
wherein the content of the first and second substances,
the included angle between the noise reference point position and the X axis and the field source connecting line is shown.
4. The method of denoising artificial source electromagnetic survey signals with noise reference traces of claim 1, wherein sparse decomposition of the observed noise signals based on orthogonal matching pursuit comprises:
constructing a redundant dictionary library, wherein the redundant dictionary library comprises a basic atom dictionary library and a dynamic dictionary library, the basic atom dictionary library comprises a series of narrow-band atoms covering the whole spectrum space of the noise signal, and the dynamic dictionary library comprises narrow-band atoms corresponding to time-frequency units with specific spectrum characteristics in the noise signal;
and according to the redundant dictionary library, performing sparse decomposition on the time-frequency unit of the noise signal by using an orthogonal matching pursuit algorithm to obtain a series of noise atoms with narrow-band characteristics to form a noise matching dictionary library of the noise signal.
5. The method of denoising artificial source electromagnetic survey signals with noise reference traces of claim 4, wherein the narrow band features in the base dictionary library are derived from a wavelet dictionary library or a low order Legendre polynomial with narrow band features db10, sym10, dct.
6. The method of claim 4, wherein the narrow-band features in the dynamic dictionary library have Chirplet wavelet, Gabor wavelet, sin or cos wavelet narrow-band features.
7. The method of claim 4, wherein the number of iterations is set when sparse decomposition is performed on the time-frequency unit of the noise signal using an orthogonal matching pursuit algorithm.
8. The method of denoising artificial source electromagnetic survey signals with noise reference channels according to claim 4, wherein performing signal-to-noise separation on the survey signals comprises:
and performing signal-noise separation on the observation signal based on a least square inversion denoising method according to the noise matching dictionary library.
9. The method of denoising artificial source electromagnetic survey signals with noise reference channels according to claim 8, wherein performing signal-to-noise separation on the survey signals comprises:
picking up boundary distribution by using an edge detection algorithm for the frequency spectrum image of each time-frequency unit of the noise signal, counting the number of the boundaries, evaluating the image complexity based on the number of the boundaries, and reserving the time-frequency units with the complexity lower than a threshold value according to a given complexity threshold value;
and for the reserved time-frequency unit, performing signal-noise separation on the observation signal by frequency division based on a least square inversion denoising method.
10. An artificial source electromagnetic surveying system comprising a set of noise observation devices, said noise observation devices collecting only noisy magnetic field signals; the system denoises an artificial source electromagnetic survey signal based on the method of any one of claims 1-9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010713236.5A CN112083508B (en) | 2020-07-22 | 2020-07-22 | Artificial source electromagnetic exploration signal denoising method and system with noise reference channel |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010713236.5A CN112083508B (en) | 2020-07-22 | 2020-07-22 | Artificial source electromagnetic exploration signal denoising method and system with noise reference channel |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112083508A true CN112083508A (en) | 2020-12-15 |
| CN112083508B CN112083508B (en) | 2021-08-13 |
Family
ID=73734697
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010713236.5A Active CN112083508B (en) | 2020-07-22 | 2020-07-22 | Artificial source electromagnetic exploration signal denoising method and system with noise reference channel |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112083508B (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113640888A (en) * | 2021-09-24 | 2021-11-12 | 山东大学 | External noise suppression method and system based on frequency domain cross-correlation of transmitted and received signals |
| CN113655532A (en) * | 2021-09-03 | 2021-11-16 | 山东大学 | Method and system for removing motion noise of non-full-time semi-aviation transient electromagnetic data |
| CN113759426A (en) * | 2021-08-02 | 2021-12-07 | 山东大学 | Artificial source electromagnetic exploration method and system based on reference channel |
| CN114764151A (en) * | 2021-01-13 | 2022-07-19 | 中国石油化工股份有限公司 | Magnetotelluric frequency division tomography inversion method |
| CN114924328A (en) * | 2022-05-24 | 2022-08-19 | 山东大学 | Urban artificial source electromagnetic exploration method and system with vertical magnetic field reference channel |
| CN114970647A (en) * | 2022-07-29 | 2022-08-30 | 中南大学 | Electromagnetic data identification method and system based on probabilistic neural network |
| CN114924328B (en) * | 2022-05-24 | 2024-05-24 | 山东大学 | Urban artificial source electromagnetic exploration method and system with vertical magnetic field reference channel |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2014028415A1 (en) * | 2012-08-15 | 2014-02-20 | Westerngeco Llc | Methods and systems for deghosting marine seismic data |
| CN107368668A (en) * | 2017-05-30 | 2017-11-21 | 中国石油大学(华东) | Seismic data noise attenuation method based on the study of dual sparse dictionary |
| CN107644406A (en) * | 2017-09-22 | 2018-01-30 | 南京理工大学 | A kind of image de-noising method based on improved orthogonal matching pursuit |
| CN110208869A (en) * | 2019-07-08 | 2019-09-06 | 湖南师范大学 | A kind of Magnetotelluric signal denoising method based on the setting of sparse decomposition threshold value |
| CN110673222A (en) * | 2019-09-30 | 2020-01-10 | 湖南师范大学 | Magnetotelluric signal noise suppression method and system based on atomic training |
| CN110865414A (en) * | 2019-11-01 | 2020-03-06 | 吉林大学 | Transient electromagnetic noise suppression method for urban underground space detection |
| CN111190234A (en) * | 2020-01-13 | 2020-05-22 | 山东大学 | Noise observation method and device for artificial electrical source frequency domain electromagnetic method |
| CN111239839A (en) * | 2020-02-10 | 2020-06-05 | 山东大学 | Frequency spectrum density calculation method and device for frequency domain electromagnetic method |
-
2020
- 2020-07-22 CN CN202010713236.5A patent/CN112083508B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2014028415A1 (en) * | 2012-08-15 | 2014-02-20 | Westerngeco Llc | Methods and systems for deghosting marine seismic data |
| CN107368668A (en) * | 2017-05-30 | 2017-11-21 | 中国石油大学(华东) | Seismic data noise attenuation method based on the study of dual sparse dictionary |
| CN107644406A (en) * | 2017-09-22 | 2018-01-30 | 南京理工大学 | A kind of image de-noising method based on improved orthogonal matching pursuit |
| CN110208869A (en) * | 2019-07-08 | 2019-09-06 | 湖南师范大学 | A kind of Magnetotelluric signal denoising method based on the setting of sparse decomposition threshold value |
| CN110673222A (en) * | 2019-09-30 | 2020-01-10 | 湖南师范大学 | Magnetotelluric signal noise suppression method and system based on atomic training |
| CN110865414A (en) * | 2019-11-01 | 2020-03-06 | 吉林大学 | Transient electromagnetic noise suppression method for urban underground space detection |
| CN111190234A (en) * | 2020-01-13 | 2020-05-22 | 山东大学 | Noise observation method and device for artificial electrical source frequency domain electromagnetic method |
| CN111239839A (en) * | 2020-02-10 | 2020-06-05 | 山东大学 | Frequency spectrum density calculation method and device for frequency domain electromagnetic method |
Non-Patent Citations (1)
| Title |
|---|
| 朱会杰 等: ""基于正交匹配追踪的强脉冲电磁干扰滤波新方法"", 《振动与冲击》 * |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114764151A (en) * | 2021-01-13 | 2022-07-19 | 中国石油化工股份有限公司 | Magnetotelluric frequency division tomography inversion method |
| CN114764151B (en) * | 2021-01-13 | 2023-06-23 | 中国石油化工股份有限公司 | Magnetotelluric frequency division chromatography inversion method |
| CN113759426A (en) * | 2021-08-02 | 2021-12-07 | 山东大学 | Artificial source electromagnetic exploration method and system based on reference channel |
| CN113655532A (en) * | 2021-09-03 | 2021-11-16 | 山东大学 | Method and system for removing motion noise of non-full-time semi-aviation transient electromagnetic data |
| CN113655532B (en) * | 2021-09-03 | 2022-05-17 | 山东大学 | Method and system for removing motion noise of non-full-time semi-aviation transient electromagnetic data |
| CN113640888A (en) * | 2021-09-24 | 2021-11-12 | 山东大学 | External noise suppression method and system based on frequency domain cross-correlation of transmitted and received signals |
| CN114924328A (en) * | 2022-05-24 | 2022-08-19 | 山东大学 | Urban artificial source electromagnetic exploration method and system with vertical magnetic field reference channel |
| CN114924328B (en) * | 2022-05-24 | 2024-05-24 | 山东大学 | Urban artificial source electromagnetic exploration method and system with vertical magnetic field reference channel |
| CN114970647A (en) * | 2022-07-29 | 2022-08-30 | 中南大学 | Electromagnetic data identification method and system based on probabilistic neural network |
| CN114970647B (en) * | 2022-07-29 | 2022-11-11 | 中南大学 | Electromagnetic data identification method and system based on probabilistic neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112083508B (en) | 2021-08-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112083508B (en) | Artificial source electromagnetic exploration signal denoising method and system with noise reference channel | |
| Zhang et al. | Unsupervised seismic random noise attenuation based on deep convolutional neural network | |
| Yiou et al. | Spectral analysis of climate data | |
| Astone et al. | A method for detection of known sources of continuous gravitational wave signals in non-stationary data | |
| Cicone et al. | Multivariate fast iterative filtering for the decomposition of nonstationary signals | |
| Cohen et al. | Mitigation of 50–60 Hz power line interference in geophysical data | |
| Li et al. | Dictionary learning and shift-invariant sparse coding denoising for controlled-source electromagnetic data combined with complementary ensemble empirical mode decomposition | |
| Capilla | Application of the Haar wavelet transform to detect microseismic signal arrivals | |
| Dong et al. | Signal-to-noise ratio enhancement for 3C downhole microseismic data based on the 3D shearlet transform and improved back-propagation neural networks | |
| CN106443183B (en) | A kind of method that frequency source quantity and fundamental frequency value are extracted from electromagnetic spectrum | |
| Li et al. | Audio magnetotelluric signal-noise identification and separation based on multifractal spectrum and matching pursuit | |
| EP2745404A1 (en) | Smart data sampling and data reconstruction | |
| CN112487604B (en) | Long-time nonlinear drift compensation method for output data of marine gravimeter | |
| Liu et al. | Irregularly sampled seismic data reconstruction using multiscale multidirectional adaptive prediction-error filter | |
| CN111597981A (en) | Magnetotelluric signal denoising method and system based on improved multi-scale diffusion entropy | |
| Kremer et al. | Review of acquisition and signal processing methods for electromagnetic noise reduction and retrieval of surface nuclear magnetic resonance parameters | |
| Zhang et al. | Separation of magnetotelluric signals based on refined composite multiscale dispersion entropy and orthogonal matching pursuit | |
| Larnier et al. | New application of wavelets in magnetotelluric data processing: reducing impedance bias | |
| Zhang et al. | Grey wolf optimization-based variational mode decomposition for magnetotelluric data combined with detrended fluctuation analysis | |
| RU2412454C2 (en) | Method to process seismic data using discrete wavelet transform | |
| CN113887398A (en) | GPR signal denoising method based on variational modal decomposition and singular spectrum analysis | |
| Govindarajan et al. | HANKEL‐EM‐SVD: A hybrid data dropout estimation technique for high voltage partial discharge signals | |
| Chen et al. | Two-dimensional nonlinear geophysical data filtering using the multidimensional EEMD method | |
| US20200233108A1 (en) | High-Resolution Processing Method For Seismic Data based on Inverse Multi-Resolution Singular Value Decomposition | |
| CN112213782A (en) | Processing method and device for sub-phase seismic data and server |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |