A LightGBM-based method for identifying anomalies in multi-source microseismic signals

Abstract

This application proposes a method for identifying abnormalities in multi-source microseismic signals based on LightGBM, including: performing noise elimination operations on multi-source microseismic signals; obtaining multiple average values and multiple trend values of candidate target areas; determining a microseismic state abnormality detection model based on LightGBM based on the multiple average values and multiple trend values, and identifying abnormal states of underground rock formations based on the microseismic state abnormality detection model. By eliminating redundancy and noise in multi-source microseismic signals, a first intermediate signal for target detection is obtained; based on target detection of the first intermediate signal, multiple average values and multiple trend values of the candidate target area are obtained; based on the LightGBM model, the average values and trend values are trained to obtain a microseismic state abnormality detection model; and based on the microseismic state abnormality detection model, abnormal states of underground rock formations are identified, thereby greatly reducing the difficulty of capturing the spatiotemporal evolution information of microseisms and significantly improving the robustness and detection accuracy of microseismic anomaly detection.

Description

Technical Field

[0001] The present application relates to the technical fields of coal mine production safety and geological disaster early warning, and in particular to a method for identifying anomalies of multi-source microseismic signals based on LightGBM. Background Art

[0002] Microseismic signals are seismic waves emitted by rock failure or fracture. These signals are collected and acquired by sensors and analyzed and processed to obtain information such as the earthquake's location, magnitude, energy, and moment. This information is then used in conjunction with seismological principles to analyze the stress and strain state, thereby determining the stability of the rock formation. However, due to the diversity of coal mine geological structures, coal seam occurrence, rock formation structure, and mining activities, monitoring using microseismic signals inevitably contains a large amount of noise, which interferes with microseismic timing analysis and anomaly identification. This makes it difficult to capture the spatiotemporal evolution of microseismic events, affecting the accuracy of microseismic anomaly identification.

[0003] It should be noted that the above introduction to the technical background is merely intended to provide a clear and complete description of the technical solutions of this application and facilitate understanding by those skilled in the art. Simply because these solutions are described in the background technology section of this application, it should not be assumed that the above technical solutions are well known to those skilled in the art. Summary of the Invention

[0004] The purpose of this application is to solve one of the above technical problems at least to a certain extent.

[0005] To this end, the first embodiment of the present application proposes a multi-source microseismic signal anomaly identification method based on LightGBM, including:

[0006] Perform noise elimination on multi-source microseismic signals to obtain the first intermediate signal of the underground rock formation

[0007] performing target detection on the first intermediate signal to obtain multiple average values and multiple trend values of candidate target areas;

[0008] According to the multiple average values and the multiple trend values, a microseismic state anomaly detection model based on LightGBM is determined, and the abnormal state of the underground rock formation is identified according to the microseismic state anomaly detection model.

[0009] The second embodiment of the present application proposes a multi-source microseismic signal anomaly identification device based on LightGBM, comprising:

[0010] a first acquisition module, configured to perform a noise elimination operation on the multi-source microseismic signals to obtain a first intermediate signal of the underground rock formation;

[0011] a second acquisition module, configured to perform target detection on the first intermediate signal and acquire multiple average values and multiple trend values of a candidate target area;

[0012] The identification module is used to determine a microseismic state anomaly detection model based on LightGBM according to the multiple average values and the multiple trend values, and to identify the abnormal state of the underground rock formation according to the microseismic state anomaly detection model.

[0013] The third aspect embodiment of the present application proposes an electronic device, comprising: a processor; a memory for storing instructions executable by the processor; wherein the processor is configured to execute the instructions to implement the multi-source microseismic signal anomaly identification method based on LightGBM proposed in the first aspect embodiment of the present application.

[0014] The fourth aspect embodiment of the present application proposes a non-temporary computer-readable storage medium, which, when the instructions in the storage medium are executed by the processor of an electronic device, enables the electronic device to execute the method proposed in the first aspect embodiment of the present application.

[0015] The fifth aspect embodiment of the present application proposes a computer program product, including a computer program, which implements the method proposed in the first aspect embodiment of the present application when executed by a processor in a communication device.

[0016] The technical solutions provided by the embodiments of this application bring at least the following beneficial effects:

[0017] By eliminating redundancy and noise in multi-source microseismic signals, a first intermediate signal for target detection is obtained; based on target detection of the first intermediate signal, multiple average values and multiple trend values of the candidate target area are obtained; based on the LightGBM model, the average values and trend values are trained to obtain a microseismic state anomaly detection model; and based on the microseismic state anomaly detection model, the abnormal state of the underground rock formation is identified, which greatly reduces the difficulty of capturing the spatiotemporal evolution information of microseisms and significantly improves the robustness and detection accuracy of microseismic anomaly detection.

[0018] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0020] Figure 1 A flowchart of a method for identifying anomalies in multi-source microseismic signals based on LightGBM provided in an embodiment of the present application;

[0021] Figure 2 A schematic diagram of the structure of a multi-source microseismic signal anomaly identification device based on LightGBM provided in an embodiment of the present application;

[0022] Figure 3 A schematic structural diagram of an electronic device provided according to an embodiment of the present application;

[0023] Figure 4 The figure is a schematic structural diagram of another electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] Exemplary embodiments are described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numbers in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all possible implementations consistent with the present invention. Rather, they are merely examples of apparatuses and methods consistent with certain aspects of the present invention, as detailed in the appended claims.

[0025] The terms used in the embodiments of this application are for the purpose of describing specific embodiments only and are not intended to limit the embodiments of this application. The singular forms "a" and "the" used in the embodiments of this application and the appended claims are also intended to include plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0026] It should be understood that although the terms first, second, third, etc. may be used to describe various information in the embodiments of the present application, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from each other. For example, without departing from the scope of the embodiments of the present application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Depending on the context, the words "if" and "if" as used herein may be interpreted as "at the time of" or "when" or "in response to a determination."

[0027] The embodiments of the present application are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be understood as limiting the present application.

[0028] It should be noted that the LightGBM-based multi-source microseismic signal anomaly identification method provided in any embodiment of the present application can be executed alone, or in combination with possible implementation methods in other embodiments, or in combination with any technical solution in the relevant technology.

[0029] The following describes a method and device for identifying anomalies of multi-source microseismic signals based on LightGBM in an embodiment of the present application with reference to the accompanying drawings.

[0030] Figure 1 This is a flow chart of a method for identifying abnormalities in multi-source microseismic signals based on LightGBM provided in an embodiment of the present application. Figure 1 As shown, the method includes but is not limited to the following steps:

[0031] Step S101 : performing a noise elimination operation on multi-source microseismic signals to obtain a first intermediate signal of an underground rock formation.

[0032] In one feasible implementation, principal component analysis (PCA) can be used to remove noise from multi-source microseismic signals. PCA, short for Principal Component Analysis, is used to reduce the dimensionality of data while preserving the key features of a dataset. Specifically, PCA transforms multi-source microseismic signals into a new set of linearly uncorrelated variables, known as principal components, through an orthogonal transformation. The principal components are then arranged in descending order of variance.

[0033] Furthermore, the multi-source microseismic signals are aligned according to the timestamps to obtain a first matrix related to the number of sample points and the number of data channels of the multi-source microseismic signals. The multi-source microseismic signals are integrated into a first matrix, which can be represented by X∈R M×N , where M is the number of data samples of the microseismic signal; N is the number of components or channels for collecting the microseismic signal (in this embodiment, the components are microseismic sensors). It should be noted that aligning multi-source microseismic signals by timestamps belongs to the category of preprocessing, which is used to align microseismic signals from different sources by timestamps, including:

[0034] Multi-source microseismic signal acquisition is used to collect microseismic signals from various sources and ensure that each microseismic signal provides time stamp information;

[0035] Microseismic signals from different sources may have different data formats and storage methods. Therefore, each microseismic signal should be formatted in a unified manner for subsequent processing.

[0036] To ensure that all timestamps use the same time zone, you can convert the timestamps to Coordinated Universal Time or other unified time standard;

[0037] Next, check the precision of the timestamp. If the precision is inconsistent, the lower precision timestamp can be converted to a higher precision timestamp.

[0038] Sort all microseismic signals by timestamp to ensure that in the subsequent alignment process, microseismic events are processed in the order in which they occurred;

[0039] Define a time window for aligning microseismic events from different sources to ensure that nearly simultaneous microseismic events are captured;

[0040] Traverse the sorted microseismic event list and, for each event, check whether there are microseismic events from other sources that occur within the event window. If matching events are found, mark them as aligned events. For unaligned events, you can choose to ignore them.

[0041] Furthermore, the first matrix is centrally processed to obtain a second matrix, wherein the second matrix can be represented as X i is the data sample of the i-th microseismic signal, where 1≤i≤M, and i is a positive integer. It should be noted that data centering is the process of adjusting data to eliminate or reduce certain statistical biases. Specifically, in multivariate statistical analysis, there may be correlation between variables, namely multicollinearity, and data centering can reduce this correlation, thereby improving the stability and accuracy of statistical analysis. If there are interaction terms between variables (i.e., the product of two or more variables), data centering can ensure that the coefficients of the data interaction terms have a more intuitive interpretation. Centered data are more reliable in the assumptions of certain statistical models (such as principal component analysis and factor analysis), thereby improving the accuracy of the model.

[0042] Furthermore, the covariance matrix of the second matrix is obtained, and the covariance matrix is subjected to eigendecomposition to obtain multiple eigenvectors, wherein the covariance matrix can be represented as for The transposed matrix of . Perform eigendecomposition on the covariance matrix C, including: C = VΛV -1 ,in, is the eigenvector matrix of the covariance matrix C, is the 1st, 2nd, ···, Nth eigenvector of the covariance matrix C; V -1 is the inverse matrix of V; Λ is a diagonal matrix, the elements on the diagonal are eigenvalues, and the elements in the other positions are 0.

[0043] Furthermore, the principal component feature space is determined based on the multiple eigenvectors. Furthermore, candidate eigenvectors are obtained from the multiple eigenvectors; a projection matrix related to the candidate eigenvectors is obtained; and the principal component feature space is determined based on the projection matrix. It should be noted that, based on the size of the eigenvalues, the first k eigenvectors are selected from the N eigenvectors as principal components, where 1<k<N. The principal component determination process is usually based on the ratio of the eigenvalue to the sum of the total eigenvalues or the required number of principal components. The first k eigenvectors are used as the principal component feature space, and the principal component feature space constitutes a projection matrix, which can be represented as

[0044] Furthermore, each element in the second matrix is mapped to the principal component feature space to obtain the first intermediate signal, wherein the first intermediate signal can be represented as Xpca, and the second matrix Mapping to the PCA feature space to remove random noise, irrelevant information, and low-energy noise in the microseismic signal. The specific mapping process is as follows:

[0045]

[0046] in, For X pca The 1st, 2nd, …, i, …, Mth samples of .

[0047] It should be added that noise removal operations on multi-source microseismic signals include but are not limited to principal component analysis. Autocorrelation functions can also be used to analyze the characteristics of signals and noise to eliminate noise. Wavelet decomposition can also be used to decompose microseismic signals into components of different frequencies, and then identify and eliminate the frequency components representing noise to reconstruct the microseismic signals. Adaptive filters can also be used to design adaptive filters based on the statistical characteristics of noise to adapt to the noise characteristics of different signal sources, and recursively adjust the parameters of the filter to minimize noise. Therefore, different microseismic signals and noise types may require different denoising methods. When selecting a denoising method, it is necessary to evaluate and select it according to the specific situation. After applying the denoising method, the processed first intermediate signal needs to be verified and evaluated to ensure that the denoising effect achieves the expected goal.

[0048] Step S102: performing target detection on the first intermediate signal to obtain multiple average values and multiple trend values of the candidate target area.

[0049] In a feasible implementation, due to the influence of the complex working conditions and production conditions of the mine, the data anomaly patterns of multi-source microseismic signals are relatively complex (data anomaly patterns refer to data features or patterns that appear in the data set that are significantly different from the expected pattern or behavior, including: point anomalies, context anomalies, collective anomalies, distribution anomalies, etc.), and the first intermediate signal obtained by the noise elimination operation of the multi-source microseismic signal usually also covers time windows of different lengths. Therefore, the fixed window-based time series data stream feature extraction and anomaly detection method is difficult to effectively detect data anomaly patterns of different window scales.

[0050] Target detection automatically locates and identifies the location and category of target objects in images or videos. Methods such as sliding windows can be used to generate a series of candidate regions that may contain the target object. These candidate regions are also called candidate target regions. Target detection of the first intermediate signal uses specific algorithms and models to identify signal features related to the target event, such as waveform, frequency, and amplitude, thereby enabling detection of the target event. Target detection of the first intermediate signal enables real-time monitoring of safety hazards such as rockbursts and mine tremors during mining, ensuring safe production in mines.

[0051] Optionally, as an example, multiple first sliding window lengths are selected, and corresponding first sliding windows are determined based on the multiple first sliding window lengths; the first sliding window is slid on the first intermediate signal according to a first preset step size to obtain multiple average values of the candidate target area. The sliding window is used to perform an averaging operation on a data stream or array, and the operation only focuses on the elements within the current window (or subarray). In the case of averaging, the sliding window can continuously traverse the data and calculate the average value of the elements within the current window at each step.

[0052] Furthermore, according to a first preset step size, a plurality of first sliding windows are respectively slid on the first intermediate signal to obtain a second intermediate signal identical to each of the sliding windows;

[0053] Obtaining a third intermediate signal by applying different weights to the first intermediate signal based on the coordinates of the second intermediate signal;

[0054] Based on the weighted average of the third intermediate signal, multiple average values of the candidate object regions are obtained.

[0055] As an example, define L1 as the first sliding window, extract X with a first preset step size (the first preset step size can be 1). pca The average value at different scales is represented by X L1 .

[0056] If L1=1, then X L1 It can be expressed as:

[0057]

[0058] is the 1st, 2nd,…,i,…,Mth sample of the data set with a window length of 1, Where M is the number of data samples of microseismic signals.

[0059] If L1=5, then X L1 It can be expressed as:

[0060]

[0061] For the 1st, 2nd, ..., i, ..., M-5th samples of the data set with a window length of 5,

[0062]

[0063] If L1=W, then X L1 It can be expressed as:

[0064]

[0065] is the 1st, 2nd, ..., i, ..., MWth samples of the data set with a window length of W,

[0066] In one feasible implementation, extracting local trend values from microseismic signals typically involves performing trend term analysis on the signal to understand the primary trend of change within its local range. If the data exhibits an upward trend, the values of the subsequent data are significantly larger than those of the preceding data. Conversely, if the data exhibits a downward trend, the values of the subsequent data are significantly smaller than those of the preceding data. The overall trend of the data is determined by the positive or negative difference between the two time periods. In microseismic signal processing, a trend term typically refers to a slowly changing or long-lasting portion of the signal, which may be related to the background of seismic activity or the geological structure.

[0067] Optionally, as an example, multiple second sliding window lengths are selected, and corresponding second sliding windows are determined based on the multiple second sliding window lengths; a trend test is performed on the first intermediate signal according to a second preset step size to obtain multiple trend values of the candidate target area.

[0068] Further, according to the second preset step size, the first intermediate signal is sampled to obtain a plurality of samples; based on the number of samples, the plurality of samples are grouped and paired to obtain a plurality of paired groups, wherein if the number of samples is an even number, the plurality of samples are grouped into If the number of samples is odd, divide the samples into pairing groups; obtain the difference sign of each pairing group, and based on the difference sign, count the first number of increasing pairing groups and the second number of decreasing pairing groups; obtain multiple trend values of the candidate target area based on the comparison operation of the first number and the second number, wherein, if the first number is greater than the second number, the first intermediate signal has an increasing trend, and the trend value of the candidate target area is the first value; if the first number is less than the second number, the first intermediate signal has a decreasing trend, and the trend value of the candidate target area is the second value; if the first number is equal to the second number, the first intermediate signal has no obvious trend, and the trend value of the candidate target area is the third value.

[0069] As an example, define L2 as the second sliding window length, extract X with a second preset step size (the second preset step size can be 1). pca Trend data sets at different scales, denoted as T L2 .

[0070] If L2=1, then T L2 It can be expressed as:

[0071]

[0072] is the 1st, 2nd,…,i,…,M-1th sample of the data set with a window length of 1, When L2=1, the data change trend of the i-th window is increasing; When L2=1, the data change trend of the i-th window is decreasing.

[0073] If L2=5, then T L2 It can be expressed as:

[0074]

[0075] are the 1st, 2nd, …, i, …, M-5th samples of the data set when the window length is 5. The data of the i-th window is In, take and Form a pair and take and As a pair. Since L2=5, take calculate and and and The changing trend of is denoted as z1 and z2, where:

[0076]

[0077] If z1>0, z2>0 or z1<0, z2<0, If z1>0, z2<0 or z1<0, z2>0, Then, under L2=5, the data change trend of the i-th window is increasing; When L2=5, the data change trend of the i-th window is decreasing; Then, when L2=5, the data in the i-th window has no obvious changing trend.

[0078] If L2=W, then T L2 It can be expressed as:

[0079]

[0080] are the 1st, 2nd,…,i,…,MWth samples of the data set when L2=W.

[0081] The data in the window Pair, take and Form a pair and take and For a pair; ...; take and is a pair. When L2 is an even number, There are n=c pairs in total; when L is an odd number, There are n=c-1 pairs. Then calculate the trend of each pair of samples and record it as

[0082] remember It means that under L2=W, the number of paired groups in the i-th window has an increasing trend. Indicates the number of paired groups with decreasing trend in the i-th window under L2=W. pos >N neg , If N pos <N neg , If N pos =N neg , Then L2=W, the data change trend of the i-th window is increasing; When L2=W, the change trend of the data in the i-th window is decreasing; Then, under L2=W, the data in the i-th window has no obvious trend of change, where:

[0083]

[0084] It should be added that obtaining multiple trend values of the candidate target area includes but is not limited to sliding window operations, and wavelet analysis can also be used to obtain trend values, including: fitting the first intermediate signal using the least squares fitting method, and approximating the trend value of the first intermediate signal by selecting an appropriate polynomial order. Wavelet analysis can also be used to obtain trend values, including: decomposing the first intermediate signal through wavelet basis functions to obtain limited-sign components at different scales and frequencies, and separating the trend value of the first intermediate signal by selecting appropriate wavelet basis functions and decomposition layers. Therefore, in practical applications, it may be necessary to consider the characteristics of the signal, the noise level, and the purpose of the analysis to select an appropriate method to obtain the trend value.

[0085] Step S103 : determining a microseismic state anomaly detection model based on LightGBM according to the multiple average values and the multiple trend values, and performing abnormal state identification on the underground rock formation according to the microseismic state anomaly detection model.

[0086] In a feasible implementation, an average value matrix composed of a plurality of the average values and a trend value matrix composed of a plurality of the trend values are obtained; based on the LightGBM model, the average value matrix is trained to obtain a first abnormal parameter of the average value matrix; based on the LightGBM model, the trend value matrix is trained to obtain a second abnormal parameter of the trend value matrix; the first abnormal parameter and the second abnormal parameter are superimposed to obtain a microseismic state abnormality detection model.

[0087] As an example, if principal component analysis is used to obtain the first intermediate signal, taking the first sliding window L1=W as an example, the R obtained in step S102 can be (M-W)×k Each column (j=1, 2, ..., k) of the mean value matrix is regarded as the first label in turn, and the remaining (k-1) columns are used as features to train k LightGBM models (denoted as ), obtain the first predicted value of the model, if the difference between the first predicted value and the first label is greater than a preset first threshold, the first label is regarded as an outlier, and the difference between the predicted value and the label is represented as the outlier degree ROj of the column where the label is located, where j=1,2,…,k;DOj is the difference quantization parameter between the first predicted value and the first label of the jth column;ThreshO is the first threshold value set in advance. W (Average abnormality RO W That is, the first abnormal parameter) is expressed as:

[0088] Taking the second sliding window L2=W as an example, the R obtained in step S102 can be (M-W)×kEach column (j=1, 2, ..., k) of the trend value matrix is regarded as the second label in turn, and the remaining (k-1) columns are used as features to train k LightGBM models (denoted as ), obtain the second predicted value of the model, if the difference between the second predicted value and the second label is greater than a preset second threshold, the second label is regarded as an outlier, and the difference between the predicted value and the label is represented as the outlier degree RTj of the column where the label is located, where DTj is the difference quantization parameter between the second predicted value and the second label of the jth column; ThreshT is the pre-set second threshold. Furthermore, the average abnormality RT of the second sliding window L2 under the condition W is W (Average abnormality RT W That is, the second abnormal parameter) is expressed as:

[0089] Based on the first abnormal parameter and the second abnormal parameter, a microseismic state anomaly detection model is constructed:

[0090] Under the conditions of the first sliding window L1 = 1, 2, ..., W and the second sliding window L2 = 1, 2, ..., W, the abnormality R of the LightGBM model on the mean and trend values is calculated, where:

[0091]

[0092] If R ≥ the abnormality threshold, the data flow of the multi-source microseismic signal is abnormal; if R < the abnormality threshold, the data flow of the multi-source microseismic signal is normal.

[0093] In summary, by eliminating redundancy and noise in multi-source microseismic signals, the first intermediate signal for target detection is obtained; based on target detection of the first intermediate signal, multiple average values and multiple trend values of the candidate target area are obtained; based on the LightGBM model, the average values and trend values are trained to obtain a microseismic state anomaly detection model; and based on the microseismic state anomaly detection model, the abnormal state of the underground rock formation is identified, which greatly reduces the difficulty of capturing the spatiotemporal evolution information of microseisms and significantly improves the robustness and detection accuracy of microseismic anomaly detection.

[0094] Figure 2 This is a schematic diagram of the structure of a multi-source microseismic signal anomaly identification device based on LightGBM provided in an embodiment of the present application. Figure 2 As shown, the LightGBM-based multi-source microseismic signal anomaly identification device 200 includes but is not limited to the following modules:

[0095] A first acquisition module 201 is configured to perform a noise elimination operation on the multi-source microseismic signals to obtain a first intermediate signal of the underground rock formation;

[0096] A second acquisition module 202 is configured to perform target detection on the first intermediate signal and obtain multiple average values and multiple trend values of the candidate target area;

[0097] The identification module 203 is used to determine a microseismic state anomaly detection model based on LightGBM according to multiple average values and multiple trend values, and to identify the abnormal state of the underground rock formation according to the microseismic state anomaly detection model.

[0098] Figure 3 The figure is a schematic diagram of the structure of an electronic device provided according to an embodiment of the present application. Figure 3 The electronic device shown is merely an example and should not limit the functions and scope of use of the embodiments of the present application.

[0099] like Figure 3 As shown, the electronic device 300 includes a processor 301, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 302 or the program loaded from the memory 306 to the random access memory (RAM) 303. Various programs and data required for the operation of the electronic device 300 are also stored in the RAM 303. The processor 301, ROM 302 and RAM 303 are connected to each other via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0100] The following components are connected to the I / O interface 305: a memory 306 including a hard disk, etc.; and a communication part 307 including a network interface card such as a LAN (Local Area Network) card, a modem, etc., which performs communication processing via a network such as the Internet; a drive 308 is also connected to the I / O interface 305 as needed.

[0101] In particular, according to an embodiment of the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program carried on a computer-readable medium, and the computer program includes program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network via the communication section 307. When the computer program is executed by the processor 301, the above-mentioned functions defined in the method of the present application are performed.

[0102] In an exemplary embodiment, a storage medium including instructions is further provided, such as a memory including instructions, and the instructions can be executed by the processor 301 of the electronic device 300 to perform the above method. Alternatively, the storage medium can be a non-transitory computer-readable storage medium, for example, a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc.

[0103] In this application, a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. Furthermore, in this application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transfer a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium may be transmitted using any suitable medium, including but not limited to wireless, wire, optical cable, RF, etc., or any suitable combination thereof.

[0104] Figure 4 The figure is a schematic structural diagram of another electronic device provided according to an embodiment of the present application. Figure 4 The electronic device shown is only an example and should not limit the functions and scope of use of the embodiments of the present application. Figure 4 As shown, the electronic device 400 includes a processor 401 and a memory 402. The memory 402 is used to store program codes, and the processor 401 is connected to the memory 402 to read the program codes from the memory 402 to implement the multi-source microseismic signal anomaly identification method based on LightGBM in the above embodiment.

[0105] Optionally, the number of processors 401 may be one or more.

[0106] Optionally, the electronic device may further include an interface 403, and the number of the interface 403 may be multiple. The interface 403 may be connected to an application program and may receive data from an external device such as a sensor.

[0107] Those skilled in the art will readily appreciate other embodiments of the present invention after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, and the true scope and spirit of the present application are indicated by the following claims.

[0108] It should be understood that the present application is not limited to the exact structures described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. A LightGBM-based method for identifying abnormalities in multi-source microseismic signals, characterized by: include: Performing noise elimination on multi-source microseismic signals to obtain a first intermediate signal of the underground rock formation; performing target detection on the first intermediate signal to obtain multiple average values and multiple trend values of candidate target areas; Obtaining an average value matrix composed of a plurality of the average values and a trend value matrix composed of a plurality of the trend values; Based on the LightGBM model, the average value matrix is trained to obtain the first abnormal parameter of the average value matrix; Based on the LightGBM model, the trend value matrix is trained to obtain a second abnormal parameter of the trend value matrix; The first abnormal parameter and the second abnormal parameter are superimposed to obtain a microseismic state abnormality detection model, and the abnormal state of the underground rock formation is identified based on the microseismic state abnormality detection model.

2. The method according to claim 1, characterized in that The performing of a noise elimination operation on the multi-source microseismic signals to obtain a first intermediate signal of the underground rock formation includes: Based on principal component analysis, dimension reduction and feature extraction are performed on the multi-source microseismic signals to obtain the first intermediate signal.

3. The method according to claim 2, characterized in that The step of performing dimensionality reduction and feature extraction on the multi-source microseismic signals based on principal component analysis to obtain the first intermediate signal includes: Performing an alignment operation on the multi-source microseismic signals according to the timestamps to obtain a first matrix related to the number of sample points and the number of data channels of the multi-source microseismic signals; Performing data centering processing on the first matrix to obtain a second matrix; Obtaining a covariance matrix of the second matrix, and performing eigendecomposition on the covariance matrix to obtain a plurality of eigenvectors; Determining a principal component feature space based on the plurality of feature vectors; Map each element in the second matrix to the principal component feature space to obtain the first intermediate signal.

4. The method according to claim 3, characterized in that Determining the principal component feature space according to the plurality of feature vectors includes: Obtaining a candidate feature vector from the plurality of feature vectors; Obtaining a projection matrix associated with the candidate eigenvector; Based on the projection matrix, a principal component feature space is determined.

5. The method according to claim 1, wherein The performing target detection on the first intermediate signal to obtain multiple average values of candidate target areas includes: selecting a plurality of first sliding window lengths, and determining corresponding first sliding windows according to the plurality of first sliding window lengths; According to a first preset step size, the first sliding window is slid on the first intermediate signal to obtain multiple average values of the candidate target area.

6. The method according to claim 5, characterized in that The step of sliding the first sliding window on the first intermediate signal according to the first preset step size to obtain multiple average values of the candidate target area includes: performing a sliding operation on each of the first sliding windows on the first intermediate signal according to a first preset step size to obtain a second intermediate signal identical to each of the sliding windows; assigning different weights to the first intermediate signal based on the coordinates of the second intermediate signal to obtain a third intermediate signal; Based on the weighted average of the third intermediate signal, multiple average values of the candidate object area are obtained.

7. The method according to claim 1, characterized in that The performing target detection on the first intermediate signal to obtain multiple trend values of the candidate target area includes: selecting a plurality of second sliding window lengths, and determining corresponding second sliding windows according to the plurality of second sliding window lengths; According to a second preset step size, a trend test is performed on the first intermediate signal to obtain multiple trend values of the candidate target area.

8. The method according to claim 7, characterized in that The step of performing a trend test on the first intermediate signal according to the second preset step size to obtain multiple trend values of the candidate target area includes: Sampling the first intermediate signal according to a second preset step size to obtain a plurality of samples; Based on the number of samples, grouping and pairing the plurality of samples to obtain a plurality of pairing groups; Obtaining a difference sign of each pairing group, and statistically increasing a first number of pairing groups and decreasing a second number of pairing groups based on the difference sign; Based on the comparison operation of the first number and the second number, multiple trend values of the candidate target area are obtained, wherein, if the first number is greater than the second number, the first intermediate signal has an increasing trend, and the trend value of the candidate target area is a first value; if the first number is less than the second number, the first intermediate signal has a decreasing trend, and the trend value of the candidate target area is a second value; if the first number = the second number, the first intermediate signal has no obvious trend, and the trend value of the candidate target area is a third value.

9. The method according to claim 8, characterized in that The step of grouping and pairing the plurality of samples based on the number of samples to obtain a plurality of pairing groups includes: If the number of samples is even, divide the samples into paired groups; If the number of samples is odd, divide the samples into A pairing group.

Images