Full waveform inversion method, device, electronic equipment and medium based on error function

Abstract

This application discloses a full-waveform inversion method, apparatus, electronic device, and medium based on an error function. The method may include: calculating the cross-correlation value between simulated records and observation records in the detection point domain; determining the simulated record corresponding to the detection point position when the cross-correlation value is maximum, and calculating the detection point position correction; determining the error function based on the simulated record and the detection point position correction; and iteratively inverting the derivative of the error function with respect to velocity as the gradient to obtain a velocity inversion model. By introducing a new data dimension, the present invention traverses other detection points near each detection point to find the detection point data that best fits the field observation record, ultimately providing a high-precision velocity model for migration imaging and seismic data interpretation.

Description

Technical Field

[0001] The present invention relates to the field of oil and gas seismic exploration, and more particularly to a full waveform inversion method, device, electronic equipment and medium based on an error function. Background Art

[0002] Full waveform inversion is data-driven and is a highly nonlinear problem. The lack of low-frequency information in the observation record will lead to frequency skipping, which seriously affects the quality of the inversion results. In addition, full waveform inversion is highly dependent on the initial velocity model. When the model accuracy is low and it is not near the global optimal solution, the inversion will fall into a local extreme value, affecting the accuracy of the inversion results.

[0003] Currently, many approaches are available to address this issue. A common method is the multiscale inversion approach proposed by Buncks (1995). This approach utilizes low-frequency information to reduce phase errors between data. This approach typically combines window and offset selection strategies with the selection of characteristic wavefields, such as latent waves, to invert the medium within a specific range, thereby reducing frequency skipping and reliance on the initial model. However, this approach is time-consuming and requires considerable human experience, making it unstable in practical applications.

[0004] Another possible solution to this problem is to expand the dimensionality of the data space. When the model is poor, multidimensional data can be introduced early in the inversion iteration. This approach, often based on model parameterization, is also known as migration velocity analysis. While this method can be equivalent to traveltime tomography in projection scenarios, the enormous computational effort limits its development and use in practical three-dimensional applications.

[0005] Therefore, it is necessary to develop a full waveform inversion method, device, electronic equipment and medium based on the extended detection point error function.

[0006] The information disclosed in the background technology section of the present invention is only intended to deepen the understanding of the general background technology of the present invention, and should not be regarded as an admission or any form of suggestion that the information constitutes the prior art already known to those skilled in the art. Summary of the Invention

[0007] The present invention proposes a full waveform inversion method, device, electronic device and medium based on an error function. By introducing a new data dimension, the method can traverse other detection points near each detection point to find the detection point data that best fits the field observation record, ultimately providing a high-precision velocity model for offset imaging and interpretation of seismic data.

[0008] In a first aspect, an embodiment of the present disclosure provides a full waveform inversion method based on an error function, comprising:

[0009] Calculate the cross-correlation value between the simulated record and the observed record in the detection point domain;

[0010] Determine the simulated record of the detection point position corresponding to the maximum cross-correlation value, and calculate the correction amount of the detection point position;

[0011] determining an error function based on the simulation record and the correction amount of the detection point position;

[0012] The velocity inversion model is obtained by iterative inversion using the derivative of the error function with respect to velocity as the gradient.

[0013] Preferably, the correction amount of the detection point position is calculated by formula (1):

[0014] Δx r =[Δx1,…,Δx N ]=[||x1-x i,1 ||,…,||x N -x i,N ||] (1)

[0015] Where Δx r is the correction value of the detection point position, x i,N The detection point position corresponding to the detection point number N that has the greatest similarity with the observation record.

[0016] Preferably, the error function is:

[0017]

[0018] Among them, u(x i ) is the simulated record of the detection point position corresponding to the maximum cross-correlation value, d(x j ) are field observation records.

[0019] Preferably, the gradient is:

[0020]

[0021] Among them, L * is the reverse continuation process, and R is the residual between the observation record and the simulation record at the detection point.

[0022] Preferably, the gradient is preprocessed by the Gauss-Newton method, and multiple iterations are performed based on the processed gradient to obtain the velocity inversion model.

[0023] As a specific implementation of the embodiment of the present disclosure,

[0024] In a second aspect, the present disclosure further provides a full waveform inversion device based on an error function, comprising:

[0025] Cross-correlation module, which calculates the cross-correlation value between the simulated record and the observed record in the detection point domain;

[0026] A calculation module determines the simulated record of the detection point position corresponding to the maximum cross-correlation value and calculates the correction amount of the detection point position;

[0027] an error function calculation module, which determines an error function based on the simulation record and the correction amount of the detection point position;

[0028] The iterative inversion module uses the derivative of the error function with respect to the velocity as a gradient and performs iterative inversion to obtain a velocity inversion model.

[0029] Preferably, the correction amount of the detection point position is calculated by formula (1):

[0030] Δx r =[Δx1,…,Δx N ]=[||x1-x i,1 ||,…,||x N -x i,N ||] (1)

[0031] Where Δx r is the correction value of the detection point position, x i,N The detection point position corresponding to the detection point number N that has the greatest similarity with the observation record.

[0032] Preferably, the error function is:

[0033]

[0034] Among them, u(x i ) is the simulated record of the detection point position corresponding to the maximum cross-correlation value, d(x j ) are field observation records.

[0035] Preferably, the gradient is:

[0036]

[0037] Among them, L * is the reverse continuation process, and R is the residual between the observation record and the simulation record at the detection point.

[0038] Preferably, the gradient is preprocessed by the Gauss-Newton method, and multiple iterations are performed based on the processed gradient to obtain the velocity inversion model.

[0039] In a third aspect, an embodiment of the present disclosure further provides an electronic device, the electronic device comprising:

[0040] a memory storing executable instructions;

[0041] A processor runs the executable instructions in the memory to implement the full waveform inversion method based on the error function.

[0042] In a fourth aspect, an embodiment of the present disclosure further provides a computer-readable storage medium, which stores a computer program, and when the computer program is executed by a processor, implements the full waveform inversion method based on the error function.

[0043] The present invention utilizes full-waveform inversion technology with an expanded detector error function to reduce inversion nonlinearity and reliance on the initial model. By introducing new data dimensions, the method traverses nearby detectors at each detector to find the detector data that best matches field observations. This addresses the problem of inversion being prone to local extrema and non-convergence when the initial velocity model is inaccurate. The ultimate goal is to develop a robust, adaptable, and highly accurate parameter modeling scheme, providing a high-precision velocity model for migration imaging and seismic data interpretation.

[0044] The methods and apparatus of the present invention have other features and advantages that will be apparent from or will be described in detail in the accompanying drawings and subsequent detailed descriptions incorporated herein, which together serve to explain the specific principles of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The above and other objects, features and advantages of the present invention will become more apparent through a more detailed description of exemplary embodiments of the present invention with reference to the accompanying drawings, wherein like reference numerals generally represent like components throughout the exemplary embodiments of the present invention.

[0046] Figure 1 A flowchart showing the steps of a full waveform inversion method based on an error function according to an embodiment of the present invention is shown.

[0047] Figure 2a and Figure 2b Schematic diagrams of an initial velocity model and a full waveform inversion velocity model based on an extended detection point error function according to an embodiment of the present invention are respectively shown.

[0048] Figure 3a and Figure 3b Schematic diagrams respectively show velocity migration results based on initial velocity migration and full waveform inversion based on an extended detection point error function according to an embodiment of the present invention.

[0049] Figure 4 A block diagram of a full waveform inversion device based on an error function according to an embodiment of the present invention is shown.

[0050] Description of reference numerals:

[0051] 201. Cross-correlation module; 202. Calculation module; 203. Error function calculation module; 204. Iterative inversion module. DETAILED DESCRIPTION

[0052] The preferred embodiments of the present invention will be described in more detail below. Although the preferred embodiments of the present invention are described below, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0053] The present invention provides a full waveform inversion method based on an error function, comprising:

[0054] Calculate the cross-correlation value between the simulated record and the observed record in the detection point domain;

[0055] Determine the simulated record of the detection point position corresponding to the maximum cross-correlation value, and calculate the correction amount of the detection point position;

[0056] Determine the error function based on the simulated records and the correction amount of the detection point position;

[0057] The velocity inversion model is obtained by iterative inversion with the derivative of the error function with respect to velocity as the gradient.

[0058] In one example, the correction amount of the detection point position is calculated by formula (1):

[0059] Δx r =[Δx1,…,Δx N ]=[||x1-x i,1 ||,…,||x N -x i,N ||] (1)

[0060] Where Δx r is the correction value of the detection point position, x i,N The detection point position corresponding to the detection point number N that has the greatest similarity with the observation record.

[0061] In one example, the error function is:

[0062]

[0063] Among them, u(x i ) is the simulated record of the detection point position corresponding to the maximum cross-correlation value, d(x j ) are field observation records.

[0064] In one example, the gradient is:

[0065]

[0066] Among them, L * is the reverse continuation process, and R is the residual between the observation record and the simulation record at the detection point.

[0067] In one example, the gradient is preprocessed using the Gauss-Newton method, and multiple iterative updates are performed based on the processed gradient to obtain a velocity inversion model.

[0068] Specifically, by introducing a new data dimension, traversing other detection points near each detection point, and finding the detection point data that best fits the field observation records, the problem that the inversion is prone to fall into local extreme values ​​and cannot converge when the initial velocity model accuracy is low is solved.

[0069] Assume that there are N detection points for a certain shot in the field observation data, and the location of each detection point is x r =[x1,…,x N ], traverse each detection point x j and other nearby detection points, there is a detection point position x i , so that the simulation record u(x i ) and observation record d(x j ) has the largest mutual correlation value, that is, it has the greatest similarity. At this time, formula (1) represents the correction amount of the detection point position between the two.

[0070] At present, the most widely used full waveform inversion technology is the conventional least squares error functional.

[0071]

[0072] Here u(x j ) is replaced by u(x i ), and at the same time, the detection point position correction is introduced into the error function as a constraint regularization term, then the original error function changes to formula (2).

[0073] The velocity inversion model is obtained through iterative inversion, using the derivative of the error function with respect to velocity as the gradient. Full waveform inversion updates the velocity model through multiple iterations, with the gradient being the direction of velocity update.

[0074] In order to improve the computational efficiency of full waveform inversion, the Gauss-Newton method is used to preprocess the gradient, and multiple iterative updates are performed based on the processed gradient to obtain the final velocity inversion model.

[0075] The present invention also provides a full waveform inversion device based on an error function, comprising:

[0076] Cross-correlation module, which calculates the cross-correlation value between the simulated record and the observed record in the detection point domain;

[0077] A calculation module determines the simulated record of the detection point position corresponding to the maximum cross-correlation value and calculates the correction amount of the detection point position;

[0078] The error function calculation module determines the error function based on the simulation record and the correction amount of the detection point position;

[0079] The iterative inversion module uses the derivative of the error function with respect to velocity as the gradient and performs iterative inversion to obtain the velocity inversion model.

[0080] In one example, the correction amount of the detection point position is calculated by formula (1):

[0081] Δx r =[Δx1,…,Δx N ]=[||x1-x i,1 ||,…,||x N -x i,N ||] (1)

[0082] Where Δx r is the correction value of the detection point position, x i,N The detection point position corresponding to the detection point number N that has the greatest similarity with the observation record.

[0083] In one example, the error function is:

[0084]

[0085] Among them, u(x i ) is the simulated record of the detection point position corresponding to the maximum cross-correlation value, d(x j ) are field observation records.

[0086] In one example, the gradient is:

[0087]

[0088] Among them, L * is the reverse continuation process, and R is the residual between the observation record and the simulation record at the detection point.

[0089] In one example, the gradient is preprocessed using the Gauss-Newton method, and multiple iterative updates are performed based on the processed gradient to obtain a velocity inversion model.

[0090] Specifically, by introducing a new data dimension, traversing other detection points near each detection point, and finding the detection point data that best fits the field observation records, the problem that the inversion is prone to fall into local extreme values ​​and cannot converge when the initial velocity model accuracy is low is solved.

[0091] Assume that there are N detection points for a certain shot in the field observation data, and the location of each detection point is x r =[x1,…,x N ], traverse each detection point x j and other nearby detection points, there is a detection point position x i , so that the simulation record u(x i ) and observation record d(x j ) has the largest mutual correlation value, that is, it has the greatest similarity. At this time, formula (1) represents the correction amount of the detection point position between the two.

[0092] At present, the most widely used full waveform inversion technology is the conventional least squares error functional.

[0093]

[0094] Here u(x j ) is replaced by u(x i ), and at the same time, the detection point position correction is introduced into the error function as a constraint regularization term, then the original error function changes to formula (2).

[0095] The velocity inversion model is obtained through iterative inversion, using the derivative of the error function with respect to velocity as the gradient. Full waveform inversion updates the velocity model through multiple iterations, with the gradient being the direction of velocity update.

[0096] In order to improve the computational efficiency of full waveform inversion, the Gauss-Newton method is used to preprocess the gradient, and multiple iterative updates are performed based on the processed gradient to obtain the final velocity inversion model.

[0097] The present invention also provides an electronic device, which includes: a memory storing executable instructions; and a processor running the executable instructions in the memory to implement the above-mentioned full waveform inversion method based on the error function.

[0098] The present invention also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the full waveform inversion method based on the error function is implemented.

[0099] To facilitate understanding of the solutions and effects of the embodiments of the present invention, four specific application examples are given below. Those skilled in the art should understand that these examples are only for facilitating understanding of the present invention, and any specific details thereof are not intended to limit the present invention in any way.

[0100] Example 1

[0101] Figure 1A flowchart showing the steps of a full waveform inversion method based on an error function according to an embodiment of the present invention is shown.

[0102] like Figure 1 As shown, the full waveform inversion method based on the error function includes: step 101, calculating the cross-correlation value between the simulated record and the observed record in the detection point domain; step 102, determining the simulated record of the detection point position corresponding to the maximum cross-correlation value, and calculating the detection point position correction amount; step 103, determining the error function based on the simulated record and the detection point position correction amount; step 104, using the derivative of the error function with respect to the velocity as the gradient, iterative inversion to obtain the velocity inversion model.

[0103] Figure 2a and Figure 2b Schematic diagrams of an initial velocity model and a full waveform inversion velocity model based on an extended detection point error function according to an embodiment of the present invention are respectively shown.

[0104] Figure 3a and Figure 3b Schematic diagrams respectively show velocity migration results based on initial velocity migration and full waveform inversion based on an extended detection point error function according to an embodiment of the present invention.

[0105] Full-waveform inversion (FWI) based on an extended geophone error function (EDEF) was tested at the Sinopec Tahe worksite, using a ray tomography model as the initial parameter model. Comparisons between the initial model and the inversion results, as well as migrations based on these two models to verify velocity model accuracy, demonstrate that the FWI velocity migration results using the EDEEF achieve excellent separation of beads, resulting in more focused energy and facilitating subsequent reservoir interpretation.

[0106] Example 2

[0107] Figure 4 A block diagram of a full waveform inversion device based on an error function according to an embodiment of the present invention is shown.

[0108] like Figure 4 As shown, the full waveform inversion device based on the error function includes:

[0109] The cross-correlation module 201 calculates the cross-correlation value between the simulated record and the observed record in the detection point domain;

[0110] The calculation module 202 determines the simulated record of the detection point position corresponding to the maximum cross-correlation value and calculates the correction amount of the detection point position;

[0111] The error function calculation module 203 determines the error function based on the simulation record and the correction amount of the detection point position;

[0112] The iterative inversion module 204 uses the derivative of the error function with respect to the velocity as the gradient to obtain a velocity inversion model through iterative inversion.

[0113] As an alternative, the correction value of the detection point position can be calculated using formula (1):

[0114] Δx r =[Δx1,…,Δx N ]=[||x1-x i,1 ||,…,||x N -x i,N ||] (1)

[0115] Where Δx r is the correction value of the detection point position, x i,N The detection point position corresponding to the detection point number N that has the greatest similarity with the observation record.

[0116] As an alternative, the error function is:

[0117]

[0118] Among them, u(x i ) is the simulated record of the detection point position corresponding to the maximum cross-correlation value, d(x j ) are field observation records.

[0119] Alternatively, the gradient is:

[0120]

[0121] Among them, L * is the reverse continuation process, and R is the residual between the observation record and the simulation record at the detection point.

[0122] As an optional solution, the gradient is preprocessed using the Gauss-Newton method, and multiple iterative updates are performed based on the processed gradient to obtain the velocity inversion model.

[0123] Example 3

[0124] The present 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-mentioned full waveform inversion method based on the error function.

[0125] An electronic device according to an embodiment of the present disclosure includes a memory and a processor.

[0126] The 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 include, for example, random access memory (RAM) and / or cache memory. The non-volatile memory may include, for example, read-only memory (ROM), a hard disk, flash memory, etc.

[0127] The processor may be a central processing unit (CPU) or other form of processing unit having data processing capability and / or instruction execution capability, and may control other components in the electronic device to perform desired functions. In one embodiment of the present disclosure, the processor is used to execute the computer-readable instructions stored in the memory.

[0128] Those skilled in the art should understand that in order to solve the technical problem of how to obtain a good user experience, this embodiment may also include well-known structures such as a communication bus and an interface, and these well-known structures should also be included in the scope of protection of this disclosure.

[0129] For detailed description of this embodiment, please refer to the corresponding description in the aforementioned embodiments, which will not be repeated here.

[0130] Example 4

[0131] An embodiment of the present disclosure provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the full waveform inversion method based on the error function is implemented.

[0132] According to an embodiment of the present disclosure, a computer-readable storage medium stores non-transitory computer-readable instructions, which, when executed by a processor, execute all or part of the steps of the aforementioned methods of the embodiments of the present disclosure.

[0133] The above-mentioned computer-readable storage media include, but are not limited to, optical storage media (e.g., CD-ROMs and DVDs), magneto-optical storage media (e.g., MOs), magnetic storage media (e.g., magnetic tapes or mobile hard disks), media with built-in rewritable non-volatile memory (e.g., memory cards), and media with built-in ROM (e.g., ROM cartridges).

[0134] Those skilled in the art should understand that the above description of the embodiments of the present invention is only for the purpose of illustrative purposes only 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 given examples.

[0135] While various embodiments of the present invention have been described above, the above description is intended to be illustrative, not exhaustive, and not 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 full waveform inversion method based on an error function, characterized in that: include: Calculate the cross-correlation value between the simulated record and the observed record in the detection point domain; Determine the simulated record of the detection point position corresponding to the maximum cross-correlation value, and calculate the correction amount of the detection point position; determining an error function based on the simulation record and the correction amount of the detection point position; Taking the derivative of the error function with respect to velocity as the gradient, performing iterative inversion to obtain a velocity inversion model; The correction amount of the detection point position is calculated by formula (1): （1） in, is the correction value of the detection point position, The detection point position corresponding to the detection point number N that has the greatest similarity with the observation record; Wherein, the error function is: （2） in, is the simulated record of the detection point position corresponding to the maximum cross-correlation value, d(x j ) are field observation records.

2. The full waveform inversion method based on the error function according to claim 1, wherein: The gradient is: （3） in, For the reverse extension process, is the residual between the observed and simulated records at the detection point.

3. The full waveform inversion method based on error function according to claim 1, wherein: The gradient is preprocessed by the Gauss-Newton method, and multiple iterations are performed based on the processed gradient to obtain the velocity inversion model.

4. A full waveform inversion device based on an error function, characterized in that: include: Cross-correlation module, which calculates the cross-correlation value between the simulated record and the observed record in the detection point domain; A calculation module determines the simulated record of the detection point position corresponding to the maximum cross-correlation value and calculates the correction amount of the detection point position; an error function calculation module, which determines an error function based on the simulation record and the correction amount of the detection point position; An iterative inversion module, which uses the derivative of the error function with respect to velocity as a gradient and performs iterative inversion to obtain a velocity inversion model; The correction amount of the detection point position is calculated by formula (1): （1） in, is the correction value of the detection point position, The detection point position corresponding to the detection point number N that has the greatest similarity with the observation record; Wherein, the error function is: （2） in, is the simulated record of the detection point position corresponding to the maximum cross-correlation value, d(x j ) are field observation records.

5. An electronic device, characterized in that: The electronic device comprises: a memory storing executable instructions; A processor, wherein the processor runs the executable instructions in the memory to implement the full waveform inversion method based on the error function according to any one of claims 1 to 3.

6. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the full waveform inversion method based on the error function according to any one of claims 1 to 3 is implemented.

Images