Method, system, equipment and medium for improving weak signal resolution

Abstract

The invention provides a method, a system, equipment and a medium for improving weak signal resolution, and the method comprises the following steps: carrying out the frequency band decomposition of a target seismic signal based on deep learning, and obtaining an effective frequency band signal and a noise frequency band signal; identifying and compensating the effective frequency band signal and the noise frequency band signal under a compressed sensing framework, and suppressing the noise; reconstructing the effective frequency band signal and the noise frequency band signal which are processed under the compressed sensing framework to obtain a compensated seismic signal; according to the method, the deep signals are decomposed according to the spectrum features of the deep signals by using the self-adaptability of deep learning, weak signal extraction and compensation are carried out on the effective frequency band under the compressed sensing framework, signals of other noise frequency bands are suppressed, the effective signals, especially deep effective weak signals, are protected, and meanwhile, strong noise is accurately suppressed, so that the accuracy of noise suppression is improved. The processed weak signal of the deep structure of the profile is effective, and the continuity of the event is enhanced.

Description

Technical Field

[0001] The invention belongs to the technical field of petroleum geophysical exploration, and in particular relates to a method, system, equipment and medium for improving weak signal resolution. Background Art

[0002] The near-surface conditions of the foreland basin in the western exploration area are extremely complex, with large-scale penetration of deep salt bodies, narrow bandwidth of seismic data in high-steep structures and carbonate reservoirs, and poor quality of seismic data. Developing recognition and adaptive compensation technology for deep weak signals and realizing adaptive recognition and energy recovery of deep seismic data are of great research significance for obtaining high-quality imaging results of medium and deep underground layers.

[0003] Q compensation can enhance the high-frequency part of the middle and deep layers and effectively improve the quality of seismic data. However, during the compensation process, the amplitude compensation term grows exponentially, especially the high-frequency growth is more serious, and the deep weak signals that are originally obliterated by noise are even more difficult to identify.

[0004] Noise suppression is one of the important ways to identify effective signals. There are many methods for suppressing strong noise, including high-order statistics, stochastic resonance theory, singular value decomposition, various sparse transformation methods (FFT, wavelet, curvelet, seislet wave, shearlet wave, etc.), empirical mode decomposition, etc. In order to improve the effect of noise suppression, many scholars have combined different methods. The compressed sensing theory that has emerged in recent years has broken the traditional sampling concept and is not limited by the signal bandwidth. The denoising method based on compressed sensing can utilize the different characteristics of effective signals and noise in the compression process, and these characteristics can be used to separate effective signals from noise through reconstruction algorithms. This method is not adaptive and cannot effectively separate weak signals from strong interference backgrounds when processing deep seismic data. Summary of the invention

[0005] In view of the problems existing in the prior art, the present invention provides a method, system, device and medium for improving the resolution of weak signals, which can accurately suppress strong noise while protecting effective signals, especially effective weak signals in deep areas.

[0006] The present invention is achieved through the following technical solutions:

[0007] A method for improving weak signal resolution comprises the following steps:

[0008] Based on deep learning, the target seismic signal is decomposed into frequency bands to obtain effective frequency band signals and noise frequency band signals;

[0009] Under the compressed sensing framework, the effective frequency band signal and the noise frequency band signal are identified and compensated, and the noise is suppressed;

[0010] The effective frequency band signal and the noise frequency band signal processed under the compressed sensing framework are reconstructed to obtain the compensated seismic signal.

[0011] Preferably, the process of performing frequency band decomposition of seismic signals based on deep learning is:

[0012] Collect multiple seismic signals and input them into the deep learning model as training samples to obtain a trained neural network model;

[0013] The target seismic signal is input into the neural network model and decomposed into different frequency bands.

[0014] Preferably, the process of obtaining the effective frequency band signal and the noise frequency band signal is:

[0015] Determine the effective frequency band signal and the noise frequency band signal based on the autocorrelation analysis method;

[0016] The effective frequency band signal is a frequency band signal whose main lobe width of autocorrelation is smaller than that of the original noisy data;

[0017] The noise band signal is a band signal whose main lobe width of autocorrelation is larger than the main lobe width of original noisy data.

[0018] Preferably, the process of identifying and compensating the effective frequency band signal and the noise frequency band signal under the compressed sensing framework is:

[0019] The quality factor Q is extracted by using the spectral ratio method or the frequency shift method and the compensation operator Λ is set;

[0020] Based on the quality factor Q, the difference between the effective signal and the perceived effective signal is obtained through the sparse promotion method, and the noise in the noise band is suppressed:

[0021] min:||Λ(e)|| 1 st||Y-ΦΨe|| 2 ≤ε;

[0022] Among them, Φ is a sparse sampling matrix that meets the RIP criterion, and the sparse sampling matrix includes a random Gaussian matrix, a Poisson sampling matrix or a Jitter sampling matrix, Ψ is a sparse transform inverse matrix, including one obtained by wavelet transform, curvelet transform and / or seislet transform, e is a sparse transform coefficient, and ε is a noise variance estimate, which is obtained by a greedy algorithm, a basis pursuit algorithm and / or an iterative threshold algorithm.

[0023] Preferably, the process of reconstructing the effective frequency band signal is:

[0024] The effective signal is reconstructed by the inverse transform of the sparse basis transform:

[0025]

[0026] The compensation operator Λ is used to increase the energy of the weak effective signal.

[0027] Preferably, the process of suppressing the effective frequency band signal and the noise frequency band signal under the compressed sensing framework is:

[0028] The difference between the effective signal and the perceived effective signal is obtained through the sparse promotion method, and the noise in the noise band is suppressed:

[0029] min:||e|| 1 st||Y-ΦΨe|| 2 ≤ε;

[0030] Among them, Φ is a sparse sampling matrix that meets the RIP criterion, and the sparse sampling matrix includes a random Gaussian matrix, a Poisson sampling matrix or a Jitter sampling matrix, Ψ is a sparse transform inverse matrix, including one obtained by wavelet transform, curvelet transform and / or seislet transform, e is a sparse transform coefficient, and ε is a noise variance estimate, which is obtained by a greedy algorithm, a basis pursuit algorithm and / or an iterative threshold algorithm.

[0031] Preferably, the process of reconstructing the noise frequency band signal is:

[0032] The effective signal in the noise band signal is reconstructed through the inverse transform of the sparse basis transform:

[0033]

[0034] A system for improving weak signal resolution, comprising:

[0035] A preprocessing module, used for performing frequency band decomposition of the target seismic signal based on deep learning to obtain an effective frequency band signal and a noise frequency band signal;

[0036] A compensation module is used to identify and compensate effective frequency band signals and noise frequency band signals under the compressed sensing framework to suppress noise;

[0037] The reconstruction module is used to reconstruct the effective frequency band signal and the noise frequency band signal processed under the compressed sensing framework to obtain the compensated seismic signal.

[0038] A computer device comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, steps of a method for improving weak signal resolution are implemented.

[0039] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of a method for improving weak signal resolution are implemented.

[0040] Compared with the prior art, the present invention has the following beneficial technical effects:

[0041] The present invention provides a method, system, device and medium for improving weak signal resolution, comprising the following steps: performing frequency band decomposition on a target seismic signal based on deep learning to obtain an effective frequency band signal and a noise frequency band signal; identifying and compensating the effective frequency band signal and the noise frequency band signal under a compressed sensing framework to suppress the noise; reconstructing the effective frequency band signal and the noise frequency band signal processed under the compressed sensing framework to obtain a compensated seismic signal; the present application uses the adaptability of deep learning to decompose the deep signal according to its spectral characteristics, extracts and compensates the weak signal of the effective frequency band under the compressed sensing framework, and suppresses the signals of the remaining noise frequency bands, accurately suppressing strong noise while protecting effective signals, especially effective weak signals in deep parts, so that the weak signals of the deep structure of the profile are effectively higher after processing, and the continuity of the event axis is enhanced. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 This is a flow chart of a method for improving weak signal resolution in an embodiment of the present invention;

[0043] Figure 2 This is a flowchart of a method for improving weak signal resolution in an embodiment of the present invention;

[0044] FIG. 3 is a comparison diagram of weak signal compensation in an embodiment of the present invention. DETAILED DESCRIPTION

[0045] The present invention is further described in detail below in conjunction with specific embodiments, which are intended to explain the present invention rather than to limit it.

[0046] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0047] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0048] This invention provides a method to improve the resolution of weak signals. Figure 1 and Figure 2 As shown, the following steps are included:

[0049] Based on deep learning, the target seismic signal is decomposed into frequency bands to obtain effective frequency band signals and noise frequency band signals;

[0050] Under the compressed sensing framework, the effective frequency band signal and the noise frequency band signal are identified and compensated, and the noise is suppressed;

[0051] The effective frequency band signal and the noise frequency band signal processed under the compressed sensing framework are reconstructed to obtain the compensated seismic signal.

[0052] Preferably, the process of performing frequency band decomposition of seismic signals based on deep learning is:

[0053] Collect multiple seismic signals and input them into the deep learning model as training samples to obtain a trained neural network model;

[0054] The target seismic signal is input into the neural network model and decomposed into different frequency bands.

[0055] Preferably, the process of obtaining the effective frequency band signal and the noise frequency band signal is:

[0056] Determine the effective frequency band signal and the noise frequency band signal based on the autocorrelation analysis method;

[0057] The effective frequency band signal is a frequency band signal whose main lobe width of autocorrelation is smaller than that of the original noisy data;

[0058] The noise band signal is a band signal whose main lobe width of autocorrelation is larger than the main lobe width of original noisy data.

[0059] Preferably, the process of identifying and compensating the effective frequency band signal and the noise frequency band signal under the compressed sensing framework is:

[0060] The quality factor Q is extracted by using the spectral ratio method or the frequency shift method and the compensation operator Λ is set;

[0061] It should be noted that the process of extracting the quality factor Q using the spectral ratio method provided in this application is as follows:

[0062] Obtain vibration data: Perform vibration tests on the vibration system to obtain data such as vibration amplitude or speed at different frequencies. This can be done using equipment such as vibration test benches and vibration meters.

[0063] Frequency spectrum analysis: Perform frequency spectrum analysis on the acquired vibration data, that is, calculate the vibration amplitude or speed at each frequency, which can be achieved through mathematical methods such as fast Fourier transform (FFT);

[0064] Determine the resonant frequency: In the frequency spectrum, there will be one or more resonant frequencies. These resonant frequencies are the frequency points where the vibration system has the maximum response to vibrations of a specific frequency. Find these resonant frequencies;

[0065] Calculate the quality factor Q: Use the following formula to calculate the quality factor Q:

[0066] Q = (f1-f2) / f1*sqrt(E1 / E2);

[0067] Where f1 and f2 are the low and high frequency limits of the resonant frequency, and E1 and E2 are the ratios of the energy stored to the energy dissipated at these limit frequencies;

[0068] For experimental data, E1 / E2 can be calculated as follows:

[0069] E1 / E2=(A1*A2) / (A2-A1)

[0070] Where A1 and A2 are the vibration amplitudes or velocities at frequencies f1 and f2;

[0071] Calculation results: Substituting f1-f2, f1, E1 and E2 into the above formula, the value of the quality factor Q can be obtained.

[0072] It should be further explained that the quality factor Q is a parameter that describes the ability of the vibration system to store and release energy, and the compensation operator is a linear operator introduced in the vibration system to improve the dynamic characteristics of the system and improve the stability of the system; in some cases, the quality factor Q and the compensation operator may influence each other; for example, when there is nonlinear friction in the vibration system, the introduction of the compensation operator may improve the dynamic characteristics of the system and may improve the quality factor Q of the system; the present application comprehensively considers the influence of the quality factor Q and the compensation operator in order to better improve the dynamic characteristics and stability of the system.

[0073] Based on the quality factor Q, the difference between the effective signal and the perceived effective signal is obtained through the sparse promotion method, and the noise in the noise band is suppressed:

[0074] min:||Λ(e)|| 1 st||Y-ΦΨe|| 2 ≤ε;

[0075] Among them, Φ is a sparse sampling matrix that meets the RIP criterion, and the sparse sampling matrix includes a random Gaussian matrix, a Poisson sampling matrix or a Jitter sampling matrix, Ψ is a sparse transform inverse matrix, including one obtained by wavelet transform, curvelet transform and / or seislet transform, e is a sparse transform coefficient, and ε is a noise variance estimate, which is obtained by a greedy algorithm, a basis pursuit algorithm and / or an iterative threshold algorithm.

[0076] Preferably, the process of reconstructing the effective frequency band signal is:

[0077] The effective signal is reconstructed by the inverse transform of the sparse basis transform:

[0078]

[0079] The compensation operator Λ is used to increase the energy of the weak effective signal.

[0080] Preferably, the process of suppressing the effective frequency band signal and the noise frequency band signal under the compressed sensing framework is:

[0081] The difference between the effective signal and the perceived effective signal is obtained through the sparse promotion method, and the noise in the noise band is suppressed:

[0082] min:||e|| 1 st||Y-ΦΨe|| 2 ≤ε,

[0083] Among them, Φ is a sparse sampling matrix that meets the RIP criterion, and the sparse sampling matrix includes a random Gaussian matrix, a Poisson sampling matrix or a Jitter sampling matrix, Ψ is a sparse transform inverse matrix, including one obtained by wavelet transform, curvelet transform and / or seislet transform, e is a sparse transform coefficient, and ε is a noise variance estimate, which is obtained by a greedy algorithm, a basis pursuit algorithm and / or an iterative threshold algorithm.

[0084] Preferably, the process of reconstructing the noise frequency band signal is:

[0085] The effective signal in the noise band signal is reconstructed through the inverse transform of the sparse basis transform:

[0086]

[0087] It should be noted that the random Gaussian matrix is ​​a commonly used matrix, each row and column of which is generated by a random Gaussian normal distribution. Its characteristics are that the sum of each row and column is zero, and the sum of squares of elements in each row or column is proportional to the average of the squares of all elements.

[0088] It should be further explained that the Poisson sampling matrix Poisson sampling matrix is ​​a method for achieving efficient sparse representation in signal processing. It utilizes the characteristics of Poisson distribution to approximately reconstruct high-dimensional signals in lower dimensions. The construction method of Poisson sampling matrix is ​​to first determine the sparsity of the desired reconstructed signal, that is, the number of non-zero elements in the signal, and then randomly select the positions of these non-zero elements in the high-dimensional signal. When selecting the positions, the probability distribution function (PDF) of Poisson distribution is used to obtain the positions of non-zero elements with a high probability of selecting the number determined by the sparsity. Poisson sampling matrix has the following advantages: efficient sparse representation: by utilizing Poisson distribution, high-dimensional signals can be reconstructed in lower dimensions, thereby reducing the computational and storage costs; applicable to various signals: Poisson sampling matrix is ​​applicable to various types of signals, including images, audio and video, etc.; strong interpretability: Poisson sampling matrix is ​​constructed based on sparsity and Poisson distribution theory, so it has a clear theoretical basis and interpretability; it should be noted that when using Poisson sampling matrix for signal reconstruction, additional techniques may be needed to optimize the reconstruction effect, such as the least squares method based on iterative threshold.

[0089] It should be further explained that the Ψ is a sparse transformation inverse matrix, which is obtained by wavelet transformation, curvelet transformation and / or seislet transformation, and the process is as follows:

[0090] The present invention provides the general form of a common wavelet transform function:

[0091]

[0092] Compare the wavelet w(t) with the beginning part of the original function f(t) (actually, perform the inner product), and calculate the coefficient C, which indicates the similarity between this part of the function and the wavelet.

[0093] Shift the wavelet to the right by k units to obtain wavelet w(tk), and repeat step 1. Repeat this step until the function f ends.

[0094] Expand wavelet w(t) to obtain wavelet w(t / 2), and repeat steps 1 and 2.

[0095] Continuously expand the wavelet and repeat the above steps to obtain the sparse transformation inverse matrix Ψ.

[0096] It should be further explained that the ε is a noise variance estimate, which is obtained by a greedy algorithm, a basis pursuit algorithm and / or an iterative threshold algorithm, and the process is:

[0097] 1. Initialize the estimation result: set the estimated signal x^ to a 0 vector and the initial residual to r=x.

[0099] 2. Select the most relevant atom: Calculate the inner product of the column vector of D and the residual r, select the atom index i* with the largest inner product value, and update the estimated signal x^, i* = argmax_i

[0100] |<d_i,r> |, where d_i is the i-th column vector of dictionary D, x^=x^+

[0101] <d_i*,r> d_i*.

[0102] 3. Update residual: calculate the new residual r = xx^.

[0103] 4. Repeat steps 2 and 3 until the stopping criterion is met (such as the residual threshold or the number of selected atoms reaches a predetermined value).

[0104] 5. Output the final estimated signal x^.

[0105] During the iteration, OMP gradually selects atoms with the largest correlation and adds them to the estimated signal, while reducing the energy of the residual, so that the original sparse signal can be gradually reconstructed.

[0106] Formula derivation is not the core of the OMP algorithm, but the update steps in the pseudocode can be derived by the following formula:

[0107] Compute the inner product between the atom and the residual:<d_i,r>

[0108] Calculate the coefficient corresponding to atom i*:<d_i*,r>

[0109] Update the estimated signal: x^=x^+<d_i*,r> d_i*

[0110] Update residual: r = r-<d_i*,r> d_i*

[0111] The present application utilizes the adaptability of deep learning to decompose deep signals according to their spectral features, extracts and compensates weak signals in effective frequency bands under a compressed sensing framework, and suppresses signals in remaining noise frequency bands, thereby protecting effective signals, especially effective weak signals in deep areas, while accurately suppressing strong noise, as shown in FIG3 , wherein FIG3(a) is a signal spectrum in which effective signals in the original data are obliterated by strong interference background, FIG3(b) is a signal spectrum in which weak signals obliterated by strong noise are effectively extracted after weak signal extraction and compensation, and FIG3(c) is a signal spectrum in which the energy axis of the weak signal is significantly enhanced after compensation.

[0112] The present invention provides a system for improving weak signal resolution, comprising:

[0113] A preprocessing module, used for performing frequency band decomposition of the target seismic signal based on deep learning to obtain an effective frequency band signal and a noise frequency band signal;

[0114] A compensation module is used to identify and compensate effective frequency band signals and noise frequency band signals under the compressed sensing framework to suppress noise;

[0115] The reconstruction module is used to reconstruct the effective frequency band signal and the noise frequency band signal processed under the compressed sensing framework to obtain the compensated seismic signal.

[0116] In another embodiment of the present invention, a computer device is provided, the computer device including a processor and a memory, the memory is used to store a computer program, the computer program includes program instructions, and the processor is used to execute the program instructions stored in the computer storage medium. The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc., which are the computing core and control core of the terminal, which are suitable for implementing one or more instructions, and are specifically suitable for loading and executing one or more instructions in a computer storage medium to implement a corresponding method flow or a corresponding function; the processor described in the embodiment of the present invention can be used for an operation of a method for improving weak signal resolution.

[0117] In another embodiment of the present invention, the present invention also provides a storage medium, specifically a computer-readable storage medium (Memory), which is a memory device in a computer device for storing programs and data. It is understandable that the computer-readable storage medium here can include both a built-in storage medium in a computer device and an extended storage medium supported by the computer device. The computer-readable storage medium provides a storage space, which stores the operating system of the terminal. In addition, one or more instructions suitable for being loaded and executed by a processor are also stored in the storage space, and these instructions can be one or more computer programs (including program codes). It should be noted that the computer-readable storage medium here can be a high-speed RAM memory or a non-volatile memory, such as at least one disk memory. The processor can load and execute one or more instructions stored in the computer-readable storage medium to implement the corresponding steps of a method for improving weak signal resolution in the above embodiment.

[0118] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented 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.

[0119] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0120] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0121] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0122] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for improving weak signal resolution, characterized in that: The following steps are involved: Based on deep learning, the target seismic signal is decomposed into frequency bands to obtain effective frequency band signals and noise frequency band signals; Under the compressed sensing framework, the effective frequency band signal and the noise frequency band signal are identified and compensated, and the noise is suppressed; The effective frequency band signal and the noise frequency band signal processed under the compressed sensing framework are reconstructed to obtain the compensated seismic signal.

2. A method for improving weak signal resolution according to claim 1, characterized in that: The process of frequency band decomposition of seismic signals based on deep learning is as follows: Collect multiple seismic signals and input them into the deep learning model as training samples to obtain a trained neural network model; The target seismic signal is input into the neural network model and decomposed into different frequency bands.

3. A method for improving weak signal resolution according to claim 1, characterized in that: The process of obtaining the effective frequency band signal and the noise frequency band signal is as follows: Determine the effective frequency band signal and the noise frequency band signal based on the autocorrelation analysis method; The effective frequency band signal is a frequency band signal whose main lobe width of autocorrelation is smaller than that of the original noisy data; The noise band signal is a band signal whose main lobe width of autocorrelation is larger than the main lobe width of original noisy data.

4. A method for improving weak signal resolution according to claim 1, characterized in that: The process of identifying and compensating effective frequency band signals and noise frequency band signals under the compressed sensing framework is as follows: The quality factor Q is extracted by using the spectral ratio method or the frequency shift method and the compensation operator Λ is set; Based on the quality factor Q, the difference between the effective signal and the perceived effective signal is obtained through the sparse promotion method, and the noise in the noise band is suppressed: min:||Λ(e)||1s.t.||Y-ΦΨe||2≤ε; Among them, Φ is a sparse sampling matrix that meets the RIP criterion, and the sparse sampling matrix includes a random Gaussian matrix, a Poisson sampling matrix or a Jitter sampling matrix, Ψ is a sparse transform inverse matrix, including one obtained by wavelet transform, curvelet transform and / or seislet transform, e is a sparse transform coefficient, and ε is a noise variance estimate, which is obtained by a greedy algorithm, a basis pursuit algorithm and / or an iterative threshold algorithm.

5. A method for improving weak signal resolution according to claim 4, characterized in that: The process of reconstructing the effective frequency band signal is: The effective signal is reconstructed by the inverse transform of the sparse basis transform: The compensation operator Λ is used to increase the energy of the weak effective signal.

6. A method for improving weak signal resolution according to claim 1, characterized in that: The process of suppressing the effective frequency band signal and the noise frequency band signal under the compressed sensing framework is as follows: The difference between the effective signal and the perceived effective signal is obtained through the sparse promotion method, and the noise in the noise band is suppressed: min:||e||1 st||Y-ΦΨe||2≤ε; Among them, Φ is a sparse sampling matrix that meets the RIP criterion, and the sparse sampling matrix includes a random Gaussian matrix, a Poisson sampling matrix or a Jitter sampling matrix, Ψ is a sparse transform inverse matrix, including one obtained by wavelet transform, curvelet transform and / or seislet transform, e is a sparse transform coefficient, and ε is a noise variance estimate, which is obtained by a greedy algorithm, a basis pursuit algorithm and / or an iterative threshold algorithm.

7. A method for improving weak signal resolution according to claim 6, characterized in that: The process of reconstructing the noise band signal is: The effective signal in the noise band signal is reconstructed through the inverse transform of the sparse basis transform:

8. A system for improving weak signal resolution, characterized in that: A method for improving weak signal resolution according to any one of claims 1 to 7, comprising: A preprocessing module, used for performing frequency band decomposition of the target seismic signal based on deep learning to obtain an effective frequency band signal and a noise frequency band signal; A compensation module is used to identify and compensate effective frequency band signals and noise frequency band signals under the compressed sensing framework to suppress noise; The reconstruction module is used to reconstruct the effective frequency band signal and the noise frequency band signal processed under the compressed sensing framework to obtain the compensated seismic signal.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method for improving weak signal resolution as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of a method for improving weak signal resolution as claimed in any one of claims 1 to 7 are implemented.

Images