A method, device, equipment and product for detecting earthquake faults

Abstract

The embodiments of this specification relate to the field of seismic signal processing and interpretation technology, and provide a method, device, equipment and product for seismic fault detection. The method comprises: obtaining original seismic data of a target oil field; performing fault enhancement processing on the original seismic data to obtain fault-enhanced seismic data; inputting the fault-enhanced seismic data into a pre-trained convolutional neural network to obtain seismic fault detection results. Through the embodiments of this specification, the accuracy of fault detection for low signal-to-noise ratio seismic data can be solved, while the efficiency of fault detection can be improved.

Description

Technical Field

[0001] The embodiments of this specification relate to the technical field of seismic signal processing and interpretation, and in particular to a method, device, equipment and product for detecting seismic faults. Background Art

[0002] Fault zones can serve as migration channels for oil and gas, promoting the migration of oil and gas from source rocks to reservoirs, thereby accumulating in traps to form oil and gas reservoirs. Faults, as migration channels for oil and gas, can not only accelerate the migration of oil and gas, but also form natural oil and gas preservation structures because fault zones are usually formed during crustal movement and may become active and closed during geological evolution. Therefore, fault detection has important guiding significance for the exploration and development of oil and gas resources.

[0003] Faults usually appear as lateral reflection discontinuities on seismic sections. The response characteristics on seismic sections mainly include obvious misalignment of seismic event axes, obvious increase or decrease of event axes, and event bending. Based on the above response characteristics, they are usually characterized by attributes such as multi-channel similarity, degree of event bending, and reflection discontinuity. However, these attributes are usually very sensitive to seismic noise and lateral changes in strata, which means that for seismic data with low signal-to-noise ratio, simply using seismic coherence, curvature and other attributes cannot meet the needs of high-precision fault detection. Therefore, there is an urgent need for a seismic fault detection method that can be applied to areas with low signal-to-noise ratio of seismic data to solve the problem of low accuracy in seismic fault detection. Summary of the invention

[0004] In view of the above-mentioned problems in the prior art, the purpose of the embodiments of this specification is to provide a method, device, equipment and product for earthquake fault detection to solve the problem of low fault detection accuracy for earthquake data with low signal-to-noise ratio in the prior art.

[0005] In order to solve the above technical problems, the specific technical solutions of the embodiments of this specification are as follows:

[0006] On the one hand, an embodiment of this specification provides a method for detecting earthquake faults, the method comprising:

[0007] Obtaining original seismic data of the target oil field;

[0008] Performing fault enhancement processing on the original seismic data to obtain fault enhanced seismic data;

[0009] The fault-enhanced seismic data is input into a pre-trained convolutional neural network to obtain earthquake fault detection results.

[0010] Preferably, performing fault enhancement processing on the original seismic data to obtain fault enhanced seismic data includes:

[0011] Calculating the formation dip of the original seismic data using a gradient structure tensor algorithm;

[0012] Calculating discontinuity values ​​of original seismic data according to the formation dip;

[0013] Determining whether the discontinuous value is greater than a preset threshold;

[0014] If the discontinuity value is less than a preset threshold, the original seismic data is smoothed and filtered along the formation dip to obtain fault-enhanced seismic data.

[0015] Preferably, the formation dip of the original seismic data is calculated using the following formula:

[0016]

[0017] Where p represents the formation dip component along the main survey line, q represents the formation dip component along the tie survey line, and v 1x 、v 1y and v 1z Represents the characteristic components of the first eigenvector of the gradient structure tensor along the x, y and z directions respectively.

[0018] Preferably, the discontinuity value of the original seismic data is calculated using the following formula:

[0019]

[0020] Where s represents the discontinuous value of the original seismic data, K represents the number of sampling points in the vertical direction of the earthquake, N represents the total number of participating seismic traces, u represents the original seismic data, and u H represents the seismic data after Hilbert transformation, t represents the longitudinal time of the seismic data, p represents the formation dip component along the main survey line, q represents the formation dip component along the tie survey line, and x n Indicates the horizontal coordinate of the nth seismic trace, y n Indicates the vertical coordinate of the nth seismic trace.

[0021] Preferably, the original seismic data is smoothed and filtered using the following formula to obtain fault-enhanced seismic data:

[0022]

[0023] where u new represents the smoothed filtered seismic data, u(t0,x n ,y n ) represents the original seismic data, t0, x0 and y0 represent the center position of the analysis window, K represents the number of sampling points in the vertical direction of the earthquake, N represents the total number of participating seismic traces, t krepresents the time corresponding to the kth sample point, x n Indicates the horizontal coordinate of the nth seismic trace, y n represents the ordinate of the nth seismic trace, v (j) is the eigenvector corresponding to the jth eigenvalue.

[0024] Preferably, the pre-trained convolutional neural network is a U-Net network, and the U-Net network includes an encoder, a decoder and a bottleneck layer;

[0025] The encoder gradually reduces the resolution of the image through convolution and pooling operations to extract feature information of the input data;

[0026] The decoder gradually restores the resolution of the image and fuses high-resolution feature information through deconvolution and upsampling operations;

[0027] The bottleneck layer is used to connect the encoder and the decoder, and transmit the feature information in the encoder to the decoder through a skip connection operation to achieve feature fusion.

[0028] On the other hand, an embodiment of this specification provides an earthquake fault detection device, the device comprising:

[0029] An acquisition module, used to acquire the original seismic data of the target oil field;

[0030] A processing module, used for performing fault enhancement processing on the original seismic data to obtain fault enhanced seismic data;

[0031] The detection module is used to input the fault-enhanced seismic data into a pre-trained convolutional neural network to obtain earthquake fault detection results.

[0032] On the other hand, an embodiment of the present specification further provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein when the computer program is executed by the processor, the instructions of any one of the above methods are executed.

[0033] On the other hand, an embodiment of the present specification further provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor of a computer device, the computer program executes instructions of any one of the above-described methods.

[0034] On the other hand, the embodiments of this specification further provide a computer program product, which, when executed by a processor of a computer device, executes instructions of any one of the above methods.

[0035] One or more technical solutions provided by some embodiments of this specification have at least the following technical effects:

[0036] The embodiment of this specification is driven by seismic data, and the dip angle calculation of the formation is carried out. Discontinuity detection is carried out under the constraint of the dip angle of the formation. The seismic data with weak discontinuity is smoothed with boundary protection to obtain fault-enhanced data with high signal-to-noise ratio. Compared with the original seismic data, the fault-enhanced seismic data has clearer imaging of the fault and crisper breakpoints, which provides good data support for subsequent fault detection. On the basis of fault-enhanced data, the full convolutional neural network three-dimensional image segmentation algorithm is used, and the balanced cross entropy loss function is used to optimize the convolutional neural network parameters. The algorithm simplifies the original U-Net by reducing the number of convolutional layers and features per layer, which not only significantly saves GPU memory and computing time, but also maintains high efficiency and high precision three-dimensional fault detection. Compared with traditional seismic coherence, curvature and other fault attributes, the fault detection in this specification has higher detection accuracy and higher computing efficiency, which effectively supports the efficient exploration and development of oil and gas.

[0037] The above description is only an overview of the technical solutions of some embodiments of this specification. In order to more clearly understand the technical means of some embodiments of this specification, they can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the embodiments of this specification more obvious and easy to understand, the following specifically cites the preferred embodiments and describes them in detail in conjunction with the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0039] Figure 1 A schematic diagram showing a flow chart of a method for detecting earthquake faults in some embodiments of this specification is shown;

[0040] Figure 2 A schematic diagram showing the steps of performing fault enhancement processing on original seismic data in some embodiments of this specification is shown;

[0041] Figure 3 A schematic diagram showing the comparison of seismic sections in original seismic data and fault enhancement data in some embodiments of this specification is shown;

[0042] Figure 4 A schematic diagram of comparing seismic planes in original seismic data and fault enhancement data in some embodiments of this specification is shown;

[0043] Figure 5A network structure diagram of U-Net in some embodiments of this specification is shown;

[0044] Figure 6 A comparison diagram showing the effect of fault detection on seismic profiles based on convolutional neural networks and traditional coherent attributes on the basis of fault enhancement in some embodiments of this specification is shown;

[0045] Figure 7 A comparison diagram showing the effect of fault detection on a seismic plane based on a convolutional neural network and traditional coherent attributes on the basis of fault enhancement in some embodiments of this specification is shown;

[0046] Figure 8 A schematic diagram of the structure of an earthquake fault detection device in some embodiments of this specification is shown;

[0047] Fig. 9 A schematic diagram of the structure of a computer device provided in some embodiments of the present specification is shown.

[0048] Description of the accompanying symbols:

[0049] 801, obtaining module;

[0050] 802, processing module;

[0051] 803, detection module;

[0052] 902. Computer equipment;

[0053] 904, processor;

[0054] 906. Memory;

[0055] 908, driving mechanism;

[0056] 910, input / output module;

[0057] 912. Input device;

[0058] 914. Output device;

[0059] 916. Presentation equipment;

[0060] 918. Graphical user interface;

[0061] 920, network interface;

[0062] 922, communication link;

[0063] 924. Communication bus. DETAILED DESCRIPTION

[0064] The following will be combined with the drawings in the embodiments of this specification to clearly and completely describe the technical solutions in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this specification.

[0065] In the existing technology, fault detection mainly uses the edge attributes, discontinuity attributes or amplitude attributes of earthquakes for identification, mainly including algorithms such as seismic coherence, curvature, edge detection and ant tracking. Faults usually appear as lateral reflection discontinuities on seismic profiles. The response characteristics on seismic profiles mainly include obvious misalignment of seismic event axes, obvious increase or decrease of event axes, and event bending. Based on the above response characteristics, multi-channel similarity (such as coherence attributes), event axis bending degree (such as curvature attributes), reflection discontinuity (such as ant body attributes) and other attributes are usually used for characterization. However, these attributes are usually also very sensitive to seismic noise and lateral changes in strata, which means that for seismic data with low signal-to-noise ratio, the detection of traditional seismic data-driven fault detection algorithms such as coherence, curvature and ant body is poor, and cannot meet the fault detection needs of complex seismic data.

[0066] In order to solve the above problems, an embodiment of the present specification provides a method for detecting earthquake faults. Figure 1 This is a flow chart of a method for detecting earthquake faults provided in an embodiment of this specification. This specification provides method operation steps as described in the embodiment or flow chart, but may include more or fewer operation steps based on conventional or non-creative labor. The order of steps listed in the embodiment is only one way of executing the order of many steps and does not represent the only execution order. When the system or device product is executed in practice, it can be executed in the order of the method shown in the embodiment or the accompanying drawings or in parallel.

[0067] It should be noted that the terms "first", "second", etc. in this specification and claims and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of this specification described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, device, product or equipment that includes a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or equipment.

[0068] Reference Figure 1As shown, an embodiment of this specification provides a method for detecting earthquake faults, the method comprising:

[0069] S101: Acquire original seismic data of the target oil field;

[0070] S102: performing fault enhancement processing on the original seismic data to obtain fault enhanced seismic data;

[0071] S103: Inputting the fault-enhanced seismic data into a pre-trained convolutional neural network to obtain earthquake fault detection results.

[0072] The embodiment of this specification is driven by seismic data, and the dip angle calculation of the formation is carried out. Discontinuity detection is carried out under the constraint of the dip angle of the formation. The seismic data with weak discontinuity is smoothed with boundary protection to obtain fault-enhanced data with high signal-to-noise ratio. Compared with the original seismic data, the fault-enhanced seismic data has clearer imaging of the fault and crisper breakpoints, which provides good data support for subsequent fault detection. On the basis of fault-enhanced data, the full convolutional neural network three-dimensional image segmentation algorithm is used, and the balanced cross entropy loss function is used to optimize the convolutional neural network parameters. The algorithm simplifies the original U-Net by reducing the number of convolutional layers and features per layer, which not only significantly saves GPU memory and computing time, but also maintains high efficiency and high precision three-dimensional fault detection. Compared with traditional seismic coherence, curvature and other fault attributes, the fault detection in this specification has higher detection accuracy and higher computing efficiency, which effectively supports the efficient exploration and development of oil and gas.

[0073] Original seismic data refers to the original single-shot records collected from the target oil field. After the migration imaging, the three-dimensional image data that can finally represent the real underground structural morphology is obtained. The data included mainly reflects the real underground structural morphology and the changes in amplitude strength. However, due to factors such as instruments, noise, environment, and human factors, it is difficult to guarantee the quality of seismic data actually collected in the field, which easily leads to problems such as blurred seismic profiles and low signal-to-noise ratio in the collected data, which interferes with the accuracy of fault detection and directly affects oil exploration and development. The purpose of fault enhancement is to further clearly characterize the fault, while improving the signal-to-noise ratio of seismic data, and providing good data support for subsequent fault detection.

[0074] Reference Figure 2 As shown, the original seismic data is subjected to fault enhancement processing to obtain fault enhanced seismic data, including:

[0075] S201: Calculating the formation dip of the original seismic data using a gradient structure tensor algorithm;

[0076] S202: Calculating discontinuous values ​​of original seismic data according to the formation dip;

[0077] S203: Determine whether the discontinuous value is greater than a preset threshold;

[0078] S204: If the discontinuity value is less than a preset threshold, the original seismic data is smoothed and filtered along the formation dip to obtain fault-enhanced seismic data.

[0079] The formation dip refers to the angle between the downward direction of the maximum dip line of the rock layer and its projection line on the horizontal plane. By calculating the formation dip, the true structural form of the formation underground can be better represented, and the seismic data can be smoothed with boundary protection using this structural form as a constraint to enhance the clarity of the fault in the seismic data and provide more reliable data for subsequent fault detection using convolutional neural networks.

[0080] In some embodiments, the gradient structure tensor algorithm (GST) is used to calculate the formation dip, where the formation dip mainly includes the formation dip of the main survey line and the connecting survey line. Since the seismic data is collected along the plane and cannot be collected perpendicular to the plane, the formation dip is calculated along the horizontal plane. Usually, the section perpendicular to the structure (i.e., the x direction) is the main survey line, and the one perpendicular to the main survey line (i.e., the y direction) is the connecting survey line. First, GST is used to obtain the amplitude vector gradient g of the seismic data, which can be characterized as the derivative of the seismic amplitude u in three directions, mainly characterized as follows:

[0081]

[0082] Where g represents the amplitude vector gradient, x, y and z represent the three-dimensional spatial directions of the seismic data, k, l and m represent the grid points of the seismic data along the three-dimensional spatial directions, and u represents the seismic amplitude.

[0083] The gradient structure tensor C at the spatial grid point (k,l,m) klm The three components of the amplitude vector gradient g are cross-correlated and autocorrelated and calculated as follows:

[0084]

[0085] It can be seen from formula (2) that when the earthquake is at a strong peak or trough, the change of the earthquake amplitude u is very slow. For the strata with relatively continuous and flat seismic phase axes, g z ≈0, and if the earthquake is subjected to Hilbert transformation, Therefore, in order to solve the problem of unstable gradient of seismic data at the peak or trough, the Hilbert transform can be combined to calculate the seismic phase The weighted gradient of is defined as:

[0086]

[0087] Where Weight represents the weighting operator and the superscript H represents the Hilbert transform. Since the derivative of the phase cannot be effectively defined at ±180°, the seismic phase is calculated as The inverse tangent partial derivative of is characterized as follows:

[0088]

[0089] in and They represent the amplitude vector gradients along the x, y and z directions after the Hilbert transform of the seismic data, respectively. x , g y and g z Respectively represent the amplitude vector gradient of the original seismic data along the x-direction, y-direction and z-direction. klm represents the amplitude envelope of the seismic data, which is the sum of the squares of the seismic data and the seismic data after Hilbert transformation, and can be expressed as:

[0090]

[0091] where u klm represents the original earthquake amplitude, represents the earthquake amplitude after Hilbert transformation. Multiply both sides of equation (4) by (e klm ) 2 , the characterization of the weighted phase gradient vector Γ can be obtained as follows:

[0092]

[0093] in and They represent the weighted phase gradient components along the x, y and z directions respectively. Therefore, using formula (6), the gradient structure tensor C can be obtained: klm The representation form of the weighted phase gradient vector Γ is:

[0094]

[0095] From this, the gradient structure tensor C can be calculated klm The average value can be expressed as:

[0096]

[0097] Where N represents the total number of participating seismic traces, and K represents the number of sampling points in the vertical direction of the seismic trace. l(n) and m(n) represent the index of the nth seismic trace along the x and y directions, respectively. The gradient structure tensor C klm Given the transformation of seismic amplitude and phase along the x, y and z directions, in order to determine the direction of the maximum change, it is necessary to transform the gradient structure tensor C klm Decomposition of eigenvectors and eigenvalues ​​is performed, and the expressions are as follows:

[0098]

[0099] Where v1, v2 and v3 represent the first, second and third eigenvectors respectively; λ1, λ2 and λ3 represent the three eigenvalues ​​corresponding to the three eigenvectors respectively. Since the gradient structure tensor quantifies the three-dimensional variation of seismic amplitude, its first eigenvector provides the assumed reflection surface normal Therefore, the first eigenvector can be expressed as:

[0100]

[0101] where v 1x 、v 1y and v 1z They represent the characteristic components of the first eigenvector along the three directions of x, y and z. Therefore, the first eigenvector v1 can calculate the formation dip along the x and y directions, that is, the formation dip along the main survey line and the connecting survey line can be expressed as follows:

[0102]

[0103] Where p represents the formation dip component along the main survey line, and q represents the formation dip component along the connecting survey line.

[0104] In some embodiments, under the constraints of the calculated stratigraphic dip data of the main survey line and the connecting survey line, driven by seismic data, the original seismic trace and the seismic trace after Hilbert transformation are combined, which can effectively avoid the problem of gradient instability caused by the slow change of seismic amplitude when the earthquake is at a strong peak or trough. The ratio of the average energy of multiple channels to the energy of a single channel is used to characterize the discontinuity of seismic data, so the discontinuity detection characterization of earthquakes can be obtained as follows:

[0105]

[0106] Where s represents the discontinuous value of the original seismic data, K represents the number of sampling points in the vertical direction of the earthquake, N represents the total number of participating seismic traces, u represents the original seismic data, and u Hrepresents the seismic data after Hilbert transformation, t represents the longitudinal time of the seismic data, p represents the formation dip component along the main survey line, q represents the formation dip component along the tie survey line, and x n Indicates the horizontal coordinate of the nth seismic trace, y n Indicates the vertical coordinate of the nth seismic trace.

[0107] Formula (12) gives the discontinuity detection attributes under the constraints of different seismic dip angles. The algorithm uses the L1 norm to solve the multi-channel energy mean and single-channel energy, because the L1 norm is insensitive to areas with good lateral continuity of seismic event axes, which can better remove the inter-layer effect of the stratum and highlight the lateral discontinuity of the stratum. Moreover, compared with the L2 norm, the L2 norm may be affected by rounding errors, resulting in poor results in the calculated discontinuity attributes.

[0108] In some embodiments, after the discontinuity of the original seismic data is calculated by the above method, the obtained discontinuity value is compared with a preset threshold. If the discontinuity value is greater than the preset threshold, it indicates that the original seismic data collected has a strong discontinuity and does not need to be subjected to fault enhancement processing. If the discontinuity value is less than the preset threshold, it is difficult to tell whether there is a fault by the naked eye, so it is necessary to perform fault enhancement processing. In order to obtain fault enhancement data with a high signal-to-noise ratio, the above discontinuity attribute is used as a constraint to make each sample point vector have a high correlation with its original seismic data u and the seismic data u after Hilbert transformation. H The covariance matrix C of the sample point is constructed as follows:

[0109]

[0110] Where K represents the number of sampling points vertically up and down the earthquake, u represents the actual earthquake data, and u H represents the seismic data after Hilbert transformation, t k Indicates the time corresponding to the kth sample point, x0 and y0 indicate the center position of the analysis window; the sample point is the time window taken for the discontinuous attribute, as shown in formula (12), K sample points are taken upward and K sample points are taken downward for the sample point, then the total number of sample points that need to be obtained for the discontinuous attribute is 2K+1. The sample point vector is the seismic data amplitude value u at the sample point where the discontinuous attribute is required. Usually, the seismic data amplitude value u is represented by time t, horizontal coordinate x and vertical coordinate y, so the sample point vector can be expressed as u(t,x,y).

[0111] The covariance matrix in formula (13) can be expressed in the following matrix form by decomposing the eigenvalue and eigenvector:

[0112]

[0113] Where N represents the total number of earthquakes involved, C represents the covariance matrix, and λ k represents the kth eigenvalue, v (k) is the kth eigenvalue λ k The corresponding feature vector.

[0114] Normalizing the eigenvector of formula (14) yields:

[0115]

[0116] Sort the eigenvalue-eigenvector pairs from large to small according to the size of the eigenvalue. The most important eigenvector corresponds to the largest eigenvalue, that is, |λ1|≥|λ2|≥…≥|λ J |.

[0117] After normalizing the eigenvectors in formula (15), the principal component analysis method is used to select eigenvectors with a correlation of more than 90%. Under the constraint of seismic discontinuity attributes, smoothing filtering with boundary protection is carried out to make the imaging accuracy of seismic data at faults higher and further improve the signal-to-noise ratio of seismic data as shown below:

[0118]

[0119] where u new represents the seismic data after improving the signal-to-noise ratio, t0, x0 and y0 represent the center position of the analysis window, K represents the number of sampling points in the vertical direction of the earthquake, and N represents the total number of tracks involved in the earthquake. Figure 3 As shown, Figure 3 (a) is a schematic diagram of the seismic profile in the original seismic data. Figure 3 (b) is a schematic diagram of the seismic profile in the fault enhancement data, which shows that the imaging accuracy of the data at the fault can be effectively improved by fault enhancement of the original seismic data. Figure 4 As shown, Figure 4 (a) is a schematic diagram of the seismic plane in the original seismic data. Figure 4 (b) is a schematic diagram of the earthquake plane in the fault enhancement data, which shows that the details of the earthquake coherent attributes are more clearly portrayed after fault enhancement.

[0120] In some embodiments, combined with the seismic data with high signal-to-noise ratio after fault enhancement, an efficient end-to-end convolutional neural network (CNN) algorithm is used to treat the fault detection of 3D seismic as a 3D image segmentation problem, that is, the fault area marked in the 3D seismic image is 1, and other areas are 0. The convolutional neural network selected in some embodiments of this specification is U-Net. Figure 5As shown in the figure, the U-Net network includes an encoder, a decoder and a bottleneck layer, wherein the encoder gradually reduces the resolution of the image through convolution and pooling operations to extract the feature information of the input data; the decoder gradually restores the resolution of the image through deconvolution and upsampling operations and fuses the high-resolution feature information; the bottleneck layer is used to connect the encoder and the decoder, and pass the feature information in the encoder to the decoder through a skip connection operation to achieve feature fusion. U-Net has shown high accuracy in image segmentation tasks, especially in edge detection, because of its skip connection mechanism, it is able to use low-level and high-level features to make accurate predictions. Figure 5 As shown in the figure, the U-Net network contains four encoders and three decoders. The encoder includes 7 convolutional layers and 3 pooling layers. ReLu activation function is used after each convolutional layer to accelerate the convergence of the neural network model and improve the generalization ability of the model. The decoder includes 6 convolutional layers and 3 upsampling layers. The encoder and decoder are connected through jump connection operations. Through jump connections, the feature maps in the pooling layer are connected to the feature maps in the corresponding upsampling layer. These connections enable the model to better capture local and global features in the image, thereby improving the performance of the model.

[0121] In 3D seismic images, the distribution of fault samples and non-fault samples is usually highly unbalanced. Therefore, the balanced cross entropy loss function is used to optimize the CNN network parameters. This algorithm simplifies the original U-Net by reducing the number of convolutional layers and features per layer, which not only significantly saves the memory and computing time of the graphics processing unit (GPU), but also maintains high efficiency and high accuracy of 3D fault detection. The expression of the balanced cross entropy loss function is:

[0122]

[0123] Where R is the balanced cross entropy loss function, S represents the number of pixels in the input 3D seismic image, and y i represents the true binary label, and β represents the ratio of non-fragmented pixels to total pixels, i.e. 1-β represents the ratio of tomographic pixels to total pixels, p i represents the predicted probability calculated from the sigmoid activation function of the last convolutional layer, and 0<p i <1. Figure 6 As shown, Figure 6 (a) is the effect diagram of fault detection on seismic profile based on convolutional neural network on the basis of fault enhancement. Figure 6(b) The effect of fault detection on seismic profiles based on fault enhancement based on traditional coherence attributes. This figure shows that fault detection using convolutional neural networks based on fault enhancement data has higher longitudinal continuity and lateral resolution than traditional coherence attributes. Figure 7 As shown, Figure 7 (a) is the effect diagram of fault detection on the seismic plane based on the convolutional neural network on the basis of fault enhancement. Figure 7 (b) is the effect diagram of fault detection on the seismic plane based on traditional coherence attributes on the basis of fault enhancement. The figure shows that the convolutional neural network image segmentation fault detection technology can depict the fault more clearly and has a higher resolution on the plane.

[0124] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. And the acquisition, storage, use, and processing of data in the technical solutions described in the embodiments of this application are in compliance with relevant regulations.

[0125] Based on the above-mentioned earthquake fault detection method, the embodiment of this specification also provides an earthquake fault detection device. The device may include a system (including a distributed system), software (application), module, component, server, client, etc. using the method described in the embodiment of this specification and a device combined with necessary implementation hardware. Based on the same innovative concept, the device in one or more embodiments provided in the embodiment of this specification is as described in the following embodiments. Since the implementation scheme and method of the device to solve the problem are similar, the implementation of the specific device in the embodiment of this specification can refer to the implementation of the aforementioned method, and the repetitions will not be repeated. As used below, the term "unit" or "module" can implement a combination of software and / or hardware of predetermined functions. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceived.

[0126] Specifically, Figure 8 is a schematic diagram of a module structure of an embodiment of an earthquake fault detection device provided in the present specification, referring to Figure 8 As shown, an earthquake fault detection device provided in an embodiment of this specification includes:

[0127] An acquisition module 801 is used to acquire original seismic data of a target oil field;

[0128] A processing module 802 is used to perform fault enhancement processing on the original seismic data to obtain fault enhanced seismic data;

[0129] The detection module 803 is used to input the fault-enhanced seismic data into a pre-trained convolutional neural network to obtain an earthquake fault detection result.

[0130] The beneficial effects obtained by the device provided in the embodiments of this specification are consistent with the beneficial effects obtained by the above method and will not be repeated here.

[0131] Reference Fig. 9 As shown, based on the above-mentioned earthquake fault detection method, a computer device 902 is also provided in an embodiment of the specification, wherein the above-mentioned method is run on the computer device 902. The computer device 902 may include one or more processors 904, such as one or more central processing units (CPUs), and each processing unit may implement one or more hardware threads. The computer device 902 may also include any memory 906, which is used to store any kind of information such as code, settings, data, etc. Non-limiting, for example, the memory 906 may include any one or more combinations of the following: any type of RAM, any type of ROM, flash memory device, hard disk, optical disk, etc. More generally, any memory can use any technology to store information. Further, any memory can provide volatile or non-volatile retention of information. Further, any memory can represent a fixed or removable component of the computer device 902. In one case, when the processor 904 executes an associated instruction stored in any memory or a combination of memories, the computer device 902 can perform any operation of the associated instruction. The computer device 902 also includes one or more drive mechanisms 908 for interacting with any storage, such as a hard disk drive mechanism, an optical disk drive mechanism, etc.

[0132] The computer device 902 may also include an input / output module 910 (I / O) for receiving various inputs (via input devices 912) and for providing various outputs (via output devices 914). A specific output mechanism may include a presentation device 916 and an associated graphical user interface (GUI) 918. In other embodiments, the input / output module 910 (I / O), the input device 912, and the output device 914 may not be included, and the computer device 902 may be used as a computer device in a network. The computer device 902 may also include one or more network interfaces 920 for exchanging data with other devices via one or more communication links 922. One or more communication buses 924 couple the components described above together.

[0133] The communication link 922 may be implemented in any manner, for example, through a local area network, a wide area network (e.g., the Internet), a point-to-point connection, etc., or any combination thereof. The communication link 922 may include any combination of hardwired links, wireless links, routers, gateway functions, name servers, etc. governed by any protocol or combination of protocols.

[0134] Corresponding to Figure 1 to Figure 2 In addition to the method shown, an embodiment of the specification also provides a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above method are executed.

[0135] The embodiment of the present specification also provides a computer-readable instruction, wherein when the processor executes the instruction, the program therein causes the processor to execute the following Figure 1 to Figure 2 The method shown.

[0136] The embodiments of the present specification also provide a computer program product, including at least one instruction or at least one program, wherein the at least one instruction or the at least one program is loaded and executed by a processor to implement the following Figure 1 to Figure 2 The method shown.

[0137] It should be understood that in the various embodiments of this specification, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this specification.

[0138] It should also be understood that in the embodiments of this specification, the term "and / or" is only a description of the association relationship of the associated objects, indicating that three relationships can exist. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this specification generally indicates that the associated objects before and after are in an "or" relationship.

[0139] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in this specification can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the composition and steps of each example have been generally described in the above description according to function. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this specification.

[0140] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0141] In the several embodiments provided in this specification, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, or it can be an electrical, mechanical or other form of connection.

[0142] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the embodiments of this specification.

[0143] In addition, each functional unit in each embodiment of this specification may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0144] If the integrated unit is implemented in the form of 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 specification is essentially 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, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of this specification. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0145] Specific embodiments are used in this specification to illustrate the principles and implementation methods of this specification. The description of the above embodiments is only used to help understand the methods and core ideas of this specification. At the same time, for those skilled in the art, according to the ideas of this specification, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on this specification.

Claims

1. A method for detecting earthquake faults, characterized in that: The method comprises: Obtaining original seismic data of the target oil field; Calculating the formation dip of the original seismic data using a gradient structure tensor algorithm, wherein the formation dip is calculated based on the original seismic data and the seismic data after Hilbert transformation; Calculating a discontinuity value of the original seismic data according to the formation dip, wherein the discontinuity value is characterized by a ratio of a multi-channel energy average to a single-channel energy, wherein the multi-channel energy average and the single-channel energy are obtained by calculating an L1 norm; Determining whether the discontinuous value is greater than a preset threshold; If the discontinuity value is less than a preset threshold, the original seismic data is smoothed and filtered along the formation dip to obtain fault-enhanced seismic data; The fault-enhanced seismic data is input into a pre-trained convolutional neural network to obtain earthquake fault detection results.

2. The method according to claim 1, characterized in that The formation dip of the original seismic data is calculated using the following formula: Where p represents the formation dip component along the main survey line, q represents the formation dip component along the tie survey line, and v 1x 、v 1y and v 1z Represents the characteristic components of the first eigenvector of the gradient structure tensor along the x, y and z directions respectively.

3. The method according to claim 1, characterized in that The discontinuity value of the original seismic data is calculated using the following formula: Where s represents the discontinuous value of the original seismic data, K represents the number of sampling points in the vertical direction of the earthquake, N represents the total number of participating seismic traces, u represents the original seismic data, and u H represents the seismic data after Hilbert transformation, t represents the longitudinal time of the seismic data, p represents the formation dip component along the main survey line, q represents the formation dip component along the tie survey line, and x n Indicates the horizontal coordinate of the nth seismic trace, y n Indicates the vertical coordinate of the nth seismic trace.

4. The method according to claim 1, characterized in that The original seismic data is smoothed and filtered using the following formula to obtain fault-enhanced seismic data: where u new represents the smoothed filtered seismic data, u(t0,x n ,y n ) represents the original seismic data, t0, x0 and y0 represent the center position of the analysis window, K represents the number of sampling points in the vertical direction of the earthquake, N represents the total number of participating seismic traces, t k represents the time corresponding to the kth sample point, x n Indicates the horizontal coordinate of the nth seismic trace, y n represents the ordinate of the nth seismic trace, v (j) is the eigenvector corresponding to the jth eigenvalue.

5. The method according to claim 1, characterized in that The pre-trained convolutional neural network is a U-Net network, and the U-Net network includes an encoder, a decoder and a bottleneck layer; The encoder gradually reduces the resolution of the image through convolution and pooling operations to extract feature information of the input data; The decoder gradually restores the resolution of the image and fuses high-resolution feature information through deconvolution and upsampling operations; The bottleneck layer is used to connect the encoder and the decoder, and transmit the feature information in the encoder to the decoder through a skip connection operation to achieve feature fusion.

6. An earthquake fault detection device, characterized in that: The device comprises: An acquisition module, used to acquire the original seismic data of the target oil field; A first calculation module is used to calculate the formation dip of the original seismic data by using a gradient structure tensor algorithm, wherein the formation dip is calculated based on the original seismic data and the seismic data after Hilbert transformation; A second calculation module is used to calculate the discontinuity value of the original seismic data according to the formation dip, wherein the discontinuity value is characterized by a ratio of a multi-channel energy average to a single-channel energy, wherein the multi-channel energy average and the single-channel energy are obtained by calculating an L1 norm; A judging module, used to judge whether the discontinuous value is greater than a preset threshold; A processing module, configured to perform smoothing and filtering processing on the original seismic data along the formation dip to obtain fault-enhanced seismic data if the discontinuity value is less than a preset threshold; The detection module is used to input the fault-enhanced seismic data into a pre-trained convolutional neural network to obtain earthquake fault detection results.

7. A computer device comprising a memory, a processor and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the method according to any one of claims 1 to 5 is implemented.

8. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 5 is implemented.

9. A computer program product, characterized in that The method comprises at least one instruction or at least one program, wherein the at least one instruction or the at least one program is loaded and executed by a processor to implement the method according to any one of claims 1 to 5.

Images