Calculation Method, Device, Equipment and Medium for Layer Velocity Data Volume

Abstract

The present application discloses a method, apparatus, device and storage medium for calculating a layer velocity data volume. The method includes: after dividing the gather data by the central angle, calculating a central angle seismic attribute data volume related to the gather data; determining different layer velocity models within a sliding time window; and establishing a central angle attribute model training set corresponding to the different layer velocity models within the sliding window according to the central angle seismic attribute data volume; training an approximate support vector machine by using the central angle attribute model training sets corresponding to the different layer velocity models; classifying different layer velocity models for an attribute training set of seismic traces within the sliding time window by using the approximate support vector machine; and reconstructing the layer velocity data for the seismic traces within the sliding time window according to the classification result to obtain the layer velocity data volume.

Description

Technical Field

[0001] This application relates to the technical field of oil and gas field geophysical exploration, specifically to the technical fields of data mining and machine learning, and provides a method, device, equipment and medium for calculating a layer velocity data volume. Background Technique

[0002] The velocity problem is a very complex problem in seismic exploration, which involves various links such as seismic data processing, interpretation and inversion. At the same time, velocity is the link directly connecting seismic data and well data, and is also a very important parameter throughout the whole process of seismic exploration processing and interpretation. Whether the velocity is correctly selected directly affects the seismic data processing imaging, and has an important role in studying low-amplitude structures, seismic hydrocarbon anomalies, fine reservoir inversion, and time-depth conversion. It directly determines the accuracy of structural interpretation and mapping, and has an important role in the interpretation of abnormal bodies, the mapping of structures, the establishment of initial models for seismic inversion and processing, and is worthy of in-depth study.

[0003] In the current oil and gas exploration in the Sichuan Basin, obtaining an accurate velocity field is quite important for discovering lithologic oil and gas traps. For example, in the Shuangyushi exploration area in western Sichuan, the surface undulates violently and the underground structure is relatively complex. In addition, the target layer (Qixia Formation) is a platform margin shoal facies deposit, and the target layer is buried relatively deep, about 7000 - 8000m, with high requirements for seismic data imaging. In the oil and gas exploration in this block, prestack depth migration technology must be used for processing to obtain accurate lithologic traps or structural forms, so as to lay a foundation for the oil and gas exploration in this area. Therefore, in the seismic data processing of the Shuangyushi exploration area, the calculation of the relevant formation layer velocity is crucial, which is related to the success or failure of prestack depth migration processing, and thus directly affects the oil and gas exploration prospects in this area.

[0004] Exemplarily, conventional velocity conversion uses well or seismic velocity spectrum data for conversion, which often results in a large error in the obtained layer velocity data volume. Conventional velocity model accuracy is relatively high after using well data to control the layer velocity at well points, but the error is large far from the well. Subsequently, the error of using only velocity spectrum conversion is large, and it is only well corrected near the well points. Due to inaccurate layer velocity at positions far from the well points, it is easy to cause structural deformation. In addition, in a horizontal layered model, the stacking velocity is approximately equal to the root mean square velocity, and the layer velocity can be obtained through the Dix formula; when the underground structure is complex, there are lithology changes, and when there are lateral velocity changes caused by abnormal bodies, it is incorrect to obtain the layer velocity according to the Dix formula.

[0005] Therefore, how to obtain an accurate layer velocity data volume through relevant calculations, lay a foundation for subsequent relevant seismic prestack depth migration imaging processing, structure mapping, inversion, etc., so as to identify the accurate position and shape of favorable lithologic bodies, is a problem that geophysicists have been researching. Summary of the Invention

[0006] This application provides a method, apparatus, device, and storage medium for calculating a layer velocity data volume. The technical solution is as follows:

[0007] According to one aspect of the present application, a method for calculating a layer velocity data volume is provided. The method includes:

[0008] After dividing the gather data by central angle, calculate the central angle seismic attribute data volume related to the gather data;

[0009] Determine different layer velocity models within a sliding time window; and establish a training set of central angle attribute models corresponding to the different layer velocity models within the sliding window according to the central angle seismic attribute data volume;

[0010] Train an approximate support vector machine using the training set of central angle attribute models corresponding to the different layer velocity models;

[0011] Use the approximate support vector machine to classify different layer velocity models for the attribute training set of seismic traces within the sliding time window; and reconstruct the layer velocity data for the seismic traces within the sliding time window according to the classification results to obtain the layer velocity data volume.

[0012] In an alternative design of the present application, after dividing the gather data by central angle and calculating the central angle seismic attributes related to the gather data, it includes:

[0013] Divide the gather data according to the designed central angle to obtain gather data for each central angle;

[0014] Perform stacking processing and migration processing on the gather data for each central angle to obtain a post-stack data volume for each central angle;

[0015] Calculate seismic attributes for the post-stack data volume for each central angle to obtain a seismic attribute data volume for each central angle;

[0016] Reconstruct the seismic attribute data volume for each central angle according to common depth point (CDP) to obtain the central angle attribute data volume related to the gather data.

[0017] In an alternative design of the present application, determining different layer velocity models within a sliding time window includes:

[0018] Divide the measured layer velocity model in the well in the study area according to the target division factor to obtain different layer velocity models within the sliding time window;

[0019] Among them, the target division factors include at least one of geological data, logging data, oil and gas test data, lithologic combination, and layer velocity value range.

[0020] In an alternative design of the present application, the method further includes:

[0021] In the case where the number of layer velocity models in the study area is less than the target number, adding the layer velocity models in the virtual wells; and / or, in the case where the number of layer velocity models in the study area is less than the target number, adding the layer velocity models measured in the wells in the adjacent area of the study area.

[0022] In an alternative design of the present application, the establishment of the central angle attribute model training set corresponding to the different layer velocity models in the sliding window according to the central angle seismic attribute data volume includes:

[0023] Determining each central angle seismic attribute curve within the sliding time window according to the central angle seismic attribute data volume;

[0024] Determining the central angle attribute model training set corresponding to the different layer velocity models according to the different layer velocity models and their corresponding central angle seismic attribute curves within the sliding time window.

[0025] In an alternative design of the present application, the determination of the central angle attribute model training set corresponding to the different layer velocity models according to the different layer velocity models and their corresponding central angle seismic attribute curves within the sliding time window includes:

[0026] Extracting morphological feature parameters from the central angle seismic attribute curves according to the different layer velocity models and their corresponding central angle seismic attribute curves within the sliding time window to form the central angle attribute training model set corresponding to the different layer velocity models;

[0027] Among them, the morphological feature parameters include at least one of the monotonicity of the curve, concavity and convexity, the number and average value of extreme points, and the number of inflection points.

[0028] In an alternative design of the present application, the reconstruction of the layer velocity data for the seismic trace data within the sliding time window according to the classification result to obtain the layer velocity data volume includes:

[0029] Classifying the attribute training set established for each sampling point of the seismic trace data on each central angle seismic attribute curve within the sliding time window through the approximate support vector machine to determine the layer velocity model corresponding to each sampling point;

[0030] After obtaining the classification results of the sampling points belonging to the different layer velocity models, reconstructive assignment is performed on the layer velocity data at the sampling points within the sliding time window to obtain the layer velocity data volume.

[0031] According to one aspect of the present application, there is provided a computing device for a layer velocity data volume, the device comprising:

[0032] A calculation module, configured to calculate a central angle seismic attribute data volume related to the gather data after dividing the gather data by a central angle;

[0033] A building module, configured to determine different layer velocity models within a sliding time window; and establish a central angle attribute model training set corresponding to the different layer velocity models within the sliding window according to the central angle seismic attribute data volume;

[0034] A training module, configured to train an approximate support vector machine by using the central angle attribute model training set corresponding to the different layer velocity models; [[ID=*14]]

[0035] A classification module, configured to classify different layer velocity models for an attribute training set of seismic trace data within the sliding time window by using the approximate support vector machine;

[0036] A reconstruction module, configured to reconstruct layer velocity data for the seismic trace data within the sliding time window according to the classification result to obtain the layer velocity data volume.

[0037] According to one aspect of the present application, there is provided a computer device, characterized in that the computer device comprises: a processor and a memory, the memory stores a computer program, and the computer program is loaded and executed by the processor to implement the method for calculating a layer velocity data volume as described above.

[0038] According to one aspect of the present application, there is provided a computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program, and the computer program is loaded and executed by a processor to implement the method for calculating a layer velocity data volume as described above.

[0039] The beneficial effects brought by the technical solution provided by the embodiments of the present application at least include:

[0040] By introducing a layer velocity calculation method based on an approximate support vector machine, the calculation method provided by the present application is more accurate than the method provided by the traditional one. Generally, compared with the traditional layer velocity calculation method, an accurate layer velocity data volume can be obtained for subsequent prestack depth migration calculation. Description of the Drawings

[0041] To more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the accompanying drawings required for the description of the embodiments. Obviously, the accompanying drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0042] Figure 1 The flowchart of the method for calculating the layer velocity data volume provided by another exemplary embodiment of the present application is shown;

[0043] Figure 2 The flowchart of the method for calculating the layer velocity data volume provided by another exemplary embodiment of the present application is shown;

[0044] Figure 3 The block diagram of the device for calculating the layer velocity data volume provided by an exemplary embodiment of the present application is shown. Detailed implementation manners

[0045] To make the objectives, technical solutions and advantages of the present application clearer, the following further describes the embodiments of the present application in detail with reference to the accompanying drawings.

[0046] An embodiment of the present application provides a method for calculating a layer velocity data volume based on an approximate support vector machine, which can accurately calculate the layer velocity data volume and provide a relatively accurate time-domain layer velocity data volume for subsequent prestack depth migration.

[0047] The layer velocity calculation technology provided by the present application includes: dividing the gather data by central angle and calculating the relevant central angle seismic attribute data volume, determining the sliding time window for layer velocity calculation and establishing the central angle attribute training set within the sliding time window; establishing the layer velocity model within the relevant sliding time window and determining the central angle attribute model training sets corresponding to different layer velocity models within the relevant sliding time window; using the central angle attribute model training sets of different layer velocity models and the attribute training set within the relevant seismic trace sliding time window, classifying the attribute training set based on the approximate support vector machine for different layer velocity models, and reconstructing and assigning the layer velocity curve to the sampling points within the data trace sliding time window, and then calculating accordingly to obtain a layer velocity data volume, thereby realizing the calculation of the time-domain layer velocity data volume.

[0048] Figure 1 The flowchart of the method for calculating the layer velocity data volume provided by an exemplary embodiment of the present application is shown. This embodiment is illustrated by taking the application of this method to a computer device as an example The method includes:

[0049] Step 102: After dividing the gather data by central angle, calculate the central angle seismic attribute data volume related to the gather data;

[0050] Step 104: Determine the velocity models of different layers within the sliding time window; and establish a training set of central angle attribute models corresponding to the velocity models of different layers within the sliding window based on the central angle seismic attribute data volume;

[0051] In actual layer velocity calculation, the number of sampling points of the seismic attribute curve within the designed sliding time window affects the calculation accuracy of the layer velocity. Therefore, the size parameter of the relevant sliding time window should be determined according to actual conditions, layer velocity calculation accuracy, test conditions, expert experience, etc. Exemplarily, the size of the sliding time window can be taken as 30 ms.

[0052] Step 106: Train an approximate support vector machine using the training set of central angle attribute models corresponding to the velocity models of different layers;

[0053] Step 108: Use the approximate support vector machine to classify the attribute training set of the seismic trace data within the sliding time window according to the velocity models of different layers;

[0054] Step 110: Reconstruct the layer velocity data for the seismic trace data within the sliding time window according to the classification results to obtain a layer velocity data volume.

[0055] In summary, the method provided in this embodiment, by introducing a layer velocity calculation method based on an approximate support vector machine, the calculation method provided in this application is more accurate than the traditional method in predicting the layer velocity data volume. Generally, compared with the traditional layer velocity calculation method, an accurate layer velocity data volume can be obtained for subsequent prestack depth migration calculation.

[0056] Figure 2 The flowchart of the calculation method of the layer velocity data volume provided by another exemplary embodiment of the present application is shown. This embodiment is illustrated by taking the application of this method to a computer device as an example. The method includes:

[0057] For step 102, after dividing the gather data by the central angle, calculate the central angle seismic attribute data volume related to the gather data, which may include the following steps:

[0058] Step 201: Divide the gather data according to the designed central angle to obtain each central angle gather data;

[0059] After dividing the gather data by the central angle and calculating its related seismic attribute data volume. Among them, the gather data refers to the gather data after conventional field static correction, prestack denoising, amplitude compensation and deconvolution, residual static correction processing and dynamic correction.

[0060] This step mainly designs the gather data with respect to the central angle, and sets a certain range of azimuth angles and incident angles centered on the central angle. Reconstruct the data within the set azimuth angle and incident angle ranges for each central angle to obtain the gather data for each central angle.

[0061] Step 202: Perform stacking processing and migration processing on the gather data for each central angle to obtain the post-stack data volume for each central angle;

[0062] Perform stacking and migration processing on the gather data for each central angle to obtain the post-stack data volume for each central angle. Both stacking and migration processing are conventional seismic processing techniques and will not be elaborated in this application. In principle, the more the number of designed central angles, the higher the calculation accuracy of the interval velocity; the fewer the number of designed central angles, the lower the calculation accuracy. The design of the central angles can be such that the increment between the central angles is equidistant or non-equidistant. Exemplarily, it is designed to be equidistant according to the number of central angles between 0° and 180°. Exemplarily, based on the symmetry principle, the central angle and azimuth angle are set to the data between 0° and 180°. Exemplarily, the number of designed central angles should be greater than or equal to three.

[0063] Exemplarily, extract the gather data within a certain azimuth angle range and incident angle range centered on the central angle. It mainly refers to extracting the gather data within the azimuth angle range of plus or minus j° centered on the designed central angle (with 0° being due north and the included angle between the clockwise direction and the due north direction) and with the incident angle less than 30°. Exemplarily, the data value of j is set to be less than or equal to 5. Exemplarily, for the set incident angle range, in principle, the maximum incident angle cannot be greater than 30°. In actual operation, the azimuth angle and incident angle ranges can be determined according to the actual situation of the seismic data and expert experience, etc. Additionally, the incident angle range and its magnitude value can be converted to the relevant offset range according to the relevant velocity data and the two-way travel time of the target layer. By obtaining the relevant offset range and azimuth angle range, the gather data is reconstructed.

[0064] In the azimuth angle design, in the embodiments of this application, the observation system direction is set with 0° being due north and rotating clockwise for 360°. Based on the symmetry principle, the 360° azimuth of the seismic data shot point - geophone point collected in the field is converted into a 180° azimuth angle. For a certain azimuth angle range, calculate its central angle, and the central angle represents the divided azimuth angle range. The calculation formula is as follows:

[0065]

[0066] In formula (1), θ i , , , ,

[0066] ,

[0065] , , is the central angle of the i-th stacked data designed, is the minimum azimuth angle of the i-th stacked data designed, The maximum azimuth angle of the i-th stacked data in the design, where i≥2.

[0067] In the embodiments of the present application, the principle for defining the incident angle range is to take a certain offset range. The magnitude of the incident angle should be based on the principle of AV°. For the target layer, the incident angle of the maximum offset should not be greater than 30°. The relevant incident angle calculation formula is as follows:

[0068] θ = arctgD / 2h (2)

[0069] In formula (2), D is the offset of the designed seismic data, θ is the designed incident angle, and h is the buried depth of the predicted target layer. The buried depth of the target layer can be calculated from the two-way reflection time of the target layer and the average layer velocity of the strata above the target layer.

[0070] Step 203: Perform seismic attribute calculations on each post-stack data volume of the central angle to obtain the seismic attribute data volume for each central angle;

[0071] Using each post-stack data volume of the central angle, and adopting seismic attribute calculation methods and parameters, perform seismic attribute calculations to obtain the seismic attribute data volume for each central angle. Exemplarily, the seismic attributes refer to seismic attributes related to amplitude, frequency, etc., or can also be attribute data obtained from relevant inversion (such as relative wave impedance, etc.).

[0072] Step 204: Reconstruct the seismic attribute data volume for each central angle according to common depth point (CDP) to obtain the central angle attribute data volume related to the gather data;

[0073] Reconstruct the seismic attribute data volume for each central angle according to the relevant CDP trace number, so as to obtain a central angle attribute data volume.

[0074] Specifically, in the embodiments of the present application, only the division of the central angle is exemplified. It is also possible to divide the gather data according to different incident angle ranges, and calculate the seismic attribute data volume for different incident angles to enter the next step.

[0075] For the above step 104, establish different layer velocity models within the sliding time window; and according to the central angle seismic attribute data volume, establish a central angle attribute model training set corresponding to different layer velocity models within the sliding window, which may optionally include the following steps:

[0076] Step 205: Divide the measured layer velocity model in the well in the study area according to the target division factors to obtain different layer velocity models within the sliding time window;

[0077] Among them, the target division factors include at least one of geological data, logging data, oil and gas test data, lithologic combination, and layer velocity value range.

[0078] Schematically, based on geological data, logging data, test data, etc., different layer velocity models in each well are divided, thereby dividing different layer velocity models.

[0079] For example, according to the actual situation, the layer velocity models in the study area can be divided as detailed as possible according to relevant lithologic combinations, fracture development, gas content, etc. that affect the layer velocity and different lithologies; for another example, the range of values of the relevant layer velocity can also be determined according to the data value size (histogram analysis) of the layer velocity, and different layer velocity models can be divided according to the range of values.

[0080] In a possible design, when the number of layer velocity models in the study area is less than the target number, the layer velocity models in the virtual wells are added; and / or, when the number of layer velocity models in the study area is less than the target number, the layer velocity models measured in the wells in the adjacent area of the study area are added.

[0081] Exemplarily, if the number of wells in the study area is not enough to divide different layer velocity models, virtual wells can be set and added to the division and establishment of the layer velocity models. Among them, the determination of the well points of the virtual wells can be determined according to relevant layer velocity prediction results, expert experience, etc. In addition, the mature layer velocity model division and establishment scheme in the adjacent area can also be added, and its relevant attribute model training set is determined to enter the next step.

[0082] For another example, according to the layer velocity, lithologic combination, density data, etc. on the well, physical models of layer velocity with multiple sliding time window sizes (thicknesses) can be designed, and relevant forward calculations of attributes are performed on the physical models of layer velocity, so as to obtain an attribute model training set. In actual operation, according to the size of the sliding time window, the layer velocity model of the relevant sampling points within the sliding time window in the well is determined, and then the relevant central angle attribute model training set is determined. Exemplarily, the designed central angle attribute model training set of the layer velocity model should be able to meet the identification of the layer velocity model for the attribute training set within the sliding time window of each seismic trace. Among them, the calculation formula for the thickness of the physical model of layer velocity is as follows:

[0083]

[0084] In formula (3), h i is the thickness of the sliding time window of the i-th physical model of layer velocity, V i is the average layer velocity of this thickness in this physical model, and T is the designed length of the sliding time window.

[0085] Step 206: Determine each central angle seismic attribute curve within the sliding time window according to the central angle seismic attribute data volume;

[0086] Determine the sliding time window size for calculating the layer velocity on the seismic data trace, and use the central angle seismic attribute data volume to determine the seismic attribute curves within the sliding time window, so as to obtain each seismic attribute curve within the sliding time window.

[0087] Step 207: According to different layer velocity models and their corresponding central angle seismic attribute curves within the sliding time window, determine the central angle attribute model training sets corresponding to different layer velocity models;

[0088] According to different layer velocity models and the corresponding central angle seismic attribute curves of different layer velocity models within the sliding time window, extract morphological feature parameters from the central angle seismic attribute curves to form the central angle attribute training model sets corresponding to different layer velocity models; among them, the morphological feature parameters include at least one of: the monotonicity of the curve, concavity and convexity, the number of extreme points, the average value, and the number of inflection points.

[0089] Exemplarily, first, according to the calibration result of the well-seismic synthetic record, obtain the time-depth relationship in the relevant wells, and use the time-depth relationship to determine the central angle attribute data values corresponding to the relevant sampling points of different layer velocity models within the sliding time window. And according to the length of the sliding time window, obtain each central angle seismic attribute curve composed of each sampling point.

[0090] Extract morphological feature parameters from each central angle seismic attribute curve, so as to form the central angle attribute training model set. Among them, the morphological feature parameters include: the monotonicity of the curve, concavity and convexity, the number of extreme points, the average value, and the number of inflection points, etc. And so on, complete the establishment of each central angle attribute training model set of different layer velocity models.

[0091] As in the above method, the training set can be expressed as:

[0092] G = {(x1, y1),..., (x k , y k )} ∈ (x * y) k (4)

[0093] Among them, k is the number of samples; G is the attribute model training set; (x1, y1),..., (x k , y k ) are the attribute items of the training set. If there are 12 feature attributes, that is: x i ∈ x = R 12

[0094] Step 208: Train an approximate support vector machine using the central angle attribute model training sets corresponding to different layer velocity models;

[0095] For step 108, use an approximate support vector machine to classify the attribute training set of seismic trace data within a sliding time window for different layer velocity models. The optional steps include the following:

[0096] Step 209: Classify the attribute training set established on each central angle seismic attribute curve for each sampling point of the seismic trace data within the sliding time window through an approximate support vector machine, and determine the layer velocity model corresponding to each sampling point;

[0097] The specific operation is to train based on an approximate support vector machine using the central angle attribute model training set of the sliding time window sizes of different layer velocity models, classify the attribute training sets extracted from each central angle seismic attribute curve within the sliding time window of each CDP point, and then discriminate different layer velocity models for the sampling points within the sliding time window of each CDP point.

[0098] Among them, due to the introduction of an approximate support vector machine to classify each central angle seismic attribute curve within the sliding time window, the approximate support vector machine greatly reduces the computational amount during the learning process in a large data volume sample set, improves the training speed, and can perform fast learning without sacrificing the recognition accuracy. Therefore, the workload of recognizing each central angle attribute curve of different layer velocity models within the sliding time window is reduced.

[0099] For step 110, after reconstructing the layer velocity data within the sliding time window of the seismic trace data according to the classification results, a layer velocity data volume is obtained. The optional steps include the following:

[0100] Step 210: After obtaining the classification results of the sampling points belonging to different layer velocity models, perform reconstruction assignment on the layer velocity data at the sampling points within the sliding time window to obtain a layer velocity data volume.

[0101] In this step, through an approximate support vector machine, classify the attribute training set established on each central angle seismic attribute curve for each sampling point within the sliding time window, determine the layer velocity model of each sampling point within the relevant sliding time window, and after classifying the layer velocity parameter characteristics of the sampling points on different layer velocity models, perform reconstruction assignment on the layer velocity data at the sampling points within the sliding time window, thereby obtaining a reconstructed data volume regarding the layer velocity. And so on, complete the reconstruction of the layer velocity data values of each sampling point within the sliding time window of each CDP point to obtain a layer velocity data volume. Among them, the layer velocity parameter characteristics of the sampling points on the layer velocity model refer to the magnitude of the layer velocity data value within the sliding time window, and the layer velocity data values at different sampling points form a layer velocity curve. Use the layer velocity curve on this layer velocity model to assign the velocity curve of this layer velocity model to the sampling points within the relevant classified sliding time window, thereby obtaining the layer velocity curve within the sliding time window.

[0102] In actual operation, the above calculations can also be performed on multiple seismic attribute data of central angles of different types (n). If multiple types of seismic attribute data of central angles (n) are involved in training and discrimination, labels are set and they are sequentially positioned as 1, 2, 3,..., n in the manner of multi-class classification; the seismic attribute curves of central angles (n) of the sampling points to be discriminated also need to go through the above calculation steps to form samples to be discriminated, and finally the discriminant obtained after training with PSVM is used to classify the samples of the object to be discriminated.

[0103] In actual operation, grid points can also be designed for the study area, and the training set of central angle attribute data within the sliding time window on the relevant grid points is extracted. The central angle attribute models of different layer velocity models and the approximate support vector machine are used to assign layer velocities to the sampling points within the relevant sliding time window; after low-frequency filtering (filtering out high-frequency components) of the layer velocity curves on the grid points, a layer velocity curve that is relatively consistent with the root-mean-square calculated layer velocity curve obtained from velocity analysis on the grid points (basically corresponding at the standard layer) is formed; a new sliding time window is set, and the correlation coefficient is calculated for the two layer velocity curves, and a correlation coefficient threshold value is set; for the layer velocity curves within the sliding time window greater than or equal to the threshold value, no modification is made to the layer velocity data, while for the layer velocity data less than the threshold value, the layer velocity model within the relevant time window should be modified (or replaced) and the training set of central angle attribute data should be updated; the above relevant steps 2 and 3 are repeated until the threshold value is met and the loop ends, so as to obtain a relatively accurate layer velocity curve on the grid points; three-dimensional spatial interpolation calculation is performed using the layer velocity curve on the grid points to obtain an optimized layer velocity data volume. Among them, the grid points refer to the intersection points on the set rectangular grid, generally the intersection point positions of n lines × m channels. In principle, the grid parameters can be set according to the actual situation, and this grid is called the set grid. The grid parameters include grid spacing and the number of grids, etc. The size of the grid parameters can be determined according to the grid distribution and accuracy requirements of the velocity analysis points to be calculated. In principle, usually, the larger the set grid spacing, the relatively lower the calculation accuracy of the layer velocity, and some characteristic information will be lost; the smaller the grid spacing, the relatively higher the calculation accuracy of the layer velocity, and the drawn results will be more detailed. The size of the set grid should be determined according to the actual situation and the accuracy requirements of layer velocity calculation, and the set grid is usually regular. In addition, the sliding time window for correlation coefficient analysis set should be n times the sliding time window for layer velocity model identification, and the specific size or length of the relevant sliding time window can be determined according to the actual situation, expert experience, etc. Among them, the calculation formula for the correlation coefficient r of each sampling point within the sliding time window of the two layer velocity curves is:

[0104]

[0105] In formula (5), X i and Y iThe i-th data value of two types of data for performing related calculations and are respectively the rank-ordered average values of the two types of data values, and the value range of r is from 0 to 1.

[0106] In a specific example, work steps are formulated according to the technical process of the present application. The embodiment is to calculate the layer velocity data volume in the time domain for the Shuangyushi 3D work area in the northwestern part of the Sichuan Basin. In the oil and gas exploration in this area, it is found that prestack depth migration is quite important, and the layer velocity data volume in the time domain is in a relatively core position in the prestack depth migration processing. The accurate layer velocity data volume lays a foundation for the subsequent prestack depth migration processing.

[0107] Determine the central angle of the gather data, the azimuth angle range centered on it, and the incidence angle range in step ①. In actual operation, according to the actual situation of the seismic data in this area and expert experience, etc., the number of central angles is determined to be 18, which are 5°, 15°, 25°, 35°, 45°, 55°, 65°, 75°, 85°, 95°, 105°, 115°, 125°, 135°, 145°, 155°, 165°, 175°, etc. An equal-increment design is adopted. The azimuth angle range is set to be plus or minus 5° of the central angle, and the incidence angle of the main target layer (Qixia Formation) is set to be less than 30°. Then, the gather data is divided according to this azimuth angle and incidence angle range, and parameters are respectively stacked and migrated to obtain the post-stack data volume for each central angle.

[0108] Secondly, perform related seismic attribute calculations on the 18 post-stack data volumes for each central angle to obtain the related central angle attribute data volumes. In actual operation, mainly calculate the starting attenuation frequency (i.e., the frequency ATN-FRQ corresponding to the main frequency) attribute, and perform attribute data reconstruction processing on each central angle attribute data volume to obtain a central angle attribute data volume.

[0109] In step ②, according to the well-seismic calibration results and the division and establishment results of different layer velocity models in the study area and adjacent areas, determine the central angle attributes within the sliding time window (40 ms) corresponding to different layer velocity models, and use these data to establish a training set for the central angle attribute models of different layer velocity models. In actual operation, according to the starting attenuation frequency attributes corresponding to the sampling points of different layer velocity models (the length within the sliding time window), establish the central angle - starting attenuation frequency attribute curves within the time window of relevant sampling points; then, the characteristic values of the relevant 18 attribute curves can be extracted to establish the relevant training set for the central angle attribute models. By analogy, the training sets for the central angle attribute models of different layer velocity models are completed. In actual operation, 2821 layer velocity models are designed within the 40 ms time window. Therefore, 2821 training sets for the central angle attribute models are obtained.

[0110] In step ③, calculate the central angle attribute training set of each sampling point within the sliding time window of the seismic data volume, and classify the central angle attribute training set into different layer velocity models based on the approximate support vector machine. The specific operation is to train based on the approximate support vector machine using the central angle attribute model training set of different layer velocity models, classify the attribute training set of 18 central angle seismic attribute curves within the sliding time window of each CDP point, and then assign the layer velocity curves of the relevant layer velocity models to the sampling points within the sliding time window of the relevant CDP points. After this step, a layer velocity data volume is obtained.

[0111] In addition, the layer velocity data volume can be further optimized to obtain an optimized layer velocity data volume. The specific operation is to design grid points, extract the calculated layer velocity curves at the grid points, perform low-frequency filtering processing, and then calculate the correlation coefficient within the designed sliding time window (window length 200 ms) with the layer velocity curve obtained from velocity analysis, and set the correlation coefficient threshold value (0.65). Use the threshold value to optimize the relevant layer velocity data. When the threshold value is less than 0.65, redesign the relevant layer velocity model within the time window (or replace the specified layer velocity model) and update the central angle attribute data set, repeat the relevant steps of calculation until the relevant conditions are met, so as to obtain an optimized layer velocity curve data at the grid points; perform three-dimensional spatial interpolation on the layer velocity curves at the grid to obtain an optimized layer velocity data volume. Using the prediction results of this application, through comparative analysis with the layer velocities of subsequent drilling data in the study area, the coincidence rate is higher than 80.2%.

[0112] By comparing the results obtained by using the technical method of this application with the results obtained by the conventional layer velocity calculation technology, it can be found that the results of this application technology are relatively more accurate and superior to the technical results of the conventional layer velocity calculation. From the comparison of relevant results, it can be proved that the technology of this application is effective and can calculate an accurate time-domain layer velocity data volume. From the analysis of the subsequent drilling data in the Shuangyushi exploration area, when comparing the results of prestack depth migration processing using the results of this application with the actual drilling results, the error between the predicted formation interface and the measured interface is small, and the deep structural form is accurate, meeting the relevant geological requirements.

[0113] The above technical solution is only one implementation manner of this application. For those skilled in the art, based on the disclosed application methods and principles, it is very easy to make various types of improvements or deformations, not limited to the methods described in the above specific implementation manner of this application. Therefore, the above-described manner is only preferred and does not have a restrictive meaning.

[0114] Figure 3 The block diagram of a calculation device for a layer velocity data volume provided by an exemplary embodiment of this application is shown. The device includes:

[0115] A calculation module 310, configured to calculate a central angle seismic attribute data volume related to the gather data after dividing the gather data by a central angle.

[0116] An establishment module 320, configured to determine different layer velocity models within a sliding time window; and establish a central angle attribute model training set corresponding to the different layer velocity models within the sliding window according to the central angle seismic attribute data volume.

[0117] A training module 330, configured to train an approximate support vector machine by using the central angle attribute model training set corresponding to the different layer velocity models.

[0118] A classification module 340, configured to classify different layer velocity models for an attribute training set of seismic trace data within the sliding time window by using the approximate support vector machine.

[0119] A reconstruction module 350, configured to reconstruct layer velocity data for the seismic trace data within the sliding time window according to a classification result to obtain a layer velocity data volume.

[0120] In an optional design of the present application, the calculation module 310 is configured to divide the gather data according to a designed central angle to obtain gather data of each central angle; perform stacking processing and migration processing on the gather data of each central angle to obtain a post-stack data volume of each central angle; perform seismic attribute calculation on the post-stack data volume of each central angle to obtain a seismic attribute data volume of each central angle; and perform data reconstruction on the seismic attribute data volume of each central angle according to common depth point (CDP) to obtain a central angle attribute data volume related to the gather data.

[0121] In an optional design of the present application, the establishment module 320 is configured to divide a layer velocity model measured in a well in a study area according to a target division factor to obtain different layer velocity models within the sliding time window.

[0122] Wherein, the target division factor includes at least one of geological data, logging data, oil and gas test data, lithologic combination, and layer velocity value range.

[0123] In an optional design of the present application, the establishment module 320 is further configured to, when the number of layer velocity models in the study area is less than a target number, add layer velocity models in virtual wells; and / or

[0124] When the number of layer velocity models in the study area is less than the target number, add layer velocity models measured in wells in an adjacent area of the study area.

[0125] In an alternative design of the present application, the establishing module 320 is further configured to determine each central angle seismic attribute curve within the sliding time window according to the central angle seismic attribute data volume; and determine a central angle attribute model training set corresponding to the different layer velocity models according to the different layer velocity models and their corresponding central angle seismic attribute curves within the sliding time window.

[0126] In an alternative design of the present application, the establishing module 320 is further configured to extract morphological feature parameters from the central angle seismic attribute curves according to the different layer velocity models and their corresponding central angle seismic attribute curves within the sliding time window, and form a central angle attribute training model set corresponding to the different layer velocity models;

[0127] Wherein, the morphological feature parameters include at least one of: monotonicity of the curve, concavity and convexity, number and average value of extreme points, and number of inflection points.

[0128] In an alternative design of the present application, the classification module 340 is configured to classify the attribute training set established on each central angle seismic attribute curve for each sampling point of the seismic trace data within the sliding time window through the approximate support vector machine, and determine the layer velocity model corresponding to each sampling point; the reconstruction module 350 is configured to, after obtaining the classification results of the sampling points belonging to different layer velocity models, perform reconstruction assignment on the layer velocity data at the sampling points within the sliding time window to obtain the layer velocity data volume.

[0129] The present application also provides a computer device, which includes a memory and a processor. At least one instruction, at least one program, a code set or an instruction set is stored in the memory, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement the calculation method of the layer velocity data volume provided in the above method embodiments.

[0130] The present application also provides a computer-readable storage medium, in which at least one instruction, at least one program, a code set or an instruction set is stored, and the at least one instruction, the at least one program, the code set or the instruction set is loaded and executed by the processor to implement the calculation method of the layer velocity data volume provided in the above method embodiments.

[0131] Optionally, the present application also provides a computer program product containing instructions, which, when running on a computer device, causes the computer device to execute the calculation method of the layer velocity data volume described in the above aspects.

[0132] The serial numbers of the above embodiments of the present application are only for description and do not represent the advantages and disadvantages of the embodiments.

[0133] Those of ordinary skill in the art can understand that all or part of the steps to implement the above embodiments can be completed by hardware, or can be completed by instructing relevant hardware through a program. The program can be stored in a computer-readable storage medium. The storage medium mentioned above can be a read-only memory, a magnetic disk, an optical disc, etc.

[0134] The above are only alternative embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included within the protection scope of the present application.

Claims

1. A method for calculating a layer velocity data volume, characterized in that, The method includes: After dividing the gather data by central angle, calculating a central angle seismic attribute data volume related to the gather data; Determining different layer velocity models within a sliding time window; and establishing a central angle attribute model training set corresponding to the different layer velocity models within the sliding time window according to the central angle seismic attribute data volume; Training an approximate support vector machine using the central angle attribute model training set corresponding to the different layer velocity models; Using the approximate support vector machine to classify different layer velocity models for an attribute training set of seismic trace data within the sliding time window; and reconstructing layer velocity data for the seismic trace data within the sliding time window according to the classification result to obtain the layer velocity data volume; After dividing the gather data by central angle, calculating central angle seismic attributes related to the gather data, including: Dividing the gather data according to the designed central angle to obtain gather data for each central angle; Performing stacking processing and migration processing on the gather data for each central angle to obtain a post-stack data volume for each central angle; Calculating seismic attributes for the post-stack data volume for each central angle to obtain a seismic attribute data volume for each central angle; Reconstructing data for the seismic attribute data volume for each central angle according to common depth point (CDP) to obtain a central angle attribute data volume related to the gather data.

2. The method according to claim 1, characterized in that, Determining different layer velocity models within a sliding time window includes: Dividing the layer velocity model measured in the wells in the study area according to the target division factor to obtain different layer velocity models within the sliding time window; Wherein, the target division factor includes at least one of geological data, logging data, oil and gas test data, lithology combination, and layer velocity value range.

3. The method according to claim 2, wherein The method further includes: Adding a layer velocity model in a virtual well when the number of layer velocity models in the study area is less than the target number; And / or Adding a layer velocity model measured in a well in an adjacent area of the study area when the number of layer velocity models in the study area is less than the target number.

4. The method according to claim 1, wherein Establishing a central angle attribute model training set corresponding to the different layer velocity models within the sliding time window according to the central angle seismic attribute data volume includes: Determining each central angle seismic attribute curve within the sliding time window according to the central angle seismic attribute data volume; Determining the central angle attribute model training set corresponding to the different layer velocity models according to the different layer velocity models and their corresponding central angle seismic attribute curves within the sliding time window.

5. The method according to claim 4, wherein [[ID=—20]]Determining the central angle attribute model training set corresponding to the different layer velocity models according to the different layer velocity models and their corresponding central angle seismic attribute curves within the sliding time window includes: Extracting morphological feature parameters from the central angle seismic attribute curves according to the different layer velocity models and their corresponding central angle seismic attribute curves within the sliding time window to form a central angle attribute training model set corresponding to the different layer velocity models. Among them, the morphological feature parameters include at least one of the monotonicity of the curve, concavity and convexity, the number and average value of extreme points, and the number of inflection points.

6. The method according to claim 5, characterized in that, Using the approximate support vector machine to classify the attribute training set of seismic trace data within the sliding time window for different layer velocity models; reconstructing the layer velocity data for the seismic trace data within the sliding time window according to the classification result to obtain the layer velocity data volume, including: Classifying the attribute training set established on the central angle seismic attribute curves of each sampling point of the seismic trace data within the sliding time window by the approximate support vector machine to determine the layer velocity model corresponding to each sampling point; After obtaining the classification results of the sampling points belonging to different layer velocity models, performing reconstruction assignment on the layer velocity data of the sampling points within the sliding time window to obtain the layer velocity data volume.

7. A computing device for a layer velocity data volume, characterized in that, The device includes: A calculation module, configured to divide the gather data according to the designed central angle and calculate the central angle seismic attribute data volume related to the gather data; A building module, configured to determine different layer velocity models within the sliding time window; and establish a central angle attribute model training set corresponding to the different layer velocity models within the sliding time window according to the central angle seismic attribute data volume; A training module, configured to train an approximate support vector machine by using the central angle attribute model training set corresponding to the different layer velocity models; A classification module, configured to use the approximate support vector machine to classify the attribute training set of seismic trace data within the sliding time window for different layer velocity models; A reconstruction module, configured to reconstruct the layer velocity data of the seismic trace data within the sliding time window according to the classification result to obtain the layer velocity data volume; The calculation module is configured to: Divide the gather data according to the designed central angle to obtain each central angle gather data; Perform stacking processing and migration processing on each central angle gather data to obtain each post-stack data volume of the central angle; Perform seismic attribute calculation on each post-stack data volume of the central angle to obtain the seismic attribute data volume of each central angle; Perform data reconstruction on the seismic attribute data volume of each central angle according to common depth point (CDP) to obtain the central angle attribute data volume related to the gather data.

8. A computer device, characterized in that, The computer device includes: a processor and a memory, the memory stores a computer program, and the computer program is loaded and executed by the processor to implement the calculation method of the layer velocity data volume as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, and the computer program is loaded and executed by a processor to implement the calculation method of the layer velocity data volume as described in any one of claims 1 to 6.

Images