DAS-VSP wave field separation method and device based on convolution automatic encoder
The DAS-VSP wavefield separation is performed based on a convolutional autoencoder method, which solves the problem of poor wavefield separation in complex geological environments and achieves higher-precision wavefield separation and improved seismic data imaging quality.
Patent Information
- Application Number
- CN202510685081.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-26
- Publication Date
- 2025-09-26
AI Technical Summary
The existing DAS-VSP wavefield separation method is not effective in complex geological environments, easily causes data spatial mixing, and is not effective in handling the overlap of noise and signal spectrum ranges.
A convolutional autoencoder-based method is adopted to construct a convolutional autoencoder model through data preprocessing, and use the training set and test set for wavefield separation. This includes static correction, energy compensation, noise suppression, and the design of convolutional layers, decoding layers, and regularization layers to constrain the directional continuity and zero-mean characteristics of upgoing and downgoing waves.
It improves the accuracy of wave field separation, reduces manual intervention, adapts to complex geological conditions, and significantly improves the imaging quality and interpretation efficiency of seismic data.
Smart Images
Figure CN120703828A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of wavefield separation technology, and more specifically, to a DAS-VSP wavefield separation method and device based on convolutional autoencoder. Background Art
[0002] Distributed fiber acoustic sensing (DAS) technology is becoming an important seismic observation technology due to its low cost, easy deployment, and high-density sampling. Traditional methods for VSP wavefield separation include median filtering and fk decomposition, but they are not suitable for complex geological environments and require a large number of manual steps. The traditional fk filtering method uses the difference in apparent velocity between interference waves and effective waves to perform wavefield separation in the frequency-wavenumber domain. However, this method may cause spatial mixing of data during processing, resulting in changes in the recorded appearance and effective wave characteristics. Median filtering is a nonlinear filtering method that is effective in handling overlap between noise and signal spectral ranges, but it requires strict wavefield alignment. Inconsistent waveforms between depth channels can significantly affect the separation effect.
[0003] Therefore, a better solution is urgently needed. Summary of the Invention
[0004] The present invention provides a DAS-VSP wavefield separation method and device based on a convolutional autoencoder to solve the technical problems of the prior art as described in the background technology. The method includes:
[0005] Acquire DAS-VSP data and pre-process the DAS-VSP data to obtain data to be separated by wavefields, wherein the data to be separated by wavefields include a training set and a test set;
[0006] Constructing a convolutional autoencoder, and training the convolutional autoencoder using the training set to obtain a convolutional autoencoder model;
[0007] The test set is subjected to wavefield separation using the convolutional autoencoder model to obtain upgoing waves and downgoing waves.
[0008] In some specific embodiments, DAS-VSP data is acquired and preprocessed to obtain data to be separated by wavefields. The data to be separated by wavefields include a training set and a test set, specifically:
[0009] performing static correction on the DAS-VSP data to eliminate static correction time difference in the DAS-VSP data;
[0010] performing energy compensation on the DAS-VSP data after static correction to obtain the energy-compensated DAS-VSP data;
[0011] performing noise suppression on the energy-compensated DAS-VSP data to reduce a signal-to-noise ratio of the energy-compensated DAS-VSP data;
[0012] The noise-suppressed DAS-VSP data are divided into training set and test set.
[0013] In some specific embodiments, a convolutional autoencoder is constructed, specifically by constructing an encoder, a decoder, and a regularization layer of the convolutional autoencoder, wherein:
[0014] The encoder specifically includes a plurality of convolutional layers and pooling layers, wherein the convolutional layers are used to perform convolution operations on local areas of the signal input to the encoder;
[0015] The decoder specifically includes a deconvolution layer, and reconstructs the signal input to the encoder after convolution calculation through the deconvolution layer;
[0016] The regularization layer includes two regularizers. The regularization layer is arranged between the encoder and the decoder and is used to constrain the directional continuity and zero mean characteristics of the uplink wave and the downlink wave.
[0017] In some specific embodiments, the convolution layer is used to perform a convolution operation on a local area of the signal input to the encoder by using a convolution operation formula, and the convolution operation formula is specifically:
[0018]
[0019] Among them, Y i is the output feature of the i-th convolutional layer, Yi-1 is the output feature of the i-1-th convolutional layer, represents the convolution operation, W i represents the weight vector connecting the i-th layer and the i-1-th layer, b i represents the bias vector of the i-th layer, and F represents the nonlinear activation function.
[0020] In some specific embodiments, the convolutional autoencoder also includes an excitation layer, which is arranged in the encoder. The excitation layer is specifically a nonlinear activation function, and the nonlinear activation function specifically selects any one of the Sigmoid function, Relu function or Tanh function.
[0021] In some specific embodiments, the convolutional autoencoder is trained using the training set to obtain a convolutional autoencoder model, specifically:
[0022] The reconstruction error is defined as a loss function, and the Adam algorithm or the SGD algorithm is used to minimize the loss function. The convolutional autoencoder is trained using the training set. During the training process, back propagation and gradient descent algorithms are used.
[0023] In some specific embodiments, the method further comprises:
[0024] The upgoing wave and the downgoing wave obtained by the convolutional autoencoder model are compared with the actual upgoing wave and the downgoing wave to obtain comparison indicators, which include signal-to-noise ratio and mean square error.
[0025] Accordingly, the present invention also proposes a DAS-VSP wavefield separation device based on a convolutional autoencoder, the device comprising:
[0026] A preprocessing module is used to obtain DAS-VSP data and preprocess the DAS-VSP data to obtain data to be separated by wavefields, wherein the data to be separated by wavefields include a training set and a test set;
[0027] A model construction module is used to construct a convolutional autoencoder and train the convolutional autoencoder using the training set to obtain a convolutional autoencoder model;
[0028] The wavefield separation module is used to perform wavefield separation on the test set through the convolutional autoencoder model to obtain upgoing waves and downgoing waves.
[0029] One embodiment of the present invention further provides a computing device, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions, which, when executed by the processor, implement the steps of the DAS-VSP wavefield separation method based on convolutional autoencoder as described above.
[0030] One embodiment of the present invention further provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the DAS-VSP wavefield separation method based on convolutional autoencoder as described above.
[0031] By applying the above technical solution, a DAS-VSP wavefield separation method based on a convolutional autoencoder is proposed. The method comprises: acquiring DAS-VSP data and preprocessing the DAS-VSP data to obtain data to be separated by wavefields, wherein the data to be separated by wavefields include a training set and a test set; constructing a convolutional autoencoder and training the convolutional autoencoder using the training set to obtain a convolutional autoencoder model; performing wavefield separation on the test set using the convolutional autoencoder model to obtain upgoing waves and downgoing waves. By constructing the convolutional autoencoder model, wavefield separation of DAS-VSP data under complex conditions is achieved, and the accuracy of the wavefield separation results is improved. BRIEF DESCRIPTION OF THE DRAWINGS
[0032] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.
[0033] Figure 1 This is a flow chart of a DAS-VSP wavefield separation method based on a convolutional autoencoder provided in an embodiment of the present application;
[0034] Figure 2 Schematic diagram of the structure of a DAS-VSP wave field separation device based on a convolutional autoencoder provided in an embodiment of the present application;
[0035] Figure 3 This is a structural block diagram of a computing device provided in an embodiment of the present application;
[0036] Figure 4 is a schematic diagram of an original wave field provided in an embodiment of the present application;
[0037] Figure 5 is a schematic diagram of an upgoing wavefield obtained after wavefield separation provided by an embodiment of the present application;
[0038] Figure 6 This is a schematic diagram of a downlink wavefield obtained after wavefield separation provided in an embodiment of the present application. DETAILED DESCRIPTION
[0039] The following description sets forth many specific details to facilitate a thorough understanding of this specification. However, this specification can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the scope of this specification. Therefore, this specification is not limited to the specific implementations disclosed below.
[0040] The terms used in one or more embodiments of this specification are for the purpose of describing specific embodiments only and are not intended to limit one or more embodiments of this specification. The singular forms "a," "an," and "the" used in one or more embodiments of this specification and the appended claims are also intended to include plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in one or more embodiments of this specification refers to and includes any or all possible combinations of one or more associated listed items.
[0041] It should be understood that although the terms first, second, etc. may be used to describe various information in one or more embodiments of this specification, such information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of one or more embodiments of this specification, the first may also be referred to as the second, and similarly, the second may also be referred to as the first. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".
[0042] like Figure 1 As shown, the present application proposes a DAS-VSP wavefield separation method based on a convolutional autoencoder, the method comprising the following steps:
[0043] Step S101 : acquiring DAS-VSP data and preprocessing the DAS-VSP data to obtain data to be separated by wavefields, wherein the data to be separated by wavefields include a training set and a test set.
[0044] In a possible implementation, DAS-VSP data is acquired and preprocessed to obtain data to be separated by wavefields. The data to be separated by wavefields include a training set and a test set. Specifically,
[0045] performing static correction on the DAS-VSP data to eliminate static correction time difference in the DAS-VSP data;
[0046] performing energy compensation on the DAS-VSP data after static correction to obtain the energy-compensated DAS-VSP data;
[0047] performing noise suppression on the energy-compensated DAS-VSP data to reduce a signal-to-noise ratio of the energy-compensated DAS-VSP data;
[0048] The noise-suppressed DAS-VSP data are divided into training set and test set.
[0049] In this embodiment, static correction is performed on the DAS-VSP data to eliminate static correction time differences caused by terrain undulations and uneven ground surfaces, thereby ensuring temporal consistency of the data.
[0050] In this embodiment, energy compensation is performed on the data to adjust the signal energy at different depths and positions so that the amplitudes remain consistent, thereby reducing energy attenuation caused by differences in propagation paths.
[0051] In this embodiment, filtering, denoising and other technologies are used to remove background noise and random interference, thereby improving the signal-to-noise ratio of the signal.
[0052] In this embodiment, the pre-processed data is divided into a training set and a test set. The training set is used for model training, and the test set is used to verify the performance and generalization ability of the model.
[0053] Step S102: construct a convolutional autoencoder, and train the convolutional autoencoder using the training set to obtain a convolutional autoencoder model.
[0054] In one possible implementation, a convolutional autoencoder is constructed, specifically by constructing an encoder, a decoder, and a regularization layer of the convolutional autoencoder, wherein:
[0055] The encoder specifically includes a plurality of convolutional layers and pooling layers, wherein the convolutional layers are used to perform convolution operations on local areas of the signal input to the encoder;
[0056] The decoder specifically includes a deconvolution layer, and reconstructs the signal input to the encoder after convolution calculation through the deconvolution layer;
[0057] The regularization layer includes two regularizers. The regularization layer is arranged between the encoder and the decoder and is used to constrain the directional continuity and zero mean characteristics of the uplink wave and the downlink wave.
[0058] In one possible implementation, the convolution layer is used to perform a convolution operation on a local area of the signal input to the encoder by using a convolution operation formula, and the convolution operation formula is specifically:
[0059]
[0060] Among them, Y i is the output feature of the i-th convolutional layer, represents the convolution operation, W i represents the weight vector connecting the i-th layer and the i-1-th layer, b i represents the bias vector of the i-th layer, and F represents the nonlinear activation function.
[0061] In this embodiment, the network structure design of the convolutional autoencoder includes:
[0062] (1) Encoder: Construct convolutional layers to gradually reduce the spatial dimension of the data while extracting the feature representation of the data. For example, multiple convolutional layers and pooling layers are used to gradually compress the dimension of the data.
[0063] (2) Decoder: Constructs deconvolution layers to gradually restore the spatial dimensions of the data and reconstruct the original data. The decoder structure is usually symmetrical with the encoder.
[0064] The convolution layer mainly uses the convolution kernel to perform convolution operations with the local area of the input signal to extract the local area features of the input signal. Unlike fully connected neural networks, the convolution layer has the characteristics of local connection and weight sharing, so it can use fewer parameters to obtain richer features. The calculation formula of the convolution layer is:
[0065]
[0066] Where Y i The output feature of the i-th convolutional layer, Yi-1 is the output feature of the i-1-th convolutional layer, represents the convolution operation, W i represents the weight vector connecting the i-th layer and the i-1-th layer, b i represents the bias vector of the i-th layer, and F represents the nonlinear activation function.
[0067] In this example, a regularization layer, such as a Dropout layer or a BatchNormalization layer, is added between the encoder and decoder to improve the model's generalization capabilities. The regularization layer prevents overfitting and ensures good performance on unseen data.
[0068] In this embodiment, two regularizers are added between the encoder and decoder to constrain the directional continuity and zero-mean characteristics of the upgoing and downgoing waves, respectively. These regularizers help the model learn more accurate wavefield separation features.
[0069] In a possible implementation, the convolutional autoencoder further includes an excitation layer, which is arranged in the encoder. The excitation layer is specifically a nonlinear activation function, and the nonlinear activation function specifically selects any one of the Sigmoid function, the Relu function or the Tanh function.
[0070] In this embodiment, the encoding part is a superposition of three convolutional layers and two excitation function layers. The mapping after the convolution operation will be stimulated by the excitation layer below. This layer is a nonlinear activation function. By introducing nonlinear factors into the model, the problem of insufficient expression ability of the linear model is solved. The activation function can retain the features and map them to the next layer. Common activation functions include Sigmoid function, Relu function and Tanh function. This patent selects the Relu function, which can be expressed by the following formula:
[0071] Relu(x)=max(0,x);
[0072] Where x represents the input.
[0073] Step S103 : performing wavefield separation on the test set using the convolutional autoencoder model to obtain upgoing waves and downgoing waves.
[0074] In one possible implementation, the convolutional autoencoder is trained using the training set to obtain a convolutional autoencoder model, specifically:
[0075] The reconstruction error is defined as a loss function, and the Adam algorithm or the SGD algorithm is used to minimize the loss function. The convolutional autoencoder is trained using the training set. During the training process, back propagation and gradient descent algorithms are used.
[0076] In this example, the Marmousi-II model was used for forward numerical simulation of synthetic VSP data and wavefield separation, totaling 2,500 images. A separate 500 images were prepared as a test set. A convolutional autoencoder was trained using the training set data, and model parameters were gradually adjusted through backpropagation and gradient descent algorithms, enabling the model to better learn the data's feature representation and reconstruction capabilities.
[0077] In this embodiment, during training, the directional continuity and zero-mean characteristics of upgoing and downgoing waves are exploited and constrained by two regularizers to achieve wavefield separation. This step ensures that the model can accurately distinguish upgoing and downgoing waves.
[0078] In a possible implementation, the method further includes:
[0079] The upgoing wave and the downgoing wave obtained by the convolutional autoencoder model are compared with the actual upgoing wave and the downgoing wave to obtain comparison indicators, which include signal-to-noise ratio and mean square error.
[0080] In this embodiment, the performance of the model is evaluated by comparing the separation results with the actual data. Various evaluation indicators, such as signal-to-noise ratio (SNR) and mean square error (MSE), can be used to measure the accuracy and reliability of the separation results. The effectiveness of the separation results is verified by numerical simulation and actual data to ensure that the model can accurately reflect the underground geological structure. The original wave field before separation is as follows: Figure 4 As shown, the upgoing wavefield and the downgoing wavefield after wavefield separation are as follows Figure 5 and Figure 6 shown.
[0081] In summary, this paper proposes a DAS-VSP wavefield separation method based on a convolutional autoencoder. The DAS-VSP data is divided into a training set and a test set. A convolutional autoencoder network is constructed, consisting of an encoder and a decoder. The encoder extracts data features, while the decoder reconstructs the data. A regularization layer is added between the layers to enhance generalization. Wavefield separation is achieved by exploiting the directional continuity and zero-mean characteristics of upgoing and downgoing waves and constraining the model with a regularizer. This method effectively separates upgoing and downgoing waves, improves separation accuracy, reduces manual intervention, adapts to complex geological conditions, and significantly enhances seismic data imaging quality and interpretation efficiency.
[0082] The present application also proposes a DAS-VSP wave field separation device based on a convolutional autoencoder, such as Figure 2 As shown, the device includes:
[0083] A preprocessing module 10 is used to acquire DAS-VSP data and preprocess the DAS-VSP data to obtain data to be separated by wavefields, wherein the data to be separated by wavefields include a training set and a test set;
[0084] A model construction module 20 is used to construct a convolutional autoencoder and train the convolutional autoencoder using the training set to obtain a convolutional autoencoder model;
[0085] The wavefield separation module 30 is configured to perform wavefield separation on the test set using the convolutional autoencoder model to obtain upgoing waves and downgoing waves.
[0086] Figure 3 The block diagram of a computing device 400 according to one embodiment of the present disclosure is shown. Components of the computing device 400 include, but are not limited to, a memory 410 and a processor 420. The processor 420 is connected to the memory 410 via a bus 430, and a database 450 is used to store data.
[0087] The computing device 400 also includes an access device 440 that enables the computing device 400 to communicate via one or more networks 460. Examples of such networks include a public switched telephone network (PSTN), a local area network (LAN), a wide area network (WAN), a personal area network (PAN), or a combination of communication networks such as the Internet. The access device 440 may include one or more of any type of network interface (e.g., a network interface card (NIC)) whether wired or wireless, such as an IEEE 802.11 wireless local area network (WLAN) wireless interface, a Worldwide Interoperability for Microwave Access (Wi-MAX) interface, an Ethernet interface, a universal serial bus (USB) interface, a cellular network interface, a Bluetooth interface, or a near field communication (NFC) interface.
[0088] In one embodiment of the present specification, the above components of the computing device 400 and Figure 3 Other components not shown in the figure may also be connected to each other, for example, via a bus. Figure 3 The computing device structure block diagram shown is for illustrative purposes only and is not intended to limit the scope of this specification. Those skilled in the art may add or replace other components as needed.
[0089] Computing device 400 may be any type of stationary or mobile computing device, including a mobile computer or mobile computing device (e.g., a tablet computer, a personal digital assistant, a laptop computer, a notebook computer, a netbook computer, etc.), a mobile phone (e.g., a smartphone), a wearable computing device (e.g., a smartwatch, smart glasses, etc.), or other types of mobile devices, or a stationary computing device such as a desktop computer or a personal computer (PC). Computing device 400 may also be a mobile or stationary server.
[0090] The processor 420 is configured to execute the following computer-executable instructions, which, when executed by the processor, implement the steps of the aforementioned DAS-VSP wavefield separation method based on a convolutional autoencoder. The above is a schematic diagram of a computing device according to this embodiment. It should be noted that the technical solution of this computing device and the technical solution of the aforementioned DAS-VSP wavefield separation method based on a convolutional autoencoder are based on the same concept. For details not described in detail in the technical solution of the computing device, please refer to the description of the technical solution of the aforementioned DAS-VSP wavefield separation method based on a convolutional autoencoder.
[0091] An embodiment of the present specification further provides a computer-readable storage medium storing computer-executable instructions. When the computer-executable instructions are executed by a processor, the steps of the above-mentioned DAS-VSP wavefield separation method based on convolutional autoencoder are implemented.
[0092] The above is an illustrative embodiment of a computer-readable storage medium. It should be noted that the technical solution of this storage medium shares the same concept as the technical solution of the DAS-VSP wavefield separation method based on a convolutional autoencoder. For details not described in detail in the technical solution of the storage medium, please refer to the description of the technical solution of the DAS-VSP wavefield separation method based on a convolutional autoencoder.
[0093] An embodiment of the present specification further provides a computer program, wherein when the computer program is executed in a computer, the computer is instructed to perform the steps of the above-mentioned DAS-VSP wavefield separation method based on convolutional autoencoder.
[0094] The above is an illustrative embodiment of a computer program. It should be noted that the technical solution of this computer program shares the same concept as the technical solution of the DAS-VSP wavefield separation method based on a convolutional autoencoder. For details not described in detail in the technical solution of the computer program, please refer to the description of the technical solution of the DAS-VSP wavefield separation method based on a convolutional autoencoder.
[0095] The foregoing description of this specification describes specific embodiments. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in an order different from that described in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or the sequential order to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0096] The computer instructions include computer program code, which may be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal, and software distribution medium, etc. It should be noted that the content contained in the computer-readable medium may be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.
[0097] It should be noted that for the aforementioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the embodiments of this specification are not limited by the order of the actions described, because according to the embodiments of this specification, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the embodiments of this specification.
[0098] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.
[0099] The preferred embodiments disclosed above are intended only to help illustrate this specification. The optional embodiments do not exhaustively describe all details, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and variations can be made based on the content of the embodiments of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the embodiments of this specification, so that those skilled in the art can better understand and utilize this specification. This specification is limited only by the claims and their full scope and equivalents.
Claims
1. A DAS-VSP wavefield separation method based on convolutional autoencoder, characterized in that: The method comprises: Acquire DAS-VSP data and pre-process the DAS-VSP data to obtain data to be separated by wavefields, wherein the data to be separated by wavefields include a training set and a test set; Constructing a convolutional autoencoder, and training the convolutional autoencoder using the training set to obtain a convolutional autoencoder model; The test set is subjected to wavefield separation using the convolutional autoencoder model to obtain upgoing waves and downgoing waves.
2. The method according to claim 1, characterized in that DAS-VSP data is acquired and preprocessed to obtain data to be separated by wavefields. The data to be separated by wavefields include a training set and a test set, specifically: performing static correction on the DAS-VSP data to eliminate static correction time difference in the DAS-VSP data; performing energy compensation on the DAS-VSP data after static correction to obtain the energy-compensated DAS-VSP data; performing noise suppression on the energy-compensated DAS-VSP data to reduce a signal-to-noise ratio of the energy-compensated DAS-VSP data; The noise-suppressed DAS-VSP data are divided into training set and test set.
3. The method according to claim 2, characterized in that Construct a convolutional autoencoder, specifically constructing the encoder, decoder and regularization layer of the convolutional autoencoder, wherein: The encoder specifically includes a plurality of convolutional layers and pooling layers, wherein the convolutional layers are used to perform convolution operations on local areas of the signal input to the encoder; The decoder specifically includes a deconvolution layer, and reconstructs the signal input to the encoder after convolution calculation through the deconvolution layer; The regularization layer includes two regularizers. The regularization layer is arranged between the encoder and the decoder and is used to constrain the directional continuity and zero mean characteristics of the uplink wave and the downlink wave.
4. The method according to claim 3, characterized in that The convolution layer is used to perform a convolution operation on a local area of the signal input to the encoder through a convolution operation formula, and the convolution operation formula is specifically: Among them, Y i is the output feature of the i-th convolutional layer, Yi-1 is the output feature of the i-1-th convolutional layer, represents the convolution operation, W i represents the weight vector connecting the i-th layer and the i-1-th layer, b i represents the bias vector of the i-th layer, and F represents the nonlinear activation function.
5. The method according to claim 3, characterized in that The convolutional autoencoder also includes an excitation layer, which is arranged in the encoder. The excitation layer is specifically a nonlinear activation function, and the nonlinear activation function specifically selects any one of the Sigmoid function, the Relu function or the Tanh function.
6. The method according to claim 5, characterized in that The convolutional autoencoder is trained using the training set to obtain a convolutional autoencoder model, specifically: The reconstruction error is defined as a loss function, and the Adam algorithm or the SGD algorithm is used to minimize the loss function. The convolutional autoencoder is trained using the training set. During the training process, back propagation and gradient descent algorithms are used.
7. The method according to claim 6, characterized in that The method further comprises: The upgoing wave and the downgoing wave obtained by the convolutional autoencoder model are compared with the actual upgoing wave and the downgoing wave to obtain comparison indicators, which include signal-to-noise ratio and mean square error.
8. A DAS-VSP wavefield separation device based on convolutional autoencoder, characterized in that: The device comprises: A preprocessing module is used to obtain DAS-VSP data and preprocess the DAS-VSP data to obtain data to be separated by wavefields, wherein the data to be separated by wavefields include a training set and a test set; A model construction module is used to construct a convolutional autoencoder and train the convolutional autoencoder using the training set to obtain a convolutional autoencoder model; The wavefield separation module is used to perform wavefield separation on the test set through the convolutional autoencoder model to obtain upgoing waves and downgoing waves.
9. A computing device, characterized in that include: memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions. When the computer-executable instructions are executed by the processor, the steps of the DAS-VSP wavefield separation method based on convolutional autoencoder according to any one of claims 1 to 7 are implemented.
10. A computer-readable storage medium, characterized in that It stores computer-executable instructions, which, when executed by a processor, implement the steps of the DAS-VSP wavefield separation method based on convolutional autoencoder according to any one of claims 1 to 7.