A method, apparatus, equipment, medium, and product for early warning of HE events.

Abstract

This application discloses a method, device, equipment, medium, and product for early warning of seismic events (HE events), relating to the field of seismic early warning. The method includes: analyzing the spatiotemporal correlation of seismic data energy based on acquired seismic monitoring data using a semi-variogram function to determine a historical data window; constructing a dynamic sliding mechanism; based on the dynamic sliding mechanism, extracting the kernel density peak value of microseismic events using principal component analysis and kernel density estimation methods, and determining the fractal dimension index using fractal dimension analysis to analyze the changes in quantified geometric complexity and obtain quantification results; based on the quantification results, optimizing the early warning parameters using grid search and determining a sliding window joint early warning model; using the sliding window joint early warning model, based on the optimized early warning parameters, real-time monitoring of the kernel density peak value and fractal dimension index to achieve early warning of HE events. This application aims to improve the accuracy and timeliness of HE event prediction.

Description

Technical Field

[0001] This application relates to the field of mine tremor early warning, and in particular to a method, device, equipment, medium and product for early warning of HE events. Background Technology

[0002] Mine tremors are geological response phenomena triggered by mining activities, characterized by rock mass fracturing, slippage, and energy release. With the continuous increase in coal mining depth (some reaching 1500 meters), the frequency and intensity of mine tremors have significantly increased, and high-energy mine tremor events (HE events) pose a serious threat to mine safety. Although mine tremor monitoring technology has made some progress, existing early warning methods mainly rely on static models and single-indicator analysis, making it difficult to fully reveal the spatiotemporal heterogeneity of mine tremors and the complexity of their precursors.

[0003] Current methods rely on single-indicator analysis based on static time windows, such as source location and time series statistical analysis. These methods analyze microseismic data within a fixed time window, extracting single features (such as event frequency, total energy, or fractal dimension) to predict seismic events (HE events). However, these methods have several limitations: First, static time windows restrict analytical flexibility, failing to adapt to the dynamic characteristics of seismic activity varying with mining depth and time, resulting in insufficient timeliness of early warnings. Second, single-indicator analysis struggles to comprehensively characterize the spatiotemporal heterogeneity of seismic events, easily leading to false alarms or missed alarms, especially with low prediction accuracy in high-energy seismic events. This is because seismic activity is influenced by complex geological conditions and mining disturbances, resulting in highly dynamic and non-uniform spatiotemporal distribution, which existing methods do not dynamically adjust by incorporating multi-dimensional features. Furthermore, traditional methods lack detailed characterization of spatial clustering features, making it difficult to accurately locate seismic occurrence areas and limiting the spatial accuracy of early warnings. These shortcomings make current methods insufficient to meet the real-time and accuracy requirements of safe mine production. Summary of the Invention

[0004] The purpose of this application is to provide a method, apparatus, equipment, medium, and product for early warning of HE events, which can improve the accuracy and timeliness of HE event prediction.

[0005] To achieve the above objectives, this application provides the following solution:

[0006] Firstly, this application provides a method for early warning of HE events, including:

[0007] Acquire mine seismic monitoring data;

[0008] The spatiotemporal correlation of the energy in the seismic monitoring data is analyzed using a semi-variogram function to determine the historical data window.

[0009] A dynamic sliding mechanism is constructed; the dynamic sliding mechanism includes: a dynamic sliding time window determined based on historical data windows and time resolution determined according to the mine production cycle;

[0010] Based on the aforementioned dynamic sliding mechanism, principal component analysis and kernel density estimation methods are used to extract the kernel density peak value of microseismic events, and fractal dimension analysis is used to determine the fractal dimension index in order to analyze and quantify the changes in geometric complexity and obtain quantification results. The quantification results include the following information: the kernel density peak value before HE events shows an upward trend, while the fractal dimension index shows a downward trend.

[0011] A sliding window joint early warning model is constructed. The sliding window joint early warning model is determined based on the quantization results, with the objective function of maximizing the F1 score, and after optimizing the early warning parameters using grid search. The early warning parameters include: kernel density peak threshold, fractal dimension threshold, and early warning window.

[0012] The process of grid search: warning window range nuclear density peak threshold range fractal dimension threshold range For each combination, calculate the F1 score, and select the parameter combination with the highest F1 score as the optimized warning parameter; where, For resolution; This is the lower limit of the kernel density peak threshold; This represents the upper limit of the peak threshold for kernel density. This is the lower limit of the fractal dimension threshold; This represents the upper limit of the fractal dimension threshold.

[0013] A sliding window joint early warning model is adopted to monitor the kernel density peak and fractal dimension index in real time based on the optimized early warning parameters, so as to realize the early warning of HE events.

[0014] The fractal dimension analysis method employs the correlation integral method; the mathematical expression corresponding to the correlation integral method is:

[0015] ;

[0016] ;

[0017] in, For the relevant integral method function; The scaling factor; This represents the total number of earthquake events. For the first The first earthquake event and the first The distance of an earthquake event; It is the Heaviside step function;

[0018] The expression for the fractal dimension index is:

[0019] ;

[0020] in, For the relevant integrals; The fractal dimension index;

[0021] Principal component analysis and kernel density estimation methods were used to extract the kernel density peak values ​​of microseismic events, specifically including:

[0022] Principal component analysis was used for dimensionality reduction, and kernel density estimation was used to quantify the spatiotemporal clustering characteristics and determine the probability density.

[0023] A Gaussian kernel density function is used to extract kernel density peaks based on the probability density to determine the spatiotemporal clustering.

[0024] The expression corresponding to the probability density is:

[0025] ;

[0026] The expression for the Gaussian kernel density function is:

[0027] ;

[0028] in, It represents the probability density; For data points; This represents the total number of data points in the earthquake event group. For bandwidth parameters; For the first Data points for each earthquake event; is the Gaussian kernel density function.

[0029] In one embodiment, the semivariogram is a mathematical function used to characterize the change in semivariability under different lag values, corresponding to the resulting curve; the formula for calculating the semivariability is:

[0030] ;

[0031] in, It is semi-variable; For the lag interval The number of data point pairs separated; For the first Earthquake data; To and Paired, lag interval is The data.

[0032] Secondly, this application provides a HE event early warning device, comprising:

[0033] The data acquisition module is used to acquire mine seismic monitoring data;

[0034] The correlation analysis module is used to analyze the spatiotemporal correlation of the energy of the seismic data based on the seismic monitoring data using a semi-variogram function, and to determine the historical data window.

[0035] A window determination module is used to construct a dynamic sliding mechanism; the dynamic sliding mechanism includes: a dynamic sliding time window determined based on historical data windows and a time resolution determined according to the mine production cycle;

[0036] The quantization module is used to extract the kernel density peak value of microseismic events based on the dynamic sliding mechanism, using principal component analysis and kernel density estimation methods, and to determine the fractal dimension index using fractal dimension analysis to analyze the changes in quantization geometric complexity and obtain quantization results. The quantization results include the regularity information that the kernel density peak value before the HE event shows an upward trend, while the fractal dimension index shows a downward trend.

[0037] The model building module is used to construct a sliding window joint early warning model. The sliding window joint early warning model is determined based on the quantization results, with the objective function of maximizing the F1 score, and after optimizing the early warning parameters using grid search. The early warning parameters include: kernel density peak threshold, fractal dimension threshold, and early warning window.

[0038] The process of grid search: warning window range nuclear density peak threshold range fractal dimension threshold range For each combination, the F1 score is calculated, and the parameter combination with the highest F1 score is selected as the optimized warning parameter; where, For resolution; This is the lower limit of the kernel density peak threshold; This represents the upper limit of the peak threshold for kernel density. This is the lower limit of the fractal dimension threshold; This represents the upper limit of the fractal dimension threshold.

[0039] The monitoring and early warning module is used to monitor the kernel density peak and fractal dimension index in real time based on the optimized early warning parameters using a sliding window joint early warning model, so as to realize the early warning of HE events.

[0040] The fractal dimension analysis method employs the correlation integral method; the mathematical expression corresponding to the correlation integral method is:

[0041] ;

[0042] ;

[0043] in, For the relevant integral method function; The scaling factor; This represents the total number of earthquake events. For the first The first earthquake event and the first The distance of an earthquake event; It is the Heaviside step function;

[0044] The expression for the fractal dimension index is:

[0045] ;

[0046] in, For the relevant integrals; The fractal dimension index;

[0047] Principal component analysis and kernel density estimation methods were used to extract the kernel density peak values ​​of microseismic events, specifically including:

[0048] Principal component analysis was used for dimensionality reduction, and kernel density estimation was used to quantify the spatiotemporal clustering characteristics and determine the probability density.

[0049] A Gaussian kernel density function is used to extract kernel density peaks based on the probability density to determine the spatiotemporal clustering.

[0050] The expression corresponding to the probability density is:

[0051] ;

[0052] The expression for the Gaussian kernel density function is:

[0053] ;

[0054] in, It represents the probability density; For data points; This represents the total number of data points in the earthquake event group. For bandwidth parameters; For the first Data points for each earthquake event; is the Gaussian kernel density function.

[0055] Thirdly, this application provides a computer device, including: 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 HE event early warning method described above.

[0056] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the HE event early warning method described above.

[0057] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the HE event early warning method described above.

[0058] According to the specific embodiments provided in this application, the following technical effects are disclosed:

[0059] This application provides a method, device, equipment, medium, and product for early warning of seismic events (HE events). It analyzes the correlation between microseismic data in mine seismic monitoring data using a semi-variogram to comprehensively consider the temporal distribution characteristics of microseismic events. Simultaneously, based on the resolution determined by the mine production cycle, a dynamic sliding time window is constructed to fully capture the spatiotemporal dynamic evolution characteristics of microseismic activity as mining progresses. The sliding window joint early warning model uses principal component analysis combined with kernel density estimation to determine spatiotemporal aggregation characteristics and extract kernel density peaks. It also uses fractal dimension analysis to generate fractal dimension indices and optimizes them using a grid search method, ensuring adaptability and stability under different geological conditions. Based on the dynamic sliding time window and the sliding window joint early warning model, it can dynamically monitor the spatiotemporal aggregation characteristics and geometric complexity trends, achieving real-time prediction of HE events and significantly improving the accuracy and timeliness of early warning. Attached Figure Description

[0060] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0061] Figure 1 A flowchart for the HE event early warning method;

[0062] Figure 2 Design a flowchart for the HE event early warning method;

[0063] Figure 3 A bar chart showing the frequency of earthquake magnitudes;

[0064] Figure 4 A schematic diagram illustrating the calculation of the local magnitude b-value and the goodness of fit for a microseismic event;

[0065] Figure 5 This is a semivariogram of the seismic energy.

[0066] Figure 6 This is a clustering diagram of seismic events preceding HE event A; where... Figure 6 (a) is a schematic diagram of the PCA transformation of datasets M and A; Figure 6 (b) is a schematic diagram of the maximum value of the probability density function obtained by kernel density estimation of datasets M and A;

[0067] Figure 7 The fractal dimension fitting plot for events A, B, and C;

[0068] Figure 8 for Time evolution diagram;

[0069] Figure 9 for D A diagram illustrating the early warning results for a single indicator;

[0070] Figure 10 For joint early warning heat map;

[0071] Figure 11 This is a schematic diagram of the joint early warning results;

[0072] Figure 12 This is a schematic diagram of a joint early warning confusion matrix;

[0073] Figure 13 This is a structural diagram of a HE event early warning device;

[0074] Figure 14 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0075] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0076] To address the shortcomings of existing solutions, there is an urgent need to develop a dynamic early warning method that combines multi-dimensional features. By introducing a dynamic sliding time window, fractal dimension analysis, and spatiotemporal clustering joint feature extraction, the prediction accuracy and timeliness of HE events can be improved, providing reliable technical support for mine safety management. This application aims to overcome the limitations of existing technologies in dynamic analysis and multi-feature fusion, thereby achieving more accurate and timely early warning of HE events.

[0077] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0078] In one exemplary embodiment, such as Figure 1 As shown, a method for early warning of HE events is provided, including:

[0079] Step 100: Obtain mine seismic monitoring data.

[0080] Step 200: Using a semi-variogram function, analyze the spatiotemporal correlation of energy in the seismic monitoring data to determine the historical data window.

[0081] Step 300: Construct a dynamic sliding mechanism. The dynamic sliding mechanism includes: a dynamic sliding time window determined based on historical data windows and a time resolution determined according to the mine production cycle.

[0082] The semivariogram is a mathematical function used to characterize the change in semivariability under different lag values, corresponding to the resulting curve. The formula for calculating semivariability is:

[0083] .

[0084] in, It is semi-variable; For the lag interval The number of data point pairs separated; For the first Earthquake data; To and Paired, lag interval is The data.

[0085] Step 400: Based on the dynamic sliding mechanism, principal component analysis and kernel density estimation methods are used to extract the kernel density peak value of microseismic events, and fractal dimension analysis is used to determine the fractal dimension index to analyze the changes in quantification geometric complexity and obtain quantification results. The quantification results include the following: the kernel density peak value before HE events shows an upward trend, while the fractal dimension index shows a downward trend.

[0086] The fractal dimension analysis method employs the correlation integral method; the mathematical expression for the correlation integral method is:

[0087] .

[0088] .

[0089] in, For the relevant integral method function; The scaling factor; This represents the total number of earthquake events. For the first The first earthquake event and the first The distance of an earthquake event; It is the Heaviside step function.

[0090] The expression for the fractal dimension index is:

[0091] .

[0092] in, For the relevant integrals; It is a fractal dimension index.

[0093] In one embodiment, principal component analysis and kernel density estimation methods are used to extract the kernel density peak value of microseismic events, specifically including:

[0094] Principal component analysis was used for dimensionality reduction, and kernel density estimation was used to quantify the spatiotemporal aggregation characteristics and determine the probability density.

[0095] A Gaussian kernel density function is used to extract kernel density peaks based on probability density in order to determine the spatiotemporal clustering.

[0096] The expression corresponding to the probability density is:

[0097] .

[0098] The expression for the Gaussian kernel density function is:

[0099] .

[0100] in, It represents the probability density; For data points; This represents the total number of data points in the earthquake event group. For bandwidth parameters; For the first Data points for each earthquake event; is the Gaussian kernel density function.

[0101] Step 500: Construct a sliding window joint early warning model. The sliding window joint early warning model is determined based on quantization results, with maximizing the F1 score as the objective function, and after optimizing the early warning parameters using grid search. The early warning parameters include: kernel density peak threshold, fractal dimension threshold, and early warning window.

[0102] The grid search process includes: warning window range. nuclear density peak threshold range fractal dimension threshold range For each combination, calculate the F1 score, and select the parameter combination with the highest F1 score as the optimized warning parameter; where, For resolution; This is the lower limit of the kernel density peak threshold; This represents the upper limit of the peak threshold for kernel density. This is the lower limit of the fractal dimension threshold; This represents the upper limit of the fractal dimension threshold.

[0103] Step 600: Using a sliding window joint early warning model based on optimized early warning parameters, the kernel density peak and fractal dimension index are monitored in real time to achieve early warning of HE events.

[0104] like Figure 2 As shown, this application mentions an early warning method for high-energy seismic events (HE events) based on spatiotemporal clustering and fractal dimension analysis. First, it analyzes microseismic monitoring data (seismic monitoring data) using a semi-variogram function, comprehensively considering the temporal distribution characteristics of microseismic events, and optimizes the length of the historical data window to ensure sufficient coverage of event sequences. Simultaneously, it sets the real-time resolution according to the actual situation of the mine's production cycle, constructing a dynamic sliding time window to fully capture the spatiotemporal dynamic evolution characteristics of microseismic activity as mining progresses. Next, it uses Principal Component Analysis (PCA) to reduce the dimensionality of the spatial distribution data of microseismic events, combined with kernel density estimation. The Kernel Density Estimation (KDE) method quantifies the spatiotemporal aggregation characteristics of microseismic events over different time periods, extracting kernel density peaks as dynamic precursor signals in the time dimension. Simultaneously, fractal dimension analysis is used to calculate the geometric complexity of microseismic events, generating a fractal dimension index to characterize the temporal trend of microseismic activity. Then, a grid search method is employed to systematically optimize the key thresholds and warning time windows of the early warning model, ensuring its adaptability and stability under different geological conditions. Finally, combining kernel density peaks and fractal dimension indices, a sliding window joint early warning model is constructed. By dynamically monitoring the spatiotemporal aggregation characteristics and geometric complexity trends of microseismic activity, real-time prediction of HE events is achieved, significantly improving the accuracy and timeliness of early warnings and providing reliable support for mine safety management in complex geological environments.

[0105] This embodiment analyzes microseismic monitoring data from the 6306 working face of a coal mine. The monitoring period was from October 1, 2023 to April 1, 2024, and a total of 3084 valid microseismic event data were acquired. The monitoring system was configured in full-waveform continuous recording mode, with a sampling frequency set to 500Hz, meeting the high-precision detection requirements for high-energy mine seismic events. Verification through calibration shot experiments showed that the system's positioning error was less than 30m, indicating high reliability of the spatiotemporal positioning results.

[0106] 1. Definition of high-energy mine earthquake events.

[0107] For a given seismic data distribution set, the first problem to solve is how to quantitatively define high-energy (HE) seismic events. According to the theory of Gutenberg and Richer, the magnitude and cumulative number of seismic events follow formula (1):

[0108] (1)

[0109] in, The total number of earthquake events, indicating magnitude. Greater than the local magnitude The number of earthquake events, and All are constants.

[0110] Figure 3 The chart shows the distribution of magnitude frequency, with the frequency peaking around -0.3. To derive the complete dataset... This application employs the Maximum Likelihood Estimation (MLE) method. This method assumes that the magnitude-frequency diagram of earthquake events follows a Poisson process. It estimates the parameter values ​​of the Poisson distribution model, thereby maximizing the likelihood of the observed values. By applying MLE, The estimated value is 2.89. To evaluate the quality of the fit, the concept of Goodness of Fit (GoF) is introduced for each earthquake magnitude interval. As shown in formula (2).

[0111] (2)

[0112] in, and These represent the number of observed earthquake events and the number of expected earthquake events, respectively. This is the mean predicted by the Poisson distribution through maximum likelihood estimation. The goodness of fit changes with... Changes with the increase, such as Figure 4 As shown. Within the relatively high-energy region, GoF gradually decreases, and in... A sudden change occurs at this point. This value indicates the error between the observed time and the predicted value, which may represent a different earthquake catalog with different magnitude-frequency responses in the high-energy range. Therefore, it will... The threshold for high-energy (HE) events was set. Table 1 shows the distribution of events with a magnitude greater than 0.96 during this monitoring period, totaling 17 high-energy seismic events, which were defined as high-energy seismic events and numbered.

[0113] Table 1 Information on High-Energy Seismic Events

[0114]

[0115] 2. Event window parameterization.

[0116] To quantitatively analyze the correlation between non-uniformly spaced microseismic data, this application employs a semivariogram, which displays the transformation of semivariability under different hysteresis values, for the location hysteresis interval. Semivariability of seismic data The calculation method is as follows:

[0117] (3)

[0118] The semivariogram is a curve representing the semivariability at different lags, typically fitted to a specific model. As the lag increases, the semivariability between data points increases until it reaches a maximum at a certain lag. This increase in semivariability indicates a gradual weakening of autocorrelation. To further describe the correlation of the semivariogram, it mainly has three parameters: Nugger represents the lag... k =0 represents the semivariability; Sill represents the maximum variability of the data array; Range represents the scale fluctuation and the critical lag time required for the semivariability to reach its maximum value.

[0119] By analyzing the semivariogram of the instances, such as Figure 5 As shown, the energy of microseismic events exhibits significant spatial correlation within a 5-day timescale. Therefore, past time windows... Set to 5 days (the time range of historical data), and simultaneously set the sliding time window. The time resolution was set to 8 hours (for real-time monitoring) to capture the spatiotemporal evolution of energy correlations in microseismic events. For the entire dataset, [the following parameters were used]. Select a set of past and current microseismic events. By sliding this time window The past and current datasets are updated sequentially. Recorded microseismic events represent future events. The recorded seismic events represent past events, and high-energy events are studied through seismic datasets within each past-to-present time window.

[0120] 3. Calculation and analysis of spatiotemporal clustering.

[0121] This application employs a spatiotemporal clustering calculation method based on principal component analysis and kernel density estimation. First, a dataset of microseismic events spanning five days is selected as the base data, and principal component analysis (PCA) is used to transform the four-dimensional spatiotemporal data (x, y, z, t) into a two-dimensional plane. Then, the kernel density estimation (KDE) algorithm is used to calculate the probability density function of the microseismic events in the PCA space.

[0122] Given a set of microseismic events KDE can estimate each data point probability density in this group :

[0123] (5)

[0124] Gaussian kernel density function Typically smooth, symmetrical, and positive, bandwidth parameter This determines the smoothness of the estimation. KDE works by adjusting the smoothness of each data point. The probability density is transformed into smaller Gaussian distribution density curves, and then these smaller density blocks are merged to obtain the final probability density estimate. The Gaussian kernel density function expression used is:

[0125] (6)

[0126] To analyze a specific HE event, we assume that the day of the HE event is taken as the time of the event, and the past dataset (microseismic event set) is used. This data represents all seismic data prior to the event. The data was transformed into a two-dimensional coordinate system using PCA for kernel density estimation, thereby calculating the probability density (PDF) of the event. The probability density is then presented using contour lines, allowing assessment of the relationship between the high-energy (HE) event and the overall microseismic events. The location of the high-energy event shows a significant spatial proximity to the region of maximum kernel density, indicating a close correlation between its occurrence and high-density clustering of microseismic events. The kernel density peak value can characterize the spatial clustering of microseismic events and serve as an effective indicator for predicting the timing of high-energy microseismic events. Figure 6 The process of analyzing the spatiotemporal clustering of a high-energy event A.

[0127] Figure 6 Clustering of mining tremor events prior to HE event A, where, Figure 6 (a) is the PCA transformation of datasets M and A; Figure 6 (b) is the maximum value of the probability density function obtained from the kernel density estimation of datasets M and A.

[0128] 4. Analysis of the temporal characteristics of fractal dimension.

[0129] A method for calculating fractal dimension based on the correlation integral method. The mathematical expression for the correlation integral method is:

[0130] (7)

[0131] (8)

[0132] H It is a Heaviside step function, which takes the value "1" for positive parameters and '0' for negative parameters.

[0133] It can be seen from formula (7) Dependent on scale exponential factor , i.e., scale distance, can be expressed by formula (7) as follows:

[0134] (9)

[0135] By plotting the relevant integrals on a logarithmic scale and scale index factor The fractal dimension index is obtained from the slope of the linear fit curve:

[0136] (10)

[0137] A dynamic sliding window mechanism is used to calculate each timestamp. t Corresponding fractal dimension index Evaluation using the coefficient of determination R² The linear fitting relationship is shown in the analysis results. Figure 7 The results show that the fractal dimensions of high-energy seismic events (such as events A, B, and C) are 1.18, 1.59, and 1.45, respectively, and their coefficients of determination R² are 0.98, 1.00, and 0.99, respectively, indicating a good fit.

[0138] 5. Optimization of early warning parameters and sliding window early warning.

[0139] To quantify the performance of the early warning indicators and determine the optimal parameters, this application adopts the following evaluation indicators: True Positive Cases (TP): the number of successful high-energy seismic events; False Positive Cases (FP): the number of false early warnings; False Negative Cases (FN): the number of unpredicted high-energy seismic events; True Negative Cases (TN): the number of time windows for correctly issuing early warnings.

[0140] Based on the above metrics, the following performance evaluation metrics are calculated: Precision (P), Recall (R), and F1 score:

[0141] .

[0142] .

[0143] .

[0144] Using maximizing the F1 score as the objective function, the warning parameters are optimized through a grid search. The grid search process is as follows: setting the grid search parameters and the warning window range. nuclear density peak threshold range fractal dimension threshold range For each combination, calculate the F1 score and select the parameter combination with the highest F1 score as the optimal parameter.

[0145] Based on the warning threshold and warning window determined through grid search, this application implements a sliding window warning function for real-time monitoring and early warning of high-energy mining tremor events. The warning window size is determined accordingly. Starting from the first data point, the warning window moves gradually, shifting by one real-time resolution each time. The kernel density peak and fractal dimension (fractal dimension index) within each window are determined by statistically analyzing the median, average, maximum, and minimum values ​​within that window.

[0146] 6. Results.

[0147] To quantify the spatiotemporal clustering characteristics of pre-seismic microseismic activity, this application first analyzes the kernel density peak value. The temporal evolution pattern before high-energy mineral seismic events. A dynamic update mechanism based on a 5-day historical data window and an 8-hour real-time resolution window. It can effectively capture changes in the spatiotemporal concentration of mining seismic events. Figure 8 Showing the entire monitoring range The evolution of the sequence is shown, with the red dashed lines marking the occurrence frequency of 17 high-energy mineral seismic events. The results indicate that before most high-energy mineral seismic events, It shows a significant upward trend and reaches a local peak 1 to 2 days before the main shock, with a certain time lag, such as events A, B, C, G, and N, but this phenomenon was not found in events P and Q.

[0148] D The fractal dimension index is an important parameter characterizing the complexity of microseismic activity. D It can effectively reflect the geometric transformation of rock mass fracture. Figure 9 This demonstrates a fractal dimension index calculated using a dynamic update mechanism based on a 5-day historical data window and an 8-hour real-time resolution window. D The geometric changes throughout the monitoring period indicate that, prior to the occurrence of high-energy mineral tremors, the fractal dimension index... D The values ​​generally show an initial downward trend, and form a clear local minimum before the HE event.

[0149] This application constructs a joint early warning model based on two indicators. Specifically, the optimal model is determined through a grid search of three parameters. and D Threshold, Figure 10 Displayed under the optimal warning window of 40 hours and D The heatmap distribution of the threshold values ​​has an optimal warning window of 40 hours. and D The threshold values ​​are 0.13 and 1.3.

[0150] Based on optimized thresholds and warning windows, the kernel density peak and fractal dimension indices are monitored in real time. When both simultaneously meet the threshold condition (kernel density peak exceeding 0.13 and...),... D An alert is triggered when the value is below 1.3.

[0151] Its warning results are as follows Figure 11 As shown, the solid green line represents The maximum value within the warning window is represented by the solid blue line. D The value is the minimum value within the warning window. The green circle indicates the location where the trigger occurred, the green triangle indicates the actual location of the HE event, the purple diamond indicates the location of the accidental trigger, and the red cross indicates the location of the HE event without error. Figure 12 The specific analysis of the results shows that, under these conditions, 14 HE events received warnings, 3 were missed, and 12 were false alarms, with precision, recall, and F1 scores of 0.538, 0.824, and 0.651, respectively.

[0152] To comprehensively evaluate the advantages of the joint early warning model over single-indicator early warning, this application compared and analyzed the key performance indicators of the three early warning methods. The results are shown in Table 2. The results show that the joint early warning model significantly reduces the false alarm rate while maintaining a high recall rate, and its precision is significantly higher than that of the single-indicator early warning model. An increase of 17.1%, relative to D The value increased by 32.1%, and the F1 score increased by 18.3% and 30.2% respectively, which confirms the effectiveness and optimization of the joint early warning strategy in mine tremor early warning.

[0153] Table 2 Performance Comparison of Three Early Warning Models

[0154]

[0155] This application employs a dynamic sliding time window optimization method, which can adjust the analysis range in real time according to the spatiotemporal distribution characteristics of microseismic activity, overcoming the limitations of traditional static time windows and thus significantly improving the timeliness of early warning and the ability to respond to rapidly evolving disasters. Secondly, this application constructs a multi-parameter joint early warning model by combining spatiotemporal clustering and fractal dimension analysis, comprehensively characterizing the spatiotemporal heterogeneity and geometric complexity changes of microseismic activity. Compared with single-index analysis methods, this effectively reduces the false alarm rate and improves prediction accuracy. Furthermore, this application proposes a systematic analysis framework, including data window optimization, feature extraction, and parameter optimization, enhancing the model's adaptability and universality under different geological conditions, and providing reliable technical support for safety management in complex mining environments. These advantages work together to enable this application to achieve more accurate and timely early warning of high-energy mining earthquake events, thereby significantly reducing mine disaster risks and economic losses.

[0156] In one exemplary embodiment, such as Figure 13 As shown, a HE event early warning device is provided, comprising:

[0157] The data acquisition module is used to acquire mine seismic monitoring data.

[0158] The correlation analysis module is used to analyze the spatiotemporal correlation of energy in seismic monitoring data using a semi-variogram function, and to determine the historical data window.

[0159] The window determination module is used to construct a dynamic sliding mechanism. The dynamic sliding mechanism includes a dynamic sliding time window determined based on historical data windows and a time resolution determined according to the mine production cycle.

[0160] The quantization module, based on the aforementioned dynamic sliding mechanism, employs principal component analysis and kernel density estimation methods to extract the kernel density peak value of microseismic events, and uses fractal dimension analysis to determine the fractal dimension index, thereby analyzing the changes in quantization geometric complexity and obtaining quantization results. The quantization results include the following: the kernel density peak value before HE events shows an upward trend, while the fractal dimension index shows a downward trend. The determination process for the kernel density peak value and the fractal dimension index has been discussed above and will not be repeated here.

[0161] The model building module is used to construct a sliding window joint early warning model. This model is determined based on quantization results, with maximizing the F1 score as the objective function, and using grid search to optimize the early warning parameters. The early warning parameters include: kernel density peak threshold, fractal dimension threshold, and early warning window.

[0162] The process of grid search: warning window range nuclear density peak threshold range fractal dimension threshold range For each combination, the F1 score is calculated, and the parameter combination with the highest F1 score is selected as the optimized warning parameter; where, For resolution; This is the lower limit of the kernel density peak threshold; This represents the upper limit of the peak threshold for kernel density. This is the lower limit of the fractal dimension threshold; This represents the upper limit of the fractal dimension threshold.

[0163] The monitoring and early warning module is used to monitor the kernel density peak and fractal dimension index in real time based on the optimized early warning parameters using a sliding window joint early warning model, so as to realize the early warning of HE events.

[0164] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 14 As shown, this computer device includes a processor, memory, input / output interfaces (I / O), and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores HE event warning data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements the HE event warning method.

[0165] Those skilled in the art will understand that Figure 14 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0166] In one exemplary embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0167] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0168] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0169] 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, data stored, data displayed, 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 the relevant data must comply with relevant regulations.

[0170] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media 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), magnetic 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 can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0171] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0172] The technical features of the above embodiments can be combined in any way. For the sake of brevity, 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.

[0173] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for early warning of HE events, characterized in that, include: Acquire mine seismic monitoring data; The spatiotemporal correlation of the energy in the seismic monitoring data is analyzed using a semi-variogram function to determine the historical data window. Construct a dynamic sliding mechanism; The dynamic sliding mechanism includes: a dynamic sliding time window determined based on historical data windows and a time resolution determined according to the mine production cycle; Based on the aforementioned dynamic sliding mechanism, principal component analysis and kernel density estimation methods are used to extract the kernel density peak value of microseismic events. Fractal dimension analysis is then employed to determine the fractal dimension index, thereby analyzing and quantifying changes in geometric complexity to obtain quantification results. These results include: a trend of increasing kernel density peak value and decreasing fractal dimension index before HE events; HE events are high-energy seismic events, and their magnitudes are... Set the threshold for HE events; A sliding window joint early warning model is constructed. The sliding window joint early warning model is determined based on the quantization results, with the objective function of maximizing the F1 score, and after optimizing the early warning parameters using grid search. The early warning parameters include: kernel density peak threshold, fractal dimension threshold, and early warning window. The process of grid search: warning window range nuclear density peak threshold range fractal dimension threshold range For each combination, the F1 score is calculated, and the parameter combination with the highest F1 score is selected as the optimized warning parameter; where, For time resolution; This is the lower limit of the kernel density peak threshold; This represents the upper limit of the peak threshold for kernel density. This is the lower limit of the fractal dimension threshold; This represents the upper limit of the fractal dimension threshold. A sliding window joint early warning model is adopted to monitor the kernel density peak and fractal dimension index in real time based on the optimized early warning parameters, so as to realize the early warning of HE events. The fractal dimension analysis method employs the correlation integral method; the mathematical expression corresponding to the correlation integral method is: ； ； in, For the relevant integral method function; The scaling factor; This represents the total number of earthquake events. For the first The first earthquake event and the first The distance of an earthquake event; It is the Heaviside step function; The expression for the fractal dimension index is: ； in, For the relevant integrals; The fractal dimension index; Principal component analysis and kernel density estimation methods were used to extract the kernel density peak values ​​of microseismic events, specifically including: Principal component analysis was used for dimensionality reduction, and kernel density estimation was used to quantify the spatiotemporal clustering characteristics and determine the probability density. A Gaussian kernel density function is used to extract kernel density peaks based on the probability density to determine the spatiotemporal clustering. The expression corresponding to the probability density is: ； The expression for the Gaussian kernel density function is: ； in, It represents the probability density; Data points within earthquake event groups divided based on historical data windows; This represents the total number of data points in the earthquake event group. For bandwidth parameters; For the first Data points for each earthquake event; is the Gaussian kernel density function.

2. The HE event early warning method according to claim 1, characterized in that, The semivariogram is a mathematical function used to characterize the change in semivariability under different lag values, corresponding to the resulting curve; the formula for calculating the semivariability is: ； in, It is semi-variable; For the lag interval The number of data point pairs separated; For the first Earthquake data; To and Paired, lag interval is The data.

3. A HE event early warning device, characterized in that, include: The data acquisition module is used to acquire mine seismic monitoring data; The correlation analysis module is used to analyze the spatiotemporal correlation of the energy of the seismic data based on the seismic monitoring data using a semi-variogram function, and to determine the historical data window. The window definition module is used to build a dynamic sliding mechanism; The dynamic sliding mechanism includes: a dynamic sliding time window determined based on historical data windows and a time resolution determined according to the mine production cycle; The quantization module, based on the dynamic sliding mechanism, uses principal component analysis and kernel density estimation methods to extract the kernel density peak value of microseismic events, and uses fractal dimension analysis to determine the fractal dimension index, in order to analyze the changes in quantization geometric complexity and obtain quantization results. The quantization results include: a trend of increasing kernel density peak value and decreasing fractal dimension index before HE events; HE events are high-energy seismic events, and the magnitude... Set the threshold for HE events; The model building module is used to construct a sliding window joint early warning model. The sliding window joint early warning model is determined based on the quantization results, with the objective function of maximizing the F1 score, and after optimizing the early warning parameters using grid search. The early warning parameters include: kernel density peak threshold, fractal dimension threshold, and early warning window. The process of grid search: warning window range nuclear density peak threshold range fractal dimension threshold range For each combination, the F1 score is calculated, and the parameter combination with the highest F1 score is selected as the optimized warning parameter; where, For time resolution; This is the lower limit of the kernel density peak threshold; This represents the upper limit of the peak threshold for kernel density. This is the lower limit of the fractal dimension threshold; This represents the upper limit of the fractal dimension threshold. The monitoring and early warning module is used to monitor the kernel density peak and fractal dimension index in real time based on the optimized early warning parameters using a sliding window joint early warning model, so as to realize the early warning of HE events. The fractal dimension analysis method employs the correlation integral method; the mathematical expression corresponding to the correlation integral method is: ； ； in, For the relevant integral method function; The scaling factor; This represents the total number of earthquake events. For the first The first earthquake event and the first The distance of an earthquake event; It is the Heaviside step function; The expression for the fractal dimension index is: ； in, For the relevant integrals; The fractal dimension index; Principal component analysis and kernel density estimation methods were used to extract the kernel density peak values ​​of microseismic events, specifically including: Principal component analysis was used for dimensionality reduction, and kernel density estimation was used to quantify the spatiotemporal clustering characteristics and determine the probability density. A Gaussian kernel density function is used to extract kernel density peaks based on the probability density to determine the spatiotemporal clustering. The expression corresponding to the probability density is: ； The expression for the Gaussian kernel density function is: ； in, It represents the probability density; Data points within earthquake event groups divided based on historical data windows; This represents the total number of data points in the earthquake event group. For bandwidth parameters; For the first Data points for each earthquake event; is the Gaussian kernel density function.

4. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the HE event early warning method according to any one of claims 1-2.

5. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the HE event early warning method as described in any one of claims 1-2.

6. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the HE event early warning method as described in any one of claims 1-2.

Images