A frequency space domain eigen-decomposition denoising method, device and analysis method

Abstract

The application belongs to the technical field of oil and gas exploration, and discloses a frequency space domain feature decomposition denoising method, a device and an analysis method. The denoising method comprises the following steps: transforming original time-space domain data to a frequency space domain; setting a frequency processing range; randomly extracting a part of proportional data as data analysis points on a frequency slice data, taking a rectangular window to expand each data analysis point into a feature analysis sample, traversing all data analysis points to obtain a feature analysis sample set; calculating a covariance matrix of the feature analysis sample set matrix, and performing feature decomposition to obtain a feature vector of an effective signal; taking a rectangular window and expanding the whole frequency slice data point by point, and calculating the effective signal feature vector through inner product to obtain the frequency slice data after denoising; and transforming the seismic effective signal in the frequency space domain back to a time-space domain to obtain the seismic data after final denoising. The application can realize amplitude-preserving denoising of the seismic signal.

Description

Technical Field

[0001] This invention belongs to the field of oil and gas exploration technology, specifically relating to a feature decomposition and denoising method, apparatus and analysis method in the frequency spatial domain. Background Technology

[0002] During seismic data acquisition, numerous complex factors, including the source, excitation method, medium characteristics at the excitation point, receiving environment, and surface and subsurface medium structures, often result in significant noise in the data, reducing the accuracy of seismic interpretation. Noise sources in seismic data are diverse and can be categorized into regular noise and random noise based on their profile characteristics and statistical patterns. Regular noise possesses a specific frequency spectrum and apparent velocity, appearing as interference waves on a certain phase axis in the seismic record. Random noise lacks a definite dominant frequency and propagation direction, manifesting as a chaotic background interference in the seismic record, but still following statistical patterns. Currently, seismic data denoising methods can be broadly classified into three categories: spatiotemporal domain-based denoising methods, transform domain-based denoising methods, and machine learning-based denoising methods. Denoising methods based on the spatiotemporal domain suppress random noise based on the correlation of the effective signal, such as Gaussian filtering, nonlocal mean filtering, and fitting filtering. These algorithms have the advantage of high computational efficiency, but often lead to excessive smoothing of seismic data, reducing the fidelity of the effective signal. Denoising methods based on the transform domain mainly rely on the distribution patterns of the effective signal and noise signal in the transform domain to suppress random noise and coherent noise, and then inversely transform back to the spatiotemporal domain to obtain the denoised seismic data, such as FK filtering, Radon transform filtering, and KL transform filtering. When the filter is short or the threshold is small, the separation between the effective signal and the noise signal is small, and there is still a lot of noise in the denoised seismic data. Denoising methods based on machine learning require the establishment of samples that can cover the statistical characteristics of the effective signal and noise. However, seismic data is often complex and variable, and its sample set often cannot meet the modeling requirements. For complex seismic data containing faults, the fidelity of the denoised data will be reduced.

[0003] Chinese patent publication number CN117555027A, entitled "A Method for Suppressing Seismic Coherent Noise Using Frequency-Spatial Domain Eigenvector Projection," describes a method comprising: transforming a seismic signal from the time-space domain to the frequency-space domain; extracting data at a fixed frequency point from the frequency-space domain data to obtain a frequency slice; selecting a prediction step size and constructing a forward prediction signal vector, data matrix, and AR model equation; calculating the covariance matrix of the data matrix and performing eigenvalue decomposition, extracting the largest eigenvalue, and projecting the corresponding eigenvector in a specified direction; obtaining the solution to the AR model equation and filtering the signal to be predicted; constructing a backward prediction signal vector, data matrix, and AR model equation, and performing reverse prediction; averaging the results of the forward and reverse predictions to obtain the processed frequency slice; processing all frequency slices; and using an inverse Fourier transform to transform the processed signal back to the time-space domain to obtain the noise-suppressed signal. This patent application's method requires filtering during the calculation process, resulting in low fidelity of the effective signal from the obtained seismic data. Summary of the Invention

[0004] To overcome the problems existing in the prior art, the present invention aims to provide a frequency spatial domain feature decomposition denoising method, apparatus and analysis method. Based on the fact that seismic signals are not affected by stratigraphic occurrence in the frequency spatial domain and the optimal function of feature decomposition, the seismic data is first transformed from the original time spatial domain to the frequency spatial domain. Then, feature decomposition denoising is performed on each frequency slice in the effective frequency range of the seismic signal. Finally, the denoised frequency spatial domain data is transformed back to the time spatial domain to achieve amplitude-preserving denoising of the seismic signal.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: In a first aspect, the present invention provides a feature decomposition and denoising method in the frequency spatial domain, comprising the following steps: Acquire raw spatiotemporal data of the earthquake; transform the raw spatiotemporal data to the frequency spatial domain; Set the range of frequency processing; On a slice of data at a certain frequency, a portion of the data is randomly selected as data analysis points. Each data analysis point is expanded into a feature analysis sample by taking a rectangular window. This process is repeated for all data analysis points to obtain a feature analysis sample set. The covariance matrix of the feature analysis sample set matrix is ​​calculated, and its eigenvalues ​​are decomposed to obtain the eigenvectors of the effective signal. The entire frequency slice data is then expanded by taking rectangular windows point by point and calculating the inner product with the eigenvectors of the effective signal to obtain the denoised frequency slice data. By traversing the set frequency processing range, the effective seismic signal in the frequency spatial domain is obtained; The effective seismic signal in the frequency spatial domain is transformed back to the time spatial domain to obtain the final denoised seismic data.

[0006] Optionally, the original spatiotemporal domain data can be transformed to the frequency spatial domain through a channel-by-channel Fourier transform.

[0007] Optionally, the formula for calculating the covariance matrix of the feature analysis sample set matrix is: Where D is the analysis sample dataset, C is the covariance matrix, and M is the number of samples. It is the i-th sample.

[0008] Optionally, in the step of performing eigenvalue decomposition to obtain the eigenvectors of the effective signal, the characteristic equation of the covariance matrix C is: ,in Here, are the eigenvalues, and E is the N-order identity matrix. These are the eigenvectors.

[0009] Optionally, the formula for calculating the denoised frequency slice data is: ,in It is a valid signal. It is the vector of the largest eigenvalue, and N is the total length of the rectangular window data after expansion.

[0010] In a second aspect, the present invention provides a frequency spatial domain feature decomposition denoising system, comprising: The data acquisition module is used to acquire raw spatiotemporal domain earthquake data and transform the raw spatiotemporal domain data to the frequency spatial domain. The range setting module is used to set the range of frequency processing; The calculation module is used to randomly select a portion of the data from a frequency slice as data analysis points. Each data analysis point is expanded into a feature analysis sample by taking a rectangular window. By traversing all data analysis points, a feature analysis sample set is obtained. The covariance matrix of the feature analysis sample set matrix is ​​calculated, and eigenvalue decomposition is performed to obtain the feature vector of the effective signal. The entire frequency slice data is then expanded by taking rectangular windows point by point and calculating the inner product with the feature vector of the effective signal to obtain the denoised frequency slice data. Finally, the set frequency processing range is traversed to obtain the effective seismic signal in the frequency spatial domain. The data transformation module is used to transform the effective seismic signal in the frequency spatial domain back to the time spatial domain to obtain the final denoised seismic data.

[0011] Thirdly, the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the feature decomposition and denoising method in the frequency spatial domain.

[0012] Fourthly, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the feature decomposition and denoising method in the frequency spatial domain.

[0013] Fifthly, the present invention provides a computer program product including a computer-readable medium, wherein computer-readable program code is included on the computer-readable medium, the program code performing the feature decomposition denoising method in the frequency spatial domain.

[0014] In a sixth aspect, the present invention provides an earthquake data analysis method, which statistically analyzes the final denoised earthquake data obtained by the frequency spatial domain feature decomposition denoising method, and performs earthquake monitoring, earthquake early warning, earthquake disaster assessment, crustal structure analysis, or earthquake prediction based on the statistical results.

[0015] Compared with the prior art, the present invention has the following beneficial effects: This invention proposes a frequency spatial domain feature decomposition denoising method that combines the frequency spatial domain's independence from stratigraphic attitude with the optimization function of feature decomposition, resulting in high amplitude preservation. First, this invention employs a lossless Fourier transform method to transform the time-space domain seismic signal to the frequency spatial domain. This transformation, based on the wavelet frequency band characteristics of the effective seismic signal, concentrates the effective information within a narrow frequency band for processing. Furthermore, the signal-to-noise separation process in the frequency spatial domain does not require consideration of stratigraphic attitude, thus avoiding the complex process of stratigraphic dip estimation.

[0016] Furthermore, this invention performs feature decomposition denoising on frequency slices within a narrow frequency band in the frequency space domain. Feature decomposition involves scaling, shearing, and rotating the signal in the original coordinate system, separating different statistical feature signals in different feature vector subspaces, thus maintaining the effectiveness of the signal without loss. Finally, unlike the traditional feature decomposition denoising based on the overall principal component method, this invention uses a small rectangular window to extract a portion of the data points as an analysis sample set. The process of calculating the covariance matrix of the analysis sample set extracted by the small window compresses the effective signal features into a low-dimensional feature space. Therefore, by performing feature decomposition on the covariance matrix of the analysis sample set, the feature subspace vector of the effective signal can be extracted. The feature subspace vector of the effective signal in this invention is the vector corresponding to the largest eigenvalue, achieving amplitude-preserving denoising of seismic signals. Attached Figure Description

[0017] The accompanying drawings described herein are for illustrative purposes only and are not intended to limit the scope of the invention in any way.

[0018] In the attached diagram: Figure 1 This is a flowchart illustrating the principle and steps of the method of the present invention; Figure 2 This is the original seismic data from Embodiment 1 of the present invention; Figure 3 (a) is a cross-section of the real part of the frequency spatial domain transformation of Embodiment 1 of the present invention; (b) is a partial enlarged view of the real part of the frequency spatial domain transformation of Embodiment 1 of the present invention. Figure 4 (a) is a cross-section of the imaginary part of the frequency spatial domain transformation of Embodiment 1 of the present invention; (b) is a partial enlarged view of the imaginary part of the frequency spatial domain transformation of Embodiment 1 of the present invention. Figure 5 (a) is the original temporal and spatial domain seismic data of Embodiment 1 of the present invention; (b) is the frequency and spatial domain seismic data volume of Embodiment 1 of the present invention; (c) is a frequency slice data of Embodiment 1 of the present invention, wherein the small rectangular window represents the extracted feature analysis sample set; Figure 6 (a) is a comparison of the frequency spatial domain feature decomposition denoising results of Embodiment 1 of the present invention. The top image is the original earthquake, the middle image is the earthquake after denoising, and the bottom image is the noise; (b) is a local magnified view of (a). Figure 7 This is a comparison of the frequency spatial domain feature decomposition and denoising spectrum of Embodiment 1 of the present invention. Detailed Implementation

[0019] To enable those skilled in the art to better understand the technical solutions of this invention, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this invention, and not all embodiments. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this invention.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the specification of this invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0021] It should be noted that any reference signs placed between parentheses in the claims should not be construed as limiting the claims. The word "comprising" does not exclude the presence of components or steps not listed in the claims. The word "a" or "an" preceding a component does not exclude the presence of a plurality of such components. This application can be implemented by means of hardware comprising several different components and by means of a suitably programmed computer. In a unit claim enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified. The present invention will now be described in detail with reference to the accompanying drawings.

[0022] The present invention provides a frequency spatial domain feature decomposition denoising method, comprising the following steps: Acquire raw spatiotemporal data of the earthquake; transform the raw spatiotemporal data to the frequency spatial domain; Set the range of frequency processing; On a slice of data at a certain frequency, a portion of the data is randomly selected as data analysis points. Each data analysis point is expanded into a feature analysis sample by taking a rectangular window. This process is repeated for all data analysis points to obtain a feature analysis sample set. The covariance matrix of the feature analysis sample set matrix is ​​calculated, and its eigenvalues ​​are decomposed to obtain the eigenvectors of the effective signal. The entire frequency slice data is then expanded by taking rectangular windows point by point and calculating the inner product with the eigenvectors of the effective signal to obtain the denoised frequency slice data. By traversing the set frequency processing range, the effective seismic signal in the frequency spatial domain is obtained; The effective seismic signal in the frequency spatial domain is transformed back to the time spatial domain to obtain the final denoised seismic data.

[0023] Traditional temporal and spatial domain seismic denoising methods mostly adopt the approach of image denoising, denoising based on the correlation of effective signals. However, denoising operators are often easily affected by topographic attitude and noise. This invention can effectively reduce the impact of stratigraphic attitude on data, and at the same time process signals in different frequency ranges more accurately, thereby improving data processing performance.

[0024] Example 1 The following is combined with Figure 1 The specific embodiments of the present invention will be further described below.

[0025] Step 1: Transform the original spatiotemporal domain data to the frequency space domain through a channel-by-channel Fourier transform; Assuming the original time-space domain seismic data Where S is the seismic data volume in the time and space domain, A single data point in the time-space domain seismic data volume, where i, j, k are the indices for time, line, and trace directions, respectively; ns is the number of samples in the time direction; nl is the number of samples in the line direction; and nc is the number of samples in the trace direction. The original seismic data is converted to the frequency-space domain by performing a Fast Fourier Transform on each trace to obtain the data volume. ,in It is a frequency-space domain seismic data volume. It is a point in the frequency spatial domain seismic data volume, nf is the number of samples in the frequency direction, and other parameters are the same as above.

[0026] The Fast Fourier Transform (FFT) is lossless, and the signal-to-noise separation process in the frequency spatial domain does not need to consider the influence of stratigraphic attitude, thus avoiding the complex process of stratigraphic dip estimation. Actual seismic data is a non-steady-state signal, so the denoising process can be divided into time windows along the time axis. The denoising process within each time window is the denoising step of this invention. Time windows can be spliced ​​together using triangular windows, which will not be elaborated here. When there are insufficient sampling points in the time direction, this invention employs upsampling and zero-placing, which achieves the effect of frequency interpolation while avoiding frequency aliasing and leakage caused by conventional zero-padding.

[0027] Figure 2 Raw earthquake data; Figure 5 It is (a) Figure 3 (a) Real part of the frequency spatial domain transformation of the profile; (b) Enlarged view of the real part of the frequency spatial domain transformation of the profile. Figure 4 (a) Imaginary part of the spatial domain transformation of the cross-section frequency, (b) Enlarged view of the imaginary part of the spatial domain transformation of the cross-section frequency. Figure 5 (a) Raw temporal-spatial domain seismic data; (b) Frequency-spatial domain seismic data volume.

[0028] Step 2: Set the frequency processing range; Constrained by the seismic wavelet frequency band, effective information is concentrated in a narrow frequency band. To denoise the frequency spatial domain of seismic data, it is only necessary to extract the effective signal within the seismic frequency band, which can be set as [a, b], where a is the starting point of frequency processing and b is the ending point of frequency processing.

[0029] Step 3: Randomly select a portion of the data from a frequency slice as data analysis points. Expand each data analysis point into a rectangular window to form a feature analysis sample. Iterate through all data analysis points to obtain the feature analysis sample set. In a frequency slice Above, for a given data point, a two-dimensional rectangular window of data is taken as an analysis sample, centered on this point. Let the side length of the rectangular window be... Then an analysis sample is By iterating through all M feature vector analysis points, the analysis sample dataset is obtained. , yes The matrix, where .

[0030] Unlike traditional eigenvalue decomposition denoising based on principal component analysis, the process of extracting the covariance matrix from the analysis sample set using a small window compresses the effective signal features into a low-dimensional feature space, which is beneficial for the extraction of effective signals.

[0031] like Figure 5 (c) is a frequency slice data, where small rectangular windows represent the extracted random sample point data.

[0032] Step 4: Calculate the covariance matrix of the feature analysis sample set matrix and perform eigenvalue decomposition to obtain the eigenvectors of the effective signal; covariance matrix Where D is the analysis sample dataset, C is the covariance matrix, and M is the number of samples. It is the i-th sample.

[0033] The covariance matrix is ​​used to describe the correlation between samples. By performing eigenvalue decomposition on the covariance matrix, the main signal energy mapping direction can be analyzed. The magnitude of the eigenvalue reflects the magnitude of the sample in that direction in the new space. Removing the directions with small eigenvalues ​​and retaining only the main directions can achieve the purpose of dimensionality reduction and noise reduction.

[0034] Eigenvalue decomposition: First, solve for the eigenvalues. The process of finding the solution to the homogeneous characteristic equation of the characteristic polynomial of the covariance matrix C is to solve for the special value. The process, regarding the characteristic equation of C:

[0035] Where E is an N-order identity matrix, and the eigenvalues ​​are calculated according to the characteristic equation. Then, the eigenvectors are obtained by back-substituting the equations. ,in each Let i represent the i-th eigenvector.

[0036] Elements in one space are linearly transformed to another space of the same dimension. Matrix eigenvalues ​​are measures of rotation and scaling of the eigenvectors in the original space, and the eigenvectors represent the new space. The significance of eigenvalue decomposition lies in identifying the aspects in which a matrix produces the greatest dispersion, reducing overlap, and meaning more information is preserved. The eigenspace vector of the effective signal in the method of this invention is the vector corresponding to the largest eigenvalue, i.e. .

[0037] Step 5: Take rectangular windows point by point from the entire frequency slice data and expand them. The inner product of the window and the effective signal feature vector is used to obtain the denoised frequency slice data. In a frequency slice The effective signal feature vector is obtained through the feature sample set covariance matrix eigenvalue decomposition process in steps three and four, and then denoising is performed point by point on the frequency slice.

[0038] The two-dimensional window data centered on each point is: d is dimensional data, d and the largest eigenvalue vector from step four. Multiplying the two signals yields the valid signal for this center point.

[0039]

[0040] It is a valid signal. It is the vector of the largest eigenvalue, and N is the total length of the rectangular window data after expansion.

[0041] The above single-point denoising process is applied to all data points on the frequency slice to obtain the denoised earthquake of that frequency slice.

[0042] Step 6: Steps 3 to 5 traverse the frequency processing range of Step 2 to obtain the effective seismic signal in the frequency spatial domain.

[0043] Step 7: The effective seismic signal in the frequency spatial domain is inversely transformed back to the time spatial domain to obtain the final denoised seismic data.

[0044] Figure 6 (a) In the comparison of the frequency spatial domain feature decomposition denoising results, the top figure is the original earthquake, the middle figure is the earthquake after denoising, and the bottom figure is the noise; (b) is a local magnified view of (a). Figure 7 Comparison of denoised spectra by frequency spatial domain feature decomposition.

[0045] Combination Figure 6 and Figure 7The present invention compares the effects of noise reduction before and after the method of the present invention. The method of the present invention has high signal-to-noise separation and high amplitude preservation of effective signal. Many traditional noise reduction methods suppress noise in the high-frequency band of seismic signals, while the method of the present invention separates the effective signal and noise in the mid-to-high frequency bands, further suppressing the influence of noise, which is beneficial to the subsequent comprehensive interpretation of seismic signals.

[0046] Example 2 Based on the frequency spatial domain feature decomposition denoising method of Embodiment 1, a frequency spatial domain feature decomposition denoising system is disclosed, comprising: The data acquisition module is used to acquire raw spatiotemporal domain earthquake data and transform the raw spatiotemporal domain data to the frequency spatial domain. The range setting module is used to set the range of frequency processing; The calculation module is used to randomly select a portion of the data from a frequency slice as data analysis points. Each data analysis point is expanded into a feature analysis sample by taking a rectangular window. By traversing all data analysis points, a feature analysis sample set is obtained. The covariance matrix of the feature analysis sample set matrix is ​​calculated, and eigenvalue decomposition is performed to obtain the feature vector of the effective signal. The entire frequency slice data is then expanded by taking rectangular windows point by point and calculating the inner product with the feature vector of the effective signal to obtain the denoised frequency slice data. Finally, the set frequency processing range is traversed to obtain the effective seismic signal in the frequency spatial domain. The data transformation module is used to transform the effective seismic signal in the frequency spatial domain back to the time spatial domain to obtain the final denoised seismic data.

[0047] Example 3 The purpose of this embodiment is to provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the feature decomposition and denoising method in the frequency spatial domain.

[0048] Example 4 The purpose of this embodiment is to provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the feature decomposition and denoising method in the frequency spatial domain.

[0049] Example 5 The purpose of this embodiment is to provide a computer program product including a computer-readable medium, wherein the computer-readable medium contains computer-readable program code, and the program code executes the feature decomposition denoising method in the frequency spatial domain.

[0050] Example 6 The purpose of this embodiment is to provide a seismic data analysis method that statistically analyzes the final denoised seismic data obtained by the frequency spatial domain feature decomposition denoising method, and performs seismic monitoring, earthquake early warning, earthquake disaster assessment, crustal structure analysis, or earthquake prediction based on the statistical results.

[0051] The steps and methods involved in the apparatus of embodiments 2, 3, 4, 5 and 6 above correspond to those of embodiment 1. For specific implementation methods, please refer to the relevant description section of embodiment 1.

[0052] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes. These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes. These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0053] Unless otherwise specified, the working methods or control methods involved in the above embodiments are conventional working methods or control methods in the art.

[0054] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application. Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Any other modifications or equivalent substitutions made by those skilled in the art to the technical solutions of the present invention, as long as they do not depart from the spirit and scope of the technical solutions of the present invention, should be covered within the scope of the claims of the present invention.

Claims

1. A feature decomposition and denoising method in the frequency spatial domain, characterized in that, Includes the following steps: Acquire raw spatiotemporal data of the earthquake; transform the raw spatiotemporal data to the frequency spatial domain; Set the range of frequency processing; On a slice of data at a certain frequency, a portion of the data is randomly selected as data analysis points. Each data analysis point is expanded into a feature analysis sample by taking a rectangular window. By traversing all data analysis points, a feature analysis sample set is obtained. The covariance matrix of the feature analysis sample set matrix is ​​calculated, and the eigenvalues ​​of the sample set matrix are decomposed to obtain the eigenvectors of the effective signal. The entire frequency slice data is then divided into rectangular windows point by point and expanded. The inner product of the rectangular windows with the eigenvectors of the effective signal is calculated to obtain the denoised frequency slice data. By traversing the set frequency processing range, the effective seismic signal in the frequency spatial domain is obtained; The effective seismic signal in the frequency spatial domain is transformed back to the time spatial domain to obtain the final denoised seismic data.

2. The frequency spatial domain feature decomposition denoising method according to claim 1, characterized in that, The original spatiotemporal domain data is transformed to the frequency space domain through a channel-by-channel Fourier transform.

3. The frequency spatial domain feature decomposition denoising method according to claim 1, characterized in that, The formula for calculating the covariance matrix of the feature analysis sample set matrix is: Where D is the analysis sample dataset, C is the covariance matrix, and M is the number of samples. It is the i-th sample.

4. The frequency spatial domain feature decomposition denoising method according to claim 1, characterized in that, In the step of obtaining the eigenvectors of the effective signal through eigenvalue decomposition, the characteristic equation of the covariance matrix C is: ,in Here, are the eigenvalues, and E is the N-order identity matrix. These are the eigenvectors.

5. The frequency spatial domain feature decomposition denoising method according to claim 1, characterized in that, The formula for calculating the denoised frequency slice data is: ,in It is a valid signal. It is the vector of the largest eigenvalue, and N is the total length of the rectangular window data after expansion.

6. A frequency spatial domain feature decomposition and denoising system, characterized in that, include: The data acquisition module is used to acquire raw spatiotemporal data of earthquakes; Transform the original spatiotemporal domain data to the frequency space domain; The range setting module is used to set the range of frequency processing; The calculation module is used to randomly extract a portion of the data from a slice of data at a certain frequency as data analysis points. Each data analysis point is expanded into a feature analysis sample by taking a rectangular window. By traversing all data analysis points, a feature analysis sample set is obtained. The covariance matrix of the feature analysis sample set matrix is ​​calculated, and its eigenvalues ​​are decomposed to obtain the eigenvectors of the effective signal. The entire frequency slice data is then divided into rectangular windows point by point and expanded. The inner product of the rectangular window and the eigenvectors of the effective signal is calculated to obtain the denoised frequency slice data. The set frequency processing range is traversed to obtain the effective seismic signal in the frequency spatial domain. The data transformation module is used to transform the effective seismic signal in the frequency spatial domain back to the time spatial domain to obtain the final denoised seismic data.

7. An electronic device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the frequency spatial domain feature decomposition denoising method according to any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the frequency spatial domain feature decomposition denoising method according to any one of claims 1-5.

9. A computer program product comprising a computer-readable medium, characterized in that, The computer-readable medium contains computer-readable program code that performs the frequency spatial domain eigenvalue decomposition denoising method according to any one of claims 1-5.

10. A seismic data analysis method, characterized in that, The final denoised seismic data of the frequency spatial domain feature decomposition denoising method described in any one of claims 1-5 are statistically analyzed, and earthquake monitoring, earthquake early warning, earthquake disaster assessment, crustal structure analysis, or earthquake prediction are performed based on the statistical results.

Images