Multi-scale fault identification method and device for three-dimensional seismic data, equipment and medium

Abstract

The invention discloses a multi-scale fault identification method and device for three-dimensional seismic data, equipment and a medium, and relates to the technical field of deep learning, and the method comprises the steps: obtaining seismic wavelets, designing different geologic models, generating seismic data based on the different geologic models and seismic wavelets, carrying out the three-dimensional enhancement and expansion of the seismic data, and carrying out the normalization processing. Obtaining processed seismic data; constructing a double-branch convolutional neural network model, and performing training, testing and hyper-parameter optimization on the double-branch convolutional neural network model by using the processed seismic data to obtain an optimized double-branch convolutional neural network model; and acquiring actual seismic data, and inputting the actual seismic data into the optimized double-branch convolutional neural network model to obtain a multi-scale fault identification result. According to the method, the problem of insufficient feature extraction in the aspect of seismic data fault interpretation can be solved, the faults in the seismic data are accurately identified and distinguished, and the three-dimensional seismic data fault identification efficiency and accuracy are improved.

Description

Technical Field

[0001] The present invention relates to the field of deep learning technology, and in particular to a multi-scale fault identification method, device, equipment and medium for three-dimensional seismic data. Background Art

[0002] Although artificial intelligence technology has made significant progress in the identification of earthquake fault information, it still faces some challenges in practical applications. First, the training of deep learning models requires a large amount of labeled data, which is time-consuming and costly to obtain in the field of earthquake data. In addition, these models also have a high demand for computing resources. The limitations of synthetic data cannot be ignored. It may limit the generalization ability of network models under different geological conditions, and the model is prone to overfitting during training. In addition, existing neural network algorithms have the problem of insufficient feature extraction in the interpretation of earthquake data faults, resulting in low accuracy and efficiency in identifying faults in earthquake data.

[0003] As can be seen from the above, how to solve the problem of insufficient feature extraction in the interpretation of seismic data faults, accurately identify and distinguish faults in seismic data, and improve the efficiency and accuracy of fault identification in three-dimensional seismic data are problems to be solved in this field. Summary of the invention

[0004] In view of this, the purpose of the present invention is to provide a multi-scale fault identification method, device, equipment and medium for three-dimensional seismic data, which can solve the problem of insufficient feature extraction in the interpretation of seismic data faults, accurately identify and distinguish faults in seismic data, and improve the efficiency and accuracy of three-dimensional seismic data fault identification. The specific scheme is as follows:

[0005] In a first aspect, the present application discloses a multi-scale fault identification method for three-dimensional seismic data, comprising:

[0006] Acquire seismic wavelets, design different geological models, generate seismic data based on the different geological models and the seismic wavelets, perform three-dimensional enhancement and expansion and normalization processing on the seismic data to obtain processed seismic data;

[0007] Constructing a dual-branch convolutional neural network model, and using the processed seismic data to train, test and optimize hyperparameters of the dual-branch convolutional neural network model to obtain an optimized dual-branch convolutional neural network model;

[0008] Actual seismic data is acquired and input into the optimized dual-branch convolutional neural network model to obtain multi-scale fault identification results.

[0009] Optionally, generating seismic data based on different geological models and seismic wavelets, and performing three-dimensional enhancement, expansion and normalization processing on the seismic data include:

[0010] convolving different geological models with the seismic wavelets respectively to generate seismic data;

[0011] The seismic data is three-dimensionally enhanced and expanded by using data enhancement technology to obtain a three-dimensional seismic data set, and the three-dimensional seismic data set is normalized.

[0012] Optionally, the constructing of a dual-branch convolutional neural network model includes:

[0013] Select loss functions and optimizers based on business needs;

[0014] A two-branch convolutional neural network model is constructed, and the hyperparameters in the two-branch convolutional neural network model are set using the loss function and the optimizer.

[0015] Optionally, the using the processed seismic data to train, test and optimize hyperparameters of the dual-branch convolutional neural network model includes:

[0016] Install and configure a database and a Python development environment for training, testing, and hyperparameter optimization, so that the processed seismic data can be used to train, test, and optimize the hyperparameters of the dual-branch convolutional neural network model in the Python development environment; the development environment includes TensorFlow and Keras.

[0017] Optionally, using the processed seismic data to train and test the dual-branch convolutional neural network model includes:

[0018] Dividing the processed seismic data according to a preset data set division ratio to obtain training data and test data;

[0019] The two-branch convolutional neural network model is trained using the training data, and the trained two-branch convolutional neural network model is tested using the test data.

[0020] Optionally, using the processed seismic data to perform hyperparameter optimization on the trained and tested dual-branch convolutional neural network model includes:

[0021] The processed seismic data is used to monitor the index parameters in the two-branch convolutional neural network model after training and testing in real time, and the hyperparameters in the two-branch convolutional neural network model after training and testing are adjusted and optimized accordingly; the index parameters include accuracy and loss value; the hyperparameters include learning rate, number of iterations and optimization algorithm parameters.

[0022] Optionally, inputting the actual seismic data into the optimized dual-branch convolutional neural network model to obtain a multi-scale fault identification result includes:

[0023] The actual seismic data is input into the optimized two-branch convolutional neural network model, and the U-Net network architecture in the optimized two-branch convolutional neural network model is used as the basic architecture to identify and segment the actual seismic data to obtain multi-scale fault characteristics, and generate multi-scale fault identification results based on the multi-scale fault characteristics.

[0024] In a second aspect, the present application discloses a multi-scale fault identification device for three-dimensional seismic data, comprising:

[0025] A data processing module is used to obtain seismic wavelets, design different geological models, generate seismic data based on the different geological models and the seismic wavelets, and perform three-dimensional enhancement, expansion and normalization processing on the seismic data to obtain processed seismic data;

[0026] A model optimization module, used for constructing a dual-branch convolutional neural network model, and using the processed seismic data to train, test and optimize hyperparameters of the dual-branch convolutional neural network model to obtain an optimized dual-branch convolutional neural network model;

[0027] The multi-scale fault identification module is used to obtain actual seismic data and input the actual seismic data into the optimized dual-branch convolutional neural network model to obtain a multi-scale fault identification result.

[0028] In a third aspect, the present application discloses an electronic device, comprising:

[0029] Memory, used to store computer programs;

[0030] A processor is used to execute the computer program to implement the aforementioned multi-scale fault identification method for three-dimensional seismic data.

[0031] In a fourth aspect, the present application discloses a computer storage medium for storing a computer program; wherein, when the computer program is executed by a processor, the steps of the multi-scale fault identification method for three-dimensional seismic data disclosed above are implemented.

[0032] It can be seen that the present application provides a multi-scale fault identification method for three-dimensional seismic data, including obtaining seismic wavelets, designing different geological models, generating seismic data based on different geological models and the seismic wavelets, performing three-dimensional enhancement expansion and normalization processing on the seismic data to obtain processed seismic data; constructing a two-branch convolutional neural network model, using the processed seismic data to train, test and optimize hyperparameters of the two-branch convolutional neural network model to obtain the optimized two-branch convolutional neural network model; obtaining actual seismic data, and inputting the actual seismic data into the optimized two-branch convolutional neural network model to obtain multi-scale fault identification results. The present application generates seismic data through different geological models and seismic wavelets, performs three-dimensional enhancement expansion and normalization processing on the seismic data, which can reduce possible deviations in the prediction process, ensure the continuity and integrity of fault information, construct a two-branch convolutional neural network model, and use the processed seismic data to train, test and optimize the hyperparameters of the two-branch convolutional neural network model. The core advantage of the two-branch convolutional neural network model lies in its unique two-branch architecture. While learning fault features with different characteristics, it can retain key fault edge information, improve the ability to distinguish between fault areas and non-fault areas in seismic data, input the actual seismic data into the optimized two-branch convolutional neural network model, and obtain multi-scale fault recognition results, which can solve the problem of insufficient feature extraction in the interpretation of seismic data faults, accurately identify and distinguish faults in seismic data, and improve the efficiency and accuracy of three-dimensional seismic data fault recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the embodiments of the present invention 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 embodiments of the present invention. For ordinary technicians in this field, other drawings can be drawn based on the provided drawings without paying creative work.

[0034] Figure 1 A flow chart of a multi-scale fault identification method for three-dimensional seismic data disclosed in this application;

[0035] Figure 2 A three-dimensional display diagram of processed seismic data disclosed in the present application;

[0036] Figure 3 A graph showing the accuracy change of a dual-branch convolutional neural network model training disclosed in this application;

[0037] Figure 4 A loss value change curve diagram disclosed in this application;

[0038] Figure 5 A multi-scale fault identification process and principle diagram disclosed in this application;

[0039] Figure 6 A fault identification effect diagram of three-dimensional seismic data disclosed in this application;

[0040] Figure 7 This is a schematic diagram of the structure of a multi-scale fault identification device for three-dimensional seismic data disclosed in this application;

[0041] Figure 8 A structural diagram of an electronic device provided for this application. DETAILED DESCRIPTION

[0042] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0043] Although artificial intelligence technology has made significant progress in the identification of earthquake fault information, it still faces some challenges in practical applications. First, the training of deep learning models requires a large amount of labeled data, and in the field of seismic data, the acquisition of this data is not only time-consuming but also costly. In addition, these models also have a high demand for computing resources. The limitations of synthetic data cannot be ignored. It may limit the generalization ability of network models under different geological conditions, and the model is prone to overfitting during the training process. In addition, the existing neural network algorithms have the problem of insufficient feature extraction in the interpretation of seismic data faults, resulting in low accuracy and efficiency in identifying faults in seismic data. As can be seen from the above, how to solve the problem of insufficient feature extraction in the interpretation of seismic data faults, accurately identify and distinguish faults in seismic data, and improve the efficiency and accuracy of three-dimensional seismic data fault identification are problems to be solved in this field.

[0044] See also Figure 1 As shown, the embodiment of the present invention discloses a multi-scale fault identification method for three-dimensional seismic data, which may specifically include:

[0045] Step S11: Acquire seismic wavelets, design different geological models, generate seismic data based on the different geological models and the seismic wavelets, perform three-dimensional enhancement expansion and normalization processing on the seismic data to obtain processed seismic data.

[0046] In this embodiment, seismic wavelets are obtained, and different geological models are designed. The different geological models are convolved with the seismic wavelets to generate seismic data. The seismic data are three-dimensionally enhanced and expanded using data enhancement technology to obtain a three-dimensional seismic data set. The three-dimensional seismic data set is normalized to obtain processed seismic data.

[0047] Specifically, different geological models are designed and convolved with seismic wavelets to obtain synthetic seismic data. The three-dimensional seismic data training data set is further expanded through data enhancement technology. The three-dimensional display of the processed seismic data is as follows: Figure 2 As shown, they are distributed on a uniform scale, which speeds up the convergence of model training and facilitates network processing.

[0048] Step S12: construct a dual-branch convolutional neural network model, and use the processed seismic data to train, test and optimize hyperparameters of the dual-branch convolutional neural network model to obtain the optimized dual-branch convolutional neural network model.

[0049] In this embodiment, a loss function and an optimizer are selected according to business needs, a two-branch convolutional neural network model is constructed, the hyperparameters in the two-branch convolutional neural network model are set using the loss function and the optimizer, and then a database and a Python development environment for training, testing and hyperparameter optimization are installed and configured, so that the processed seismic data can be used in the Python development environment to train, test and optimize the hyperparameters of the two-branch convolutional neural network model to obtain the optimized two-branch convolutional neural network model; the development environment includes TensorFlow and Keras.

[0050] Among them, the process of training, testing and hyperparameter optimization of the dual-branch convolutional neural network model is as follows: the processed seismic data is divided according to a preset data set division ratio to obtain training data and test data; the dual-branch convolutional neural network model is trained using the training data, and the trained dual-branch convolutional neural network model is tested using the test data; the indicator parameters in the trained and tested dual-branch convolutional neural network model are monitored in real time using the processed seismic data, and the hyperparameters in the trained and tested dual-branch convolutional neural network model are adjusted and optimized accordingly; the indicator parameters include accuracy and loss value; the hyperparameters include learning rate, number of iterations and optimization algorithm parameters.

[0051] Specifically, balance the complexity and performance of the network model. According to the available computing resources and training time limit, balance the complexity and performance of the network, determine the depth and parameters of the network to balance the complexity and performance of the network model, and ensure that the network can learn richer features within a reasonable training time; select the appropriate loss function and optimizer according to business needs. The loss function defines the optimization goal during the model training process and measures the difference between the model prediction and the actual result. Binary cross entropy is simple and practical, and is often used for classification problems. The optimizer is responsible for adjusting the parameters of the model during the training process, adaptively adjusting the learning rate, and minimizing the loss function; designing the architecture of the dual-branch convolutional neural network model, including the convolution layer, pooling layer, network depth, etc.; setting the parameters of each layer (convolution kernel size, step size, etc.), selecting the appropriate loss function and optimizer, designing the network module for information extraction, and designing the network module for information extraction. The two expansion paths on the dual-branch architecture can extract different fault features, thereby improving the performance of the network model; then configuring the environment, installing and developing necessary libraries, such as TensorFlow, Keras, etc., and configuring the GPU (Graphics Processing Unit) configuration, and configuring the Python development environment; then training the model, using the training data to train the neural network model, supervising indicators such as the loss function and accuracy during the training process; then testing the model, using the test data to evaluate the final performance of the model to ensure that the model can also perform well on unseen data.

[0052] The principle of the dual-branch convolutional neural network model is as follows: The convolutional neural network uses convolution operations to extract feature information, and the convolution features are controlled by information such as the number, stride, and padding of the convolution kernels. The convolution kernel slides on the image data to generate a feature map, where the calculation formula for the convolution operation is:

[0053] ;

[0054] Among them, T is the input feature map, (i, j) is the position of the feature map, K is the convolution kernel, (m, n) is the dimension of the convolution kernel, and O is the output feature map;

[0055] The activation function generally chooses ReLU (Rectified Linear Unit), which has high computational efficiency and is widely used in deep learning to alleviate the problem of gradient disappearance. When the input x is greater than 0, the output is equal to the input x; when the input x is less than or equal to 0, the output is 0.

[0056] ;

[0057] The present invention designs the architecture of the network model according to the rules of convolutional neural networks, improves the feature extraction effect of the neural network, achieves the accuracy and generalization performance of fault identification, adopts the U-Net network architecture as the basic architecture to segment the fault part in the seismic image, learns the nonlinear mapping relationship from seismic data to the target fault, accurately identifies the fault characteristics in the seismic data, and realizes end-to-end rapid prediction of fault information.

[0058] Neural networks generally use the back propagation algorithm, which is a supervised learning algorithm used to calculate the gradient of the neural network weights and update the weights to minimize the loss function. Back propagation will first perform forward propagation to calculate the output of the network. For a neural network with L layers, let is the activation input of layer l, is the weighted input of the lth layer, is the weight of the lth layer, is the bias of the lth layer, Is the activation function, then there is a formula:

[0059] ;

[0060] ;

[0061] Then calculate the loss. If y is the true label and C is the loss function, then for the Nth training sample, the loss formula is as follows:

[0062] ;

[0063] Next comes back propagation. The goal of back propagation is to calculate the gradient of the loss function with respect to each parameter. Starting from the last layer, the back propagation algorithm uses the chain rule to calculate the gradient. For the Lth layer (output layer), the gradient formula is as follows:

[0064] ;

[0065] ;

[0066] ;

[0067] For the lth layer ( ), the gradient formula is as follows:

[0068] ;

[0069] ;

[0070] ;

[0071] Then the weights are updated using the gradient descent algorithm:

[0072] ;

[0073] ;

[0074] in, is the learning rate;

[0075] The model automatically repeats the above steps until the stopping condition is met, such as the gradient is small enough or the maximum number of iterations is reached.

[0076] In this embodiment, the accuracy of the dual-branch convolutional neural network model training changes as follows: Figure 3 As shown in Figure 2, the loss value change curve is as follows: Figure 4 As shown in the figure, it can be seen that the changing trend of the curve conforms to the training law of the neural network. As the training progresses, the accuracy of the model gradually increases and the loss value gradually decreases, indicating that the network model is effectively learning fault feature information from the data. The relevant evaluation indicators are shown in Table 1. The AUC (Area Under Curve) value is the area under the ROC (Receiver Operating Characteristic curve). The larger the value, the higher the accuracy.

[0077] Table 1

[0078]

[0079] Step S13: Acquire actual seismic data, and input the actual seismic data into the optimized dual-branch convolutional neural network model to obtain multi-scale fault recognition results.

[0080] In this embodiment, actual seismic data is first obtained and input into an optimized dual-branch convolutional neural network model. The U-Net network architecture in the optimized dual-branch convolutional neural network model is used as a basic architecture to identify and segment the actual seismic data to obtain multi-scale fault features, and multi-scale fault identification results are generated based on the multi-scale fault features.

[0081] This application uses the PyCharm development environment to train a two-branch convolutional neural network model, and dynamically adjusts hyperparameters such as learning rate and optimization algorithm during the training process to optimize the performance of the model, and obtains an optimized two-branch convolutional neural network model. The optimized two-branch convolutional neural network model is applied to actual seismic data for fault identification and evaluation. The multi-scale fault identification process and principle are as follows Figure 5 As shown in the figure, the multi-scale fault recognition results are finally obtained. The fault recognition effect of 3D seismic data is shown in Figure 6 shown.

[0082] This application includes the following key steps: setting up the environment and preparing the data; designing a neural network architecture suitable for the fault segmentation task, setting hyperparameters such as the learning rate, number of iterations, and optimization algorithm; then using the three-dimensional enhanced, expanded, and normalized seismic data to train a dual-branch convolutional neural network model, and monitoring the accuracy and other indicators during the training process in real time; timely optimizing the model structure or hyperparameter configuration according to the results of the model test to improve the accuracy and robustness of the model; and finally applying the optimized dual-branch convolutional neural network model to the actual seismic data to obtain fault identification results. The advantage of the present invention is that it can extract multi-scale fault information, reduce possible deviations in the prediction process, and ensure the continuity and integrity of the fault information. Through this method, the efficiency and accuracy of the fault interpretation of three-dimensional seismic data can be improved, and faults in seismic data can be more accurately identified and distinguished, providing a new method for the structural interpretation of three-dimensional seismic data.

[0083] In this embodiment, firstly, seismic wavelets are obtained, and different geological models are designed. Seismic data are generated based on the different geological models and seismic wavelets. The seismic data is subjected to three-dimensional enhancement expansion and normalization processing to obtain processed seismic data; a two-branch convolutional neural network model is constructed, and the processed seismic data is used to train, test and optimize the hyperparameters of the two-branch convolutional neural network model to obtain the optimized two-branch convolutional neural network model; actual seismic data is obtained, and the actual seismic data is input into the optimized two-branch convolutional neural network model to obtain multi-scale fault identification results. This application generates seismic data through different geological models and seismic wavelets, performs three-dimensional enhancement expansion and normalization processing on the seismic data, can reduce the possible deviations in the prediction process, ensure the continuity and integrity of fault information, construct a two-branch convolutional neural network model, and use the processed seismic data to train, test and optimize the hyperparameters of the two-branch convolutional neural network model. The core advantage of this dual-branch convolutional neural network model lies in its unique dual-branch architecture. While learning fault features with different characteristics, it can retain key fault edge information, improve the ability to distinguish between fault areas and non-fault areas in seismic data, and input actual seismic data into the optimized dual-branch convolutional neural network model to obtain multi-scale fault recognition results. It can solve the problem of insufficient feature extraction in the interpretation of seismic data faults, accurately identify and distinguish faults in seismic data, and improve the efficiency and accuracy of fault recognition in three-dimensional seismic data.

[0084] See also Figure 7 As shown, the embodiment of the present invention discloses a multi-scale fault identification device for three-dimensional seismic data, which may specifically include:

[0085] The data processing module 11 is used to obtain seismic wavelets, design different geological models, generate seismic data based on the different geological models and the seismic wavelets, and perform three-dimensional enhancement, expansion and normalization processing on the seismic data to obtain processed seismic data;

[0086] A model optimization module 12 is used to construct a dual-branch convolutional neural network model, and use the processed seismic data to train, test and optimize hyperparameters of the dual-branch convolutional neural network model to obtain an optimized dual-branch convolutional neural network model;

[0087] The multi-scale fault identification module 13 is used to obtain actual seismic data and input the actual seismic data into the optimized dual-branch convolutional neural network model to obtain a multi-scale fault identification result.

[0088] In this embodiment, seismic wavelets are obtained, and different geological models are designed. Seismic data are generated based on the different geological models and seismic wavelets. The seismic data is subjected to three-dimensional enhancement expansion and normalization processing to obtain processed seismic data; a two-branch convolutional neural network model is constructed, and the processed seismic data is used to train, test and optimize the hyperparameters of the two-branch convolutional neural network model to obtain the optimized two-branch convolutional neural network model; actual seismic data is obtained, and the actual seismic data is input into the optimized two-branch convolutional neural network model to obtain multi-scale fault identification results. This application generates seismic data through different geological models and seismic wavelets, performs three-dimensional enhancement expansion and normalization processing on the seismic data, can reduce the possible deviations in the prediction process, ensure the continuity and integrity of fault information, construct a two-branch convolutional neural network model, and use the processed seismic data to train, test and optimize the hyperparameters of the two-branch convolutional neural network model. The core advantage of this dual-branch convolutional neural network model lies in its unique dual-branch architecture. While learning fault features with different characteristics, it can retain key fault edge information, improve the ability to distinguish between fault areas and non-fault areas in seismic data, and input actual seismic data into the optimized dual-branch convolutional neural network model to obtain multi-scale fault recognition results. It can solve the problem of insufficient feature extraction in the interpretation of seismic data faults, accurately identify and distinguish faults in seismic data, and improve the efficiency and accuracy of fault recognition in three-dimensional seismic data.

[0089] In some specific embodiments, the data processing module 11 may specifically include:

[0090] A seismic data generation module, used for convolving different geological models with the seismic wavelets respectively to generate seismic data;

[0091] The three-dimensional enhancement expansion and normalization processing module is used to perform three-dimensional enhancement expansion on the seismic data using data enhancement technology to obtain a three-dimensional seismic data set, and perform normalization processing on the three-dimensional seismic data set.

[0092] In some specific embodiments, the model optimization module 12 may specifically include:

[0093] The screening module is used to screen loss functions and optimizers according to business needs;

[0094] A setting module is used to construct a two-branch convolutional neural network model, and use the loss function and the optimizer to set the hyperparameters in the two-branch convolutional neural network model.

[0095] In some specific embodiments, the model optimization module 12 may specifically include:

[0096] The installation and configuration module is used to install and configure a database and a Python development environment for training, testing and hyperparameter optimization, so that the processed seismic data can be used to train, test and optimize the hyperparameters of the dual-branch convolutional neural network model in the Python development environment; the development environment includes TensorFlow and Keras.

[0097] In some specific embodiments, the model optimization module 12 may specifically include:

[0098] A data partitioning module, used to partition the processed seismic data according to a preset data set partitioning ratio to obtain training data and test data;

[0099] A testing module is used to train the two-branch convolutional neural network model using the training data, and to test the trained two-branch convolutional neural network model using the test data.

[0100] In some specific embodiments, the model optimization module 12 may specifically include:

[0101] A hyperparameter adjustment and optimization module is used to use the processed seismic data to monitor the indicator parameters in the two-branch convolutional neural network model after training and testing in real time, and to adjust and optimize the hyperparameters in the two-branch convolutional neural network model after training and testing accordingly; the indicator parameters include accuracy and loss value; the hyperparameters include learning rate, number of iterations and optimization algorithm parameters.

[0102] In some specific embodiments, the multi-scale fault identification module 13 may specifically include:

[0103] The recognition result generation module is used to input the actual seismic data into the optimized dual-branch convolutional neural network model, and use the U-Net network architecture in the optimized dual-branch convolutional neural network model as the basic architecture to recognize and segment the actual seismic data to obtain multi-scale fault characteristics, and generate multi-scale fault recognition results based on the multi-scale fault characteristics.

[0104] Figure 8 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device 20 may specifically include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 is used to store a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the multi-scale fault identification method for three-dimensional seismic data performed by the electronic device disclosed in any of the aforementioned embodiments.

[0105] In this embodiment, the power supply 23 is used to power various hardware devices on the electronic device 20; the communication interface 24 is used to create a data transmission channel between the electronic device 20 and external devices, and the communication protocol it follows is any communication protocol that can be applied to the technical solution of the present application, and is not specifically limited here; the input and output interface 25 is used to obtain external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs and is not specifically limited here.

[0106] In addition, the memory 22, as a carrier for storing resources, can be a read-only memory, a random access memory, a disk or an optical disk, etc. The resources stored thereon include an operating system 221, a computer program 222 and data 223, etc. The storage method can be temporary storage or permanent storage.

[0107] The operating system 221 is used to manage and control the hardware devices and computer programs 222 on the electronic device 20, so as to realize the operation and processing of the data 223 in the memory 22 by the processor 21, which can be Windows, Unix, Linux, etc. In addition to including a computer program that can be used to complete the multi-scale fault identification method of three-dimensional seismic data performed by the electronic device 20 disclosed in any of the aforementioned embodiments, the computer program 222 can also further include a computer program that can be used to complete other specific tasks. In addition to including data transmitted from an external device received by the multi-scale fault identification device for three-dimensional seismic data, the data 223 can also include data collected by its own input and output interface 25, etc.

[0108] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0109] Furthermore, an embodiment of the present application also discloses a computer-readable storage medium, in which a computer program is stored. When the computer program is loaded and executed by a processor, the steps of the multi-scale fault identification method for three-dimensional seismic data disclosed in any of the aforementioned embodiments are implemented.

[0110] Finally, it should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0111] The above is a detailed introduction to the multi-scale fault identification method, device, equipment and storage medium for three-dimensional seismic data provided by the present invention. Specific examples are used in this article to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core idea; at the same time, for general technical personnel in this field, according to the idea of ​​the present invention, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present invention.

Claims

1. A multi-scale fault identification method for three-dimensional seismic data, characterized in that: include: Acquire seismic wavelets, design different geological models, generate seismic data based on the different geological models and the seismic wavelets, perform three-dimensional enhancement and expansion and normalization processing on the seismic data to obtain processed seismic data; Constructing a dual-branch convolutional neural network model, and using the processed seismic data to train, test and optimize hyperparameters of the dual-branch convolutional neural network model to obtain an optimized dual-branch convolutional neural network model; Actual seismic data is acquired and input into the optimized dual-branch convolutional neural network model to obtain multi-scale fault identification results.

2. The multi-scale fault identification method for three-dimensional seismic data according to claim 1, characterized in that: The generating of seismic data based on the different geological models and the seismic wavelets, and performing three-dimensional enhancement, expansion and normalization processing on the seismic data include: convolving different geological models with the seismic wavelets respectively to generate seismic data; The seismic data is three-dimensionally enhanced and expanded by using data enhancement technology to obtain a three-dimensional seismic data set, and the three-dimensional seismic data set is normalized.

3. The multi-scale fault identification method for three-dimensional seismic data according to claim 1, characterized in that: The dual-branch convolutional neural network model is constructed, comprising: Select loss functions and optimizers based on business needs; A two-branch convolutional neural network model is constructed, and the hyperparameters in the two-branch convolutional neural network model are set using the loss function and the optimizer.

4. The multi-scale fault identification method for three-dimensional seismic data according to claim 1, characterized in that: The method of using the processed seismic data to train, test and optimize hyperparameters of the dual-branch convolutional neural network model includes: Install and configure a database and a Python development environment for training, testing, and hyperparameter optimization, so that the processed seismic data can be used to train, test, and optimize the hyperparameters of the dual-branch convolutional neural network model in the Python development environment; the development environment includes TensorFlow and Keras.

5. The multi-scale fault identification method for three-dimensional seismic data according to claim 1, characterized in that: The processed seismic data is used to train and test the dual-branch convolutional neural network model, including: Dividing the processed seismic data according to a preset data set division ratio to obtain training data and test data; The two-branch convolutional neural network model is trained using the training data, and the trained two-branch convolutional neural network model is tested using the test data.

6. The multi-scale fault identification method for three-dimensional seismic data according to claim 1, characterized in that: The processed seismic data is used to perform hyperparameter optimization on the trained and tested dual-branch convolutional neural network model, including: The processed seismic data is used to monitor the index parameters in the two-branch convolutional neural network model after training and testing in real time, and the hyperparameters in the two-branch convolutional neural network model after training and testing are adjusted and optimized accordingly; the index parameters include accuracy and loss value; the hyperparameters include learning rate, number of iterations and optimization algorithm parameters.

7. The multi-scale fault identification method for three-dimensional seismic data according to any one of claims 1 to 6, characterized in that: The inputting the actual seismic data into the optimized dual-branch convolutional neural network model to obtain a multi-scale fault identification result includes: The actual seismic data is input into the optimized two-branch convolutional neural network model, and the U-Net network architecture in the optimized two-branch convolutional neural network model is used as the basic architecture to identify and segment the actual seismic data to obtain multi-scale fault characteristics, and generate multi-scale fault identification results based on the multi-scale fault characteristics.

8. A multi-scale fault identification device for three-dimensional seismic data, characterized in that: include: A data processing module is used to obtain seismic wavelets, design different geological models, generate seismic data based on the different geological models and the seismic wavelets, and perform three-dimensional enhancement, expansion and normalization processing on the seismic data to obtain processed seismic data; A model optimization module, used for constructing a dual-branch convolutional neural network model, and using the processed seismic data to train, test and optimize hyperparameters of the dual-branch convolutional neural network model to obtain an optimized dual-branch convolutional neural network model; The multi-scale fault identification module is used to obtain actual seismic data and input the actual seismic data into the optimized dual-branch convolutional neural network model to obtain a multi-scale fault identification result.

9. An electronic device, characterized in that: include: Memory, used to store computer programs; A processor is used to execute the computer program to implement the multi-scale fault identification method for three-dimensional seismic data as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: Used to store a computer program; wherein, when the computer program is executed by a processor, the multi-scale fault identification method for three-dimensional seismic data as described in any one of claims 1 to 7 is implemented.

Images