A method, apparatus, device and medium for synthesizing seismic records

Abstract

The application provides a seismic record synthesis method, device, equipment and medium, relates to the technical field of seismic data processing, and comprises the following steps: obtaining a measured seismic record of a target area; according to a configured processing rule, pre-processing the measured seismic record, target seismic data; wherein the processing rule comprises an effective waveform-noise binary identification rule, a partition denoising threshold, an amplitude relative relationship locking parameter, a mapping parameter and a sampling rate calibration parameter; inputting the target seismic data into a pre-trained seismic record synthesis model of the target area to obtain multiple initial synthetic seismic records; and performing full-dimensional feature optimization on each initial synthetic seismic record to obtain a set of synthetic seismic records of the target area. The application generates and optimizes the synthetic seismic record in full dimensions through a special model of the target area, significantly improves the amplitude fidelity, geological consistency and waveform continuity of the synthetic seismic record, and effectively supports high-precision seismic interpretation and reservoir prediction.

Description

Technical Field

[0001] This application relates to the field of earthquake data processing technology, and more specifically, to a method, apparatus, device, and medium for synthesizing earthquake records. Background Technology

[0002] In seismic exploration, synthetic seismic records are a core tool connecting well logging data with surface seismic data, and are widely used in key stages such as stratigraphic calibration, reservoir identification, and geological modeling. Traditional methods typically calculate reflection coefficient sequences based on sonic and density logging curves, and then generate synthetic records through convolution using Ricker wavelets or extracted seismic wavelets. However, such methods assume that the subsurface medium is a one-dimensional, linear, and attenuation-free ideal model, making it difficult to accurately characterize the complex lithological variations, heterogeneous structures, and physical effects such as frequency-varying attenuation and multiple interference during the propagation of real seismic waves in the target area. This results in significant differences between the synthetic results and the measured seismic data in terms of amplitude, phase, and waveform continuity.

[0003] With the development of artificial intelligence technology, data-driven deep learning models have been introduced into the field of seismic record synthesis, which can improve synthesis efficiency and generalization ability to a certain extent. However, existing models are mostly built based on general training sets and lack targeted adaptation to the geological conditions, sedimentary environment and seismic response characteristics of specific target areas. At the same time, the initial synthetic records directly output by the models often have problems such as amplitude distortion, geological logic inconsistencies and discontinuities in phase axes, making it difficult to meet the actual needs of high-precision seismic interpretation and quantitative reservoir prediction. Summary of the Invention

[0004] The purpose of this application is to provide a method, apparatus, device, and medium for synthesizing seismic records, which solves the above-mentioned problems existing in the prior art and can significantly improve the regional adaptability and physical authenticity of synthesized seismic records.

[0005] Firstly, a method for synthesizing seismic records is provided, which may include: Obtain measured seismic records for the target area; According to the configured processing rules, the measured seismic records are preprocessed to obtain the target seismic data; The target seismic data is input into a pre-trained seismic record synthesis model for the target area to obtain multiple initial synthetic seismic records; Full-dimensional feature optimization is performed on each initial synthetic seismic record to obtain a set of synthetic seismic records for the target region.

[0006] In an optional implementation, the processing rules include effective waveform-noise binary identification rules, partition denoising thresholds, amplitude relative relationship locking parameters, mapping parameters, and sampling rate calibration parameters; According to the configured processing rules, the measured seismic records are preprocessed to obtain the target seismic data, including: Using effective waveform-noise binary identification rules and partition denoising thresholds, the measured seismic records are subjected to differential denoising processing to obtain the first seismic data; The first seismic data is normalized using the amplitude relative relationship locking parameters and mapping parameters to obtain the second seismic data; Using the sampling rate calibration parameter, the second seismic data is resampled by time-series interpolation to obtain the target seismic data.

[0007] In an optional implementation, the seismic record synthesis model includes: The input layer is used to perform amplitude extremum smoothing and linear mapping on the target seismic data to obtain seismic feature vectors. The encoding layer is used to downsample the seismic feature tensor, extract convolutional features, and perform spatiotemporal attention weighting to obtain a deep global feature tensor. The fusion layer is used to perform 1×1 convolution dimensionality reduction and global feature fusion on the deep global feature tensor to obtain the fused feature tensor. The decoding layer is used to upsample, deconvolve feature restoration, and perform geological constraint linear correction on the fused feature tensor to obtain the restored feature tensor. The output layer is used to perform convolutional calibration, Sigmoid linear mapping, and Gaussian smoothing on the reconstructed feature tensor to obtain the initial synthetic seismic record.

[0008] In an optional implementation, the encoding layer includes: The downsampling convolution module is used to perform progressive convolution downsampling, batch normalization, and residual connection on the seismic feature tensor to obtain a multi-scale geological feature tensor set; The spatiotemporal attention weighting module is used to assign and weight the multi-scale geological feature tensor set with weights on both the time axis and the spatial axis to obtain the deep global feature tensor.

[0009] In an optional implementation, the decoding layer includes: The upsampling deconvolution module is used to perform progressive transpose convolution upsampling and channel number adaptation dimensionality reduction on the fused feature tensor to obtain a multi-scale upsampling feature tensor set; The jump connection feature fusion module is used to perform channel splicing and convolution fusion on the multi-scale upsampled feature tensor set and the geological feature tensor set of the same scale in the coding layer to obtain the upsampled feature tensor. The geological constraint linear correction module is used to perform feature-geological parameter linear mapping, overbound verification, and inverse mapping correction on the upsampled feature tensor to obtain the restored feature tensor.

[0010] In an optional implementation, after obtaining multiple initial synthetic seismic records, the method further includes: The initial synthetic seismic record is subjected to amplitude feature verification, geological law matching verification, and waveform continuity verification to obtain the verified synthetic seismic record.

[0011] In an optional implementation, any initial synthetic record includes: synthetic seismic wave amplitude values ​​corresponding to different synthetic depths, different synthetic sampling times, and different synthetic traces; Full-dimensional feature optimization is performed on each initial synthetic seismic record to obtain the synthetic seismic record of the target area, including: For any verified synthetic seismic record, the amplitude values ​​of the synthetic seismic waves in the verified synthetic seismic record are calibrated to obtain the first initial synthetic seismic record; Geological features were enhanced on the first initial synthetic seismic record to obtain a second initial synthetic seismic record; The waveform continuity enhancement process is applied to the second initial synthetic seismic record to obtain the synthetic seismic record of the target area.

[0012] Secondly, a seismic record synthesis apparatus is provided, which may include: Acquisition unit, used to acquire measured seismic records of the target area; The processing unit is used to preprocess the measured seismic records according to the configured processing rules to obtain the target seismic data; The synthesis unit is used to input the target seismic data into a pre-trained seismic record synthesis model of the target area to obtain multiple initial synthetic seismic records; The optimization unit is used to perform full-dimensional feature optimization on each initial synthetic seismic record to obtain a set of synthetic seismic records for the target region.

[0013] Thirdly, an electronic device is provided, which includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; When a processor executes a program stored in memory, it implements any of the steps described in the first aspect above.

[0014] Fourthly, a computer-readable storage medium is provided, wherein a computer program is stored therein, and when executed by a processor, the computer program implements the steps of any of the methods described in the first aspect above.

[0015] This application generates multiple initial synthetic seismic records by inputting pre-processed measured seismic data into a pre-trained target area-specific synthetic model. This effectively integrates regional geological characteristics and seismic response patterns, improving the physical rationality and regional adaptability of the synthetic results. Furthermore, it performs full-dimensional feature optimization on each initial synthetic record, systematically correcting amplitude response, strengthening the consistency of geological patterns, and enhancing the lateral continuity of waveforms, significantly reducing the feature deviation between the synthetic records and measured seismic data. The final target area synthetic seismic record set significantly outperforms traditional methods in terms of amplitude fidelity, geological interpretability, and waveform structure integrity. It not only provides high-precision support for automatic stratigraphic tracking and well-seismic joint calibration but also serves as reliable input for subsequent reservoir inversion, attribute prediction, and integrated seismic-geological modeling, significantly improving the interpretation accuracy and decision-making reliability of oil and gas exploration and development. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 An architecture diagram of a seismic record synthesis control system provided for embodiments of this application; Figure 2 A schematic flowchart illustrating a method for synthesizing seismic records provided in an embodiment of this application; Figure 3 A schematic diagram of a seismic record synthesis device provided in an embodiment of this application; Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application. Unless otherwise defined, the technical or scientific terms used in this application should have the ordinary meaning understood by those skilled in the art. The words "first," "second," and similar terms used in this application do not indicate any order, quantity, or importance, but are only used to distinguish different components. The words "comprising" or "including," etc., mean that the element or object preceding the word covers the element or object listed after the word and its equivalents, but do not exclude other elements or objects. The words "connected," "coupled," or "connected," etc., are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. "Up," "down," "left," "right," etc., are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0019] The method for synthesizing seismic records provided in this application embodiment can be applied to... Figure 1 In the system architecture shown, such as Figure 1 As shown, the system may include: a server and data acquisition equipment; Data acquisition equipment is used to acquire measured seismic records of the target area; The server is used to execute the method for synthesizing earthquake records provided in the embodiments of this application. The server can be a physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0020] The preferred embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit this application. Furthermore, the embodiments and features in the embodiments of this application can be combined with each other without conflict.

[0021] Figure 2 This is a schematic flowchart illustrating a method for synthesizing seismic records according to an embodiment of this application. Figure 2 As shown, the method may include: Step S210: Obtain the measured seismic records of the target area.

[0022] In practice, based on the configured acquisition parameters, seismic acquisition equipment in the target area is used to acquire simulated underground reflected seismic wave signals in a shot-by-scan, channel-by-channel excitation-reception mode. After analog-to-digital conversion, raw digital acquisition data with acquisition location and depth attributes are obtained. Invalid data is removed according to the initial quality screening criteria, and the valid data is supplemented with geological correlation attributes and its coordinates are calibrated. Then, standardized packaging is completed according to the standard format. The packaged data is classified, stored, and an index table is generated to obtain the measured seismic record. The measured seismic record can include seismic wave amplitude sequences at different locations in the target area at different sampling time points. The depths at different locations can be the same or different.

[0023] In one embodiment of this application, after obtaining the measured seismic records of the target area, the method may further include: Feature extraction is performed on the measured seismic records of the target area to obtain the seismic features of the target area. The seismic features may include: amplitude extreme range, dominant frequency range, noise type and distribution ratio, sampling rate, data format and depth range. From the configured seismic feature processing rule library, match the target region with the target processing rule corresponding to the seismic feature; wherein, the seismic feature processing rule library may include feature intervals and corresponding processing rules for different regions and different seismic features; the target processing rule may include: mapping parameters, effective waveform-noise binary recognition rules, zonal denoising threshold, amplitude relative relationship locking parameters, and sampling rate calibration parameters; The target processing rule is determined as the processing rule configured for the target region.

[0024] Step S220: Preprocess the measured seismic records according to the configured processing rules to obtain the target seismic data.

[0025] In practice Using the matched effective waveform-noise binary identification rules and zonal denoising thresholds, the measured seismic records are subjected to zonal differential denoising processing to obtain the first seismic data. Specifically, based on the configured geological exploration zoning standards of the target area (such as shallow / intermediate / deep, fault-developed areas / smooth stratigraphic areas), the measured seismic records are divided into several processing zones according to acquisition depth or spatial location, with each zone corresponding one-to-one with the matched zonal denoising threshold. For the measured seismic records within each processing zone, effective waveform-noise binary identification is performed point-by-point: based on the preset effective waveform feature boundaries of the target area in the binary identification rules (such as the frequency range, amplitude threshold, and waveform morphology characteristics of the effective waveform), the signal type of each sampling point is determined. For valid seismic waveforms or noise, generate signal type identification masks for each processing zone (the masks mark the coordinates of valid waveform sampling points and noise sampling points). For each processing zone, call its corresponding denoising threshold and perform thresholding filtering on the noise sampling points marked in the mask: only remove noise signals with amplitudes exceeding the zone denoising threshold (e.g., remove high-amplitude surface waves in shallow zones and high-amplitude multiple waves in deep zones), and retain weak noise signals with amplitudes below the threshold; do not modify the valid waveform sampling points marked in the mask, and completely retain their original characteristics; seamlessly stitch all the denoised seismic records from all processing zones in their original spatial or depth order to restore the original data structure of the records (inter-trace and inter-section continuity of gathers or profiles). Using amplitude relative relationship locking parameters and mapping parameters, the first seismic data is normalized to obtain the second seismic data. Specifically, layer-level amplitude features are extracted from the first seismic data to identify the reflection wave groups corresponding to each standard layer in the target area, and core amplitude features such as amplitude extrema, amplitude mean, and peak-to-trough amplitude ratio of each wave group are extracted. The amplitude relative relationship locking parameters are then used to rigidly lock the extracted amplitude features: through coefficient calibration, the relative amplitude ratios between different layers and different reflection wave groups in the first seismic data are fixed to ensure that the amplitude strength relationship of each wave group is consistent with the geological response law of the target area and does not cause distortion. The model input interval mapping parameters are then used to perform linear normalization calculations on the first seismic data after locking the amplitude relative relationship. The normalized data is then subjected to amplitude range verification to ensure that no values ​​exceed the limit. If there are values ​​exceeding the limit, secondary fine-tuning is performed through the mapping parameters to ensure that the data fully adapts to the model input requirements. Using the sampling rate calibration parameter, the second seismic data is resampled using temporal interpolation to obtain the target seismic data. Specifically, the original sampling rate of the second seismic data is read and compared with the target standard sampling rate in the sampling rate calibration parameter to determine whether calibration is needed. The feature-preserving interpolation algorithm in the sampling rate calibration parameter is called. Using the temporal interpolation step size coefficient as a reference, the second seismic data is resampled by channel-by-channel or profile-by-profile temporal interpolation: interpolation points are inserted between the original sampling points. The amplitude of the interpolation points is calculated by the cubic spline interpolation algorithm based on the amplitude and waveform trend of adjacent original sampling points, ensuring that the temporal evolution of the waveform after interpolation is consistent with the original. The original sampling points are screened according to the interpolation step size coefficient, retaining the core feature points of the waveform (peaks, troughs, phase abrupt change points) and eliminating redundant sampling points to ensure that the waveform is not distorted after downsampling. The sampling rate of the resampled data is checked to confirm that its sampling rate is completely consistent with the target standard sampling rate. At the same time, the waveform feature integrity of the resampled data is verified to ensure that the phase, frequency, and morphological features of the core effective waveform are not lost or distorted.

[0026] In another embodiment of this application, the method may further include: Using the configured data verification rules for the target area, the target seismic data is automatically verified for format, numerical value, and feature integrity. If the verification passes, the preprocessing is completed; if the verification fails, the pre-stored fixed area-specific correction rules are triggered for one-click correction until the verification passes.

[0027] Step S230: Input the target seismic data into the pre-trained seismic record synthesis model of the target area to obtain multiple initial synthetic seismic records.

[0028] The target seismic data may include depth data, sampling time, and corresponding seismic wave amplitude values; the seismic record synthesis model is trained using three types of core data: measured seismic data of the target area, corresponding geological exploration calibration data, and seismic wave forward modeling data.

[0029] In one embodiment of this application, the seismic record synthesis model may include: The input layer is used to smooth the amplitude extrema and perform linear mapping on the target seismic data to obtain seismic feature vectors. Specifically, it iterates through all amplitude values ​​in the target seismic data, identifies isolated extrema points (amplitude points that deviate from the mean of the 5 neighboring sampling points by ±3σ), and smooths the amplitude values ​​of the extrema points using the mean replacement method of the 5 neighboring points to obtain multiple smoothed amplitude values. The multiple smoothed amplitude values, along with the corresponding depth data and sampling time, are used as the processed seismic data tensor. The amplitude values ​​in the processed seismic data tensor are linearly mapped to the model adaptation interval [-0.99, 0.99] using a linear mapping formula to complete the numerical interval standardization constraint. The mapped amplitude values, along with the corresponding depth data and sampling time, are used as the seismic feature tensor. The encoding layer is used to downsample the seismic feature tensor, extract convolutional features, and perform spatiotemporal attention weighting to obtain a deep global feature tensor. The fusion layer is used to perform 1×1 convolutional dimensionality reduction and global feature fusion on the deep global feature tensor to obtain a fused feature tensor. Specifically, a 1×1 pre-trained convolutional kernel with no stride and no padding is used to perform a convolution operation on the deep global feature tensor. This process calibrates the small amplitude deviations generated during the upsampling and correction process of the decoding layer point by point, ensuring the accuracy and consistency of the amplitude values, resulting in a dimensionality-reduced deep feature tensor. The dimensionality-reduced deep feature tensor is then subjected to spatiotemporal global average pooling to extract the global mean features of multiple channels, generating a mean feature tensor. The mean feature tensor is then restored by dimensionality replication and added element-wise with the dimensionality-reduced deep feature tensor to achieve the fusion of global mean patterns and local deep features, resulting in a fused feature tensor. The decoding layer is used to upsample, deconvolve feature restoration, and perform geological constraint linear correction on the fused feature tensor to obtain the restored feature tensor. The output layer performs convolution calibration, Sigmoid linear mapping, and Gaussian smoothing on the reconstructed feature tensor to obtain multiple initial synthetic seismic records containing synthetic seismic wave amplitude values ​​corresponding to different synthesis depths, synthesis sampling times, and synthetic seismic traces. Specifically, a 1×1 pre-trained convolutional kernel with no stride and no padding is used to perform a convolution operation on the reconstructed feature tensor, calibrating the small amplitude deviations generated during the upsampling and correction process in the decoding layer point by point to ensure the accuracy and consistency of amplitude values, resulting in a calibrated feature tensor. The calibrated amplitude values ​​are precisely mapped from any interval to the standard interval [-1,1] using the Sigmoid linear mapping formula, perfectly matching the numerical interval of the input target seismic data, resulting in an interval-standardized feature tensor. A 5×5 pre-trained Gaussian smoothing window is used to perform point-by-point multi-neighbor Gaussian weighted mean smoothing on the interval-standardized feature tensor to eliminate waveform glitches generated by model inference, ensuring the continuity of amplitude temporal sequence, and obtaining the initial synthetic seismic records.

[0030] In one embodiment of this application, the encoding layer may include: The downsampling convolution module is used to perform progressive convolutional downsampling, batch normalization, and residual connection on the seismic feature tensor to obtain a multi-scale geological feature tensor set. Specifically, a 4-layer progressive convolutional downsampling is used, with each layer using a 3×3 pre-trained convolution kernel to sequentially convolve the seismic feature tensor. After each layer of convolution, batch normalization and residual connection are superimposed to obtain a multi-scale geological feature tensor set. The spatiotemporal attention weighting module is used to assign and weight the multi-scale geological feature tensor set with weights along both the time and spatial axes to obtain a deep global feature tensor. Specifically, a pre-trained spatiotemporal dual-axis attention weight matrix is ​​used to perform spatiotemporal dual-axis attention weighting on each layer of feature tensor to obtain a deep global feature tensor. This application assigns weights to the sampling time and depth dimensions of seismic data separately to accurately enhance core geological features such as stratigraphy and faults, solving the problem of non-targeting in traditional convolutional feature extraction.

[0031] In another embodiment of this application, the decoding layer may include: The upsampling deconvolution module is used to perform progressive transposed convolution upsampling and channel number adaptation dimensionality reduction on the fused feature tensor to obtain a multi-scale upsampled feature tensor set. Specifically, four layers of progressive transposed convolution upsampling are used, with each layer using a 3×3 pre-trained transposed convolution kernel to perform progressive transposed convolution upsampling on the fused feature tensor to obtain a multi-scale upsampled feature tensor set. The jump connection feature fusion module is used to perform channel concatenation and convolution fusion on the multi-scale upsampled feature tensor set and the geological feature tensor set of the same scale in the coding layer to obtain the upsampled feature tensor of fused details. Specifically, the feature tensors of each upsampled layer are concatenated element-wise with the tensor of the same scale in the coding layer, and then the number of channels after concatenation is restored to the original number of channels after upsampling through 1×1 convolution, so as to realize the fusion of the detailed features of the coding layer and the global features of the decoding layer to obtain the upsampled feature tensor of fused details. The geological constraint linear correction module is used to perform feature-geological parameter linear mapping, overbound verification, and inverse mapping correction on the upsampled feature tensor of the fused details to obtain the restored feature tensor. Specifically, it is based on the configured geological constraint parameters of the target area, which may include the range of stratum thickness, fault dip angle, and wave velocity. The tensor amplitude values ​​in the upsampled feature tensor are transformed into interpretable geological parameters using the feature-geological parameter linear mapping formula. The interpretable geological parameters are checked point by point to see if they are within the range of geological constraint parameters, and the out-of-boundary points are marked. The nearest boundary value correction method is used for the out-of-boundary points to correct the interpretable geological parameters to the boundary values ​​of the constraint range. Then, the inverse mapping formula is used to restore them to amplitude values, which replace the original out-of-boundary amplitude values, and the restored feature tensor is obtained.

[0032] In another embodiment of this application, after obtaining the initial synthetic seismic record, the method may further include: The initial synthetic seismic record is validated for amplitude characteristics, geological regularity matching, and waveform continuity to obtain a validated initial synthetic seismic record. Specifically, for any amplitude tensor in the initial synthetic seismic record, core amplitude characteristic indicators are extracted channel by channel and depth by depth, including amplitude extrema, amplitude mean, amplitude variance, and reflected wave group amplitude ratio, forming an amplitude characteristic indicator set. The extracted amplitude characteristic indicator set is compared one by one with the configured amplitude characteristic validation threshold for the target area, and amplitude anomalies exceeding the threshold are marked. For the marked amplitude anomalies, interpolation correction is performed using the measured amplitude characteristic mean of the target area, and invalid data segments are discarded. Afterwards, seamless splicing of effective segments is performed to ensure the integrity of amplitude data; amplitude feature indicators of the corrected records are extracted again, and it is confirmed that all indicators meet the verification threshold, resulting in a synthetic seismic record with qualified amplitude verification; the synthetic seismic record with qualified amplitude verification is then converted into interpretable quantitative geological parameters, including stratum thickness, fault dip angle, seismic wave velocity, and layer reflection intensity, through a pre-trained feature-geological parameter linear mapping formula, forming a geological parameter set; the converted geological parameter set is then compared one by one with the geological regularity matching verification interval of the target area, and points of geological regularity mismatch outside the interval are marked; for geological regularity mismatches... The data segments are filtered, with only invalid data segments containing more than 5 consecutive mismatched points being removed. The remaining isolated mismatched points are retained (to avoid over-correction that could damage the original waveform characteristics). After data segment filtering, seamless stitching is performed. It is confirmed that the geological parameter set recorded after stitching falls within the geological regularity matching interval of the target area, resulting in a synthetic seismic record with acceptable geological regularity matching. For the synthetic seismic record with acceptable geological regularity matching, waveform continuity features are extracted along the sampling point dimension, including amplitude differences between adjacent sampling points, phase continuity of reflected wave groups, and inter-trace waveform correlation coefficients, forming a continuity feature index set. This continuity feature index set is then compared with the waveform continuity determination of the target area. Standards (such as amplitude difference between adjacent sampling points ≤ 0.1, waveform correlation coefficient between traces ≥ 0.85) are used for comparison to mark waveform discontinuities with abrupt amplitude changes, phase breaks, and inter-trace discrepancies. For the marked waveform discontinuities, a 3-point neighborhood Gaussian weighted smoothing method is used for correction, which preserves the core waveform features while eliminating abrupt changes and breaks, ensuring the temporal continuity and spatial consistency of the seismic wave waveform. The corrected synthetic seismic record is then fully checked again for amplitude characteristics, geological patterns, and waveform continuity to confirm that there are no abnormal indicators and that all features meet the target area verification benchmark, thus obtaining the first synthetic seismic record.

[0033] Step S240: Optimize each initial synthetic seismic record to obtain a set of synthetic seismic records for the target area.

[0034] The synthetic seismic records may include: synthetic seismic wave amplitude values ​​corresponding to different synthetic depths, different synthetic sampling times, and different synthetic seismic traces; In practice, for any verified synthetic seismic record, the amplitude values ​​of the synthetic seismic waves in the verified synthetic seismic record are calibrated to obtain an initial synthetic seismic record with calibrated amplitude values. Specifically, based on this initial synthetic seismic record, an amplitude value matrix is ​​constructed. The amplitude value matrix is ​​then compared point-by-point with the amplitude standard features of the configured target area to generate an amplitude deviation matrix. For amplitude deviation points in the amplitude deviation matrix that are not less than a first deviation threshold and not greater than a second deviation threshold, the mean amplitude of the neighborhood is used to weight and correct the amplitude deviation point. Taking the deviation point as the center, values ​​from the same channel and... The standard amplitude mean of five adjacent synthetic sampling times at the same depth is used to calculate the corrected amplitude value by weighting it 0.7 (standard mean) + 0.3 (original amplitude value). The first deviation threshold can be 0.01, and the second deviation threshold can be 0.05. For amplitude deviation points greater than the second deviation threshold, the standard amplitude value of the target area is directly replaced with the measured statistical standard amplitude value of the corresponding synthetic depth and synthetic sampling time interval. All corrected amplitude values ​​are then re-locked to the standard interval [-1,1] through linear mapping to obtain the synthetic seismic record with calibrated amplitude values. Geological feature enhancement is applied to the first initial synthetic seismic record to obtain the second initial synthetic seismic record. Specifically, based on the configured geological-amplitude mapping parameters of the target area, the distribution pattern of the amplitude values ​​of the first synthetic seismic waves in the first initial synthetic seismic record is analyzed to locate the core amplitude value segments of the peak amplitude value clusters (corresponding to layer reflections), amplitude value abrupt change segments (corresponding to fault structures), and amplitude value plateau segments (corresponding to thick interbedded strata) in the synthetic depth direction. The distribution location and amplitude variation of the core amplitude value segments are compared with the configured geological structure matching requirements of the target area, and mismatched amplitude value segments are marked. For the matching core geological structure amplitude value segments, amplitude value weighting enhancement is applied (the corresponding synthetic seismic wave amplitude value segments are weighted and enhanced). (Multiply the seismic wave amplitude value by a specific enhancement coefficient of 1.02~1.05). For mismatched amplitude value segments, perform linear adjustment of the amplitude value gradient according to geological structure requirements to correct amplitude changes and distribution locations, and conform to the geological structure of the target area. Verify the variation trend of synthetic seismic wave amplitude values ​​corresponding to different synthetic seismic traces for the same geological structure, eliminate inter-trace dispersion, ensure consistent geological response, and obtain the second initial synthetic seismic record. The second initial synthetic seismic record may include the second synthetic seismic wave amplitude values ​​corresponding to different synthetic depths, synthetic sampling times, and synthetic seismic traces after optimization. The amplitude value distribution pattern is completely matched with the geological structure of the target area, and the identification of core geological features is improved. Waveform continuity enhancement processing is performed on the second initial synthetic seismic record to obtain the synthetic seismic record of the target area; specifically, For the amplitude values ​​of the second synthetic seismic wave in the second initial synthetic seismic record, three types of continuity indices were calculated and anomalies were marked: (1) Temporal continuity: the difference in amplitude values ​​between adjacent synthetic sampling times under the same synthetic seismic trace and synthetic depth, and marking the temporal abrupt change points with a difference > 0.1; (2) Depth continuity: the gradient of amplitude value changes between adjacent synthetic depths under the same synthetic seismic trace and synthetic sampling time, and marking the depth anomaly segments with gradient abrupt changes; (3) Inter-trace continuity: the correlation coefficient of amplitude values ​​between different synthetic seismic traces under the same synthetic depth and synthetic sampling time, and marking the inter-trace discrete segments with a correlation coefficient < 0.9. Categorized smoothing correction: Temporal abrupt change points are smoothed using a 3-point neighborhood Gaussian weighted smoothing method (with the abrupt change point as the center, the amplitude values ​​of one sampling point before and after the abrupt change point are taken, and the smoothed amplitude values ​​are calculated with a weight of 0.5 for the center and 0.25 for both sides); depth anomaly segments are smoothed using linear interpolation, and interpolated amplitude values ​​are added in the synthetic depth interval of the abrupt change to conform to the gradient change law; inter-trace discrete segments are smoothed using adaptive fusion of inter-trace amplitude values, and the mean amplitude values ​​of discrete segments and adjacent normal segments are taken for linear correction to improve the inter-trace correlation coefficient; three types of continuity indicators are recalculated, and all indicators are verified to meet the waveform continuity judgment threshold of the target area to obtain the synthetic seismic record of the target area.

[0035] In one embodiment of this application, after obtaining the synthetic seismic record of the target area, the method may further include: This process is used to perform dimensional calibration, attribute completion, and format encapsulation on optimized synthetic seismic records to obtain standardized synthetic seismic records. Specifically, based on the configured seismic exploration industry data standards and specifications, full-dimensional attribute annotations are added to the synthetic seismic records of the target area, including core metadata such as synthetic seismic trace number, synthetic depth range, synthetic sampling time interval, amplitude value range, and geological code of the target area. The attribute-annotated synthetic seismic records are converted into a general seismic exploration data format, retaining the three-dimensional structure of [synthetic seismic trace C, synthetic depth H, synthetic sampling time W] and all synthetic seismic wave amplitude values. The encapsulated data undergoes format verification to confirm that there is no missing metadata, the format fully matches the industry standard, and there are no data structure errors. According to the configured application scenario requirements, the synthetic seismic wave amplitude values ​​of the synthetic seismic records are adjusted in a targeted manner, such as slicing according to the measured data volume in data augmentation scenarios and scaling according to sample specifications in model training scenarios. The process involves cropping the data to retain core geological response features; supplementing the synthetic seismic records with scenario-specific attribute annotations (such as data augmentation sample numbers, model training batch identifiers, and exploration scheme simulation parameters); verifying the matching degree between the adapted data and scenario requirements, confirming that amplitude values, dimensional structure, and attribute annotations all conform to actual application requirements, and obtaining scenario-adapted synthetic seismic records; comparing the amplitude values ​​of the scenario-adapted synthetic seismic records with measured seismic data from the same region and dimension, calculating the overall similarity of the synthetic seismic wave amplitude values ​​and the matching degree of the amplitude segments corresponding to the core geological structures; deploying the synthetic seismic records into specific application scenarios for on-the-ground testing, and verifying the effectiveness of the application according to scenario effect evaluation indicators (such as verifying the exploration interpretation accuracy after the completion of measured data in data augmentation scenarios, and verifying the training accuracy of the geological identification model in model training scenarios); and making local fine-tuning corrections to minor deviations found during verification to ensure that the on-the-ground application effect meets the standards.

[0036] Corresponding to the above method, embodiments of this application also provide a seismic record synthesis apparatus, such as... Figure 3 As shown, the device includes: Acquisition unit 310 is used to acquire measured seismic records of the target area; The processing unit 320 is used to preprocess the measured seismic records according to the configured processing rules to obtain the target seismic data; Synthesis unit 330 is used to input target seismic data into a pre-trained seismic record synthesis model of the target area to obtain multiple initial synthetic seismic records; The optimization unit 340 is used to perform full-dimensional feature optimization on each initial synthetic seismic record to obtain a set of synthetic seismic records for the target area.

[0037] The functions of each functional unit of the seismic record synthesis device provided in the above embodiments of this application can be implemented through the above method steps. Therefore, the specific working process and beneficial effects of each unit in the seismic record synthesis device provided in the embodiments of this application will not be repeated here.

[0038] This application also provides an electronic device, such as... Figure 4 As shown, it includes a processor 410, a communication interface 420, a memory 430, and a communication bus 440, wherein the processor 410, the communication interface 420, and the memory 430 communicate with each other through the communication bus 440.

[0039] Memory 430 is used to store computer programs; When the processor 410 executes the program stored in the memory 430, it performs the following steps: Obtain measured seismic records for the target area; According to the configured processing rules, the measured seismic records are preprocessed to obtain the target seismic data; The target seismic data is input into a pre-trained seismic record synthesis model for the target area to obtain multiple initial synthetic seismic records; Full-dimensional feature optimization is performed on each initial synthetic seismic record to obtain a set of synthetic seismic records for the target region.

[0040] The communication bus mentioned above can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not mean that there is only one bus or one type of bus.

[0041] The communication interface is used for communication between the aforementioned electronic devices and other devices.

[0042] The memory may include random access memory (RAM) or non-volatile memory (NVM), such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.

[0043] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0044] The implementation methods and beneficial effects of the various components of the electronic device in the above embodiments for solving the problem can be found in [reference needed]. Figure 2 The steps in the illustrated embodiments are used to implement the electronic device. Therefore, the specific working process and beneficial effects of the electronic device provided in this application will not be repeated here.

[0045] In another embodiment provided in this application, a computer-readable storage medium is also provided, which stores instructions that, when executed on a computer, cause the computer to perform the method for synthesizing seismic records as described in any of the above embodiments.

[0046] In another embodiment provided in this application, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to perform the method for synthesizing seismic records as described in any of the above embodiments.

[0047] Those skilled in the art will understand that the embodiments in this application can be provided as methods, systems, or computer program products. Therefore, the embodiments in this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the embodiments in this application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0048] This application describes embodiments of methods, apparatus (systems), and computer program products according to embodiments of this application with reference to flowchart illustrations and / or block diagrams. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0049] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0050] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0051] Although preferred embodiments have been described in this application, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of the embodiments of this application.

[0052] Obviously, those skilled in the art can make various modifications and variations to the embodiments of this application without departing from the spirit and scope of the embodiments of this application. Therefore, if these modifications and variations to the embodiments of this application fall within the scope of this application and its equivalents, then these modifications and variations are also intended to be included in the embodiments of this application.

Claims

1. A method for synthesizing seismic records, characterized in that, The method includes: Obtain measured seismic records for the target area; According to the configured processing rules, the measured seismic records are preprocessed to obtain the target seismic data; The target seismic data is input into a pre-trained seismic record synthesis model for the target area to obtain multiple initial synthetic seismic records; Full-dimensional feature optimization is performed on each initial synthetic seismic record to obtain a set of synthetic seismic records for the target region.

2. The method as described in claim 1, characterized in that, The processing rules include effective waveform-noise binary identification rules, partition denoising threshold, amplitude relative relationship locking parameters, mapping parameters, and sampling rate calibration parameters; According to the configured processing rules, the measured seismic records are preprocessed to obtain the target seismic data, including: Using effective waveform-noise binary identification rules and partition denoising thresholds, the measured seismic records are subjected to differential denoising processing to obtain the first seismic data; The first seismic data is normalized using the amplitude relative relationship locking parameters and mapping parameters to obtain the second seismic data; Using the sampling rate calibration parameter, the second seismic data is resampled by time-series interpolation to obtain the target seismic data.

3. The method as described in claim 1, characterized in that, The seismic record synthesis model includes: The input layer is used to perform amplitude extremum smoothing and linear mapping on the target seismic data to obtain seismic feature vectors. The encoding layer is used to downsample the seismic feature tensor, extract convolutional features, and perform spatiotemporal attention weighting to obtain a deep global feature tensor. The fusion layer is used to perform 1×1 convolution dimensionality reduction and global feature fusion on the deep global feature tensor to obtain the fused feature tensor. The decoding layer is used to upsample, deconvolve feature restoration, and perform geological constraint linear correction on the fused feature tensor to obtain the restored feature tensor. The output layer is used to perform convolutional calibration, Sigmoid linear mapping, and Gaussian smoothing on the reconstructed feature tensor to obtain the initial synthetic seismic record.

4. The method as described in claim 3, characterized in that, The coding layer includes: The downsampling convolution module is used to perform progressive convolution downsampling, batch normalization, and residual connection on the seismic feature tensor to obtain a multi-scale geological feature tensor set; The spatiotemporal attention weighting module is used to assign and weight the multi-scale geological feature tensor set with weights on both the time axis and the spatial axis to obtain the deep global feature tensor.

5. The method as described in claim 3, characterized in that, The decoding layer includes: The upsampling deconvolution module is used to perform progressive transpose convolution upsampling and channel number adaptation dimensionality reduction on the fused feature tensor to obtain a multi-scale upsampling feature tensor set; The jump connection feature fusion module is used to perform channel splicing and convolution fusion on the multi-scale upsampled feature tensor set and the geological feature tensor set of the same scale in the coding layer to obtain the upsampled feature tensor. The geological constraint linear correction module is used to perform feature-geological parameter linear mapping, overbound verification, and inverse mapping correction on the upsampled feature tensor to obtain the restored feature tensor.

6. The method as described in claim 1, characterized in that, After obtaining multiple initial synthetic seismic records, the method further includes: The initial synthetic seismic record is subjected to amplitude feature verification, geological law matching verification, and waveform continuity verification to obtain the verified synthetic seismic record.

7. The method as described in claim 6, characterized in that, Any initial synthetic record includes: synthetic seismic wave amplitude values ​​corresponding to different synthetic depths, different synthetic sampling times, and different synthetic traces; Full-dimensional feature optimization is performed on each initial synthetic seismic record to obtain the synthetic seismic record of the target area, including: For any verified synthetic seismic record, the amplitude values ​​of the synthetic seismic waves in the verified synthetic seismic record are calibrated to obtain the first initial synthetic seismic record; Geological features were enhanced on the first initial synthetic seismic record to obtain a second initial synthetic seismic record; The waveform continuity enhancement process is applied to the second initial synthetic seismic record to obtain the synthetic seismic record of the target area.

8. A device for synthesizing seismic records, characterized in that, The device includes: Acquisition unit, used to acquire measured seismic records of the target area; The processing unit is used to preprocess the measured seismic records according to the configured processing rules to obtain the target seismic data; The synthesis unit is used to input the target seismic data into a pre-trained seismic record synthesis model of the target area to obtain multiple initial synthetic seismic records; The optimization unit is used to perform full-dimensional feature optimization on each initial synthetic seismic record to obtain a set of synthetic seismic records for the target region.

9. An electronic device, characterized in that, The electronic device includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; A processor, when executing a program stored in memory, implements the method of any one of claims 1-7.

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

Images