Mining earthquake risk early warning methods, systems, equipment and products

Abstract

This invention discloses a method, system, equipment, and product for early warning of mine seismic risks, belonging to the field of mine seismic data processing technology. It aims to address the problems of weak physical interpretability and insufficient reliability in existing mine seismic early warning methods. The method includes: calculating statistical parameters in multiple dimensions (time, depth, and planar space) based on microseismic event data to construct a statistical mine seismic risk index; establishing a numerical stress model of the target area and verifying the physical consistency between the high-risk area indicated by the statistical risk index and the high-stress area in the model; generating a consistency coefficient based on the verification results and using this coefficient to correct the statistical mine seismic risk index to obtain a corrected risk index; and finally, issuing an early warning based on the corrected risk index. This invention enhances the physical interpretability of the early warning results and improves the reliability and accuracy of the early warning by introducing independent physical consistency verification and correction steps.

Description

Technical Field

[0001] This application relates to the field of mine seismic data processing, and in particular to a method, system, equipment and product for mine seismic risk early warning. Background Technology

[0002] During the mining of deep mineral resources, high ground stress and strong mining disturbances lead to frequent high-energy mine tremors, seriously threatening the safety of personnel and equipment. Current technologies for mine tremor risk early warning mainly rely on statistical analysis of massive amounts of microseismic monitoring data, particularly applying the Gutenberg-Richter relationship to identify precursors by analyzing abnormal changes in the b-value (reflecting magnitude distribution) and a-value (reflecting seismic activity frequency). For example, a decrease in the b-value is often considered a sign of stress concentration.

[0003] However, this type of method, which relies solely on statistical analysis, has inherent flaws: First, there is a lack of direct physical correlation between statistical anomalies and actual rock mechanics processes, resulting in weak physical interpretability of the early warning results and difficulty in distinguishing between actual stress accumulation and random data fluctuations, leading to insufficient early warning accuracy; second, microseismic data itself has problems such as positioning errors and incomplete magnitude measurements, which affect the accuracy of statistical parameters and also lead to insufficient early warning accuracy; finally, traditional spatiotemporal analysis methods are difficult to fully capture the non-uniformity and nonlinear characteristics of stress evolution under complex geological conditions, which easily leads to false alarms or missed alarms.

[0004] In summary, the data analysis methods used in existing mine tremor risk early warning systems are insufficient to analyze the data sufficiently and accurately, resulting in problems with the accuracy of early warning systems and making them prone to false alarms and missed alarms. Summary of the Invention

[0005] The purpose of this application is to provide a method, system, equipment and product for early warning of mine tremor risks, which aims to improve the accuracy of risk warning and reduce the probability of false alarms / missed alarms.

[0006] To achieve the above objectives, this application provides the following solution: Firstly, this application provides a method for early warning of mine tremor risks, including: Based on microseismic event data of the target area, statistical parameters characterizing microseismic activity are calculated in the time domain, depth domain, and planar spatial domain, respectively. A unified statistical mine tremor risk index is constructed based on multiple statistical parameters from different dimensions. A numerical stress model of the target area is established to obtain its stress distribution field, and the high-risk area indicated by the statistical mine seismic risk index is compared with the high-stress area in the stress distribution field to perform physical consistency verification. A consistency coefficient is generated based on the results of the physical consistency verification, and the statistical mine tremor risk index is corrected using the consistency coefficient to obtain a corrected risk index. Early warnings are issued based on the aforementioned modified risk index.

[0007] Secondly, this application provides a mine tremor risk early warning system, including: The statistical parameter calculation module calculates statistical parameters characterizing microseismic activity in the time domain, depth domain, and planar spatial domain, based on microseismic event data of the target area. The risk index calculation module is used to construct a unified statistical mine tremor risk index based on multiple statistical parameters of different dimensions. The physical consistency verification module is used to establish a numerical stress model of the target area to obtain its stress distribution field, and compare the high-risk area indicated by the statistical mine seismic risk index with the high-stress area in the stress distribution field to perform physical consistency verification. The risk index correction module is used to generate a consistency coefficient based on the result of the physical consistency verification, and to use the consistency coefficient to correct the statistical mine tremor risk index to obtain a corrected risk index. The risk warning module is used to issue warnings based on the modified risk index.

[0008] 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 steps of the mine tremor risk early warning method described in any one of the above.

[0009] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the mine tremor risk early warning method described in any one of the above descriptions.

[0010] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the mine tremor risk early warning method described in any one of the above descriptions.

[0011] According to the specific embodiments provided in this application, the following technical effects are disclosed: This application provides a method, system, equipment, and product for early warning of mine tremor risks. Through a sequential process of "multi-domain fusion - physical verification - risk correction," it can effectively filter out false statistical anomalies caused solely by data fluctuations without any actual physical basis. By correcting the statistical mine tremor risk index using a consistency coefficient, the false alarm rate is significantly reduced. At the same time, the fusion of multi-dimensional information avoids the risk of missed alarms caused by the instability of a single indicator, making the final early warning results more stable and reliable.

[0012] Furthermore, in the physical verification, the data-driven statistical results are retrospectively validated with the rock mechanics model, providing a clear physical and mechanical basis for the early warning results. This method overcomes the shortcomings of traditional statistical methods, such as unclear physical meaning and the "black box" nature of integrated models, ensuring that every step of the early warning process has a clear physical meaning. Attached Figure Description

[0013] 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.

[0014] Figure 1 This is a flowchart illustrating a mine tremor risk early warning method in one embodiment of this application; Figure 2 This is a schematic diagram of the microseismic monitoring system and the spatial distribution of events in an application case of this application; Figure 3 This is a schematic diagram illustrating the magnitude-frequency relationship and the determination of integrity magnitude in an application case of this application; Figure 4 This is a schematic diagram of multi-domain a / b precursor extraction in an application case of this application; Figure 5 This is a graph showing the results of TRI risk threshold calibration and cross-validation based on ROC curves in an application case of this application; Figure 6 This is a consistency verification diagram of statistical precursor parameters and numerical stress surrogate quantities in an application case of this application; Figure 7 This is a diagram illustrating the effect of TRI early warning in an application case of this application. Detailed Implementation

[0015] 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.

[0016] 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.

[0017] The mine tremor risk early warning method in this embodiment is executed by a computer device, which can be a terminal computing device or a server. The terminal computing device can be, but is not limited to, various desktop computers, laptops, smartphones, tablets, IoT devices, etc. The server can be a standalone server or a server cluster composed of multiple servers, or it can be a cloud server.

[0018] Reference Figure 1 In one embodiment of the present invention, the mine tremor risk early warning method specifically includes: Step S110: Based on the microseismic event data of the target area (e.g., the mining face of a deep mine), calculate the statistical parameters characterizing microseismic activity in the time domain, depth domain, and planar spatial domain, respectively.

[0019] Existing technologies, relying solely on single-dimensional analysis in the time or spatial domains, are susceptible to random data fluctuations, leading to unstable precursor identification. This invention, by performing calculations across multiple dimensions, can capture signs of mine tremors from different perspectives, providing a more comprehensive information foundation for subsequent integrated judgment. This multi-dimensional data analysis solves the uncertainty problem inherent in single-dimensional analysis, laying the foundation for improving the stability and reliability of precursor identification.

[0020] Step S120: Construct a unified statistical mine tremor risk index based on multiple statistical parameters of different dimensions.

[0021] Its purpose is to integrate precursory parameters from different dimensions and with varying physical meanings into a single, quantifiable indicator. This avoids the complexity and subjectivity of manually interpreting multiple parameter charts, achieving automated and standardized risk level assessment. By constructing a unified risk index, it solves the problem of difficult comprehensive decision-making when multiple parameters coexist, and realizes the quantitative expression of mine tremor risk.

[0022] However, risk indices constructed solely based on statistical data may contain spurious anomalies with unclear physical meaning. To address this issue, step S130 is introduced.

[0023] Step S130: Establish a numerical stress model of the target area to obtain its stress distribution field, and compare the high-risk areas indicated by the statistical mine seismic risk index with the high-stress areas in the stress distribution field to perform physical consistency verification.

[0024] Among them, high-risk areas refer to areas where the statistical mine seismic risk index is higher than the preset risk threshold, and high-stress areas refer to areas where the stress is higher than the preset stress threshold.

[0025] The core idea of ​​this step is that the statistical manifestations of mine seismic risks with real physical significance (such as anomalies in microseismic activity) should spatially correspond to their physical causes (such as high stress concentration). By introducing an independent physical model for posterior review, statistical noise lacking a physical basis can be effectively filtered out, thus solving the fundamental defect of unclear physical meaning in traditional statistical methods.

[0026] Step S140: Generate a consistency coefficient based on the results of the physical consistency check, and use the consistency coefficient to correct the statistical mine tremor risk index to obtain the corrected risk index.

[0027] This correction process is not a simple "yes" or "no" judgment, but rather uses a quantified consistency coefficient to "score" or "weight" statistical risk, assigning higher risk weights to areas with better physical consistency. This design ensures that the final risk assessment results retain the sensitivity of statistical data while incorporating the determinism of the physical model, achieving a deep integration of data-driven approaches and physical mechanisms. This correction mechanism solves the problem of how to quantify physical verification results and feed them back into the risk assessment model, significantly improving the reliability of early warning results.

[0028] Step S150: Issue an early warning based on the modified risk index.

[0029] This step triggers corresponding early warnings by comparing the value of the revised risk index with a pre-set risk threshold. Since the revised risk index has undergone physical verification, the risks it indicates are more certain. Therefore, early warnings issued based on this index will have effectively controlled false alarm and false negative rates, providing a solid basis for mines to take precise and targeted disaster prevention and mitigation measures.

[0030] In summary, this invention introduces an independent physical consistency verification step to perform posterior verification between data-driven statistical results and rock mechanics models, providing a clear physical and mechanical basis for early warning results. This method overcomes the shortcomings of traditional statistical methods, such as unclear physical meaning and the "black box" nature of integrated models, ensuring that every step of the early warning process has a clear physical meaning.

[0031] Furthermore, the sequential process of "multi-domain fusion - physical verification - risk correction" effectively filters out false statistical anomalies caused solely by data fluctuations without any actual physical basis. Correcting the statistical mine tremor risk index using a consistency coefficient significantly reduces the false alarm rate. Simultaneously, the fusion of multi-dimensional information avoids the risk of missed reports caused by the instability of a single indicator, making the final early warning results more stable and reliable.

[0032] In one specific implementation, the statistical parameters are the a-value and b-value of the Gutenberg-Richard relation.

[0033] The α-value and b-value are classic parameters in seismology used to describe seismic activity. The b-value is usually negatively correlated with the regional stress level, and its decrease is widely considered an indicator of stress concentration. The α-value, on the other hand, reflects the frequency of microseismic events in the region. Choosing these two parameters with clear physical meaning as a foundation provides a solid theoretical basis for subsequent risk analysis. Compared to using some general statistical characteristics without clear physical meaning, the precursor identification in this scheme is more targeted and interpretable.

[0034] Furthermore, the construction steps of the statistical mine tremor risk index include: constructing a time sub-index based on the changing trend of statistical parameters in the time domain; constructing a depth sub-index based on the vertical gradient of statistical parameters in the depth domain; constructing a spatial sub-index based on the spatial distribution of statistical parameters in the planar spatial domain; and performing linear weighted fusion of the time sub-index, depth sub-index, and spatial sub-index to obtain the statistical mine tremor risk index.

[0035] This divide-and-conquer strategy allows the model to clearly characterize risk contributions from different dimensions and flexibly adapt to early warning needs under different geological conditions and mining stages by adjusting weighting coefficients. By decomposing risk into multi-dimensional sub-indicators and then fusing them, the problem of how to systematically integrate multi-dimensional precursory information is solved, making the composition of the risk index clear and adjustable.

[0036] Specifically, the formula for constructing the time sub-indicator is as follows: ; in, For time sub-indicators, and These are the trend terms for the values ​​of a and b in the time domain, respectively. This is the normalization function; The formula for constructing the depth sub-metric is: ; in, As a depth sub-metric, and These are the gradients of the values ​​a and b in the depth domain in the vertical direction, respectively. This is the normalization function; The formula for constructing the spatial sub-index is: ; ; in, For spatial sub-indices, For empirical quantile normalization, As an indicator of spatial hazard, and These represent the spatial distributions of the values ​​of a and b in the planar spatial domain, respectively. This is a standardized function.

[0037] The above formula clarifies the optimal calculation method for each sub-indicator.

[0038] In one specific implementation, the weight coefficients of the linear weighted fusion are determined using the Analytic Hierarchy Process (AHP). AHP is a mature multi-objective decision analysis method that can decompose complex decision problems into multiple levels and transform expert qualitative judgments into quantitative weights through pairwise comparisons. Using this method to determine weights makes the weight allocation process more scientific and systematic, avoiding the arbitrariness of subjective setting based on experience, thereby improving the objectivity and robustness of the risk index model.

[0039] Based on the above implementation method, a preferred setting is that the weighting coefficient of the time sub-indicator is greater than that of the spatial sub-indicator. This setting is derived from extensive practical experience in mine tremor early warning. In short-term mine tremor early warning, the temporal evolution trend of the risk state (such as a rapid decrease in the b-value) is often more indicative than the slow changes in its spatial distribution. Therefore, assigning a higher weight to the time sub-indicator makes the risk index more sensitive to impending dangers, thereby improving the timeliness of the early warning.

[0040] In another alternative implementation, the weighting coefficients of the linear weighted fusion can be dynamic weighting coefficients. Specifically, the dynamic weighting coefficients are adaptively adjusted based on the preset risk level of the statistical seismic risk index at the previous moment. For example, when the risk level is low, the model can focus more on the initial signs of spatial anomalies; while when the risk level increases, the model can dynamically increase the weight of the time sub-indicators to more sensitively capture rapidly evolving pre-earthquake precursors. This adaptive adjustment mechanism makes the early warning model more intelligent, enabling it to automatically switch its focus according to different stages of risk evolution, thereby achieving optimal early warning performance throughout the entire early warning period.

[0041] In one specific implementation, the results of the physical consistency check include the overlap ratio, center distance, and intensity ranking consistency of high-risk and high-stress areas; the consistency coefficient is determined based on the overlap ratio, center distance, and intensity ranking consistency.

[0042] By decomposing the abstract concept of "consistency" into three quantifiable indicators—spatial overlap, proximity, and strong correlation—the consistency verification process becomes more concrete and objective, solving the problem of how to quantify the degree of fit between statistical anomalies and physical models.

[0043] For example, the formula for calculating the stress consistency coefficient is: ; ; ; in, The stress consistency coefficient, This is the spatial overlap coefficient, used to characterize the overlap ratio between high-risk and high-stress areas; This is the distance consistency coefficient, used to characterize the degree of proximity between the center of a high-risk area and the center of a high-stress area; The intensity ranking consistency coefficient is used to characterize the degree of consistency between the intensity ranking of high-risk areas and the intensity ranking of high-stress areas. , , Let be the weight coefficient, and satisfy... + + =1, This is a high-risk area. This is a high-stress area. The center-to-center distance between high-risk and high-stress areas. This is the spatial decay scale.

[0044] In one specific correction method, the correction is achieved by multiplying the statistical seismic risk index by the consistency coefficient. This method is simple and intuitive, with the consistency coefficient acting as a "discount factor" or "confidence level." If the statistical anomaly is completely inconsistent with the physical model (consistency coefficient tends to 0), the statistical seismic risk index will be significantly reduced; if it is completely consistent (consistency coefficient tends to 1), the statistical seismic risk index is retained. This correction method can effectively filter out spurious anomalies with unclear physical meaning.

[0045] In another parallel correction method, the correction is achieved by weighting and summing the statistical seismic risk index and the consistency coefficient. Compared to multiplicative correction, the weighted summation method provides a mechanism for balancing statistical risk and physical consistency. By adjusting the weights of the two, it is possible to control whether the final corrected risk index leans more towards statistical results or the judgment of the physical model, providing greater flexibility for risk decision-making in different scenarios.

[0046] Therefore, the corrected formula for the statistical mine tremor risk index is: or ; in, and These are the statistical mine tremor risk indices before and after the correction. The consistency coefficient, For statistical risk weights.

[0047] In one specific implementation, the method further includes using historical data to determine risk thresholds for early warning and high-risk area classification by constructing receiver operating characteristic (ROC) curves and calculating the Youden index. ROC curve analysis is a classic method in signal detection theory, and the thresholds determined by this method achieve an optimal balance between the sensitivity (no missed detections) and specificity (no false alarms) of early warning. This solves the problem of early warning thresholds being set based on experience and lacking scientific basis, enabling the entire early warning process to achieve closed-loop and optimization. The invention will be further illustrated by an application example below.

[0048] In one application example of the present invention, the mine tremor risk early warning method includes the following steps: Step 1: Create a dataset. This step mainly details the process of acquiring or generating microseismic event data for the target area.

[0049] First, the spatial range and mining condition parameters of the study area (target area) are determined, and the geological structural characteristics and mining conditions of the study area are obtained, including fault structure, fold morphology, coal seam distribution and roof and floor strata structure. At the same time, the overlying strata structure and its mechanical parameters as well as in-situ stress state parameters are obtained to characterize the stress redistribution characteristics during mining and to provide a basis for subsequent physical consistency verification.

[0050] The 6306 working face of the No. 6 mining area in a coal mine in a certain region was selected as the study area. A microseismic monitoring system was deployed within the study area, with several three-component and single-component seismic stations set up to continuously acquire microseismic waveform signals generated during mining operations. The sampling rate of the monitoring system was 500 Hz, the effective frequency band of the three-component stations was 0.5–100 Hz, and the effective frequency band of the single-component stations was 1–600 Hz. The deployment method of the monitoring system and the spatial distribution of the stations are as follows. Figure 2 As shown. Figure 2 This is used to illustrate the monitoring data sources and spatial coverage conditions upon which the present invention is based. Figure 2 The medium triangle represents an earthquake or microseismic station, the gray dot represents a localized microseismic event, and the red star represents a high-energy mineral seismic event. (Through...) Figure 2 It can determine the coverage area of ​​the monitoring system in the study area, the spatial clustering area of ​​events, and the spatial relationship between high-energy mine tremors and working faces, faults, and goaf areas, providing a basis for subsequent calculation of planar spatial domain GR parameters and location of risk areas.

[0051] The acquired microseismic waveform data were processed for event identification and source location. Based on P-wave arrival time information, the Geiger iterative method was used for source location, and the data was corrected by combining it with a layered P-wave velocity model of the study area. This yielded microseismic event data containing information on event occurrence time, spatial location (x, y, z), and magnitude or energy. The spatial uncertainty of the source location results was preferably controlled within the range of 15–20 m horizontally and 5–10 m vertically.

[0052] The microseismic event data were organized, a microseismic event catalog was constructed, and parameters such as time, spatial location, magnitude, and energy of each event were recorded.

[0053] Data quality control and screening are performed on the microseismic event catalog to ensure the reliability of statistical analysis. The maximum curvature method can be used to determine the integrity magnitude. Mc Only earthquakes with a magnitude of not less than 100 are retained. Mc The events are used in subsequent analysis. Event data that does not meet the magnitude requirements are removed from the original microseismic event data to form the final microseismic event data used for mine earthquake risk early warning.

[0054] Step 2: Multi-domain GR parameter estimation. This step mainly details the process of "calculating statistical parameters characterizing microseismic activity in the time domain, depth domain, and planar spatial domain, respectively."

[0055] The number of microseismic events retained from the microseismic event data was statistically analyzed, a frequency-magnitude distribution relationship was established, and the Gutenberg-Richter power-law relationship was used to describe it. ; in, NFor earthquakes with a magnitude greater than or equal to M The cumulative number of microseismic events; only in M ≥ Mc Statistical calculations are performed under the condition of (integrity magnitude); the a value characterizes the frequency of seismic activity, while the b value reflects the distribution characteristics of rupture events.

[0056] The value of b is calculated using the maximum likelihood estimation method: ; in, To meet M ≥ Mc The average magnitude of microseismic events under certain conditions. The magnitude box width is set to 0.1; and the standard error of the b-value is calculated using the following formula: .

[0057] Based on this, multi-domain calculations are performed on the Gutenberg–Richter relation parameters, specifically including: (1) Within the time domain, event data is statistically analyzed on a daily basis, and only when a single day meets the requirements... M ≥ Mc The number of microseismic events N When the value is ≥30, calculate the corresponding a and b values, and remove data that do not meet the conditions.

[0058] (2) In the depth domain, a sliding window is constructed vertically, with the range from the source depth of the high-energy mine earthquake to the coal seam as the analysis interval. The half height of the window is set to 0.01 km. The a and b values ​​of the events within the window are calculated, and the window is gradually slid along the depth direction to obtain the parameter distribution.

[0059] (3) Within the planar spatial domain, the study area is divided into 50 m × 50 m grid cells, and the parameters satisfying the conditions within each grid cell are statistically analyzed. M ≥ Mc Microseismic events, only the number of events N The a and b values ​​are calculated for grids with a grid size of ≥30; grids with insufficient event counts are not calculated. The G–R relationship and its typical frequency-magnitude distribution characteristics are as follows: Figure 3 As shown.

[0060] like Figure 3 As shown, Figure 3 (a) and Figure 3 (b) Establish magnitude-frequency distribution relationships for events in the long-term background catalog and the target time window, respectively, and determine the integrity magnitude using the maximum curvature method. Mc Only if the event magnitude meets the following conditions. M ≥ McOnly when a and b values ​​are detected will they be included in the calculation of GR parameters, in order to avoid the omission of low-magnitude events and the resulting bias in the estimation of GR parameters. Figure 3 It is also used to verify whether the number of events and magnitude distribution within a short-term warning window possess statistical stability. If the target window meets the following conditions... M ≥ Mc If the number of microseismic events is lower than the set threshold, the GR parameters and the Mine Seismic Risk Index (TRI) results for that time window will not be output.

[0061] Step 3: Multi-domain precursor feature extraction and normalization. This step mainly details the construction of temporal, depth, and spatial sub-indicators.

[0062] To transform multi-domain G-R parameters into actionable precursor features, the time series parameters a and b are first smoothed to reduce the impact of daily fluctuations on short-term early warning while maintaining sensitivity to stress accumulation and crack propagation. Specifically, a five-day exponentially weighted moving average (EWMA) is used, and its smoothing formula is as follows: ; in, Here are the smoothed parameter values ​​at time t. The original parameter values ​​(a or b) at time t; This is the smoothed value of the previous time step t-1; It is a smoothing factor with a value range of [0,1], corresponding to a 5-day time scale.

[0063] To quantify the trend of parameter changes, the standardized lag difference is calculated: ; in, The trend of parameter change at time t; The parameter value is lagging by 5 time steps; This is the standard deviation of the parameter sequence, used for standardization.

[0064] The a and b values ​​obtained after the above preprocessing can then be used to construct time sub-indicators, depth sub-indicators, and spatial sub-indicators.

[0065] Within the time domain, scissor-like coupling precursors are extracted based on the changing trends of the a and b values. Scissor-like coupling precursors refer to a situation where, within the same warning time window, the b value shows a continuous decreasing trend while the a value shows a continuous increasing trend. Specifically, a decrease in the b value indicates enhanced stress concentration in the rock mass, increased fracture scale, and improved structural locking; an increase in the a value indicates enhanced microfracture activity, accelerated fracture propagation, and enhanced damage accumulation. When both occur simultaneously, the study area is determined to be in a coupled evolution stage of "enhanced stress concentration—accelerated damage activity," and its risk of mine-seismic gestation is higher than in the case of only a single parameter anomaly.

[0066] The formula for constructing the time sub-indicator is: ; in, This is a time-based sub-index used to characterize the stress accumulation rate and crack evolution intensity. and These are the trend terms for the values ​​of a and b in the time domain, respectively. For normalization functions, the Min–Max normalization method can be used to map variables to the [0,1] interval.

[0067] Within the depth domain, the mechanical differences between rock layers are characterized by calculating the vertical gradients of the a and b values.

[0068] The formula for constructing the depth sub-metric is: ; in, This is a depth sub-index used to reflect the mechanical differences and stress migration path characteristics between different rock strata. It can reflect the non-uniformity of stress and fracture distribution in rock strata at different depths. and These are the gradients of the values ​​a and b in the depth domain in the vertical direction, respectively. This is the normalization function.

[0069] The formula for constructing the spatial sub-index is: ; ; in, This is a spatial sub-index, characterizing the spatial clustering of potential fracturing areas and the distribution of hazardous areas. For empirical quantile normalization, This is a spatial hazard index, specifically representing the hazard level at spatial location (x, y). and These represent the spatial distribution of the values ​​of a and b in the planar spatial domain, specifically the values ​​of a and b at the spatial location (x, y). The function is standardized using Z-score.

[0070] During the calculation process, a coarse grid (50 m × 50 m, number of events N≥50) can be used for statistical analysis first, and then extended to a fine grid (10 m × 10 m) through bilinear interpolation to improve the spatial resolution.

[0071] The above steps completed the construction of the time sub-indicator, depth sub-indicator, and spatial sub-indicator, as shown in the following figure. Figure 4 As shown.

[0072] Figure 4 (a) indicates the synchronous evolution characteristics of the a and b values ​​in the time domain. When the b value continues to decrease while the a value increases synchronously, it indicates that the rock mass has entered the scissor-like precursor stage of "intensified stress concentration - accelerated microfracture activity". Figure 4 (b) represents the distribution of the a / b parameter along the vertical depth direction, used to identify mechanical differences and potential fracture zones near hard roofs, coal seams and interlayer interfaces; Figure 4 (c) and Figure 4 (d) represents the anomalous distribution of a and b values ​​in planar space, where high a-value areas indicate enhanced micro-fracture activity, and low b-value areas indicate stress concentration and an increased proportion of large-scale fractures. If high a-value areas and low b-value areas overlap spatially, they are identified as potential seismic nucleation zones and used as the main source of spatial sub-indicators.

[0073] Step 4: Construction of a statistical mine tremor risk index.

[0074] After obtaining the time, depth, and spatial sub-indices, each sub-indicator is normalized and then a weighted linear fusion is used to construct the mine seismic risk index TRI, the calculation formula of which is as follows: ; in, The mine earthquake risk index; It is a time-based sub-indicator used to characterize time-evolutionary features; This is a depth sub-index used to characterize vertical structural differences; It is a spatial sub-index used to characterize the spatial distribution characteristics of a plane; , and These are the time, depth, and spatial weighting coefficients, respectively.

[0075] The weights satisfy the constraints: ; The Analytic Hierarchy Process (AHP) is used to determine the weights, and considering the dominant role of temporal evolution in short-term early warning, the weights are set as follows: =0.50、 =0.25、 =0.25. The mine tremor risk index is a continuous real-time indicator with a value range of [0,1]. It is used to quantify the level of mine tremor risk, where a larger value indicates a higher mine tremor risk.

[0076] This step integrates multi-domain precursor information such as time trends, vertical gradients, and spatial anomalies into a single risk indicator, enabling the coordinated expression of multi-scale information.

[0077] Step 5: Risk threshold calibration.

[0078] Based on the constructed mine earthquake risk index TRI, a risk level classification is established.

[0079] During the threshold calibration process, two types of binary classification tasks are established respectively: The first type is the discrimination between "high risk H and non-high risk M + B", which is used to determine the high risk threshold TH; The second type is the discrimination between "medium-high risk H + M and background risk B", which is used to determine the medium risk threshold TM. TH and TM are determined based on the ROC curve and Youden index respectively. In a certain scenario, TH = 0.683 and TM = 0.445. The area where TRI ≥ TH is the high risk area; the area where TM ≤ TRI < TH is the medium risk area; the area where TRI < TM is the low risk area. During the threshold calibration process, the BlockedCV (Blocked Cross-Validation) strategy can be adopted to divide the data according to consecutive time periods, ensuring that the training set and the test set are strictly independent in time, so as to maintain the continuity and integrity of the seismic time series and avoid the impact of forward-looking bias on the model performance.

[0080] The discrimination ability of TRI is evaluated through the Receiver Operating Characteristic (ROC) curve and the Precision-Recall (PR) curve, and the Area Under the Curve (AUC) is used as the evaluation index. Among them, the ROC curve reflects the relationship between the true positive rate and the false positive rate; the PR curve reflects the relationship between the precision rate and the recall rate; the AUC value is used to measure the overall discrimination ability of the model, and its value range is [0, 1].

[0081] As Figure 5 shown, two types of risk discrimination tasks are constructed respectively based on the results of blocked cross-validation. Among them, Figure 5 (a) Taking the high risk level H as the positive category and the medium risk level M and the background level B as the negative categories, which is used to evaluate the recognition ability of TRI for the pregnant state of high risk mine earthquakes; Figure 5 (b) Taking the medium-high risk level H + M as the positive category and the background level B as the negative category, which is used to evaluate the recognition ability of TRI for the early rise of the risk state. Referring to Figure 5 , for each fold of cross-validation, the ROC curve is generally close to the upper left corner, indicating that TRI has a good discrimination ability for the risk state. Figure 5 The circles in

[0082] In one validation test, the mean AUC for the high-risk level H validation task was 0.910, with a 95% confidence interval of 0.887–0.929; the mean AUC for the medium-to-high-risk level H+M validation task was 0.904, with a 95% confidence interval of 0.883–0.930. These results demonstrate that the constructed TRI maintains stable risk discrimination capability across different time periods and can be used to determine high-risk and medium-risk thresholds.

[0083] based on Figure 5 The ROC validation results shown further determine the optimal risk threshold by optimizing the balance between sensitivity and false alarm rate, preferably using the Youden index as the criterion: ; in, The Youden index; Sensitivity Sensitivity; Specificity For specificity. When When the maximum value is taken, the corresponding threshold is used as the optimal risk boundary point.

[0084] Based on the high-risk threshold, the TRI calculated in real time can be classified and judged, and the corresponding risk level can be output to identify high-risk areas.

[0085] Step Six: Physical Constraint Consistency Verification and Result Correction.

[0086] like Figure 6 As shown, Figure 6 (a) shows the relationship between the measured local b-values ​​and the simulated deviatoric stress index. As the simulated deviatoric stress index increases, the local b-values ​​generally show a decreasing trend, indicating that the low b-value regions correspond to a higher stress concentration state; Figure 6 (b) shows the relationship between the measured local a-value and the simulated deviatoric stress index. As the simulated deviatoric stress index increases, the local a-value generally shows an increasing trend, indicating that the high a-value area corresponds to a more active loading area or damage activity area.

[0087] The above results indicate that the low b-value and high a-value anomalies extracted by statistical analysis are not random noise, but rather have a consistent mechanical significance with the stress concentration and enhanced damage activity reflected in the numerical simulation. Therefore, high-risk areas that simultaneously satisfy "low b-value and high a-value" and correspond to high-stress areas can be identified as mechanically consistent high-risk areas.

[0088] Based on the consistency verification, a stress consistency coefficient is further constructed to characterize the degree of consistency between statistical high-risk areas and high-stress areas.

[0089] The formula for calculating the stress consistency coefficient is: ; ; ; in, The stress consistency coefficient, This is the spatial overlap coefficient, used to characterize the overlap ratio between high-risk and high-stress areas; This is the distance consistency coefficient, used to characterize the degree of proximity between the center of a high-risk area and the center of a high-stress area; The intensity ranking consistency coefficient is used to characterize the degree of consistency between the intensity ranking of high-risk areas and the intensity ranking of high-stress areas. , , Let be the weight coefficient, and satisfy... + + =1, This is a high-risk area. This is a high-stress area. The center-to-center distance between high-risk and high-stress areas. This is the spatial decay scale.

[0090] If the stress consistency coefficient of a high-risk area is greater than a set threshold, the high-risk area is determined to be a physically consistent high-risk area, its high-risk level is retained, and an early warning is issued. If the stress consistency coefficient of a high-risk area is less than a set threshold, the high-risk area is determined to be a statistically abnormal area awaiting verification, and its risk level is reduced or further verification based on on-site engineering information is required.

[0091] The corrected formula for the statistical mine tremor risk index is: or ; in, and These are the statistical mine tremor risk indices before and after the correction. The consistency coefficient, For statistical risk weights.

[0092] In this application case, the early warning effect is as follows: Figure 7 As shown, on January 12, 2025, the TRI (Triplet Response Index) first exceeded the high-risk threshold TH, issuing a high-risk warning approximately 72 hours earlier than the ML2.65 high-energy mineral tremor that occurred on January 15, 2025. At the decision-making time on January 14, the TRI reached 0.73, maintaining a high-risk status. The spatial high-risk envelope range is... Figure 4 Planar a / b anomaly region and Figure 6The stress consistency verification results jointly determined that the epicenter of the subsequent high-energy mine earthquake was located within the aforementioned high-risk spatial range. This indicates that the method of the present invention can not only output the occurrence time of mine earthquake risk, but also identify potential energy release spatial areas, thereby providing a basis for directional stress relief, enhanced support, or local shutdown of mining.

[0093] In summary, the present invention has the following technical effects: 1. Enhanced Physical Interpretability: This invention introduces an independent physical consistency verification step, performing posterior validation between data-driven statistical results and rock mechanics models, thus providing a clear physical and mechanical basis for the early warning results. This method overcomes the shortcomings of traditional statistical methods, such as unclear physical meaning and the "black box" nature of integrated models, ensuring that every step of the early warning process has a clear physical meaning.

[0094] 2. Improve the reliability of early warnings: Through a sequential process of "multi-domain fusion - physical verification - risk correction," false statistical anomalies caused solely by data fluctuations without any actual physical basis can be effectively filtered out. Correcting the statistical mine tremor risk index using a consistency coefficient significantly reduces the false alarm rate. Simultaneously, the fusion of multi-dimensional information avoids the risk of missed reports caused by the instability of a single indicator, making the final early warning results more stable and reliable.

[0095] 3. More accurate precursor identification: Through comprehensive analysis of multi-dimensional information in time, depth and space, especially the quantitative identification of "scissor-like" coupling precursors that characterize stress concentration and damage acceleration, the complex evolutionary characteristics in the process of mine seismic gestation can be captured more comprehensively and accurately, thus improving the accuracy of precursor identification.

[0096] 4. The early warning information is more instructive: This invention can not only predict the time probability of risk occurrence by modifying the time series of the risk index, but also refine the delineation of high-risk spatial areas with physical significance through the results of spatial sub-indicators and stress verification, providing technical reference for mines to take targeted disaster prevention and mitigation measures such as local reinforcement of support and regional pressure relief.

[0097] This invention also provides a mine tremor risk early warning system, comprising: The statistical parameter calculation module calculates statistical parameters characterizing microseismic activity in the time domain, depth domain, and planar spatial domain, based on microseismic event data of the target area.

[0098] The risk index calculation module is used to construct a unified statistical mine tremor risk index based on multiple statistical parameters of different dimensions.

[0099] The physical consistency verification module is used to establish a numerical stress model of the target area to obtain its stress distribution field, and compare the high-risk areas indicated by the statistical seismic risk index with the high-stress areas in the stress distribution field to perform physical consistency verification.

[0100] The risk index correction module is used to generate a consistency coefficient based on the results of the physical consistency verification, and to use the consistency coefficient to correct the statistical mine tremor risk index to obtain the corrected risk index.

[0101] The risk warning module is used to issue warnings based on the modified risk index.

[0102] The modules in the above system are used to execute the various steps of the mine tremor risk early warning method. For a detailed description of the functions of each module, please refer to the description of the mine tremor risk early warning method in this invention.

[0103] 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.

[0104] 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.

[0105] 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.

[0106] Those skilled in the art will understand that all or part of the processes in 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. When executed, the computer program can include the processes of the embodiments described above. 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).

[0107] 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.

[0108] 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.

[0109] 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 mine tremor risks, characterized in that, include: Based on microseismic event data of the target area, statistical parameters characterizing microseismic activity are calculated in the time domain, depth domain, and planar spatial domain, respectively. A unified statistical mine tremor risk index is constructed based on multiple statistical parameters from different dimensions. A numerical stress model of the target area is established to obtain its stress distribution field, and the high-risk area indicated by the statistical mine seismic risk index is compared with the high-stress area in the stress distribution field to perform physical consistency verification. A consistency coefficient is generated based on the results of the physical consistency verification, and the statistical mine tremor risk index is corrected using the consistency coefficient to obtain a corrected risk index. Early warnings are issued based on the aforementioned modified risk index.

2. The method for early warning of mine tremor risk according to claim 1, characterized in that, The statistical parameters are the a-value and b-value of the Gutenberg-Richard relation.

3. The mine tremor risk early warning method according to claim 2, characterized in that, The steps for constructing the statistical mine tremor risk index include: Based on the changing trend of the statistical parameters in the time domain, a time sub-indicator is constructed; A depth sub-index is constructed based on the vertical gradient of the statistical parameters in the depth domain; Based on the spatial distribution of the statistical parameters in the planar spatial domain, a spatial sub-index is constructed; The time sub-indicator, the depth sub-indicator, and the spatial sub-indicator are linearly weighted and fused to obtain a statistical mine seismic risk index.

4. The mine tremor risk early warning method according to claim 3, characterized in that, The formula for constructing the time sub-indicator is as follows: ； in, For the aforementioned time sub-indicator, and These are the trend terms for the values ​​of a and b in the time domain, respectively. This is the normalization function; The formula for constructing the depth sub-index is as follows: ； in, The depth sub-index is... and These are the gradients of the values ​​a and b in the depth domain in the vertical direction, respectively. This is the normalization function; The formula for constructing the spatial sub-index is as follows: ； ； in, The spatial sub-index, For empirical quantile normalization, As an indicator of space hazard level, and These refer to the spatial distributions of the values ​​a and b on the planar spatial domain, respectively. This is a standardized function.

5. The mine tremor risk early warning method according to claim 3, characterized in that, The weighting coefficients of the linear weighted fusion are determined by the analytic hierarchy process (AHP).

6. The method for early warning of mine tremor risks according to claim 1, characterized in that, The results of the physical consistency verification include the overlap ratio, center distance, and intensity ranking consistency of the high-risk area and the high-stress area. The consistency coefficient is determined based on the overlap ratio, the center distance, and the intensity sorting consistency.

7. The method for early warning of mine tremor risk according to claim 1, characterized in that, The correction formula for the statistical mine tremor risk index is as follows: or ; in, and These are the statistical mine tremor risk indices before and after the correction, respectively. The consistency coefficient is... For statistical risk weights.

8. A mine tremor risk early warning system, characterized in that, include: The statistical parameter calculation module calculates statistical parameters characterizing microseismic activity in the time domain, depth domain, and planar spatial domain, based on microseismic event data of the target area. The risk index calculation module is used to construct a unified statistical mine tremor risk index based on multiple statistical parameters of different dimensions. The physical consistency verification module is used to establish a numerical stress model of the target area to obtain its stress distribution field, and compare the high-risk area indicated by the statistical mine seismic risk index with the high-stress area in the stress distribution field to perform physical consistency verification. The risk index correction module is used to generate a consistency coefficient based on the result of the physical consistency verification, and to use the consistency coefficient to correct the statistical mine tremor risk index to obtain a corrected risk index. The risk warning module is used to issue warnings based on the modified risk index.

9. 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 mine tremor risk early warning method according to any one of claims 1-7.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the mine tremor risk early warning method according to any one of claims 1-7.

Images