A method for advance blasting pressure relief based on rock burst prediction

Abstract

The application discloses a kind of ahead of time blasting pressure relief method based on rock burst prediction, which realizes the accurate identification and efficient intervention to rock burst risk by combining microseismic monitoring technology, energy thermal diagram modeling and intelligent algorithm.The method analyzes microseismic data in depth, identifies abnormal activity area, constructs energy thermal diagram, intuitively displays stress concentration situation, and predicts risk level combined with historical data, thereby improving the accuracy and response timeliness of early warning, uses density-based clustering algorithm to automatically identify the optimal blast hole position in high-risk area, and implements graded and directional blasting pressure relief according to risk level, effectively releases local stress and reduces the probability of rock burst.The overall process is highly automated, integrating monitoring, prediction, decision-making and intervention, and is suitable for various complex coal mine geological environments.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of rock burst prevention, in particular to a rock burst prediction-based advanced blasting pressure relief method. BACKGROUND

[0002] Rock burst is a common dynamic disaster in the process of coal mining, mainly occurring in deep and high-stress mining areas. Due to the sudden instability of rock strata, a large amount of energy is released instantaneously, causing coal and rock to violently burst out, accompanied by severe vibration, which seriously threatens the safety of miners and equipment. Its occurrence mechanism is complex and is affected by many factors such as geological structure, stress distribution and mining method. Prevention of rock burst requires the comprehensive use of monitoring and early warning, regional management and on-site control.

[0003] However, in the process of coal mining, the conventional monitoring method of rock burst relies on a single data source, lacks effective multi-dimensional data fusion means, resulting in low positioning accuracy of microseismic abnormal areas, and lacks in-depth analysis and fine processing of microseismic data, making it difficult to accurately identify potential risk areas of rock burst, limiting its application effect in high-precision early warning. SUMMARY

[0004] The purpose of the present application is to provide a rock burst prediction-based advanced blasting pressure relief method to solve the problems raised in the background.

[0005] To achieve the above purpose, the present application provides the following technical solution: a rock burst prediction-based advanced blasting pressure relief method, comprising the following steps:

[0006] S1, arranging microseismic detection nodes on the four structural surfaces of the roof, left side, right side and floor of the coal mine roadway to form a closed distribution around the cross section of the roadway and extend forward along the direction of roadway excavation, for collecting original microseismic data, optimizing the collected original microseismic data, and extracting short-time energy, spectral centroid, peak amplitude, signal kurtosis and duration to construct a multi-dimensional feature vector;

[0007] S2, based on the constructed feature vector, constructing an energy threshold judgment model for identifying microseismic abnormal areas, comparing and calibrating abnormal nodes with the set threshold, and calculating the suspected source location based on the energy weighting of abnormal node coordinates;

[0008] S3, according to the suspected source location and its energy value, using Gaussian diffusion function to simulate energy in three-dimensional space to construct an energy distribution thermal map for representing local stress concentration areas;

[0009] S4, adopt a spatio-temporal graph neural network risk prediction model to predict the microseismic risk probability at a future time, and dynamically construct an early warning level according to the predicted probability, the early warning level being divided into three levels of low risk, medium risk and high risk;

[0010] S5, when the early warning level is higher than the low risk, a DBSCAN clustering algorithm is used to automatically locate the main pressure relief hole based on the stress concentration area in the heat map, and a blasting hole group is arranged on both sides of the main pressure relief hole in a fan shape according to the direction of the coal mine roadway and the stress direction;

[0011] S6, the blasting hole group is first detonated, and after 30-60 minutes, the local cracked coal structure is broken, then the water jet device is placed in the main hole, and water pressure impact is performed for 10-20 minutes, the water jet pressure is 20-30 MPa, after the water pressure impact is completed, the blasting hole group is detonated again after 10-15 minutes, and the fracture zone is expanded;

[0012] S7, the microseismic data in the area after blasting is collected again, the local stress energy state is re-evaluated, if the energy decreases below the set baseline within 24 hours, it is determined that the intervention is successful, if it is determined that the intervention is unsuccessful, further supplementary blasting or adjustment of the hole arrangement is performed.

[0013] Preferably, in the step S1, the formula for extracting the short-time energy is: wherein, is the original signal, is the sampling point index in the sliding window, is the energy density of the signal at time ;

[0014] The algorithm for extracting the spectral centroid is: wherein, is the frequency variable, is the Fourier transform of the signal , is the power spectral density of the signal at frequency .

[0015] Preferably, in the step S1, when the original microseismic data is subjected to data optimization processing, the data is subjected to wavelet transform denoising processing, and the low-frequency noise below and the high-frequency interference above are filtered out by using a band-pass filter.

[0016] Preferably, in the step S2, the energy threshold is the historical average energy value plus twice the standard deviation, and the suspected source position is obtained by a weighted centroid positioning algorithm based on a plurality of energy anomaly nodes.

[0017] Preferably, in the step S3, the energy distribution heat map is generated by the following Gaussian diffusion function: wherein, is the source energy, is the source position coordinate, represents the target point to the square of the position distance of the first source point, is the diffusion radius.

[0018] Preferably, in the step S4, the method for constructing the spatio-temporal graph neural network risk prediction model is: wherein, is the current graph structure, is the weight matrix corresponding to the graph structure, is the historical time series feature, is the weight of the time series module, is the Sigmoid function, is the model bias term, used to adjust the baseline offset of the prediction result.

[0019] Preferably, the current graph structure is constructed by: taking a plurality of microseismic detection nodes as nodes of the graph, constructing an adjacency relationship based on the spatial distance between the nodes satisfying when the nodes and establish an edge; calculating the correlation weight based on the microseismic feature sequence in the historical time window to form a weighted adjacency matrix of the graph , and together with the node feature matrix to form the graph structure , the weight matrix corresponding to the current graph structure is a trainable parameter matrix for node feature transformation, acting on the node feature matrix .

[0020] Preferably, in the step S5, the DBSCAN clustering algorithm is: wherein, is the heat map point set, is the neighborhood radius, is the minimum point number.

[0021] Preferably, in the step S6, the blast hole group arranged on both sides of the main pressure relief hole has a diameter of 90mm and an included angle with the main pressure relief hole of ≥45°.

[0022] ​Preferably, in the step S6, the main pressure relief hole has a diameter of 150 mm and a depth of 10-14 m, and the distance between each hole is 3-5 m.

[0023] The present application realizes accurate positioning of the microseismic abnormal area by fusing microseismic data recognition, energy thermal map modeling, risk prediction and directional blasting intervention, and intuitively reflects the stress concentration situation through the energy distribution map, and predicts the risk level at the corresponding stress concentration position, thereby improving the accuracy and timeliness of early warning. In addition, the present application automatically locates the blasting hole based on the DBSCAN algorithm, implements hierarchical blasting intervention, effectively controls the rock burst risk, and has high automation degree of the overall process, and is suitable for early warning and management of rock burst in complex coal mine environment. BRIEF DESCRIPTION OF DRAWINGS

[0024] Figure 1 A flowchart of the advanced blasting pressure relief method based on rock burst prediction of the embodiments of the present application is shown in the figure.

[0025] Figure 2 A thermal map representing a local stress concentration area in the advanced blasting pressure relief method based on rock burst prediction of the embodiments of the present application. DETAILED DESCRIPTION

[0026] 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.

[0027] Please refer to Figure 1 The embodiments of the present application provide an advanced blasting pressure relief method based on rock burst prediction, which comprises the following steps:

[0028] S1, microseismic detection nodes are arranged on the four structural surfaces of the roof, left side, right side and floor of the coal mine roadway to form a closed distribution around the cross section of the roadway, and are arranged in the extension direction in front of the roadway, so that the distance between the nodes in each direction is 5 m (specifically, the distance between the roof and floor nodes in the vertical direction is 5 m, the distance between the left and right side nodes in the horizontal transverse direction is 5 m, and the distance along the longitudinal direction of the roadway is 5 m), forming a closed monitoring network. Each node is equipped with a wideband acceleration sensor (frequency response range 0.1 Hz-2 kHz) for collecting original microseismic data. The original microseismic data collected is optimized and processed, and the short-time energy, spectral centroid, peak amplitude, signal kurtosis and duration thereof are extracted to construct a multi-dimensional feature vector.

[0029] S11, in step S1, the formula for extracting short-time energy is: (Formula 1) Where, The original signal, For the index of the sampling points within the sliding window, For the signal at time The energy density is calculated using a sliding window method (window length 50ms, step size 25ms), and the energy of each window is calculated according to the short-time energy extraction formula. The algorithm for extracting the spectral centroid is as follows: (Formula 2) Where, For frequency variables, For signal Fourier transform, For the signal at frequency The power spectral density at that location.

[0030] In step S12, during the data optimization process of the original microseismic data, wavelet transform denoising is performed on the data, and bandpass filtering is used to filter out noise. The following low-frequency noise and The above high-frequency interference.

[0031] Furthermore, the multidimensional feature vector includes, but is not limited to, short-time energy, spectral centroid, peak amplitude, signal kurtosis, duration, and other feature indicators. By splicing features from multiple time windows, a fixed-length multidimensional feature vector is formed.

[0032] Specifically, in step S12, after the original microseismic data is denoised by wavelet transform (using db4 wavelet basis), low-frequency mechanical vibration and high-frequency electromagnetic interference are eliminated by a 10Hz-1kHz bandpass filter.

[0033] S2. Based on the constructed feature vector, an energy threshold judgment model is built to identify microseismic anomaly areas. Anomaly nodes are calibrated by comparing with a set threshold, and the suspected seismic source location is calculated based on the energy weighting of the anomaly node coordinates.

[0034] In step S21, the energy threshold is the historical average energy value plus twice the standard deviation, and the suspected epicenter location is obtained by weighted centroid localization algorithm on multiple energy anomaly nodes.

[0035] In the calculation of the source location using the weighted centroid localization algorithm, the energy values ​​of anomalous nodes are logarithmically weighted to reduce the weight bias of high-energy points.

[0036] S3. Based on the suspected epicenter location and its energy value, energy simulation is performed in three-dimensional space using the Gaussian diffusion function to construct an energy distribution heatmap to represent local stress concentration areas.

[0037] S31. In step S3, the energy distribution heatmap is generated using the following Gaussian diffusion function: ( Formula 3 ) Wherein, is the energy of the source, is the coordinate of the source position, represents the target point to the square of the position distance of the first source point, is the diffusion radius.

[0038] Referring to Figure 2 , a Gaussian diffusion function with a diffusion radius is selected to generate a three-dimensional energy heat map, and by adjusting to , the display range of the stress concentration area can be expanded, which is suitable for large-scale stress monitoring scenarios.

[0039] S4, adopt a spatio-temporal graph neural network risk prediction model to predict the microseismic risk probability at a future time, and dynamically construct an early warning level according to the predicted probability, the early warning level is divided into three levels of low risk, medium risk and high risk.

[0040] S41, in step S4, the spatio-temporal graph neural network includes a time convolution layer, a graph convolution layer and a full connection layer, a sliding window mechanism is used to input a graph sequence, a time convolution layer (TCN, convolution kernel size 5) is used to extract time sequence features, a graph convolution layer (GCN, adjacency matrix based on node spatial distance) is used to extract spatial correlation, and finally a full connection layer is used to fuse features and output a microseismic risk probability .

[0041] In this embodiment, the low risk: only records data without triggering intervention, the medium risk: starts local monitoring, and triggers a blasting pressure relief process within a predetermined time, and the high risk: immediately triggers a blasting pressure relief process.

[0042] S42, in step S4, the method for constructing the spatio-temporal graph neural network risk prediction model is: ( Formula 4 ) Wherein, is the current graph structure, is the weight matrix corresponding to the graph structure, is the historical time sequence feature, is the weight of the time sequence module, is a Sigmoid function, is a model bias term, used to adjust the baseline offset of the prediction result.

[0043] In this embodiment, a plurality of microseismic detection nodes arranged in a coal mine tunnel are used as graph nodes of the graph neural network, the spatial positions of the nodes can be obtained through actual arrangement coordinates, and according to the spatial distance between the nodes, a distance threshold is set If , a non-directed edge is established between node and node to reflect their spatial proximity relationship, and an initial adjacency matrix is obtained.

[0044] Further, to improve the modeling ability of the graph structure on the correlation of the monitoring data, the microseismic feature sequence of each node is extracted within a set time window, including but not limited to microseismic amplitude, energy, frequency and other indicators, and the Pearson correlation coefficient is used to calculate the feature similarity between each pair of nodes. The edges in the graph are weighted according to the similarity results, and a weighted adjacency matrix is formed. The specific Pearson correlation coefficient formula is: (5) Wherein, represents the microseismic feature sequence of node in the historical time window , and the adjacency matrix obtained by the above method and the corresponding node feature matrix form a graph structure together, which is used for subsequent risk prediction modeling using a spatio-temporal graph neural network model. The weight matrix corresponding to the graph structure is a learnable parameter matrix in the graph neural network, which is used for linear transformation of the node feature matrix , combined with the normalized adjacency matrix , to extract graph structure features through graph convolution operation. The specific graph convolution formula is: (6) Wherein, is an activation function, is the output graph structure encoding, which is used for subsequent time series modeling module. This method can effectively utilize the spatial structure information between microseismic monitoring points and improve the expression ability of the risk prediction model.

[0045] S5, when the warning level is higher than the low risk, the DBSCAN clustering algorithm is used to automatically locate the main pressure relief hole based on the stress concentration area in the heat map, and the blasting hole group is arranged on both sides of the main pressure relief hole in a fan shape according to the direction of the coal mine roadway and the stress direction.

[0046] S51, in step S5, the DBSCAN clustering algorithm is: (7) Wherein, is the point set of the heat map, is the neighborhood radius, is the minimum number of points.

[0047] S6, first detonate the blast hole group, detonate after 30-60 minutes, locally break the coal structure, then put the water jet device into the main hole, and carry out water pressure impact for 10-20 minutes, the water jet pressure is 20-30 MPa, after the water pressure impact is completed, the blast hole group is detonated again after 10-15 minutes, and the fracture zone is expanded.

[0048] S61, in step S6, the blast hole group arranged on both sides of the main pressure relief hole has a diameter of 90mm and an included angle of greater than or equal to 45 degrees with the main pressure relief hole.

[0049] S62, in step S6, the main pressure relief hole has a diameter of 150mm and a depth of 10-14m, and the distance between each hole is 3-5m.

[0050] S7, the microseismic data in the area after blasting is collected again, the local stress energy state is re-evaluated, if the energy decreases to below the set baseline within 24 hours, it is determined that the intervention is successful, if it is determined that the intervention is unsuccessful, further supplementary blasting or adjustment of the hole arrangement is carried out.

[0051] In another embodiment, in the hard rock geological condition, the K-Means clustering algorithm is used instead of the DBSCAN clustering algorithm, the number of clusters is set to 3, and the three main pressure relief hole positions of the stress concentration area are automatically located. When the blast hole group is arranged, the fan-shaped included angle on both sides of the main pressure relief hole is adjusted to 60 degrees, and the hole spacing is encrypted to 2m.

[0052] In summary: through the fusion of microseismic data identification, energy thermal diagram modeling, risk prediction and directional blasting intervention, the accurate positioning of the microseismic abnormal area is realized, the stress concentration situation is directly reflected through the energy distribution diagram, the risk level is predicted in the corresponding stress concentration position, and the accuracy and timeliness of the early warning are improved.

[0053] In addition, based on the DBSCAN algorithm, the blast hole position is automatically located, the hierarchical blasting intervention is implemented, the rock burst risk is effectively controlled, the overall process has high automation degree, and is suitable for the early warning and management of rock burst in complex coal mine environment.

[0054] 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 method for predicting rock burst based on advanced blasting pressure relief, characterized in that, The method comprises the following steps: S1, arranging microseismic detection nodes on four structural surfaces of the roof, left side, right side and floor of the coal mine tunnel, and extending and arranging them in the tunnel driving direction, for collecting original microseismic data, optimizing the collected original microseismic data, and extracting short-time energy, spectral centroid, peak amplitude, signal kurtosis and duration to construct a multi-dimensional feature vector; S2, based on the constructed feature vector, an energy threshold judgment model is constructed to identify microseismic abnormal areas, and the abnormal nodes are calibrated by comparing with the set threshold, and the suspected source position is calculated based on the energy weighted calculation of the abnormal node coordinates; In the step S2, the energy threshold is the average energy value plus twice the standard deviation, and the suspected source position is obtained by a weighted centroid positioning algorithm for multiple energy abnormal nodes; S3, according to the suspected source position and its energy value, a Gaussian diffusion function is used to simulate energy in a three-dimensional space to construct an energy distribution thermal map for representing a local stress concentration area; In the step S3, the energy distribution thermal map is generated by the following Gaussian diffusion function: wherein, is the source energy, is the source position coordinate, denotes the target point to the square of the position distance of the first source point, is the diffusion radius; S4, a spatio-temporal graph neural network risk prediction model is used to predict the microseismic risk probability at a future time, and an early warning level is dynamically constructed according to the predicted probability; S5, when the early warning level shows medium risk or high risk, a DBSCAN clustering algorithm is used to automatically locate the main pressure relief hole based on the stress concentration area in the thermal map, and a blasting hole group is arranged on both sides of the main pressure relief hole according to the direction of the coal mine tunnel and the stress direction, and the blasting hole group is arranged in a fan shape on both sides of the main pressure relief hole; S6, two blasting hole groups are arranged, the holes are subjected to water pressure impact after the first blasting, and the second blasting is performed after completion to expand the fracture zone; S7, microseismic data in the area after blasting is collected again, the local stress energy state is re-evaluated, if the energy decreases below the set baseline within 24 hours, it is determined that the intervention is successful, if it is determined that the intervention is not successful, further supplementary blasting or adjustment of the hole arrangement is performed.

2. The method of claim 1, wherein: In the step S1, when performing data optimization processing on the original microseismic data, the data is subjected to wavelet transform denoising processing, and the high-frequency interference is filtered out by using a band-pass filter The following low-frequency noise and The above high-frequency interference.

3. The method of claim 1, wherein: In the step S4, the method for constructing the spatio-temporal graph neural network risk prediction model is: wherein, is the current graph structure, is the weight matrix corresponding to the graph structure, is the historical time series feature, is the weight of the time series module, is the Sigmoid function, is the model bias term, used to adjust the baseline offset of the prediction result.

4. The method of claim 3, wherein: The current graph structure The construction method comprises the following steps: a plurality of microseismic detection nodes are taken as nodes of a graph, spatial distances between the nodes are calculated, and adjacency relationships are constructed based on the spatial distances. The adjacency relationships satisfy When the spatial distance is less than or equal to a preset threshold, an edge is established between the nodes. And ​ Calculating correlation weights based on microseismic feature sequences within a historical time window , , calculating correlation weights constituting a weighted adjacency matrix of a graph , and a node feature matrix constituting a graph structure , the current graph structure corresponding weight matrix is a trainable parameter matrix for node feature transformation, acting on the node feature matrix .

5. The method of claim 1, wherein: In the step S5, the DBSCAN clustering algorithm is: wherein, is a set of heat map points, is a neighborhood radius, is a minimum number of points.

6. The method of claim 1, wherein: In the step S6, the diameter of the blasting hole group arranged on both sides of the main pressure relief hole is 90mm, and the included angle with the main pressure relief hole is greater than or equal to 45°.

7. The method of claim 1, wherein: In the step S6, the diameter of the main pressure relief hole is 150mm, the depth is 10-14m, and the distance between each hole is 3-5m.

Images