A method, device, medium and product for predicting the source mechanism of coal mine earthquake events

Abstract

The present application discloses a method, device, medium and product for predicting the focal mechanism of a coal mine earthquake event, and relates to the field of focal mechanism prediction. The method comprises obtaining waveform data of a preset number of sampling points of a target coal mine earthquake event; preprocessing the waveform data to obtain processed waveform data; using a focal mechanism prediction model to determine the Gaussian distribution prediction curve of the focal mechanism of the target coal mine earthquake event according to the processed waveform data; wherein the focal mechanism prediction model is obtained by training the initial focal mechanism prediction model using a training data set; the initial focal mechanism prediction model comprises a graph convolutional neural network and a fully convolutional neural network connected in sequence; according to the Gaussian distribution prediction curve of the focal mechanism, the focal mechanism of the target coal mine earthquake event is determined; the focal mechanism is the angle corresponding to the highest probability value in each Gaussian distribution curve. The present application improves the accuracy and efficiency of focal mechanism prediction.

Description

Technical Field

[0001] The present application relates to the field of earthquake source mechanism prediction, and in particular to a method, equipment, medium and product for predicting the earthquake source mechanism of a coal mine earthquake event. Background Art

[0002] As the depth of coal mining increases, mine earthquake disasters have become a major problem threatening mine safety production. Due to the large ground stress and complex geological structure in deep coal mines, the frequency and intensity of mine earthquakes have increased significantly. This vibration not only damages the structure and equipment of the mine, but also threatens the lives of miners. Therefore, early warning and prevention of mine earthquake disasters have become a key link in mine safety production.

[0003] The occurrence of mine earthquakes is closely related to the focal mechanism, which describes the fault sliding mode and energy release direction of earthquakes, and reveals the cause and nature of mine earthquakes. By studying the focal mechanism, we can gain a deeper understanding of the occurrence process of mine earthquakes and their impact on mines, thus providing a basis for formulating scientific earthquake prevention and disaster reduction measures. However, due to the complexity of the coal mining environment, it has always been a challenge to accurately determine the focal mechanism of mine earthquakes.

[0004] The rapid development of deep learning technology has provided new possibilities for the automatic identification and analysis of focal mechanisms. By constructing neural network models, especially graph convolutional neural networks, the characteristics of focal mechanisms can be effectively extracted from mine earthquake waveform data. Compared with traditional focal mechanism analysis methods, deep learning models have stronger nonlinear expression capabilities and data-driven characteristics, and can more accurately capture the complex information in mine earthquake waveforms, thereby improving the accuracy of focal mechanism identification.

[0005] At present, the research on focal mechanism mainly relies on theoretical models and traditional seismic data analysis methods, such as time-domain full waveform inversion and first-arrival polarity inversion methods. Although these methods have achieved certain results in conventional seismic analysis, they are often limited in the complex geological environment of deep coal mines. In addition, existing mine earthquake analysis tools usually require manual labeling and expert experience, are inefficient and easily affected by human factors. Summary of the invention

[0006] The purpose of this application is to provide a method, device, medium and product for predicting the focal mechanism of coal mine earthquake events, so as to improve the accuracy and efficiency of focal mechanism prediction.

[0007] To achieve the above objectives, this application provides the following solutions:

[0008] In a first aspect, the present application provides a method for predicting the focal mechanism of a coal mine earthquake event, comprising:

[0009] Acquire waveform data of a preset number of sampling points of a target coal mine earthquake event;

[0010] Preprocessing the waveform data to obtain processed waveform data; the preprocessing includes noise reduction processing, normalization processing based on initial arrival polarity, and waveform alignment processing;

[0011] According to the processed waveform data, a focal mechanism prediction model is used to determine a Gaussian distribution prediction curve of the focal mechanism of the target coal mine earthquake event; wherein the focal mechanism prediction model is obtained by training an initial focal mechanism prediction model using a training data set; the initial focal mechanism prediction model includes a graph convolutional neural network and a full convolutional neural network connected in sequence; the training data set includes waveform data of a preset number of sampling points of the mine earthquake event of the training coal mine and corresponding labels; the labels include a Gaussian distribution curve of a strike angle, a Gaussian distribution curve of a dip angle, and a Gaussian distribution curve of a slip angle;

[0012] The focal mechanism of the target coal mine earthquake event is determined according to the Gaussian distribution prediction curve of the focal mechanism; the focal mechanism is the angle corresponding to the highest probability value in each Gaussian distribution curve.

[0013] Optionally, it also includes:

[0014] According to the focal mechanism of the target coal mine earthquake event, the moment tensor of the mine earthquake event is calculated, and the corresponding focal mechanism beach ball is displayed.

[0015] Optionally, the initial prediction model of the focal mechanism is trained using the training data set, specifically including:

[0016] Get the training dataset;

[0017] Construct an initial prediction model of focal mechanism;

[0018] The waveform data of a preset number of sampling points of the mine earthquake events in the training coal mine are used as input, the corresponding labels are used as output, and the initial prediction model of the focal mechanism is trained in combination with the mean square error loss function to obtain the focal mechanism prediction model.

[0019] Optionally, the waveform data of a preset number of sampling points of the mine earthquake event of the training coal mine is used as input, the corresponding labels are used as output, and the initial prediction model of the focal mechanism is trained in combination with a mean square error loss function to obtain the focal mechanism prediction model, specifically including:

[0020] Preprocessing the waveform data of a preset number of sampling points of the mine earthquake event of the training coal mine to obtain processed waveform data;

[0021] Inputting the processed waveform data into the current initial prediction model of the focal mechanism to obtain a Gaussian distribution prediction curve of the focal mechanism of the mine earthquake event in the training coal mine;

[0022] According to the Gaussian distribution prediction curve of the focal mechanism of the mine earthquake event in the training coal mine and the corresponding label, the loss function value is determined by using the mean square error loss function;

[0023] Determine whether the training end condition is met; the training end condition is that the maximum number of iterations is reached or the loss function value is less than a set value;

[0024] If yes, the current initial prediction model of focal mechanism is used as the prediction model of focal mechanism;

[0025] If not, adjust the model parameters of the current focal mechanism initial prediction model according to the loss function value, and return to "input the processed waveform data into the current focal mechanism initial prediction model to obtain the Gaussian distribution prediction curve of the focal mechanism of the mine earthquake event in the training coal mine."

[0026] Optionally, the graph convolutional neural network includes a first graph convolutional layer, a second graph convolutional layer, a third graph convolutional layer and a fourth graph convolutional layer connected in sequence; the first graph convolutional layer and the second graph convolutional layer each include 256 neurons; the third graph convolutional layer and the fourth graph convolutional layer each include 128 neurons.

[0027] Optionally, the fully convolutional neural network includes a first convolutional layer, a second convolutional layer, a third convolutional layer, a maximum pooling layer, a first transposed convolutional layer, a second transposed convolutional layer and a third transposed convolutional layer connected in sequence; the first convolutional layer and the second transposed convolutional layer each include 64 neurons; the second convolutional layer and the first transposed convolutional layer each include 32 neurons; the third convolutional layer includes 16 neurons; the third transposed convolutional layer includes 100 neurons.

[0028] In a second aspect, the present application provides a computer device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement any of the above-described methods for predicting the source mechanism of coal mine earthquake events.

[0029] In a third aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for predicting the focal mechanism of coal mine earthquake events as described in any one of the above.

[0030] In a fourth aspect, the present application provides a computer program product, including a computer program, which, when executed by a processor, implements any of the above-mentioned methods for predicting the focal mechanism of coal mine earthquake events.

[0031] According to the specific embodiments provided in this application, this application has the following technical effects:

[0032] The present application provides a method, device, medium and product for predicting the focal mechanism of a coal mine earthquake event, which obtains waveform data of a preset number of sampling points of a target coal mine earthquake event; pre-processes the waveform data to obtain processed waveform data; based on the processed waveform data, uses a focal mechanism prediction model to determine the Gaussian distribution prediction curve of the focal mechanism of the target coal mine earthquake event; wherein the focal mechanism prediction model is obtained by training an initial focal mechanism prediction model using a training data set; the initial focal mechanism prediction model includes a graph convolutional neural network and a fully convolutional neural network connected in sequence; according to the Gaussian distribution prediction curve of the focal mechanism, the focal mechanism of the target coal mine earthquake event is determined; the focal mechanism is the angle corresponding to the highest probability value in each Gaussian distribution curve. The present application improves the accuracy and efficiency of focal mechanism prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0034] Figure 1 A schematic diagram of a flow chart of a method for predicting the focal mechanism of a coal mine earthquake event provided in one embodiment of the present application;

[0035] Figure 2 The basic flow chart for training and prediction of the GFFMNet model of this application;

[0036] Figure 3 This is a schematic diagram of the station distribution of the coal mine earthquake monitoring system of a coal mine;

[0037] Figure 4 It is a schematic diagram of the horizontal projection of the monitoring area grid;

[0038] Figure 5 This is the column chart of the top and bottom plate drilling of the 6306 working surface;

[0039] Figure 6 This is the GFFMNet model architecture diagram;

[0040] Figure 7 This is a graph showing how the loss function changes with the number of training rounds during training;

[0041] Figure 8 This is the bar graph of Kagan angle analysis results;

[0042] Fig. 9 This is a schematic diagram of the beach ball prediction results of the focal mechanism of the 6306 working face;

[0043] Fig.10 A schematic diagram of the structure of a computer device provided in one embodiment of the present application. DETAILED DESCRIPTION

[0044] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0045] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0046] In an exemplary embodiment, Figure 1 As shown, a method for predicting the focal mechanism of a coal mine earthquake event is provided, comprising the following steps:

[0047] S1: Acquire waveform data of a preset number of sampling points of a target coal mine earthquake event.

[0048] S2: Preprocessing the waveform data to obtain processed waveform data; the preprocessing includes noise reduction processing, normalization processing based on initial arrival polarity, and waveform alignment processing.

[0049] S3: According to the processed waveform data, a focal mechanism prediction model is used to determine a Gaussian distribution prediction curve of the focal mechanism of the target coal mine earthquake event; wherein, the focal mechanism prediction model is obtained by training an initial focal mechanism prediction model using a training data set; the initial focal mechanism prediction model includes a graph convolutional neural network and a fully convolutional neural network connected in sequence; the training data set includes waveform data of a preset number of sampling points of mine earthquake events in the coal mine used for training and corresponding labels; the labels include a Gaussian distribution curve of a strike angle, a Gaussian distribution curve of a dip angle, and a Gaussian distribution curve of a slip angle.

[0050] S4: Determine the focal mechanism of the target coal mine earthquake event according to the Gaussian distribution prediction curve of the focal mechanism; the focal mechanism is the angle corresponding to the highest probability value in each Gaussian distribution curve.

[0051] As an optional implementation, it also includes:

[0052] According to the focal mechanism of the target coal mine earthquake event, the moment tensor of the mine earthquake event is calculated, and the corresponding focal mechanism beach ball is displayed.

[0053] This application provides a model called GFFMNet (i.e., initial prediction model of focal mechanism) for the prediction of focal mechanism of coal mine earthquake events. The model combines the graph convolutional network (GCN) architecture with the fully convolutional neural network (FCN). The model can accurately identify the focal mechanism of coal mine earthquake events and shows high accuracy and computational efficiency. The basic process of training and prediction of the model is as follows: Figure 2 shown.

[0054] This application is generally divided into three steps: creating a data set, building and training a model (using Convolutional Neural Networks (CNN) to extract spatial features and Transformer to extract temporal features), and predicting the source mechanism of mine earthquakes.

[0055] As an optional implementation method, the initial prediction model of the focal mechanism is trained using the training data set, specifically including:

[0056] Step 1: Get the training dataset.

[0057] In actual applications, a coal mine in Shandong Province, China, has frequent mine earthquakes and a risk of rock burst. A mine earthquake monitoring system was deployed in the mine. The monitoring system consists of 6 three-component stations and 12 single-component microseismic sensors. The single-component microseismic sensors are installed perpendicular to the ground with a sampling rate of 500Hz. The plane distribution of mine earthquake stations and the plane diagram of the mine excavation project are shown in the figure below. Figure 3 As shown, the squares represent three-component stations, the numbers 1-6 correspond to their station names (three-component station 1-three-component station 6), and the circles represent the locations of single-component microseismic sensors 1-single-component microseismic sensors 12. The coordinates of each station are shown in Table 1.

[0058] Table 1 Station coordinates

[0059]

[0060] The key monitoring area is located in the 6306 coal mining face waiting area surrounded by single-component microseismic sensors. The vibration events mainly occur in the range of -550m to -700m above sea level. The waiting area of ​​the working face is a rectangle with an area of ​​about 700m×250m. Therefore, the geological body of 700m×250m×150m is divided into 14×5×3 grids. The length, width and height of each small grid are 50m, and there are 210 small grids in total. The horizontal projection of the grid is as follows Figure 4 As shown, the dots represent the locations of single-component sensors.

[0061] The focal mechanism includes strike angle, dip angle and rake angle. The strike angle (Strike) ranges from 0 to 360 degrees, the dip angle (Dip) ranges from 0 to 90 degrees, and the rake angle (Rake) ranges from 0 to 180 degrees. Based on the above three angles, the moment tensor of a mine earthquake event can be calculated and its corresponding beach ball can be displayed. This application samples every 36 degrees in the strike angle range, every 9 degrees in the dip angle range, and every 18 degrees in the rake angle range to construct samples. Therefore, the simulated source at the center of each small grid corresponds to 10×10×10, a total of 1000 focal mechanisms at different angles. Based on the simulated source location and 1000 different focal mechanism angles for each grid, the FK method is used to calculate the theoretical waveforms of the vertical components of 18 stations corresponding to each focal mechanism as samples, and then the theoretical waveforms corresponding to each event are normalized. If the initial polarity of the waveform is positive, the amplitude range is converted to between 0 and 1. If the initial polarity is negative, the amplitude range is converted to between -1 and 0. This method is used to assign the initial polarity to the samples. The real noise in the historical events of the mine is proportionally increased to the normalized waveform to improve the generalization ability of the model. Then based on the drill hole histogram of the area (the drill hole histogram is shown in Figure 5 The names and parameters of the drilling layers are shown in Table 2. The names of the drilling layers from top to bottom are basic top, direct top, 3 上 Coal, false bottom, direct bottom, 3 下 Coal and basic bottom) to build a layered velocity model, according to the theoretical arrival time of each station to offset, so that all waveforms are aligned, each waveform intercepted 1 second in length, that is, 500 sampling points. The sample label is a Gaussian distribution sequence of 100 sampling points in length constructed based on different strike angles, dip angles, and sliding angles corresponding to the theoretical waveform. The calculation formula of Gaussian distribution is as follows:

[0062] (1)

[0063] in, is the mean of the distribution, is the standard deviation of the distribution, which determines the width of the distribution. Here, this application uses an adaptive standard deviation , the value is one tenth of the angle range, that is:

[0064] For the strike angle, =36; for the inclination angle, =9; for the sliding angle, = 18. The angle corresponding to the highest probability value of the Gaussian distribution of each angle is the predicted value.

[0065] Therefore, the entire grid has a total of 14×5×3×10×10×10, totaling 210,000 sample data. Each sample includes waveform data of 18×500 sampling points, and the labels are Gaussian distribution sequences corresponding to the three angles of the focal mechanism.

[0066] Table 2 Columnar drilling parameters

[0067]

[0068] Step 2: Construct an initial prediction model of the focal mechanism.

[0069] The GFFMNet model is divided into two parts, the first part is GCN and the second part is FCN. The model hyperparameters are set based on the Bayesian Optimization (BO) method.

[0070] The graph convolutional neural network includes a first graph convolutional layer, a second graph convolutional layer, a third graph convolutional layer and a fourth graph convolutional layer connected in sequence; the first graph convolutional layer and the second graph convolutional layer each include 256 neurons; the third graph convolutional layer and the fourth graph convolutional layer each include 128 neurons.

[0071] The fully convolutional neural network includes a first convolutional layer, a second convolutional layer, a third convolutional layer, a maximum pooling layer, a first transposed convolutional layer, a second transposed convolutional layer and a third transposed convolutional layer connected in sequence; the first convolutional layer and the second transposed convolutional layer each include 64 neurons; the second convolutional layer and the first transposed convolutional layer each include 32 neurons; the third convolutional layer includes 16 neurons; the third transposed convolutional layer includes 100 neurons.

[0072] Model architecture details are as follows Figure 6 As shown. The GCN part first needs to build a graph structure according to the station location. Each station corresponds to a node of the graph, and the weight of the edge between the nodes corresponds to the spatial straight-line distance between the stations. An 18×18 adjacency matrix is ​​constructed based on the nodes and edges. The waveform length corresponding to each node is 500, so the size of the feature matrix is ​​18×500. The GCN part can fully extract the spatial relationship between stations. After the graph structure is built, the waveform data enters 4 GCN convolutional layers, and each layer has special parameters to extract and abstract spatial features from the data. The first two layers have 256 neurons respectively, and the last two layers have 128 neurons respectively. After being processed by the graph convolutional neural network, the input data size changes from 18×500 to 18×128.

[0073] The FCN part performs 3 layers of convolution, 1 layer of pooling, and 3 layers of transposed convolution on the 18×128 data. The final output size becomes 3×100, which corresponds to the Gaussian distribution of three angles. The angle corresponding to the highest probability is selected to obtain the final focal mechanism beach ball. Each convolution layer applies a set of learnable convolution kernels, and then introduces nonlinearity through the ReLU activation function, so that the model can capture the complex spatial relationships in the data. For the input earthquake feature X, the first l The convolution operation of the layer is defined as follows:

[0074] (2)

[0075] In the proposed model, Indicates l th The input of the layer, and Respectively represent l th The weights and biases of the convolutional layer of the layer. f represents the nonlinear activation function, i.e., ReLU, which means to perform the operation . It is l th This series of convolution operations gradually extracts and abstracts spatial features from the seismic data features extracted by GCN, and adjusts the output data to 3×100 size.

[0076] The batch size is set to 64, the Adam optimizer is used, and the initial learning rate is set to 1e-4. As the training progresses, a learning rate warm-up strategy is implemented to gradually increase the learning rate to 1e-4 in the initial period, and then fine-tune the training process by gradually reducing the learning rate based on the performance feedback of the validation set.

[0077] The loss function uses MSE (Mean Squared Error) to calculate the residual of the probability value corresponding to each angle to promote the generalization ability of the model. A total of 40 cycles of training were carried out. If the performance of the model on the validation set exceeds all previous cycles, the model is saved as the current best model. This ensures that we always have the best performing version of the model.

[0078] Step 3: Taking the waveform data of a preset number of sampling points of the mine earthquake event of the training coal mine as input and the corresponding labels as output, the initial prediction model of the focal mechanism is trained in combination with the mean square error loss function to obtain the focal mechanism prediction model, specifically including:

[0079] The waveform data of a preset number of sampling points of the mine earthquake event in the training coal mine are preprocessed to obtain processed waveform data.

[0080] The processed waveform data is input into the current initial prediction model of the focal mechanism to obtain a Gaussian distribution prediction curve of the focal mechanism of the mine earthquake event in the training coal mine.

[0081] According to the Gaussian distribution prediction curve of the focal mechanism of the mine earthquake event in the training coal mine and the corresponding labels, the loss function value is determined using the mean square error loss function.

[0082] Determine whether the training end condition is met; the training end condition is that the maximum number of iterations is reached or the loss function value is less than a set value.

[0083] If yes, the current focal mechanism initial prediction model is used as the focal mechanism prediction model.

[0084] If not, adjust the model parameters of the current focal mechanism initial prediction model according to the loss function value, and return to "input the processed waveform data into the current focal mechanism initial prediction model to obtain the Gaussian distribution prediction curve of the focal mechanism of the mine earthquake event in the training coal mine."

[0085] Figure 7 The figure shows the changes in the loss function during the training process. It can be observed that in the early stages of training, the loss function drops rapidly, indicating that the model can quickly learn and effectively extract features from the data, thereby quickly reducing the prediction error. After about 7 cycles, the downward trend of the loss function tends to stabilize, indicating that the model's learning of the training data has reached saturation. Throughout the training process, the loss function generally shows a downward trend and remains basically unchanged after 36 cycles, reaching an ideal state.

[0086] To evaluate the overall performance and estimation error of the model, a test dataset of 500 unseen synthetic samples was randomly generated, which simulated a diversity of normal faults, strike-slip faults, and reverse fault mechanisms. Using the trained model, focal mechanisms were predicted on this test dataset. For these predicted focal mechanisms, Kagan angle analysis was used to quantify the estimation error, where each Kagan angle quantifies the difference in rotation angle between the true focal mechanism and the predicted focal mechanism. Figure 8 As shown in the figure, from the results of the Kagan angle distribution, 96.6% of the Kagan angles are within 15 degrees, and only about 3.4% of the estimation errors exceed 15 degrees. Investigating the remaining 3.4%, this may be due to the equivalence of the two nodes of the focal mechanism (the true and auxiliary nodes are equivalent). Nevertheless, this test shows that the model proposed in this application can stably predict most events with acceptable estimation errors when testing a variety of unseen data. In addition, this test also verifies that the model of this application has learned the ability to predict diverse focal mechanisms.

[0087] After the training, the best model was applied to the prediction of focal mechanism. Six large-energy mine earthquake events with magnitudes between 1.8 and 2.5 were selected, which occurred on the 6306 working face. The waveform data of these events were of high quality. Fig. 9 As shown, the gray dots represent these 6 events. After the waveforms of all events are denoised, normalized based on the initial polarity, and processed by waveform alignment and interception based on the theoretical travel time, the focal mechanism beach ball prediction is performed. The beach ball on the side close to the working face is the prediction result of the GFFMNet model, and the beach ball on the side far from the working face is the calculation result of gCAP. It can be found from observation that the calculation results of the two are very close. The focal mechanism prediction results of the two mine earthquake events near the working face mining area are reverse faults, which is consistent with the result that the coal rock is squeezed and fractured due to the advance support pressure near the mining area, which verifies the accuracy of the GFFMNet model of this application. On average, the prediction of each event only takes 0.12s, which is much better than the traditional gCAP algorithm.

[0088] The GFFMNet model of this application combines the graph convolutional neural network (GCN) architecture with the fully convolutional neural network (FCN), which can effectively capture the complex characteristics of the focal mechanism. While ensuring the accuracy of the inversion, it shows superior performance in prediction efficiency over traditional methods. By integrating spatial information and node features, GFFMNet has significant advantages in identifying and classifying a variety of focal mechanisms. On a variety of data sets, GFFMNet demonstrates excellent generalization capabilities and improves the accuracy of the model while reducing computational costs. With more sufficient training data, GFFMNet has the potential to further reduce prediction errors, thereby improving the reliability of focal mechanism predictions. Optimizing this capability will be a focus of future research.

[0089] In an exemplary embodiment, a computer device is provided. The computer device may be a server or a terminal. The internal structure diagram thereof may be as follows: Fig.10As shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, referred to as I / O) and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a method for predicting the source mechanism of a coal mine earthquake event is implemented.

[0090] Those skilled in the art will understand that Fig.10 The structure shown in the figure is only a block diagram of a part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components. In an exemplary embodiment, a computer device is provided, including a memory and a processor, the memory stores a computer program, and the processor implements the above-mentioned coal mine earthquake event source mechanism prediction method when executing the computer program.

[0091] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, which, when executed by a processor, implements the above-mentioned method for predicting the focal mechanism of coal mine earthquake events.

[0092] In an exemplary embodiment, a computer program product is provided, including a computer program, which, when executed by a processor, implements the above-mentioned method for predicting the focal mechanism of coal mine earthquake events.

[0093] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0094] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM may be in various forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0095] The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. The non-relational database may include a distributed database based on blockchain, etc., but is not limited thereto. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but is not limited thereto.

[0096] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0097] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A method for predicting the focal mechanism of coal mine earthquake events, characterized in that: include: Acquire waveform data of a preset number of sampling points of a target coal mine earthquake event; Preprocessing the waveform data to obtain processed waveform data; The preprocessing includes noise reduction processing, normalization processing based on initial arrival polarity and waveform alignment processing; According to the processed waveform data, a focal mechanism prediction model is used to determine a Gaussian distribution prediction curve of the focal mechanism of the target coal mine earthquake event; wherein the focal mechanism prediction model is obtained by training an initial focal mechanism prediction model using a training data set; the initial focal mechanism prediction model includes a graph convolutional neural network and a full convolutional neural network connected in sequence; the training data set includes waveform data of a preset number of sampling points of the mine earthquake event of the training coal mine and corresponding labels; the labels include a Gaussian distribution curve of a strike angle, a Gaussian distribution curve of a dip angle, and a Gaussian distribution curve of a slip angle; The initial prediction model of focal mechanism is trained using the training data set, including: Get the training dataset; Constructing an initial prediction model of focal mechanism; wherein the graph convolutional neural network includes a first graph convolutional layer, a second graph convolutional layer, a third graph convolutional layer and a fourth graph convolutional layer connected in sequence; the first graph convolutional layer and the second graph convolutional layer each include 256 neurons; the third graph convolutional layer and the fourth graph convolutional layer each include 128 neurons; the full convolutional neural network includes a first convolutional layer, a second convolutional layer, a third convolutional layer, a maximum pooling layer, a first transposed convolutional layer, a second transposed convolutional layer and a third transposed convolutional layer connected in sequence; the first convolutional layer and the second transposed convolutional layer each include 64 neurons; the second convolutional layer and the first transposed convolutional layer each include 32 neurons; the third convolutional layer includes 16 neurons; the third transposed convolutional layer includes 100 neurons; Taking the waveform data of a preset number of sampling points of the mine earthquake event of the training coal mine as input and the corresponding labels as output, the initial prediction model of the focal mechanism is trained in combination with the mean square error loss function to obtain the focal mechanism prediction model, which specifically includes: Preprocessing the waveform data of a preset number of sampling points of the mine earthquake event of the training coal mine to obtain processed waveform data; Inputting the processed waveform data into the current initial prediction model of the focal mechanism to obtain a Gaussian distribution prediction curve of the focal mechanism of the mine earthquake event in the training coal mine; According to the Gaussian distribution prediction curve of the focal mechanism of the mine earthquake event in the training coal mine and the corresponding label, the loss function value is determined by using the mean square error loss function; Determine whether the training end condition is met; the training end condition is that the maximum number of iterations is reached or the loss function value is less than a set value; If yes, the current initial prediction model of focal mechanism is used as the focal mechanism prediction model; If not, the model parameters of the current focal mechanism initial prediction model are adjusted according to the loss function value, and "the processed waveform data is input into the current focal mechanism initial prediction model to obtain the Gaussian distribution prediction curve of the focal mechanism of the mine earthquake event in the training coal mine" is returned; The focal mechanism of the target coal mine earthquake event is determined according to the Gaussian distribution prediction curve of the focal mechanism; the focal mechanism is the angle corresponding to the highest probability value in each Gaussian distribution curve.

2. The method for predicting the focal mechanism of coal mine earthquake events according to claim 1, characterized in that: Also includes: According to the focal mechanism of the target coal mine earthquake event, the moment tensor of the mine earthquake event is calculated, and the corresponding focal mechanism beach ball is displayed.

3. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method for predicting the focal mechanism of coal mine earthquake events as described in any one of claims 1 to 2.

4. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for predicting the focal mechanism of a coal mine earthquake event described in any one of claims 1 to 2 is implemented.

5. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for predicting the focal mechanism of a coal mine earthquake event described in any one of claims 1 to 2 is implemented.

Images