Frequency division wave impedance inversion method and device, electronic equipment and medium

Abstract

The invention discloses a frequency division wave impedance inversion method and device, electronic equipment and a medium. The method comprises the following steps: converting seismic data and logging data into the same domain, and acquiring a plurality of frequency division data volumes according to the seismic data; obtaining a wave impedance curve according to the logging data; establishing a sample data set according to the frequency division data volume and the wave impedance curve; obtaining a wave impedance inversion result of each frequency division data volume based on the sample data set; and obtaining a full-band seismic wave impedance inversion result based on a firefly algorithm according to the wave impedance inversion result. According to the method, on the basis of fully utilizing high, medium and low frequency information in an effective frequency band of seismic data, the relation between amplitude and frequency is introduced into a wave impedance inversion process, and full-band seismic wave impedance inversion is realized by taking a GRU neural network and a firefly algorithm as technical means.

Description

Technical Field

[0001] This invention relates to the field of oil and gas geophysical exploration, and more specifically, to a frequency-division impedance inversion method, apparatus, electronic equipment, and medium. Background Technology

[0002] Seismic impedance inversion refers to the seismic interpretation technique that uses seismic data to invert subsurface impedance information. It is one of the important methods for reservoir prediction and oil reservoir characteristic description, and it is a research hotspot and challenge in geophysical exploration.

[0003] Seismic impedance is one of the important parameters that can reflect changes in reservoir properties, and its accurate prediction can help in the exploration and development of oil and gas. However, current seismic impedance inversion relies on seismic wavelets and low-frequency models, resulting in limited inversion accuracy.

[0004] Currently, the main methods for wave impedance inversion based on seismic data are as follows:

[0005] (1) Wave impedance inversion method based on convolution model. This type of method is greatly affected by low frequency model and wavelet, and the inversion result has multiple solutions.

[0006] (2) Wave impedance inversion method based on wave equation. This type of method has complex algorithms, is sensitive to noise, and is difficult to obtain a stable solution.

[0007] Therefore, it is necessary to develop a frequency-division impedance inversion method, device, electronic equipment, and medium.

[0008] The information disclosed in the background section of this invention is intended only to enhance the understanding of the general background of this invention, and should not be construed as an admission or in any way implying that such information constitutes prior art known to those skilled in the art. Summary of the Invention

[0009] This invention proposes a frequency-division wave impedance inversion method, device, electronic equipment, and medium, which comprehensively utilizes the powerful nonlinear fitting ability of the GRU neural network and the global optimization ability of the firefly algorithm. The calculation results are stable, and the integration of seismic data and well data improves the accuracy and reliability of wave impedance inversion, with high wave impedance inversion consistency.

[0010] In a first aspect, embodiments of this disclosure provide a frequency-division wave impedance inversion method, including:

[0011] Seismic data and well logging data are converted into the same domain, and multiple frequency-division data volumes are obtained based on the seismic data;

[0012] The wave impedance curve is obtained based on the well logging data;

[0013] A sample dataset is established based on the frequency division data and the wave impedance curve;

[0014] Based on the sample dataset, the wave impedance inversion results of each frequency-division data volume are obtained;

[0015] Based on the impedance inversion results, the full-band seismic impedance inversion results are obtained using the Firefly algorithm.

[0016] As a specific implementation of this disclosure, acquiring multiple frequency-division data volumes based on seismic data includes:

[0017] Based on the analysis of the spectral characteristics of seismic data, multiple frequency-division data volumes are obtained using time-frequency analysis methods.

[0018] As a specific implementation of this disclosure, obtaining the wave impedance curve based on the logging data includes:

[0019] The wave impedance curve in the well logging data is resampled to obtain a wave impedance curve consistent with the seismic frequency band.

[0020] As a specific implementation of this disclosure, obtaining the wave impedance inversion results of each frequency-division data volume based on the sample dataset includes:

[0021] Based on the sample dataset, the wave impedance inversion model of each frequency division data is obtained using the GRU neural network, and then the wave impedance inversion result of each frequency division data volume is obtained.

[0022] As a specific implementation of this disclosure, the network structure of the GRU neural network includes a reset gate and an update gate, wherein the calculation formulas for the reset gate, the update gate, the node state, and the node output are as follows:

[0023] Z t =σ(W z ·[y t-1 ,X t ])

[0024] r t =σ(W r ·[y t-1 ,X t ])

[0025]

[0026] Among them, X t For input information, y t-1 Output information at time t-1, y t Output information at time t, r t To update the gate state, z t To reset the door state, Ht Let σ represent the current node state, tanh be the activation functions, and W be the shared parameters. For y t Mean.

[0027] As a specific implementation of this disclosure, obtaining full-band seismic wave impedance inversion results based on the wave impedance inversion results and the Firefly algorithm includes:

[0028] After obtaining the wave impedance inversion results of each frequency division data, the optimal weight coefficients of each frequency division wave impedance inversion results are obtained based on the relationship between amplitude and frequency and the firefly algorithm, thereby obtaining the full-band seismic wave impedance inversion results.

[0029] As one specific implementation of this disclosure, the firefly algorithm includes:

[0030] Set parameters and initialize the firefly positions;

[0031] Update the position of individual fireflies using a position update formula;

[0032] Substitute the updated position vector of each individual into the objective function to be optimized, and update the brightness of all firefly individuals.

[0033] Wherein, if the brightness of firefly i at time t is less than that of firefly j, then the position update formula is:

[0034] W i (t+1)=W i (t)+β(d ij )*{W j (t)-W i (t)}+α*(rand-1 / 2)

[0035] If firefly i has the highest brightness at time t, then the position update formula is:

[0036] W i (t+1)=W i (t)+α*(rand-1 / 2)

[0037] Where β0 represents the maximum attraction, and γ is the brightness attenuation coefficient; d ij W represents the distance between firefly i and firefly j; i (t), W i (t+1) represents the position of the i-th firefly at time t and time t+1; β(d ij The number () represents the attractiveness of firefly i to firefly j.

[0038] Secondly, embodiments of this disclosure also provide a frequency-division impedance inversion device, comprising:

[0039] The frequency division module converts seismic data and well logging data into the same domain and obtains multiple frequency division data volumes based on the seismic data.

[0040] The calculation module obtains the wave impedance curve based on the well logging data;

[0041] The dataset creation module creates a sample dataset based on the frequency division data volume and the wave impedance curve.

[0042] The training module obtains the wave impedance inversion results of each frequency-division data volume based on the sample dataset;

[0043] The inversion module obtains full-band seismic wave impedance inversion results based on the wave impedance inversion results and the Firefly algorithm.

[0044] As a specific implementation of this disclosure, acquiring multiple frequency-division data volumes based on seismic data includes:

[0045] Based on the analysis of the spectral characteristics of seismic data, multiple frequency-division data volumes are obtained using time-frequency analysis methods.

[0046] As a specific implementation of this disclosure, obtaining the wave impedance curve based on the logging data includes:

[0047] The wave impedance curve in the well logging data is resampled to obtain a wave impedance curve consistent with the seismic frequency band.

[0048] As a specific implementation of this disclosure, obtaining the wave impedance inversion results of each frequency-division data volume based on the sample dataset includes:

[0049] Based on the sample dataset, the wave impedance inversion model of each frequency division data is obtained using the GRU neural network, and then the wave impedance inversion result of each frequency division data volume is obtained.

[0050] As a specific implementation of this disclosure, the network structure of the GRU neural network includes a reset gate and an update gate, wherein the calculation formulas for the reset gate, the update gate, the node state, and the node output are as follows:

[0051] Z t =σ(W z ·[y t-1 ,X t ])

[0052] r t =σ(W r ·[y t-1 ,Xt ])

[0053]

[0054] Among them, X t For input information, y t-1 Output information at time t-1, y t Output information at time t, r t To update the gate state, z t To reset the door state, H t Let σ represent the current node state, tanh be the activation functions, and W be the shared parameters. For y t Mean.

[0055] As a specific implementation of this disclosure, obtaining full-band seismic wave impedance inversion results based on the wave impedance inversion results and the Firefly algorithm includes:

[0056] After obtaining the wave impedance inversion results of each frequency division data, the optimal weight coefficients of each frequency division wave impedance inversion results are obtained based on the relationship between amplitude and frequency and the firefly algorithm, thereby obtaining the full-band seismic wave impedance inversion results.

[0057] As one specific implementation of this disclosure, the firefly algorithm includes:

[0058] Set parameters and initialize the firefly positions;

[0059] Update the position of individual fireflies using a position update formula;

[0060] Substitute the updated position vector of each individual into the objective function to be optimized, and update the brightness of all firefly individuals.

[0061] Wherein, if the brightness of firefly i at time t is less than that of firefly j, then the position update formula is:

[0062] W i (t+1)=W i (t)+β(d ij )*{W j (t)-W i (t)}+α*(rand-1 / 2)

[0063] If firefly i has the highest brightness at time t, then the position update formula is:

[0064] W i (t+1)=W i (t)+α*(rand-1 / 2)

[0065] Where β0 represents the maximum attraction, and γ is the brightness attenuation coefficient; d ij W represents the distance between firefly i and firefly j; i (t), W i (t+1) represents the position of the i-th firefly at time t and time t+1; β(d ij The number () represents the attractiveness of firefly i to firefly j.

[0066] Thirdly, embodiments of this disclosure also provide an electronic device, the electronic device comprising:

[0067] Memory, which stores executable instructions;

[0068] A processor that executes the executable instructions in the memory to implement the frequency division impedance inversion method.

[0069] Fourthly, embodiments of this disclosure also provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the frequency division impedance inversion method.

[0070] Its beneficial effects are as follows:

[0071] (1) This invention uses GRU neural network to perform frequency division data volume impedance inversion, which can make full use of the powerful nonlinear fitting capability of GRU neural network to obtain the frequency division impedance inversion results of each frequency division data volume.

[0072] (2) Based on the firefly algorithm, this invention can obtain the optimal weighting coefficients for different frequency division wave impedance inversion, and further obtain the conversion relationship between the full-band wave impedance and each frequency division wave impedance.

[0073] (3) This invention provides a technical process for frequency-division wave impedance inversion. By introducing the relationship between amplitude and frequency into the wave impedance inversion process, the full-band wave impedance inversion is realized by using the GRU neural network and the firefly algorithm.

[0074] The methods and apparatus of the present invention have other features and advantages that will be apparent from or will be set forth in detail in the accompanying drawings and following detailed description, which together serve to explain the particular principles of the invention. Attached Figure Description

[0075] The above and other objects, features and advantages of the present invention will become more apparent from the more detailed description of exemplary embodiments of the invention in conjunction with the accompanying drawings, wherein the same reference numerals generally represent the same parts.

[0076] Figure 1 A flowchart illustrating the steps of a frequency-division impedance inversion method according to an embodiment of the present invention is shown.

[0077] Figure 2 A schematic diagram of a typical wellbore vibration calibration according to an embodiment of the present invention is shown.

[0078] Figure 3 A schematic diagram of a 5-15Hz frequency-division seismic data profile is shown according to an embodiment of the present invention.

[0079] Figure 4 A schematic diagram of a raw seismic data profile is shown according to an embodiment of the present invention.

[0080] Figure 5 A schematic diagram of a 5-15Hz frequency division impedance inversion profile according to an embodiment of the present invention is shown.

[0081] Figure 6 A schematic diagram of a full-band wave impedance inversion profile according to an embodiment of the present invention is shown.

[0082] Figure 7 A block diagram of a frequency division impedance inversion device according to an embodiment of the present invention is shown.

[0083] Explanation of reference numerals in the attached figures:

[0084] 201. Frequency division module; 202. Calculation module; 203. Dataset creation module; 204. Training module; 205. Inversion module. Detailed Implementation

[0085] Preferred embodiments of the invention will now be described in more detail. While preferred embodiments of the invention are described below, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0086] To facilitate understanding of the solutions and effects of the embodiments of the present invention, six specific application examples are given below. Those skilled in the art should understand that these examples are merely for the purpose of understanding the present invention, and any specific details therein are not intended to limit the present invention in any way.

[0087] Example 1

[0088] Figure 1 A flowchart illustrating the steps of a frequency-division impedance inversion method according to an embodiment of the present invention is shown.

[0089] like Figure 1 As shown, the frequency division impedance inversion method includes:

[0090] Step 101: Convert the seismic data and well logging data to the same domain, and obtain multiple frequency-division data volumes based on the seismic data;

[0091] Step 102: Obtain the wave impedance curve based on the well logging data;

[0092] Step 103: Establish a sample dataset based on the frequency division data volume and wave impedance curve;

[0093] Step 104: Obtain the wave impedance inversion results for each frequency division data volume based on the sample dataset;

[0094] Step 105: Based on the wave impedance inversion results, obtain the full-band seismic wave impedance inversion results using the Firefly algorithm.

[0095] In one example, obtaining multiple frequency-division data volumes from seismic data includes:

[0096] Based on the analysis of the spectral characteristics of seismic data, multiple frequency-division data volumes are obtained using time-frequency analysis methods.

[0097] In one example, the wave impedance curve obtained from well logging data includes:

[0098] The wave impedance curves in the well logging data are resampled to obtain wave impedance curves consistent with the seismic frequency band.

[0099] In one example, the wave impedance inversion results for each frequency-division data volume obtained based on the sample dataset include:

[0100] Based on the sample dataset, the wave impedance inversion model of each frequency division data is obtained using the GRU neural network, and then the wave impedance inversion results of each frequency division data volume are obtained.

[0101] In one example, the network structure of the GRU neural network includes a reset gate and an update gate, wherein the calculation formulas for the reset gate, the update gate, the node state, and the node output are as follows:

[0102] z t =σ(W z ·[y t-1 ,X t ])

[0103] r t =σ(W r ·[y t-1 ,X t ])

[0104]

[0105] Among them, X t For input information, y t-1Output information at time t-1, y t Output information at time t, r t To update the gate state, z t To reset the door state, H t Let σ represent the current node state, tanh be the activation functions, and W be the shared parameters. For y t Mean.

[0106] In one example, based on the wave impedance inversion results, and using the Firefly algorithm, the full-band seismic wave impedance inversion results are obtained, including:

[0107] After obtaining the wave impedance inversion results of each frequency division data, the optimal weight coefficients of each frequency division wave impedance inversion results are obtained based on the relationship between amplitude and frequency and the firefly algorithm, thereby obtaining the full-band seismic wave impedance inversion results.

[0108] In one example, the firefly algorithm includes:

[0109] Set parameters and initialize the firefly positions;

[0110] Update the position of individual fireflies using a position update formula;

[0111] Substitute the updated position vector of each individual into the objective function to be optimized, and update the brightness of all firefly individuals.

[0112] Wherein, if the brightness of firefly i at time t is less than that of firefly j, then the position update formula is:

[0113] W i (t+1)=W i (t)+β(d ij )*{W j (t)-W i (t)}+α*(rand-1 / 2)

[0114] If firefly i has the highest brightness at time t, then the position update formula is:

[0115] W i (t+1)=W i (t)+α*(rand-1 / 2)

[0116] Where β0 represents the maximum attraction, and γ is the brightness attenuation coefficient; d ij W represents the distance between firefly i and firefly j; i (t), W i (t+1) represents the position of the i-th firefly at time t and time t+1; β(d ij The number () represents the attractiveness of firefly i to firefly j.

[0117] Specifically, seismic data and well logging data are converted into the same domain. Based on the analysis of the spectral characteristics of seismic data, multiple frequency-division data volumes are obtained using high-precision time-frequency analysis methods. The well logging impedance curves are resampled to obtain impedance curves consistent with the seismic frequency band.

[0118] Extract the wellbore side channel frequency-division data and create a sample dataset with the wellbore acoustic impedance. Based on the sample dataset, obtain the acoustic impedance inversion model for each frequency-division data using a GRU neural network, and further obtain the acoustic impedance inversion results for each frequency-division data volume.

[0119] The GRU neural network is an improvement on the LSTM neural network. It simplifies the network structure while preserving long-term memory, thus improving computational efficiency. Its network structure mainly consists of reset gates and update gates. The calculation formulas for the reset gate, update gate, node state, and node output are as follows:

[0120] Z t =σ(W z ·[y t-1 ,X t ])

[0121] r t =σ(W r ·[y t-1 ,X t ])

[0122]

[0123] Among them, X t For input information, y t-1 Output information at time t-1, y t Output information at time t, r t To update the gate state, z t To reset the door state, H t Let σ represent the current node state, tanh be the activation functions, and W be the shared parameters. For y t Mean.

[0124] After obtaining the wave impedance inversion results of each frequency division data, the optimal weight coefficients of each frequency division wave impedance inversion results are obtained by using the amplitude-frequency relationship and the firefly algorithm, and then the full-band seismic wave impedance inversion results are obtained.

[0125] The core idea of ​​the firefly algorithm is that weakly fluorescent fireflies are attracted to strongly fluorescent ones. Position updates are viewed as an iterative process, with multiple iterations used to find the optimal position. The algorithm has three main stages: initialization, position update, and brightness update. The initialization stage involves parameter setting and initializing the firefly positions. The position update stage calculates the position of each individual firefly according to the defined position update formula. The brightness update stage substitutes the updated position vectors of each individual firefly into the objective function to update the brightness of all fireflies. If the brightness of firefly i at time t is less than that of firefly j, then the position update formula is:

[0126] W i (t+1)=W i (t)+β(d ij )*{W j (t)-W i (t)}+α*(rand-1 / 2)

[0127] If firefly i has the highest brightness at time t, then the position update formula is:

[0128] W i (t+1)=W i (t)+α*(rand-1 / 2)

[0129] Where β0 represents the maximum attraction, and γ is the brightness attenuation coefficient; d ij W represents the distance between firefly i and firefly j; i (t), W i (t+1) represents the position of the i-th firefly at time t and time t+1; β(d ij The number () represents the attractiveness of firefly i to firefly j.

[0130] Example 2

[0131] The present invention also provides a frequency-division impedance inversion device, comprising:

[0132] The frequency division module converts seismic data and well logging data into the same domain and obtains multiple frequency division data volumes based on the seismic data;

[0133] The calculation module obtains the wave impedance curve based on the well logging data;

[0134] The dataset creation module establishes a sample dataset based on the frequency division data volume and wave impedance curve.

[0135] The training module obtains the wave impedance inversion results for each frequency-division data volume based on the sample dataset;

[0136] The inversion module obtains full-band seismic wave impedance inversion results based on the wave impedance inversion results and the Firefly algorithm.

[0137] In one example, obtaining multiple frequency-division data volumes from seismic data includes:

[0138] Based on the analysis of the spectral characteristics of seismic data, multiple frequency-division data volumes are obtained using time-frequency analysis methods.

[0139] In one example, the wave impedance curve obtained from well logging data includes:

[0140] The wave impedance curves in the well logging data are resampled to obtain wave impedance curves consistent with the seismic frequency band.

[0141] In one example, the wave impedance inversion results for each frequency-division data volume obtained based on the sample dataset include:

[0142] Based on the sample dataset, the wave impedance inversion model of each frequency division data is obtained using the GRU neural network, and then the wave impedance inversion results of each frequency division data volume are obtained.

[0143] In one example, the network structure of the GRU neural network includes a reset gate and an update gate, wherein the calculation formulas for the reset gate, the update gate, the node state, and the node output are as follows:

[0144] z t =σ(W z ·[y t-1 ,X t ])

[0145] r t =σ(W r ·[y t-1 ,X t ])

[0146]

[0147] Among them, X t For input information, y t-1 Output information at time t-1, y t Output information at time t, r t To update the gate state, z t To reset the door state, H t Let σ represent the current node state, tanh be the activation functions, and W be the shared parameters. For y t Mean.

[0148] In one example, based on the wave impedance inversion results, and using the Firefly algorithm, the full-band seismic wave impedance inversion results are obtained, including:

[0149] After obtaining the wave impedance inversion results of each frequency division data, the optimal weight coefficients of each frequency division wave impedance inversion results are obtained based on the relationship between amplitude and frequency and the firefly algorithm, thereby obtaining the full-band seismic wave impedance inversion results.

[0150] In one example, the firefly algorithm includes:

[0151] Set parameters and initialize the firefly positions;

[0152] Update the position of individual fireflies using a position update formula;

[0153] Substitute the updated position vector of each individual into the objective function to be optimized, and update the brightness of all firefly individuals.

[0154] Wherein, if the brightness of firefly i at time t is less than that of firefly j, then the position update formula is:

[0155] W i (t+1)=W i (t)+β(d ij )*{W j (t)-W i (t)}+α*(rand-1 / 2)

[0156] If firefly i has the highest brightness at time t, then the position update formula is:

[0157] W i (t+1)=W i (t)+α*(rand-1 / 2)

[0158] Where β0 represents the maximum attraction, and γ is the brightness attenuation coefficient; d ij W represents the distance between firefly i and firefly j; i (t), W i (t+1) represents the position of the i-th firefly at time t and time t+1; β(d ij The number () represents the attractiveness of firefly i to firefly j.

[0159] Specifically, seismic data and well logging data are converted into the same domain. Based on the analysis of the spectral characteristics of seismic data, multiple frequency-division data volumes are obtained using high-precision time-frequency analysis methods. The well logging impedance curves are resampled to obtain impedance curves consistent with the seismic frequency band.

[0160] Extract the wellbore side channel frequency-division data and create a sample dataset with the wellbore acoustic impedance. Based on the sample dataset, obtain the acoustic impedance inversion model for each frequency-division data using a GRU neural network, and further obtain the acoustic impedance inversion results for each frequency-division data volume.

[0161] The GRU neural network is an improvement on the LSTM neural network. It simplifies the network structure while preserving long-term memory, thus improving computational efficiency. Its network structure mainly consists of reset gates and update gates. The calculation formulas for the reset gate, update gate, node state, and node output are as follows:

[0162] z t =σ(W z ·[y t-1 ,X t ])

[0163] r t =σ(W r ·[y t-1 ,X t ])

[0164]

[0165] Among them, X t For input information, y t-1 Output information at time t-1, y t Output information at time t, r t To update the gate state, z t To reset the door state, H t Let σ represent the current node state, tanh be the activation functions, and W be the shared parameters. For y t Mean.

[0166] After obtaining the wave impedance inversion results of each frequency division data, the optimal weight coefficients of each frequency division wave impedance inversion results are obtained by using the amplitude-frequency relationship and the firefly algorithm, and then the full-band seismic wave impedance inversion results are obtained.

[0167] The core idea of ​​the firefly algorithm is that weakly fluorescent fireflies are attracted to strongly fluorescent ones. Position updates are viewed as an iterative process, with multiple iterations used to find the optimal position. The algorithm has three main stages: initialization, position update, and brightness update. The initialization stage involves parameter setting and initializing the firefly positions. The position update stage calculates the position of each individual firefly according to the defined position update formula. The brightness update stage substitutes the updated position vectors of each individual firefly into the objective function to update the brightness of all fireflies. If the brightness of firefly i at time t is less than that of firefly j, then the position update formula is:

[0168] W i (t+1)=W i (t)+β(d ij )*{W j (t)-W i(t)}+α*(rand-1 / 2)

[0169] If firefly i has the highest brightness at time t, then the position update formula is:

[0170] W i (t+1)=W i (t)+α*(rand-1 / 2)

[0171] Where β0 represents the maximum attraction, and γ is the brightness attenuation coefficient; d ij W represents the distance between firefly i and firefly j; i (t), W i (t+1) represents the position of the i-th firefly at time t and time t+1; β(d ij The number () represents the attractiveness of firefly i to firefly j.

[0172] Example 3

[0173] Taking data from a specific work area in China as an example, and integrating seismic logging data, we conducted a wave impedance inversion study based on frequency-division data.

[0174] Figure 2 A schematic diagram of a typical wellbore vibration calibration according to an embodiment of the present invention is shown.

[0175] By converting seismic and well logging data to the same domain, the calibration results of typical wells in the study area are as follows: Figure 2 As shown, the target well-seismic matching rate is high.

[0176] Figure 3 A schematic diagram of a 5-15Hz frequency-division seismic data profile is shown according to an embodiment of the present invention.

[0177] Figure 4 A schematic diagram of a raw seismic data profile is shown according to an embodiment of the present invention.

[0178] Based on the analysis of the spectral characteristics of seismic data, multiple frequency-divided data volumes are obtained using a high-precision time-frequency analysis method (matching pursuit algorithm). Figure 3 and Figure 4 The 5-15Hz frequency-divided seismic data and the original seismic data profile are displayed.

[0179] The well logging impedance curves were resampled to obtain impedance curves consistent with the seismic frequency band.

[0180] Frequency-divided data from wellbore access channels were extracted and combined with surface impedance data to create a sample dataset. This dataset was then normalized and divided into training and test sets at a 3:7 ratio. Data from eight wells in the study area was complete; six wells were randomly selected as the training set, and two as the test set.

[0181] Figure 5 A schematic diagram of a 5-15Hz frequency division impedance inversion profile according to an embodiment of the present invention is shown.

[0182] Based on the sample set, and using a GRU neural network, the wave impedance inversion model for each frequency-division data is obtained, and the wave impedance inversion results for each frequency-division data volume are further obtained. Figure 5 The impedance inversion results of 5-15Hz frequency-division seismic data are presented, by Figure 5 It can be seen that the overall trend of the wave impedance inversion results based on 5-15Hz frequency-division seismic data is consistent with that of the well, but there are some deviations in the details.

[0183] Figure 6 A schematic diagram of a full-band wave impedance inversion profile according to an embodiment of the present invention is shown.

[0184] After obtaining the wave impedance inversion results of each frequency division data, the optimal weight coefficients of each frequency division wave impedance inversion results are obtained by using the amplitude-frequency relationship and the firefly algorithm, and then the full-band seismic wave impedance inversion results are obtained. Figure 6 The full-band wave impedance inversion results are shown. As can be seen from the figure, the inversion results are in high agreement with the wellbore wave impedance results, and the inversion results are accurate, proving that the wave impedance inversion accuracy of the present invention is high.

[0185] Example 4

[0186] Figure 7 A block diagram of a frequency division impedance inversion device according to an embodiment of the present invention is shown.

[0187] like Figure 7 As shown, the frequency division impedance inversion device includes:

[0188] Frequency division module 201 converts seismic data and well logging data into the same domain and obtains multiple frequency division data volumes based on the seismic data;

[0189] Calculation module 202 obtains the wave impedance curve based on the well logging data;

[0190] The dataset creation module 203 creates a sample dataset based on the frequency division data volume and wave impedance curve.

[0191] Training module 204 obtains the wave impedance inversion results of each frequency division data volume based on the sample dataset;

[0192] Inversion module 205 obtains full-band seismic wave impedance inversion results based on the wave impedance inversion results and the Firefly algorithm.

[0193] In one example, obtaining multiple frequency-division data volumes from seismic data includes:

[0194] Based on the analysis of the spectral characteristics of seismic data, multiple frequency-division data volumes are obtained using time-frequency analysis methods.

[0195] In one example, the wave impedance curve obtained from well logging data includes:

[0196] The wave impedance curves in the well logging data are resampled to obtain wave impedance curves consistent with the seismic frequency band.

[0197] In one example, the wave impedance inversion results for each frequency-division data volume obtained based on the sample dataset include:

[0198] Based on the sample dataset, the wave impedance inversion model of each frequency division data is obtained using the GRU neural network, and then the wave impedance inversion results of each frequency division data volume are obtained.

[0199] In one example, the network structure of the GRU neural network includes a reset gate and an update gate, wherein the calculation formulas for the reset gate, the update gate, the node state, and the node output are as follows:

[0200] z t =σ(W z ·[y t-1 ,X t ])

[0201] r t =σ(W r ·[y t-1 ,X t ])

[0202]

[0203] Among them, X t For input information, y t-1 Output information at time t-1, y t Output information at time t, r t To update the gate state, z t To reset the door state, H t Let σ represent the current node state, tanh be the activation functions, and W be the shared parameters. For y t Mean.

[0204] In one example, based on the wave impedance inversion results, and using the Firefly algorithm, the full-band seismic wave impedance inversion results are obtained, including:

[0205] After obtaining the wave impedance inversion results of each frequency division data, the optimal weight coefficients of each frequency division wave impedance inversion results are obtained based on the relationship between amplitude and frequency and the firefly algorithm, thereby obtaining the full-band seismic wave impedance inversion results.

[0206] In one example, the firefly algorithm includes:

[0207] Set parameters and initialize the firefly positions;

[0208] Update the position of individual fireflies using a position update formula;

[0209] Substitute the updated position vector of each individual into the objective function to be optimized, and update the brightness of all firefly individuals.

[0210] Wherein, if the brightness of firefly i at time t is less than that of firefly j, then the position update formula is:

[0211] W i (t+1)=W i (t)+β(d ij )*{W j (t)-W i (t)}+α*(rand-1 / 2)

[0212] If firefly i has the highest brightness at time t, then the position update formula is:

[0213] W i (t+1)=W i (t)+α*(rand-1 / 2)

[0214] Where β0 represents the maximum attraction, and γ is the brightness attenuation coefficient; d ij W represents the distance between firefly i and firefly j; i (t), W i (t+1) represents the position of the i-th firefly at time t and time t+1; β(d ij The number () represents the attractiveness of firefly i to firefly j.

[0215] Example 5

[0216] This disclosure provides an electronic device, comprising: a memory storing executable instructions; and a processor executing the executable instructions in the memory to implement the above-described frequency division impedance inversion method.

[0217] An electronic device according to an embodiment of the present disclosure includes a memory and a processor.

[0218] This memory is used to store non-transitory computer-readable instructions. Specifically, the memory may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may, for example, include random access memory (RAM) and / or cache memory. The non-volatile memory may, for example, include read-only memory (ROM), hard disk, flash memory, etc.

[0219] The processor may be a central processing unit (CPU) or other form of processing unit with data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device to perform desired functions. In one embodiment of this disclosure, the processor is used to execute computer-readable instructions stored in the memory.

[0220] Those skilled in the art will understand that, in order to solve the technical problem of how to achieve a good user experience, this embodiment may also include well-known structures such as communication buses and interfaces, and these well-known structures should also be included within the protection scope of this disclosure.

[0221] For a detailed description of this embodiment, please refer to the corresponding descriptions in the foregoing embodiments, which will not be repeated here.

[0222] Example 6

[0223] This disclosure provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the frequency division impedance inversion method.

[0224] A computer-readable storage medium according to embodiments of the present disclosure stores non-transitory computer-readable instructions. When these non-transitory computer-readable instructions are executed by a processor, all or part of the steps of the methods described in the foregoing embodiments of the present disclosure are performed.

[0225] The aforementioned computer-readable storage media include, but are not limited to: optical storage media (e.g., CD-ROM and DVD), magneto-optical storage media (e.g., MO), magnetic storage media (e.g., magnetic tape or portable hard drive), media with built-in rewritable non-volatile memory (e.g., memory card), and media with built-in ROM (e.g., ROM cartridge).

[0226] Those skilled in the art should understand that the above description of the embodiments of the present invention is only intended to illustrate the beneficial effects of the embodiments of the present invention, and is not intended to limit the embodiments of the present invention to any of the examples given.

[0227] The various embodiments of the present invention have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments.

Claims

1. A method for frequency-division wave impedance inversion, characterized in that, include: Seismic data and well logging data are converted into the same domain, and multiple frequency-division data volumes are obtained based on the seismic data; The wave impedance curve is obtained based on the well logging data; A sample dataset is established based on the frequency division data and the wave impedance curve; Based on the sample dataset, the wave impedance inversion results of each frequency-division data volume are obtained; Based on the impedance inversion results, the full-band seismic impedance inversion results are obtained using the Firefly algorithm.

2. The frequency division wave impedance inversion method according to claim 1, wherein, Multiple frequency-division data volumes were obtained from earthquake data, including: Based on the analysis of the spectral characteristics of seismic data, multiple frequency-division data volumes are obtained using time-frequency analysis methods.

3. The frequency division wave impedance inversion method according to claim 1, wherein, The wave impedance curve obtained from the well logging data includes: The wave impedance curve in the well logging data is resampled to obtain a wave impedance curve consistent with the seismic frequency band.

4. The frequency division wave impedance inversion method according to claim 1, wherein, The wave impedance inversion results obtained based on the aforementioned sample dataset for each frequency-division data volume include: Based on the sample dataset, the wave impedance inversion model of each frequency division data is obtained using the GRU neural network, and then the wave impedance inversion result of each frequency division data volume is obtained.

5. The frequency division wave impedance inversion method according to claim 4, wherein, The network structure of the GRU neural network includes a reset gate and an update gate, wherein the calculation formulas for the reset gate, the update gate, the node state, and the node output are as follows: z t =σ(W z ·[y t-1 ,X t ]) r t =σ(W r ·[y t-1 ,X t ]) Among them, X t For input information, y t-1 Output information at time t-1, y t Output information at time t, r t To update the gate state, z t To reset the door state, H t Let σ represent the current node state, tanh be the activation functions, and W be the shared parameters. For y t Mean.

6. The frequency division wave impedance inversion method according to claim 1, wherein, Based on the impedance inversion results, and using the firefly algorithm, the full-band seismic impedance inversion results are obtained, including: After obtaining the wave impedance inversion results of each frequency division data, the optimal weight coefficients of each frequency division wave impedance inversion results are obtained based on the relationship between amplitude and frequency and the firefly algorithm, thereby obtaining the full-band seismic wave impedance inversion results.

7. The frequency division wave impedance inversion method according to claim 1, wherein, The firefly algorithm includes: Set parameters and initialize the firefly positions; Update the position of individual fireflies using a position update formula; Substitute the updated position vector of each individual into the objective function to be optimized, and update the brightness of all firefly individuals. Wherein, if the brightness of firefly i at time t is less than that of firefly j, then the position update formula is: W i (t+1)=W i (t)+β(d ij )*{W j (t)-W i (t)}+α*(rand-1 / 2) If firefly i has the highest brightness at time t, then the position update formula is: W i (t+1)=W i (t)+α*(rand-1 / 2) Where β0 represents the maximum attraction, and γ is the brightness attenuation coefficient; d ij W represents the distance between firefly i and firefly j; i (t), W i (t+1) represents the position of the i-th firefly at time t and time t+1; β(d ij The number () represents the attractiveness of firefly i to firefly j.

8. A frequency-division impedance inversion device, characterized in that, include: The frequency division module converts seismic data and well logging data into the same domain and obtains multiple frequency division data volumes based on the seismic data. The calculation module obtains the wave impedance curve based on the well logging data; The dataset creation module creates a sample dataset based on the frequency division data volume and the wave impedance curve. The training module obtains the wave impedance inversion results of each frequency-division data volume based on the sample dataset; The inversion module obtains full-band seismic wave impedance inversion results based on the wave impedance inversion results and the Firefly algorithm.

9. An electronic device, characterized in that, The electronic device includes: Memory, which stores executable instructions; A processor that executes the executable instructions in the memory to implement the frequency division impedance inversion method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the frequency division impedance inversion method according to any one of claims 1-7.

Images