Quantitative evaluation method for coal mine rock burst risk

Abstract

The invention discloses a coal mine rock burst risk quantitative evaluation method, and the method achieves the unified organization and expression of monitoring data, geological conditions and mining parameters through constructing a three-dimensional grid data model combining space coordinates and time information, provides a structured data basis for risk evaluation, and improves the risk evaluation efficiency. According to the method, multi-dimensional feature vector extraction and causal association graph modeling are combined, causal paths and strength among monitoring parameters, geological factors and mining conditions are described, it is ensured that an evaluation result has a traceable logic basis, historical time sequence data are utilized to train a dynamic causal evaluation model, conditional probability propagation is executed in a real-time monitoring stage, and a dynamic causal evaluation result is obtained. The danger probability value of each grid unit is dynamically output, the space migration and time evolution rule of danger along with mining disturbance can be reflected, and the defects that the result is static and updating is lagged in a traditional method are overcome.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of risk evaluation of coal mine rock burst, in particular to a coal mine rock burst risk quantitative evaluation method. BACKGROUND

[0002] Rock burst is one of the most dangerous dynamic disasters in the process of coal mining, and its occurrence mechanism is complex, which is controlled by the coupling of multiple factors such as geological structure, coal and rock physical properties, mining disturbance and stress evolution. However, the existing technology excessively relies on static geological parameters and qualitative experience indexes, lacks systematic integration of geological conditions, mining activities and real-time monitoring data, which makes it difficult to establish effective correlation logic between various parameters, at the same time, the evaluation process cannot reflect the dynamic evolution characteristics of mining stress, resulting in poor interpretability and weak adaptability of the evaluation results, which seriously restricts the accuracy and reliability of engineering application. SUMMARY

[0003] The purpose of the present application is to provide a coal mine rock burst risk quantitative evaluation method and system to solve the problems raised in the background technology.

[0004] According to one aspect of the present application, a coal mine rock burst risk quantitative evaluation method, the method comprising the following steps: S1, based on the geological exploration data and roadway layout of the target mining area, a three-dimensional grid data model is constructed along the strike, dip and normal of the coal seam, wherein the grid cell size is dynamically optimized and adjusted according to the roof rock thickness characteristic parameters, geological structure complexity and mining stress influence range of the unit location; S2, for each grid cell, the geological static parameters, mining technology parameters and real-time monitoring parameters are fused to form a multi-dimensional feature vector representing the rock burst risk of the unit, wherein the geological static parameters include coal seam impact dip index and roof and floor rock strength ratio, the mining technology parameters include coal pillar effective width and spatial relationship index with goaf, and the real-time monitoring parameters include microseismic energy release spatiotemporal gradient and stress change rate; S3, taking the multi-dimensional feature vector as the node attribute, and the spatial adjacency relationship and geomechanics connection between grid cells as the edge, a spatiotemporal causal graph model is constructed, and the model is trained using the monitoring data sequence before and after the historical rock burst event to learn the dynamic causal relationship between node features and rock burst risk; S4, the real-time monitoring data at the current time is mapped to the three-dimensional grid data model and input into the trained spatiotemporal causal graph model to obtain the rock burst risk probability value of each grid cell in the future set time period; S5, generating a rock burst danger level spatial distribution map of the whole mining area according to the danger probability values of each grid unit and combining a preset probability-level mapping relationship.

[0005] Preferably, the three-dimensional gridding data model is constructed, including: Based on the boundary coordinate range of the target mining area, x, y and z directions are divided into layers, and the size of the grid unit in each direction is dynamically adjusted according to the geological complexity parameter in the direction, The geological structure complexity is determined by the fault density, coal seam thickness variation coefficient and roof stratum thickness characteristic parameters near the position, and the size of the grid unit is: Wherein, is the boundary length of the target area in x, y and z directions, is the number of basic unit division, is the complexity adjustment factor of each direction, represents the geological complexity coefficient of the i,j,k layer in each direction, which is calculated by the fault density, coal seam thickness variation coefficient and roof stratum thickness characteristic parameters near the position.

[0006] Preferably, the multi-dimensional feature vector in step S2 further includes a rock burst danger state level evaluation comprehensive index Wt obtained by a comprehensive index method, and the calculation method is: Wherein, is the rock burst danger state level evaluation comprehensive index, is the influence degree of geological factors on rock burst, is the influence index of mining technology factors, and are obtained by weighting and normalizing the evaluation indexes of the selected multiple geological factors and mining technology factors, Wherein, is the evaluation index of the i-th geological factor, is the maximum possible value of the evaluation index, is the maximum possible value of the evaluation index, is the evaluation value corresponding to the j-th mining technology condition factor, is the maximum possible value of the factor.

[0007] Preferably, the comprehensive index Wt is input into the space-time causal diagram model as a key node attribute, which is used for coupling analysis with real-time monitoring parameters to correct the evaluation error caused by simple monitoring data fluctuation.

[0008] Preferably, the construction of the space-time causal diagram model in step S3 includes: S31, define each grid cell as a node, and the node attribute is the multi-dimensional feature vector of the cell; S32, based on the mechanism of stress transmission in rock strata, if two grid cells are adjacent in space and are in the same key rock strata or stress transmission path, a directed edge is established between them, and the direction of the edge represents the dominant direction of the principal stress transmission or energy accumulation; S33, use a structure learning algorithm combined with time series analysis to learn the causal relationship between node attribute changes and shock danger from historical data, and quantify the causal strength weight of each directed edge.

[0009] Preferably, in step S33, the stress transmission path in the key rock strata is determined by analyzing the spatio-temporal migration law of the historical microseismic event sequence, thereby optimizing the connection and direction of the directed edge in step S32.

[0010] Preferably, the node attribute in the spatio-temporal causal graph model includes the original three-dimensional stress parameters of the coal body determined based on the mining depth H; wherein H is the actual depth of the working face, γ is the gravity density of the rock strata, and μ is the Poisson's ratio, is the vertical stress, , is the horizontal stress.

[0011] Preferably, the node attribute also includes an elastic energy judgment parameter, which is calculated based on the total elastic energy U of the coal and rock mass, and the total elastic energy U is the sum of the volume deformation energy and the shape deformation energy. The elastic energy judgment parameter is used to represent the size relationship of the energy relative to the critical energy threshold: , , wherein E is the elastic modulus of the medium, and μ is the Poisson's ratio.

[0012] Preferably, in step S4, the danger probability value of each grid cell is output, which specifically includes the following steps: S41, map the microseismic event energy and location, the readings of each stress online monitoring point, and the electromagnetic radiation intensity value at the current time to the corresponding three-dimensional grid cell according to their spatial coordinates, and update the real-time monitoring parameter sequence in the node attribute vector of the cell; S42, use the updated node attribute as the observation evidence to perform probability reasoning on the spatio-temporal causal graph model, and use the confidence propagation algorithm to iteratively calculate the posterior probability of each node representing "being in a shock danger state" along the causal path defined by the directed edges in the model; S43, introduce a time decay factor λ (0 < λ < 1), and increase the initial prior probability of the grid node that has occurred a microseismic event with energy greater than a set threshold within a predetermined time according to its time proximity, to reflect the persistent effect of stress disturbance; S44. After a preset number of iterations converge, the posterior probability value calculated by each node is output as the rock burst hazard probability value of the grid cell.

[0013] Preferably, the step S5 of determining the hazard level according to the hazard probability value comprises the following steps: S51. A piecewise function is used to define the mapping relationship between the probability value P and the hazard level L, wherein a specific level is divided in a preset high hazard probability interval, when 0 < P≤ 0.25, it is determined as no hazard L0, when 0.25 < P≤ 0.5, it is determined as weak hazard L1, when 0.5 < P≤ 0.6, it is determined as moderate hazard L2, when 0.6 < P≤ 0.8, it is determined as strong hazard L3, when P > 0.8, it is determined as severe hazard L4; S52. For any grid cell determined as L3 or L4 level, the level of its spatial adjacent cell is checked, if the adjacent cell is L2 or above, the level of the cell is confirmed, if the adjacent cell is L1 or L0 level, the level of the cell is temporarily downgraded by one level and an artificial review flag is triggered to filter the misjudgment caused by transient abnormal monitoring data; S53. The final determined hazard level result is filled in the corresponding three-dimensional grid cell of the mining engineering plan with different colors or patterns to generate a rock burst hazard level spatial distribution map.

[0014] According to another aspect of the present application, an electronic device is provided, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor executes the program to implement the coal mine rock burst hazard quantitative evaluation method.

[0015] According to another aspect of the present application, a computer readable storage medium is provided, which stores a computer program, wherein the program is executed by a processor to implement the coal mine rock burst hazard quantitative evaluation method.

[0016] The present application aims at the defects of the existing rock burst hazard evaluation method, such as relying on static geological parameters, lacking causal logic support and insufficient dynamic evolution description, and proposes a quantitative evaluation method combining multi-dimensional spatio-temporal data modeling and causal reasoning. Through this method, the accuracy and verifiability of hazard identification can be significantly improved, providing reliable support for engineering safety under deep mining conditions.

[0017] Specifically, the present application is directed to the problem of relying on static geological parameters and qualitative experience determination in coal mine rock burst risk assessment, lacking causal logic support, and being difficult to dynamically reflect risk evolution. A quantitative evaluation method combining three-dimensional grid modeling and dynamic causal reasoning is proposed, which realizes the quantification and dynamic of risk identification, and improves the physical interpretation and engineering reliability of the evaluation results.

[0018] Specifically, the present application constructs a three-dimensional grid data model combining spatial coordinates and time information, realizes the unified organization and expression of monitoring data, geological conditions and mining parameters, provides a structured data basis for risk assessment, extracts multi-dimensional feature vectors and builds causal association graphs, describes the causal path and strength between monitoring parameters, geological factors and mining conditions, ensures that the evaluation results have traceable logical basis, trains a dynamic causal evaluation model using historical time series data, propagates conditional probability in real-time monitoring stage, and dynamically outputs the risk probability value of each grid unit, which can reflect the spatial migration and time evolution law of risk with mining disturbance, overcome the shortcomings of static results and update lag of traditional methods, and introduce three-dimensional stress parameters and elastic energy judgment parameters based on mining depth H in node attributes, so as to add coal and rock mechanics mechanism constraint on the basis of data driven, and improve the sensitive capture ability of sudden danger. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 The block diagram of the coal mine rock burst risk quantitative evaluation method of the embodiments of the present application. DETAILED DESCRIPTION

[0020] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0021] Please refer to Figure 1 According to one embodiment of the present application, a coal mine rock burst risk quantitative evaluation method is provided. Specifically, the method comprises the following steps S1-S5.

[0022] In step S1, based on the geological exploration data and roadway layout map of the target mining area, a three-dimensional grid data model is constructed along the strike, dip and normal of the coal seam, wherein the grid cell size is dynamically optimized and adjusted according to the roof rock thickness characteristic parameter, geological structure complexity and mining stress influence range of the unit location. The multi-dimensional space-time data set includes spatial coordinates and time label information.

[0023] According to the technical scheme, the three-dimensional gridding data model can unify the spatial boundary information and the geological complexity information of the mining area into a standardized unit structure, thereby providing a basis support for subsequent spatial mapping of monitoring data and risk characteristics.

[0024] In one embodiment, specifically, the three-dimensional gridding data model is constructed, including: Based on the boundary coordinate range of the target mining area, x, y and z directions are divided into layers, and the grid unit size of each direction is dynamically adjusted according to the geological complexity parameter in the direction, and the geological structure complexity is determined by the fault density, the coal seam thickness variation coefficient and the roof stratum thickness characteristic parameter near the position, The grid unit size is: Wherein, is the boundary length of the target area in the x, y and z directions, is the number of basic units, is the complexity adjustment factor of each direction, represents the geological complexity coefficient of the i,j,k layer in each direction, which is calculated by the fault density, the coal seam thickness variation coefficient and the roof stratum thickness characteristic parameter near the position.

[0025] In step S2, for each grid unit, the geological static parameters, the mining technology parameters and the real-time monitoring parameters are fused to form a multi-dimensional feature vector representing the impact risk of the unit; the geological static parameters include the coal seam impact tendency index and the roof and floor lithology strength ratio; the mining technology parameters include the coal pillar effective width and the spatial relationship index with the goaf; the real-time monitoring parameters include the microseismic energy release space-time gradient and the stress change rate.

[0026] Geological static parameters are determined at the mine development and stope design stage and are relatively stable, being the "inherent" basis of rockburst hazard. Specifically, the coal seam impact liability index is an inherent attribute index obtained by testing the mechanical properties of field-drilled raw coal samples in the laboratory. According to national or industry standards (e.g. "Method for Determining Coal Seam Impact Liability Classification and Index"), uniaxial compression tests are performed on the coal samples to determine key indicators such as dynamic failure time (DT), elastic energy index (WET), and impact energy index (KE). Through weighted calculation or comprehensive judgment, the coal seam is finally classified into grades such as no impact liability, weak impact liability, medium impact liability, and strong impact liability, and is assigned a corresponding quantitative index. The roof-to-floor lithology strength ratio is based on core sample analysis and geophysical logging data from geological exploration boreholes. First, the uniaxial compressive strength of the immediate roof and floor rock is obtained through laboratory experiments. Then, the ratio of roof strength to floor strength is calculated. A higher ratio (i.e. a hard roof relative to a soft floor) means that the roof is less likely to collapse during mining, and is more likely to accumulate a large amount of bending elastic energy, which is suddenly released upon breaking, causing strong disturbance to the coal seam and significantly increasing the rockburst hazard.

[0027] Mining technical parameters are dynamically changed with the progress of mining activities, reflecting the degree of human engineering activities on the stress environment of surrounding rock, and are key factors inducing rockburst. Specifically, the effective width of coal pillars is based on the mine mining engineering plan and actual measurement data, and the coal pillars are left between the goaf edges or roadways, whose stability is closely related to the width. By measuring or obtaining the actual geometric width of the coal pillar from the design drawings, and combining with the height to calculate the width-height ratio. When the effective width (or width-height ratio) of the coal pillar is too small, the coal pillar itself will enter a plastic failure state due to bearing too high concentrated stress, lose the supporting capacity, and become a high stress area and a potential rockburst trigger source. The goaf spatial relationship index is a relative position index obtained by spatial analysis of the three-dimensional mining model or mining engineering drawings. This index is used to quantitatively evaluate the relative position relationship between the current grid cell and the adjacent goaf. Generally, the closer the unit is to the edge of the goaf, the higher the hazard. A quantitative model can be established, for example: the distance from the unit to the nearest goaf boundary, whether the unit is located in the lateral support pressure influence area of the goaf, and whether it is in the multiple goaf superimposed influence area, etc. are comprehensively scored to generate a spatial relationship index. The higher the index, the greater the influence of stress concentration caused by the goaf.

[0028] Real-time monitoring parameters are obtained by a downhole monitoring system in real time, directly reflecting the current activity state and energy dynamics of the surrounding rock system, and are the direct basis for dynamic early warning. Specifically, the spatio-temporal gradient of microseismic energy release is continuously recorded by the microseismic monitoring system arranged in the well. The system first locates the occurrence position (space) and occurrence time (time) of each microseismic event, and calculates the energy released thereby. The calculation of the spatio-temporal gradient contains two layers of meanings. First, in a certain time window (such as 24 hours), it is evaluated whether the total energy of microseismic events or the number of large energy events in a certain range around the current grid cell appears a significant aggregation, and second, it is compared with the total energy of microseismic in the previous window to calculate the energy change rate. A rapidly rising gradient (i.e., a sharp increase in microseismic activity in time and space) is an important precursor to the intensification of surrounding rock rupture, stress readjustment, and possible approach to the critical state of instability.

[0029] The stress change rate is measured in real time by stress monitoring sensors (such as vibrating wire stress meters, grating sensors, etc.) installed in the roadway surrounding rock or deep in the borehole. The sensor directly monitors the absolute value or relative change of stress inside the coal and rock mass. The stress change rate refers to the increase or decrease rate of stress value per unit time. A continuously and rapidly increasing stress change rate indicates that the region is rapidly accumulating elastic energy, the stress concentration degree is intensifying, and the impact risk is significantly increasing. Conversely, a sudden decrease in stress may indicate that the energy has been released or local damage has occurred.

[0030] Finally, the above six types (or more) of parameters are standardized to eliminate the dimensional influence, and are sequentially combined into a multi-dimensional feature vector, such as [impact tendency index, roof-to-floor strength ratio, effective coal pillar width, spatial relationship index, microseismic energy gradient, stress change rate]. The vector is input into a machine learning algorithm or a comprehensive index evaluation model, and through weighting, clustering or deep learning, etc., the impact risk level of each grid cell is finally output.

[0031] According to the above technical solution, the multi-dimensional feature vector can uniformly encapsulate static geological parameters, mining technical parameters and dynamic monitoring parameters in the node attribute, thereby ensuring the completeness and comparability of the input of the subsequent causal model.

[0032] In one embodiment, specifically, the multi-dimensional feature vector in step S2 further includes an impact ground pressure risk state level evaluation comprehensive index W obtained by a comprehensive index method t , and the calculation method is: wherein, W is the impact ground pressure risk state level evaluation comprehensive index, is the influence degree of geological factors on impact ground pressure, is the mining technology factor influence index, and respectively weighted and normalized by the evaluation indexes of the selected geological factors and mining technical factors, wherein, is the evaluation index of the jth geological factor, is the maximum possible value of the evaluation index, is the evaluation value corresponding to the jth mining technical factor, is the maximum possible value of the factor.

[0033] In one embodiment, specifically, the comprehensive index Wt t As a key node attribute input into the spatio-temporal causal graph model, it is used for coupling analysis with real-time monitoring parameters to correct the evaluation error caused by simple fluctuations in monitoring data.

[0034] According to the above technical solution, the comprehensive index Wt can be input into the causal model as part of the node attribute, which is used to reflect the weighted risk level of multiple factors, and provides a stable reference quantity for risk discrimination under monitoring data fluctuation conditions.

[0035] In step S3, a multi-dimensional feature vector is used as a node attribute, and a spatial adjacency relationship and a geomechanical connection between grid cells are used as edges to construct a spatio-temporal causal graph model, and historical monitoring data sequences before and after the rock burst event are used to train the model to learn the dynamic causal relationship between node features and rock burst risk.

[0036] According to the above technical solution, the feature vector is used as the node attribute, and the geomechanical connection is used to establish the causal graph model, which can make the model structure consistent with the physical action process of the coal and rock mass, thereby ensuring that the causal reasoning has engineering interpretation.

[0037] In one embodiment, specifically, the construction of the spatio-temporal causal graph model in step S3 includes: S31, defining each grid cell as a node, and the node attribute is a multi-dimensional feature vector of the cell.

[0038] According to the above technical solution, the grid cell is defined as a node and the feature vector is used as the node attribute, which can ensure that the model granularity is consistent with the spatial resolution of the monitoring data, thereby facilitating the positioning of the dangerous unit in the spatial dimension.

[0039] S32, based on the mechanism of rock stratum stress transmission, if two grid cells are adjacent in space and are in the same key rock stratum or stress transmission path, a directed edge is established between them, and the direction of the edge represents the dominant direction of principal stress transmission or energy accumulation.

[0040] ​According to the technical solution, the directed edge is established based on a stress transmission mechanism of a rock stratum, can reflect a mechanical action direction between adjacent units, and provides a path structure in line with physical logic for a propagation calculation of a danger probability.

[0041] In S33, a structure learning algorithm combined with time series analysis is used to learn a causal relationship between a node attribute change and a shock danger from historical data, and to quantify a causal strength weight of each directed edge.

[0042] According to the technical solution, the causal relationship between the node attribute change and the danger is obtained through time series structure learning, which can automatically calibrate the causal strength under the driving of historical data, thereby improving adaptability to future risk evolution.

[0043] In one embodiment, specifically, in S33, a stress transmission path in a key rock stratum is determined by analyzing a spatiotemporal migration law of a historical microseismic event sequence, thereby optimizing connection and direction of the directed edge in S32.

[0044] According to the technical solution, the connection and direction of the edge are optimized by using the migration law of the historical microseismic event, which can ensure that the causal graph is more in line with actual stress propagation characteristics in structure, thereby enhancing accuracy of the causal relationship.

[0045] In one embodiment, specifically, a node attribute in the spatiotemporal causal graph model includes a coal body original three-dimensional stress parameter determined based on a mining depth H. wherein the mining depth H is an actual buried depth of a working face, the gamma is a rock stratum gravity density, and the mu is a Poisson's ratio, is a vertical stress, 、 is a horizontal stress.

[0046] According to the technical solution, the three-dimensional stress parameter based on the mining depth H is used as the node attribute, which can directly map a coal body original stress level to a node in the causal model, thereby enabling the model to capture an influence of a buried depth change on the danger.

[0047] In one embodiment, specifically, the node attribute further includes an elastic energy judgment parameter, and the mining depth H satisfies the following formula: wherein the Rc is a coal body uniaxial compressive strength, is a coefficient greater than 1.

[0048] According to the technical solution, the elastic energy judgment parameter in the node attribute is used to represent an elastic potential energy level accumulated in the medium under the condition of the mining depth H.

[0049] The node attribute further includes an elastic energy determination parameter, the elastic energy determination parameter is calculated based on total elastic energy U of the coal rock mass, the total elastic energy U is the sum of volume deformation energy and shape deformation energy, and the elastic energy determination parameter is used to represent the size relationship of the energy relative to a critical energy threshold value: , wherein E is the elastic modulus of the medium, μ is the Poisson's ratio, σ1, σ2, σ3 are three-directional stresses borne by the medium, and are determined by the mining depth H and the gravity density γ of the rock stratum, i.e.: The total elastic energy U is: The elastic energy determination parameter is a normalized value of U or a comparison result with the critical energy threshold value. When the mining depth H satisfies the following condition: It is considered that the elastic energy at the position of the node has reached or exceeded a critical state, and the determination parameter is marked as "high risk", which is an important evidence input for enhancing the impact danger.

[0050] In step S4, the real-time monitoring data at the current time is mapped to the three-dimensional gridding data model and input into the trained spatiotemporal causal graph model to obtain a dangerous probability value of each grid element in a future set time period.

[0051] According to the above technical solution, after the real-time monitoring data is mapped to the three-dimensional grid element and input into the causal model, the coupling of the spatiotemporal dynamic state and the causal relationship model can be realized, thereby supporting real-time updating of the dangerous probability.

[0052] In one embodiment, specifically, the dangerous probability value of each grid element is output in step S4, and specifically includes the following steps: S41, the energy and position of the microseismic event at the current time, the readings of each stress online monitoring point, and the electromagnetic radiation intensity value are mapped to the corresponding three-dimensional grid element according to the spatial coordinates, and the real-time monitoring parameter sequence in the node attribute vector of the element is updated.

[0053] According to the above technical solution, the real-time monitoring data is mapped and updated to the corresponding grid element node attribute, which can ensure that each node reflects the current physical state, thereby improving the immediacy of subsequent probability reasoning.

[0054] S42, taking the updated node attribute as observation evidence, performing probability reasoning on the spatiotemporal causal graph model, the probability reasoning adopts a belief propagation algorithm, and iteratively calculates the posterior probability of each node representing "being in an impact danger state" along the causal path defined by the directed edges in the model.

[0055] According to the technical solution, the node observation evidence is propagated along the directed edges based on the updated node attributes to perform the probabilistic inference on the causal graph, so that the posterior probability of each unit being in a dangerous state is obtained.

[0056] In step S43, a time decay factor λ (0<λ<1) is introduced, and the initial prior probability of a grid node in which a microseismic event with energy greater than a set threshold occurs within a predetermined time is increased according to the time proximity, so as to reflect the sustained effect of stress disturbance.

[0057] According to the technical solution, the time decay factor is introduced to weight the influence of the microseismic event, so that the dangerous inference result is decayed over time, thereby being more consistent with the physical process of gradual dissipation of stress disturbance.

[0058] In step S44, after the preset number of iterations converges, the posterior probability value of each node calculated is output as the rock burst dangerous probability value of the grid unit.

[0059] According to the technical solution, the posterior probability value of each node is output after the iteration converges, so that a spatially distributed probability result is formed, thereby providing a quantitative basis for subsequent grade division.

[0060] In step S5, according to the dangerous probability value of each grid unit, a rock burst dangerous grade spatial distribution map of the entire mining area is generated in combination with a preset probability-grade mapping relationship.

[0061] According to the technical solution, the dangerous grade distribution map is generated according to the dangerous probability value, so that the quantitative probability is corresponded to the visual spatial result, thereby facilitating intuitive identification of a high-danger area on site.

[0062] In one embodiment, specifically, the step of determining the dangerous grade according to the dangerous probability value in step S5 includes the following steps: In step S51, a piecewise function is used to define the mapping relationship between the probability value P and the dangerous grade L, wherein a specific grade is divided in a preset high-danger probability interval, when 0<P≤0.25, it is determined as no danger L0, when 0.25<P≤0.5, it is determined as weak danger L1, when 0.5<P≤0.6, it is determined as medium danger L2, when 0.6<P≤0.8, it is determined as strong danger L3, when P>0.8, it is determined as severe danger L4. S52, for any grid cell determined as L3 or L4 level, check its spatial adjacent cell level, if its adjacent cell is L2 and above level, confirm the cell level, if its adjacent cell is L1 or L0 level, temporarily downgrade the cell level by one level and trigger the manual review flag to filter the false positive caused by the transient abnormal monitoring data; S53, fill the final determined danger level result in different color or pattern in the corresponding three-dimensional grid cell of the mining engineering plan to generate the spatial distribution map of rock burst danger level.

[0063] The parts not involved in the present application are the same as or can be realized by the prior art. Although the embodiments of the present application have been shown and described, it can be understood by those skilled in the art that various changes, modifications, replacements and variations can be made to the embodiments without departing from the principles and spirits of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A quantitative evaluation method for the risk of rockburst in coal mines, characterized in that, The method includes the following steps: S1. Based on the geological exploration data and roadway layout map of the target mining area, a three-dimensional gridded data model is constructed along the strike, dip and normal of the coal seam. The size of the grid cell is dynamically optimized and adjusted according to the characteristic parameters of the roof strata thickness, the complexity of the geological structure and the range of influence of mining stress at the location of the cell. S2. For each grid cell, geological static parameters, mining technical parameters, and real-time monitoring parameters are integrated and processed to form a multi-dimensional feature vector characterizing the impact hazard of the cell. The geological static parameters include the coal seam impact tendency index and the lithological strength ratio of the roof and floor. The mining technical parameters include the effective width of the coal pillar and the spatial relationship index with the goaf. The real-time monitoring parameters include the spatiotemporal gradient of microseismic energy release and the stress change rate. S3. Using multidimensional feature vectors as node attributes and spatial adjacency relationships and geomechanical connections between grid cells as edges, a spatiotemporal causal graph model is constructed. The spatiotemporal causal graph model is trained using monitoring data sequences before and after historical rockburst events to learn the dynamic causal relationship between node characteristics and rockburst risk. S4. Map the real-time monitoring data of the current moment to the three-dimensional gridded data model, input it into the trained spatiotemporal causal graph model, and obtain the risk probability value of rockburst occurring in each grid cell within a set time period in the future. S5. Based on the hazard probability value of each grid unit and combined with the preset probability-level mapping relationship, generate a spatial distribution map of the rockburst hazard level of the entire mining area.

2. The method according to claim 1, characterized in that, in, Constructing the three-dimensional meshed data model includes: Based on the boundary coordinate range of the target mining area, the area is divided into three layers in the x, y, and z directions. The grid cell size in each direction is dynamically adjusted according to the geological complexity parameter in that direction. The complexity of the geological structure is determined by a combination of factors, including the fault density, coal seam thickness variation coefficient, and roof strata thickness characteristics in the vicinity of the location. The grid cell size is: in, Let x be the boundary length of the target region in the x, y, and z directions. The number of basic unit divisions, These are complexity adjustment factors for each direction. This represents the geological complexity coefficient of the i, j, k layers in each direction. Its value is calculated by combining the fault density, coal seam thickness variation coefficient, and roof strata thickness characteristic parameters near that location.

3. The method according to claim 1, characterized in that, The multidimensional feature vector mentioned in step S2 also includes the comprehensive index W for assessing the risk level of rockburst, obtained through the comprehensive index method. t The calculation method is as follows: in, The comprehensive index for assessing the risk level of rockburst. The degree of influence of geological factors on rockburst. The index is influenced by mining technology factors. and The results were obtained by weighted normalization of the evaluation indices of the selected geological and mining technology factors. in, The first influence of geological factors Evaluation index, This represents the maximum probable value for the assessment item, and its maximum probable value is the upper limit of the geological factor obtained from historical monitoring data. This represents the evaluation value corresponding to the j-th mining technology condition factor. This represents the maximum possible value for this factor.

4. The method according to claim 3, characterized in that, The comprehensive index W t As key node attributes, they are input into the spatiotemporal causal graph model for coupling analysis with real-time monitoring parameters to correct evaluation errors caused solely by fluctuations in monitoring data.

5. The method according to claim 1, characterized in that, The construction of the spatiotemporal causal graph model in step S3 includes: S31. Define each grid cell as a node, and the node attribute is the multidimensional feature vector of that cell. S32. Based on the mechanism of stress transmission in rock strata, if two grid cells are spatially adjacent and located in the same key rock strata or stress transmission path, a directed edge is established between them, and the direction of the edge represents the dominant direction of principal stress transmission or energy accumulation. S33. A structure learning algorithm combining time series analysis is adopted to learn the causal relationship between node attribute changes and impact risk from historical data, and to quantify the causal strength weight of each directed edge.

6. The method according to claim 5, characterized in that, In step S33, by analyzing the spatiotemporal migration patterns of historical microseismic event sequences, the stress transmission paths in key rock strata are determined, thereby optimizing the connection and direction of the directed edges in step S32.

7. The method according to claim 5, characterized in that, The node attributes in the spatiotemporal causal graph model include the original triaxial stress parameters of the coal body determined based on the mining depth H; Where H is the actual burial depth of the working face, γ is the density of the rock strata, and μ is Poisson's ratio. For vertical stress, , It is horizontal stress.

8. The method according to claim 7, characterized in that, The node attributes further include an elastic energy determination parameter, which is calculated based on the total elastic energy U of the coal and rock mass. The total elastic energy U is the sum of the volume deformation energy and the shape deformation energy. The elastic energy determination parameter is used to characterize the magnitude relationship of this energy relative to the critical energy threshold: , , Where E is the elastic modulus of the media, and μ is Poisson's ratio.

9. The method according to claim 1, characterized in that, The output of the hazard probability values of each grid cell in step S4 specifically includes the following steps: S41. Map the energy and location of the microseismic event, the readings of each stress online monitoring point, and the electromagnetic radiation intensity value at the current moment to the corresponding three-dimensional grid cells according to their spatial coordinates, and update the real-time monitoring parameter sequence in the node attribute vector of the cell; S42. Using the updated node attributes as observed evidence, perform probability inference on the spatio-temporal causal graph model. The probability inference uses the belief propagation algorithm and iteratively calculates the posterior probability that each node represents "being in a rock burst hazard state" along the causal path defined by the directed edges in the model; S43. Introduce a time decay factor λ, and for the grid nodes where microseismic events with energy greater than the set threshold have occurred within a predetermined time, increase their initial prior probability according to their time distance to reflect the continuous effect of stress disturbance; S44. After the preset number of iterative convergences, output the posterior probability value calculated for each node as the rock burst hazard probability value of the grid cell.

10. The method according to claim 1, characterized in that, The determination of the hazard level according to the hazard probability value in step S5 includes the following steps: S51. Define the mapping relationship between the probability value P and the hazard level L using a piecewise function, where specific levels are divided in the preset high hazard probability interval. When 0 < P ≤ 0.25, it is determined as no hazard L0. When 0.25 < P ≤ 0.5, it is determined as weak hazard L1. When 0.5 < P ≤ 0.6, it is determined as medium hazard L2. When 0.6 < P ≤ 0.8, it is determined as strong hazard L3. When P > 0.8, it is determined as severe hazard L4. S52. For any grid cell determined to be at level L3 or L4, check the levels of its spatially adjacent cells. If its adjacent cells are at level L2 or above, confirm the level of the cell. If its adjacent cells are at level L1 or L0, temporarily lower the level of the cell by one level and trigger the manual review flag to filter out misjudgments caused by instantaneous anomalies in monitoring data; S53. Fill the corresponding three-dimensional grid cells on the mining engineering plan with the finally determined hazard level results in different colors or patterns to generate a spatial distribution map of the rock burst hazard level.

Images