Mode component analysis seismic abnormal amplitude suppression method and device based on pattern recognition, and storage medium

Abstract

The invention discloses a form component analysis seismic abnormal amplitude suppression method and device based on pattern recognition and a storage medium, and belongs to the technical field of oil exploration seismic data processing. According to the seismic abnormal amplitude suppression method based on morphological component analysis of pattern recognition provided by the invention, a proper sparse dictionary is found for seismic data containing abnormal amplitude by using a morphological component analysis technology, so that the coefficient of the seismic data in the dictionary is sparse; separating the seismic data into different signal components, namely effective signals and noise (abnormal amplitude) according to the difference between the seismic data by using a signal separation theory, calculating the similarity between the background record containing the abnormal amplitude and the separated noise through pattern recognition, and determining the reliability and suppression degree of the separated abnormal amplitude, so as to achieve the purpose of noise suppression. Finally, large-range and strong-energy abnormal amplitude of seismic data can be effectively suppressed, and effective reflection signals can be protected to the greatest extent. The method is suitable for suppressing large-area and strong-energy seismic abnormal amplitude, the signal-to-noise ratio of seismic data can be effectively improved, and the processing precision of the seismic data is improved.

Description

Technical Field

[0001] This invention belongs to the field of seismic data processing technology for petroleum exploration, and relates to a method for suppressing seismic anomaly amplitudes. Specifically, it is a method, device and storage medium for suppressing seismic anomaly amplitudes based on pattern recognition morphological component analysis. Background Technology

[0002] As oil and gas exploration and development place increasingly higher demands on detailed imaging of complex subsurface structures and media, the signal-to-noise ratio (SNR) of seismic data becomes increasingly critical. Therefore, improving the SNR of seismic data remains a hot topic in seismic data processing research. In particular, anomalous amplitudes in raw seismic data, such as strong energy anomalies generated by strong field vibrations, can mask effective reflection records, reduce the SNR, and cause an imbalance in amplitude energy in the longitudinal and transverse directions of the seismic trace. This can lead to arcing phenomena during pre-stack migration, thus affecting seismic data processing and geological interpretation.

[0003] Accurately identifying anomalous amplitude noise and separating it from the effective signal are common methods for suppressing anomalous amplitudes. However, existing methods, such as manual trace editing, frequency division suppression, two-dimensional median filtering, and adaptive amplitude suppression, are only suitable for situations where the noise pattern does not change significantly with time and space. Anomalous amplitude noise signals typically exhibit substantial changes in time and space, thus these methods are prone to losing effective signals during noise suppression, exhibiting limitations. Furthermore, they are ineffective when large-area, high-energy anomalous amplitudes appear in seismic data. Therefore, it is necessary to research methods suitable for suppressing large-area, high-energy anomalous amplitudes to maximize the signal-to-noise ratio and processing effectiveness of seismic data. Summary of the Invention

[0004] The purpose of this invention is to provide a method and apparatus for suppressing seismic anomaly amplitudes based on pattern recognition morphological component analysis, in order to solve the problem of reduced signal-to-noise ratio and deteriorated processing effect of seismic data caused by large-area, high-energy anomaly amplitudes.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] A method for suppressing seismic anomaly amplitude based on pattern recognition morphological component analysis includes the following steps:

[0007] S1. Use morphological component analysis to obtain noise-free signal vectors and pseudo-inverse matrices of noise matrices from seismic data containing anomalous amplitudes;

[0008] S2. Calculate the noise component using the following formula:

[0009]

[0010] In the formula, x(t) represents seismic data containing anomalous amplitudes; s(t) represents a noise-free signal vector; A + The pseudo-inverse of the noise matrix is ​​n(t); n(t) is the noise signal; k i n i (t) represents the noise component;

[0011] S3. Perform pattern recognition calculations on the noise components and the background recording data containing anomalous amplitudes to obtain the AAAS exponential function:

[0012]

[0013] In the formula, N is the number of samples of the noise component or background record containing abnormal amplitude; l(P,Z) is the brightness coefficient; α is the brightness weighting index; c(P,Z) is the contrast coefficient; β is the contrast weighting index; s(P,Z) is the structural measurement coefficient; γ is the structural measurement weighting index.

[0014]

[0015] In the formula, u p The average amplitude value of the noise component; u z C1 is the average amplitude value of the background record containing abnormal amplitudes; C2 is the brightness damping coefficient.

[0016]

[0017] In the formula, σ p σ represents the variance of the noise components. z C1 represents the variance of the background record containing anomalous amplitudes; C2 is the contrast damping coefficient.

[0018]

[0019] In the formula, σ p σ represents the variance of the noise components. z σ represents the variance of the background records containing anomalous amplitudes. pz C1 is the covariance between the noise component and the background record containing anomalous amplitude; C2 is the structural measurement damping coefficient.

[0020] S4. By suppressing the noise component using the AAAS exponential function, the seismic data after suppressing abnormal amplitudes is obtained, thus completing the abnormal amplitude suppression process:

[0021]

[0022] In the formula, x'(t) represents the seismic data after suppressing anomalous amplitudes; s(t) is the noise-free signal vector; M is the number of morphological components; f(AAAS) i) is the effective information extraction function, which extracts the effective information from the noise components that have been incorrectly suppressed and adds it back into the final result; k i n i (t) represents the noise component.

[0023] As a limitation, step S1 specifically includes:

[0024] S11. Obtain a sparse dictionary D from seismic data containing anomalous amplitudes using morphological component analysis. K and the corresponding sparsity coefficient a k ;

[0025] S12. Calculation Initial Lagrange multipliers are set based on experience, and the coefficients a are updated according to the following formula. opt :

[0026]

[0027] In the formula, a opt A is the effective signal under the optimal Lagrange multipliers; λ is the Lagrange multiplier; A K + D is the pseudo-inverse of the noise matrix; K For a sparse dictionary; ||·||1 is the L1 norm; a k is the sparse coefficient; A is the noise matrix; arg min is the value of the independent variable when the function is at its minimum; Thresh is the convergence threshold identifier;

[0028] S13. Update the noise matrix A and the Lagrange multiplier λ according to the following formulas, and normalize the noise matrix A to obtain the noise-free signal vector and the pseudo-inverse matrix of the noise matrix:

[0029]

[0030] s(t)=a opt

[0031] In the formula, ||·||1 is the L1 norm; a k λ is the sparsity coefficient; λ is the Lagrange multiplier; D K A is a sparse dictionary; A is a noise matrix; arg min is the value of the independent variable when the function reaches its minimum value; a opt s(t) is the effective signal under the optimal Lagrange multipliers; s(t) is the noise-free signal vector.

[0032] As a further definition, the sparse dictionary is a combination of Shearlet transform, discrete cosine transform, and Curvelet dictionary.

[0033] The present invention also discloses a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-described method for suppressing seismic anomaly amplitude based on pattern recognition morphological component analysis.

[0034] The present invention also discloses a computer-readable storage medium storing a computer program that performs the above-described method for suppressing seismic anomaly amplitude based on pattern recognition morphological component analysis.

[0035] By adopting the above technical solution, the technical progress achieved by this invention compared with the prior art is as follows:

[0036] This invention provides a method for suppressing seismic anomalous amplitudes based on pattern recognition morphological component analysis. By utilizing morphological component analysis technology, a suitable sparse dictionary is found for seismic data containing anomalous amplitudes, making the coefficients of the seismic data sparse under this dictionary. Then, using signal separation theory, the seismic data is separated into different signal components according to their differences, namely effective signal and noise (anomalous amplitude). Finally, the similarity between the background record containing anomalous amplitudes and the separated noise components is calculated by pattern recognition to determine the reliability and suppression degree of the separated anomalous amplitudes. Ultimately, this method can effectively suppress large-scale, high-energy anomalous amplitudes in seismic data while maximizing the protection of effective reflection signals, thereby effectively improving the signal-to-noise ratio of seismic data and improving the accuracy of seismic data processing. Attached Figure Description

[0037] Figure 1 This is a diagram illustrating the effect of the pattern recognition-based morphological component analysis method for suppressing seismic anomaly amplitudes in Example 1. Figure 1 'a' represents the seismic data before the suppression of abnormal amplitude. Figure 1 b represents the seismic data after suppressing abnormal amplitude. Detailed Implementation

[0038] The present invention will be further described in detail below through specific embodiments. It should be understood that the described embodiments are only for explaining the present invention and do not limit the present invention.

[0039] Example 1

[0040] This embodiment discloses a method for suppressing seismic anomaly amplitude based on pattern recognition morphological component analysis, specifically including the following steps performed sequentially:

[0041] S1. Using morphological component analysis, obtain the noise-free signal vector and the pseudo-inverse matrix of the noise matrix from the seismic data containing anomalous amplitudes.

[0042] S11. Obtain a sparse dictionary D from seismic data containing anomalous amplitudes using morphological component analysis.K and the corresponding sparsity coefficient a k

[0043] According to morphological component analysis theory, seismic data can be viewed as a linear combination of multiple components with different morphologies. Each component has a dictionary that can only sparsely represent the corresponding morphological component, but not the other components. Given a 2D or 3D seismic dataset containing anomalous amplitudes, assuming it consists of a linear combination of K components with different morphologies, its sparse representation is as follows:

[0044]

[0045] In the formula, x(t) represents seismic data containing anomalous amplitudes; S K For the Kth morphological component; D K For a sparse dictionary; a k The sparsity coefficient;

[0046] Read in seismic data containing anomalous amplitudes and obtain its sparse dictionary. Typically, Shearlet transform dictionaries, Discrete Cosine Transform (DCT) dictionaries, Curvelet dictionaries, Dirac dictionaries, Fourier dictionaries, wavelet dictionaries, and wavelet packet dictionaries are used in combination. This embodiment uses a combination of Shearlet transform, Discrete Cosine Transform, and Curvelet dictionaries to obtain the sparse dictionary: D K =[D1,D2,…,D K ], and the corresponding sparsity coefficient a k .

[0047] S12. Calculation The initial Lagrange multipliers λ are set empirically, and the coefficients a are updated according to the following formula. opt :

[0048]

[0049] In the formula, a opt For λ = λ K Effective signal under optimal Lagrange multipliers; λ K A is the optimal Lagrange multiplier; K + D is the pseudo-inverse of the noise matrix; K For a sparse dictionary; ||·||1 is the L1 norm; a k λ is the sparsity coefficient; λ is the Lagrange multiplier; D K A is a sparse dictionary; A is a noise matrix; arg min is the value of the independent variable when the function is at its minimum; Thresh is the convergence threshold identifier;

[0050] S13. Update the noise matrix A and the Lagrange multiplier λ according to the following formulas, and normalize the noise matrix A to obtain the noise-free signal vector s(t) and the pseudo-inverse matrix A of the noise matrix. + :

[0051]

[0052] s(t)=a opt

[0053] In the formula, ||·||1 is the L1 norm; a k λ is the sparsity coefficient; λ is the Lagrange multiplier; D K A is a sparse dictionary; A is a noise matrix; arg min is the value of the independent variable when the function reaches its minimum value; a opt For λ = λ K The effective signal under the optimal Lagrange multipliers; s(t) is a noise-free signal vector.

[0054] S2. For seismic data x(t) containing anomalous amplitudes, it can be regarded as a linear superposition of noise-free and noise signals, that is, the noise component can be calculated using the following formula:

[0055]

[0056] In the formula, x(t) represents seismic data containing anomalous amplitudes; s(t) represents a noise-free signal vector; A + The pseudo-inverse of the noise matrix is ​​n(t); n(t) is the noise signal; k i n i (t) represents the noise component;

[0057] S3. Perform pattern recognition calculations on the noise components and background recording data containing anomalous amplitudes to obtain the AAAS exponential function.

[0058] In actual noise separation, the process is not as ideal as the formula above suggests. Especially when seismic signals are mixed with large-area, high-energy anomalous amplitude interference, the noise signal calculated according to morphological component analysis theory may contain effective reflection signals. If the noise components are directly suppressed in this case, the effective signals will be lost. Therefore, it is necessary to further confirm the predicted noise components to avoid the problem of excessive effective information in the noise components being incorrectly removed. Therefore, it is necessary to perform pattern recognition calculations on multiple predicted noise components and background record data containing anomalous amplitudes to confirm the specific details of the anomalous amplitudes contained therein. The algorithm is as follows:

[0059] (1) Luminance coefficient:

[0060] The predicted average amplitude of the noise component is:

[0061]

[0062] The average amplitude of the background recorded data containing anomalous amplitudes is:

[0063]

[0064] In the formula, u p The average amplitude value of the noise component; u z P represents the average amplitude value of the background record containing anomalous amplitude; N is the noise component or the number of samples in the background record containing anomalous amplitude; i Z represents the amplitude value of each sampling point of the noise component. i Record the amplitude values ​​of each sampling point for a background containing abnormal amplitudes;

[0065] Therefore, the luminance coefficient is:

[0066]

[0067] In the formula, u p The average amplitude value of the noise component; u z C1 is the average amplitude value of the background record containing abnormal amplitudes; C2 is the brightness damping coefficient.

[0068] (2) Contrast coefficient:

[0069] The variance of the predicted noise component is:

[0070]

[0071] The variance of the background recorded data containing anomalous amplitudes is:

[0072]

[0073] In the formula, σ p σ represents the variance of the noise components. z P represents the variance of the background records containing anomalous amplitudes; N is the number of noise components or the number of samples in the background records containing anomalous amplitudes; i u represents the amplitude value of each sampling point of the noise component. p Z represents the average amplitude of the noise component. i Record the amplitude values ​​of each sampling point for a background containing abnormal amplitudes; u z The average amplitude value of the background records containing abnormal amplitudes;

[0074] Therefore, the contrast ratio is:

[0075]

[0076] In the formula, σ p σ represents the variance of the noise components.z C1 represents the variance of the background record containing anomalous amplitudes; C2 is the contrast damping coefficient.

[0077] (3) Structural measurement coefficients:

[0078] The covariance between the predicted noise component and the background recorded data containing anomalous amplitudes is:

[0079]

[0080] In the formula, σ PZ is the covariance between the noise component and the background record containing anomalous amplitude; N is the number of samples of the noise component or the background record containing anomalous amplitude; Pi is the amplitude value of each sample of the noise component; u p Z represents the average amplitude of the noise component. i Record the amplitude values ​​of each sampling point for a background containing abnormal amplitudes; u z The average amplitude value of the background records containing abnormal amplitudes;

[0081] Therefore, the structural measurement coefficient is:

[0082]

[0083] In the formula, σ p σ represents the variance of the noise components. z σ represents the variance of the background records containing anomalous amplitudes. pz C1 is the covariance between the noise component and the background record containing anomalous amplitude; C2 is the structural measurement damping coefficient.

[0084] Then, by combining the three functions, we obtain the AAAS (Adaptive Abnormal Amplitude Attenuation Suppression) exponential function for the noise component and the background recorded data containing anomalous amplitudes:

[0085]

[0086] In the formula, N is the number of samples of the noise component or background record containing abnormal amplitude; l(P,Z) is the brightness coefficient; α is the brightness weighting index; c(P,Z) is the contrast coefficient; β is the contrast weighting index; s(P,Z) is the structural measurement coefficient; γ is the structural measurement weighting index.

[0087] S4. By suppressing the noise component using the AAAS exponential function, the seismic data after suppressing anomalous amplitudes is obtained, thus completing the anomalous amplitude suppression process. The results are as follows: Figure 1 As shown:

[0088]

[0089] In the formula, x'(t) represents the seismic data after suppressing anomalous amplitudes; s(t) is the noise-free signal vector; M is the number of morphological components; f(AAAS) i ) is the effective information extraction function, which extracts the effective information from the noise components that have been incorrectly suppressed and adds it back into the final result; k i n i (t) represents the noise component.

[0090] contrast Figure 1 a and Figure 1 As shown in b, after suppressing the abnormal amplitude, the abnormal amplitude noise on the right side of the figure is well suppressed, and the abnormal amplitude noise on the left side is not only well suppressed, but the effective signal at the bottom of the abnormal amplitude on the left side is also well preserved.

[0091] Example 2

[0092] This embodiment provides a computer device, which includes a memory, a processor, and a computer program stored in the memory and capable of running on the processor, to implement the above-described method for suppressing seismic anomaly amplitude based on pattern recognition morphological component analysis.

[0093] This memory is used to store non-transitory computer-readable instructions. Specifically, the memory may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may, for example, include random access memory (RAM) and / or cache memory. The non-volatile memory may, for example, include read-only memory (ROM), hard disk, flash memory, etc.

[0094] The processor may be a central processing unit (CPU) or other form of processing unit with data processing and / or instruction execution capabilities, and may control other components in the electronic device to perform desired functions. The processor is used to execute computer-readable instructions stored in the memory.

[0095] Those skilled in the art will understand that, in order to solve the technical problem of how to achieve a good user experience, this embodiment may also include well-known structures such as communication buses and interfaces, and these well-known structures should also be included within the protection scope of this disclosure.

[0096] For a detailed description of this embodiment, please refer to the corresponding descriptions in the foregoing embodiments, which will not be repeated here.

[0097] Example 3

[0098] This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned method for suppressing seismic anomaly amplitude based on pattern recognition morphological component analysis.

[0099] The computer-readable storage medium stores non-transitory computer-readable instructions thereon. When the non-transitory computer-readable instructions are executed by a processor, all or part of the steps of the methods of the foregoing embodiments are performed.

[0100] The aforementioned computer-readable storage media include, but are not limited to: optical storage media (e.g., CD-ROM and DVD), magneto-optical storage media (e.g., MO), magnetic storage media (e.g., magnetic tape or portable hard drive), media with built-in rewritable non-volatile memory (e.g., memory card), and media with built-in ROM (e.g., ROM cartridge).

Claims

1. A method for seismic anomaly amplitude suppression based on pattern recognition of morphological components, characterized in that, The method comprises the following steps: S1. obtaining a noise-free signal vector and a pseudo-inverse matrix of a noise matrix by using a morphological component analysis method on seismic data containing abnormal amplitudes; S2. calculating a noise component by using the following formula: where x(t) is seismic data containing abnormal amplitude; s(t) is a signal vector without noise; A + is a pseudo-inverse matrix of noise matrix; n(t) is a noise signal; k i n i (t) is a noise component; S3. calculating a pattern recognition on the noise component and background record data containing abnormal amplitudes to obtain an AAAS exponential function: In the formula, N is the number of samples of the noise component or the background record containing abnormal amplitudes; l(P, Z) is a brightness coefficient; a is a brightness weight index; c(P, Z) is a contrast coefficient; β is a contrast weight index; s(P, Z) is a structure measurement coefficient; and γ is a structure measurement weight index; where u p is the average amplitude value of the noise component; u z is the average amplitude value of the background record containing the abnormal amplitude; C1 is the luminance damping coefficient; where σ p is the variance of the noise component; σ z is the variance of the background record containing the anomalous amplitude; and C2 is a contrast damping coefficient. where σ p is the variance of the noise component; σ z is the variance of the background record containing the anomalous amplitude; σ pz is the covariance of the noise component and the background record containing the anomalous amplitude; and C3 is a structure measurement damping coefficient. S4. suppressing the noise component according to the AAAS exponential function to obtain seismic data after suppressing abnormal amplitudes, that is, the abnormal amplitude suppression processing is completed: In the formula, x'(t) represents the seismic data after suppressing anomalous amplitudes; s(t) is the noise-free signal vector; M is the number of morphological components; f(AAAS) i ) is the effective information extraction function; k i n i (t) represents the noise component.

2. The method according to claim 1, wherein, The step S1 specifically comprises: S11. Obtain sparse dictionary D using morphological component analysis on seismic data containing abnormal amplitude K and corresponding sparse coefficients a k ; S12. Compute The initial Lagrange multiplier is set empirically, and the coefficient a is updated according to the following formula opt : where a opt is the effective signal under the optimal Lagrange multiplier; λ is the Lagrange multiplier; A K + is the pseudo-inverse matrix of the noise; D K is the sparse dictionary; ||·||1 is the L1 norm; a k is the sparse coefficient; A is the noise matrix; arg min is the argument value when the function value is the minimum; Thresh is the convergence threshold identifier; S13. updating the noise matrix A and the Lagrange multiplier λ respectively according to the following formula, normalizing the noise matrix A, and obtaining a noise-free signal vector and a pseudo-inverse matrix of a noise matrix: s(t) = a opt where ||·||1 is the L1 norm; a k is a sparse coefficient; λ is a Lagrange multiplier; D K is a sparse dictionary; A is a noise matrix; arg min is the argument of the minimum value of a function; a opt is the effective signal under the optimal Lagrange multiplier; s(t) is a signal vector without noise.

3. The method according to claim 2, wherein the method is a pattern recognition based morphological component analysis seismic anomaly amplitude suppression method, characterized in that, The sparse dictionary is a combination of a Shearlet transform, a discrete cosine transform and a Curvelet dictionary.

4. A computer device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor implements the pattern recognition based morphological component analysis seismic abnormal amplitude suppression method in any one of claims 1-3 when executing the computer program.

5. A computer readable storage medium, characterized in that, The computer readable storage medium stores the computer program for executing the pattern recognition based morphological component analysis seismic abnormal amplitude suppression method in any one of claims 1-3.

Images