Seismic exploration method and system based on data enhancement
By placing geophones at favorable mineralization locations and using a closed-loop design for data enhancement and borehole verification via residual shrinkage networks, the problems of low signal-to-noise ratio and missing data in seismic exploration were solved, achieving efficient enhancement and accurate interpretation of seismic data.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- AEROSPACE INFORMATION TECH UNIV
- Filing Date
- 2026-03-03
- Publication Date
- 2026-04-21
AI Technical Summary
In seismic exploration, conventional signal enhancement techniques are insufficient to effectively improve the signal-to-noise ratio of seismic data, resulting in unsatisfactory imaging quality, especially under complex geological conditions where data loss and severe noise are common problems.
A data augmentation method based on residual shrinkage networks is adopted, which combines multi-source data fusion and borehole verification closed-loop design. Seismic exploration data is acquired by arranging receiver points or shot points in favorable mineralization locations, data augmentation is performed using residual shrinkage networks, and the interpreted data is optimized through 3D interpretation and borehole verification to achieve closed-loop iterative optimization of the data.
It improved the signal-to-noise ratio of seismic data, enhanced the resolution of effective reflection signals, solved the problem of missing data, ensured the accuracy and continuity of interpreted data, and improved exploration efficiency and interpretation accuracy.
Smart Images
Figure CN121899902A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of geophysical exploration technology, and in particular to a data-augmented seismic exploration method and system. Background Technology
[0002] Seismic exploration is an indispensable technology in geological exploration and resource development, widely used in oil and gas exploration, mineral resource development, and earthquake prediction. With the continuous development of Earth science, the complexity and diversity of seismic signals are increasing. Enhancing seismic data to improve its signal-to-noise ratio and thus image quality is a crucial step in processing seismic exploration data in complex areas. Currently, in the field of Earth geography exploration technology, conventional signal enhancement techniques are used to enhance seismic data. For example, during the acquisition phase, optimizing acquisition parameters and improving acquisition equipment are used to improve the quality of raw seismic data. However, even after a series of refined processing steps, the seismic data obtained in actual production still contains noise and missing data. Conventional signal enhancement techniques are not very effective in enhancing seismic data, resulting in unsatisfactory image quality based on the enhanced seismic data. Summary of the Invention
[0003] The purpose of this invention is to provide a data-enhanced seismic exploration method and system to solve the problems mentioned in the background art, enhance the acquired seismic exploration data, and improve the quality of the seismic data.
[0004] To achieve the above objectives, the present invention provides the following solution: a data-augmented seismic exploration method, the specific steps of which include the following:
[0005] Obtain geological exploration data of the target area, analyze and determine the favorable mineralization locations in the target area;
[0006] Seismic exploration data are collected by setting up geophones or shot points in favorable mineralization locations.
[0007] The acquired seismic exploration data is augmented using a residual shrinkage network to obtain enhanced seismic data;
[0008] Interpret the enhanced seismic data and output three-dimensional seismic interpretation data;
[0009] The accuracy of the 3D seismic interpretation data was verified and corrected by designing borehole locations, thus completing the 3D seismic exploration.
[0010] Preferably, the specific steps for arranging geophone points or shot points at favorable mineralization locations to acquire seismic exploration data are as follows:
[0011] The grid information is determined based on the construction area information, and the grid information includes grid spacing values and grid shape parameters;
[0012] Based on the grid information, detector points or shot points are randomly deployed within each grid.
[0013] Based on the type of ore deposit in the favorable mineralization location, matching observation system parameters are designed. The observation system parameters include channel spacing, shot spacing, coverage times, and receiver array length. Different ore deposit types correspond to different parameter combinations.
[0014] The artificial seismic source located at the grid node is activated, and the actual detection data of all the aforementioned receiver points are obtained based on the observation system parameters.
[0015] Preferably, the excitation mode of the artificial seismic source is determined through experiments based on geological conditions.
[0016] Preferably, the method further includes correcting the receiver point or shot point. Specifically, the method is as follows: using the observation system parameters and seismic trace data file as initial conditions, scanning the observation system graph and seismic trace graph, judging and determining the shot point or receiver point that needs to be corrected based on the results of linear dynamic correction, gradually correcting the position coordinates of the shot point or receiver point that needs to be corrected until the linear dynamic correction accuracy meets the requirements, ending the correction process, and outputting the corrected and modified observation system data file.
[0017] Preferably, the acquired seismic exploration data is first preprocessed and then augmented based on a residual shrinkage network.
[0018] Preferably, the step of preprocessing the seismic exploration data is as follows:
[0019] To address the various noises present in the seismic exploration data, a combination of pre-stack one-dimensional filtering and two-dimensional filtering is used to suppress interference and improve the signal-to-noise ratio.
[0020] The surface-consistent deconvolution technique is used to correct the seismic wavelet, eliminating the influence of excitation and reception on the seismic wavelet;
[0021] Based on the residual static correction, predictive deconvolution is performed to improve the longitudinal resolution.
[0022] Preferably, the specific steps for data augmentation based on residual shrinkage networks to obtain augmented seismic data are as follows:
[0023] After converting the format of the seismic exploration data, gather extraction is performed to extract valid seismic gather data;
[0024] To label the noise type in the training data, synthetic noise is injected into clean seismic data to generate training samples;
[0025] Design a residual shrinking network, which includes a residual module, a soft thresholding layer, and an attention mechanism; wherein, the residual module includes a convolutional layer, batch normalization, and ReLU activation module, supporting cross-layer feature propagation; a soft thresholding layer is inserted in the residual path, and the noise threshold of each sample is dynamically calculated through the sub-network to shrink irrelevant features to zero;
[0026] The residual shrinkage network is trained using the training samples, with mean squared error loss, using high signal-to-noise ratio labeled seismic data as the ground truth, and constraining the difference between the model output and the ground truth to obtain the trained residual shrinkage network.
[0027] The effective seismic gather data is input into the trained residual shrinkage network, which outputs the enhanced seismic data.
[0028] On the other hand, a data-augmented seismic exploration system is provided, including a location determination module, a seismic exploration module, a data augmentation module, an interpretation module, and a verification module; wherein,
[0029] The location determination module is used to acquire geological exploration data of the target area, analyze and determine the favorable mineralization locations in the target area;
[0030] The seismic exploration module is used to set up geophones or shot points in favorable mineralization locations to collect seismic exploration data.
[0031] The data augmentation module is used to augment the acquired seismic exploration data based on a residual shrinkage network to obtain enhanced seismic data.
[0032] The interpretation module is used to interpret the enhanced seismic data and output three-dimensional seismic interpretation data;
[0033] The verification module is used to verify the accuracy of the three-dimensional seismic interpretation data by designing borehole locations, and to make corrections to complete the three-dimensional seismic exploration.
[0034] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:
[0035] (1) Through the closed-loop design of “target area precise focusing - intelligent data enhancement - interpretation verification closed loop”, compared with traditional seismic exploration methods, multiple improvements have been achieved in exploration efficiency, data quality and interpretation accuracy;
[0036] (2) Data augmentation has the dual effect of noise suppression and effective signal enhancement. The residual shrinkage network has attention mechanism and soft threshold shrinkage characteristics, which can accurately identify and suppress interference signals such as surface waves, multiple waves and random noise in seismic data. At the same time, it enhances the effective reflection signals related to mineralization such as ore layer interfaces and fault zones. After enhancement, the signal-to-noise ratio of seismic data is improved, and the resolution of effective reflection layers is improved.
[0037] (3) In response to the problems of missing data and incomplete gathers caused by the complex geological conditions of the metallogenic belt, the residual shrinkage network can complete the data through feature learning, avoid the distortion problem of traditional interpolation methods, ensure the geological continuity and physical rationality of the enhanced data, and provide a complete and reliable data foundation for three-dimensional interpretation.
[0038] (4) Iterative optimization of borehole verification closed loop can quantitatively evaluate the interpretation error and reverse the seismic inversion model, improve the matching degree between the corrected interpretation data and the actual geological conditions, and effectively avoid the problem of theoretical interpretation being out of touch with actual geology. Attached Figure Description
[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0040] Figure 1 This is a flowchart of the method of the present invention;
[0041] Figure 2 This is a system structure diagram of the present invention. Detailed Implementation
[0042] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0043] The purpose of this invention is to provide a data-augmented seismic exploration method, such as... Figure 1 As shown, the specific steps include the following:
[0044] S1. Obtain geological exploration data of the target area, analyze and determine the favorable mineralization locations in the target area;
[0045] S2. Set up geophones or shot points in favorable mineralization locations to collect seismic exploration data.
[0046] S3. The acquired seismic exploration data is augmented using a residual shrinkage network to obtain enhanced seismic data;
[0047] S4. Interpret enhanced seismic data and output 3D seismic interpretation data;
[0048] S5. Verify the accuracy of the 3D seismic interpretation data by designing borehole locations, make corrections, and complete the 3D seismic exploration.
[0049] Furthermore, multi-source geological exploration data of the target area is obtained in S1. The data is analyzed and quantitatively judged through a multi-source data fusion model to determine the favorable mineralization locations in the target area. The multi-source data fusion model adopts the evidence weight method or fuzzy comprehensive evaluation method, sets a quantitative threshold for mineralization favorableness, and defines areas with a comprehensive score ≥ the preset threshold as favorable mineralization locations.
[0050] Furthermore, in S2, high-density geophones are deployed only in favorable primary mineralization areas, while sparse verification geophones are deployed in favorable secondary mineralization areas. The specific steps for deploying geophones or shot points in favorable mineralization areas for seismic exploration data acquisition are as follows:
[0051] S21. Determine the grid information based on the construction area information. The grid information includes the grid spacing value and the grid shape parameters.
[0052] S22. Randomly deploy receiver points or shot points within each grid based on grid information;
[0053] S23. Based on the type of ore deposit in the favorable mineralization location, design matching observation system parameters. The observation system parameters include track spacing, shot spacing, coverage times and receiver array length. Different ore deposit types correspond to different parameter combinations.
[0054] S24. Activate the artificial seismic source located at the grid node. The excitation mode of the artificial seismic source is determined through experiments based on geological conditions. Obtain the actual detection data of all receiver points based on the observation system parameters.
[0055] Furthermore, it also includes the correction of receiver points or shot points. The specific method is as follows: using the observation system parameters and seismic trace data file as initial conditions, scanning the observation system graph and seismic trace graph, judging and determining the shot points or receiver points that need to be corrected based on the results of linear dynamic correction, and gradually correcting the position coordinates of the shot points or receiver points that need to be corrected until the linear dynamic correction accuracy meets the requirements, ending the correction process, and outputting the corrected and modified observation system data file.
[0056] Furthermore, the acquired seismic exploration data undergoes preprocessing before data augmentation based on a residual shrinkage network. The steps for preprocessing the seismic exploration data are as follows:
[0057] S311. To address various noises in seismic exploration data, a combination of pre-stack one-dimensional filtering and two-dimensional filtering is used to suppress interference and improve the signal-to-noise ratio.
[0058] S312. The surface-consistent deconvolution technique is used to correct the seismic wavelet and eliminate the influence of excitation and reception on the seismic wavelet.
[0059] S313. Based on the residual static correction, perform predictive deconvolution to improve the longitudinal resolution.
[0060] The method combining pre-stack one-dimensional filtering and two-dimensional filtering processes seismic data in stages, utilizing the differences between interfering waves and effective waves in different dimensions to achieve suppression. One-dimensional filtering is usually based on frequency domain processing and is suitable for single-channel signals, while two-dimensional filtering is extended to multi-channel data, utilizing spatial information such as apparent velocity. Combining the two can improve the targeting and effectiveness of denoising.
[0061] One-dimensional filtering methods primarily rely on frequency filtering. They convert a single-channel seismic signal from the time domain to the frequency domain using Fourier transform, identifying and attenuating interfering frequency bands within the frequency domain. For example, for signals containing periodic interference, the interfering frequency points (such as 50Hz power frequency interference) can be located. By designing a frequency domain mask to zero or attenuating the correlation coefficient, the signal can be reconstructed using inverse Fourier transform, thus preserving the effective wave components. This method is computationally simple and suitable for handling local noise, but it neglects the spatial correlation of the signal.
[0062] Two-dimensional filtering methods utilize techniques such as apparent velocity filtering to process multichannel seismic data. Based on the difference in apparent velocity (or spatial frequency) between interfering and effective waves, it maps the data to the frequency-wavenumber domain using a two-dimensional Fourier transform (2D-FFT), constructing an apparent velocity filter mask to retain components matching the apparent velocity of the effective wave and suppressing interference. For example, in pre-stack seismic data, linear interfering waves often have a specific apparent velocity direction. This direction can be identified using an automatic tracking algorithm, and then the interfering signal can be reconstructed and subtracted using small window singular value decomposition (SVD) to reduce residual interference.
[0063] Furthermore, the specific steps for data augmentation based on residual shrinkage networks to obtain enhanced seismic data are as follows:
[0064] S321. After converting the format of the seismic exploration data, perform gather extraction to extract valid seismic gather data;
[0065] S322. Label the noise type in the training data, inject synthetic noise into the clean seismic data, and generate training samples;
[0066] S323. Design a residual shrinkage network, which includes a residual module, a soft thresholding layer, and an attention mechanism. The residual module includes convolutional layers, batch normalization, and ReLU activation modules, supporting cross-layer feature propagation. A soft thresholding layer is inserted into the residual path, dynamically calculating the noise threshold for each sample through sub-networks to shrink irrelevant features to zero. The attention mechanism integrates the SENet module, adaptively weighting important feature channels to improve noise suppression performance.
[0067] S324. Train the residual shrinkage network using training samples, employ mean squared error loss, use high signal-to-noise ratio labeled seismic data as the true value, constrain the difference between the model output and the true value, and obtain the trained residual shrinkage network.
[0068] S325. Input the effective seismic gather data into the trained residual shrinkage network and output enhanced seismic data.
[0069] The data augmentation process also includes an enhancement effect evaluation step: calculating the signal-to-noise ratio (SNR) and root mean square error (RMSE) of the seismic exploration data before and after augmentation. The requirements are that the SNR is improved by ≥15dB and the RMSE is ≤5%. If these requirements are not met, the residual shrinkage network parameters are adjusted and the data augmentation is performed again.
[0070] Furthermore, in S4, the enhanced seismic data is subjected to three-dimensional seismic interpretation, and the output includes three-dimensional seismic interpretation data containing structural distribution, lithological interfaces and ore layer parameters; the ore layer parameters include at least the ore layer burial depth, thickness and wave impedance.
[0071] Furthermore, in step S5, borehole locations are designed based on the 3D seismic interpretation data. The accuracy of the 3D seismic interpretation data is verified using borehole measurement data. If the interpretation error exceeds a preset threshold, the 3D seismic interpretation model is corrected based on the borehole measurement data. Steps S4-S5 are repeated until the interpretation error meets the requirements, thus completing the 3D seismic exploration. The specific method for correcting the 3D seismic interpretation model is as follows: the borehole-measured ore layer depth and thickness data are substituted into the 3D seismic interpretation model to optimize the model's velocity parameters and reflection layer calibration coefficients. Based on the optimized model, the 3D seismic interpretation is re-executed.
[0072] On the other hand, a data-augmented seismic exploration system is provided, such as... Figure 2 As shown, it includes a location determination module, a seismic exploration module, a data augmentation module, an interpretation module, and a verification module; among which,
[0073] The location determination module is used to acquire geological exploration data of the target area, analyze and determine the favorable mineralization locations in the target area;
[0074] The seismic exploration module is used to set up geophones or shot points in favorable mineralization locations to collect seismic exploration data.
[0075] The data augmentation module is used to augment the acquired seismic exploration data based on a residual shrinkage network to obtain enhanced seismic data.
[0076] The interpretation module is used to interpret the enhanced seismic data and output three-dimensional seismic interpretation data;
[0077] The verification module is used to verify the accuracy of the three-dimensional seismic interpretation data by designing borehole locations, and to make corrections to complete the three-dimensional seismic exploration.
[0078] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.
Claims
1. A data-augmented seismic exploration method, characterized in that, The specific steps include the following: Obtain geological exploration data of the target area, analyze and determine the favorable mineralization locations in the target area; Seismic exploration data are collected by setting up geophones or shot points in favorable mineralization locations. The acquired seismic exploration data is augmented using a residual shrinkage network to obtain enhanced seismic data; Interpret the enhanced seismic data and output three-dimensional seismic interpretation data; The accuracy of the 3D seismic interpretation data was verified and corrected by designing borehole locations, thus completing the 3D seismic exploration.
2. The data-augmented seismic exploration method according to claim 1, characterized in that, The specific steps for collecting seismic exploration data by arranging geophones or shot points in favorable mineralization locations are as follows: The grid information is determined based on the construction area information, and the grid information includes grid spacing values and grid shape parameters; Based on the grid information, detector points or shot points are randomly deployed within each grid. Based on the type of ore deposit in the favorable mineralization location, matching observation system parameters are designed. The observation system parameters include channel spacing, shot spacing, coverage times, and receiver array length. Different ore deposit types correspond to different parameter combinations. The artificial seismic source located at the grid node is activated, and the actual detection data of all the aforementioned receiver points are obtained based on the observation system parameters.
3. The data-augmented seismic exploration method according to claim 2, characterized in that, The excitation mode of the artificial seismic source is determined through experiments based on geological conditions.
4. The data-augmented seismic exploration method according to claim 2, characterized in that, It also includes correcting the receiver points or shot points. Specifically, the method is as follows: using the observation system parameters and seismic trace data file as initial conditions, scanning the observation system graph and seismic trace graph, judging and determining the shot points or receiver points that need to be corrected based on the results of linear dynamic correction, and gradually correcting the position coordinates of the shot points or receiver points that need to be corrected until the linear dynamic correction accuracy meets the requirements, ending the correction process, and outputting the corrected and modified observation system data file.
5. The data-augmented seismic exploration method according to claim 1, characterized in that, The collected seismic exploration data is first preprocessed and then augmented based on a residual shrinkage network.
6. The data-augmented seismic exploration method according to claim 5, characterized in that, The steps for preprocessing the seismic exploration data are as follows: To address the various noises present in the seismic exploration data, a combination of pre-stack one-dimensional filtering and two-dimensional filtering is used to suppress interference and improve the signal-to-noise ratio. The surface-consistent deconvolution technique is used to correct the seismic wavelet, eliminating the influence of excitation and reception on the seismic wavelet; Based on the residual static correction, predictive deconvolution is performed to improve the longitudinal resolution.
7. The data-augmented seismic exploration method according to claim 1, characterized in that, The specific steps for data augmentation based on residual shrinkage networks to obtain augmented seismic data are as follows: After converting the format of the seismic exploration data, gather extraction is performed to extract valid seismic gather data; To label the noise type in the training data, synthetic noise is injected into clean seismic data to generate training samples; Design a residual shrinking network, which includes a residual module, a soft thresholding layer, and an attention mechanism; wherein, the residual module includes a convolutional layer, batch normalization, and ReLU activation module, supporting cross-layer feature propagation; a soft thresholding layer is inserted in the residual path, and the noise threshold of each sample is dynamically calculated through the sub-network to shrink irrelevant features to zero; The residual shrinkage network is trained using the training samples, with mean squared error loss, using high signal-to-noise ratio labeled seismic data as the ground truth, and constraining the difference between the model output and the ground truth to obtain the trained residual shrinkage network. The effective seismic gather data is input into the trained residual shrinkage network, which outputs the enhanced seismic data.
8. A data-augmented seismic exploration system, characterized in that, It includes a location determination module, a seismic exploration module, a data augmentation module, an interpretation module, and a verification module; among which, The location determination module is used to acquire geological exploration data of the target area, analyze and determine the favorable mineralization locations in the target area; The seismic exploration module is used to set up geophones or shot points in favorable mineralization locations to collect seismic exploration data. The data augmentation module is used to augment the acquired seismic exploration data based on a residual shrinkage network to obtain enhanced seismic data. The interpretation module is used to interpret the enhanced seismic data and output three-dimensional seismic interpretation data; The verification module is used to verify the accuracy of the three-dimensional seismic interpretation data by designing borehole locations, and to make corrections to complete the three-dimensional seismic exploration.