Special geologic body identification method based on RPCA background modeling

Through RPCA background modeling, seismic data is decomposed into sparse, sharp noise and low-rank structure information matrix, which solves the problems of low efficiency and poor accuracy of special geological bodies in the existing technology, and realizes efficient and accurate automatic recognition, suitable for complex Geological conditions and low signal-to-noise ratio data.

CN118915133AActive Publication Date: 2024-11-08CHINA NAT PETROLEUM CORP +2
View PDF 7 Cites 0 Cited by

Patent Information

Application Number
CN202310506913.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2023-05-08
Publication Date
2024-11-08
Estimated Expiration
2043-05-08

AI Technical Summary

Technical Problem

The existing technology is inefficient and has poor accuracy when identifying special geological bodies such as slit holes, volcanoes, rivers, etc., especially in the case of low signal-to-noise ratio and complex structures. Manual interpretation is time-consuming and errors are prone to occur. The intelligent identification method has not yet achieved relatively accurate results. automatic interpretation of results.

Method used

Using a method based on RPCA background modeling, the seismic data is decomposed into sparse sharp noise matrix and low-rank structure information matrix. The matrix is ​​updated through SVD decomposition and soft threshold algorithm until the residuals meet the set threshold, and the sparse sharp noise matrix is ​​output as a special Geological data, low-rank structure information matrix serves as the residual stratigraphic background.

Benefits of technology

It improves the automatic identification efficiency and accuracy of special geological bodies, is suitable for complex geological conditions and low seismic data quality, reduces computing resource consumption, and improves computing efficiency and applicability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118915133A_ABST
    Figure CN118915133A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of seismic signal processing and interpretation, and particularly discloses a special geologic body recognition method based on RPCA background modeling, which comprises the following steps: S1, constructing a special geologic body recognition model based on an RPCA algorithm, and decomposing a seismic data matrix into a sparse sharp noise matrix and a low-rank structure information matrix; s2, calculating a low-rank structure information matrix through an SVD decomposition algorithm; calculating a sparse sharp noise matrix through a soft threshold algorithm; the low-rank structure information matrix and the sparse sharp noise matrix are updated through iteration residual errors until the residual errors between the seismic data two-dimensional matrix and the sparse sharp noise matrix and between the seismic data two-dimensional matrix and the low-rank structure information matrix meet set conditions; s3, outputting a sparse sharp noise matrix, namely special geologic body data; and outputting a low-rank structure information matrix, namely the residual stratum background. According to the method, the special geologic body can be identified, and the interpretation efficiency, accuracy and applicability of the special geologic body are improved. The method is suitable for identification of special geologic bodies.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of seismic signal processing and interpretation, and in particular is a special geological body recognition method based on RPCA background modeling. Background Art

[0002] With the increase of energy demand and the continuous development of oil and gas exploration and development, most of the large structural oil and gas reservoirs such as anticline traps and fault structural traps in the world have been discovered, and geological exploration personnel have begun to shift their attention to identifying hidden oil and gas reservoirs in special geological bodies such as fractures, caves, and rivers. Traditional interpretation of special geological bodies is mostly done manually, but with the continuous deepening of oil and gas exploration, seismic data has also increased, and the efficiency of manual interpretation can no longer meet production needs.

[0003] At present, in existing technologies and software, interpreters generally use multi-attribute body coherence, seismic phase identification, wave dynamics characteristics and lithology to assist in the interpretation of special geological bodies, but the efficiency of manual interpretation is very low and requires a lot of time and effort. For low signal-to-noise ratio and complex structural conditions, identification is more difficult and more prone to large errors. After identification, manual secondary corrections are often required, which greatly reduces work efficiency. In recent years, some deep learning recognition methods based on forward models as labels have been developed and have begun to be applied to seismic phase interpretation. However, since intelligent methods are still in their infancy and are highly targeted, the currently known fracture and cave identification cannot achieve relatively accurate automatic interpretation results, and no intelligent recognition methods specifically targeted at volcanoes and rivers have been introduced. Summary of the invention

[0004] The purpose of the present invention is to provide a special geological body identification method based on RPCA background modeling to achieve accurate automatic identification and interpretation of special geological bodies such as fractures, volcanoes, and rivers.

[0005] In order to achieve the above-mentioned purpose, the technical methods adopted by the present invention are as follows:

[0006] A special geological body identification method based on RPCA background modeling includes the following steps performed in sequence:

[0007] S1. Based on the RPCA algorithm, a special geological body recognition model is constructed. The seismic data of the target work area is decomposed into a two-dimensional matrix of seismic data, which is input into the special geological body recognition model. The two-dimensional matrix of seismic data is decomposed into a sparse sharp noise matrix and a low-rank structural information matrix.

[0008] S2. Calculate the low-rank structural information matrix by SVD decomposition algorithm; calculate the sparse sharp noise matrix by soft threshold algorithm; update the low-rank structural information matrix and the sparse sharp noise matrix by iterating the residuals of the two-dimensional matrix of seismic data and the sparse sharp noise matrix and the low-rank structural information matrix until the residuals of the two-dimensional matrix of seismic data and the sparse sharp noise matrix and the low-rank structural information matrix meet the set threshold;

[0009] S3. Output the sparse sharp noise matrix and the low-rank structural information matrix respectively. The output sparse sharp noise matrix is ​​the special geological body data, and the low-rank structural information matrix is ​​the remaining stratum background.

[0010] As a limitation: in step S1, the seismic data of the target work area is decomposed into a two-dimensional matrix of seismic data along the longitudinal or transverse slices of the data body, and the special geological body recognition model is:

[0011]

[0012] In the formula, ||A|| * is the nuclear norm of matrix A, ||E||1 is the 1-norm of matrix E, E is the sparse sharp noise matrix, A is the low-rank structural information matrix, D is the two-dimensional matrix of seismic data in the target area; λ is the weight of the sparse sharp noise related to the modeling intensity parameter; Wherein, m is the number of columns of the seismic data matrix of the target work area, n is the number of rows of the seismic data matrix of the target work area, a is the modeling intensity parameter, 0<a<1.

[0013] As a further limitation: in step S2, when the computer hardware performance cannot bear the data volume of the two-dimensional matrix of the seismic data of the target work area, before calculating the low-rank structural information matrix and the sparse sharp noise matrix, the two-dimensional matrix of the seismic data of the target work area is first divided into multiple matrix blocks along the vertical direction, and there is an overlapping part between adjacent matrix blocks. Each matrix block of the seismic data of the target work area is calculated, and each matrix block is decomposed into a sparse sharp noise matrix and a low-rank structural information matrix; when the computer hardware performance can bear the data volume of the two-dimensional matrix of the seismic data of the target work area, the two-dimensional matrix of the seismic data of the target work area is directly decomposed into a sparse sharp noise matrix and a low-rank structural information matrix.

[0014] As a further limitation: Step S3 is specifically as follows: If the two-dimensional matrix of seismic data of the target work area is processed in blocks in step S2, the calculation results of the seismic data of each matrix block obtained in step S2 are superimposed, the weight of the non-overlapping part of each block is 1, the weight of the superimposed part of the left block of adjacent blocks from left to right is 1-g / G, and the weight of the superimposed part of the right block of adjacent blocks from left to right is g / G, G is the number of horizontal sampling points in the overlapping part of adjacent blocks, 0≤g≤G; the sparse sharp noise matrix calculated for each block is multiplied by its corresponding weight and then added to output the sparse sharp noise matrix E, and the low-rank structural information matrix calculated for each block is multiplied by its corresponding weight and then added to output the low-rank structural information matrix A; if the two-dimensional matrix of seismic data of the target work area is not processed in blocks in step S2, the calculated sparse sharp noise matrix is ​​the sparse sharp noise matrix E, and the calculated low-rank structural information matrix is ​​the low-rank structural information matrix A; the output sparse sharp noise matrix E is the special geological body data, and the low-rank structural information matrix A is the residual stratigraphic background.

[0015] As another limitation: In step S2, the low-rank structural information matrix is ​​calculated by the SVD decomposition algorithm, specifically:

[0016] If m≤n, then define the matrix X m*n ,

[0017] X m*n =D m*n -E m*n (1)

[0018] The matrix X m*n Multiply it with its own transposed matrix to get the symmetric matrix C m*m ,

[0019] X m*n *X m*n T =C m*m (2)

[0020] C m*m Do finite eigenvalue decomposition to get the first k main eigenvectors U m*k and the main eigenvalue D k ,

[0021]

[0022] For the main eigenvalue D k Take the square root of each point in and get S k ,

[0023]

[0024] X m*n Transpose and multiply by Um*k , we get V′ n*k ,

[0025] V′ n*k =X m*n T *U m*k (5)

[0026] V′ n*k Divide all n values ​​in each row by S k The number of the corresponding position in

[0027]

[0028] V′ n*k After all rows of V are calculated according to formula (6), V n*k , U m*k Multiply S k Multiply V n*k Transpose, get X m*n The low-rank information matrix

[0029]

[0030] A m*n ,

[0031]

[0032] In the above formula, D k It is C m*m The main eigenvalue of k yes All eigenvalues ​​of U m*k It is C m*m The main eigenvector of U m*k yes All eigenvectors of U m*k yes The left singular vector, S k yes The singular values ​​of V n*k T yes The right singular vectors of V′ n*k Each row has n values, Indicates S k Zhongyu The number of corresponding positions; k = 0.3 × min (m, n), represents the 30% eigenvalues ​​before decomposition;

[0033] If m>n, then define the matrix X n*m ,

[0034] T

[0035] X n*m =(D m*n -E m*n ) (9)

[0036] The matrix X n*m Multiply it with its own transposed matrix to get the symmetric matrix C n*n ,

[0037] X n*m *X n*m T =C n*n (10)

[0038] C n*n Do finite eigenvalue decomposition to get the first k main eigenvectors U n*k and the main eigenvalue D k ,

[0039]

[0040] For the main eigenvalue D k Take the square root of each point in and get S k ,

[0041]

[0042] X n*m Transpose and multiply by U n*k , we get V′ m*k ,

[0043] V′ m*k =X n*m T *U n*k (13)

[0044] V′ m*k Divide all m values ​​in each row by S k The number of the corresponding position in

[0045]

[0046] V′ m*k After all rows of formula (14) are calculated, V m*k , U n*k Multiply S k Multiply V m*k Transpose, get X n*m The low-rank information matrix

[0047]

[0048] The transpose of is A m*n ,

[0049]

[0050] In the above formula, D k It is C n*n The main eigenvalue of k yes All eigenvalues ​​of U n*k It is C n*n The main eigenvector of U n*k yes All eigenvectors of U n*k yes The left singular vector, S k yes The singular values ​​of V m*k T yes The right singular vectors of V′ m*k Each row has m values, Indicates S k Zhongyu The number of corresponding positions; k = 0.3 × min (m, n), represents the 30% eigenvalues ​​before decomposition;

[0051] Calculate the sparse sharp noise matrix E by soft thresholding algorithm m*n , specifically:

[0052] (When P1<0, P1=0; when P1>0, keep the original value of P1)

[0053] (When P2>0, P2=0; when P2<0, keep the original value of P2)

[0054] E m*n =P1+P2

[0055] Among them, S Y is the weighted sum of each round of iteration Y, where Y is the residual, Y = D m*n -A m*n -E m*n ; S0 is D m*n The maximum component of the singular value decomposition, x is the number of iterations;

[0056] By iterating the residuals of the two-dimensional matrix of seismic data and the sparse sharp noise matrix and the low-rank structural information matrix, the low-rank structural information matrix and the sparse sharp noise matrix are updated until the residual Y of the two-dimensional matrix of seismic data and the sparse sharp noise matrix and the low-rank structural information matrix meets the set conditions.

[0057] As a further limitation: the residual Y of the two-dimensional matrix of seismic data and the sparse sharp noise matrix and the low-rank structural information matrix in step S2 satisfies the setting condition: in, S d is the matrix D m*n The standard deviation of each point in the statistics.

[0058] As another limitation: the method further includes step S4, which specifically comprises: resynthesizing the two-dimensional special geological body data and the two-dimensional residual stratum background obtained in step S3 into a three-dimensional special geological body and a three-dimensional residual stratum background respectively.

[0059] Due to the adoption of the above scheme, the present invention has the following beneficial effects compared with the prior art:

[0060] The present invention provides a special geological body identification method based on RPCA background modeling, which decomposes the seismic data of the target work area into a two-dimensional matrix of seismic data along the longitudinal or transverse slices of the data body, and can identify the body attributes or surface attributes of special geological bodies such as fractures, volcanoes, and rivers. When decomposing along the longitudinal slices, it can analyze separately in two directions of the line or combine the results of the two directions for analysis or analyze along a specified arbitrary line direction; by adjusting the modeling intensity parameters, it can be applied to actual seismic data of different work areas and data characteristics; by SVD decomposition of some eigenvalues ​​plus some simple matrix operations, the size and amount of subsequent operation matrices are reduced, thereby reducing the consumption of hardware resources and greatly improving the computing efficiency. Compared with the existing deep learning method, the efficiency, accuracy and applicability of special geological body interpretation are improved, and it has application capabilities for complex geological conditions and low seismic data quality, providing reliable data for subsequent geological modeling and reservoir prediction.

[0061] The invention is suitable for identifying special geological bodies. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0063] Figure 1 A flowchart of a method for identifying a special geological body based on RPCA background modeling according to an embodiment of the present invention;

[0064] Figure 2 The effect of layer leveling on the results of the embodiment of the present invention;

[0065] Figure 3 This is a schematic diagram of a block operation weighted superposition method according to an embodiment of the present invention;

[0066] Figure 4 A schematic diagram showing the influence of the modeling strength parameters on the contrast strength between a special geological body and the remaining stratum background in a seismic data slice according to an embodiment of the present invention;

[0067] Figure 5 A schematic diagram of a partial seismic data slice inputted into a work area and a partial special geological body outputted in a work area according to an embodiment of the present invention;

[0068] Figure 6 It is a schematic diagram of partial seismic data slices input into a work area and partial special geological bodies outputted in a work area according to an embodiment of the present invention. DETAILED DESCRIPTION

[0069] The present invention is further described below in conjunction with embodiments, but those skilled in the art should understand that the present invention is not limited to the following embodiments, and any improvements and equivalent changes made based on the specific embodiments of the present invention are within the scope of protection of the claims of the present invention.

[0070] Embodiment A special geological body identification method based on RPCA background modeling

[0071] A special geological body identification method based on RPCA background modeling, such as Figure 1 As shown, the following steps are performed in sequence:

[0072] S1. Based on the RPCA algorithm, a special geological body recognition model is constructed to decompose the seismic data of the target area along the vertical or horizontal slices of the data body into a two-dimensional matrix D. Before the seismic data of the target area are decomposed along the vertical slices of the data body, they can be flattened along the layers to cope with complex geological conditions, such as Figure 2 As shown in the figure, the two-dimensional matrix D of seismic data of the target work area is input into the special geological body recognition model, and the two-dimensional matrix D is decomposed into a sparse sharp noise matrix E and a low-rank structural information matrix A;

[0073] The special geological body identification model is:

[0074]

[0075] In the formula, ||A|| * is the nuclear norm of matrix A, ||E||1 is the 1-norm of matrix E, E is the sparse sharp noise matrix, A is the low-rank structural information matrix, and D is the seismic data matrix of the target work area; λ is the weight of the sparse sharp noise related to the modeling intensity parameter; Wherein, m is the number of columns of the seismic data matrix of the target work area, n is the number of rows of the seismic data matrix of the target work area, a is the modeling intensity parameter, 0<a<1;

[0076] S2. When the computer hardware performance cannot bear the data volume of the two-dimensional matrix of seismic data of the target work area, the two-dimensional matrix of seismic data of the target work area is divided into blocks along the vertical direction. The block division method is as follows: Figure 3 As shown, it is divided into multiple matrix blocks. The part between adjacent solid lines is a matrix block, and the part between adjacent dotted lines is a matrix block. There is a 50% overlap between adjacent matrix blocks. Each matrix block of the seismic data of the target work area is calculated. Figure 3 The dashed and solid lines in the upper part represent the values Figure 3 The weights corresponding to the lower matrix blocks, each matrix block is decomposed into a sparse sharp noise matrix and a low-rank structural information matrix; when the computer hardware performance can bear the data volume of the two-dimensional matrix of seismic data in the target work area, the two-dimensional matrix of seismic data in the target work area is directly decomposed into a sparse sharp noise matrix and a low-rank structural information matrix;

[0077] The low-rank structural information matrix is ​​calculated by the SVD decomposition algorithm, specifically:

[0078] If m≤n, then define the matrix X m*n ,

[0079] X m*n =D m*n -E m*n (1)

[0080] The matrix X m*n Multiply it with its own transposed matrix to get the symmetric matrix C m*m ,

[0081] X m*n *X m*n T =C m*m (2)

[0082] C m*m Do finite eigenvalue decomposition to get the first k main eigenvectors U m*k and the main eigenvalue D k ,

[0083]

[0084] For the main eigenvalue D k Take the square root of each point in and get S k ,

[0085]

[0086] X m*nTranspose and multiply by U m*k , we get V′ n*k ,

[0087] V′ n*k =X m*n T *U m*k (5)

[0088] V′ n*k Divide all n values ​​in each row by S k The number of the corresponding position in

[0089]

[0090] V′ n*k After all rows of V are calculated according to formula (6), V n*k , U m*k Multiply S k Multiply V n*k Transpose, get X m*n The low-rank information matrix

[0091]

[0092] A m*n ,

[0093]

[0094] In the above formula, D k It is C m*m The main eigenvalue of k yes All eigenvalues ​​of U m*k It is C m*m The main eigenvector of U m*k yes All eigenvectors of U m*k yes The left singular vector, S k yes The singular values ​​of V n*k T yes The right singular vectors of V′ n*k Each row has n values, Indicates S k Zhongyu The number of corresponding positions; k = 0.3 × min (m, n), represents the 30% eigenvalues ​​before decomposition;

[0095] If m>n, then define the matrix X n*m ,

[0096] T

[0097] X n*m =(D m*n -E m*n ) (9)

[0098] The matrix X n*m Multiply it with its own transposed matrix to get the symmetric matrix C n*n ,

[0099] X n*m *X n*m T =C n*n (10)

[0100] C n*n Do finite eigenvalue decomposition to get the first k main eigenvectors U n*k and the main eigenvalue D k ,

[0101]

[0102] For the main eigenvalue D k Take the square root of each point in and get S k ,

[0103]

[0104] X n*m Transpose and multiply by U n*k , we get V′ m*k ,

[0105] V′ m*k =X n*m T *U n*k (13)

[0106] V′ m*k Divide all m values ​​in each row by S k The number of the corresponding position in

[0107]

[0108] V′ m*k After all rows of formula (14) are calculated, V m*k , U n*k Multiply S k Multiply V m*k Transpose, get X n*m The low-rank information matrix

[0109]

[0110] The transpose of is A m*n ,

[0111]

[0112] In the above formula, D k It is C n*n The main eigenvalue of k yes All eigenvalues ​​of U n*k It is C n*n The main eigenvector of U n*k yes All eigenvectors of U n*k yes The left singular vector, S k yes The singular values ​​of V m*k T yes The right singular vectors of V′ m*k Each row has m values, Indicates S k Zhongyu The number of corresponding positions; k = 0.3 × min (m, n), represents the 30% eigenvalues ​​before decomposition;

[0113] Calculate the sparse sharp noise matrix E by soft thresholding algorithm m*n , specifically:

[0114] (When P1<0, P1=0; when P1>0, keep the original value of P1)

[0115] (When P2>0, P2=0; when P2<0, keep the original value of P2)

[0116] E m*n =P1+P2

[0117] Among them, S Y is the weighted sum of each round of iteration Y, where Y is the residual, Y = D m*n -A m*n -E m*n ; S0 is D m*n The maximum component of the singular value decomposition, x is the number of iterations;

[0118] By iterating the residuals of the two-dimensional matrix of seismic data, the sparse sharp noise matrix, and the low-rank structural information matrix, the low-rank structural information matrix and the sparse sharp noise matrix are updated until the residuals of the two-dimensional matrix of seismic data, the sparse sharp noise matrix, and the low-rank structural information matrix meet in, S d is the matrix D m*n The standard deviation of each point is obtained by counting;

[0119] S3. If the two-dimensional matrix of seismic data of the target work area is processed in blocks in step S2, the calculation results of the seismic data of each matrix block obtained in step S2 are superimposed, the weight of the non-overlapping part of each block is 1, the weight of the superimposed part of the left block of the adjacent block from left to right is 1-g / G, and the weight of the superimposed part of the right block of the adjacent block from left to right is g / G, where G is the number of horizontal sampling points of the overlapping part of the adjacent blocks, 0≤g≤G; the sparse sharp noise matrix E calculated for each block m*n Multiply the corresponding weights and then add the corresponding output sparse sharp noise matrix E, and the low-rank structural information matrix A calculated by each block m*n Multiply the corresponding weights and then add them up to output the low-rank structural information matrix A; if the two-dimensional matrix of seismic data of the target work area is not processed in blocks in step S2, the sparse sharp noise matrix E m*n That is, the sparse sharp noise matrix E, the calculated low-rank structural information matrix A m*n That is the low-rank structural information matrix A; the output sparse sharp noise matrix E is the special geological body data, and the low-rank structural information matrix A is the remaining stratum background;

[0120] S4, resynthesizing the two-dimensional special geological body data and the two-dimensional remaining stratum background obtained in step S3 into a three-dimensional special geological body and a three-dimensional remaining stratum background respectively.

[0121] The special geological body identification method based on RPCA background modeling of this embodiment is applied to a certain work area to identify special geological bodies. First, the seismic data slices of the work area are input, and the modeling intensity parameter a is continuously adjusted until the contrast intensity between the special geological body and the remaining stratigraphic background in the seismic data slices of the work area meets the requirements of the interpreter. The effect of adjusting the modeling intensity parameter a on the contrast intensity between the special geological body and the remaining stratigraphic background is shown in the following figure. Figure 4 As shown, according to the above steps, the special geological body data and the remaining stratum background are output. Figure 5 The upper part is the input seismic data slice of the work area, and the lower part is the output of some special geological bodies, the special geological bodies are fractures and caves. Figure 6The left half is the input seismic data slice of part of the work area, and the right half is the output part of the special geological body, which is a volcano. Finally, the calculated two-dimensional special geological body data and the two-dimensional residual stratigraphic background are resynthesized into three-dimensional special geological body and three-dimensional residual stratigraphic background respectively, and displayed in three dimensions through GeoEast software, and the display quality control of the special geological body is carried out to obtain the interpretation results.

Claims

1. A special geological body identification method based on RPCA background modeling, It is characterized in that The process includes the following steps: S1. Based on the RPCA algorithm, a special geological body recognition model is constructed. The seismic data of the target work area is decomposed into a two-dimensional matrix of seismic data, which is input into the special geological body recognition model. The two-dimensional matrix of seismic data is decomposed into a sparse sharp noise matrix and a low-rank structural information matrix. S2. Calculate the low-rank structural information matrix by SVD decomposition algorithm; calculate the sparse sharp noise matrix by soft threshold algorithm; update the low-rank structural information matrix and the sparse sharp noise matrix by iterating the residuals of the two-dimensional matrix of seismic data and the sparse sharp noise matrix and the low-rank structural information matrix until the residuals of the two-dimensional matrix of seismic data and the sparse sharp noise matrix and the low-rank structural information matrix meet the set conditions; S3. Output the sparse sharp noise matrix and the low-rank structural information matrix respectively. The output sparse sharp noise matrix is ​​the special geological body data, and the low-rank structural information matrix is ​​the remaining stratum background.

2. A special geological body identification method based on RPCA background modeling according to claim 1, It is characterized in that In step S1, the seismic data of the target work area is decomposed into a two-dimensional matrix of seismic data along the vertical or horizontal slices of the data body. The special geological body recognition model is: In the formula, ||A|| * is the nuclear norm of matrix A, ||E|| 1 is the 1-norm of matrix E, E is the sparse sharp noise matrix, A is the low-rank structural information matrix, and D is the two-dimensional matrix of seismic data in the target work area; λ is the weight of the sparse sharp noise associated with the modeled intensity parameter; Wherein, m is the number of columns of the seismic data matrix of the target work area, n is the number of rows of the seismic data matrix of the target work area, a is the modeling intensity parameter, 0<a<1.

3. A special geological body identification method based on RPCA background modeling according to claim 2, It is characterized in that In step S2, when the computer hardware performance cannot bear the data volume of the two-dimensional matrix of the seismic data of the target work area, before calculating the low-rank structural information matrix and the sparse sharp noise matrix, the two-dimensional matrix of the seismic data of the target work area is first divided into multiple matrix blocks along the vertical direction, and there is an overlapping part between adjacent matrix blocks. Each matrix block of the seismic data of the target work area is calculated, and each matrix block is decomposed into a sparse sharp noise matrix and a low-rank structural information matrix; when the computer hardware performance can bear the data volume of the two-dimensional matrix of the seismic data of the target work area, the two-dimensional matrix of the seismic data of the target work area is directly decomposed into a sparse sharp noise matrix and a low-rank structural information matrix.

4. A special geological body identification method based on RPCA background modeling according to claim 3, It is characterized in that Step S3 is specifically as follows: if the two-dimensional matrix of seismic data of the target work area is processed in blocks in step S2, the calculation results of the seismic data of each matrix block obtained in step S2 are superimposed, the weight of the non-overlapping part of each block is 1, the weight of the superimposed part of the left block of adjacent blocks from left to right is 1-g / G, and the weight of the superimposed part of the right block of adjacent blocks from left to right is g / G, G is the number of horizontal sampling points in the overlapping part of adjacent blocks, 0≤g≤G; the sparse sharp noise matrix calculated by each block is multiplied by its corresponding weight and then added to output the sparse sharp noise matrix E, and the low-rank structural information matrix calculated by each block is multiplied by its corresponding weight and then added to output the low-rank structural information matrix A; if the two-dimensional matrix of seismic data of the target work area is not processed in blocks in step S2, the calculated sparse sharp noise matrix is ​​the sparse sharp noise matrix E, and the calculated low-rank structural information matrix is ​​the low-rank structural information matrix A; the output sparse sharp noise matrix E is the special geological body data, and the low-rank structural information matrix A is the residual stratigraphic background.

5. A special geological body identification method based on RPCA background modeling according to any one of claims 1 to 4, It is characterized in that In step S2, the low-rank structural information matrix is ​​calculated by the SVD decomposition algorithm, specifically: If m≤n, then define the matrix X m*n , X m*n =D m*n -E m*n (1) The matrix X m*n Multiply it with its own transposed matrix to get the symmetric matrix C m*m , X m*n *X m*n T =C m*m (2) C m*m Do finite eigenvalue decomposition to get the first k main eigenvectors U m*k and the main eigenvalue D k , For the main eigenvalue D k Take the square root of each point in and get S k , X m*n Transpose and multiply by U m*k , we get V′ n*k , V′ n*k =X m*n T *U m*k (5) V′ n*k Divide all n values ​​in each row by S k The number of the corresponding position in V′ n*k After all rows of V are calculated according to formula (6), V n*k , U m*k Multiply S k Multiply V n*k Transpose, get X m*n The low-rank information matrix A m*n , In the above formula, D k It is C m*m The main eigenvalue of k yes All eigenvalues ​​of U m*k It is C m*m The main eigenvector of U m*k yes All eigenvectors of U m*k yes The left singular vector, S k yes The singular values ​​of V n*k T yes The right singular vectors of V′ n*k Each row has n values, Indicates S k Zhongyu The number of corresponding positions; k = 0.3 × min (m, n), represents the 30% eigenvalues ​​before decomposition; If m>n, then define the matrix X n*m , X n*m = (D m*n -E m*n ) T (9) The matrix X n*m Multiply it with its own transposed matrix to get the symmetric matrix C n*n , X n*m *X n*m T =C n*n (10) C n*n Do finite eigenvalue decomposition to get the first k main eigenvectors U n*k and the main eigenvalue D k , For the main eigenvalue D k Take the square root of each point in and get S k , X n*m Transpose and multiply by U n*k , we get V′ m*k , V′ m*k =X n*m T *U n*k (13) V′ m*k Divide all m values ​​in each row by S k The number of the corresponding position in V′ m*k After all rows of formula (14) are calculated, V m*k , U n*k Multiply S k Multiply V m*k Transpose, get X n*m The low-rank information matrix The transpose of is A m*n , In the above formula, D k It is C n*n The main eigenvalue of k yes All eigenvalues ​​of U n*k It is C n*n The main eigenvector of U n*k yes All eigenvectors of U n*k yes The left singular vector, S k yes The singular values ​​of V m*k T yes The right singular vectors of V′ m*k Each row has m values, Indicates S k Zhongyu The number of corresponding positions; k = 0.3 × min (m, n), represents the 30% eigenvalues ​​before decomposition; Calculate the sparse sharp noise matrix E by soft thresholding algorithm m*n , specifically: (P 1 <0 time P 1 =0,P 1 >0 when P is retained 1 Original value) (P 2 >0 time P 2 =0,P 2 <0, keep P 2 Original value) E m*n =P 1 +P 2 Among them, S Y is the weighted sum of each round of iteration Y, where Y is the residual, Y = D m*n -A m*n -E m*n ; S 0 D m*n The maximum component of the singular value decomposition, x is the number of iterations; By iterating the residuals of the two-dimensional matrix of seismic data and the sparse sharp noise matrix and the low-rank structural information matrix, the low-rank structural information matrix and the sparse sharp noise matrix are updated until the residual Y of the two-dimensional matrix of seismic data and the sparse sharp noise matrix and the low-rank structural information matrix meets the set conditions.

6. A special geological body identification method based on RPCA background modeling according to claim 5, It is characterized in that In step S2, the residual Y between the two-dimensional matrix of seismic data and the sparse sharp noise matrix and the low-rank structural information matrix satisfies the setting condition: in, S d is the matrix D m*n The standard deviation of each point in the statistics.

7. A special geological body identification method based on RPCA background modeling according to any one of claims 1 to 4 and 6, It is characterized in that The method further comprises step S4, which specifically comprises: resynthesizing the two-dimensional special geological body data and the two-dimensional residual stratum background obtained in step S3 into a three-dimensional special geological body and a three-dimensional residual stratum background respectively.

8. The special geological body identification method based on RPCA background modeling according to claim 5, It is characterized in that The method further comprises step S4, which specifically comprises: resynthesizing the two-dimensional special geological body data and the two-dimensional residual stratum background obtained in step S3 into a three-dimensional special geological body and a three-dimensional residual stratum background respectively.

Citation Information

Patent Citations

  • Seismic attribute multi-source information fusion method

    CN103837889A

  • Seismic fault enhancement method based on Ho-RPCA

    CN109471164A

  • Shale reservoir fracture identification method and system cooperating with variant ant body

    CN112444859A

  • Intelligent road cavity online detection system and detection method

    CN115220036A

  • Early warning-oriented strong earthquake magnitude sparse classification method

    CN115577250A