Three-dimensional fault identification method, device, and medium

Abstract

The application provides a three-dimensional fault identification method, device and medium. The method comprises the following steps: according to a preset offset distance requirement, determining a first seismic trace in a first offset distance domain and a second seismic trace in a second offset distance domain in a first common point trace set, and a third seismic trace in the first offset distance domain and a fourth seismic trace in the second offset distance domain in a second common point trace set; respectively performing superposition processing on the first seismic trace and the third seismic trace, and the second seismic trace and the fourth seismic trace, to obtain a first offset distance domain data set and a second offset distance domain data set; inputting the first offset distance domain data set and the second offset distance domain data set into a fault identification model respectively, to obtain a first fault identification result and a second fault identification result; and determining a target fault identification result of a to-be-identified region according to the first fault identification result and the second fault identification result. The method provided by the application improves the fault identification precision and avoids missing part of the fault.

Description

Technical Field

[0001] This application relates to the field of seismic data interpretation technology, and in particular to a three-dimensional fault identification method, device and medium. Background Technology

[0002] Traditional earthquake fault identification mainly relies on earthquake interpreters to manually interpret geological information such as fault pattern distribution and tectonic formation mechanism. The interpretation results are often constrained by factors such as noise in earthquake data, complexity of geological structure and resolution of earthquake imaging, which is time-consuming, labor-intensive and inevitably subject to human interpretation.

[0003] Existing technologies use artificial intelligence to identify and process post-stack seismic data, thereby improving the accuracy and efficiency of fault identification.

[0004] However, existing fault identification technologies suffer from insufficient accuracy, leading to the loss of some faults. Summary of the Invention

[0005] This application provides a three-dimensional fault identification method, device, and medium to solve the problem of partial fault loss in the prior art.

[0006] In a first aspect, this application provides a tomographic identification method, comprising:

[0007] Obtain the first common center point gather and the second common center point gather of the region to be identified, wherein the offset domain of the first common center point gather and the offset domain of the second common center point gather are the same.

[0008] According to the preset offset requirements, the first seismic trace in the first common center point trace set and the second seismic trace in the second offset point trace set are determined, as well as the third seismic trace in the second common center point trace set and the fourth seismic trace in the second offset point trace set, wherein the first offset point trace set and the second offset point trace set are different.

[0009] The first and third seismic traces, as well as the second and fourth seismic traces, are overlaid to obtain the first offset domain dataset and the second offset domain dataset.

[0010] The first offset domain dataset and the second offset domain dataset are respectively input into the fault identification model to obtain the first fault identification result and the second fault identification result. The fault identification model is trained based on the three-dimensional seismic data samples and the fault label data corresponding to the three-dimensional seismic data samples. The first fault identification result represents the identification probability that the three-dimensional point in the area to be identified in the first offset domain is in the fault, and the second fault identification result represents the identification probability that the three-dimensional point in the area to be identified in the second offset domain is in the fault.

[0011] Based on the results of the first and second fault identifications, the target fault identification results for the area to be identified are determined.

[0012] In this application, before inputting the first offset domain dataset and the second offset domain dataset into the fault identification model to obtain the first fault identification result and the second fault identification result, the method further includes:

[0013] Acquire 3D seismic data samples;

[0014] Three-dimensional seismic data samples are input into the model to be trained for fault identification processing to obtain training identification results;

[0015] Based on the training recognition results and the tomographic label data of the training region, the balanced cross-entropy of the model to be trained is determined, wherein the balanced cross-entropy satisfies:

[0016]

[0017] Where Y is the model to be trained, x i For the i-th 3D seismic data sample, y i Let be the fault label data corresponding to the i-th 3D seismic data sample, m be the parameters of the model to be trained, N be the number of 3D seismic data samples, and β be the balance coefficient.

[0018] Based on the balanced cross-entropy, the model to be trained is adjusted to obtain the tomography identification model.

[0019] In this application, obtaining three-dimensional seismic data samples includes:

[0020] Determine the fault label data;

[0021] Based on fault label data, convolutional forward modeling is performed to generate three-dimensional synthetic seismic data;

[0022] Based on the three-dimensional synthetic seismic data, a three-dimensional seismic data sample is obtained.

[0023] In this application, the target fault identification result of the region to be identified is determined based on the first fault identification result and the second fault identification result, including:

[0024] Based on the first fault identification result and the second fault identification result, determine the first identification probability of the target three-dimensional point being located in the fault in the first fault identification result and the second identification probability of the target three-dimensional point being located in the fault in the second fault identification result. The target three-dimensional point is a three-dimensional point in the area to be identified.

[0025] The target recognition probability of the target 3D point is determined based on the comparison between the first recognition probability and the second recognition probability.

[0026] The target recognition probabilities of each target 3D point are summarized and processed to obtain the target tomographic recognition results of the area to be identified.

[0027] In this application, selecting the target recognition probability for a target three-dimensional point from either the first tomographic recognition result or the second tomographic recognition result based on the comparison result between the first recognition probability and the second recognition probability includes:

[0028] If the first recognition probability is greater than the second recognition probability, then the first recognition probability shall be used as the target recognition probability of the target 3D point.

[0029] If the first recognition probability is less than the second recognition probability, then the second recognition probability shall be used as the target recognition probability of the target three-dimensional point.

[0030] If the first recognition probability is equal to the second recognition probability, then the first recognition probability or the second recognition probability is taken as the target recognition probability of the target three-dimensional point.

[0031] In this application, after determining the target fault identification result of the area to be identified based on the first fault identification result and the second fault identification result, the method further includes:

[0032] Extract planar and profile data from the fault identification results;

[0033] Based on the preset requirements for joint evaluation of planar and profile data, the planar data and profile data are evaluated to obtain the first evaluation result;

[0034] Present the results of the first assessment.

[0035] In this application, after determining the target fault identification result of the area to be identified based on the first fault identification result and the second fault identification result, the method further includes:

[0036] Acquire exploration data for the area to be identified, including at least one of well logging data and drilling data;

[0037] The fault identification results are evaluated based on the exploration data to obtain a second evaluation result;

[0038] Present the results of the second assessment.

[0039] Secondly, this application provides a three-dimensional tomographic identification device, comprising:

[0040] The acquisition module is used to acquire the first common center point gather and the second common center point gather of the region to be identified, wherein the offset range of the first common center point gather and the offset range of the second common center point gather are the same.

[0041] The first determining module is used to determine, according to the preset offset distance requirements, the first seismic trace in the first common center point trace set is located within the first offset distance domain and the second seismic trace is located within the second offset distance domain, and the third seismic trace in the second common center point trace set is located within the first offset distance domain and the fourth seismic trace is located within the second offset distance domain, wherein the first offset distance domain and the second offset distance domain are different.

[0042] The overlay module is used to overlay the first and third seismic traces, as well as the second and fourth seismic traces, to obtain the first offset domain dataset and the second offset domain dataset.

[0043] The identification module is used to input the first offset domain dataset and the second offset domain dataset into the fault identification model respectively to obtain the first fault identification result and the second fault identification result. The fault identification model is trained based on the three-dimensional seismic data samples and the fault label data corresponding to the three-dimensional seismic data samples. The first fault identification result represents the identification probability that the three-dimensional point in the area to be identified in the first offset domain is in the fault, and the second fault identification result represents the identification probability that the three-dimensional point in the area to be identified in the second offset domain is in the fault.

[0044] The second determining module is used to determine the target fault identification result of the area to be identified based on the first fault identification result and the second fault identification result.

[0045] Thirdly, this application provides an electronic device, including: a processor, and a memory communicatively connected to the processor;

[0046] The memory stores the instructions that the computer executes;

[0047] The processor executes computer execution instructions stored in memory to implement the method provided in this application.

[0048] Fourthly, this application provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the method provided in this application.

[0049] Fifthly, this application provides a computer program product, which includes a computer program that, when executed by a processor, is used to implement the method provided in this application.

[0050] The three-dimensional fault identification method, device, and medium provided in this application extract seismic traces from each common center point gather of the area to be identified according to the offset requirements, and then superimpose the seismic traces from different offset domains to obtain multiple offset domain datasets. Then, each offset domain dataset is input into the fault identification model for fault identification processing. This can obtain fault identification results corresponding to different offset domains. Finally, the target fault identification result of the area to be identified is obtained based on the fault identification results of different offset domains, thereby improving the fault identification accuracy of the area to be identified and avoiding the loss of some faults. Attached Figure Description

[0051] 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.

[0052] Figure 1 A schematic diagram of a three-dimensional tomographic identification scenario provided for an embodiment of this application;

[0053] Figure 2 A flowchart illustrating a three-dimensional tomographic identification method provided in this application embodiment. Figure 1 ;

[0054] Figure 3 A schematic diagram of a common center point gather provided in an embodiment of this application;

[0055] Figure 4 A schematic diagram of the tomographic label data provided in the embodiments of this application;

[0056] Figure 5 A schematic diagram of a three-dimensional seismic data sample provided in an embodiment of this application;

[0057] Figure 6 A cross-sectional schematic diagram of the fault identification results provided in an embodiment of this application;

[0058] Figure 7 A flowchart illustrating a three-dimensional tomographic identification method provided in this application embodiment. Figure 2 ;

[0059] Figure 8 This is a schematic diagram of the structure of a three-dimensional tomographic identification device provided in an embodiment of this application;

[0060] Figure 9 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0061] The accompanying drawings have illustrated specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to specific embodiments. Detailed Implementation

[0062] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0063] To clearly understand the technical solution of this application, the solutions of the prior art will be described in detail first.

[0064] Traditional earthquake fault identification mainly relies on earthquake interpreters to manually interpret geological information such as fault pattern distribution and tectonic formation mechanism. The interpretation results are often constrained by factors such as noise in earthquake data, complexity of geological structure and resolution of earthquake imaging, which is time-consuming, labor-intensive and inevitably subject to human interpretation.

[0065] Existing technologies use artificial intelligence to identify and process post-stack seismic data, thereby improving the accuracy and efficiency of fault identification.

[0066] However, most existing fault identification results are based on post-stack 3D seismic data, without taking into account the rich fault features contained in pre-stack seismic data. This results in insufficient identification accuracy and the loss of some faults.

[0067] To address the aforementioned issue of partial fault loss, the inventors discovered in their research that seismic traces can be aggregated according to the offset domain for different common center point gathers, forming different offset domain datasets. Fault identification can then be performed on these offset datasets to obtain identification results corresponding to different offset datasets. Finally, the identification results can be summarized to obtain the fault identification results.

[0068] Among them, a fault can refer to a structure in which the earth's crust fractures under stress, resulting in significant relative displacement of rock blocks on both sides of the fracture surface.

[0069] Common center point gathers can refer to a collection of seismic traces that share the same center point.

[0070] A seismic trace can refer to the seismic wave energy received by a geophone from a shot point in a certain three-dimensional region. Here, the shot point can refer to the point where the seismic wave is generated. The seismic wave energy can be detected by a geophone.

[0071] The center point of a seismic trace can refer to the center point between the receiver point and the shot point.

[0072] Offset distance can refer to the distance between the receiver point and the shot point.

[0073] The offset range can refer to the range of offsets, which can be a continuous interval or a discrete interval. In some implementations, the offset range can also be called the offset range group.

[0074] The application scenarios of the tomography identification method provided in the embodiments of this application are described below.

[0075] Figure 1 This is a schematic diagram of a three-dimensional tomographic recognition scenario provided in an embodiment of this application, such as... Figure 1 As shown, the execution subject of this fault identification method can be a server, which can be a device such as a mobile phone, tablet, or computer. This embodiment does not impose any particular restrictions on the implementation method of the execution subject, as long as the execution subject can obtain the first common center point gather and the second common center point gather of the area to be identified, wherein the offset domains of the first and second common center point gathers are the same; according to the preset offset requirements, the first seismic trace within the first offset domain and the second seismic trace within the second offset domain in the first common center point gather are determined, as well as the third and fourth seismic traces within the first offset domain and the second offset domain in the second common center point gather, wherein the first and second offset domains are different; the first and third seismic traces, and the second and fourth seismic traces are then superimposed respectively. The process involves obtaining a first offset domain dataset and a second offset domain dataset. These datasets are then input into a fault identification model to obtain a first fault identification result and a second fault identification result. The fault identification model is trained based on 3D seismic data samples and corresponding fault label data. The first fault identification result represents the probability that a 3D point in the area to be identified within the first offset domain is located within a fault, and the second fault identification result represents the probability that a 3D point in the area to be identified within the second offset domain is located within a fault. Based on the first and second fault identification results, the target fault identification result for the area to be identified can be determined.

[0076] Figure 2 This is a flowchart illustrating a three-dimensional tomographic identification method provided in an embodiment of this application, as shown below. Figure 2 As shown, the method includes:

[0077] S201. Obtain the first common center point gather and the second common center point gather of the region to be identified, wherein the offset domain of the first common center point gather and the offset domain of the second common center point gather are the same.

[0078] The area to be identified can refer to the area that needs to be identified by tomography.

[0079] The center point of the first common center point set and the center point of the second common center point set are different.

[0080] In some implementations, in order to make the extraction of seismic traces more uniform, the number of seismic traces in the first common center point gather is the same as the number of seismic traces in the second common center point gather, and the offset domain of the first common center point gather is the same as the offset domain of the second common center point gather.

[0081] For example, Figure 3 This is a schematic diagram of a common center point gather provided in an embodiment of this application. The vertical axis of the diagram represents time depth, and the horizontal axis represents the main survey line or connecting survey line, denoted by CDP. Figure 3 The dataset includes five common center point gathers, each containing 25 seismic traces. Within the same common center point gather, each trace has a different offset, and the traces are arranged according to their offset. The offset range is the same for all common center point gathers (refer to the horizontal axis). The main survey lines or connecting survey lines differ between common center point gathers, indicating that each gather was measured at different center point locations. For example, a center point between 220 and 222 was selected, and measurements were taken at 25 offset receiver and shot points, resulting in 25 seismic traces.

[0082] Since fault identification is mainly obtained by analyzing the changes in seismic wave energy, and the pre-stack raw data obtained from seismic trace measurements are common midpoint gathers, obtaining the first and second common midpoint gathers of the area to be identified is faster and more direct, which can improve the efficiency of fault identification.

[0083] S202. Based on the preset offset distance requirements, determine the first seismic trace in the first common center point trace set that is within the first offset distance domain and the second seismic trace in the second offset distance domain, as well as the third seismic trace in the second common center point trace set that is within the first offset distance domain and the fourth seismic trace in the second offset distance domain, wherein the first offset distance domain and the second offset distance domain are different.

[0084] The preset offset distance requirement can refer to the preset requirement for the size of each offset distance domain, thereby determining the first offset distance domain and the second offset distance domain.

[0085] The first offset domain and the second offset domain can refer to two offset domains divided according to the offset requirements, or they can be two offset domains arbitrarily selected from multiple divided offset domains.

[0086] In this embodiment, the method for dividing and extracting seismic traces in the common center point trace set represented by the offset requirement according to different offsets can be based on the range between the maximum and minimum offsets of the seismic traces in the entire common center point trace set. Alternatively, in order to make the data distribution in each trace set more uniform, the offset can be divided and extracted using a box-plot algorithm, that is, by determining the number of upper-middle quartiles and lower quartiles in the offset for division and extraction.

[0087] Since the seismic wave energy at different offsets is measured from different locations in the area to be identified, by analyzing the seismic wave energy at different offsets, compared with the existing technology of superimposing seismic data at multiple offsets for fault identification, this embodiment can obtain the fault situation of the area to be identified from multiple angles, thereby improving the fault identification effect of the area to be identified.

[0088] S203. The first and third seismic traces, as well as the second and fourth seismic traces, are superimposed to obtain the first offset domain dataset and the second offset domain dataset.

[0089] Among them, overlay processing can refer to placing multiple seismic traces in the same three-dimensional coordinate system according to the three-dimensional coordinates of each seismic trace.

[0090] The first offset domain dataset can refer to the result after superimposing the first and third seismic traces; the second offset domain dataset can refer to the result after superimposing the second and fourth seismic traces.

[0091] In some implementations, before the superposition process, the first, second, third, and fourth seismic channels can be screened according to the seismic channel requirements, and seismic channels that do not meet the requirements can be removed. The seismic channel requirements can be the threshold values ​​that the seismic amplitude in the seismic channel needs to reach.

[0092] S204. Input the first offset domain dataset and the second offset domain dataset into the fault identification model respectively to obtain the first fault identification result and the second fault identification result. The fault identification model is trained based on the three-dimensional seismic data samples and the fault label data corresponding to the three-dimensional seismic data samples. The first fault identification result represents the identification probability that the three-dimensional point in the area to be identified in the first offset domain is in the fault, and the second fault identification result represents the identification probability that the three-dimensional point in the area to be identified in the second offset domain is in the fault.

[0093] Among them, the fault identification model can refer to the model used for fault identification based on seismic trace datasets.

[0094] Three-dimensional seismic data samples can refer to the data used to train the model to be trained. Three-dimensional seismic data samples can be obtained from the three-dimensional synthetic seismic data volume of convolutional forward modeling.

[0095] Three-dimensional synthetic seismic data volumes can be used to characterize seismic data in forward models. The dimensions of a three-dimensional synthetic seismic data volume include the main seismic line (Inline), the connecting seismic line (Crossline), and the time depth (Time / ms).

[0096] Fault label data can be three-dimensional fault data that includes fault location and fault structure.

[0097] For example, Figure 4 This is a schematic diagram of the tomographic label data provided in an embodiment of this application, wherein, Figure 4 The yellow areas in the diagram represent identified faults. Figure 5 A schematic diagram of a three-dimensional seismic data sample provided in an embodiment of this application, wherein, Figure 5 Different colors in the image represent the energy level of the earthquake signal. Figure 5 for Figure 4 This was obtained from forward modeling of the convolution.

[0098] The results of the first and second fault identifications are presented in the form of a three-dimensional image. The dimensions of the three-dimensional image include the main survey line, the connecting survey line, and the time depth. The three-dimensional image includes three-dimensional points and the identification probability of the three-dimensional points being in the fault. The combination of multiple three-dimensional points constitutes the image of the fault. The identification probability is presented in the form of different colors of the three-dimensional points or in the form of a 0-1 value.

[0099] Three-dimensional points can refer to pixels in the first and second tomographic identification results, or pixels in a three-dimensional image representing the region to be identified.

[0100] In this embodiment of the application, before inputting the first offset domain dataset and the second offset domain dataset into the tomography identification model to obtain the first tomography identification result and the second tomography identification result, the method further includes:

[0101] Acquire 3D seismic data samples;

[0102] Three-dimensional seismic data samples are input into the model to be trained for fault identification processing to obtain training identification results;

[0103] Based on the training recognition results and the tomographic label data of the training region, the balanced cross-entropy of the model to be trained is determined, wherein the balanced cross-entropy satisfies:

[0104]

[0105] Where Y is the model to be trained, xi For the i-th 3D seismic data sample, y i Let be the fault label data corresponding to the i-th 3D seismic data sample, m be the parameters of the model to be trained, N be the number of 3D seismic data samples, and β be the balance coefficient.

[0106] Based on the balanced cross-entropy, the model to be trained is adjusted to obtain the tomography identification model.

[0107] The model to be trained can be a neural network model.

[0108] For example, the neural network model used in this embodiment can adopt a three-dimensional multi-scale convolutional neural network architecture, mainly composed of an encoder, a decoder, and a fusion unit. The encoder function is mainly responsible for extracting seismic features of different levels and semantic meanings from the input data. The fusion unit function is responsible for fusing features from different levels to ensure that too much semantic information is not lost. The decoder function mainly reconstructs the extracted seismic features to restore the complete seismic data. In addition, a multi-scale convolutional architecture is used inside the neural network. This architecture can effectively extract fault features from the seismic data by using convolutional kernels of different sizes in parallel.

[0109] The balance coefficient refers to the coefficient used to control the weight of positive and negative samples in the total loss. The balance coefficient can be set to 0.75.

[0110] The expression for the learned tomographic identification model in this embodiment can be:

[0111] Y = 3DMSNet(X,m)

[0112] In this model, 3DMSNet is a three-dimensional multi-scale convolutional neural network in deep learning. X is the input offset domain dataset, m are the parameters of the neural network, and Y is the tomography model. By using the Adam (Adaptive Moment Estimation) optimization algorithm as the network parameter optimizer, the three-dimensional multi-scale convolutional neural network is trained and its parameters are optimized. The loss function of the model is iteratively solved to obtain the optimal model parameters and the final tomography model. The Adam algorithm is a combination of Momentum gradient descent and RMSprop (Root Mean Square Propagation) algorithm. It designs an adaptive learning rate for each parameter by calculating the first and second moment estimates of the gradient.

[0113] Balanced cross entropy is a performance metric used to calculate classification models. Traditional tomography models often use cross entropy loss functions, which, while achieving the desired overall results, have limitations due to the extremely low proportion of tomographic data. To alleviate the imbalance between positive and negative samples in tomography, this embodiment uses balanced cross entropy as the loss function for the model being trained, thereby improving the training effect.

[0114] In this embodiment of the application, obtaining three-dimensional seismic data samples includes:

[0115] Determine the fault label data;

[0116] Fault label data are subjected to convolutional forward modeling to generate three-dimensional synthetic seismic data;

[0117] Based on the three-dimensional synthetic seismic data, a three-dimensional seismic data sample is obtained.

[0118] Among them, the convolution forward modeling operation can refer to performing a convolution operation on a seismic wavelet (a known or hypothetical waveform) and a sequence of reflection coefficients (a sequence describing the reflection characteristics of underground rock strata) to simulate the propagation and reflection process of seismic waves underground.

[0119] Three-dimensional synthetic seismic data can refer to seismic data obtained through convolution forward modeling.

[0120] In this embodiment, since convolution forward modeling can simulate three-dimensional synthetic seismic data based on various types of fault label data, the three-dimensional seismic data samples obtained from the three-dimensional synthetic seismic data are richer and more diverse, which can improve the training effect of the model to be trained.

[0121] In this embodiment, convolutional forward modeling is used to make the data for training the model more multifaceted. Therefore, 3D synthetic seismic data is evaluated and screened to improve the model training effect. Specifically, this includes:

[0122] 1. Data integrity check:

[0123] Verify that the synthetic seismic record is complete and has no missing or damaged parts.

[0124] Ensure that all necessary parameters (such as reflection coefficient sequence, seismic wavelet, etc.) have been correctly calculated and applied in the convolution forward modeling process.

[0125] 2. Data quality assessment:

[0126] Evaluate the signal-to-noise ratio (SNR) of synthetic seismic records. A high SNR indicates a strong effective signal and a low noise level in the record, which is beneficial for model training.

[0127] Examine the resolution of the synthetic seismic record, i.e., whether it can clearly reflect the reflection characteristics of the subsurface rock strata. High-resolution records help the model identify the stratigraphic structure more accurately.

[0128] 3. Geological rationality verification:

[0129] The synthetic seismic record is compared with known geological models to verify its geological plausibility. For example, it is examined whether the reflection horizons in the synthetic seismic record correspond to the stratigraphic boundaries in the geological model.

[0130] Ensure that the reflection coefficient sequence in the synthetic seismic record is consistent with the physical properties (such as density and velocity) of the underground rock strata.

[0131] 4. Representative Screening:

[0132] Representative samples are selected from multiple synthetic seismic records for training. These representative samples should cover different geological conditions and stratigraphic structures to ensure the model can handle various scenarios.

[0133] In this embodiment, since the offset domain dataset contains a large amount of data, manual analysis is time-consuming and labor-intensive. Therefore, using a tomographic identification model to perform tomographic identification on the offset domain dataset can improve identification efficiency and accuracy. Secondly, by inputting offset domain datasets from different offset domains into the tomographic identification model, tomographic identification results of the region to be identified in different offset domains can be obtained, thereby improving the accuracy of tomographic identification.

[0134] S205. Based on the first fault identification result and the second fault identification result, determine the target fault identification result for the area to be identified.

[0135] The target fault identification result can be used to characterize the faults in the region to be identified. The presentation method of the target fault identification result is the same as that of the first fault identification result or the second fault identification result. The faults included in the target fault identification result are the combined result of the faults in the first fault identification result and the faults in the second fault identification result.

[0136] The method for determining the target fault identification results of the area to be identified can overlap and merge the first fault identification results and the second fault identification results in the same three-dimensional coordinate system. That is, when two data points appear at a certain three-dimensional point, the larger data point is used to cover the smaller data point, thereby preserving the identification details of the first and second fault identification results.

[0137] For example, Figure 6 This is a cross-sectional schematic diagram of the fault identification results provided in the embodiments of this application. In the diagram, colored lines represent different types of rock strata, black lines represent faults, and the shade of black lines represents the identification probability of the fault. Figure 6 a is a schematic cross-sectional view of the first fault identification result. Figure 6 b is a schematic cross-sectional view of the second fault identification results. Figure 6 c is a cross-sectional schematic diagram of the third fault identification results. As can be seen from the three diagrams, the identification probability of the point indicated by the arrow is different in each of the three diagrams. Figure 6 d is Figure 6 a, Figure 6 b and Figure 6 c. A schematic cross-sectional view of the summarized target fault identification results. Figure 6 e is a cross-sectional schematic diagram of the fault identification results of existing technology, which shows that... Figure 6 d compared to Figure 6 e has more and more diverse faults.

[0138] In this embodiment of the application, the method of summarizing the fault features of the first fault identification result and the second fault identification result to obtain the target fault identification result in the area to be identified not only preserves the identification details of each fault identification result, but also facilitates viewing and analysis.

[0139] This application provides a three-dimensional fault identification method. It extracts seismic traces from each common center point gather of the region to be identified according to offset requirements, then superimposes seismic traces from different offset domains to obtain multiple offset domain datasets. These datasets are then input into a fault identification model for fault identification processing. This yields fault identification results corresponding to different offset domains. Finally, the target fault identification result for the region to be identified is obtained based on the fault identification results from different offset domains, improving the fault identification accuracy of the region to be identified and avoiding the loss of some faults.

[0140] Figure 7 A flowchart illustrating a three-dimensional tomographic identification method provided in this application embodiment. Figure 2 ,like Figure 7 As shown, the method includes:

[0141] S701. Obtain the first common center point gather and the second common center point gather of the region to be identified, wherein the offset domain of the first common center point gather and the offset domain of the second common center point gather are the same.

[0142] S702. According to the preset offset distance requirements, determine the first seismic trace in the first common center point trace set that is within the first offset distance domain and the second seismic trace in the second offset distance domain, and the third seismic trace in the second common center point trace set that is within the first offset distance domain and the fourth seismic trace in the second offset distance domain, wherein the first offset distance domain and the second offset distance domain are different.

[0143] S703. The first and third seismic traces, as well as the second and fourth seismic traces, are superimposed to obtain the first offset domain dataset and the second offset domain dataset.

[0144] S704. Input the first offset domain dataset and the second offset domain dataset into the fault identification model respectively to obtain the first fault identification result and the second fault identification result. The fault identification model is trained based on the three-dimensional seismic data samples and the fault label data corresponding to the three-dimensional seismic data samples. The first fault identification result represents the identification probability that the three-dimensional point in the area to be identified in the first offset domain is in the fault. The second fault identification result represents the identification probability that the three-dimensional point in the area to be identified in the second offset domain is in the fault.

[0145] S705. Based on the first fault identification result and the second fault identification result, determine the first identification probability of the target three-dimensional point being located in the fault in the first fault identification result and the second identification probability of the target three-dimensional point being located in the fault in the second fault identification result. The target three-dimensional point is a three-dimensional point in the area to be identified.

[0146] The method for determining the first recognition probability is to determine the location point of the target three-dimensional point in the first fault recognition result, then determine the recognition probability of that location point, and finally determine the recognition probability of that location point as the first recognition probability. Similarly, the second recognition probability can be obtained.

[0147] S706. Determine the target recognition probability of the target three-dimensional point based on the comparison result between the first recognition probability and the second recognition probability.

[0148] The comparison result can refer to the comparison result of the magnitude of the first recognition probability and the second recognition probability.

[0149] In the embodiments of this application,

[0150] Based on the comparison between the first and second recognition probabilities, the target recognition probability for the target three-dimensional point is selected from either the first or second tomographic recognition result, including:

[0151] If the first recognition probability is greater than the second recognition probability, then the first recognition probability shall be used as the target recognition probability of the target 3D point.

[0152] If the first recognition probability is less than the second recognition probability, then the second recognition probability shall be used as the target recognition probability of the target three-dimensional point.

[0153] If the first recognition probability is equal to the second recognition probability, then the first recognition probability or the second recognition probability is taken as the target recognition probability of the target three-dimensional point.

[0154] Specifically, if the first recognition probability is greater than the second recognition probability, it means that the fault in which the target 3D point is located is more obvious in the first fault recognition result, so the first recognition probability is used as the target recognition probability of the target 3D point; if the first recognition probability is less than the second recognition probability, it means that the fault in which the target 3D point is located is more obvious in the second fault recognition result, so the second recognition probability is used as the target recognition probability of the target 3D point; if the first recognition probability is equal to the second recognition probability, it means that the fault in which the target 3D point is located has the same degree of obviousness in the first and second fault recognition results, so either the first or the second recognition probability can be used as the target recognition probability of the target 3D point. By comparing the first and second recognition probabilities, the recognition results of the target fault can be integrated, making the target recognition result more complete.

[0155] In some implementations, when the offset distance requirement is multiple offset distance domains and multiple fault identification results are obtained accordingly, the method for determining the target fault identification result of the area to be identified can also be to first determine the target three-dimensional point in the area to be identified, then determine the identification probability of the target three-dimensional point in the multiple fault identification results, and finally select the identification probability with the largest value as the target identification probability of the target three-dimensional point, so that the target fault identification result can be obtained according to the target identification probability of each target three-dimensional point.

[0156] S707. The target recognition probabilities of each target 3D point are summarized and processed to obtain the target tomographic recognition result of the area to be identified.

[0157] The summary processing can refer to summarizing the target recognition probabilities of each target 3D point in the same coordinate system to obtain the target tomographic recognition result of the area to be identified.

[0158] In this embodiment, by summarizing the results of each target fault identification, the fault features of each target three-dimensional point can be extracted into the same image. This facilitates the analysis of the interaction between each target fault, whether there are any contradictions, and the quality of the obtained fault identification results can be evaluated, thereby improving the accuracy of the fault identification results.

[0159] S708. Extract planar and profile data from the fault identification results.

[0160] Among them, planar data can refer to the planes where the main survey line and connecting survey line are located in the three-dimensional fault identification results.

[0161] Profile data can refer to the time depth and the plane where the main survey line is located, or the time depth and the plane where the connecting survey line is located, in the three-dimensional fault identification results.

[0162] In this embodiment, in order to evaluate the quality of the fault identification results, planar data and profile data are extracted from the fault identification results, and the planar data and profile data are combined and analyzed according to the fault characteristics.

[0163] S709. Based on the preset requirements for joint evaluation of planar and cross-sectional data, evaluate the planar data and cross-sectional data to obtain the first evaluation result.

[0164] Among them, the requirements for joint assessment of planar and profile data can refer to the fault characteristics that the fault must meet in both planar and profile data. For example, the profile of the fault cannot cross a reliable reflection wave phase axis; on adjacent parallel profiles, the morphology, dip angle, and fault position of the same fault plane are basically consistent with the fault properties.

[0165] In this embodiment, the fault identification results are evaluated at the level of the fault features according to the geological characteristics of the fault, which can further determine the accuracy of the fault identification results and thus improve the fault identification effect.

[0166] S710, Present the results of the first assessment.

[0167] The first assessment results will be visualized for easy viewing by users.

[0168] In this embodiment of the application, after determining the target fault identification result of the area to be identified based on the first fault identification result and the second fault identification result, the method further includes:

[0169] Acquire exploration data for the area to be identified, including at least one of well logging data and drilling data;

[0170] The fault identification results are evaluated based on the exploration data to obtain a second evaluation result;

[0171] Present the results of the second assessment.

[0172] Well logging data can refer to physical parameters such as the density of underground rocks, the speed of sound wave propagation, and the intensity of natural gamma radiation, measured by these parameters.

[0173] Drilling data refers to information such as the morphology, location, direction, and width of faults in underground rock core samples that are obtained through drilling.

[0174] In some implementations, based on the evaluation results of the first or second assessment, the tomographic identification and evaluation operations can be repeated using electronic devices until the obtained first assessment result meets the evaluation requirements.

[0175] In other embodiments, the parameters of the tomography model can be adjusted by electronic devices based on the evaluation results of the first or second assessment, thereby changing the recognition accuracy of the tomography model and improving the effect of tomography recognition.

[0176] In this embodiment, by making full use of well logging data and drilling data, the fault identification results are verified from different dimensions, thereby improving the accuracy of the fault identification results.

[0177] The fault identification method provided in this embodiment determines different offset domains based on offset requirements. Then, seismic traces are extracted from the common center point trace set according to different offset domains to obtain different offset domain datasets. The offset domain datasets are then input into the fault identification model to obtain different fault identification results. The identification probability of the target 3D point in each fault identification result is determined, and the identification probability with the highest value is taken as the target identification probability of the target 3D point. The target identification probabilities of each target 3D point are then summarized to obtain the target fault identification result of the area to be identified. Finally, the fault identification results are verified or quality evaluated by the joint evaluation requirements of planar profiles and exploration data to improve the accuracy of the fault identification results and ultimately improve the precision of fault identification, avoiding the loss of some faults.

[0178] Figure 8 This is a schematic diagram of the structure of a three-dimensional tomographic identification device provided in an embodiment of this application, as shown below. Figure 8 As shown, the device includes: an acquisition module 801, a first determination module 802, an overlay module 803, an identification module 804, and a second determination module 805, wherein:

[0179] The acquisition module 801 is used to acquire the first common center point gather and the second common center point gather of the region to be identified, wherein the offset range of the first common center point gather and the offset range of the second common center point gather are the same.

[0180] The first determining module 802 is used to determine, according to the preset offset distance requirements, the first seismic trace in the first common center point trace set is located within the first offset distance domain and the second seismic trace is located within the second offset distance domain, and the third seismic trace in the second common center point trace set is located within the first offset distance domain and the fourth seismic trace is located within the second offset distance domain, wherein the first offset distance domain and the second offset distance domain are different.

[0181] The overlay module 803 is used to overlay the first seismic trace and the third seismic trace, as well as the second seismic trace and the fourth seismic trace, to obtain the first offset domain dataset and the second offset domain dataset.

[0182] The identification module 804 is used to input the first offset domain dataset and the second offset domain dataset into the fault identification model respectively to obtain the first fault identification result and the second fault identification result. The fault identification model is trained based on the three-dimensional seismic data samples and the fault label data corresponding to the three-dimensional seismic data samples. The first fault identification result represents the identification probability that the three-dimensional point in the area to be identified in the first offset domain is in the fault, and the second fault identification result represents the identification probability that the three-dimensional point in the area to be identified in the second offset domain is in the fault.

[0183] The second determining module 805 is used to determine the target fault identification result of the area to be identified based on the first fault identification result and the second fault identification result.

[0184] In this embodiment of the application, the second determining module 805 is further configured to:

[0185] Based on the first fault identification result and the second fault identification result, determine the first identification probability of the target three-dimensional point being located in the fault in the first fault identification result and the second identification probability of the target three-dimensional point being located in the fault in the second fault identification result. The target three-dimensional point is a three-dimensional point in the area to be identified.

[0186] The target recognition probability of the target 3D point is determined based on the comparison between the first recognition probability and the second recognition probability.

[0187] The target recognition probabilities of each target 3D point are summarized and processed to obtain the target tomographic recognition results of the area to be identified.

[0188] In this embodiment of the application, the second determining module 805 is further configured to:

[0189] If the first recognition probability is greater than the second recognition probability, then the first recognition probability shall be used as the target recognition probability of the target 3D point.

[0190] If the first recognition probability is less than the second recognition probability, then the second recognition probability shall be used as the target recognition probability of the target three-dimensional point.

[0191] If the first recognition probability is equal to the second recognition probability, then the first recognition probability or the second recognition probability is taken as the target recognition probability of the target three-dimensional point.

[0192] Figure 9 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 9 As shown, the electronic device 90 includes:

[0193] The electronic device 90 may include a processor 901 with one or more processing cores, a memory 902 with one or more computer-readable storage media, a communication component 903, and other components. The processor 901, memory 902, and communication component 903 are connected via a bus 904.

[0194] In the specific implementation process, at least one processor 901 executes computer execution instructions stored in memory 902, causing at least one processor 901 to execute the above-mentioned three-dimensional tomographic identification method.

[0195] The specific implementation process of processor 901 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0196] In the above Figure 9 In the illustrated embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0197] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0198] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0199] In some embodiments, a computer program product is also provided, including a computer program or instructions that, when executed by a processor, implement the steps in any of the three-dimensional tomographic identification methods described above.

[0200] For details on the implementation of each of the above operations, please refer to the previous examples, which will not be repeated here.

[0201] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be performed by instructions, or by instructions controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor.

[0202] Therefore, embodiments of this application provide a computer-readable storage medium storing a plurality of instructions that can be loaded by a processor to execute the steps in any of the three-dimensional tomographic identification methods provided in embodiments of this application.

[0203] The storage medium may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0204] According to one aspect of this application, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium.

[0205] Since the instructions stored in the storage medium can execute the steps of any of the three-dimensional tomographic identification methods provided in the embodiments of this application, the beneficial effects that any of the three-dimensional tomographic identification methods provided in the embodiments of this application can achieve can be realized. For details, please refer to the previous embodiments, which will not be repeated here.

[0206] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the following claims.

[0207] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A three-dimensional tomographic identification method, characterized in that, include: Obtain the first common center point gather and the second common center point gather of the region to be identified, wherein the offset range of the first common center point gather and the offset range of the second common center point gather are the same. According to the preset offset distance requirements, the first seismic trace in the first common center point trace set and the second seismic trace in the second offset distance set are determined, as well as the third seismic trace in the second common center point trace set and the fourth seismic trace in the second offset distance set, wherein the first offset distance set and the second offset distance set are different. The first seismic trace and the third seismic trace, as well as the second seismic trace and the fourth seismic trace, are superimposed to obtain a first offset domain dataset and a second offset domain dataset; wherein, the superimposition process is used to indicate that multiple seismic traces are placed in the same three-dimensional coordinate system according to the three-dimensional coordinates of each seismic trace; The first offset domain dataset and the second offset domain dataset are respectively input into the fault identification model to obtain the first fault identification result and the second fault identification result. The fault identification model is trained based on three-dimensional seismic data samples and fault label data corresponding to the three-dimensional seismic data samples. The first fault identification result represents the identification probability that the three-dimensional point in the area to be identified in the first offset domain is in the fault, and the second fault identification result represents the identification probability that the three-dimensional point in the area to be identified in the second offset domain is in the fault. Based on the first fault identification result and the second fault identification result, a first identification probability that the target three-dimensional point is located in the fault in the first fault identification result and a second identification probability that it is located in the fault in the second fault identification result are determined, and the target three-dimensional point is a three-dimensional point in the region to be identified; If the first recognition probability is greater than the second recognition probability, then the first recognition probability is used as the target recognition probability of the target 3D point; if the first recognition probability is less than the second recognition probability, then the second recognition probability is used as the target recognition probability of the target 3D point; if the first recognition probability is equal to the second recognition probability, then either the first recognition probability or the second recognition probability is used as the target recognition probability of the target 3D point. The target recognition probabilities of each of the target three-dimensional points are summarized to obtain the target tomographic recognition result of the region to be identified.

2. The method according to claim 1, characterized in that, Before inputting the first offset domain dataset and the second offset domain dataset into the tomography identification model to obtain the first tomography identification result and the second tomography identification result, the method further includes: Acquire 3D seismic data samples; The three-dimensional seismic data samples are input into the model to be trained for fault identification processing to obtain the training identification results; Based on the training recognition results and the fault label data corresponding to the 3D seismic data samples, the balanced cross-entropy of the model to be trained is determined, wherein the balanced cross-entropy satisfies: ； in, For the model to be trained, For the first A sample of three-dimensional seismic data. For the first Fault label data corresponding to each 3D seismic data sample The parameters of the model to be trained are... The number of 3D seismic data samples. This is called the balance coefficient; Based on the balanced cross-entropy, the model to be trained is adjusted to obtain the tomography identification model.

3. The method according to claim 2, characterized in that, The acquisition of three-dimensional seismic data samples includes: Determine the fault label data; Based on the fault label data, a convolution forward modeling operation is performed to generate three-dimensional synthetic seismic data; Based on the aforementioned three-dimensional synthetic seismic data, a three-dimensional seismic data sample is obtained.

4. The method according to claim 1, characterized in that, After determining the target fault identification result of the region to be identified based on the first fault identification result and the second fault identification result, the method further includes: Extract planar and profile data from the fault identification results; According to the preset planar and profile joint evaluation requirements, the planar data and the profile data are evaluated to obtain a first evaluation result; The results of the first evaluation are displayed.

5. The method according to claim 1, characterized in that, After determining the target fault identification result of the region to be identified based on the first fault identification result and the second fault identification result, the method further includes: Obtain exploration data for the area to be identified, wherein the exploration data includes at least one of well logging data and drilling data; The fault identification results are evaluated based on the exploration data to obtain a second evaluation result; The results of the second evaluation are presented.

6. A three-dimensional tomographic identification device, characterized in that, include: The acquisition module is used to acquire the first common center point gather and the second common center point gather of the region to be identified, wherein the offset range of the first common center point gather and the offset range of the second common center point gather are the same. The first determining module is used to determine, according to a preset offset distance requirement, a first seismic trace in the first common center point trace set that is within a first offset distance domain and a second seismic trace in the second offset distance domain, and a third seismic trace in the second common center point trace set that is within a first offset distance domain and a fourth seismic trace in the second offset distance domain, wherein the first offset distance domain and the second offset distance domain are different. The overlay module is used to overlay the first seismic trace and the third seismic trace, as well as the second seismic trace and the fourth seismic trace, to obtain a first offset domain dataset and a second offset domain dataset; wherein, the overlay process is used to instruct that multiple seismic traces be placed in the same three-dimensional coordinate system according to the three-dimensional coordinates of each seismic trace; The identification module is used to input the first offset domain dataset and the second offset domain dataset into the fault identification model to obtain the first fault identification result and the second fault identification result. The fault identification model is trained based on three-dimensional seismic data samples and fault label data corresponding to the three-dimensional seismic data samples. The first fault identification result represents the identification probability that a three-dimensional point in the area to be identified in the first offset domain is in a fault, and the second fault identification result represents the identification probability that a three-dimensional point in the area to be identified in the second offset domain is in a fault. The second determining module is used to determine, based on the first fault identification result and the second fault identification result, a first identification probability that the target three-dimensional point is located in a fault in the first fault identification result and a second identification probability that it is located in a fault in the second fault identification result, wherein the target three-dimensional point is a three-dimensional point in the region to be identified; If the first recognition probability is greater than the second recognition probability, then the first recognition probability is used as the target recognition probability of the target 3D point; if the first recognition probability is less than the second recognition probability, then the second recognition probability is used as the target recognition probability of the target 3D point; if the first recognition probability is equal to the second recognition probability, then either the first recognition probability or the second recognition probability is used as the target recognition probability of the target 3D point. The target recognition probabilities of each of the target three-dimensional points are summarized to obtain the target tomographic recognition result of the region to be identified.

7. An electronic device, characterized in that, include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory to implement the method as described in any one of claims 1-5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-5.

Images