A method for suppressing non-physical event axis noise of acoustic passive source virtual shot gather

Abstract

The present invention relates to a method for suppressing non-physical event noise of virtual shot collection of acoustic passive source, including obtaining virtual shot collection data after interference, obtaining the common detection point gather morphology thereof, splitting the common detection point data and the common shot point data into data blocks, block matching, obtaining a 3D data body, obtaining a preliminary filtering result, obtaining a filtered result, obtaining a final filtering result, obtaining iterative input data, obtaining a denoising result, and judging whether the final result meets the accuracy requirement. The present invention distinguishes between effective signals and non-physical event, and introduces collaborative filtering into the suppression process of non-physical event, identifying and suppressing noise by comparing various signal patterns; modifying the collaborative filtering framework so that it can perform filtering operations on non-physical event, the formed suppression method combines the advantages of noise characteristics and collaborative filtering, effectively enhances the signal-to-noise ratio and accuracy of the expected signal, and greatly supplements the robustness of the non-physical event noise suppression algorithm.

Description

Technical Field

[0001] The invention belongs to the technical field of seismic exploration, and in particular relates to a method for suppressing non-physical event axis noise of acoustic passive source virtual shot gathers. Background Art

[0002] Passive source seismic exploration methods mainly use environmental noise in nature for exploration, such as microseismic geological activities, and produce seismic responses similar to active sources through seismic interferometry. Compared with active source seismic exploration methods, it avoids dependence on explosive sources or other artificial sources, has the advantages of low construction cost and small environmental impact during the exploration process, provides a new method for seismic data acquisition, and has become an indispensable technology in seismic exploration. In recent years, with the development of passive source seismic exploration technology, it has been widely used in various fields such as resource exploration, earthquake disaster assessment, and reservoir management and optimization. Although passive source seismic exploration has developed rapidly, there are still serious defects in practical applications.

[0003] Passive source seismic exploration has strict requirements on the layout of the detectors. Under ideal conditions, the detectors are required to completely enclose the passive source that generates vibrations. In addition, the amplitude and main frequency band of the passive source in the field cannot be controlled. This leads to a large amount of non-physical event noise interference in the virtual shot set obtained by seismic interferometry. The presence of non-physical event in the virtual shot set often makes many subsequent seismic data processing methods unable to meet the basic assumptions, making it impossible for traditional processing to obtain satisfactory processing results. Since non-physical event belongs to coherent noise and has signal characteristics similar to effective reflection, it is difficult to obtain a good noise suppression effect using traditional filters. At present, in passive source seismic exploration, there is still a lack of a mature method for suppressing non-physical event noise in passive source virtual shot sets. Summary of the invention

[0004] The purpose of the present invention is to provide a method for suppressing non-physical event axis noise of acoustic passive source virtual shot gathers, so as to solve the problem of suppressing coherent noise generated in virtual shot gathers due to failure to meet the basic assumptions of passive source seismic interferometry. The method effectively enhances the signal-to-noise ratio and accuracy of virtual shot gathers, and improves the usability of passive source seismic data in the field of exploration.

[0005] The objective of the present invention is achieved through the following technical solutions:

[0006] A method for suppressing non-physical event axis noise of acoustic passive source virtual shot gathers comprises the following steps:

[0007] A. Based on the passive source wave field data, obtain the virtual shot gather data after interference;

[0008] B. Rearrange the common shot point data and obtain the common detection point gather form by formula (2);

[0009]

[0010] in, Indicates detection point x j The common receiver gathers at Indicates that the earthquake source is located at x N The j-th data of the N-th shot gather;

[0011] C. The common detection point data D CRG and total shot point data D CSG Perform data segmentation and split it into data blocks;

[0012]

[0013] in, Represents the split data block, x1, x2, ... x M Indicates the midpoint coordinates of each data block after splitting;

[0014] D. Build a noise model for pseudo waves and implement block matching: For the split common shot point gather data blocks, calculate the similarity between each block;

[0015] E. Sort the similarities between the data blocks, and perform 3D transformation on the data blocks with high similarity to obtain the 3D data volume, i.e., the spectrum coefficient after 3D transformation

[0016] F. The spectrum coefficient is calculated by formula (10) Perform thresholding to obtain preliminary filtering results

[0017] in,

[0018] G. The result of the common shot gather after preliminary filtering by formula (11) With the original common detection point gather data Perform Wiener filtering and obtain the filtered result:

[0019]

[0020] Among them, CRG Represents the noise density spectrum of the common detection point gather;

[0021] H. Perform an inverse transformation on the result of collaborative filtering to restore it to a 2D image, thereby obtaining the final filtering result, namely:

[0022]

[0023] Where Q represents the inverse transform;

[0024] I. Record the number of iterations k = 1, and record the kth denoising result as Y k , superimpose the denoising result with the original data to obtain the input data D of the k+1th iteration k+1 :

[0025] D k+1 =(Y k +D) (13)

[0026] J. Replace the input data with D k+1 Repeat steps B to H, and calculate the difference between the result obtained and the result of the previous iteration to obtain the denoising result Y k+1 ;

[0027] Y k+1 =C(D k+1 )-Y k (14)

[0028] Among them, C represents the denoising process;

[0029] K. Determine whether the final result meets the accuracy requirement. If not, k=k+1 and return to step I.

[0030] Furthermore, in step A, the virtual shot data after interference is obtained based on the interference method, and the formula is as follows:

[0031]

[0032] Where D represents the virtual shot gather after interferometric measurement, <·> represents the overall spatial average, ρ is the density, c is the propagation velocity, and the inverse of the product of density and velocity represents the amplitude correction factor. is the passive source wave field received at two points A and B.

[0033] Furthermore, in step D, for the split common shot point gather data blocks, the similarity between each block is calculated by formula (5):

[0034]

[0035] in, Represents the reference data block split above and The similarity between them, || ||2 means to find the 2-norm, Represents variance.

[0036] Furthermore, in step D, the specific process of obtaining the variance is:

[0037]

[0038] Where F represents Fourier transform, represents the 2D Fourier transform basis function, is an array of the same size as the data block z, except for the coordinates x1, x2, …x M Outside the , it is zero everywhere, the value at the coordinate is the 1D Fourier transform basis function, Ψ represents the power density spectrum of the noise, and its calculation formula is:

[0039] Ψ=|D CSG ||F(g)| 2 (7)

[0040] Where g is the linear noise convolution kernel:

[0041] g = cos(x (1) +x (2) )G 10 (x (1) ,x (2) ) (8)

[0042] Among them, x (1) and x (2) represents horizontal and vertical integer coordinates, G 10 represents a Gaussian function centered at the origin with standard deviation 10.

[0043] Furthermore, in step E, the process of performing 3D transformation to obtain a 3D data volume can be understood as first performing 2D transformation on the data block and then performing 1D transformation:

[0044]

[0045] in, represents the spectral coefficient after 3D transformation, T2 represents the 2D Fourier transform, and its basis function is T1 is 1D Fourier transform, and the basis function is

[0046] Furthermore, in step F, only the common shot point gather data are preliminarily filtered.

[0047] Furthermore, in step G, Ψ CRG It is obtained by formula (7).

[0048] Compared with the prior art, the present invention has the following beneficial effects:

[0049] 1. The present invention utilizes the characteristics of different similarities between signals and noises recorded in different gathers of passive source virtual shots to distinguish effective signals from non-physical events, and introduces collaborative filtering into the suppression process of non-physical events to identify and suppress noise by comparing various signal patterns;

[0050] 2. By modifying the framework of collaborative filtering so that it can perform filtering operations on non-physical event axes, the present invention ultimately forms a targeted virtual shot set noise suppression method, which combines the noise characteristics with the advantages of collaborative filtering, effectively enhances the signal-to-noise ratio and accuracy of the expected signal, and greatly supplements the robustness of the non-physical event axis noise suppression algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0052] Figure 1 It is a flow chart of the method for suppressing non-physical event axis noise of acoustic passive source virtual shot gather according to the present invention;

[0053] Figure 2 This is a schematic diagram of data partitioning;

[0054] Figure 3 To show the processing effect, Figure 3 (a) is the original data, Figure 3 (b) is the denoising result. Figure 3 (c) is the difference profile. DETAILED DESCRIPTION

[0055] The present invention will be further described below in conjunction with embodiments:

[0056] The present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It is to be understood that the specific embodiments described herein are only used to explain the present invention, rather than to limit the present invention. It should also be noted that, for ease of description, only parts related to the present invention, rather than all structures, are shown in the accompanying drawings.

[0057] It should be noted that similar reference numerals and letters represent similar items in the following drawings, so once an item is defined in one drawing, it does not need to be further defined and explained in the subsequent drawings. At the same time, in the description of the present invention, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.

[0058] The present invention utilizes the large difference between the similarity of passive source virtual shot gather recording signals and the similarity of noise in different channel gather displays to distinguish between effective signals and non-physical event axes. In addition, collaborative filtering is introduced into the suppression process of non-physical event axes, which greatly supplements the robustness of the non-physical event axis noise suppression algorithm. The present invention can effectively suppress the coherent noise generated in the virtual shot gather due to the failure to meet the basic assumptions of passive source seismic interference, effectively enhance the signal-to-noise ratio and accuracy of the virtual shot gather, and improve the availability of passive source seismic data in the exploration field.

[0059] The present invention provides a method for suppressing non-physical event axis noise of acoustic passive source virtual shot gather, comprising the following steps:

[0060] The first step is to obtain the virtual shot collection data after interference based on the interference method;

[0061]

[0062] Where D represents the virtual shot gather after interferometric measurement, <·> represents the overall spatial average, ρ is the density, and c is the propagation velocity; the inverse of the product of density and velocity represents the amplitude correction factor. is the passive source wave field received at two points A and B.

[0063] The second step is to rearrange the common shot point data to obtain the common receiver point gather form;

[0064]

[0065] in, Indicates detection point x j The common receiver gathers at Indicates that the earthquake source is located at x N The j-th data of the N-th shot collection.

[0066] The third step is to convert the common detection point data D CRG and total shot point data D CSG All are split into data blocks;

[0067]

[0068] in, Represents the split data block, x1, x2, ... x M Indicates the midpoint coordinates of each data block after splitting.

[0069] The fourth step is block matching: for the split common shot point gather data blocks, the similarity between each block is calculated respectively. The formula can be expressed as:

[0070]

[0071] in, Represents the reference data block split above and The similarity between them, || ||2 means to find the 2-norm, Represents variance.

[0072] It should be noted that the noise characteristics of non-physical event axes are improved here, and the variance calculation formula is different from the conventional one. The specific calculation process is:

[0073]

[0074] Where F represents Fourier transform, represents the 2D Fourier transform basis function, is an array of the same size as the data block z, except for the coordinates x1, x2, …x M Outside the , it is zero everywhere, and the value at the coordinate is the 1D Fourier transform basis function. Ψ represents the power density spectrum of the noise, and its calculation formula is:

[0075] Ψ=|D CSG ||F(g)| 2 (7)

[0076] Where g is the linear noise convolution kernel:

[0077] g = cos(x (1) +x (2) )G 10 (x (1) ,x (2) ) (8)

[0078] Among them, x (1) and x (2) represents horizontal and vertical integer coordinates, G 10 represents a Gaussian function centered at the origin with standard deviation 10.

[0079] In the fifth step, the similarities between the data blocks can be sorted through the above steps, and the data blocks with high similarity can be transformed into 3D to obtain a 3D data volume. This process can be understood as first performing a 2D transformation on the data block and then a 1D transformation.

[0080]

[0081] in, represents the spectral coefficient after 3D transformation, T2 represents the 2D Fourier transform, and its basis function is T1 is 1D Fourier transform, and the basis function is

[0082] The sixth step is to use formula (10) to calculate the spectral coefficient Perform thresholding to obtain preliminary filtering results

[0083] in, It should be noted that only the common shot point gather data are initially filtered here. For the sake of distinction, the data of this initial filtering is recorded as:

[0084] Step 7: The common shot gather results after preliminary filtering With the original common detection point gather data Perform Wiener filtering and obtain the filtered result:

[0085]

[0086] Among them, CRG The noise density spectrum of the common detection point gather can be obtained by formula (7).

[0087] In the eighth step, the result after collaborative filtering is inversely transformed to restore it to a 2D image, thereby obtaining the final filtering result, namely:

[0088]

[0089] Wherein, Q represents the inverse transform.

[0090] In order to further improve the denoising effect and the accuracy of the results, an enhancement algorithm is introduced, and the original denoising is modified to iterative acquisition. The specific process is as follows:

[0091] Step 9: Record the number of iterations k = 1 and record the kth denoising result as Y k , superimpose the denoising result with the original data to obtain the input data D of the k+1th iteration k+1 :

[0092] D k+1 =(Y k +D) (13)

[0093] Step 10: Replace the input data with D k+1 Repeat steps 2 to 8, and calculate the difference between the result obtained and the result of the previous iteration to obtain the denoising result Y k+1 ;

[0094] Y k+1 =C(D k+1 )-Y k (14)

[0095] Wherein, C represents the denoising process proposed by the present invention.

[0096] Step 11: Determine whether the final result meets the accuracy requirement. If not, k=k+1 and return to step 9.

[0097] Example 1

[0098] A method for suppressing the non-physical event axis noise of acoustic passive source virtual shot gathers, such as Figure 1 As shown, it is applied to data testing, and the two-dimensional data received by 250 detectors are tested.

[0099] The first step is to obtain the virtual shot data after interference based on the interference method. Specifically, the sampling time is 2.8s, the sampling interval is 0.004s, and the shot channels are the same, both of which are 250.

[0100]

[0101] Where D represents the virtual shot gather after interferometric measurement, <·> represents the overall spatial average value, ρ is the density, c is the propagation velocity, which is 1400 m / s in this example, and the inverse of the product of density and velocity represents the amplitude correction factor. is the passive source wave field received at two points A and B.

[0102] The second step is to rearrange the common shot point data to obtain the common receiver point gather form;

[0103]

[0104] in, represents the common detection point gather at detection point j, The detection point represents the j-th data of the N-th shot gather.

[0105] The third step is to convert the common detection point data D CRG and total shot point data D CSG are split into data blocks. The data block process is as follows Figure 2 As shown, in this example, the data block size is selected as 9.

[0106]

[0107] in, Represents the split data block, x1, x2, ... x M Indicates the midpoint coordinates of each data block after splitting.

[0108] The fourth step is block matching: for the split common shot point gather data blocks, the similarity between each block is calculated respectively. The formula can be expressed as:

[0109]

[0110] in, Indicates a reference data block and data blocks The similarity between them, || ||2 means to find the 2-norm, Here, the noise characteristics of non-physical event axes are improved, and the variance calculation formula is different from the conventional one. The specific calculation process is:

[0111]

[0112] Where F represents Fourier transform, represents the 2D Fourier transform basis function, is an array of the same size as the data block z, except for the coordinates x1, x2, …x M Outside the , it is zero everywhere, and the value at the coordinate is the 1D Fourier transform basis function. Ψ represents the power density spectrum of the noise, and its calculation formula is:

[0113] Ψ=|D CSG ||F(g)| 2 (7)

[0114] Where g is the linear noise convolution kernel:

[0115] g = cos(x (1) +x (2) )G 10 (x (1) ,x (2) ), (8)

[0116] Among them, x (1) and x (2) represents horizontal and vertical integer coordinates, G 10 represents a Gaussian function centered at the origin with standard deviation 10.

[0117] Step 5: Through the above steps, the similarities between the data blocks can be sorted, and the data blocks with high similarity can be transformed into 3D to obtain a 3D data volume. This process can be understood as: first perform a 2D transformation on the data block and then a 1D transformation:

[0118]

[0119] in, represents the spectral coefficient after 3D transformation, T2 represents the 2D Fourier transform, and its basis function is T1 is 1D Fourier transform, and the basis function is

[0120] The sixth step is to perform threshold processing on the transformed 3D data volume. In practical applications, the threshold parameters can be automatically obtained through the above steps without being specifically given to obtain the preliminary filtering results.

[0121] in,

[0122] It should be noted that only the common shot point gather data are initially filtered here. For the sake of distinction, we record this initial filtered data as:

[0123] Step 7: The common shot gather results after preliminary filtering With the original common detection point gather data Perform Wiener filtering and obtain the filtered result:

[0124]

[0125] Among them, CRG The noise density spectrum of the common detection point gather can be obtained by formula (7).

[0126] The eighth step is to perform an inverse transformation on the result of collaborative filtering to restore it to a 2D image, thereby obtaining the final filtering result. That is:

[0127]

[0128] Wherein, Q represents the inverse transform.

[0129] In order to further improve the denoising effect and the accuracy of the results, an enhancement algorithm is introduced, and the original denoising is modified to iterative acquisition. The specific process is as follows:

[0130] Step 9: Record the number of iterations k = 1 and record the kth denoising result as Y k , superimpose the denoising result with the original data to obtain the input data D of the k+1th iteration k+1 :

[0131] D k+1 =(Y k +D) (13)

[0132] Step 10: Replace the input data with D k+1 Repeat steps 2 to 8, and calculate the difference between the result obtained and the result of the previous iteration to obtain the denoising result Y k+1 :

[0133] Y k+1 =C(D k+1 )-Y k (14)

[0134] Wherein, C represents a single denoising process proposed by the present invention, including steps 2 to 8, that is, represents a process in an iterative cycle.

[0135] Step 11: Determine whether the final result meets the accuracy requirement. If not, k=k+1 and return to step 9. Figure 3 As shown, it can be seen that the non-physical event axis noise in the virtual shot set is well suppressed.

[0136] Note that the above are only preferred embodiments of the present invention and the technical principles used. Those skilled in the art will understand that the present invention is not limited to the specific embodiments described herein, and that various obvious changes, readjustments and substitutions can be made by those skilled in the art without departing from the scope of protection of the present invention. Therefore, although the present invention has been described in more detail through the above embodiments, the present invention is not limited to the above embodiments, and may include more other equivalent embodiments without departing from the concept of the present invention, and the scope of the present invention is determined by the scope of the appended claims.

Claims

1. A method for suppressing non-physical event axis noise of acoustic passive source virtual shot gather, characterized in that: The following steps are involved: A. Based on the passive source wave field data, obtain the interferometric virtual shot gather data; B. Rearrange the common shot point data and obtain the common detection point gather form by formula (2); in, Indicates detection point x j The common receiver gathers at Indicates that the earthquake source is located at x N The j-th data of the N-th shot gather; C. The common detection point data D CRG and total shot point data D CSG Perform data segmentation and split it into data blocks; in, Represents the split data block, x1, x2, ... x M Indicates the midpoint coordinates of each data block after splitting; D. Build a noise model for pseudo waves and implement block matching: For the split common shot point gather data blocks, calculate the similarity between each block; E. Sort the similarities between the data blocks, and perform 3D transformation on the data blocks with high similarity to obtain the 3D data volume, i.e., the spectrum coefficient after 3D transformation F. The spectrum coefficient is calculated by formula (10) Perform threshold processing to obtain preliminary filtering results That is, through the piecewise function Spectral coefficient Perform 0 / 1 system filtering to obtain preliminary filtering results in, G. The result of the common shot gather after preliminary filtering by formula (11) With the original common detection point gather data Perform Wiener filtering and obtain the filtered result: Among them, CRG Represents the noise density spectrum of the common detection point gather; H. Perform an inverse transformation on the result of collaborative filtering to restore it to a 2D image, thereby obtaining the final filtering result, namely: Where Q represents the inverse transform; I. Record the number of iterations k = 1, and record the kth denoising result as Y k , superimpose the denoising result with the original data to obtain the input data D of the k+1th iteration k+1 : D k+1 =(Y k +D) (13) J. Replace the input data with D k+1 Repeat steps B to H, and calculate the difference between the result obtained and the result of the previous iteration to obtain the denoising result Y k+1 ; Y k+1 =C(D k+1 )-Y k (14) Among them, C represents the denoising process; K. Determine whether the final result meets the accuracy requirement. If not, k=k+1 and return to step I.

2. A method for suppressing non-physical event axis noise of acoustic passive source virtual shot gather according to claim 1, characterized in that: Step A: The virtual shot data after interference is obtained based on the interference method, and the formula is as follows: Where D represents the virtual shot gather after interferometric measurement, <·> represents the overall spatial average, ρ is the density, c is the propagation velocity, and the inverse of the product of density and velocity represents the amplitude correction factor. is the passive source wave field received at two points A and B.

3. A method for suppressing non-physical event axis noise of acoustic passive source virtual shot gather according to claim 1, characterized in that: Step D: For the split common shot point gather data blocks, the similarity between each block is calculated by formula (5): in, Represents the reference data block split above and The similarity between them, || ||2 means to find the 2-norm, Represents variance.

4. A method for suppressing non-physical event axis noise of acoustic passive source virtual shot gather according to claim 3, characterized in that: Step D: The specific process of obtaining the variance is as follows: Where F represents Fourier transform, represents the 2D Fourier transform basis function, is an array of the same size as the data block z, except for the coordinates x1, x2, …x M Outside the , it is zero everywhere, the value at the coordinate is the 1D Fourier transform basis function, Ψ represents the power density spectrum of the noise, and its calculation formula is: Ψ=|D CSG ||F(g)| 2 (7) Where g is the linear noise convolution kernel: g=cos(x (1) +x (2) )G 10 (x (1) ,x (2) ) (8) Among them, x (1) and x (2) represents horizontal and vertical integer coordinates, G 10 represents a Gaussian function centered at the origin with standard deviation 10.

5. The method for suppressing non-physical event axis noise of acoustic passive source virtual shot gather according to claim 1, characterized in that: Step E, perform 3D transformation. The process of obtaining a 3D data volume can be understood as first performing a 2D transformation on the data block and then performing a 1D transformation: in, represents the spectral coefficient after 3D transformation, T2 represents the 2D Fourier transform, and its basis function is T1 is 1D Fourier transform, and the basis function is 6. The method for suppressing non-physical event axis noise of acoustic passive source virtual shot gather according to claim 1, characterized in that: Step F, only perform preliminary filtering on the common shot point gather data.

7. A method for suppressing non-physical event axis noise of acoustic passive source virtual shot gather according to claim 4, characterized in that: Step G, Ψ CRG It is obtained by formula (7).

Images