Method for processing and identifying thin interbedded targets based on q-compensation compressed sensing

Abstract

The application provides a thin interbedded target processing and identification method based on Q compensation compressive sensing, which comprises the following steps: step 1, extracting a wavelet of an original seismic record; step 2, Q value estimation of the original seismic record; step 3, constructing a frequency domain diagonal wavelet measurement matrix, a time-frequency attenuation sparse matrix and an attenuation sensing matrix; step 4, Fourier transform of single-channel seismic data to convert the single-channel seismic data into frequency domain data, and calculating a time-frequency attenuation factor at each sampling point and an attenuation sensing matrix of the whole data according to step 3, and then carrying out sparse inversion of reflection coefficients; step 5, single-channel cyclic processing of all seismic data to obtain sparse inversion results of reflection coefficients corresponding to the all seismic data, and then carrying out convolution to obtain seismic data with improved resolution. The thin interbedded target processing and identification method based on Q compensation compressive sensing is used for improving thin interbedded seismic resolution and improving subsequent thin reservoir identification and prediction accuracy, and provides a basis for seismic development.

Description

Technical Field

[0001] This invention relates to the field of seismic data target processing and reservoir description, and in particular to a method for processing and identifying thin interbedded targets based on Q-compensated compressed sensing. Background Technology

[0002] Compared to thick, massive sandstone reservoirs, reservoirs with sand bodies around 5 meters thick interbedded with mudstone experienced lower levels of early exploitation. However, as the water cut of the main reservoir sand bodies continues to rise, these thin sand bodies (around 5 meters thick) are increasingly becoming the focus of exploration and development in the old oilfields of eastern China, representing a major potential target area for increasing oil reserves and production. The original resolution of 3D seismic data in the 1500-2500 m depth range in the old oilfields of eastern China is only 7-15 m, and conventional geological profiles and slices cannot meet the requirements for the detailed identification and description of thin, interbedded sand bodies. Therefore, improving seismic resolution and enhancing the ability to identify and characterize thin, interbedded sand bodies is becoming increasingly important.

[0003] Considering the dispersion and thin-layer tuning phenomena in seismic data, Q-filtering is commonly used to compensate for medium absorption attenuation and dispersion. Then, based on a forward model, instantaneous class and amplitude type attributes are extracted to predict thin interbedded sand bodies. This method has strong predictive ability for sand bodies with a thickness greater than 8m and a distribution range greater than 100m. Since the original seismic data is a comprehensive response of strata, lithofacies, pores, fluids, and faults, even Q-compensated and corrected seismic data is limited by signal-to-noise ratio and resolution. Seismic attributes are less effective in identifying thin interbedded sand bodies with a distribution range less than 80m and a thickness less than 7m. Traditional compressed sensing sparse signal sampling and reconstruction theory provides a basis for reconstructing broadband sparse reflection coefficients from band-limited seismic data. However, traditional compressed sensing methods for improving seismic resolution do not consider the absorption and attenuation of seismic waves by strata and pore fluids, leading to distortion of weak seismic signals after processing, which is detrimental to subsequent thin interbedded identification and reservoir prediction.

[0004] Chinese patent application CN201810679707.8 discloses a time-varying sparse deconvolution method and system. The method includes: Step 1: Inputting seismic records, initial seismic wavelet, and sparsity; Step 2: Estimating the Q-value vector and constructing an attenuation matrix; Step 3: Establishing an optimized reflection coefficient model and a seismic wavelet model based on the attenuation matrix; Step 4: Obtaining optimized reflection coefficients based on the initial seismic wavelet, sparsity, and the optimized reflection coefficient model; Step 5: Obtaining optimized seismic wavelets based on the optimized reflection coefficients and the optimized seismic wavelet model; Step 6: Repeating steps 4-5 until the reflection coefficient difference is less than a reflection coefficient difference threshold or the seismic wavelet difference is less than a seismic wavelet difference threshold. This invention performs absorption attenuation estimation, iteratively optimizing the reflection coefficient and source wavelet, obtaining more accurate results, improving seismic data resolution, better maintaining the relative amplitude and phase deviation of the reflection coefficient, and exhibiting good fault tolerance and noise resistance.

[0005] Chinese patent application CN201710994606.5 discloses a method for acoustic impedance inversion with Q absorption attenuation compensation. Impedance inversion is one of the important methods for seismic oil and gas resource exploration. This invention integrates the formation Q absorption attenuation dispersion factor into the convolution model and directly obtains high-resolution impedance results from the non-steady-state seismic record data with absorption attenuation through iterative inversion. This invention avoids the error accumulation effect in the traditional recursive process and avoids the numerical instability problem of the traditional inverse Q filtering method. That is, the Q-value compensation technology of this invention is unconditionally stable and can compensate for higher frequency components than traditional technologies, further improving the resolution of seismic exploration data.

[0006] Chinese patent application CN201910815949.X discloses a method for band extension processing of seismic data based on compressed sensing. The method includes: S1) characterizing seismic signals based on the reflection coefficients of underground strata; S2) transforming the seismic signals to obtain a spectrum corresponding to the seismic signals; S3) performing discrete operations on the spectral equation of the spectrum to establish an equation for the effective selection spectrum of the reflection coefficients; and S4) optimizing the solution of the reflection coefficients to obtain a broadband sparse pulse reflection coefficient sequence.

[0007] The above-mentioned existing technologies are all quite different from the present invention and have failed to solve the technical problem we want to solve. Therefore, we have invented a new method for thin interlayer target processing and recognition based on Q-compensated compressed sensing. Summary of the Invention

[0008] The purpose of this invention is to provide a method for processing and identifying thin interbedded targets based on Q-compensated compressed sensing, which improves the seismic resolution of thin interbedded layers, enhances the accuracy of subsequent thin reservoir identification and prediction, and provides a basis for actual seismic development.

[0009] The objective of this invention can be achieved through the following technical measures: a method for processing and recognizing thin interlayer targets based on Q-compensated compressed sensing, comprising:

[0010] Step 1: Extract the wavelet from the original seismic record;

[0011] Step 2: Estimate the Q-value of the original seismic record;

[0012] Step 3: Construct the frequency domain diagonal wavelet measurement matrix, the time-frequency attenuation sparse matrix, and the attenuation sensing matrix;

[0013] Step 4: Perform Fourier transform on the single-channel seismic data to convert it into frequency domain data, and calculate the time-frequency attenuation factor at each sampling point and the attenuation sensing matrix of the whole channel data according to Step 3. Then perform sparse inversion of reflection coefficients.

[0014] Step 5: Perform single-channel loop processing on all seismic data to obtain sparse inversion results of reflection coefficients corresponding to all seismic data, and then perform convolution to obtain seismic data with improved resolution.

[0015] The objective of this invention can also be achieved through the following technical measures:

[0016] In step 1, basic information about the seismic data is obtained by performing spectral and waveform feature analysis, and then the wavelet of the original seismic record is extracted using the polyspectral method.

[0017] Step 1 also includes adjusting the obtained wavelet frequency band and phase by sliding a time window based on the frequency band and phase of the data.

[0018] In step 1, the time window should not be too large or too small, but should be designed to preserve the wavelet information at the origin of the complex spectrum while removing the reflection coefficient information at the non-origin points.

[0019] In step 1, the wavelet of the original seismic record is extracted using the complex spectrum method. The specific steps are as follows: First, a time window range is selected for the original seismic record. Then, the seismic traces within this range are subjected to Fourier transform to obtain their amplitude spectrum. Next, the amplitude spectrum is subjected to logarithmic transform and inverse Fourier transform to obtain the complex spectrum result. A low-pass filter is designed by setting the width of the passband to filter the complex spectrum result and perform Fourier and inverse logarithmic transforms. Finally, the extracted wavelet of the original seismic record is obtained.

[0020] In step 2, the quality factor Q of the original seismic record is estimated using the spectral ratio method or the amplitude envelope method. Then, the estimated quality factor is smoothed to increase the stability of the subsequent sparse inversion process.

[0021] In step 2, the Q-value of the original seismic record is estimated using the amplitude envelope method. The specific steps are as follows:

[0022] First, set the time window size and time window sliding step size, calculate the corresponding amplitude spectrum A(f) for the time window, and perform a certain weighted average on adjacent channels; then, convert the obtained amplitude spectrum into decibel form B(f), as shown in equation (1), and then use the least squares method to calculate the slope k of the envelope in decibel form and estimate the Q value, as shown in equation (2). Finally, interpolate the estimated Q value so that each sample point has a Q value:

[0023] B(f)=20lg(A(f)) (1)

[0024] B(f)=-27.3fQ -1 t

[0025] Q = 27.3·t / |k| (2)

[0026] In the formula, A(f) is the amplitude spectrum, B(f) is the decibel form of A(f), f is the frequency, t is the center time of the time window, k is the slope of the envelope, and Q is the quality factor.

[0027] In step 3, based on the wavelet obtained in step 1 and the quality factor Q obtained in step 2, a frequency domain diagonal wavelet measurement matrix and a time-frequency attenuation sparse matrix are constructed to further obtain the attenuation sensing matrix.

[0028] In step 3, the frequency domain diagonal wavelet measurement matrix W is constructed based on the calculated seismic wavelet, as shown in equation (3):

[0029]

[0030] In the formula, W is the measurement matrix, diag is the diagonal function, FT is the forward Fourier transform, and w(t) is the extracted wavelet. For the wavelet w(t) with different frequency components f1, f2, ..., f M The corresponding spectral value, M is the sample number corresponding to the highest frequency component.

[0031] In step 3, based on the quality factor Q estimated in step S2, the time-frequency attenuation factor is calculated for each trace to obtain the attenuation sparse matrix G of the seismic data for that trace, as shown in equation (4). The attenuation sensing matrix A is further obtained by combining the measurement matrix W and the attenuation sparse matrix G, where A = GW.

[0032]

[0033] In the formula, i represents an imaginary number, f is the frequency, f0 is the reference frequency, t is the time, and Q(t) is the quality factor at time t.

[0034] In step 4, the single-channel seismic data is converted into frequency domain data by Fourier transform, and the time-frequency attenuation factor at each sampling point and the attenuation sensing matrix of the whole channel data are calculated according to step 3. Then, the sparse inversion of reflection coefficients is performed using the fast threshold shrinkage iterative algorithm.

[0035] In step 4, the single-channel seismic data s(t) is subjected to Fourier transform to convert it into frequency domain data S(f), and the attenuation sensing matrix of the data is calculated according to step 3. The inversion objective function is constructed as shown in equation (5).

[0036]

[0037] In the formula, |||2 represents the 2-norm, |||1 represents the 1-norm, A is the attenuation sensing matrix, R is the reflection coefficient to be solved, S is the frequency domain single-channel seismic data, and λ is the regularization factor.

[0038] In step 5, all seismic data are processed in a single loop to obtain sparse inversion results of reflection coefficients corresponding to all seismic data. Then, these results are convolved with a broadband seismic wavelet to finally obtain seismic data with improved resolution.

[0039] In step 5, the broadband seismic wavelet adopts the higher dominant frequency Ricker wavelet.

[0040] The thin interbedded layer target processing and identification method based on Q-compensated compressed sensing in this invention is mainly used to improve the seismic resolution of thin interbedded layers, enhance the accuracy of subsequent thin reservoir identification and prediction, and provide a basis for actual seismic development. The beneficial technical effects of this invention are:

[0041] This invention provides high-resolution processing for thin-layered and interbedded thin-layered seismic profiles. Building upon traditional compressed sensing for improving seismic resolution, it considers the absorption and attenuation effects of the actual medium on seismic waves. By introducing a time-frequency attenuation factor to construct an attenuation sensing matrix, it performs sparse inversion of seismic reflection coefficients and convolution with broadband wavelets to reasonably compensate for bandwidth and increase the dominant frequency, resulting in a high-resolution seismic profile based on Q-compensation. This invention effectively improves the seismic resolution of interbedded thin-layered reservoirs, enhances the identification and prediction accuracy of these reservoirs, and provides a basis for actual seismic development. This invention has significant theoretical and practical value. Attached Figure Description

[0042] Figure 1This is a flowchart of a specific embodiment of the thin interlayer target processing and recognition method based on Q-compensated compressed sensing of the present invention;

[0043] Figure 2 This is a diagram of a one-dimensional theoretical earthquake attenuation forward model in a specific application example of the present invention;

[0044] Figure 3 This is a schematic diagram of the high-resolution processing result (noise-free) of a one-dimensional theoretical model in a specific application example of the present invention;

[0045] Figure 4 This is a schematic diagram of the high-resolution processing result (including noise) of a one-dimensional theoretical model in a specific application example of the present invention;

[0046] Figure 5 This is a schematic diagram of the high-resolution processing result (10% random noise) of the two-dimensional theoretical model in a specific application example of the present invention;

[0047] Figure 6 This is a plan view of a work area in a specific application example of the present invention;

[0048] Figure 7 This is an original seismic profile of a series of wells in a specific application example of the present invention;

[0049] Figure 8 This is a high-resolution seismic profile of a series of wells in a specific application example of the present invention;

[0050] Figure 9 This is a comparison of the spectrum before and after the well profile processing in a specific application example of the present invention. Detailed Implementation

[0051] It should be noted that the following detailed descriptions are exemplary and intended to provide further illustration of the invention. Unless otherwise specified, 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.

[0052] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments of the present invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, and / or combinations thereof.

[0053] The present invention discloses a method for processing and identifying thin interbedded reservoirs based on Q-compensated compressed sensing, comprising the following steps: S1 extracting wavelets from the original seismic records using the complex spectrum method; S2 estimating the quality factor Q of the original seismic records using the amplitude envelope method; S3 constructing a frequency domain diagonal wavelet measurement matrix and a time-frequency attenuation sparse matrix based on the wavelets and the quality factor Q, thereby obtaining an attenuation sensing matrix; S4 converting the single-channel seismic data into frequency domain data by performing a Fourier transform, and calculating the time-frequency attenuation factor at each sampling point and the attenuation sensing matrix of the entire channel data according to step S3, followed by sparse inversion of reflection coefficients; S5 performing single-channel cyclic processing on all seismic data to obtain sparse inversion results of reflection coefficients corresponding to all seismic data, followed by convolution to finally obtain seismic data with improved resolution. This invention can effectively improve the seismic resolution of thin interbedded reservoirs and enhance the subsequent identification capability of thin interbedded reservoirs.

[0054] The following are several specific embodiments of the application of the present invention.

[0055] Example 1

[0056] In a specific embodiment 1 of the present invention, the thin interlayer target processing and recognition method based on Q-compensated compressed sensing includes the following steps:

[0057] S1 obtains basic information about seismic data by performing spectral and waveform feature analysis, and then uses the complex spectrum method to extract wavelets from the original seismic record.

[0058] Based on the frequency band and phase of the data, a sliding time window is used to adjust the obtained wavelet frequency band and phase.

[0059] The time window should not be too large or too small. It should be designed to preserve the wavelet information at the origin of the semi-final spectrum while removing the reflection coefficient information at the non-origin points.

[0060] The wavelet of the original seismic record is extracted using the complex spectrum method. The specific steps are as follows: First, select a time window range for the original seismic record. Then, perform a Fourier transform on the seismic traces within this range to obtain their amplitude spectrum. Next, perform a logarithmic transform and an inverse Fourier transform on the amplitude spectrum to obtain the complex spectrum result. By setting the width of the passband, a low-pass filter is designed to filter the complex spectrum result and perform Fourier and inverse logarithmic transforms. Finally, the extracted wavelet of the original seismic record is obtained.

[0061] S2 estimates the quality factor Q of the original seismic record using either the spectral ratio method or the amplitude envelope method. The specific steps for estimating the Q value of the original seismic record using the amplitude envelope method are as follows:

[0062] First, set the time window size and time window sliding step size, calculate the corresponding amplitude spectrum A(f) for the time window, and perform a certain weighted average on adjacent channels; then, convert the obtained amplitude spectrum into decibel form B(f), as shown in equation (1), and then use the least squares method to calculate the slope k of the envelope in decibel form and estimate the Q value, as shown in equation (2). Finally, interpolate the estimated Q value so that each sample point has a Q value:

[0063] B(f)=20lg(A(f)) (1)

[0064] B(f)=-27.3fQ -1 t

[0065] Q = 27.3 t / |k (2)

[0066] In the formula, A(f) is the amplitude spectrum, B(f) is the decibel form of A(f), f is the frequency, t is the center time of the time window, k is the slope of the envelope, and Q is the quality factor.

[0067] The estimated quality factor is then appropriately smoothed to increase the stability of the subsequent sparse inversion process.

[0068] S3 uses the wavelet obtained in step S1 and the quality factor Q obtained in step S2 to construct the frequency domain diagonal wavelet measurement matrix and the time-frequency attenuation sparse matrix, and further obtains the attenuation sensing matrix; the calculated seismic wavelet is used to construct the frequency domain diagonal wavelet measurement matrix W, as shown in equation (3).

[0069]

[0070] In the formula, W is the measurement matrix, diag is the diagonal function, FT is the forward Fourier transform, and w(t) is the extracted wavelet. For the wavelet w(t) with different frequency components f1, f2, ..., f M The corresponding spectral value, M is the sample number corresponding to the highest frequency component.

[0071] Based on the quality factor Q estimated in step S2, the time-frequency attenuation factor is calculated for each trace to obtain the attenuation sparse matrix G of the seismic data for that trace, as shown in equation (4). The attenuation sensing matrix A is further obtained by combining the measurement matrix W and the attenuation sparse matrix G, where A = GW.

[0072]

[0073] In the formula, i represents an imaginary number, f is the frequency, f0 is the reference frequency, t is the time, and Q(t) is the quality factor at time t.

[0074] S4 performs a Fourier transform on the single-channel seismic data to convert it into frequency domain data, and calculates the time-frequency attenuation factor at each sampling point and the attenuation sensing matrix of the entire channel data according to step S3. Then, the sparse inversion of the reflection coefficient can be performed using the fast threshold shrinkage iterative algorithm.

[0075] The single-channel seismic data s(t) is transformed into frequency domain data S(f) by Fourier transform, and the attenuation sensing matrix of the data is calculated according to step S3. The inversion objective function is constructed as shown in equation (5).

[0076]

[0077] In the formula, |||2 represents the 2-norm, |||1 represents the 1-norm, A is the attenuation sensing matrix, R is the reflection coefficient to be solved, S is the frequency domain single-channel seismic data, and λ is the regularization factor.

[0078] The sparse inversion of reflection coefficients is performed using a fast threshold shrinkage iterative algorithm.

[0079] After S5, all seismic data are processed in a single loop to obtain sparse inversion results of reflection coefficients corresponding to all seismic data. Then, these results are convolved with a broadband seismic wavelet to finally obtain seismic data with improved resolution.

[0080] Broadband seismic wavelets employ higher-frequency Reck wavelets.

[0081] Example 2

[0082] In a specific embodiment 2 of the present invention, the present invention is applied. Figure 1 The flowchart below shows the thin interlayer target processing and recognition method based on Q-compensated compressed sensing of the present invention. The specific steps are as follows:

[0083] Step S1: Extract the wavelet w(t) of the original seismic record using the complex spectrum method.

[0084] First, a suitable time window range is selected for the original seismic record, preferably in an area with a stable phase axis and relatively slow changes. Then, the seismic traces in this area are subjected to Fourier transform to obtain their amplitude spectrum. Next, the amplitude spectrum is subjected to logarithmic transform and inverse Fourier transform to obtain the complex spectrum result. By setting the width of the passband, a low-pass filter is designed to filter the complex spectrum result and perform Fourier and inverse logarithmic transforms, finally obtaining the extracted wavelet of the original seismic record.

[0085] Step S2: Estimate the Q value of the original seismic record using the amplitude envelope method.

[0086] First, set the time window size and time window sliding step size. The corresponding amplitude spectrum A(f) for time-division window calculation can be weighted and averaged on adjacent channels (e.g., the weights of three channels can be 0.5:1:0.5). Next, convert the obtained amplitude spectrum into decibel form B(f) (as shown in equation (1)). Then, the least squares method can be used to calculate the slope k of the envelope in decibel form and estimate the Q value (as shown in equation (2)). Finally, the estimated Q value is interpolated so that each sample point has a Q value.

[0087] B(f)=20lg(A(f)) (1)

[0088] B(f)=-27.3fQ -1 t

[0089] Q = 27.3 t / |k (2)

[0090] In the formula, A(f) is the amplitude spectrum, B(f) is the decibel form of A(f), f is the frequency, t is the center time of the time window, k is the envelope slope, and Q is the quality factor.

[0091] Step S3: Construct the frequency domain diagonal wavelet measurement matrix, the time-frequency attenuation sparse matrix, and the attenuation sensing matrix.

[0092] Based on the seismic wavelet calculated in step S1, construct the frequency domain diagonal wavelet measurement matrix W (as shown in equation (3)).

[0093]

[0094] In the formula, W is the measurement matrix, diag is the diagonal function, FT is the forward Fourier transform, and w(t) is the extracted wavelet. For the wavelet w(t) with different frequency components f1, f2, ..., f M The corresponding spectral value, M is the sample number corresponding to the highest frequency component.

[0095] Based on the estimated quality factor Q calculated in step S2, the time-frequency attenuation factor is calculated for each trace to obtain the attenuation sparse matrix G of the seismic data for that trace (as shown in equation (4)). The attenuation sensing matrix A (A = GW) is further obtained by combining the measurement matrix W and the attenuation sparse matrix G.

[0096]

[0097] In the formula, i represents an imaginary number, f is the frequency, f0 is the reference frequency (usually the main frequency), t is the time, and Q(t) is the quality factor at time t.

[0098] Step S4: Perform Fourier transform on the single-channel seismic data s(t) to convert it into frequency domain data S(f), and calculate the attenuation sensing matrix of the data according to step S3. Construct the inversion objective function (as shown in Equation 5). Then, the sparse inversion of the reflection coefficient can be performed using the fast threshold shrinkage iterative algorithm.

[0099]

[0100] In the formula, |||2 represents the 2-norm, |||1 represents the 1-norm, A is the attenuation sensing matrix, R is the reflection coefficient to be solved, S is the frequency domain single-channel seismic data, and λ is the regularization factor, which can generally be set to 0.1.

[0101] Step S5: Then, perform single-channel cyclic processing on all seismic data to obtain sparse inversion results of reflection coefficients corresponding to all seismic data. Then, it can be convolved with a broadband seismic wavelet (such as a higher dominant frequency Ricker wavelet) to finally obtain seismic data with improved resolution.

[0102] Example 3

[0103] In a specific embodiment 3 of the present invention, the present invention will be further described in conjunction with a specific application example:

[0104] The method of this invention was first applied to a theoretical model to test its effectiveness and applicability, such as Figure 2 The results are for a one-dimensional forward model with 2ms sampling, 255 sampling points, and a 25Hz Ricker wavelet. Figure 3 The results are noise-free inversion and high-resolution processing results; Figure 4 The results are from the noise-added inversion and high-resolution processing. Figure 5 The results are shown in the 2D noisy inversion and high-resolution processing, where (a) shows the original 2D seismic record and (b) shows the 2D extended-spectrum seismic record. Figures 2 to 5 The processing results show that the present invention can achieve sparse inversion of reflection coefficients relatively accurately while ensuring a certain signal-to-noise ratio, which can effectively improve the resolution of seismic data and provide solid theoretical model support for practical applications.

[0105] Next, the invention was applied to a thin interbedded layer section in a certain work area, as shown in the work area plan. Figure 6 As shown, the original seismic profile of a certain well-connecting main survey line is as follows: Figure 7 As shown, sampling point 301, sampling interval 2 milliseconds. It can be seen that the original seismic profile has low resolution, the reservoir spontaneous potential response at the target layer is sensitive, but its seismic response energy is weak, the consistency between the reflected phase axis and the logging curve is low, and the thin interbedded information is superimposed on the waveform, failing to effectively represent it. Using the Q-compensated compressed sensing-based thin interbedded target processing and identification method of this invention, a higher resolution profile is obtained, as shown... Figure 8As shown, after target processing, the overall profile structure remains unchanged, while the ability to identify thin interbedded layers is improved, weak amplitude energy is strengthened, channel sandstone reservoirs are highlighted, lateral continuity is good, and the well-seismic correlation is good, proving the reliability of this method. Multichannel spectral analysis shows that ( Figure 9 The processing results can better expand high-frequency information and improve the high-frequency effective energy reflecting channel sand bodies, which is conducive to the further interpretation of thin interbedded reservoirs. This invention can provide a more accurate basis for the exploration and development of thin interbedded oil and gas reservoirs.

[0106] Finally, it should be noted that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0107] Except for the technical features described in the specification, all other technologies are known to those skilled in the art.

Claims

1. A method for processing and recognizing thin interlayer targets based on Q-compensated compressed sensing, characterized in that, This method for processing and recognizing thin interlayer targets based on Q-compensated compressed sensing includes: Step 1: Extract the wavelet from the original seismic record; Step 2: Estimate the Q-value of the original seismic record; Step 3: Construct the frequency domain diagonal wavelet measurement matrix, the time-frequency attenuation sparse matrix, and the attenuation sensing matrix; Step 4: Perform Fourier transform on the single-channel seismic data to convert it into frequency domain data, and calculate the time-frequency attenuation factor at each sampling point and the attenuation sensing matrix of the whole channel data according to Step 3. Then perform sparse inversion of reflection coefficients. Step 5: Perform single-channel loop processing on all seismic data to obtain sparse inversion results of reflection coefficients corresponding to all seismic data, and then perform convolution to obtain seismic data with improved resolution. In step 3, based on the wavelet obtained in step 1 and the quality factor Q obtained in step 2, a frequency domain diagonal wavelet measurement matrix and a time-frequency attenuation sparse matrix are constructed to further obtain the attenuation sensing matrix. In step 3, the frequency domain diagonal wavelet measurement matrix W is constructed based on the calculated seismic wavelet, as shown in equation (3): (3) In the formula, W is the measurement matrix, diag is the diagonal function, FT is the forward Fourier transform, and w(t) is the extracted wavelet. For different frequency components of wavelet w(t) The corresponding spectral value, M is the sample number corresponding to the highest frequency component; In step 3, based on the quality factor Q estimated in step S2, the time-frequency attenuation factor is calculated for each trace to obtain the attenuation sparse matrix G of the seismic data for that trace, as shown in equation (4). The attenuation sensing matrix A is further obtained by combining the measurement matrix W and the attenuation sparse matrix G, where A = GW. (4) In the formula, i represents an imaginary number, For frequency, Here, t is the reference frequency, and t is time. Let be the quality factor at time t.

2. The method for processing and recognizing thin interlayer targets based on Q-compensated compressed sensing according to claim 1, characterized in that, In step 1, basic information about the seismic data is obtained by performing spectral and waveform feature analysis, and then the wavelet of the original seismic record is extracted using the polyspectral method.

3. The method for processing and recognizing thin interlayer targets based on Q-compensated compressed sensing according to claim 2, characterized in that, Step 1 also includes adjusting the obtained wavelet frequency band and phase by sliding a time window based on the frequency band and phase of the data.

4. The thin-layer target processing and recognition method based on Q-compensated compressed sensing according to claim 3, characterized in that, In step 1, the time window should not be too large or too small, but should be designed to preserve the wavelet information at the origin of the complex spectrum while removing the reflection coefficient information at the non-origin points.

5. The method for processing and recognizing thin interlayer targets based on Q-compensated compressed sensing according to claim 2, characterized in that, In step 1, the wavelet of the original seismic record is extracted using the complex spectrum method. The specific steps are as follows: First, a time window range is selected for the original seismic record. Then, the seismic traces within this range are subjected to Fourier transform to obtain their amplitude spectrum. Next, the amplitude spectrum is subjected to logarithmic transform and inverse Fourier transform to obtain the complex spectrum result. A low-pass filter is designed by setting the width of the passband to filter the complex spectrum result and perform Fourier and inverse logarithmic transforms. Finally, the extracted wavelet of the original seismic record is obtained.

6. The method for processing and recognizing thin interlayer targets based on Q-compensated compressed sensing according to claim 1, characterized in that, In step 2, the quality factor Q of the original seismic record is estimated using the spectral ratio method or the amplitude envelope method. Then, the estimated quality factor is smoothed to increase the stability of the subsequent sparse inversion process.

7. The method for processing and recognizing thin interlayer targets based on Q-compensated compressed sensing according to claim 6, characterized in that, In step 2, the Q-value of the original seismic record is estimated using the amplitude envelope method. The specific steps are as follows: First, set the time window size and time window sliding step size, then calculate the corresponding amplitude spectrum for the time-division window. A weighted average is then applied to adjacent channels; subsequently, the resulting amplitude spectrum is converted into decibel representation. As shown in equation (1), the slope of the envelope in decibel form is then calculated using the least squares method. The calculation and Q-value estimation are performed as shown in equation (2). Finally, the estimated Q-values ​​are interpolated to ensure that each sample point has a Q-value. (1) (2) In the formula, For amplitude spectrum, for In decibel form, For frequency, The center time of the time window Let be the slope of the envelope. This is the quality factor.

8. The method for processing and recognizing thin interlayer targets based on Q-compensated compressed sensing according to claim 1, characterized in that, In step 4, the single-channel seismic data is converted into frequency domain data by Fourier transform, and the time-frequency attenuation factor at each sampling point and the attenuation sensing matrix of the whole channel data are calculated according to step 3. Then, the sparse inversion of reflection coefficients is performed using the fast threshold shrinkage iterative algorithm.

9. The method for processing and recognizing thin interlayer targets based on Q-compensated compressed sensing according to claim 8, characterized in that, In step 4, the single-channel seismic data... Perform a Fourier transform to convert it into frequency domain data. And calculate the attenuation sensing matrix of the data according to step 3, and construct the inversion objective function as shown in equation (5). (5) In the formula, |||2 represents the 2-norm, |||1 represents the 1-norm, A is the attenuation sensing matrix, and R is the solution for the reflection coefficient. This is frequency domain single-channel seismic data. It is a regularization factor.

10. The method for processing and recognizing thin interlayer targets based on Q-compensated compressed sensing according to claim 1, characterized in that, In step 5, all seismic data are processed in a single loop to obtain sparse inversion results of reflection coefficients corresponding to all seismic data. Then, these results are convolved with a broadband seismic wavelet to finally obtain seismic data with improved resolution.

11. The method for processing and recognizing thin interlayer targets based on Q-compensated compressed sensing according to claim 10, characterized in that, In step 5, the broadband seismic wavelet adopts the higher dominant frequency Ricker wavelet.

Images