Seismic data feature extraction and clustering method and device based on deep autoencoder

Abstract

The present invention provides a method and device for extracting and clustering seismic data features based on a deep autoencoder, the method comprising: acquiring seismic data, extracting inter-layer data between target layers according to the seismic data and target layer information; bringing the inter-layer data into a deep autoencoder network for unsupervised training to obtain a feature model; extracting feature data from the seismic data using the feature model; performing cluster analysis on the feature data to obtain feature points, and restoring the feature points to the original two-dimensional plane to obtain waveform clustering results.

Description

Technical Field

[0001] The present invention relates to seismic signal processing and interpretation technology, and in particular to a seismic data feature extraction and clustering method and device based on a deep autoencoder. Background Art

[0002] Traditionally, when performing waveform clustering, a time window is generally selected along the layer in the time domain, and a section of data is taken out for waveform clustering, or Fourier transform is performed on it in the frequency domain for clustering analysis. However, the data waveform taken out by the selected time window may not be representative and is often not a true representation of the target layer; at the same time, there are too many feature points in the single-channel waveform data taken out, and there is serious data redundancy, which increases the instability and complexity of the waveform clustering algorithm; the spectrum method is easily affected by noise and has a greater impact on data with low signal-to-noise ratio. Summary of the invention

[0003] The purpose of the present invention is to provide a method and device for seismic data feature extraction and clustering based on a deep autoencoder, so as to realize seismic data compression and feature extraction which are important bases for seismic phase analysis, and improve the waveform clustering effect.

[0004] To achieve the above-mentioned purpose, the present invention provides a seismic data feature extraction and clustering method based on a deep autoencoder, which specifically includes: acquiring seismic data, extracting inter-layer data between target layers according to the seismic data and target layer information; bringing the inter-layer data into a deep autoencoder network for unsupervised training to obtain a feature model; extracting feature data from the seismic data through the feature model; performing cluster analysis on the feature data to obtain feature points, and restoring the feature points to the original two-dimensional plane to obtain waveform clustering results.

[0005] In the above-mentioned seismic data feature extraction and clustering method based on deep autoencoder, extracting inter-layer data between target layers according to the seismic data and target level information includes: extracting top-level data and bottom-level data according to the seismic data and target level; obtaining coordinate data of top-level data and bottom-level data in the seismic data according to the top-level data and bottom-level data; obtaining inter-layer data between target layers in the seismic data according to the top-level data, the bottom-level data and the corresponding coordinate data.

[0006] In the above-mentioned seismic data feature extraction and clustering method based on deep autoencoder, preferably, calculating the inter-layer data between target layers based on the top-level data, the bottom-level data and the corresponding coordinate data includes: obtaining a whole block of data in the seismic data with the coordinate data corresponding to the top-level data and the coordinate data corresponding to the bottom-level data as boundaries; setting the data above the top-level and below the bottom-level in each data track of the whole block of data to zero to obtain the inter-layer data between target layers.

[0007] In the above-mentioned seismic data feature extraction and clustering method based on deep autoencoder, preferably, calculating the inter-layer data between target layers based on the top-level data, the bottom-level data and the corresponding coordinate data includes: obtaining the minimum inter-layer spacing between the top-level data and the bottom-level data based on the coordinate data corresponding to the top-level data and the coordinate data corresponding to the bottom-level data; performing adaptive pooling processing on each data between the top-level data and the bottom-level data using the minimum inter-layer spacing as the adaptive pooling output value to obtain the inter-layer data between the target layers.

[0008] In the above-mentioned seismic data feature extraction and clustering method based on deep autoencoder, preferably, calculating the inter-layer data between target layers based on the top-level data, the bottom-level data and the corresponding coordinate data includes: obtaining the maximum inter-layer spacing between the top-level data and the bottom-level data based on the coordinate data corresponding to the top-level data and the coordinate data corresponding to the bottom-level data; performing interpolation / resampling processing on each data between the top-level data and the bottom-level data to obtain multiple output data; comparing the output data with the maximum inter-layer spacing, and when the comparison result is consistent in size, obtaining the inter-layer data between the target layers based on the output data.

[0009] In the above-mentioned seismic data feature extraction and clustering method based on deep autoencoder, preferably, bringing the inter-layer data into the deep autoencoder network for unsupervised training to obtain the feature model includes: bringing the inter-layer data into a convolutional autoencoder or a fully connected autoencoder for unsupervised training to obtain the feature model.

[0010] The present invention also provides a seismic data feature extraction and clustering device based on a deep autoencoder, the device comprising a data extraction module, a model training module, a feature extraction module and a clustering module; the data extraction module is used to obtain seismic data, and extract inter-layer data between target layers according to the seismic data and target layer information; the model training module is used to bring the inter-layer data into a deep autoencoder network for unsupervised training to obtain a feature model; the feature extraction module is used to extract feature data from the seismic data through the feature model; the clustering module is used to perform cluster analysis on the feature data to obtain feature points, and restore the feature points to the original two-dimensional plane to obtain waveform clustering results.

[0011] In the above-mentioned seismic data feature extraction and clustering device based on deep autoencoder, preferably, the data extraction module includes an inter-layer data extraction unit, and the inter-layer data extraction unit is used to extract top-level data and bottom-level data according to the seismic data and the target level; obtain the coordinate data of the top-level data and the bottom-level data in the seismic data according to the top-level data and the bottom-level data; obtain the inter-layer data between the target layers according to the top-level data, the bottom-level data and the corresponding coordinate data in the seismic data.

[0012] In the above-mentioned seismic data feature extraction and clustering device based on deep autoencoder, preferably, the inter-layer data extraction unit is used to obtain a whole block of data in the seismic data with the coordinate data corresponding to the top layer data and the coordinate data corresponding to the bottom layer data as boundaries; the data above the top layer and below the bottom layer in each data of the whole block of data is set to zero to obtain the inter-layer data between the target layers.

[0013] In the above-mentioned seismic data feature extraction and clustering device based on deep autoencoder, preferably, the inter-layer data extraction unit is used to obtain the minimum inter-layer spacing between the top-level data and the bottom-level data according to the coordinate data corresponding to the top-level data and the coordinate data corresponding to the bottom-level data; and adaptive pooling processing is performed on each data between the top-level data and the bottom-level data using the minimum inter-layer spacing as the adaptive pooling output value to obtain inter-layer data between target layers.

[0014] In the above-mentioned seismic data feature extraction and clustering device based on deep autoencoder, preferably, the inter-layer data extraction unit is used to obtain the maximum inter-layer spacing between the top-level data and the bottom-level data according to the coordinate data corresponding to the top-level data and the coordinate data corresponding to the bottom-level data; interpolate / resampling each data between the top-level data and the bottom-level data to obtain multiple output data; compare the output data with the maximum inter-layer spacing, and when the comparison result is consistent in size, obtain the inter-layer data between the target layers according to the output data.

[0015] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above method when executing the computer program.

[0016] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program for executing the above method.

[0017] The beneficial technical effects of the present invention are: based on an unsupervised deep neural network, using an autoencoder network, seismic data is compressed and feature extracted, data dimension is reduced, data redundancy is removed, and stratigraphic features are extracted, especially for seismic data with unequal thickness between layers. The extracted feature data is clustered and analyzed to provide reliable support for subsequent seismic phase analysis, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings described herein are used to provide a further understanding of the present invention, constitute a part of the present application, and do not constitute a limitation of the present invention. In the drawings:

[0019] Figure 1 A schematic diagram of a flow chart of a method for extracting and clustering seismic data features based on a deep autoencoder provided in one embodiment of the present invention;

[0020] Figure 2 A schematic diagram of a process for obtaining inter-layer data provided by an embodiment of the present invention;

[0021] Figure 3 A schematic diagram of a process of obtaining inter-layer data using a zero-filling method provided by an embodiment of the present invention;

[0022] Figure 4 A schematic diagram of a process of obtaining inter-layer data using an adaptive pooling method provided by an embodiment of the present invention;

[0023] Figure 5 A schematic diagram of a process for obtaining inter-layer data using an interpolation / resampling method provided by an embodiment of the present invention;

[0024] Figure 6 A schematic diagram of an application flow of a seismic data feature extraction and clustering method based on a deep autoencoder provided in one embodiment of the present invention;

[0025] Figure 7 A schematic diagram of the structure of an encoder provided by an embodiment of the present invention;

[0026] Figure 8 A schematic diagram of a feature point compression result provided by an embodiment of the present invention;

[0027] Fig. 9 A schematic diagram of a feature point clustering result provided by an embodiment of the present invention;

[0028] Fig.10 A schematic diagram of a waveform clustering result provided by an embodiment of the present invention;

[0029] Fig.11 A schematic diagram of the structure of a seismic data feature extraction and clustering device based on a deep autoencoder provided in one embodiment of the present invention;

[0030] Fig.12 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention;

[0031] Fig.13 A schematic diagram of a network structure provided by an embodiment of the present invention;

[0032] Fig.14 A schematic diagram of waveform clustering results provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0033] The following will describe the implementation methods of the present invention in detail with reference to the accompanying drawings and embodiments, so that the implementation process of how the present invention applies technical means to solve technical problems and achieve technical effects can be fully understood and implemented accordingly. It should be noted that as long as there is no conflict, the various embodiments of the present invention and the various features in the embodiments can be combined with each other, and the technical solutions formed are all within the protection scope of the present invention.

[0034] In addition, the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be performed in an order different from that shown here.

[0035] Please refer to Figure 1 As shown, the present invention provides a seismic data feature extraction and clustering method based on a deep autoencoder, which specifically includes:

[0036] S101 acquires seismic data, and extracts inter-layer data between target layers according to the seismic data and target layer information;

[0037] S102 brings the inter-layer data into a deep autoencoder network for unsupervised training to obtain a feature model;

[0038] S103 extracts characteristic data from the seismic data using the characteristic model;

[0039] S104 performs cluster analysis on the feature data to obtain feature points, and restores the feature points to the original two-dimensional plane to obtain waveform clustering results.

[0040] For further information, please refer to Figure 2 As shown, in the above embodiment, extracting the target layer interlayer data according to the seismic data and the target layer information includes:

[0041] S201 extracts top layer data and bottom layer data according to the seismic data and the target layer;

[0042] S202 obtains coordinate data of the top layer data and the bottom layer data in the seismic data according to the top layer data and the bottom layer data and the seismic data;

[0043] S203 obtains inter-layer data between target layers from the seismic data according to the top layer data, the bottom layer data and the corresponding coordinate data.

[0044] In actual work, the seismic data feature extraction and clustering method based on deep autoencoder provided by the present invention can be mainly divided into the following three parts. The overall process is as follows: Figure 6 As shown:

[0045] 1. Inter-layer data extraction:

[0046] In order to perform waveform clustering, it is first necessary to extract the waveform of each trace between target layers from the seismic data. Different from the traditional method of extracting seismic trace waveforms by opening a time window, the present invention mainly provides the following three methods to extract inter-layer data in one embodiment:

[0047] Please refer to Figure 3 As shown, in one embodiment of the present invention, calculating and obtaining the inter-layer data between target layers according to the top layer data, the bottom layer data and the corresponding coordinate data includes:

[0048] S301, taking the coordinate data corresponding to the top layer data and the coordinate data corresponding to the bottom layer data as boundaries, obtaining a whole block of data from the seismic data;

[0049] S302 sets the data above the top layer and below the bottom layer in each data of the entire block of data to zero, and obtains inter-layer data between target layers.

[0050] Specifically, the above embodiment is a zero-filling method, that is, after two layers are given, the zero-filling method is used to extract the data between the given layers; first, the data of the top layer and the bottom layer of the target area and the coordinates in the corresponding matrix are read. The entire block of data is taken out with the minimum coordinate of the top layer and the maximum coordinate of the bottom layer as the boundary; for the data taken out, the data above the top layer and below the bottom layer are set to zero in each channel to obtain the required inter-layer data.

[0051] Please refer to Figure 4 As shown, in one embodiment of the present invention, calculating and obtaining the inter-layer data between target layers according to the top layer data, the bottom layer data and the corresponding coordinate data includes:

[0052] S401 obtains the minimum layer spacing between the top layer data and the bottom layer data according to the coordinate data corresponding to the top layer data and the coordinate data corresponding to the bottom layer data;

[0053] S502 performs adaptive pooling processing on each data between the top layer data and the bottom layer data using the minimum layer spacing as an adaptive pooling output value to obtain inter-layer data between target layers.

[0054] Specifically, the above embodiment is an adaptive pooling method, that is, after two layers are given, adaptive pooling is used to process the data between the given layers; first, the minimum distance between the top and bottom layers is obtained as (or smaller) as the size of the adaptive pooling output, and each channel between the top and bottom layers is adaptively pooled to get the same size, and the data between the top and bottom layers can be obtained. Adaptive pooling can pool tensors of different sizes into the same size, and its principle is:

[0055] kernel_size=(input_size+2*padding)-(output_size-1)*stride;

[0056] Among them, kernel_size is the pooling window size, stride is the step size of the window movement, padding is the number of layers of 0 added to each input variable, input_size is the size of the input tensor, and output_size is the size of the output tensor.

[0057] Please refer to Figure 5 As shown, in one embodiment of the present invention, calculating and obtaining the inter-layer data between target layers according to the top layer data, the bottom layer data and the corresponding coordinate data includes:

[0058] S501 obtains the maximum layer spacing between the top layer data and the bottom layer data according to the coordinate data corresponding to the top layer data and the coordinate data corresponding to the bottom layer data;

[0059] S502 performs interpolation / resampling processing on each data between the top layer data and the bottom layer data to obtain a plurality of output data;

[0060] S503 compares the output data with the maximum inter-layer distance, and when the comparison result shows that the maximum inter-layer distance is consistent, obtains inter-layer data between target layers according to the output data.

[0061] Specifically, the above embodiment is an interpolation / resampling method, that is, after two layers are given, the interpolation / resampling method is used to process the given inter-layer data; first, the maximum (or larger) distance between the top and the bottom is obtained as the output size, and each channel between the top and the bottom is processed by interpolation / resampling to obtain data with the same output size, which is the required inter-layer data; interpolation / resampling can use a variety of methods, which will not be described in detail here.

[0062] 2. Data compression and feature extraction:

[0063] Using a deep autoencoder network to build a feature model is to bring the inter-layer data into the deep autoencoder network for unsupervised training to obtain the feature model, which includes: bringing the inter-layer data into a convolutional autoencoder or a fully connected autoencoder for unsupervised training to obtain the feature model; in actual work, the autoencoder includes an encoder and a decoder. The above-mentioned convolutional autoencoder and fully connected autoencoder both have good results; among them, the fully connected autoencoder, (such as Figure 7 As shown in the figure, it consists of a fully connected layer and an activation layer. The convolutional autoencoder consists of a convolutional layer, a pooling layer, an activation layer, a fully connected layer, etc.

[0064] Next, the inter-layer data obtained by the above calculation is put into the autoencoder for unsupervised training, and the output of the encoder is set to a specified size to obtain compressed data (such as Figure 8 As shown in the figure, the compressed data contains the features of the original data and removes data redundancy; wherein, after the network training is stable, the output of the encoding part is the compressed seismic data, which contains the feature data extracted from the original data.

[0065] 3. Cluster analysis:

[0066] The compressed feature data is clustered using clustering methods (K-means, Mean Shift, etc.) Fig. 9 ), and project the result back to the original two-dimensional plane to obtain the waveform clustering result (as shown in Fig.10 shown).

[0067] In order to further improve the clustering effect, we introduced an adaptive Gaussian mixture model layer to perform clustering analysis on the data. The adaptive Gaussian mixture model can automatically determine the number of clustering categories and can be combined with the autoencoder structure in 2 to form a new network (such as Fig.13 ), dynamically adjust the trainable parameters of the autoencoder and clustering layer to improve the accuracy of waveform clustering (as shown in Fig.14 shown).

[0068] The adaptive Gaussian mixture model can be defined by the following formula:

[0069]

[0070] where q ij (x|θ i ) is the kernel for measuring similarity, m is the mean vector, ∑ j is the covariance matrix, L cis the loss function of the clustering layer, and KL stands for KL divergence.

[0071] To more clearly illustrate the combined application of the above embodiments, please refer to Figure 6 As shown, the specific implementation process of the seismic data feature extraction and clustering method based on deep autoencoder provided by the present invention can be as follows:

[0072] 1. Obtain the data of the top layer h1 and the bottom layer h2 and their coordinates in the seismic data;

[0073] 2. Get inter-layer data. Take out the entire block of data with the minimum coordinate of the top layer and the maximum coordinate of the bottom layer as the boundary, and set the data above the top layer and below the bottom layer to zero in each channel to get the required inter-layer data; or take the minimum distance between the top and bottom layers as the size of the adaptive pooling output (or smaller), and get the same size for each channel between the top and bottom layers through adaptive pooling; or take the maximum (or larger) distance between the top and bottom as the output size, and get data with the same output size through interpolation / resampling for each channel between the top and bottom;

[0074] 3. Put the inter-layer data into the autoencoder network for training;

[0075] 4. Use the trained autoencoder network model to output the encoder results, input them into the clustering layer, and dynamically adjust the encoding results and clustering results through cyclic iterative training;

[0076] 5. Conduct cluster analysis;

[0077] 6. Restore the clustered points to the original two-dimensional plane to obtain the waveform clustering result.

[0078] Please refer to Fig.11 As shown, the present invention also provides a seismic data feature extraction and clustering device based on a deep autoencoder, the device comprising a data extraction module, a model training module, a feature extraction module and a clustering module; the data extraction module is used to obtain seismic data, and extract inter-layer data between target layers according to the seismic data and target layer information; the model training module is used to bring the inter-layer data into a deep autoencoder network for unsupervised training to obtain a feature model; the feature extraction module is used to extract feature data from the seismic data through the feature model; the clustering module is used to perform cluster analysis on the feature data to obtain feature points, and restore the feature points to the original two-dimensional plane to obtain waveform clustering results.

[0079] In the above embodiment, the data extraction module includes an interlayer data extraction unit, and the interlayer data extraction unit is used to extract top-level data and bottom-level data according to the seismic data and the target level; obtain the coordinate data of the top-level data and the bottom-level data in the seismic data according to the top-level data and the bottom-level data; obtain the interlayer data between the target layers according to the top-level data, the bottom-level data and the corresponding coordinate data. Wherein, the interlayer data extraction unit is used to obtain the whole block of data in the seismic data with the coordinate data corresponding to the top-level data and the coordinate data corresponding to the bottom-level data as the boundary; set the data above the top-level data and below the bottom-level data in each data of the whole block of data to zero, and obtain the interlayer data between the target layers. Or, obtain the minimum interlayer spacing between the top-level data and the bottom-level data according to the coordinate data corresponding to the top-level data and the coordinate data corresponding to the bottom-level data; use the minimum interlayer spacing as the adaptive pooling output value to perform adaptive pooling processing on each data between the top-level data and the bottom-level data, and obtain the interlayer data between the target layers. Or, obtain the maximum layer spacing between the top-level data and the bottom-level data based on the coordinate data corresponding to the top-level data and the coordinate data corresponding to the bottom-level data; perform interpolation / resampling processing on each data between the top-level data and the bottom-level data to obtain multiple output data; compare the output data with the maximum layer spacing, and when the comparison result is consistent in size, obtain inter-layer data between target layers based on the output data.

[0080] The beneficial technical effects of the present invention are: based on an unsupervised deep neural network, using an autoencoder network, seismic data is compressed and feature extracted, data dimension is reduced, data redundancy is removed, and stratigraphic features are extracted, especially for seismic data with unequal thickness between layers. The extracted feature data is clustered and analyzed to provide reliable support for subsequent seismic phase analysis, etc.

[0081] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the above method when executing the computer program.

[0082] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program for executing the above method.

[0083] like Fig.12 As shown, the electronic device 600 may further include: a communication module 110, an input unit 120, an audio processor 130, a display 160, and a power supply 170. It is worth noting that the electronic device 600 does not necessarily have to include Fig.12 In addition, the electronic device 600 may also include Fig.12 For components not shown, reference may be made to the prior art.

[0084] like Fig.12 As shown, the central processor 100 is sometimes also referred to as a controller or an operation control, and may include a microprocessor or other processor devices and / or logic devices. The central processor 100 receives inputs and controls the operations of various components of the electronic device 600.

[0085] The memory 140 may be, for example, one or more of a cache, a flash memory, a hard drive, a removable medium, a volatile memory, a non-volatile memory or other suitable devices. The above-mentioned information related to the failure may be stored, and a program for executing the relevant information may also be stored. The CPU 100 may execute the program stored in the memory 140 to implement information storage or processing.

[0086] The input unit 120 provides input to the CPU 100. The input unit 120 is, for example, a key or a touch input device. The power supply 170 is used to provide power to the electronic device 600. The display 160 is used to display display objects such as images and text. The display may be, for example, an LCD display, but is not limited thereto.

[0087] The memory 140 may be a solid-state memory, such as a read-only memory (ROM), a random access memory (RAM), a SIM card, etc. It may also be a memory that saves information even when the power is off, can be selectively erased, and is provided with more data, examples of which are sometimes referred to as EPROMs, etc. The memory 140 may also be some other type of device. The memory 140 includes a buffer memory 141 (sometimes referred to as a buffer). The memory 140 may include an application / function storage unit 142, which is used to store application programs and function programs or processes for executing the operation of the electronic device 600 through the central processor 100.

[0088] The memory 140 may also include a data storage unit 143 for storing data, such as contacts, digital data, pictures, sounds, and / or any other data used by the electronic device. The driver storage unit 144 of the memory 140 may include various drivers for communication functions of the electronic device and / or for executing other functions of the electronic device (such as messaging applications, address book applications, etc.).

[0089] The communication module 110 is a transmitter / receiver 110 that transmits and receives signals via an antenna 111. The communication module 110 (transmitter / receiver) is coupled to the central processor 100 to provide input signals and receive output signals, which may be the same as the case of a conventional mobile communication terminal.

[0090] Based on different communication technologies, multiple communication modules 110 may be provided in the same electronic device, such as a cellular network module, a Bluetooth module and / or a wireless LAN module. The communication module 110 (transmitter / receiver) is also coupled to a speaker 131 and a microphone 132 via an audio processor 130 to provide an audio output via the speaker 131 and receive an audio input from the microphone 132, thereby realizing a common telecommunication function. The audio processor 130 may include any suitable buffer, decoder, amplifier, etc. In addition, the audio processor 130 is also coupled to the central processor 100, so that recording can be performed on the local machine through the microphone 132, and the sound stored on the local machine can be played through the speaker 131.

[0091] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0092] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0093] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0094] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0095] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A seismic data feature extraction and clustering method based on deep autoencoder, characterized in that: The method comprises: Acquire seismic data, extract top layer data and bottom layer data according to the seismic data and target layer information, and extract inter-layer data between target layers according to the top layer data and the bottom layer data; Bringing the inter-layer data into a deep autoencoder network for unsupervised training to obtain a feature model; Extracting characteristic data from the seismic data using the characteristic model to obtain characteristic data; Performing cluster analysis on the characteristic data to obtain characteristic points, and restoring the characteristic points to the original two-dimensional plane to obtain waveform clustering results; Calculating the inter-layer data between target layers based on the top-level data, the bottom-level data and the corresponding coordinate data includes: obtaining the maximum inter-layer spacing between the top-level data and the bottom-level data based on the coordinate data corresponding to the top-level data and the coordinate data corresponding to the bottom-level data; performing interpolation / resampling processing on each data between the top-level data and the bottom-level data to obtain multiple output data; comparing the output data with the maximum inter-layer spacing, and when the comparison result is consistent in size, obtaining the inter-layer data between target layers based on the output data.

2. The method for extracting and clustering seismic data features based on deep autoencoders according to claim 1, characterized in that: Extracting inter-layer data between target layers based on top-level data and bottom-level data includes: Obtaining coordinate data of the top layer data and the bottom layer data in the seismic data according to the top layer data and the bottom layer data and the seismic data; The inter-layer data of the target layers are obtained from the seismic data according to the top layer data, the bottom layer data and the corresponding coordinate data.

3. The method for extracting and clustering seismic data features based on deep autoencoders according to claim 2, characterized in that: Calculating the inter-layer data of the target layer according to the top layer data, the bottom layer data and the corresponding coordinate data includes: Taking the coordinate data corresponding to the top layer data and the coordinate data corresponding to the bottom layer data as boundaries, obtaining a whole block of data from the seismic data; The data above the top layer and below the bottom layer in each data of the entire block of data are set to zero to obtain inter-layer data between target layers.

4. The method for extracting and clustering seismic data features based on deep autoencoders according to claim 2, characterized in that: Calculating the inter-layer data of the target layer according to the top layer data, the bottom layer data and the corresponding coordinate data includes: Obtaining a minimum layer spacing between the top layer data and the bottom layer data according to the coordinate data corresponding to the top layer data and the coordinate data corresponding to the bottom layer data; Adaptive pooling is performed on each data between the top layer data and the bottom layer data using the minimum layer spacing as an adaptive pooling output value to obtain inter-layer data between target layers.

5. The method for extracting and clustering seismic data features based on deep autoencoders according to claim 1, characterized in that: Bringing the inter-layer data into a deep autoencoder network for unsupervised training to obtain a feature model includes: bringing the inter-layer data into a convolutional autoencoder or a fully connected autoencoder for unsupervised training to obtain a feature model.

6. A seismic data feature extraction and clustering device based on deep autoencoder, characterized in that: The device comprises a data extraction module, a model training module, a feature extraction module and a clustering module; The data extraction module is used to obtain seismic data, extract top layer data and bottom layer data according to the seismic data and target layer information, and extract inter-layer data between target layers according to the top layer data and bottom layer data; The model training module is used to bring the inter-layer data into the deep autoencoder network for unsupervised training to obtain a feature model; The feature extraction module is used to extract feature data from the seismic data using the feature model; The clustering module is used to perform cluster analysis on the feature data to obtain feature points, and restore the feature points to the original two-dimensional plane to obtain waveform clustering results; The data extraction module includes an inter-layer data extraction unit, which is used to obtain the maximum inter-layer spacing between the top-level data and the bottom-level data based on the coordinate data corresponding to the top-level data and the coordinate data corresponding to the bottom-level data; perform interpolation / resampling processing on each data between the top-level data and the bottom-level data to obtain multiple output data; compare the output data with the maximum inter-layer spacing, and when the comparison result is consistent in size, obtain the inter-layer data between the target layers based on the output data.

7. The seismic data feature extraction and clustering device based on deep autoencoder according to claim 6, characterized in that: The inter-layer data extraction unit is used to obtain the coordinate data of the top layer data and the bottom layer data in the seismic data based on the top layer data and the bottom layer data; and to obtain the inter-layer data between target layers in the seismic data based on the top layer data, the bottom layer data and the corresponding coordinate data.

8. The seismic data feature extraction and clustering device based on deep autoencoder according to claim 7, characterized in that: The interlayer data extraction unit is used to obtain a whole block of data in the seismic data with the coordinate data corresponding to the top layer data and the coordinate data corresponding to the bottom layer data as boundaries; the data above the top layer and below the bottom layer in each data track of the whole block of data is set to zero to obtain the interlayer data between the target layers.

9. The seismic data feature extraction and clustering device based on deep autoencoder according to claim 7, characterized in that: The inter-layer data extraction unit is used to obtain the minimum layer spacing between the top-level data and the bottom-level data according to the coordinate data corresponding to the top-level data and the coordinate data corresponding to the bottom-level data; and to perform adaptive pooling processing on each data between the top-level data and the bottom-level data using the minimum layer spacing as the adaptive pooling output value to obtain the inter-layer data between the target layers.

10. An electronic 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.

11. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program for executing the method according to any one of claims 1 to 5.

Images