A method, apparatus, equipment, and medium for optimizing seismic data sample sets.

Abstract

This application discloses a method, apparatus, device, and medium for optimizing a seismic data sample set. The method includes: acquiring a seismic data sample set; copying at least one seismic data sample from the seismic data sample set; processing the at least one seismic data sample using a target processing method to obtain a new seismic data sample; determining an augmented seismic data sample set based on the seismic data sample set and the new seismic data sample, and training a machine learning model based on the augmented seismic data sample set; the target processing method includes at least one of the following: flipping processing, noise addition processing, high-pass filtering processing, and low-pass filtering processing. This technical solution augments the number of samples in the seismic data sample set, thereby improving the generalization ability and other performance characteristics of the machine learning model trained based on the augmented seismic data sample set.

Description

Technical Field

[0001] This invention relates to the field of machine learning technology, and in particular to a method, apparatus, device, and medium for optimizing a seismic data sample set. Background Technology

[0002] In recent years, artificial intelligence technologies, represented by deep learning, have been increasingly applied to the processing and interpretation of seismic exploration data. The quality of seismic data sample sets directly affects the accuracy of the model's interpretation of seismic data. How to optimize seismic data samples is a technical problem that urgently needs to be solved by those skilled in the art.

[0003] Because different earthquake data collection sites have different surface conditions or underground structures, the characteristics of the collected earthquake data samples vary greatly. How to augment smaller earthquake data sample sets with insufficient features to increase their scale and feature richness, and improve the generalization ability of the trained network model, is an urgent technical problem to be solved. Summary of the Invention

[0004] This invention provides a method, apparatus, device, and medium for optimizing seismic data sample sets, which can augment seismic data samples to improve the quantity and quality of the seismic data sample sets, so that the augmented seismic data sample sets can be used to train machine learning models with better generalization ability.

[0005] According to one aspect of the present invention, a method for optimizing a seismic data sample set is provided, the method comprising:

[0006] Obtain a set of earthquake data samples, and copy at least one earthquake data sample from the set of earthquake data samples;

[0007] The at least one seismic data sample is processed using a target processing method to obtain a new seismic data sample;

[0008] Based on the earthquake data sample set and the newly added earthquake data samples, an augmented earthquake data sample set is determined, and a machine learning model is trained based on the augmented earthquake data sample set.

[0009] The target processing method includes at least one of the following: inversion processing, noise addition processing, high-pass filtering processing, and low-pass filtering processing;

[0010] If the seismic data sample set is a first seismic data sample set used to solve the regional segmentation problem, then copying at least one seismic data sample from the seismic data sample set includes: copying at least one first seismic data sample from the first seismic data sample set.

[0011] If the seismic data sample set is a second seismic data sample set used to solve the classification problem, then copying at least one seismic data sample from the seismic data sample set includes: in the second seismic data sample set, if the number of second seismic data samples corresponding to the target category is less than a quantity threshold, then copying at least one second seismic data sample from the second seismic data samples corresponding to the target category.

[0012] According to another aspect of the present invention, an optimization apparatus for a seismic data sample set is provided, comprising:

[0013] The seismic data sample copying module is used to acquire a seismic data sample set and copy at least one seismic data sample from the seismic data sample set;

[0014] The seismic data sample processing module is used to process the at least one seismic data sample in a target processing manner to obtain a new seismic data sample;

[0015] The augmented seismic data sample set determination module is used to determine the augmented seismic data sample set based on the seismic data sample set and the newly added seismic data samples, so as to train a machine learning model based on the augmented seismic data sample set.

[0016] The target processing method includes at least one of the following: inversion processing, noise addition processing, high-pass filtering processing, and low-pass filtering processing;

[0017] The seismic data sample copying module includes: a first seismic data sample copying unit, used to copy at least one first seismic data sample from the first seismic data sample set if the seismic data sample set is a first seismic data sample set used to solve the regional segmentation problem;

[0018] The seismic data sample copying module includes a second seismic data sample copying unit, configured to copy at least one second seismic data sample from the second seismic data sample set corresponding to the target category if the seismic data sample set is a second seismic data sample set used to solve a classification problem.

[0019] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0020] At least one processor; and

[0021] A memory communicatively connected to the at least one processor; wherein,

[0022] The memory stores a computer program that can be executed by the at least one processor, which enables the at least one processor to perform the method for optimizing the seismic data sample set according to any embodiment of the present invention.

[0023] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the method for optimizing the seismic data sample set according to any embodiment of the present invention.

[0024] The technical solution of this application embodiment includes: acquiring a seismic data sample set; copying at least one seismic data sample from the seismic data sample set; processing the at least one seismic data sample in a target processing manner to obtain a new seismic data sample; determining an augmented seismic data sample set based on the seismic data sample set and the new seismic data sample, and training a machine learning model based on the augmented seismic data sample set; the target processing manner includes at least one of the following: flipping processing, noise addition processing, high-pass filtering processing, and low-pass filtering processing; if the seismic data sample set is a first seismic data sample set used to solve a region segmentation problem, then copying at least one seismic data sample from the seismic data sample set includes: copying at least one first seismic data sample from the first seismic data sample set; if the seismic data sample set is a second seismic data sample set used to solve a classification problem, then copying at least one seismic data sample from the seismic data sample set includes: in the second seismic data sample set, if the number of second seismic data samples corresponding to the target category is less than a number threshold, then copying at least one second seismic data sample from the second seismic data samples corresponding to the target category. This technical solution processes earthquake data samples in a target-oriented manner to obtain new earthquake data samples, thereby increasing the quantity and quality of earthquake data samples. This improves the generalization ability and other performance aspects of the machine learning model after training it based on the augmented earthquake data sample set.

[0025] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments of this application, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 This is a flowchart of an optimization method for a seismic data sample set according to Embodiment 1 of this application;

[0028] Figure 2 This is a flowchart of a method for optimizing a seismic data sample set according to Embodiment 2 of this application;

[0029] Figure 3 These are screenshots of augmented software based on the first seismic data sample set provided in Embodiment 2 of this application;

[0030] Figure 4 This is a schematic diagram of the software interface for imbalance optimization provided in Embodiment 2 of this application;

[0031] Figure 5 This is a network structure diagram provided according to Embodiment 2 of this application;

[0032] Figure 6 This is a schematic diagram of a sample with a high degree of imbalance and its label provided in Embodiment 2 of this application;

[0033] Figure 7 This is a flowchart of a method for optimizing a seismic data sample set according to Embodiment 3 of this application;

[0034] Figure 8 These are screenshots of software used to augment second seismic data samples, provided in Embodiment 3 of this application.

[0035] Figure 9 This is a schematic diagram of the flipping process of the layer interpretation sample provided in Embodiment 3 of this application;

[0036] Figure 10 This is a schematic diagram of high-pass filtering processing of the layer interpretation sample provided in Embodiment 3 of this application;

[0037] Figure 11 This is a schematic diagram of high-pass filtering processing of seismic facies segmentation samples according to Embodiment 3 of this application;

[0038] Figure 12 This is a schematic diagram of the structure of an optimization device for a seismic data sample set according to Embodiment 4 of this application;

[0039] Figure 13 This is a schematic diagram of the structure of an electronic device that implements an optimization method for a seismic data sample set according to an embodiment of this application. Detailed Implementation

[0040] To enable those skilled in the art to better understand the present invention, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0041] It should be noted that the terms "first," "second," "target," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0042] Example 1

[0043] Figure 1 This application provides a flowchart of a method for optimizing a seismic data sample set, as shown in Embodiment 1. This embodiment is applicable to situations involving the augmentation of seismic data samples. The method can be executed by a seismic data sample set optimization device, which can be implemented in hardware and / or software and can be configured in an electronic device with data processing capabilities. Figure 1 As shown, the method includes:

[0044] S110, Obtain a set of earthquake data samples, and copy at least one earthquake data sample from the set of earthquake data samples.

[0045] The seismic data sample set includes seismic data samples used to train machine learning models. These samples can be pre-labeled, and their labels will vary depending on the context. For example, a sample might reflect data collected in formation A, with the label "formation A." A sample might reflect data collected in regions B, C, and D, with the label including the region corresponding to each data point. For instance, a seismic data sample can consist of seismic amplitude data arranged in a matrix, corresponding to seismic traces and observation points. The sample can be two-dimensional or three-dimensional. A two-dimensional sample reflects seismic data collected on a geological profile, while a three-dimensional sample reflects seismic data collected in a specific region.

[0046] Specifically, the earthquake data sample set includes multiple earthquake data samples, and at least one earthquake data sample can be selected from the earthquake data sample set either randomly or according to selection criteria; the selection of at least one earthquake data sample according to selection criteria can be: selecting the earthquake data sample corresponding to the target label according to the sample label.

[0047] S120, the at least one seismic data sample is processed in a target processing manner to obtain a new seismic data sample.

[0048] The target processing method refers to the way the seismic data sample is transformed. After processing the seismic data sample according to the target processing method, the processed seismic data sample will become a new seismic data sample. The target processing method includes at least one of the following: flipping processing, noise addition processing, high-pass filtering processing, and low-pass filtering processing.

[0049] Furthermore, the flipping process refers to flipping the seismic data samples. Specifically, if the seismic data is two-dimensional, the distribution of the two-dimensional data is rotated, for example, by rotating it 180 degrees around the center point, to obtain new seismic data samples.

[0050] Noise addition processing refers to adding noise to each earthquake data in at least one earthquake data sample to obtain a new earthquake data sample.

[0051] High-pass filtering refers to filtering each earthquake data in at least one earthquake data sample, retaining the high-frequency components and removing the low-frequency components to obtain new earthquake data samples.

[0052] Low-pass filtering refers to filtering each earthquake data in at least one earthquake data sample, retaining the low-frequency components and discarding the high-frequency components to obtain new earthquake data samples.

[0053] Specifically, after obtaining at least one seismic data sample, each of the at least one seismic data sample is processed using different target processing methods to ensure that the newly added seismic data sample is different from the original seismic data sample and other newly added seismic data samples.

[0054] S130, based on the earthquake data sample set and the newly added earthquake data samples, determine the augmented earthquake data sample set, and train the machine learning model based on the augmented earthquake data sample set.

[0055] After obtaining the newly added seismic data samples, the existing seismic data sample set and the newly added seismic data samples are merged to obtain an augmented seismic data sample set. For example, if the existing seismic data sample set contains 10 samples and the newly added seismic data samples contain 5 samples, then the augmented seismic data sample set contains 15 samples. Furthermore, a machine learning model is trained based on the augmented seismic data sample set to improve the generalization ability and other performance aspects of the machine learning model.

[0056] In this embodiment of the application, if the seismic data sample set is a first seismic data sample set used to solve the regional segmentation problem, then copying at least one seismic data sample from the seismic data sample set includes: copying at least one first seismic data sample from the first seismic data sample set; if the seismic data sample set is a second seismic data sample set used to solve the classification problem, then copying at least one seismic data sample from the seismic data sample set includes: in the second seismic data sample set, if the number of second seismic data samples corresponding to the target category is less than a quantity threshold, then copying at least one second seismic data sample from the second seismic data samples corresponding to the target category.

[0057] The technical solution of this application embodiment includes: acquiring a seismic data sample set; copying at least one seismic data sample from the seismic data sample set; processing the at least one seismic data sample in a target processing manner to obtain a new seismic data sample; determining an augmented seismic data sample set based on the seismic data sample set and the new seismic data sample, and training a machine learning model based on the augmented seismic data sample set; the target processing manner includes at least one of the following: flipping processing, noise addition processing, high-pass filtering processing, and low-pass filtering processing; if the seismic data sample set is a first seismic data sample set used to solve a region segmentation problem, then copying at least one seismic data sample from the seismic data sample set includes: copying at least one first seismic data sample from the first seismic data sample set; if the seismic data sample set is a second seismic data sample set used to solve a classification problem, then copying at least one seismic data sample from the seismic data sample set includes: in the second seismic data sample set, if the number of second seismic data samples corresponding to the target category is less than a number threshold, then copying at least one second seismic data sample from the second seismic data samples corresponding to the target category. This technical solution processes earthquake data samples in a target-oriented manner to obtain new earthquake data samples, thereby increasing the quantity and quality of earthquake data samples. This improves the generalization ability and other performance aspects of the machine learning model after training it based on the augmented earthquake data sample set.

[0058] Example 2

[0059] Figure 2 This is a flowchart illustrating a method for optimizing a seismic data sample set according to Embodiment 2 of this application. This embodiment is an optimization based on the above embodiment. The seismic data sample set described in this embodiment is a first seismic data sample set used to solve the regional segmentation problem.

[0060] like Figure 2 As shown, the method in this embodiment of the application specifically includes the following steps:

[0061] S210, Obtain the first earthquake data sample set, and copy at least one first earthquake data sample from the first earthquake data sample set.

[0062] In this embodiment of the application, optionally, the seismic data sample set is a first seismic data sample set used to solve the regional segmentation problem. The first seismic data sample set consists of seismic data arranged in matrix form, and the seismic data arranged in matrix form corresponds to each region on the geological profile. The labels corresponding to the first seismic data sample set include the segmentation region corresponding to each seismic data set.

[0063] The first seismic data sample is used to solve the problem of region segmentation. That is, the machine learning model trained based on the first seismic data sample is able to identify which segmented regions are included in the seismic data input into the model. For example, taking the first seismic data sample in two-dimensional form as an example, the first seismic data sample consists of multiple seismic data arranged in a matrix (the seismic data is the sample points). The seismic data in the horizontal and vertical directions of this matrix can be seismic amplitude data collected from each observation point of each seismic trace. The seismic data arranged in a matrix in the first seismic data sample correspond to the regions on the geological profile. The label of the first seismic data sample includes the segmented region corresponding to each seismic data. For example, a first seismic data sample includes 6 sample points at positions (1,1), (1,2), (1,3), (2,1), (2,2), and (2,3) (obviously there would be many more sample points in reality, but a small number of sample points are used here for example). The label of the first seismic data sample reflects that the sample points at positions (1,1) and (1,2) correspond to region A, the sample points at positions (1,3) and (2,1) correspond to region B, and the sample points at positions (2,2) and (2,3) correspond to region C.

[0064] Specifically, copying at least one first earthquake data sample from the first earthquake data sample set includes: randomly copying at least one first earthquake data sample from the first earthquake data sample set.

[0065] S220, the at least one first seismic data sample is processed in a target processing manner to obtain a new seismic data sample.

[0066] The target processing method includes at least one of the following: inversion processing, noise addition processing, high-pass filtering processing, and low-pass filtering processing.

[0067] S230, based on the first earthquake data sample set and the newly added earthquake data samples, determine the augmented earthquake data sample set, and train the machine learning model based on the augmented first earthquake data sample set.

[0068] Specifically, training a machine learning model based on the augmented first seismic data sample set includes: training a machine learning model for identifying segmented regions based on the augmented first seismic data sample set, so that the machine learning model for identifying segmented regions can more accurately identify segmented regions.

[0069] The technical solution of this application includes: acquiring a first seismic data sample set; copying at least one first seismic data sample from the first seismic data sample set; processing the at least one first seismic data sample in a target processing manner to obtain a new seismic data sample; determining an augmented seismic data sample set based on the first seismic data sample set and the new seismic data sample, and training a machine learning model based on the augmented first seismic data sample set. This technical solution augments the first seismic data sample set to obtain an augmented first seismic data sample set, thus achieving the effect of increasing the number of samples.

[0070] In a specific implementation process, based on the corresponding sample augmentation software ( Figure 3 (Screenshot of the augmentation software for the first earthquake data sample set). To implement the augmentation function for the first earthquake data sample set, please refer to the following steps:

[0071] 1) Select the sample set directory and label directory: All sample files and corresponding label files of the first earthquake data sample set are stored in two folders named "train" and "label" in the same directory. You need to select the specific folder where the sample files and label files are located.

[0072] 2) Select sample dimension: Choose whether the samples in the sample set to be augmented are two-dimensional or three-dimensional samples.

[0073] 3) Set the sample size: For two-dimensional samples, you need to set the number of channels and the number of sample points. For three-dimensional samples, you need to set the number of lines, channels, and sample points.

[0074] 4) Select label data type: Select whether the sample labels in the sample set to be augmented are integers or floating-point numbers.

[0075] 5) Select the sample augmentation processing method and set the number of executions and parameters: This step first randomly selects one or more of the four sample augmentation processing methods: rotation (corresponding to flipping processing), noise addition (corresponding to noise addition processing), high frequency extraction (corresponding to high-pass filtering processing), and low frequency extraction (corresponding to low-pass filtering processing). Then, set the number of executions for each processing method, i.e., the number of samples augmented by each processing method. Finally, set the sample augmentation processing parameters. If high frequency extraction or low frequency extraction is selected, the high-pass threshold frequency range and low-pass threshold frequency range need to be set.

[0076] The conventional random sample augmentation method for augmenting small-scale seismic data sets involves the following steps: First, all samples in the sample set are randomly sorted. Then, one or more augmentation methods are randomly selected from four options: rotation, noise addition, high-frequency extraction, and low-frequency extraction. The number of times each method is executed (i.e., the number of samples augmented by each method) is set, and the parameters for each method are configured according to specified rules. Finally, the samples and their labels in the randomly sorted sample set are augmented sequentially according to the set augmentation methods and their execution counts to obtain the augmented sample set. If the total number of augmentation operations is greater than or equal to the original sample set size, the augmented sample set size will double. If the total number of augmentation operations is less than the original sample set size, the augmented sample set size will be the sum of the original sample set size and the total number of augmentation operations.

[0077] The results of sample collection and augmentation for a certain work area are as follows: The stratigraphic interpretation sample set for this work area contains 225 samples, each with a size of 80*160 (80 traces, 160 sampling points). Due to the small number of samples in the sample set, the deep learning network model is prone to overfitting during training, resulting in limited predictive and generalization capabilities. Augmentation was performed on the first seismic data sample set used to solve the regional segmentation problem. The target processing methods for sample augmentation were flipping, noise addition, and high-pass filtering. The execution times for the three processing methods were set to 2, 111, and 112, respectively, resulting in an augmented first seismic data sample set containing 450 samples.

[0078] This invention designs a U-shaped fully convolutional neural network structure. The network model was trained using a first seismic data sample set and an augmented first seismic data sample set, and the accuracy improvement was compared. The network designed in this invention adopts a symmetrical structure, consisting of an encoder network with 5 convolutional blocks on the left and a decoder network with 5 convolutional blocks on the right, named FCN-U5. Each convolutional block in both the encoder and decoder networks contains 2-3 sets of "Conv+BN+ReLU" operations. The direct connection between corresponding convolutional blocks of the two networks can fuse multi-scale first-arrival features and propagate max-pooling indices, achieving better first-arrival upper and lower region segmentation performance. Furthermore, the upsampling operation of the decoder network uses simple unpooling instead of the computationally complex deconvolution, accelerating the convergence speed of the network model training. The network structure diagram of this invention is shown below. Figure 5 As shown. In this invention, the hardware environment for training and testing the FCN-U5 network model is a Tesla P100 GPU with 16GB of memory, and the deep learning framework is Keras 2.3.1 installed on the Red Hat Linux 9.0 operating system.

[0079] Twenty samples were randomly selected from the first earthquake data sample set as the validation set for evaluating network performance, and the remaining samples were used as the training set for training the network model. The FCN-U5 network model was trained using mini-batch gradient descent with a batch size of 50. The loss function used was multi-class cross-entropy, which was minimized during training. The gradient descent optimization algorithm was Adam, with parameters β1 and β2 set to default values ​​of 0.9 and 0.999, respectively. A dynamic learning rate was used during model training, halving every 20 iterations to prevent the gradient from getting trapped in a local minimum and halting training. An early stopping strategy was employed: training was stopped when the loss value on the validation set increased within five consecutive iterations.

[0080] The FCN-U5 network model for layer interpretation was trained using the first seismic data sample set and completed the training process within 32 iterations, with a final validation accuracy of 92.21%. The FCN-U5 network model for layer interpretation was trained using the augmented first seismic data sample set and completed the training process within 56 iterations, with a final validation accuracy of 95.77%. After the sample set optimization, the validation accuracy of the FCN-U5 network model improved by 3.56%.

[0081] In this embodiment of the application, optionally, the method further includes: calculating the imbalance degree of each initial seismic data sample for the initial seismic data sample set used to solve the regional segmentation problem; the imbalance degree reflects the proportion of each segmented region in the sample; deleting samples in the initial seismic data sample set whose imbalance degree is greater than the imbalance degree threshold, and using the remaining samples as the first seismic data sample set.

[0082] In this embodiment of the application, for the initial seismic data sample set used to solve the regional segmentation problem, since the detailed processing and interpretation of seismic data requires a lot of time and effort from the staff, the initial seismic data sample set with a large scale is often of low quality and contains samples with high imbalance, which will affect the regional segmentation ability of the deep learning network. This method targets two-dimensional or three-dimensional initial seismic data sample sets for region segmentation problems. Based on the labels of the initial seismic data samples, it counts the number of sample points corresponding to each segmentation region in the labels, and then calculates the proportion of each segmentation region in the total number of regions, that is, the proportion of the number of sample points corresponding to the segmentation region in the total number of sample points (for example, in an initial seismic data sample with 100×80 sample points arranged in matrix form, the label of the sample indicates that there are 2000 sample points in segmentation region A, 4000 sample points in segmentation region B, and 2000 sample points in segmentation region C, then segmentation region A accounts for 25%, segmentation region B accounts for 50%, and segmentation region C accounts for 25%). Based on the proportion of the number of sample points corresponding to the segmentation region in the total number of sample points, the variance of the sample is calculated, that is, the imbalance of the sample. Finally, samples with an imbalance greater than the imbalance threshold are removed. This can significantly reduce the overall imbalance of the sample set and improve the region segmentation ability of the deep learning network model.

[0083] The imbalance threshold can be determined based on actual conditions, and this application embodiment does not impose any limitation on it. The imbalance of the initial seismic data sample is determined using the following formula:

[0084]

[0085] Where, σ 2 X represents the imbalance of the initial seismic data samples. i Let μ be the proportion of the sample points in the i-th segmented region to the total sample points, μ be the average proportion of the sample points in each segmented region to the total sample points, and n be the number of segmented regions.

[0086] In a concrete implementation, the imbalance checking and optimization function is implemented for the initial seismic data sample set used to solve the region segmentation problem through the following steps:

[0087] 1) Select the candidate earthquake data sample set label directory: All sample files and corresponding label files of the candidate earthquake data sample set are stored in two folders in the same directory, named "train" and "label" respectively. You need to select the specific folder where the label files are located.

[0088] 2) Select sample dimension: Select the sample dimension of the sample set that needs to be optimized.

[0089] 3) Select the type of label data: The sample set imbalance check and optimization function supports integer and floating-point labels, and you need to select the specific data type.

[0090] 4) Set sample size: For 2D samples, the number of channels and sample points need to be set; for 3D samples, the number of lines, channels, and sample points need to be set. Figure 4 (Illustrative diagram of the software interface optimized for imbalance)

[0091] 5) Perform sample set imbalance check: Based on the sample dimension, sample size, and label values ​​in the labels (different segmentation regions correspond to different label values), first, iterate through all label files in the sample set label directory for each label value, count the number of label values ​​(i.e., the number of regions), then count the proportion of each label value in each label file (i.e., the proportion of the number of samples corresponding to each segmentation region in the total number of samples), and calculate the variance of the proportion of each segmentation region (i.e., the imbalance of that sample). Finally, calculate the imbalance of each sample. In this function, the proportion of each segmentation region and its imbalance in each sample, as well as the proportion of each segmentation region and its imbalance in the entire sample set, will be displayed scrolling in the results display area.

[0092] 6) Sample set optimization: First, set the imbalance threshold, then iterate through the imbalance value of each sample, and finally remove each sample file and its corresponding label file with an imbalance value greater than the threshold from the sample set, thus completing the sample set imbalance optimization process.

[0093] Implementation case of imbalance optimization for the initial seismic data sample set used to solve the region segmentation problem:

[0094] The initial seismic data sample set for a certain work area contains 12,000 samples, with a size of 80*160 (80 traces, 160 sampling points). Because the samples in the set were randomly selected, they contain many samples with high imbalance, resulting in insufficient overall quality. By setting an imbalance threshold of 0.2 and optimizing the sample set by removing samples with an imbalance greater than 0.2, the overall imbalance of the sample set decreased from 0.0196 to 0.0174.

[0095] To verify the quality improvement of the sample set after imbalance optimization, the FCN-U5 network model designed in this invention was trained using the first seismic data sample set and the optimized first seismic data sample set, respectively, and the accuracy improvement was compared. 200 samples were randomly selected from the two seismic data sample sets as the validation set for evaluating network performance, and the remaining samples were used as the training set for training the network model. The FCN-U5 network model was trained using mini-batch gradient descent with a batch size of 70. The loss function used was binary cross-entropy, which was minimized during training. The gradient descent optimization algorithm, dynamic learning rate, and early stopping strategy were set the same as those used in the previous layer-by-layer explanation of the FCN-U5 network model training.

[0096] The FCN-U5 network model with initial arrival picking was trained using the first earthquake data sample set, completing the training process in 54 iterations, with a final validation accuracy of 97.28%. The FCN-U5 network model with initial arrival picking was trained using the first earthquake data sample set with imbalance optimization, completing the training process in 42 iterations, with faster convergence and a final validation accuracy of 99.31%. After sample set optimization, the validation accuracy of the FCN-U5 network model improved by 2.03%, getting closer to 100%.

[0097] Figure 6 This is a schematic diagram of a sample with a high degree of imbalance and its labels. In the diagram: the right side is the label, the proportion of sample points with a value of 1 is 99.35%, the proportion of sample points with a value of 0 is 0.65%, the variance of the proportion of each type of sample is 0.24, that is, the imbalance of the sample is 0.24.

[0098] Example 3

[0099] Figure 7 This is a flowchart illustrating a method for optimizing a seismic data sample set according to Embodiment 3 of this application. This embodiment is an optimization based on the above embodiment. The seismic data sample set mentioned in this embodiment is a second seismic data sample set used to solve the classification problem.

[0100] like Figure 7 As shown, the method in this embodiment of the application specifically includes the following steps:

[0101] S310, Obtain a second earthquake data sample set. If the number of second earthquake data samples corresponding to the target category in the second earthquake data sample set is less than the number threshold, then copy at least one second earthquake data sample from the second earthquake data samples corresponding to the target category.

[0102] The target category can be any category, such as any category other than the category with the largest number of samples.

[0103] The step of copying at least one second seismic data sample from the second seismic data samples corresponding to the target category can be a target number of second seismic data samples. This target number is determined based on the number of samples in the target category and a quantity threshold.

[0104] In this embodiment of the application, optionally, the label of the second seismic data sample reflects the category corresponding to the second seismic data sample; the quantity threshold is determined based on the number of samples included in the category in the second seismic data sample set that reaches the preset requirement.

[0105] (For example, the sum of the number of targets and the number of samples in the target category equals the number threshold; the number threshold equals the number of samples in the category with the largest number of samples in the second seismic data sample set.)

[0106] In this embodiment of the application, optionally, the seismic data sample set is a second seismic data sample set used to solve the classification problem, and the label of the second seismic data sample reflects the category corresponding to the second seismic data sample.

[0107] Specifically, in the second seismic data sample set, for two-dimensional samples, each sample still consists of sample points arranged in a matrix. Each second seismic data sample corresponds to a category, which can be a stratigraphic category. For example, if a second seismic data sample is acquired in stratigraphic A, then the sample is pre-labeled with the category corresponding to stratigraphic A.

[0108] S320, the copied second seismic data sample is processed by randomly and non-repeatedly selecting the target processing method to obtain a new second seismic data sample corresponding to the target category.

[0109] The target processing method includes at least one of the following: inversion processing, noise addition processing, high-pass filtering processing, and low-pass filtering processing.

[0110] For example, if the largest category in the second earthquake data sample set includes 10 samples, category A includes 8 samples, and category B includes 5 samples, then both category A and category B can be target categories. For category A, 2 samples are copied from category A; for category B, 5 samples are copied from category B; and the copied samples are processed randomly and without repetition using the target processing method to obtain new second earthquake data samples corresponding to the target categories.

[0111] It should be noted that when processing with the target processing method, the target processing method can be randomly selected for each sample; or a target processing method that has not been used before can be selected for each sample. For example, four samples can be processed by flipping, adding noise, high-pass filtering and low-pass filtering respectively.

[0112] S330, based on the second seismic data sample corresponding to the target category and the newly added second seismic data sample corresponding to the target category, determine the augmented second seismic data sample corresponding to the target category.

[0113] Specifically, after obtaining the newly added second earthquake data samples corresponding to the target category, the number of samples corresponding to the target category reaches the quantity threshold. Then, each category is traversed to ensure that the number of samples corresponding to each category reaches the quantity threshold.

[0114] S340, based on the augmented second seismic data samples corresponding to each category, determine the augmented second seismic data sample set, and train the machine learning model based on the augmented seismic data sample set.

[0115] Specifically, a machine learning model for solving classification problems is trained based on the augmented seismic data sample set, so that the machine learning model for solving classification problems can accurately identify the category corresponding to the input seismic data.

[0116] In this embodiment of the application, optionally, in the second seismic data sample set, if the number of second seismic data samples corresponding to the target category is less than a quantity threshold, then at least one second seismic data sample is copied from the second seismic data samples corresponding to the target category, including: calculating the number of second seismic data samples corresponding to each category in the second seismic data sample set, and taking the maximum value among the number of second seismic data samples corresponding to each category as the quantity threshold; if the number of second seismic data samples corresponding to the target category is less than the quantity threshold, then determining the sequence number corresponding to each second seismic data sample of the target category; the sequence number is an integer starting from 0; establishing a target array corresponding to the target category; the target array includes randomly arranged integers, and the value of each integer is from 0 to a first value; the first value is equal to the quantity threshold minus 1; dividing each value in the target array by a second value and taking the remainder; the second The numerical value equals the number of second seismic data samples included in the target category; the remainder is used as a sequence number, and if the same remainder appears twice or more, the corresponding second seismic data sample is copied each time the remainder appears repeatedly; correspondingly, the at least one seismic data sample is processed using the target processing method to obtain a new seismic data sample, including: randomly and non-repeatingly selecting the target processing method to process the copied second seismic data sample to obtain a new second seismic data sample corresponding to the target category; correspondingly, based on the seismic data sample set and the new seismic data samples, the augmented seismic data sample set is determined, including: based on the second seismic data samples corresponding to the target category and the new second seismic data samples corresponding to the target category, determining the augmented second seismic data sample corresponding to the target category; based on the augmented second seismic data samples corresponding to each category, determining the augmented second seismic data sample set.

[0117] In one specific embodiment, if the second seismic data sample set includes three categories, the samples in each category are numbered starting from 0. For example, the first category corresponds to three samples with IDs 0, 1, and 2; the second category corresponds to two samples with IDs 0 and 1; and the third category corresponds to five samples with IDs 0, 1, 2, 3, and 4. Then, the five numbers 0, 1, 2, 3, and 4 are randomly arranged three times to obtain three arrays. Then, the number in each array is divided by the number of samples included in the corresponding category to obtain three sets of remainders. Obviously, each set has 5 remainders (for example, in the first array, the five numbers are randomly arranged as 0, 4, 2, 1, and 3; each number is divided by 3 and the remainder is 0). 1, 2, 1, 0); For a set of remainders, the remainders are used as IDs to obtain the samples corresponding to each remainder. If a remainder appears twice or more, the samples corresponding to the repeated remainders are the samples that need to be processed using the target method (for example, when the remainders are 0, 1, 2, 1, 0, the samples corresponding to the last two remainders are the samples that need to be processed using the target method); The samples that need to be processed using the target method are processed using different target processing methods, or a processing method (flipping processing, noise addition processing, high-pass filtering processing, and low-pass filtering processing) is randomly and non-repeatedly selected for processing to obtain the newly added second seismic data samples corresponding to the target category.

[0118] In a specific implementation process, based on the corresponding sample augmentation software ( Figure 8 (Screenshot of software used for augmenting the second earthquake data sample set). The following steps can be used to implement the augmentation function for the second earthquake data sample set:

[0119] 1) Select the sample set description file: The sample set description file is an Excel file containing two columns. The first column is the file name of the second seismic data sample containing the path, and the second column is the label value, i.e., the category of the sample.

[0120] 2) Select sample dimension: Choose whether the samples in the sample set to be augmented are two-dimensional or three-dimensional samples.

[0121] 3) Set the sample size: For two-dimensional samples, you need to set the number of channels and the number of sample points. For three-dimensional samples, you need to set the number of lines, channels, and sample points.

[0122] 4) Select label data type: Select whether the sample labels in the sample set to be augmented are integers or floating-point numbers.

[0123] 5) Select the sample augmentation processing method and set the number of executions and parameters: This step first randomly selects one or more of the four sample augmentation processing methods: rotation (corresponding to flipping processing), noise addition (corresponding to noise addition processing), high frequency extraction (corresponding to high-pass filtering processing), and low frequency extraction (corresponding to low-pass filtering processing). Then, set the number of executions for each processing method, i.e., the number of samples augmented by each processing method. Finally, set the sample augmentation processing parameters. If high frequency extraction or low frequency extraction is selected, the high-pass threshold frequency range and low-pass threshold frequency range need to be set.

[0124] 6) Select the sample set description file to be augmented (including the target number of second seismic data samples). The sample set description file to be augmented should have the same format as the original sample set description file. In addition to the original sample file name and label value, it should also include the file name and label value of the newly generated sample after augmentation.

[0125] 7) Perform class-balanced random augmentation (i.e., the specific process of S310-S330) on the sample set that needs augmentation (target number of second seismic data samples): This step first performs the following 4 sub-steps: ① Assuming there are N categories of samples in the sample set, count the number of samples of each category in the sample set to obtain the array [#C0,#C1,…,#C N-1 ], where the maximum number of samples in the array is denoted by M; ② Randomly arrange the sequence (0,1,…,M-1) to generate N sets of sequences; ③ Divide the N sets of sequences into arrays [#C0,#C1,…,#C N-1 ] Take the remainder to obtain N new sequences representing sample indices; ④ Select duplicate samples from N categories of samples based on the N new sequences to obtain N categories of samples with M samples each; then perform random non-repeating sample augmentation on the duplicate samples in the N categories of samples according to the augmentation processing method, execution number and parameters set in step 5); finally merge the N categories of samples with the same number of samples after augmentation into the final sample set and generate the augmented sample set description file.

[0126] When performing random, non-repeating sample augmentation on the second earthquake data sample set, the parameter setting rules for the four processing methods are the same as those for the conventional random sample augmentation method. However, it is required that the total number of sample processing executions be greater than or equal to the product of the maximum number of samples by category and the number of sample categories in the second earthquake data sample set, minus the number of samples in the second earthquake data sample set. Otherwise, the number of samples in each category cannot reach the maximum number of samples in the second earthquake data sample set, that is, random sample augmentation with category balance cannot be achieved.

[0127] Implementation case of category-based sample augmentation for a second seismic data sample set used to solve classification problems:

[0128] The seismic facies classification sample set (second seismic data sample set) for a certain work area contains 153 samples, with a size of 80*80*80 (80 lines, 80 traces, 80 sampling points). It includes samples of five seismic facies, named Seismic Facies 1, Seismic Facies 2, Seismic Facies 3, Seismic Facies 4, and Seismic Facies 5, with 81, 17, 8, 12, and 35 samples respectively. The label values ​​for the five seismic facies types are 0, 1, 2, 3, and 4. Due to the small sample size and significant differences in the number of samples among different types, the categories are severely imbalanced, resulting in weak predictive and generalization abilities of the trained network model. The sample set was augmented according to the above invention. The number of samples augmented for the four types of seismic phases 2, 3, 4 and 5 were 64, 73, 69 and 46 respectively. The sample augmentation processing methods selected were sample rotation, sample noise addition, high frequency extraction and low frequency extraction. The number of executions for the four processing methods were set to 6, 80, 84 and 82 respectively, and finally an augmented sample set containing 405 samples was obtained.

[0129] This invention utilizes the publicly available VGG-16 deep learning network to perform multi-label classification on the second seismic data sample set and the augmented second seismic data sample set, achieving classification accuracies of 93.78% and 97.52%, respectively. After augmenting the seismic facies classification sample set, the classification accuracy of the VGG-16 network improved by 3.74%.

[0130] In this embodiment of the application, optionally, the target processing method is flipping processing; correspondingly, processing the seismic data sample in the target processing method includes: if the seismic data sample is a two-dimensional seismic data sample, then performing horizontal flipping and / or vertical flipping processing on the two-dimensional seismic data sample; if the seismic data sample is a three-dimensional seismic data sample, then performing horizontal flipping and / or vertical flipping processing based on any of the following planes; the planes include: the InLine-Xline plane, the InLine-time plane, and the XLine-time plane.

[0131] Accordingly, seismic data samples are processed using a target-oriented processing method, including:

[0132] If the seismic data sample is a two-dimensional seismic data sample, then the data in the j-th column and the (mj)-th column of the two-dimensional seismic data sample are interchanged; and / or, the data in the i-th row and the (pi)-th row are interchanged; where m is the total number of columns in the two-dimensional seismic data sample and p is the total number of rows in the two-dimensional seismic data sample;

[0133] If the seismic data sample is a three-dimensional seismic data sample, then the three-dimensional seismic data sample shall be subjected to any one or more of the following processes:

[0134] Interchange the data of the a-th and (ba)-th planes of the InLine-Xline plane;

[0135] Interchange the data on the c-th and (dc)-th planes of the InLine-time plane;

[0136] Interchange the data of the e-th and (fe)-th planes of the XLine-time plane;

[0137] Wherein, b is the total number of faces in the InLine-Xline plane, d is the total number of faces in the InLine-Time plane, and f is the total number of faces in the XLine-Time plane.

[0138] It should be noted that since the flipping of 2D seismic data samples involves both horizontal and vertical flipping, the number of times the sample rotation is performed on 2D seismic data must be within 2 (for example, if three samples are flipped, two samples will have the same flipping effect, and the effect of these two samples on model training is similar to that of one sample; the same principle applies when limiting the number of times each processing method is performed throughout the text); 3D seismic data samples can be flipped horizontally and vertically in three planes: the InLine-XLine plane, the InLine-time plane, and the XLine-time plane, so the number of times the sample rotation is performed on 3D seismic data samples must be within 6.

[0139] In this embodiment of the application, optionally, the target processing method is noise addition processing; correspondingly, processing the seismic data samples in the target processing method includes: determining the values ​​of each sample point (each seismic data) in the i-th seismic data sample using the following formula:

[0140]

[0141] Among them, A n A represents the values ​​of the sample points after adding noise. r A represents the values ​​of the sample points before noise was added. max A represents the maximum value of all samples in the earthquake data sample. min Let be the minimum value of all samples in the seismic data sample, and 'i' be the execution order of the noise addition process, i.e., the i-th noise addition process. This setting ensures that the random noise value added to each sample is different during noise addition.

[0142] In this embodiment of the application, optionally, the target processing method is high-pass filtering; correspondingly, processing the seismic data samples in the target processing method includes: determining the value of the target sample after high-pass filtering based on the portion of the target sample's frequency that is greater than or equal to a high-pass threshold frequency threshold; the target processing method is low-pass filtering; correspondingly, processing the seismic data samples in the target processing method includes: determining the value of the target sample after low-pass filtering based on the portion of the target sample's frequency that is less than or equal to a low-pass threshold frequency threshold; wherein, the high-pass threshold frequency threshold is determined using the following formula:

[0143] ω hc =ω hmin +(ω hmax -ω hmin ) / (N h )*i;

[0144] Where, ω hc ω is the high-pass threshold frequency. hmin ω is the minimum value of the high-pass threshold frequency range. hmax N represents the maximum value of the high-pass threshold frequency range. h N represents the number of times high-frequency data was extracted from the samples, and i represents the execution order of high-frequency data extraction; (for example, if 5 samples were subjected to 5 high-pass filtering processes, N...) h The number of times high frequencies are extracted from the sample is 5, where i represents the i-th high frequency extraction.

[0145] The low-pass threshold frequency is determined using the following formula:

[0146] ω lc =ω lmin +(ω lmax -ω lmin ) / (N l )*i;

[0147] Where, ω lc ω is the low-pass threshold frequency. lmin ω is the minimum value of the low-pass threshold frequency range. lmax N represents the maximum value of the low-pass threshold frequency range. l denoted as the number of times low-frequency samples are extracted, and i represents the execution order of low-frequency sample extraction.

[0148] For example, high-pass filtering is performed on each channel in the seismic data sample to extract high-frequency information. The frequency response of the high-pass filter is as follows:

[0149]

[0150] Where H(ω) is the frequency response function of the high-pass filter, ω is the frequency, and ω ch This is the high-pass threshold frequency.

[0151] For low-pass filtering: A low-pass filter is used for low-frequency extraction from the samples. Each channel in the seismic data sample is subjected to low-pass filtering to extract low-frequency information. The frequency response of the low-pass filter is:

[0152]

[0153] In the formula: L(ω) is the frequency response function of the low-pass filter, ω is the frequency, and ω cl This is the low-pass threshold frequency.

[0154] Figure 9 A schematic diagram of the flipping process of stratified interpretation samples. In the figure, (a) is stratified interpretation sample A; (b) is the label of sample A; (c) is sample B obtained by horizontally flipping sample A; and (d) is the label of sample B.

[0155] Figure 10 A schematic diagram of high-pass filtering processing of the stratified interpretation samples; (a) is the stratified interpretation sample C; (b) is the label of sample C; (c) is the high-frequency extraction processing of sample C to obtain sample D (high-pass filtering threshold frequency: 48Hz); (d) is the label of sample D.

[0156] Figure 11 A schematic diagram of high-pass filtering processing for seismic facies segmentation samples; (a) is seismic facies segmentation sample A (labeled 0, indicating a certain seismic facies); (b) is sample B obtained by extracting high frequencies from sample A (sample B is also labeled 0, high-pass filtering threshold frequency: 76Hz).

[0157] Example 4

[0158] Figure 12 This is a schematic diagram of a seismic data sample set optimization device provided in Embodiment 4 of this application. This device can execute the seismic data sample set optimization method provided in any embodiment of this invention, and possesses the corresponding functional modules and beneficial effects for executing the method. Figure 12 As shown, the device includes:

[0159] The earthquake data sample copying module 410 is used to acquire an earthquake data sample set and copy at least one earthquake data sample from the earthquake data sample set.

[0160] The seismic data sample processing module 420 is used to process the at least one seismic data sample in a target processing manner to obtain a new seismic data sample;

[0161] The augmented earthquake data sample set determination module 430 is used to determine the augmented earthquake data sample set based on the earthquake data sample set and the newly added earthquake data samples, so as to train a machine learning model based on the augmented earthquake data sample set.

[0162] The target processing method includes at least one of the following: inversion processing, noise addition processing, high-pass filtering processing, and low-pass filtering processing;

[0163] The seismic data sample copying module 410 includes: a first seismic data sample copying unit, used to copy at least one first seismic data sample from the first seismic data sample set if the seismic data sample set is a first seismic data sample set used to solve the regional segmentation problem;

[0164] The seismic data sample copying module 410 includes a second seismic data sample copying unit, which is used to copy at least one second seismic data sample from the second seismic data sample corresponding to the target category if the seismic data sample set is a second seismic data sample set used to solve a classification problem.

[0165] The technical solution of this application embodiment includes: a seismic data sample copying module 410, used to acquire a seismic data sample set and copy at least one seismic data sample from the seismic data sample set; a seismic data sample processing module 420, used to process the at least one seismic data sample in a target processing mode to obtain a new seismic data sample; and an augmented seismic data sample set determination module 430, used to determine an augmented seismic data sample set based on the seismic data sample set and the new seismic data sample, so as to train a machine learning model based on the augmented seismic data sample set; the target processing mode includes at least one of the following: flipping processing, noise addition processing, high-pass filtering processing, and low-pass filtering processing. This technical solution, by processing the seismic data samples in a target processing mode to obtain new seismic data samples, thereby increasing the number of seismic data samples, and improving the generalization ability and other performance aspects of the machine learning model after training it based on the augmented seismic data sample set.

[0166] In this embodiment of the application, optionally, the first seismic data sample consists of seismic data arranged in a matrix, wherein the seismic data arranged in a matrix corresponds to each region on the geological profile, and the label corresponding to the first seismic data sample includes the segmented region corresponding to each seismic data.

[0167] In this embodiment of the application, optionally, the label of the second seismic data sample reflects the category corresponding to the second seismic data sample; the quantity threshold is determined based on the number of samples included in the category in the second seismic data sample set that reaches the preset requirement.

[0168] In this embodiment of the application, optionally, the target processing method is flipping processing;

[0169] Correspondingly, the seismic data sample processing module 420 includes:

[0170] A two-dimensional seismic data sample processing unit is used to perform horizontal and / or vertical flipping processing on the two-dimensional seismic data sample if the seismic data sample is a two-dimensional seismic data sample.

[0171] A three-dimensional seismic data sample processing unit is used to perform horizontal and / or vertical flipping processing based on any of the following planes if the seismic data sample is a three-dimensional seismic data sample; the planes include: the InLine-Xline plane, the InLine-time plane, and the XLine-time plane.

[0172] In this embodiment of the application, optionally, the target processing method is noise addition processing;

[0173] Correspondingly, the seismic data sample processing module 420 includes:

[0174] The sample point value determination unit is used to determine the value of each sample point in the i-th seismic data sample using the following formula:

[0175]

[0176] Among them, A n A represents the values ​​of the sample points after adding noise. r A represents the values ​​of the sample points before noise was added. max A represents the maximum value of all samples in the earthquake data sample. min is the minimum value of all sample points in the seismic data sample, and i is the index of the seismic data sample.

[0177] In this embodiment of the application, optionally, the target processing method is high-pass filtering processing;

[0178] Correspondingly, the seismic data sample processing module 420 includes:

[0179] The high-pass filter unit is used to determine the value of the target sample after high-pass filtering based on the part of the target sample whose frequency is greater than or equal to the high-pass threshold frequency.

[0180] The target processing method is low-pass filtering; correspondingly, the seismic data sample processing module 420 includes:

[0181] The low-pass filter unit is used to determine the value of the target sample after low-pass filtering based on the part of the target sample whose frequency is less than or equal to the low-pass threshold frequency.

[0182] The high-pass threshold frequency is determined using the following formula:

[0183] ω hc =ω hmin +(ω hmax -ω hmin ) / (N h )*i;

[0184] Where, ω hc ω is the high-pass threshold frequency. hmin ω is the minimum value of the high-pass threshold frequency range. hmax N represents the maximum value of the high-pass threshold frequency range. h denoted as the number of times high-frequency samples are extracted, and i represents the execution order of high-frequency sample extraction.

[0185] The low-pass threshold frequency is determined using the following formula:

[0186] ω lc =ω lmin +(ω lmax -ω lmin ) / (N l )*i;

[0187] Where, ω lc ω is the low-pass threshold frequency. lmin ω is the minimum value of the low-pass threshold frequency range. lmax N represents the maximum value of the low-pass threshold frequency range. l denoted as the number of times low-frequency samples are extracted, and i represents the execution order of low-frequency sample extraction.

[0188] In this embodiment of the application, optionally, the device further includes:

[0189] The imbalance calculation unit is used to calculate the imbalance of each initial seismic data sample for the initial seismic data sample set used to solve the regional segmentation problem; the imbalance reflects the proportion of each segmented region in the sample.

[0190] The imbalance optimization unit is used to delete samples in the initial seismic data sample set whose imbalance degree is greater than the imbalance threshold, and use the remaining samples as the first seismic data sample set.

[0191] In this embodiment of the application, the optional second seismic data sample replication unit includes:

[0192] The sample number calculation subunit is used to calculate the number of second earthquake data samples corresponding to each category in the second earthquake data sample set, and the maximum value of the number of second earthquake data samples corresponding to each category is used as the number threshold.

[0193] The sample sequence number determination subunit is used to determine the sequence number corresponding to each second seismic data sample of the target category if the number of second seismic data samples corresponding to the target category is less than the number threshold; the sequence number is an integer starting from 0.

[0194] A target array creation sub-unit is used to create a target array corresponding to the target category; the target array includes randomly arranged integers, and the value of each integer ranges from 0 to a first value; the first value is equal to the quantity threshold minus 1;

[0195] The remainder determination sub-unit is used to divide each value in the target array by the second value and take the remainder; the second value is equal to the number of second seismic data samples included in the target category;

[0196] The second seismic data sample replication sub-unit is used to use the remainder as the sequence number. If the same remainder appears twice or more, the corresponding second seismic data sample is replicated each time the remainder is repeated.

[0197] Correspondingly, the seismic data sample processing module 420 includes:

[0198] The second seismic data sample processing unit is used to randomly and non-repeatedly select the target processing method to process the copied second seismic data sample, so as to obtain a new second seismic data sample corresponding to the target category.

[0199] Accordingly, the augmented seismic data sample set determination module 430 includes:

[0200] The augmented second seismic data sample determination unit corresponding to the target category is used to determine the augmented second seismic data sample corresponding to the target category based on the second seismic data sample corresponding to the target category and the newly added second seismic data sample corresponding to the target category.

[0201] The augmented second seismic data sample set determination unit is used to determine the augmented second seismic data sample set based on the augmented second seismic data samples corresponding to each category.

[0202] The seismic data sample set optimization device provided in this application embodiment can execute the seismic data sample set optimization method provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the method execution.

[0203] Example 5

[0204] Figure 13 A schematic diagram of an electronic device 10 that can be used to implement embodiments of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0205] like Figure 13 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 may also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0206] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0207] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as optimization methods for seismic data sample sets.

[0208] In some embodiments, the method for optimizing the seismic data sample set may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the method for optimizing the seismic data sample set described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to execute the method for optimizing the seismic data sample set by any other suitable means (e.g., by means of firmware).

[0209] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0210] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0211] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0212] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0213] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or computing systems that include middleware components (e.g., application servers), or computing systems that include frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0214] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0215] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0216] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for optimizing a seismic data sample set, characterized in that, include: Obtain a set of earthquake data samples, and copy at least one earthquake data sample from the set of earthquake data samples; The at least one seismic data sample is processed using a target processing method to obtain a new seismic data sample; Based on the earthquake data sample set and the newly added earthquake data samples, an augmented earthquake data sample set is determined, and a machine learning model is trained based on the augmented earthquake data sample set. The target processing method includes at least one of the following: inversion processing, noise addition processing, high-pass filtering processing, and low-pass filtering processing; If the seismic data sample set is a first seismic data sample set used to solve the regional segmentation problem, then copying at least one seismic data sample from the seismic data sample set includes: copying at least one first seismic data sample from the first seismic data sample set. If the earthquake data sample set is a second earthquake data sample set used to solve the classification problem, then copying at least one earthquake data sample from the earthquake data sample set includes: in the second earthquake data sample set, if the number of second earthquake data samples corresponding to the target category is less than a number threshold, then copying at least one second earthquake data sample from the second earthquake data samples corresponding to the target category. The first seismic data sample consists of seismic data arranged in a matrix, which corresponds to different regions on a geological profile. The labels of the first seismic data sample include the segmented regions corresponding to each seismic data.

2. The method according to claim 1, characterized in that, The label of the second earthquake data sample reflects the category corresponding to the second earthquake data sample; the quantity threshold is determined based on the number of samples included in the category that reaches the preset requirement in the second earthquake data sample set.

3. The method according to claim 1, characterized in that, The target processing method is flipping; Accordingly, seismic data samples are processed using a target-oriented processing method, including: If the seismic data sample is a two-dimensional seismic data sample, then the two-dimensional seismic data sample is horizontally flipped and / or vertically flipped. If the seismic data sample is a three-dimensional seismic data sample, then it is horizontally flipped and / or vertically flipped based on any of the following planes; the planes include: InLine-Xline plane, InLine-time plane, and XLine-time plane.

4. The method according to claim 1, characterized in that, The target processing method is noise addition processing; Accordingly, seismic data samples are processed using a target-oriented processing method, including: The values ​​of each sample point in the i-th seismic data sample are determined using the following formula: ； in, These are the values ​​of the sample points after adding noise. These are the values ​​of the sample points before noise was added. This represents the maximum value of all sample points in the earthquake data sample. is the minimum value of all sample points in the seismic data sample, and i is the index of the seismic data sample.

5. The method according to claim 1, characterized in that, The target processing method is high-pass filtering; Accordingly, seismic data samples are processed using a target-oriented processing method, including: The value of the target sample after high-pass filtering is determined based on the portion of the target sample whose frequency is greater than or equal to the high-pass threshold frequency. The target processing method is low-pass filtering; correspondingly, the seismic data samples are processed using the target processing method, including: The value of the target sample after low-pass filtering is determined based on the portion of the target sample whose frequency is less than or equal to the low-pass threshold frequency. The high-pass threshold frequency is determined using the following formula: ； in, This is the high-pass threshold frequency. This is the minimum value of the high-pass threshold frequency range. This represents the maximum value within the high-pass threshold frequency range. For high-frequency extraction of samples, The execution order for extracting high-frequency data from samples; The low-pass threshold frequency is determined using the following formula: ； in, This is the low-pass threshold frequency threshold. This is the minimum value of the low-pass threshold frequency range. This represents the maximum value within the low-pass threshold frequency range. Extract low-frequency times from the sample. The execution order for extracting low-frequency samples.

6. The method according to claim 1, characterized in that, The method further includes: For the initial seismic data sample set used to solve the regional segmentation problem, the imbalance degree of each initial seismic data sample is calculated; the imbalance degree reflects the proportion of each segmented region in the sample. In the initial seismic data sample set, samples with an imbalance degree greater than the imbalance threshold are deleted, and the remaining samples are used as the first seismic data sample set.

7. The method according to claim 2, characterized in that, In the second seismic data sample set, if the number of second seismic data samples corresponding to the target category is less than the number threshold, then at least one second seismic data sample is copied from the second seismic data samples corresponding to the target category, including: Calculate the number of second earthquake data samples corresponding to each category in the second earthquake data sample set, and take the maximum value of the number of second earthquake data samples corresponding to each category as the quantity threshold; If the number of second seismic data samples corresponding to the target category is less than the number threshold, then the sequence number corresponding to each second seismic data sample of the target category is determined; the sequence number is an integer starting from 0. Establish a target array corresponding to the target category; the target array includes randomly arranged integers, each integer ranging from 0 to a first value; the first value is equal to the quantity threshold minus 1; Divide each value in the target array by the second value and take the remainder; the second value is equal to the number of second seismic data samples included in the target category; Use the remainder as the sequence number. If the same remainder appears twice or more, copy the corresponding second seismic data sample each time the remainder appears. Accordingly, the at least one seismic data sample is processed using a target processing method to obtain a new seismic data sample, including: The copied second seismic data sample is processed by randomly and non-repeatedly selecting the target processing method to obtain a new second seismic data sample corresponding to the target category; Accordingly, based on the aforementioned seismic data sample set and the newly added seismic data samples, the augmented seismic data sample set is determined, including: Based on the second seismic data sample corresponding to the target category and the newly added second seismic data sample corresponding to the target category, determine the augmented second seismic data sample corresponding to the target category; Based on the augmented second seismic data samples corresponding to each category, determine the augmented second seismic data sample set.

8. An optimization device for a seismic data sample set, characterized in that, include: The seismic data sample copying module is used to acquire a seismic data sample set and copy at least one seismic data sample from the seismic data sample set; The seismic data sample processing module is used to process the at least one seismic data sample in a target processing manner to obtain a new seismic data sample; The augmented seismic data sample set determination module is used to determine the augmented seismic data sample set based on the seismic data sample set and the newly added seismic data samples, so as to train a machine learning model based on the augmented seismic data sample set. The target processing method includes at least one of the following: inversion processing, noise addition processing, high-pass filtering processing, and low-pass filtering processing; The seismic data sample copying module includes: a first seismic data sample copying unit, used to copy at least one first seismic data sample from the first seismic data sample set if the seismic data sample set is a first seismic data sample set used to solve the regional segmentation problem; The seismic data sample copying module includes: a second seismic data sample copying unit, used to copy at least one second seismic data sample in the second seismic data sample corresponding to the target category if the seismic data sample set is a second seismic data sample set used to solve a classification problem; The first seismic data sample consists of seismic data arranged in a matrix, which corresponds to different regions on a geological profile. The labels of the first seismic data sample include the segmented regions corresponding to each seismic data.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the optimization method for the seismic data sample set according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the method for optimizing the seismic data sample set according to any one of claims 1-7.

Images