Seismic event identification method and device based on machine learning, equipment and medium

Abstract

The invention relates to the field of seismic exploration, in particular to a seismic event identification method and device based on machine learning, equipment and a medium, and the method comprises the steps: obtaining a seismic signal: carrying out the forward numerical simulation of the seismic signal through a thin interbed velocity model, and obtaining to-be-processed training data; performing single-channel segmentation and data recombination on the to-be-processed training data to obtain the training data; performing feature processing on the training data to obtain a feature vector; performing feature screening on the feature vectors to obtain screened feature vectors; learning the screened feature vectors by using a gradient boosting decision tree learning method to obtain a seismic event recognition model; and obtaining test data, and identifying the test data by using the seismic event identification model to obtain seismic event data. According to the method, the seismic event data can be accurately and rapidly extracted from a large number of seismic signals, and the processing efficiency and accuracy of the seismic event identification model are improved.

Description

Technical Field

[0001] The present invention relates to the field of seismic exploration, and in particular to a seismic event axis identification method, device, equipment and medium based on machine learning. Background Art

[0002] Seismic exploration is an important technology in earth science, which is usually used to study the structure and properties of the earth's interior. By analyzing the propagation and reflection characteristics of seismic waves in the earth's crust, scientists can reveal key information such as the earth's structure, rock layer distribution, and crustal movement.

[0003] In the field of seismic exploration, automatic extraction of seismic events is a crucial step in the data processing process. However, in this step, since the seismic event signal is usually weak and easily masked by background noise, it is difficult to accurately detect the first arrival of the signal, and it also brings certain difficulties to determine the precise location of the signal. Therefore, how to quickly and accurately identify the seismic event data in a large amount of seismic data has become a difficult problem that needs to be solved urgently. Summary of the invention

[0004] In view of the above problems, embodiments of the present invention provide a method, device, equipment and medium for seismic event identification based on machine learning.

[0005] In a first aspect, an embodiment of the present invention provides a method for identifying seismic events based on machine learning, comprising:

[0006] Acquire seismic signals, and use a thin interlayer velocity model to perform forward numerical simulation on the seismic signals to obtain training data to be processed;

[0007] Performing single-channel segmentation and data reorganization on the training data to be processed to obtain training data;

[0008] Performing feature processing on the training data to obtain a feature vector;

[0009] Performing feature screening on the feature vector to obtain a screened feature vector;

[0010] Using a gradient boosting decision tree learning method to learn the screening feature vector, to obtain a seismic event recognition model;

[0011] Acquire test data, and use the seismic event recognition model to identify the test data to obtain seismic event data.

[0012] According to an embodiment of the present invention, a thin interbed velocity model is used to perform forward numerical simulation on the seismic signal to obtain training data to be processed, including:

[0013] Calculating the seismic signal according to the finite difference method and the thin interlayer velocity model to obtain the wave field condition;

[0014] Extracting position information of the wave field to obtain forward synthetic shot record data;

[0015] The forward synthetic shot record data is divided into preset values ​​to obtain training data.

[0016] According to an embodiment of the present invention, the step of performing single-channel segmentation and data reorganization on the training data to be processed to obtain the training data includes:

[0017] Segmenting the training data to obtain segmented data;

[0018] Performing equal division processing on the segmented data to obtain reorganized data;

[0019] Noise is added to the reorganized data to obtain training data.

[0020] According to an embodiment of the present invention, the step of performing feature processing on the training data to obtain a feature vector includes:

[0021] Performing single-channel feature extraction on the training data to obtain singular feature data;

[0022] Performing gray-level co-occurrence matrix feature extraction on the training data to obtain gray-level co-occurrence matrix feature data;

[0023] Extracting inter-channel features from the training data to obtain inter-channel feature data;

[0024] Constructing feature vectors for the gray-level co-occurrence matrix feature data, the single-channel feature extraction, and the inter-channel feature data to obtain feature data;

[0025] The feature data is standardized to obtain a feature vector.

[0026] According to an embodiment of the present invention, the step of performing feature screening on the feature vector to obtain the screened feature vector includes:

[0027] Performing univariate feature analysis on the feature vector using variance analysis to obtain a univariate feature vector;

[0028] The single variable feature vector is subjected to feature selection by using an external model of a support vector machine to obtain a screening feature vector.

[0029] According to an embodiment of the present invention, the method of performing univariate feature analysis on the feature vector by using variance analysis to obtain the univariate feature vector includes:

[0030] Calculating the sum of squares of deviations between the eigenvectors to obtain vector deviations;

[0031] Calculating the square sum of the deviations of the feature vector to obtain a training deviation;

[0032] Calculating test statistics of the feature vectors one by one according to the training deviation and the vector deviation;

[0033] The test statistics are sorted in descending order, and features corresponding to the test statistics that meet preset conditions are screened out as univariate feature vectors.

[0034] According to an embodiment of the present invention, the step of performing feature selection on the univariate feature vector using an external model of a support vector machine to obtain a screening feature vector includes:

[0035] Calculate the importance score of each of the univariate features using a preset support vector external model;

[0036] Univariate features are screened out according to their importance scores to obtain the screened feature vector.

[0037] In a second aspect, an embodiment of the present invention provides a seismic event identification device based on machine learning, characterized in that it includes:

[0038] A data acquisition module is used to acquire seismic signals, and to perform forward numerical simulation on the seismic signals using a thin interlayer velocity model to obtain training data to be processed;

[0039] A data processing module, used for performing single-channel segmentation and data reorganization on the training data to be processed to obtain training data;

[0040] A feature processing module, used to perform feature processing on the training data to obtain a feature vector;

[0041] A feature screening module, used for screening the feature vector to obtain a screened feature vector;

[0042] A model training module is used to learn the screening feature vector using a gradient boosting decision tree learning method to obtain a seismic event recognition model;

[0043] The model application module is used to obtain test data, identify the test data using the seismic event recognition model, and obtain seismic event data.

[0044] In a third aspect, an embodiment of the present invention provides an electronic device, comprising:

[0045] processor;

[0046] a memory for storing instructions executable by the processor;

[0047] Wherein, the processor is configured to execute the instructions to implement the steps of a method for seismic event axis identification based on machine learning as described in the first aspect above.

[0048] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of a method for seismic event axis identification based on machine learning as described in the first aspect.

[0049] Compared with the prior art, the above technical solution of the present invention has the following beneficial effects:

[0050] The embodiment of the present invention performs data segmentation on training data to obtain segmented data, performs equal division processing on the segmented data to obtain reorganized data, and adds noise to the reorganized data to obtain training data, thereby reducing the latitude of the obtained training data and improving the recognition efficiency of the seismic event recognition model; obtains a decision tree model, initializes the decision tree model to obtain a strong learner, calculates a negative gradient based on the strong learner and a screened feature vector, obtains a weak learner based on negative gradient training, and trains and optimizes the strong learner based on the weak learner, the strong learner and the screened feature vector, which can gradually reduce the residual to reduce the variance of the model, thereby reducing the risk of overfitting, so that the obtained seismic event recognition model can quickly process low signal-to-noise ratio data or event seismic signals submerged by background noise, thereby improving the processing efficiency and accuracy of the seismic event recognition model. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0052] Figure 1 A flowchart showing a method for identifying seismic event based on machine learning according to the first embodiment of the present invention is shown;

[0053] Figure 2 The thin interbed velocity model used for forward simulation in the first embodiment of the present invention is shown;

[0054] Figure 3 A graph showing synthetic shot record data obtained by forward modeling in the first embodiment of the present invention is shown;

[0055] Figure 4A training data graph obtained by performing single-channel segmentation and data reorganization in Embodiment 1 of the present invention is shown;

[0056] Figure 5 The univariate characteristic analysis diagram obtained by using variance analysis method in the first embodiment of the present invention is shown;

[0057] Figure 6 A schematic diagram of the external model of a support vector machine according to the first embodiment of the present invention is shown;

[0058] Figure 7 A test data diagram for testing and optimizing a model according to the first embodiment of the present invention is shown;

[0059] Figure 8 It shows the seismic event data diagram obtained by testing the seismic event identification model based on the test data of the first embodiment of the present invention;

[0060] Fig. 9 A comparison diagram of single-channel shot records of the first embodiment of the present invention is shown;

[0061] Fig.10 A functional module diagram of a seismic event identification device based on machine learning according to a third embodiment of the present invention is shown;

[0062] Fig.11 A schematic diagram of the composition structure of an electronic device for implementing the seismic event axis identification method based on machine learning according to Embodiment 5 of the present invention is shown. DETAILED DESCRIPTION

[0063] The present disclosure is further described below in conjunction with the embodiments shown in the accompanying drawings.

[0064] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0065] This paper proposes a seismic event recognition method based on machine learning. Based on the model forward theory and combined with the feature screening method, a seismic event recognition model is built to realize the recognition of seismic event data. Compared with traditional methods, the seismic phase analysis technology based on deep learning is more efficient and reduces human subjectivity. It has great potential and application prospects in the field of seismic exploration.

[0066] Embodiment 1

[0067] like Figure 1 As shown, the present invention proposes a seismic event identification method based on machine learning, comprising the following steps:

[0068] S1. Acquire seismic signals, and use a thin interlayer velocity model to perform forward numerical simulation on the seismic signals to obtain training data to be processed.

[0069] In an embodiment of the present invention, by using a thin interlayer velocity model to perform forward numerical simulation on the seismic signal, unprocessed training data that can be used to train monitoring and prediction models is generated, and the data set of training data is enriched, so that the obtained unprocessed training data is more reliable.

[0070] In an optional embodiment of the present invention, the finite difference method can be used for forward numerical simulation; when the finite difference method is used for forward numerical simulation, since the basic idea of ​​the finite difference method is relatively simple and easy to implement, the grid size and time step can be flexibly adjusted to balance numerical stability and simulation accuracy, so that the accuracy of the training data to be processed is higher. In another optional embodiment of the present invention, the finite element method can also be used for forward numerical simulation; when the finite element method is used for forward numerical simulation, since the finite element method uses a matrix expression, the problem can be described very simply, making the method of solving the problem faster, so the efficiency of obtaining the training data to be processed is higher.

[0071] In the embodiment of the present invention, the seismic signal is forward modeled by using a thin interlayer velocity model to obtain training data to be processed, including:

[0072] Calculating the seismic signal according to the finite difference method and the thin interlayer velocity model to obtain the wave field condition;

[0073] Extracting position information of the wave field to obtain forward synthetic shot record data;

[0074] The forward synthetic shot record data is divided into preset values ​​to obtain training data.

[0075] See also Figure 2 As shown, it is the thin interlayer velocity model used for forward simulation, wherein the thin interlayer refers to the repeated appearance of two rock layers. The thin interlayer velocity model is a discrete grid structure, which decomposes the underground rock layer into a series of thin layers and specifies a velocity value for each layer. The middle 10 layers are low-speed thin interlayer bodies.

[0076] See also Figure 3 As shown, this is a synthetic gun record data diagram obtained through forward simulation. Through the synthetic gun record data diagram, more richness and diversity can be provided for the training data to cover more different situations, thereby improving the robustness of the trained model.

[0077] In an embodiment of the present invention, forward numerical simulation of the designed seismic signal can obtain more accurate data. By dividing the data into signal and noise data according to preset values, the noise data in the training set can be excluded, thereby reducing the model's learning and adaptation to noise, thereby simplifying the model's training process and improving training efficiency.

[0078] S2. Segment the training data to be processed into single channels and reorganize the data to obtain training data.

[0079] In the embodiment of the present invention, by performing single-channel segmentation and data reorganization on the training data to be processed, the dimension of the obtained training data can be reduced, thereby improving the efficiency of subsequent feature processing.

[0080] In an embodiment of the present invention, the step of performing single-channel segmentation and data reorganization on the training data to be processed to obtain the training data includes:

[0081] Segmenting the training data to obtain segmented data;

[0082] Performing equal division processing on the segmented data to obtain reorganized data;

[0083] Noise is added to the reorganized data to obtain training data.

[0084] See also Figure 4 As shown, it is a training data diagram used in the present invention. By segmenting, reorganizing and adding random noise to the training data to be processed, the data to be processed can be enhanced to prevent the risk of data overfitting.

[0085] In an embodiment of the present invention, a threshold-based segmentation method can be used to perform data segmentation processing; when the threshold-based segmentation method is used for segmentation, since the threshold-based segmentation method can segment the training data based on specific features, the accuracy of the segmented data obtained is higher.

[0086] In the embodiment of the present invention, by performing single-channel segmentation and data reorganization on the training data to be processed, the dimension of the data can be reduced, the complexity of the data can be reduced, the calculation efficiency of the data can be improved, and the risk of overfitting can be reduced.

[0087] S3. Perform feature processing on the training data to obtain a feature vector.

[0088] In the embodiment of the present invention, by performing feature processing on the data, the distribution of the data can be adjusted and the key information of the data can be captured, thereby ensuring that subsequent model training and analysis can be carried out effectively.

[0089] In an embodiment of the present invention, the minimum-maximum standardization method can be used to perform feature processing; when the minimum-maximum standardization method is used to perform feature processing, since the minimum-maximum standardization method maps the training data to a range between 0 and 1, the method is simple, and thus the obtained feature vector is more efficient.

[0090] In the embodiment of the present invention, the step of performing feature processing on the training data to obtain a feature vector includes:

[0091] Performing single-channel feature extraction on the training data to obtain singular feature data;

[0092] Performing gray-level co-occurrence matrix feature extraction on the training data to obtain gray-level co-occurrence matrix feature data;

[0093] Extracting inter-channel features from the training data to obtain inter-channel feature data;

[0094] Constructing feature vectors for the gray-level co-occurrence matrix feature data, the single-channel feature extraction, and the inter-channel feature data to obtain feature data;

[0095] The feature data is standardized to obtain a feature vector.

[0096] In the embodiment of the present invention, the characteristic data may be standardized using the following formula:

[0097]

[0098] Wherein, x represents the feature vector, x' represents the feature vector, min(x) represents the minimum value in the feature data, and max(x) represents the maximum value in the feature data.

[0099] In the embodiment of the present invention, referring to Appendix 1, the preset feature table includes 191 features, including 63 single-channel features such as mean, energy, skewness and kurtosis, and 128 inter-channel features, gray-level co-occurrence matrix contrast, correlation, energy and homogeneity.

[0100] Schedule 1

[0101]

[0102]

[0103] In an embodiment of the present invention, by constructing new features, the nonlinear relationship in the original data to be trained can be converted into a linear relationship, so that the subsequent model can better fit the data. Standardizing the feature data can ensure that the data is within a uniform distribution range, thereby reducing the subsequent model training error.

[0104] S4. Perform feature screening on the feature vector to obtain a screened feature vector.

[0105] In an embodiment of the present invention, by performing feature screening on the feature vector, extraction can be performed based on the feature vector, reducing the dimension of the feature space, removing irrelevant or redundant features, and improving the efficiency of data utilization and the efficiency of subsequent model training.

[0106] In the embodiment of the present invention, the step of performing feature screening on the feature vector to obtain the screened feature vector includes:

[0107] Performing univariate feature analysis on the feature vector using variance analysis to obtain a univariate feature vector;

[0108] The single variable feature vector is subjected to feature selection by using an external model of a support vector machine to obtain a screening feature vector.

[0109] In an embodiment of the present invention, the support vector machine external model is a classified supervised learning algorithm, including a mathematical expression part of a hyperplane and a mathematical expression part of a constraint; in the present invention, it is used to perform a regression task on the univariate feature vector to obtain a contribution value of each of the univariate features, and to perform screening based on the contribution value to obtain a screening vector feature; in the present invention, the feature vector is assigned corresponding weights through the recursive support vector machine external model, and features are selected from the recursively reduced feature set to obtain a screening feature vector.

[0110] In the embodiment of the present invention, the method of performing univariate feature analysis on the feature vector by using variance analysis to obtain the univariate feature vector includes:

[0111] Calculating the sum of squares of deviations between the eigenvectors to obtain vector deviations;

[0112] Calculating the square sum of the deviations of the feature vector to obtain a training deviation;

[0113] Calculating test statistics of the feature vectors one by one according to the training deviation and the vector deviation;

[0114] The test statistics are sorted in descending order, and features corresponding to the test statistics that meet preset conditions are screened out as univariate feature vectors.

[0115] In the embodiment of the present invention, the test statistic can be calculated using the following formula:

[0116]

[0117] Wherein, F represents the test feature statistic, S A represents the first A vector deviations, SE represents the first E training deviations, r is the total number of the feature vectors, and n is a preset fixed constant.

[0118] See attached Figure 5 As shown in the figure, it is a univariate feature analysis chart, which expresses the score result of univariate feature selection. The variance analysis is used to calculate the test statistic F value, and the univariate feature selection is performed on the features. The dimension of the obtained features is reduced to reduce the probability of overfitting; by calculating the F value between the features, the feature vectors with the first 30% of the F value scores can be selected as the univariate feature vectors.

[0119] In an embodiment of the present invention, the variance analysis method can be used to perform univariate feature analysis. Since the variance analysis method is based on whether different levels of feature vectors have different effects on classification, the variance analysis method can accurately determine whether each feature vector has a significant effect on the classification accuracy of the training data, thereby achieving the purpose of analyzing the univariate feature vector. Therefore, the univariate feature vector obtained by the variance analysis method has higher accuracy.

[0120] In the embodiment of the present invention, the use of the external model of the support vector machine to perform feature selection on the univariate feature vector to obtain the screening feature vector includes:

[0121] Calculate the importance score of each of the univariate features using a preset support vector external model;

[0122] Univariate features are screened out according to their importance scores to obtain the screened feature vector.

[0123] In the embodiment of the present invention, the basic formula of the support vector machine is:

[0124]

[0125] y i (w T x i +b)≥1-ξ i

[0126] ξ i ≥0, i=1, 2, …, l

[0127] Wherein, w is the importance score of the univariate feature, T is the transposition operation, C is the first preset constant, ξ i is the slack variable of the ith univariate feature, y i is the first preset label of the ith univariate feature, x i is the ith univariate feature, b is the preset error amount, is the judgment condition of the support vector machine, that is, when When , the optimization is stopped, and l is the number of the univariate features.

[0128] See attached Figure 6 As shown, it is a schematic diagram of the external model of the support vector machine. The support vector machine calculates the contribution weight of each univariate feature vector externally, and simulates a classification hyperplane according to the univariate feature, that is, =0. According to the classification hyperplane and each contribution value weight, the unilateral feature farthest from the classification hyperplane is screened out, thereby realizing feature selection. In the figure, x1 and x2 are the univariate features, and the two univariate features constitute a two-dimensional plane. Each univariate feature can be regarded as a point in space. In the figure, + represents the positive class, - represents the negative class, and the positive class and the negative class are the preset labels. The solid line in the figure is the classification hyperplane. In the figure, γ is the margin, and the dotted line in the figure is the interval boundary. According to the interval boundary, the hyperplane can realize the classification of all the screening feature vectors. In the embodiment of the present invention, the univariate feature is obtained by using the variance analysis method and the external model of the support vector machine is used to select the univariate feature vector, remove irrelevant or redundant features, and improve the accuracy and data utilization of the screening feature vector.

[0129] S5. Using a gradient boosting decision tree learning method to learn the screened feature vectors, to obtain a seismic event recognition model.

[0130] In the embodiment of the present invention, the use of the gradient boosting decision tree learning method to learn the training data can enable the obtained seismic event recognition model to adaptively handle the problem of seismic data being masked due to weak signals, thereby improving the recognition accuracy of the seismic event recognition model.

[0131] In an embodiment of the present invention, the gradient boosting decision tree learning method is used to learn the training data to obtain a seismic event recognition model, including:

[0132] Obtaining a decision tree model, initializing parameters according to the decision tree model, and obtaining a strong learner;

[0133] Performing function construction on the strong learner and the training data to obtain a strong loss function of the strong learner;

[0134] Calculating the training data according to the strong loss function and the strong prediction value to obtain a negative gradient of the training data;

[0135] Using the negative gradient as a target label, and training a weak learner according to the target label;

[0136] Calculating a weak prediction value of the weak learner, and weightedly adding the weak prediction value to the prediction value of the strong learner to obtain a new strong learner;

[0137] Constructing a loss function of the new strong learner;

[0138] If the loss function of the new strong learner is greater than a preset threshold, returning to the step of constructing a function of the strong learner and the training data to obtain a strong loss function of the strong learner;

[0139] If the loss function of the new strong learner is less than a threshold, a recognition model is obtained;

[0140] The identification model is optimized using preset verification data to obtain a seismic event identification model.

[0141] In the embodiment of the present invention, the decision tree model is a tree structure, and is used as a basic model in the gradient boosting decision tree learning method. The model corrects each error through iteration, and finally obtains a seismic event recognition model.

[0142] In the embodiment of the present invention, the weak learner and the strong learner may be weightedly accumulated using the following formula:

[0143] F m (x) = F m-1 (x)+β m h(x;a m )

[0144] Among them, F m is the strong prediction value of the strong learner, β m is the preset learning constant, h(x; a m ) is the weak prediction value of the weak trainer, a m is the parameter of the weak learner in the mth round, x is the screening feature vector, and m is the preset number of cycles.

[0145] In the embodiment of the present invention, the weak learner parameters can be determined using the following formula:

[0146]

[0147] Among them, a m is the parameter of the weak learner in the mth round, α, β are the second preset constant and the third preset constant respectively, g m (x i ) is the i-th screening feature vector x i The prediction result on the mth weak learner, x i is the i-th selected feature vector, the h(x; a m) is the weak prediction value of the weak trainer, m is the preset number of cycles, and i is the screening feature vector.

[0148] In the embodiment of the present invention, the negative gradient value of the new strong learner can be obtained using the following formula:

[0149]

[0150] Among them, k m (x) represents the negative gradient value obtained at the mth iteration, x is the screened feature vector, F(x) is the accumulated prediction function of the strong learner, L(y, F(x)) is the loss function, z is the second preset label, and E is the mathematical expectation corresponding to the screened feature vector.

[0151] In the embodiment of the present invention, the training data is learned by using the gradient boosting decision tree learning method. Since each step in the gradient boosting decision tree learning method performs error training on previous errors, thereby reducing errors, the obtained seismic event recognition model has high accuracy and stability during recognition, can effectively capture the characteristics of seismic event, quickly and accurately discover seismic event data, and improve the recognition efficiency and recognition accuracy of seismic event data.

[0152] S6. Acquire test data, and use the seismic event recognition model to identify the test data to obtain seismic event data.

[0153] In the embodiment of the present invention, the test data is identified by using the seismic event recognition model to obtain the identified event data, which can quickly identify the event data in a large amount of data, thereby improving the recognition efficiency of event seismic data.

[0154] In the embodiment of the present invention, enhanced data is obtained by segmenting and enhancing the test data, and the seismic event recognition model is used to further identify the test data to obtain the seismic event data.

[0155] See attached Figure 7 As shown, Figure 7 (a) and Figure 7 (b) is the test data used in the present invention, wherein the horizontal axis is the distance and the vertical axis is the speed. The test data is input into the seismic event axis identification model for testing to obtain the test result diagram, see the attached Figure 8 As shown, Figure 8 (a) The precision rate for medium noise is 96%, the recall rate is 99%, and the precision rate for event axis is 92%, the recall rate is 70%; Figure 8(b) The noise precision is 96% and the recall is 99%, while the event precision is 91% and the recall is 69%. Figure 8 It can be seen from the test results that no matter it is a direct wave or a reflected wave, the effective signal of the coaxial line can be well detected.

[0156] In an actual application scenario of the present invention, in order to compare the detection effect of the seismic event recognition model, data sets are extracted respectively: single-channel shot record at an offset of 560m, single-channel shot record at an offset of 1760m, and single-channel shot record at an offset of 2860m. Fig. 9 As shown, it is a single-channel shot record comparison diagram, wherein the ordinate is amplitude and the abscissa is time. It can be seen from the figure that using the seismic event axis recognition model to identify the test data can more quickly identify the seismic event axis data.

[0157] Embodiment 2

[0158] In order to understand the present invention more clearly, a second embodiment is used below to further explain the situation in which the feature vector is subjected to feature screening to obtain the screened feature vector in the embodiment of the present invention.

[0159] In the embodiment of the present invention, the preset L1 regularization model is a linear regression model. According to the preset L1 regularization model, the weights of unimportant features can be gradually reduced to zero, thereby achieving feature screening.

[0160] In the embodiment of the present invention, the step of performing feature screening on the feature vector to obtain the screened feature vector includes:

[0161] Using a preset L1 regularization model to perform weight extraction on the feature vector to obtain weight data;

[0162] Extract the eigenvector corresponding to the non-zero weight in the weight data to obtain the regular eigenvector;

[0163] Calculating the covariance matrix of the regular eigenvector;

[0164] According to the eigenvalues ​​of the covariance matrix and the corresponding matrix vector;

[0165] Sorting the eigenvalues ​​in descending order to obtain a sorting result, and selecting matrix vectors corresponding to a preset number of eigenvalues ​​as principal components according to the sorting result;

[0166] The regular eigenvector is projected onto the principal component to obtain a screening eigenvector.

[0167] In the present invention, by using the preset L1 regularization to perform feature screening to obtain a regularized feature vector, and then calculating the principal component of the regularized feature vector, the screening accuracy of the screened screening features can be greatly improved, and the validity of the data can be improved.

[0168] Embodiment 3

[0169] like Fig.10 As shown, the present embodiment also provides a functional module diagram of a seismic event identification device based on machine learning. The seismic event identification device 100 based on machine learning described in the present embodiment can be installed in an electronic device. According to the implemented functions, the seismic event identification device 100 based on machine learning may include a data acquisition module 101, a data processing module 102, a feature processing module 103, a feature screening module 104, a model training module 105 and a model application module 106. The module described in the present invention may also be referred to as a unit, which refers to a series of computer program segments that can be executed by an electronic device processor and can complete fixed functions, which are stored in the memory of the electronic device.

[0170] In this embodiment, the functions of each module / unit are as follows:

[0171] The data acquisition module 101 is used to acquire seismic signals, and to perform forward numerical simulation on the seismic signals using a thin interlayer velocity model to obtain training data to be processed;

[0172] The data processing module 102 is used to perform single-channel segmentation and data reorganization on the training data to be processed to obtain training data;

[0173] A feature processing module 103 is used to perform feature processing on the training data to obtain a feature vector;

[0174] A feature screening module 104 is used to perform feature screening on the feature vector to obtain a screened feature vector;

[0175] A model training module 105 is used to learn the screening feature vector using a gradient boosting decision tree learning method to obtain a seismic event recognition model;

[0176] The model application module 106 is used to obtain test data, identify the test data using the seismic event identification model, and obtain seismic event data.

[0177] In detail, each module described in the seismic event identification device 100 based on machine learning described in the embodiment of the present invention adopts the same technical means as the seismic event identification method based on machine learning described in Example 1 and Example 2 when used, and can produce the same technical effects, which will not be repeated here.

[0178] Embodiment 4

[0179] like Fig.11 As shown, this embodiment also provides a computer electronic device, which may include a processor 10, a memory 11, a communication bus 12 and a communication interface 13, and may also include a computer program stored in the memory 11 and executable on the processor 10, such as a seismic event axis identification program based on machine learning.

[0180] In some embodiments, the processor 10 may be composed of an integrated circuit, for example, a single packaged integrated circuit, or a plurality of packaged integrated circuits with the same or different functions, including one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and combinations of various control chips. The processor 10 is the control core (ControlUnit) of the electronic device, and uses various interfaces and lines to connect various components of the entire electronic device, and executes or executes programs or modules stored in the memory 11 (for example, executing a seismic event axis identification program based on machine learning, etc.), and calls data stored in the memory 11 to execute various functions of the electronic device and process data.

[0181] The memory 11 includes at least one type of readable storage medium, and the readable storage medium includes a flash memory, a mobile hard disk, a multimedia card, a card-type memory (for example, an SD or DX memory, etc.), a magnetic memory, a disk, an optical disk, etc. In some embodiments, the memory 11 may be an internal storage unit of an electronic device, such as a mobile hard disk of the electronic device. In other embodiments, the memory 11 may also be an external storage device of an electronic device, such as a plug-in mobile hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device. Further, the memory 11 may also include both an internal storage unit of the electronic device and an external storage device. The memory 11 may not only be used to store application software and various types of data installed in the electronic device, such as the code of a seismic event axis identification program based on machine learning, but may also be used to temporarily store data that has been output or is to be output.

[0182] The communication bus 12 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. The bus is configured to realize connection and communication between the memory 11 and at least one processor 10, etc.

[0183] The communication interface 13 is used for communication between the above-mentioned electronic device and other devices, including a network interface and a user interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a WI-FI interface, a Bluetooth interface, etc.), which is generally used to establish a communication connection between the electronic device and other electronic devices. The user interface may be a display (Display), an input unit (such as a keyboard (Keyboard)), and optionally, the user interface may also be a standard wired interface, a wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, and an OLED (Organic Light-Emitting Diode, organic light-emitting diode) touch device, etc. Among them, the display may also be appropriately referred to as a display screen or a display unit, which is used to display information processed in the electronic device and to display a visual user interface.

[0184] The figure only shows an electronic device with components. Those skilled in the art will understand that the structure shown in the figure does not constitute a limitation on the electronic device, and may include fewer or more components than shown in the figure, or combine certain components, or arrange the components differently.

[0185] For example, although not shown, the electronic device may also include a power source (such as a battery) for supplying power to each component. Preferably, the power source may be logically connected to the at least one processor 10 through a power management device, so that the power management device can realize functions such as charging management, discharging management, and power consumption management. The power source may also include one or more DC or AC power sources, recharging devices, power failure detection circuits, power converters or inverters, power status indicators, and other arbitrary components. The electronic device may also include a variety of sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be repeated here.

[0186] It should be understood that the embodiment is for illustration only and the scope of the patent application is not limited to this structure.

[0187] The seismic event recognition program based on machine learning stored in the memory 11 in the electronic device is a combination of multiple instructions, and when running in the processor 10, it can achieve:

[0188] Acquire seismic signals, and use a thin interlayer velocity model to perform forward numerical simulation on the seismic signals to obtain training data to be processed;

[0189] Performing single-channel segmentation and data reorganization on the training data to be processed to obtain training data;

[0190] Performing feature processing on the training data to obtain a feature vector;

[0191] Performing feature screening on the feature vector to obtain a screened feature vector;

[0192] Using a gradient boosting decision tree learning method to learn the screening feature vector, to obtain a seismic event recognition model;

[0193] Acquire test data, and use the seismic event recognition model to identify the test data to obtain seismic event data.

[0194] Specifically, the specific implementation method of the processor 10 for the above instructions can refer to the description of the relevant steps in the corresponding embodiment of the accompanying drawings, which will not be repeated here.

[0195] Furthermore, if the module / unit integrated in the electronic device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. The computer-readable storage medium can be volatile or non-volatile. For example, the computer-readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a mobile hard disk, a magnetic disk, an optical disk, a computer memory, and a read-only memory (ROM).

[0196] Embodiment 5

[0197] This embodiment provides a storage medium storing a computer program. When the computer program is executed by a processor, the steps of the seismic event identification method based on machine learning as described above are implemented.

[0198] These program codes can also be loaded onto a computer or other programmable data processing device so that a series of operation steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions executed on the computer or other programmable device for implementing the process. Figure 1 The steps of a specified function in a process or multiple processes.

[0199] Storage media include permanent and non-permanent, removable and non-removable media, and can be implemented by any method or technology to store information. Information can be computer-readable instructions, data structures, modules of programs or other data. Examples of storage media can include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission medium that can be used to store information that can be accessed by a computing device.

[0200] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. When the terms "include" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or their combinations.

[0201] It is to be understood that the terms so used are interchangeable under appropriate circumstances such that the embodiments of the application described herein are, for example, capable of operation in sequences other than those illustrated or described herein.

[0202] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of the modules is only a logical function division, and there may be other division methods in actual implementation.

[0203] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0204] In addition, each functional module in each embodiment of the present invention may be integrated into one processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0205] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0206] Therefore, no matter from which point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the appended claims rather than the above description, so it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any attached figure mark in the claims should not be regarded as limiting the claims involved.

[0207] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0208] In addition, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in the system claim can also be implemented by one unit or device through software or hardware. The words first, second, etc. are used to indicate names, and do not indicate any particular order.

[0209] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A method for identifying seismic event axes based on machine learning, characterized in that: The method comprises: Acquire seismic signals, and use a thin interlayer velocity model to perform forward numerical simulation on the seismic signals to obtain training data to be processed; Performing single-channel segmentation and data reorganization on the training data to be processed to obtain training data; Performing feature processing on the training data to obtain a feature vector; Performing feature screening on the feature vector to obtain a screened feature vector; Using a gradient boosting decision tree learning method to learn the screening feature vector, a seismic event recognition model is obtained; Acquire test data, and use the seismic event recognition model to identify the test data to obtain seismic event data.

2. The seismic event identification method based on machine learning according to claim 1, characterized in that: The method of using the thin interlayer velocity model to perform forward numerical simulation on the seismic signal to obtain training data to be processed includes: Calculating the seismic signal according to the finite difference method and the thin interlayer velocity model to obtain the wave field condition; Extracting position information of the wave field to obtain forward synthetic shot record data; The forward synthetic shot record data is divided into preset values ​​to obtain training data.

3. The seismic event identification method based on machine learning according to claim 1, characterized in that: The step of performing single-channel segmentation and data reorganization on the training data to be processed to obtain training data includes: Segmenting the training data to obtain segmented data; Performing equal division processing on the segmented data to obtain reorganized data; Noise is added to the reorganized data to obtain training data.

4. The seismic event identification method based on machine learning according to claim 1, characterized in that: The performing feature processing on the training data to obtain a feature vector includes: Performing single-channel feature extraction on the training data to obtain singular feature data; Performing gray-level co-occurrence matrix feature extraction on the training data to obtain gray-level co-occurrence matrix feature data; Extracting inter-channel features from the training data to obtain inter-channel feature data; Constructing feature vectors for the gray-level co-occurrence matrix feature data, the single-channel feature extraction, and the inter-channel feature data to obtain feature data; The feature data is standardized to obtain a feature vector.

5. The seismic event identification method based on machine learning according to claim 1, characterized in that: The step of performing feature screening on the feature vector to obtain a screened feature vector includes: Performing univariate feature analysis on the feature vector using variance analysis to obtain a univariate feature vector; The single variable feature vector is subjected to feature selection by using an external model of a support vector machine to obtain a screening feature vector.

6. The seismic event identification method based on machine learning according to claim 5, characterized in that: The method of performing univariate feature analysis on the feature vector by using the variance analysis method to obtain the univariate feature vector includes: Calculating the sum of squares of deviations between the eigenvectors to obtain vector deviations; Calculating the square sum of the deviations of the feature vector to obtain a training deviation; Calculating test statistics of the feature vectors one by one according to the training deviation and the vector deviation; The test statistics are sorted in descending order, and features corresponding to the test statistics that meet preset conditions are screened out as univariate feature vectors.

7. The seismic event identification method based on machine learning according to claim 5, characterized in that: The step of performing feature selection on the univariate feature vector using the external model of the support vector machine to obtain the screening feature vector includes: Calculate the importance score of each of the univariate features using a preset support vector external model; Univariate features are screened out according to the importance scores to obtain the screened feature vector.

8. A seismic event identification device based on machine learning, characterized in that: The device comprises: A data acquisition module is used to acquire seismic signals, and to perform forward numerical simulation on the seismic signals using a thin interlayer velocity model to obtain training data to be processed; A data processing module, used for performing single-channel segmentation and data reorganization on the training data to be processed to obtain training data; A feature processing module, used to perform feature processing on the training data to obtain a feature vector; A feature screening module, used for screening the feature vector to obtain a screened feature vector; A model training module is used to learn the screening feature vector using a gradient boosting decision tree learning method to obtain a seismic event recognition model; The model application module is used to obtain test data, identify the test data using the seismic event recognition model, and obtain seismic event data.

9. An electronic device, characterized in that: include: processor; a memory for storing instructions executable by the processor; Wherein, the processor is configured to execute the instructions to implement the steps of the method for seismic event axis identification based on machine learning as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the program is executed by a processor, the steps of the method for seismic event axis identification based on machine learning as described in any one of claims 1 to 7 are implemented.

Images