A low-frequency compensation method, apparatus, equipment, medium, and product

Abstract

This invention discloses a low-frequency compensation method, apparatus, device, medium, and product. The method includes: acquiring target post-stack seismic data; inputting the target post-stack seismic data into a target low-frequency compensation model to obtain low-frequency compensated seismic data. The target low-frequency compensation model is obtained by iteratively training an initial low-frequency compensation model using a target sample set, where the target sample set includes: post-stack seismic data samples and corresponding well logging curves. The technical solution of this invention can improve the reliability of low-frequency compensated seismic data, providing a data foundation for improving the quality of post-stack inversion.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of exploration technology, and in particular to a low-frequency compensation method, apparatus, equipment, medium and product. Background Technology

[0002] Low-frequency seismic information is crucial for improving imaging accuracy, inversion quality, and the accuracy of oil and gas detection results. However, due to factors such as field acquisition and various noises, the low-frequency signal energy in seismic data is weak or even completely absent. Therefore, low-frequency signal compensation plays a very important role in seismic exploration, especially in seismic interpretation and inversion.

[0003] Traditional low-frequency compensation methods mainly use well logging data interpolated along the stratigraphic plane to establish a low-frequency model. However, these methods suffer from the following problems: the constructed low-frequency model is limited by the accuracy of well logging calibration and stratigraphic interpretation, making it difficult to guarantee the reliability of the low-frequency components. Summary of the Invention

[0004] This invention provides a low-frequency compensation method, apparatus, equipment, medium, and product to improve the reliability of low-frequency compensated seismic data and provide a data foundation for improving post-stack inversion quality.

[0005] According to one aspect of the present invention, a low-frequency compensation method is provided, comprising:

[0006] Acquire target post-stack seismic data;

[0007] The target post-stack seismic data is input into the target low-frequency compensation model to obtain low-frequency compensated seismic data. The target low-frequency compensation model is obtained by iteratively training an initial low-frequency compensation model using a target sample set. The target sample set includes: post-stack seismic data samples and well logging curves corresponding to the post-stack seismic data samples.

[0008] According to another aspect of the present invention, a low-frequency compensation device is provided, the low-frequency compensation device comprising:

[0009] The target post-stack seismic data acquisition module is used to acquire target post-stack seismic data;

[0010] The low-frequency compensated seismic data determination module is used to input the target post-stack seismic data into the target low-frequency compensation model to obtain the low-frequency compensated seismic data. The target low-frequency compensation model is obtained by iteratively training an initial low-frequency compensation model using a target sample set. The target sample set includes: post-stack seismic data samples and well logging curves corresponding to the post-stack seismic data samples.

[0011] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0012] At least one processor; and

[0013] A memory communicatively connected to the at least one processor; wherein,

[0014] The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the low-frequency compensation method according to any embodiment of the present invention.

[0015] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the low-frequency compensation method according to any embodiment of the present invention.

[0016] According to another aspect of the present invention, a computer program product is provided, which, when executed by a processor, implements the low-frequency compensation method as described in any of the embodiments of the present invention.

[0017] This invention, in its embodiments, acquires target post-stack seismic data; inputs the target post-stack seismic data into a target low-frequency compensation model to obtain low-frequency compensated seismic data. The target low-frequency compensation model is obtained by iteratively training an initial low-frequency compensation model using a target sample set, which includes: post-stack seismic data samples and corresponding well logging curves. This improves the reliability of the low-frequency compensated seismic data, providing a data foundation for improving the quality of post-stack inversion.

[0018] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart of a low-frequency compensation method according to an embodiment of the present invention;

[0021] Figure 2 This is a schematic diagram of the structure of a low-frequency compensation device according to an embodiment of the present invention;

[0022] Figure 3This is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0023] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0024] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0025] It is understood that before using the technical solutions disclosed in the various embodiments of this disclosure, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in this disclosure in an appropriate manner in accordance with relevant laws and regulations, and user authorization should be obtained.

[0026] Example 1

[0027] Figure 1 This is a flowchart of a low-frequency compensation method provided in an embodiment of the present invention. This embodiment is applicable to the case of low-frequency compensation of post-stack seismic data. The method can be executed by the low-frequency compensation device in this embodiment of the invention, which can be implemented in software and / or hardware, such as... Figure 1 As shown, the method specifically includes the following steps:

[0028] S110, acquire target post-stack seismic data.

[0029] In this embodiment, the target post-stack seismic data can be obtained by either acquiring the post-stack seismic data to be compensated for low frequency, or by acquiring the post-stack seismic data required for inversion.

[0030] S120, input the target post-stack seismic data into the target low-frequency compensation model to obtain the low-frequency compensated seismic data.

[0031] In this embodiment, the target low-frequency compensation model is obtained by iteratively training the initial low-frequency compensation model using a target sample set, which includes: post-stack seismic data samples and well logging curves corresponding to the post-stack seismic data samples.

[0032] In this embodiment, the method for inputting the target post-stack seismic data into the target low-frequency compensation model to obtain low-frequency compensated seismic data can be as follows: normalize the target post-stack seismic data to obtain normalized target post-stack seismic data; input the normalized target post-stack seismic data into the target low-frequency compensation model to obtain low-frequency compensated seismic data.

[0033] In this embodiment, the method of iteratively training the initial low-frequency compensation model through the target sample set can be as follows: input the post-stack seismic data samples from the target sample set into the initial low-frequency compensation model to obtain the predicted low-frequency compensated seismic data, and train the parameters of the initial low-frequency compensation model based on the objective function formed by the predicted low-frequency compensated seismic data and the post-stack seismic data samples.

[0034] Optionally, an initial low-frequency compensation model is iteratively trained using the target sample set, including:

[0035] Establish an initial low-frequency compensation model.

[0036] In this embodiment, the initial low-frequency compensation model can be a neural network model. For example, the initial low-frequency compensation model can be the U-NET model, or commonly used image segmentation network architectures in deep learning, such as SegNet, Deplab, etc.

[0037] The post-stack seismic data samples from the target sample set are input into the initial low-frequency compensation model to obtain the predicted low-frequency compensated seismic data.

[0038] It should be noted that before inputting the post-stack seismic data samples from the target sample set into the initial low-frequency compensation model to obtain the predicted low-frequency compensated seismic data, the post-stack seismic data samples from the target sample set need to be normalized in advance, and then the normalized post-stack seismic data samples are input into the initial low-frequency compensation model to obtain the predicted low-frequency compensated seismic data.

[0039] In this embodiment, the method for normalizing the post-stack seismic data samples in the target sample set can be as follows: normalize the post-stack seismic data samples according to the mean and standard deviation of the post-stack seismic data samples to obtain normalized post-stack seismic data samples.

[0040] The parameters of the initial low-frequency compensation model are trained based on the objective function formed by the predicted low-frequency compensated seismic data and the post-stack seismic data samples.

[0041] In this embodiment, the objective function includes at least one of a self-supervised loss function, a weakly supervised loss function, and a structural constraint loss function.

[0042] In this embodiment, the parameters of the initial low-frequency compensation model can be trained using the objective function formed by the predicted low-frequency compensated seismic data and the post-stack seismic data samples as follows: Based on the predicted low-frequency compensated seismic data, determine the low-frequency seismic data and local gradient vector at the well logging location; determine the low-frequency logging data based on the logging curves corresponding to the post-stack seismic data samples; determine the reflection feature vector based on the post-stack seismic data samples; and train the parameters of the initial low-frequency compensation model using the objective function formed by the predicted low-frequency compensated seismic data, the post-stack seismic data samples, the low-frequency seismic data at the well logging location, the local gradient vector, the low-frequency logging data, and the reflection feature vector.

[0043] It should be noted that, based on the predicted low-frequency compensated seismic data, the method for determining the low-frequency seismic data of the well location and the local gradient vector can be as follows: extract the low-frequency seismic data of the well location from the predicted low-frequency compensated seismic data, and obtain the local gradient vector of the predicted low-frequency compensated seismic data.

[0044] In this embodiment, the method for determining low-frequency logging data based on the logging curves corresponding to the post-stack seismic data samples can be as follows: determine the logging reflection coefficient based on the logging curves corresponding to the post-stack seismic data samples; and determine the low-frequency logging data based on the logging reflection coefficients.

[0045] In this embodiment, the parameters of the initial low-frequency compensation model are trained using an objective function formed from the predicted low-frequency compensated seismic data, post-stack seismic data samples, low-frequency seismic data at well locations, local gradient vectors, low-frequency well logging data, and reflection feature vectors. This can be achieved by: forming a self-supervised loss function based on the post-stack seismic data samples and the predicted low-frequency compensated seismic data; forming a weakly supervised loss function based on the low-frequency well logging data and the low-frequency seismic data at well locations within the predicted low-frequency compensated seismic data; forming a structural constraint loss function based on the reflection feature vector and the local gradient vector of the predicted low-frequency compensated seismic data; and training the parameters of the initial low-frequency compensation model using the objective function formed by the self-supervised loss function, the weakly supervised loss function, and the structural constraint loss function. The process then returns to inputting the post-stack seismic data samples from the target sample set into the initial low-frequency compensation model to obtain the predicted low-frequency compensated seismic data, repeating this process until the target low-frequency compensation model is obtained.

[0046] Optionally, the parameters of the initial low-frequency compensation model are trained based on the objective function formed by the predicted low-frequency compensated seismic data and post-stack seismic data samples, including:

[0047] Extract low-frequency seismic data of well locations from the predicted low-frequency compensated seismic data.

[0048] In this embodiment, the method for extracting low-frequency seismic data of well locations from the predicted low-frequency compensated seismic data can be as follows: extracting low-frequency seismic data from the predicted low-frequency compensated seismic data through a low-pass filter function, and then extracting low-frequency seismic data of well locations from the predicted low-frequency compensated seismic data according to a mask matrix. The mask matrix is ​​a 0-1 matrix, in which well locations are 1 and the rest are 0.

[0049] In a specific example, low-frequency seismic data for well locations in the predicted low-frequency compensated seismic data are extracted based on the following formula:

[0050] mask×Lf l (xout);

[0051] In this embodiment, mask is a mask matrix, which is a matrix of 0s and 1s. In the mask matrix, the logging position is 1 and the rest are 0. l () is the low-pass filter function, and xout is the predicted low-frequency compensated seismic data.

[0052] The well logging reflection coefficient is determined based on the well logging curves corresponding to the post-stack seismic data samples.

[0053] In this embodiment, the well logging reflection coefficient can be determined based on the well logging curves corresponding to the post-stack seismic data samples by performing well logging calibration on the well logging curves corresponding to the post-stack seismic data samples, and then processing the calibrated well logging curves to obtain the well logging reflection coefficients.

[0054] Low-frequency logging data is determined based on the logging reflection coefficient.

[0055] In this embodiment, the method for determining low-frequency logging data based on the logging reflection coefficient can be as follows: extract low-frequency data from the logging reflection coefficient based on a low-pass filter function, and then multiply the extracted low-frequency data with the matching coefficient to determine the low-frequency logging data.

[0056] In a specific example, low-frequency logging data is determined based on the following formula:

[0057] k×Lf l (r);

[0058] In this embodiment, k is the matching coefficient, and Lf l() is the low-pass filter function, and r is the logging reflection coefficient.

[0059] Based on the post-stack seismic data samples, the reflection feature vector is determined.

[0060] In this embodiment, the reflection feature vector can be determined based on the post-stack seismic data samples by: obtaining a structural feature tensor based on the post-stack seismic data samples; and extracting the reflection feature vector from the structural feature tensor.

[0061] Obtain the local gradient vector of the predicted low-frequency compensated seismic data.

[0062] A self-supervised loss function is formed based on the post-stack seismic data sample, the predicted low-frequency compensated seismic data, and the first formula.

[0063] In this embodiment, the first formula is:

[0064]

[0065] In this embodiment, Loss self This is the self-monitored loss value. Let x be the high-pass filter function. in For post-stack seismic data samples, x out To predict seismic data after low-frequency compensation.

[0066] A weakly supervised loss function is formed based on the low-frequency logging data, the low-frequency seismic data of the well locations in the predicted low-frequency compensated seismic data, and the second formula.

[0067] In this embodiment, the second formula is:

[0068]

[0069] In this embodiment, Loss sup For the weakly supervised loss value, k×L fl (r) represents low-frequency logging data, mask×Lf l (xout) represents the low-frequency seismic data used to predict well locations in the low-frequency compensated seismic data, k is the matching coefficient, and Lf l () is the low-pass filter function, r is the logging reflection coefficient, and mask is the mask matrix, where the logging position in the mask matrix is ​​1 and the rest are 0.

[0070] The structural constraint loss function is formed based on the reflection feature vector, the local gradient vector of the predicted low-frequency compensated seismic data, and the third formula.

[0071] In this embodiment, the third formula is:

[0072]

[0073] Among them, Loss constraint Let g be the structural constraint loss value, and g be the local gradient vector [g1, g2]. T Let g1 be the horizontal component of the local gradient vector, g2 be the vertical component of the local gradient vector, and the reflection eigenvector u = [u1, u2]. T u1 is the horizontal component of the feature vector, and u2 is the vertical component of the feature vector.

[0074] The parameters of the initial low-frequency compensation model are trained based on the objective function formed by the self-supervised loss function, the weakly supervised loss function, the structural constraint loss function, the weights corresponding to the self-supervised loss function, the weights corresponding to the weakly supervised loss function, and the weights corresponding to the structural constraint loss function.

[0075] In this embodiment, the objective function is:

[0076]

[0077] In this embodiment, β is the weight of the structural constraint loss function. Through technical practice, this weight is generally selected between 0.1 and 0.01, depending on the noise level of the seismic data itself. When the signal-to-noise ratio of the input post-stack seismic data is low, a higher β is generally selected.

[0078] Optional,

[0079] The first formula is:

[0080]

[0081] Where Lossself is the self-supervised loss value, Hf h () represents the high-pass filter function, xin represents the post-stack seismic data sample, and x out To predict seismic data after low-frequency compensation;

[0082] The second formula is:

[0083]

[0084] Among them, Loss sup For the weakly supervised loss value, k×L fl (r) represents low-frequency logging data, mask×L fl (x out ) represents low-frequency seismic data used to predict well locations in low-frequency compensated seismic data, where K is the matching coefficient and Lf is the low-frequency seismic data. l() is the low-pass filter function, r is the logging reflection coefficient, and mask is the mask matrix. In the mask matrix, the logging position is 1 and the rest are 0.

[0085] The third formula is:

[0086]

[0087] Among them, Loss constraint Let g be the structural constraint loss value, and g be the local gradient vector [g1, g2]. T Let g1 be the horizontal component of the local gradient vector, g2 be the vertical component of the local gradient vector, and the reflection eigenvector u = [u1, u2]. T u1 is the horizontal component of the feature vector, and u2 is the vertical component of the feature vector.

[0088] Optionally, low-frequency logging data can be determined based on the logging reflection coefficient, including:

[0089] The well logging reflection coefficient is bandpass filtered to obtain the passband signal of the well logging reflection coefficient.

[0090] In this embodiment, the method of obtaining the passband signal of the well logging reflection coefficient by bandpass filtering can be as follows: the intersection of the high-frequency signal and the low-frequency signal of the well logging reflection coefficient is determined as the passband signal of the well logging reflection coefficient.

[0091] Wellside seismic trace data are extracted from the well logging curves corresponding to the post-stack seismic data samples.

[0092] Bandpass filtering is performed on the well-side seismic trace data to obtain the passband signal of the well-side seismic trace data.

[0093] In this embodiment, the method of obtaining the passband signal of the well-side seismic trace data by performing bandpass filtering on the well-side seismic trace data can be as follows: the intersection of the high-frequency signal and the low-frequency signal of the well-side seismic trace data is determined as the passband signal of the well-side seismic trace data.

[0094] The matching coefficient is determined based on the passband signal of the well logging reflection coefficient, the passband signal of the well-side seismic data, and the fourth formula;

[0095] The fourth formula is:

[0096]

[0097] In this embodiment, scale is a compensation factor, and rb an d represents the passband signal of the logging reflection coefficient, sb and represents the passband signal of the seismic trace data near the well;

[0098] Low-frequency logging data are determined based on the matching coefficient and the logging reflection coefficient.

[0099] In a specific example, determining the matching coefficient is crucial because the normalized post-stack seismic data samples and the predicted low-frequency compensated seismic data will exhibit overall differences in energy and polarity compared to the low-frequency logging data. To match this energy difference, a passband signal is first obtained by bandpass filtering of the logging reflection coefficients, and simultaneously, the passband signal of the well-side seismic trace s is obtained. From this, the matching coefficient k can be determined as:

[0100]

[0101] In this embodiment, scale is generally set to 1.0, but it can be selected according to the actual situation to adjust the amount of low-frequency energy compensation. By introducing the matching operator k, the energy relationship between the high-frequency data and the low-frequency compensated seismic data can be kept consistent.

[0102] Optionally, the target post-stack seismic data is input into the target low-frequency compensation model to obtain low-frequency compensated seismic data, including:

[0103] Obtain the mean and standard deviation of the target post-stack seismic data;

[0104] The target post-stack seismic data is normalized based on the mean and standard deviation of the target post-stack seismic data to obtain normalized target post-stack seismic data.

[0105] The normalized target post-stack seismic data is input into the target low-frequency compensation model to obtain low-frequency compensated seismic data.

[0106] In a specific example, the objective function includes: a self-supervised loss function, a weakly supervised loss function, and a structural constraint loss function. Iteratively training the initial low-frequency compensation model using the target sample set involves the following steps:

[0107] Self-supervised loss function: Unlike the one-dimensional signal modeling method for logging and well-side channels in existing well-seismic joint intelligent low-frequency extension, the self-supervised loss function uses a two-dimensional convolutional network model for the initial low-frequency compensation model, which facilitates the implementation of overall structural constraints. Before inputting the post-stack seismic data samples into the initial low-frequency compensation model, the post-stack seismic data samples need to be normalized.

[0108] P s =(P-μ P ) / σ P ;

[0109] Where P represents the post-stack seismic data sample before normalization, P s For the normalized post-stack seismic data samples, μ P σ is the mean of the post-stack seismic data samples before normalization. P denoted as μ, representing the standard deviation of the post-stack seismic data samples before normalization. Because low-frequency information is missing in the post-stack seismic data samples, μ is generally not considered normal. P =0. The initial low-frequency compensation model uses a U-net two-dimensional convolutional network structure as its basic framework. The input to the model is the normalized post-stack seismic data sample P. s The extracted seismic profile data is processed into a tensor structure, resulting in seismic profile data x. in If the size is H×W, then the input tensor size of the initial low-frequency compensation model is (B, H, W, 1) or (B, 1, H, W), where B is the number of samples input to the network in a single training iteration, i.e., the batch size. Note: For clarity, the input tensor is defined as (B, 1, H, W) here and thereafter, where 1 represents the number of channels. Correspondingly, the output of the initial low-frequency compensation model is the predicted low-frequency compensated seismic data x. out The data has the same dimensions (B, 1, H, W). First, a self-supervised constraint is introduced, ensuring that the mid-to-high frequencies of the output data remain consistent with the mid-to-high frequencies of the original data. The high-pass filter function Hf... h (), the cutoff frequency is fh (due to the effective frequency band characteristics of seismic data, generally f h =10Hz (or can be selected based on the actual seismic data), then the self-supervised loss function is:

[0110]

[0111] The self-supervised loss function aims to maintain the consistency of mid-to-high frequency information within the effective frequency band of seismic data. Furthermore, this loss function can be applied to all seismic traces, whether they are well-side or not.

[0112] Weakly Supervised Loss Function: To model the relationship between high-frequency and low-frequency logging in well-side seismic traces, a weakly supervised loss function is used to constrain the low-frequency output to match the low-frequency logging for well-side locations. Due to the location characteristics of well-side traces within the seismic profile, a masking method is used to implement the loss constraint. The extracted well-side profile x... in The seismic sampling marker at the location of the seismic trace near the well is set to 1, and the sampling marker at other non-well-side trace locations is set to 0. This yields a mask tensor mask with the same dimensions (B, 1, H, W).

[0113] Low-pass filter function Lf l (), the low-pass cutoff frequency is fl (the low-pass cutoff frequency fl is greater than the high-pass cutoff frequency f). h For ease of calculation, f is generally set.l =f h +5Hz). Furthermore, the pre-compensated low-frequency component, Lf, is calculated from the logging reflection coefficient r. l (r), combined with the mask, yields the weakly supervised loss value, and the weakly supervised loss function is:

[0114]

[0115] In this embodiment, k is a matching coefficient used to match the energy consistency between the low-frequency components of the well logging data and the low-frequency components of the network output. It can be seen that the weakly supervised loss function is used to learn the energy for low-frequency compensation from the low-frequency well logging data.

[0116] To further constrain the results of low-frequency earthquake compensation and avoid anomalies in network predictions, structural constraint loss is applied to the prediction results, taking into account the incorporation of post-stack earthquake reflection structures into the model training.

[0117] Based on the seismic structure tensor method, and using post-stack seismic data samples x in Calculate the reflection eigenvector u = [u1, u2] T u1 is the horizontal component of the eigenvector, and u2 is the vertical component of the eigenvector. Then, the model output is used to calculate the predicted low-frequency compensated seismic data x. out Calculate the local gradient vector g = [g1, g2] T The local gradient vector is the gradient vector at the sampling point, g1 is the horizontal component of the local gradient vector, and g2 is the vertical component of the local gradient vector. Therefore, based on the geometric consistency of the reflection between the input and output, the structural constraint loss function can be obtained as follows (in practical applications, vectors u and g are normalized):

[0118]

[0119] The overall model training process is a multi-task training process: Task 1 is a self-supervised loss, which can be achieved by randomly sampling seismic profile data during training; Task 2 requires sampling well-crossing profile data or arbitrary lines passing through wells as model input, and calculating low-frequency logging data based on the logging reflection coefficients of the wells traversed by the well-crossing profile data. Task 3 involves structural constraint loss, which can be achieved by randomly sampling profile data and calculating local reflection features and structural gradients. The overall multi-task training loss is:

[0120]

[0121] In this embodiment, β represents the weight of the structural constraints. Through technical practice, this weight is generally selected between 0.1 and 0.01, depending on the noise level of the seismic data itself. When the signal-to-noise ratio of the input post-stack seismic data is low, a higher β weight is generally selected.

[0122] For the low-frequency compensation process, simply input the target post-stack seismic data to be compensated into the target low-frequency compensation model to obtain the low-frequency compensated result. As can be seen from the method's principle, this technique can be easily extended to three-dimensional convolutional network models, with the input and output set accordingly to three-dimensional post-stack seismic data.

[0123] The technical solution provided in this invention, compared with commonly used deep learning-based well-seismic joint learning models, does not require the preparation of a large number of samples in advance, while the network has high generalization ability and stable output results. By integrating self-supervised, weakly supervised, and structurally constrained multi-objective training modes, low-frequency components from well logging are compensated into seismic records. The compensated seismic data has richer and more accurate low-frequency information, laying the foundation for improving inversion quality.

[0124] The technical solution of this embodiment involves acquiring target post-stack seismic data; inputting the target post-stack seismic data into a target low-frequency compensation model to obtain low-frequency compensated seismic data. The target low-frequency compensation model is obtained by iteratively training an initial low-frequency compensation model using a target sample set. The target sample set includes: post-stack seismic data samples and well logging curves corresponding to the post-stack seismic data samples. This can improve the reliability of the low-frequency compensated seismic data and provide a data foundation for improving the quality of post-stack inversion.

[0125] Example 2

[0126] Figure 2 This is a schematic diagram of a low-frequency compensation device provided in an embodiment of the present invention. This embodiment is applicable to low-frequency compensation applications. The device can be implemented using software and / or hardware methods and can be integrated into any device that provides low-frequency compensation functionality, such as… Figure 2 As shown, the low-frequency compensation device specifically includes: a target post-stack seismic data acquisition module 210 and a low-frequency compensated seismic data determination module 220.

[0127] Among them, the target post-stack seismic data acquisition module is used to acquire target post-stack seismic data;

[0128] The low-frequency compensated seismic data determination module is used to input the target post-stack seismic data into the target low-frequency compensation model to obtain the low-frequency compensated seismic data. The target low-frequency compensation model is obtained by iteratively training an initial low-frequency compensation model using a target sample set. The target sample set includes: post-stack seismic data samples and well logging curves corresponding to the post-stack seismic data samples.

[0129] The above-described products can perform the methods provided in any embodiment of the present invention, and have the corresponding functional modules and beneficial effects for performing the methods.

[0130] Example 3

[0131] Figure 3 A schematic diagram of an electronic device 10 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0132] like Figure 3 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 may also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0133] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0134] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as low-frequency compensation methods.

[0135] In some embodiments, the low-frequency compensation method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or mounted on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the low-frequency compensation method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the low-frequency compensation method by any other suitable means (e.g., by means of firmware).

[0136] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0137] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0138] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0139] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0140] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0141] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0142] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0143] This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the low-frequency compensation method according to any embodiment of the invention.

[0144] In implementing the computer program product, computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof. Programming languages ​​include object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0145] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A low-frequency compensation method, characterized in that, include: Acquire target post-stack seismic data; The target post-stack seismic data is input into the target low-frequency compensation model to obtain low-frequency compensated seismic data. The target low-frequency compensation model is obtained by iteratively training an initial low-frequency compensation model using a target sample set. The target sample set includes: post-stack seismic data samples and well logging curves corresponding to the post-stack seismic data samples.

2. The method according to claim 1, characterized in that, The initial low-frequency compensation model is trained iteratively using the target sample set, including: Establish an initial low-frequency compensation model; Input the post-stack seismic data samples from the target sample set into the initial low-frequency compensation model to obtain the predicted low-frequency compensated seismic data; The parameters of the initial low-frequency compensation model are trained based on the objective function formed by the predicted low-frequency compensated seismic data and the post-stack seismic data samples. The objective function includes at least one of the following: a self-supervised loss function, a weakly supervised loss function, and a structural constraint loss function. Return to the operation of inputting the post-stack seismic data samples from the target sample set into the initial low-frequency compensation model to obtain the predicted low-frequency compensated seismic data, until the target low-frequency compensation model is obtained.

3. The method according to claim 2, characterized in that, The parameters of the initial low-frequency compensation model are trained based on the objective function formed by the predicted low-frequency compensated seismic data and post-stack seismic data samples, including: Extract low-frequency seismic data of well locations from the predicted low-frequency compensated seismic data; The well logging reflection coefficient is determined based on the well logging curves corresponding to the post-stack seismic data samples. Low-frequency logging data is determined based on the logging reflection coefficient; Based on the post-stack seismic data samples, the reflection feature vector is determined; Obtain the local gradient vector of the predicted low-frequency compensated seismic data; A self-supervised loss function is formed based on the post-stack seismic data sample, the predicted low-frequency compensated seismic data, and the first formula. A weakly supervised loss function is formed based on the low-frequency logging data, the low-frequency seismic data of the well locations in the predicted low-frequency compensated seismic data, and the second formula. The structural constraint loss function is formed based on the reflection feature vector, the local gradient vector of the predicted low-frequency compensated seismic data, and the third formula. The parameters of the initial low-frequency compensation model are trained based on the objective function formed by the self-supervised loss function, the weakly supervised loss function, the structural constraint loss function, the weights corresponding to the self-supervised loss function, the weights corresponding to the weakly supervised loss function, and the weights corresponding to the structural constraint loss function.

4. The method according to claim 3, characterized in that, The first formula is: Among them, Loss self This is the self-monitored loss value. Let x be the high-pass filter function. in For post-stack seismic data samples, x out To predict seismic data after low-frequency compensation; The second formula is: Among them, Loss sup For weak supervision loss value, For low-frequency logging data, For low-frequency seismic data used to predict well locations in low-frequency compensated seismic data, k is the matching coefficient. is a low-pass filter function, r is the logging reflection coefficient, and mask is a mask matrix in which the logging position is 1 and the rest are 0; The third formula is: Among them, Loss constraint Let g be the structural constraint loss value, and g be the local gradient vector [g1, g2]. T Let g1 be the horizontal component of the local gradient vector, g2 be the vertical component of the local gradient vector, and the reflection eigenvector u = [u1, u2]. T u1 is the horizontal component of the feature vector, and u2 is the vertical component of the feature vector.

5. The method according to claim 3, characterized in that, Low-frequency logging data is determined based on the logging reflection coefficient, including: The well logging reflection coefficient is bandpass filtered to obtain the passband signal of the well logging reflection coefficient; Extract well-side seismic traces from the post-stack seismic data samples based on the well logging curves corresponding to the post-stack seismic data samples; Bandpass filtering is performed on the well-side seismic trace data to obtain the passband signal of the well-side seismic trace data; The matching coefficient is determined based on the passband signal of the well logging reflection coefficient, the passband signal of the well-side seismic data, and the fourth formula; The fourth formula is: Where scale is the compensation factor, rb an d represents the passband signal of the logging reflection coefficient, sb an d represents the passband signal of the seismic trace data near the well; Low-frequency logging data are determined based on the matching coefficient and the logging reflection coefficient.

6. The method according to claim 1, characterized in that, The target post-stack seismic data is input into the target low-frequency compensation model to obtain low-frequency compensated seismic data, including: Obtain the mean and standard deviation of the target post-stack seismic data; The target post-stack seismic data is normalized based on the mean and standard deviation of the target post-stack seismic data to obtain normalized target post-stack seismic data. The normalized target post-stack seismic data is input into the target low-frequency compensation model to obtain low-frequency compensated seismic data.

7. A low-frequency compensation device, characterized in that, include: The target post-stack seismic data acquisition module is used to acquire target post-stack seismic data; The low-frequency compensated seismic data determination module is used to input the target post-stack seismic data into the target low-frequency compensation model to obtain the low-frequency compensated seismic data. The target low-frequency compensation model is obtained by iteratively training an initial low-frequency compensation model using a target sample set. The target sample set includes: post-stack seismic data samples and well logging curves corresponding to the post-stack seismic data samples.

8. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the low-frequency compensation method according to any one of claims 1-6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the low-frequency compensation method according to any one of claims 1-6.

10. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the low-frequency compensation method according to any one of claims 1-6.

Images