A fault detection method and related device
Patent Information
- Application Number
- CN202510565520.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-30
- Publication Date
- 2026-10-30
AI Technical Summary
现有的智能断层自动识别方法并没有充分考虑这一点，对于断裂复杂、尺度各异的断层在识别精度不高
[0027]根据上述技术方案，利用经训练的第一模型对地震数据进行多尺度断层检测，以获得多个不同的断层特征数据；基于多个不同的断层特征数据进行模型训练，以获得用于断层预测融合的第二模型；利用第二模型对地震数据进行断层结果融合，以获得包含多尺度特征的断层融合体数据；基于断层融合体数据，确定断层检测结果。由此，本申请同时考虑断层检测的准确性和计算的效率，可以在保证计算效率的前提下，得到更准确的断层检测结果。特别是对复杂地质条件以及微小断层发育的地区有良好的效果。本申请与传统方法相比，提高了断层解释效率与准确率，为后续地质建模及储层预测提供可靠数据。
Smart Images
Figure CN123010048A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of oil and gas exploration technology, and more specifically, to a fault detection method, a fault detection device, an electronic device, and a storage medium. Background Technology
[0002] Fault and fracture identification is a crucial step in the structural interpretation workflow of seismic data, and an important part of structural interpretation and reservoir characterization. The accuracy of fault identification results significantly impacts the exploration and development of oil and gas reservoirs. Therefore, fault identification is an indispensable part of seismic data interpretation. However, accurately and efficiently identifying faults from seismic data has become a major challenge. Traditional manual fault interpretation methods rely heavily on the interpreter's experience and understanding of specific geological structures. Manual fault interpretation is not only time-consuming and labor-intensive, but also prone to overlooking small faults. The rise of artificial intelligence has led to a greater variety of automatic fault identification methods. However, due to different karst influences, faults and fractures exhibit varying scales and sizes. Existing intelligent automatic fault identification methods do not fully consider this, resulting in low accuracy for faults with complex fractures and varying scales.
[0003] Therefore, a new technical solution is urgently needed to solve the above-mentioned technical problems. Summary of the Invention
[0004] The summary section introduces a series of simplified concepts, which will be further explained in detail in the detailed description section. The summary section of this invention is not intended to limit the key features and essential technical features of the claimed technical solution, nor is it intended to determine the scope of protection of the claimed technical solution.
[0005] In a first aspect, the present invention proposes a fault detection method, comprising:
[0006] The trained first model is used to perform multi-scale fault detection on seismic data to obtain multiple different fault feature data.
[0007] The model is trained based on multiple different fault feature data to obtain a second model for fault prediction fusion.
[0008] The second model is used to fuse fault results from seismic data to obtain fault fusion data containing multi-scale features;
[0009] Based on the fault fusion data, the fault detection results are determined.
[0010] In some implementations, the second model includes a multi-level backbone network, which includes a first convolutional layer and a second convolutional layer. The first and second convolutional layers have different scales. Each level in the first convolutional layer is connected to any level in the second convolutional layer, and each level in the second convolutional layer is connected to any level in the first convolutional layer.
[0011] In some implementations, the second model further includes a max pooling layer, a hidden layer, and an output layer. The scale of the max pooling layer is the same as that of the second convolutional layer. Each level in the max pooling layer is connected to any level of the second convolutional layer. Each level in the second convolutional layer is connected to any level of the max pooling layer. Each level in the max pooling layer is connected to any level of the hidden layer. The hidden layer is connected to the output layer.
[0012] In some implementations, the first model includes an attention mechanism module for extracting fault features and analyzing those features.
[0013] In some implementations, the method further includes:
[0014] Classify fault feature data;
[0015] Based on the classification results, the tomographic fusion results are determined.
[0016] In some implementations, the fault detection results are determined based on fault fusion body data, including:
[0017] The fault fusion data is input into the third model, and the output feature compression result is used as the fault detection result. The third model includes multiple serially connected long short-term memory network modules.
[0018] In some implementations, the method further includes:
[0019] A multi-level fault feature extraction network was trained using 3D synthetic seismic data to obtain a trained first model.
[0020] Secondly, a fault detection device is also proposed, comprising:
[0021] The feature extraction unit is used to perform multi-scale fault detection on seismic data using a trained first model to obtain multiple different fault feature data.
[0022] The training unit is used to train the model based on multiple different fault feature data to obtain a second model for fault prediction fusion.
[0023] The fusion unit is used to fuse fault results of seismic data using the second model to obtain fault fusion body data containing multi-scale features.
[0024] The detection unit is used to determine the fault detection results based on the fault fusion body data.
[0025] Thirdly, an electronic device is also proposed, including a processor and a memory, wherein the memory stores computer program instructions, which are executed by the processor to perform the tomographic detection method described above.
[0026] Fourthly, a storage medium is also proposed, on which program instructions are stored. When the program instructions are run, they are used to execute the tomographic detection method described above.
[0027] According to the above technical solution, a trained first model is used to perform multi-scale fault detection on seismic data to obtain multiple different fault feature data. Based on these multiple different fault feature data, a second model is trained to obtain a fault prediction fusion model. The second model is then used to fuse the fault results of the seismic data to obtain fault fusion data containing multi-scale features. Based on the fault fusion data, the fault detection results are determined. Therefore, this application simultaneously considers the accuracy of fault detection and computational efficiency, achieving more accurate fault detection results while ensuring computational efficiency. It is particularly effective in areas with complex geological conditions and the development of micro-faults. Compared with traditional methods, this application improves the efficiency and accuracy of fault interpretation, providing reliable data for subsequent geological modeling and reservoir prediction.
[0028] The tomographic detection method of the present invention, and other advantages, objectives and features of the present invention will be apparent in part from the following description, and in part from the understanding of those skilled in the art through study and practice of the present invention. Attached Figure Description
[0029] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.
[0030] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0031] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of exemplary embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the scope of this application. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings:
[0032] Figure 1This is a schematic flowchart of a tomography detection method proposed in an embodiment of this application;
[0033] Figure 2 This is a schematic diagram of the network structure of a second model proposed in an embodiment of this application;
[0034] Figure 3 This is a schematic diagram of the model structure of an end-to-end convolutional neural network proposed in an embodiment of this application;
[0035] Figure 4 This is a schematic diagram of another end-to-end convolutional neural network model structure proposed in an embodiment of this application;
[0036] Figure 5 This is a schematic diagram of a seismic profile proposed in an embodiment of this application;
[0037] Figure 6 This is a schematic diagram of a tomographic detection result proposed in an embodiment of this application;
[0038] Figure 7 This is a schematic block diagram of a tomography detection device proposed in an embodiment of this application;
[0039] Figure 8 A schematic block diagram of an electronic device according to an embodiment of the present invention is shown. Detailed Implementation
[0040] To better understand the above-mentioned objectives, features, and advantages of this application, the solution of this application will be further described below. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.
[0041] The terms "first," "second," "third," "fourth," etc. (if present) in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus. The technical solutions of the embodiments of this application will now be clearly and completely described in conjunction with the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them.
[0042] Many specific details are set forth in the following description in order to provide a full understanding of this application, but this application may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only some embodiments of this application, and not all embodiments.
[0043] To improve recognition accuracy, different methods can be used to identify faults and cracks at different scales. Traditional manual picking and machine learning-based fault detection methods are being replaced by deep learning methods. Typical deep learning-based fault detection techniques are based on an end-to-end approach using the U-Net neural network. This method uses 3D forward modeling data as training samples to train the network model, and then directly applies the trained model to predict fault data. This approach achieves good results with real-world data that has a high signal-to-noise ratio and is relatively simple in construction, but it performs poorly for identifying faults with low signal-to-noise ratios and small faults.
[0044] To address the aforementioned technical problems, a fault detection method is proposed according to the first aspect of this application. Figure 1 This is a schematic flowchart illustrating a tomographic detection method proposed in an embodiment of this application. For example, as shown... Figure 1 As shown, the method may include the following steps:
[0045] Step S110: Use the trained first model to perform multi-scale fault detection on the seismic data to obtain multiple different fault feature data.
[0046] For example, in some implementations, a multi-level fault feature extraction network can be trained using 3D synthetic seismic data to obtain a deep learning model required for 3D fault prediction, i.e., the trained first model. The 3D synthetic seismic data can be obtained through seismic data acquisition technology, followed by data processing. Generally, 3D fault identification based on convolutional neural networks often uses multiple convolutional layers to extract features from seismic data. End-to-end networks can concatenate different convolutional layers, using connection layers to link high- and low-level features, but this results in significant accuracy loss. Traditional convolutional neural networks have two main problems: first, how to effectively acquire and utilize the spatial information of feature maps at different scales to enrich the fault feature space; and second, channel attention or spatial attention can only effectively capture local information and cannot establish long-term dependencies to improve prediction accuracy. In this application, based on channel attention, the idea of multi-scale local feature extraction and fusion is introduced, and a "quasi-pyramid" segmentation attention module is used to extract and analyze fault feature data. For example, the first model includes an attention mechanism module for extracting and analyzing fault features. Specifically, the attention mechanism module can replace the 3×3 convolution operator in a conventional convolutional neural network to obtain a new convolutional module. This module can thus be used as a "plug-and-play" module, significantly improving the performance of the convolutional neural network, effectively extracting finer-grained multi-scale spatial information, and establishing longer-distance channel dependencies. For example, by inputting seismic data into the trained first model, multi-scale and multi-level fault detection can be performed, thereby outputting multiple different fault feature data.
[0047] Step S120: Train the model based on multiple different fault feature data to obtain a second model for fault prediction fusion.
[0048] For example, after obtaining multiple different fault feature data, a multi-scale, multi-level classification and fusion network can be trained using seismic data and multiple different fault feature data to obtain a second model, which is used for multi-scale fault prediction fusion.
[0049] Step S130: Use the second model to fuse the fault results of the seismic data to obtain fault fusion data containing multi-scale features.
[0050] For example, a pre-trained multi-scale, multi-level classification and fusion network, i.e., the second model, is used to fuse fault results from seismic data, thereby obtaining fault fusion data containing multi-scale, multi-level features. In some implementations, the second model includes a multi-level backbone network, which includes a first convolutional layer and a second convolutional layer. The first and second convolutional layers have different scales, and each level in the first convolutional layer is connected to any level in the second convolutional layer, and each level in the second convolutional layer is connected to any level in the first convolutional layer. Figure 2 This is a schematic diagram of a network structure for a second model proposed in an embodiment of this application. For example, as shown... Figure 2 As shown, the second model is a network structure that connects large-scale features and small-scale features across levels. Figure 2 The lines shown can represent cross-level connections, including the extension of large-scale features to small-scale features and the compatibility of small-scale features to large-scale features. Figure 2 The first convolutional layer shown is a 5×5 convolutional layer, and the second convolutional layer is a 3×3 convolutional layer. Each level in the first convolutional layer is connected to every level in the second convolutional layer. For example, the first level in the first convolutional layer is connected to all levels in the second convolutional layer, the second level in the first convolutional layer is also connected to all levels in the second convolutional layer, and so on. See [link to relevant documentation]. Figure 2 We can determine the connection relationships between each level in the first and second convolutional layers.
[0051] This allows for better preservation of feature information and improved resolution of prediction results. Multiple networks with varying depths, levels of resolution, and features can be used to enhance prediction accuracy. It should be noted that network depth and size can be set based on experience or actual test data. Utilizing a second model with the aforementioned structure can accommodate different responses from both large-scale and small-scale faults, significantly improving the model's prediction accuracy.
[0052] In some implementations, such as Figure 2 As shown, the second model also includes a max-pooling layer, hidden layers, and an output layer. The max-pooling layer has a scale of 3×3, the same as the scale of the second convolutional layer. Each level in the max-pooling layer is connected to any level in the second convolutional layer; specifically, each level in the max-pooling layer is connected to every level in the second convolutional layer. For example, the first level in the max-pooling layer is connected to all levels in the second convolutional layer, the second level in the max-pooling layer is also connected to all levels in the second convolutional layer, and so on. See [link to documentation]. Figure 2 We can determine the connection relationships between the max-pooling layer and each level in the second convolutional layer. Similarly, each level in the second convolutional layer is connected to any level of the max-pooling layer. See also Figure 2Each level in the second convolutional layer is connected to each level in the max pooling layer. For example, the first level in the second convolutional layer is connected to all levels in the max pooling layer, the second level in the second convolutional layer is also connected to all levels in the max pooling layer, and so on. See [link to relevant documentation]. Figure 2 We can determine the connection relationships between the second convolutional layer and each level in the max-pooling layer. Each level in the max-pooling layer is connected to any level in the hidden layer, and each level in the max-pooling layer is connected to every level in the hidden layer. For example, the first level in the max-pooling layer is connected to all levels in the hidden layer, the second level in the max-pooling layer is also connected to all levels in the hidden layer, and so on. See [link to documentation]. Figure 2 We can determine the connection relationships between the max pooling layer and each level in the hidden layer. The hidden layer is connected to the output layer.
[0053] In some implementations, the method further includes: classifying the fault feature data; and determining the fault fusion result based on the classification result.
[0054] For example, as described above, the first model can calculate multiple fault feature data. Small-scale fault features have high spatial resolution and a good response to small fractures, but poor analytical ability for large-scale fault features, resulting in poor lateral and longitudinal continuity of the fault features. Large-scale fault features, on the other hand, have strong lateral and longitudinal continuity but low spatial resolution. For example, these multiple fault feature data can be input into a clustering network to define the location of the faults. The clustering network can segment the faults into different categories and reduce the impact of noise and outliers. For example, network recognition can be performed on multiple data volumes to extend the high spatial resolution and continuity features of a single result. The multi-scale fault feature information classification network includes a multi-input single-output end-to-end convolutional neural network. Figure 3 This is a schematic diagram of an end-to-end convolutional neural network model structure proposed in an embodiment of this application. Multiple tomographic data features can be used as input to this neural network. Through the end-to-end network, a mapping structure can be obtained, ultimately outputting the tomographic fusion result, i.e., the tomographic fusion body data.
[0055] Step S140: Determine the fault detection results based on the fault fusion body data.
[0056] In some implementations, fault fusion data can be input into a third model to output feature compression results as fault detection results. Figure 4 This is a schematic diagram of another end-to-end convolutional neural network model structure proposed in an embodiment of this application. For example, as shown... Figure 4As shown, the third model can include an end-to-end convolutional neural network, which includes multiple serially connected Long Short-Term Memory (LSTM) network modules. The final squeezing result can be obtained through this end-to-end convolutional neural network. Figure 5 This is a schematic diagram of a seismic profile proposed in an embodiment of this application. Since LSTM computation is slow and requires a large amount of space, seismic data can be segmented into two-dimensional profile data along the inline direction for processing and prediction. Then, the refined seismic data is stitched together with the output of a high-precision two-dimensional network to obtain a three-dimensional result. The above process is then repeated along the crossline direction to obtain the final feature compression result. Thus, by utilizing fault attribute enhancement features, fault fusion data is fused to obtain the final fault detection result. Figure 6 This is a schematic diagram of a tomographic detection result proposed in an embodiment of this application.
[0057] It should be noted that this application is based on a local feature extraction neural network, which extracts fault features from 3D seismic data and predicts fault attributes through a multi-scale quasi-pyramid convolutional network, ultimately obtaining a multi-level fused fault data volume. A fault feature discrimination network is introduced to enhance the fault feature information, thereby obtaining the fused fault volume data.
[0058] According to the above technical solution, a trained first model is used to perform multi-scale fault detection on seismic data to obtain multiple different fault feature data. Based on these multiple different fault feature data, a second model is trained to obtain a fault prediction fusion model. The second model is then used to fuse the fault results of the seismic data to obtain fault fusion data containing multi-scale features. Based on the fault fusion data, the fault detection results are determined. Therefore, this application simultaneously considers the accuracy of fault detection and computational efficiency, achieving more accurate fault detection results while ensuring computational efficiency. It is particularly effective in areas with complex geological conditions and the development of micro-faults. Compared with traditional methods, this application improves the efficiency and accuracy of fault interpretation, providing reliable data for subsequent geological modeling and reservoir prediction.
[0059] According to a second aspect of this application, a fault detection device is also proposed. Figure 7 This is a schematic block diagram of a tomography device according to an embodiment of this application. For example, as shown... Figure 7 As shown, the device 700 may include: a feature extraction unit 710, a training unit 720, a fusion unit 730, and a detection unit 740.
[0060] The feature extraction unit 710 is used to perform multi-scale fault detection on seismic data using a trained first model to obtain multiple different fault feature data.
[0061] Training unit 720 is used to train the model based on multiple different fault feature data to obtain a second model for fault prediction fusion.
[0062] Fusion unit 730 is used to fuse fault results of seismic data using the second model to obtain fault fusion body data containing multi-scale features;
[0063] The detection unit 740 is used to determine the fault detection results based on the fault fusion body data.
[0064] According to a third aspect of this application, an electronic device is also proposed. Figure 8 A schematic block diagram of an electronic device according to an embodiment of the present invention is shown. Figure 8 As shown, the electronic device 800 may include a processor 810 and a memory 820. The memory 820 stores computer program instructions, which are executed by the processor 810 to perform the tomographic detection method described above.
[0065] According to a fourth aspect of this application, a storage medium is also provided, on which program instructions are stored, which, when executed, perform the tomographic detection method described above. The storage medium may, for example, include a storage component of a tablet computer, a hard disk of a computer, a read-only memory (ROM), an erasable programmable read-only memory (EPROM), a portable compact disc read-only memory (CD-ROM), a USB memory, or any combination of the above storage media. The computer-readable storage medium may be any combination of one or more computer-readable storage media.
[0066] Those skilled in the art can understand the specific details and beneficial effects of the tomographic detection device, electronic equipment, and storage medium by reading the above description of the tomographic detection method, and will not be repeated here for the sake of brevity.
[0067] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and / or device can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.
[0068] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0069] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.
[0070] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0071] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.
Claims
1. A method for detecting faults, characterized in that, include: The trained first model is used to perform multi-scale fault detection on seismic data to obtain multiple different fault feature data. Model training is performed based on the multiple different fault feature data to obtain a second model for fault prediction fusion. The second model is used to fuse the fault results of the earthquake data to obtain fault fusion data containing multi-scale features; Based on the fault fusion data, the fault detection results are determined.
2. The fault detection method according to claim 1, characterized in that, The second model includes a multi-level backbone network, which includes a first convolutional layer and a second convolutional layer. The first convolutional layer and the second convolutional layer have different scales. Each level in the first convolutional layer is connected to any level in the second convolutional layer, and each level in the second convolutional layer is connected to any level in the first convolutional layer.
3. The fault detection method according to claim 1, characterized in that, The second model further includes a max pooling layer, a hidden layer, and an output layer. The scale of the max pooling layer is the same as that of the second convolutional layer. Each level in the max pooling layer is connected to any level of the second convolutional layer. Each level in the second convolutional layer is connected to any level of the max pooling layer. Each level in the max pooling layer is connected to any level of the hidden layer. The hidden layer is connected to the output layer.
4. The fault detection method according to claim 1, characterized in that, The first model includes an attention mechanism module for extracting fault features and analyzing the fault features.
5. The fault detection method according to claim 1, characterized in that, The method further includes: The fault feature data are classified; Based on the classification results, the tomographic fusion results are determined.
6. The fault detection method according to claim 1, characterized in that, The determination of the fault detection result based on the fault fusion body data includes: The fault fusion data is input into a third model, and the output feature compression result is used as the fault detection result. The third model includes multiple serially connected long short-term memory network modules.
7. The fault detection method according to claim 1, characterized in that, The method further includes: The multi-level fault feature extraction network is trained using three-dimensional synthetic seismic data to obtain the trained first model.
8. A fault detection device, characterized in that, include: The feature extraction unit is used to perform multi-scale fault detection on seismic data using a trained first model to obtain multiple different fault feature data. A training unit is used to train a model based on the multiple different fault feature data to obtain a second model for fault prediction fusion. The fusion unit is used to fuse the fault results of the seismic data using the second model to obtain fault fusion body data containing multi-scale features. The detection unit is used to determine the fault detection result based on the fault fusion body data.
9. An electronic device, characterized in that, It includes a processor and a memory, wherein the memory stores computer program instructions, which, when executed by the processor, are used to perform the tomographic detection method as described in any one of claims 1 to 7.
10. A storage medium storing program instructions that, when executed, perform the tomographic detection method as claimed in any one of claims 1 to 7.