A multi-scale fault identification method based on sandstone-type uranium mine frequency division seismic attribute

Abstract

The present application belongs to the technical field of sandstone type uranium mine seismic exploration interpretation, and particularly relates to a multi-scale fault identification method based on sandstone type uranium mine frequency division seismic attribute, which comprises the following steps: step 1, inputting original three-dimensional post-stack seismic data volume; step 2, pre-decomposing the original three-dimensional post-stack seismic data volume; step 3, performing VMD on each trace of the whole data volume; step 4, calculating the coherence volume of each IMF frequency division seismic data by using C3 algorithm; and step 5, frequency division outputting the coherence volume and finely identifying multi-scale faults. The method can realize multi-scale fine identification of different scale faults in target layer section in sandstone type uranium exploration on low signal-to-noise ratio seismic post-stack three-dimensional data volume.

Description

Technical Field

[0001] This invention belongs to the field of seismic exploration and interpretation technology for sandstone-type uranium deposits, specifically relating to a multi-scale fault identification method based on frequency-division seismic attributes of sandstone-type uranium deposits. Background Technology

[0002] In sandstone-type uranium exploration, deep faults and fissures are one of the main channels for the upward escape of deep oil and gas in the area. The vicinity of deep faults and unconformities is often a favorable location for the enrichment of sandstone-type uranium deposits. Therefore, effective identification of faults at different scales is a key research objective in seismic interpretation. Generally, fault identification techniques in oil and gas seismic exploration can be directly applied, mainly discontinuity detection methods in seismic attributes. However, on the one hand, faults at different scales cannot be clearly imaged on a single data volume and attribute volume; on the other hand, the data itself is affected by noise, making fault identification at small and medium scales relatively difficult. Therefore, it is considered to decompose seismic data at multiple scales and denoise it, and then perform discontinuity detection on the data at each scale (frequency band), such as the extraction and analysis of attributes like coherence.

[0003] Multi-scale decomposition employs spectral decomposition. Currently, commonly used methods for seismic data spectral decomposition include Short-Time Fourier Transform, Wavelet Transform, S-Transform, Matching Pursuit Decomposition, and Empirical Mode Decomposition (EMD), among others. This study uses Variational Mode Decomposition (VMD), a method similar to EMD but fundamentally different in principle, to obtain seismic data volumes in different frequency bands. Compared to EMD-like methods, VMD has a more complete mathematical foundation, avoids endpoint effects and mode aliasing, and exhibits higher noise resistance. However, VMD requires a pre-defined number of modes K; how to adaptively obtain the parameter K is a key issue.

[0004] Coherence attributes are extracted from the denoised single-band data volume. Coherence algorithms have undergone several generations of development: the first generation was based on crosscorrelation (C1 algorithm), the second generation on multi-channel similarity (C2 algorithm), and the third generation on matrix eigenvalues ​​and feature structure (C3 algorithm). Relatively speaking, the C3 algorithm has higher lateral resolution and stronger noise resistance than the previous two generations, hence it is adopted in this study. Coherence attributes are mainly used for interpreting geological structures and sedimentary environments. Utilizing the principle of finding commonalities and differences, it can more clearly identify subsurface discontinuities or discontinuities. Points with lower coherence values ​​correspond to geological discontinuities (such as faults, boundaries of special lithological bodies). Slicing the coherence data volume (isochronous slices, stratigraphic slices) can reveal geological phenomena such as faults, lithological body boundaries, and unconformities. Summary of the Invention

[0005] The purpose of this invention is to provide a multi-scale fault identification method based on frequency-division seismic attributes of sandstone-type uranium deposits. This method can achieve multi-scale fine identification of faults of different scales in target segments during sandstone-type uranium deposit exploration on low signal-to-noise ratio post-stack 3D seismic data volumes.

[0006] Technical solution to achieve the purpose of this invention:

[0007] A multi-scale fault identification method based on frequency-division seismic attributes of sandstone-type uranium deposits, the method comprising:

[0008] Step 1: Input the original 3D post-stack seismic data volume;

[0009] Step 2: Pre-decompose the original 3D post-stack seismic data volume;

[0010] Step 3: Perform VMD on each channel of the entire data volume;

[0011] Step 4: Use the C3 algorithm to calculate the coherence volume of each IMF frequency-division seismic data;

[0012] Step 5: Analyze the frequency-divided coherent volume to accurately identify multi-scale faults.

[0013] Step 2 includes:

[0014] Step 2.1: Select several seismic traces from the data volume on average, and perform a random selection of each seismic trace s. i Perform VMD pre-decomposition, calculate the root mean square energy of each original seismic trace and the decomposed IMF, and the energy difference between each decomposition;

[0015] Step 2.2: Based on the energy conservation criterion of VMD decomposition, determine the optimal IMF decomposition number K for each seismic trace signal. i ;

[0016] Step 2.3: Optimal IMF decomposition number K for all seismic trace signals. i The average is calculated and rounded to obtain the VMD decomposition parameter K of the overall 3D data volume.

[0017] Step 2.1 specifically involves: initializing K=2, and for each seismic trace s i Perform VMD pre-decomposition, calculate the root mean square energy of each original seismic trace and the decomposed IMF, and the energy difference between each decomposition;

[0018] The energy calculation formula for each original seismic trace is:

[0019]

[0020] The formula for calculating the root mean square energy of the decomposed IMF is:

[0021]

[0022] The formula for calculating the energy difference in each decomposition is:

[0023]

[0024] In the formula, E(S) i The energy of each original seismic trace;

[0025] E(IMF i (m) ) represents the root mean square energy of the decomposed IMF;

[0026] j represents the j-th sampling point in each channel;

[0027] s(x i ,y i ,t j ) represents the seismic amplitude data corresponding to the j-th sampling point of the i-th seismic trace;

[0028] N T For record length;

[0029] IMF i (m) The m-th IMF derived from the i-th seismic trace;

[0030] (x i ,y i ,t j () represents the j-th sampling point of the i-th seismic trace;

[0031] m is the mth IMF obtained from the decomposition;

[0032] K iThe total number of IMFs extracted from the i-th seismic data;

[0033] δ represents the energy difference in the decomposition.

[0034] Step 2.2 specifically involves: applying the energy conservation criterion based on VMD decomposition. If VMD is under-decomposed or properly decomposed, then δ≈0; let K=K+1, and perform the next calculation; if VMD is over-decomposed, non-existent components will appear, then... The δ value can change abruptly; the more it is over-decomposed, the larger δ becomes. Therefore, when the δ value changes abruptly, the loop ends, and K = K-1 is set as the optimal IMF decomposition number K for that signal. i .

[0035] Step 3 specifically involves adding a pair of white noises of equal length to the signal, opposite sign, mean of 0, and standard deviation to the data volume of each seismic trace, to obtain the IMF. + IMF - For the IMF + IMF - The IMF values ​​are averaged to obtain the final IMF sequence for each seismic trace data volume.

[0036] Step 4 includes:

[0037] Step 4.1: Define the coherence window;

[0038] Step 4.2: Calculate the correlation matrix;

[0039] Step 4.3: Calculate the coherence value at the sample point.

[0040] The coherence window in step 4.1 is:

[0041]

[0042] D is the coherence window, containing J data points, with a window length of N and d. NJ Let be the amplitude value of the Nth sample point in the Jth channel.

[0043] The formula for calculating the correlation matrix in step 4.2 is as follows:

[0044]

[0045] The formula for calculating the coherence value at the sample point in step 4.3 is as follows:

[0046]

[0047] In the formula, Tr(C) is the trace of matrix C. λ j Let λ be the j-th eigenvalue among J eigenvalues. maxE is the largest eigenvalue among them. c C represents the coherence value obtained from the sample points. jj These are the values ​​of the elements on the diagonal of the correlation matrix C.

[0048] Step 5 specifically involves outputting K IMF single-band coherent data volumes, extracting the survey line profile and time slice of the same target layer segment in sequence, or setting different slices to be transparent and superimposed to display the spatial distribution of faults at different levels in the target layer segment as a whole, thereby achieving multi-scale fine identification of faults.

[0049] The beneficial technical effects of this invention are as follows:

[0050] This invention provides a multi-scale fault identification method based on frequency-division seismic attributes of sandstone-type uranium deposits. Combining an improved variational mode decomposition method and coherent attribute extraction technology, it utilizes seismic data to better characterize the distribution information of multi-scale faults in the sandstone-type uranium deposit study area. While denoising the post-stack 3D seismic data, it can more clearly identify key structural information related to the enrichment of sandstone-type uranium deposits, such as fractures, fissures, cracks, and karst caves of different levels. Attached Figure Description

[0051] Figure 1 The flowchart illustrates a multi-scale fault identification method based on frequency-division seismic attributes of sandstone-type uranium deposits provided by this invention. Detailed Implementation

[0052] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0053] like Figure 1 As shown, the present invention provides a multi-scale fault identification method based on frequency-division seismic attributes of sandstone-type uranium deposits, which specifically includes the following steps:

[0054] Step 1: Input the original 3D post-stack seismic data volume s(x,y,t), with a record length of N. T .

[0055] Step 2: Pre-decompose the original 3D post-stack seismic data volume:

[0056] Step 2.1: Select several seismic traces s from the data volume on average. i =s(x i ,y i ,t)(i=1,…,I), generally I does not need to be too large. Initialize K=2, for each seismic trace s i Perform VMD pre-decomposition and calculate the root mean square energy of each original seismic trace and the decomposed IMF:

[0057]

[0058] In the formula, E(S) i The energy of each original seismic trace;

[0059] E(IMF i (m) ) represents the root mean square energy of the decomposed IMF;

[0060] j represents the j-th sampling point in each channel;

[0061] s(x i ,y i ,t j ) represents the seismic amplitude data corresponding to the j-th sampling point of the i-th seismic trace;

[0062] N T For record length

[0063] IMF i (m) The m-th IMF derived from the i-th seismic trace;

[0064] (x i ,y i ,t j () represents the j-th sampling point of the i-th seismic trace;

[0065] m is the mth IMF obtained from the decomposition;

[0066] K i denoted as the total number of IMFs derived from the i-th seismic data.

[0067] Calculate the energy difference for each decomposition:

[0068]

[0069] In the formula, δ represents the energy difference of the decomposition.

[0070] Step 2.2: Energy conservation criterion based on VMD decomposition. If VMD is under-decomposed or properly decomposed, then δ≈0; let K = K+1, and perform the next calculation. If VMD is over-decomposed, non-existent components will appear. The δ value can change abruptly; the more it is over-decomposed, the larger δ becomes. Therefore, when the δ value changes abruptly, the loop ends, and K = K-1 is set as the optimal IMF decomposition number K for that signal. i .

[0071] Step 2.3: For the selected I channels of data, a set of optimal K values ​​{K} is finally obtained. i}(i=1,…,I), take the average of them and round down. The VMD decomposition K-parameters are used as the total 3D data volume.

[0072] Step 3: Perform VMD on each channel of the entire data volume:

[0073] Considering the possibility of introducing noise under a uniform K parameter, a pair of white noise n(t) with the same length as the signal, opposite sign, mean of 0, and standard deviation is added to each data point s(t):

[0074]

[0075] Among them, s + s(t) and s-(t) are the results of adding or subtracting the same white noise sequence n(t) from the original data, respectively.

[0076] get One IMF data volume:

[0077]

[0078] The average of the positive and negative IMFs is used to obtain the final IMF sequence for each data point:

[0079]

[0080] s respectively + (t) and s - (t) represents the component obtained after VMD.

[0081] Step 4: Using the C3 algorithm, calculate the coherence volume of each IMF frequency-divided seismic data set:

[0082] Step 4.1: Define the coherence window

[0083]

[0084] D is the coherence window, containing J data points, with a window length of N and d. NJ Let be the amplitude value of the Nth sample point in the Jth trace. The window size varies from small to large for different fault levels to specifically identify faults of different scales.

[0085] Step 4.2: Calculate the correlation matrix

[0086]

[0087] Step 4.3: Calculate the coherence value E at sample point s(x,y,t). c for

[0088]

[0089] Where Tr(C) is the trace of matrix C. λ jLet λ be the j-th eigenvalue among J eigenvalues. max E is the largest eigenvalue among them. c The range of values ​​for C is [1 / J, 1]. jj E represents the element values ​​on the diagonal of the correlation matrix C. When the waveforms within the time window are indistinguishable, i.e., there are no tomographic faults, E... c =1; When there are differences in the waveform, i.e., when there is discontinuity, E c <1. And the greater the waveform difference between channels, the higher E c The smaller the value.

[0090] Step 5, Output the results:

[0091] K IMF single-band coherent data volumes can be output separately, and the survey line profile and time slice of the same target segment can be extracted in turn. Different slices can also be set to transparent and superimposed for display. The spatial distribution of faults at different levels of the target segment can be displayed as a whole, thereby realizing multi-scale fine identification of faults.

[0092] The present invention has been described in detail above with reference to the accompanying drawings and embodiments. However, the present invention is not limited to the above embodiments, and various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention. All contents not described in detail in the present invention can be derived from existing technologies.

Claims

1. A multi-scale fault identification method based on frequency-division seismic attributes of sandstone-type uranium deposits, characterized in that, The method includes: Step 1: Input the original 3D post-stack seismic data volume; Step 2: Pre-decompose the original 3D post-stack seismic data volume; Step 3: Perform VMD on each channel of the entire data volume; Step 4: Use the C3 algorithm to calculate the coherence volume of each IMF frequency-division seismic data; Step 5: Analyze the frequency-divided coherent volume for precise identification of multi-scale faults; Step 2 includes: Step 2.1: Select several seismic traces from the data volume on average, and process each seismic trace... Perform VMD pre-decomposition, calculate the root mean square energy of each original seismic trace and the decomposed IMF, and the energy difference between each decomposition; Step 2.2: Based on the energy conservation criterion of VMD decomposition, determine the optimal number of IMF decompositions for each seismic trace signal. K i ; Step 2.3: Optimal IMF decomposition number for all seismic trace signals. K i The average is calculated and rounded to obtain the VMD decomposition parameter K of the entire 3D data volume; Step 2.1 specifically involves: initialization K =2, for each seismic trace Perform VMD pre-decomposition, calculate the root mean square energy of each original seismic trace and the decomposed IMF, and the energy difference between each decomposition; The energy calculation formula for each original seismic trace is: ； The formula for calculating the root mean square energy of the decomposed IMF is: ； The formula for calculating the energy difference in each decomposition is: ； In the formula, E(S i ) The energy of each original seismic trace; E(IMF i (m) ) The root mean square energy of the decomposed IMF; j This refers to the j-th sampling point in each channel; s(x i ,y i ,t j ) This refers to the seismic amplitude data corresponding to the j-th sampling point of the i-th seismic trace; N T For record length; IMF i (m) The m-th IMF derived from the i-th seismic trace; (x i ,y i ,t j ) This refers to the j-th sampling point of the i-th seismic trace; m This is the m-th IMF derived from the decomposition; K i The total number of IMFs extracted from the i-th seismic data; δ represents the energy difference in the decomposition process; Step 2.2 specifically involves: applying the energy conservation criterion based on VMD decomposition. If VMD is under-decomposed or decomposed appropriately, then ;make K = K +1, execute the next calculation; if VMD over-decomposes, non-existent components may appear. , The value can change abruptly; the more it is over-decomposed, the more... The larger, therefore when When a value changes abruptly, end the loop and set... K = K -1 is taken as the optimal IMF decomposition number for this signal. K i .

2. The multi-scale fault identification method based on frequency-division seismic attributes of sandstone-type uranium deposits according to claim 1, characterized in that, Step 3 specifically involves adding a pair of white noises of equal length to the signal, opposite sign, mean of 0, and standard deviation to the data volume of each seismic trace, to obtain the IMF. + IMF - For the IMF + IMF - The IMF values ​​are averaged to obtain the final IMF sequence for each seismic trace data volume.

3. The multi-scale fault identification method based on frequency-division seismic attributes of sandstone-type uranium deposits according to claim 2, characterized in that, Step 4 includes: Step 4.1: Define the coherence window; Step 4.2: Calculate the correlation matrix; Step 4.3: Calculate the coherence value at the sample point.

4. The multi-scale fault identification method based on frequency-division seismic attributes of sandstone-type uranium deposits according to claim 3, characterized in that, The coherence window in step 4.1 is: ， D is the coherent window, and the window contains a total of J Data, time window length is N , d NJ For the first J Dao Di N The amplitude value of each sample point.

5. The multi-scale fault identification method based on frequency-division seismic attributes of sandstone-type uranium deposits according to claim 4, characterized in that, The formula for calculating the correlation matrix in step 4.2 is as follows: 。 6. The multi-scale fault identification method based on frequency-division seismic attributes of sandstone-type uranium deposits according to claim 5, characterized in that, The formula for calculating the coherence value at the sample point in step 4.3 is as follows: ， In the formula, Let C be the trace of matrix C. for J The th eigenvalue j 1 eigenvalue, The largest eigenvalue among them, E c The coherence values ​​are those obtained from the sample points. C jj These are the values ​​of the elements on the diagonal of the correlation matrix C.

7. The multi-scale fault identification method based on frequency-division seismic attributes of sandstone-type uranium deposits according to claim 6, characterized in that, Step 5 specifically involves: outputting K IMF single-band coherent data volumes respectively, extracting the survey line profile and time slice of the same target layer segment in turn, setting different slices to be transparent and superimposing them to display the spatial distribution of faults of different levels in the target layer segment as a whole, thereby realizing multi-scale fine identification of faults.

Images