A close coal seam mine pressure analysis method, electronic equipment and storage medium

Abstract

The present application belongs to the technical field of coal mines, and particularly relates to a close-range coal seam mine pressure analysis method, an electronic device and a storage medium. The method comprises: collecting first time sequence data in the current working face mining process; obtaining second time sequence data containing roof laser radar point cloud data; obtaining a data segment affected by the current working face operation in the first time sequence data and performing feature fusion as third time sequence data; obtaining the relative distance between the working face operation position and the close-range coal seam operation position corresponding to the data in the third time sequence data; performing feature extraction on the second time sequence data according to the time stamp of the third time sequence data to obtain the roof change feature; constructing a vector sequence; inputting the vector sequence into a self-attention neural network to obtain the contribution weight of close-range coal seam mining to the current working face roof deformation.

Description

Technical Field

[0001] This invention belongs to the field of coal mining technology, specifically relating to a method for close-range coal seam pressure analysis, electronic equipment, and storage medium. Background Technology

[0002] During the mining of closely spaced coal seams, the movement of the upper or lower coal seams can significantly alter the stress distribution of adjacent coal seams, leading to abnormal responses such as roof deformation, mine pressure manifestation, and roof delamination at the current working face. In severe cases, this can trigger dynamic disasters such as roof falls and roof collapses. Therefore, accurately analyzing the impact of the movement of closely spaced coal seams on the stability of the current working face roof is of great significance for safe and efficient mining.

[0003] Currently, there are various methods for monitoring and analyzing mine pressure, such as monitoring coal and rock mass fractures by deploying microseismic sensors, monitoring roof pressure using the working resistance of hydraulic supports, and observing internal movement of rock strata using roof delamination instruments. These methods typically rely on a single data source or simple threshold judgments, making it difficult to achieve comprehensive analysis and quantitative attribution of multi-source heterogeneous data. Summary of the Invention

[0004] One object of the present invention is to provide a method, electronic device and storage medium for close-range coal seam pressure analysis, which can solve the above-mentioned technical problems in the prior art.

[0005] According to a first aspect of the present invention, a method for close-range coal seam pressure analysis is provided, comprising the following steps:

[0006] Collect the first time-series data during the current mining process at the working face. The first time-series data includes vibration data, mine pressure data, and delamination data.

[0007] The top plate is periodically scanned using lidar to obtain second time-series data containing lidar point cloud data of the top plate;

[0008] Based on the current work trajectory and the timestamp corresponding to the second time series data, the data segments within the first time series data affected by the current work operation are obtained and their features are fused to form the third time series data.

[0009] Based on the timestamp of the third time series data and the collected operation records of the nearby coal seam, the relative distance between the working face operation position and the nearby coal seam operation position corresponding to the data in the third time series data is obtained.

[0010] Based on the timestamp of the third time series data, feature extraction is performed on the second time series data to obtain the roof change features;

[0011] A vector sequence is constructed based on the third time-series data, the relative distance, and the roof change characteristics;

[0012] The vector sequence is input into the self-attention neural network to obtain the contribution weight of the nearby coal seam mining to the current working face roof deformation.

[0013] Optionally, the extraction process of the roof change features includes:

[0014] The second time series data of the current time is registered and compared with the data of the previous period in three dimensions to obtain the point cloud difference clusters on the roof surface caused by coal and rock spalling.

[0015] Spatial clustering analysis of point cloud difference clusters is used to obtain information on the exfoliated coal seam, and roof change characteristics are obtained based on the information on the exfoliated coal and rock.

[0016] Optionally, obtaining the roof variation features based on point cloud difference clusters specifically includes:

[0017] Spatial clustering of point cloud difference clusters yields multiple discrete point clusters;

[0018] Each point cluster is identified as exfoliated coal and rock, and the three-dimensional geometric dimensions of each exfoliated coal and rock are calculated based on the shape of each point cluster.

[0019] Based on the preset particle size range, the roof variation characteristics are obtained according to the quantity, volume, or projected area distribution of the exfoliated coal and rock.

[0020] Optionally, the roof deformation characteristics also include local deflection rate, non-uniform settlement coefficient, and roof surface roughness.

[0021] Optionally, the self-attention network uses the current working face's operational status features as a query vector, the distance to the nearest coal seam and the roof status features as key vectors, and the roof change features as value vectors to obtain the impact of the nearest coal seam operation on the current working face.

[0022] Optionally, a one-dimensional convolutional neural network can be used to fuse the first time series data to obtain the third time series data.

[0023] Optionally, the process of acquiring the third time-series data includes:

[0024] Based on the position coordinates of the coal mining machine during the mining process and the first distance threshold, the first time series data is filtered to remove time series data with working positions greater than a preset distance.

[0025] Optionally, the first time-series data may further include at least one of the following: ambient temperature and humidity, support stress, or coal mining machine load data.

[0026] According to a second aspect of the present invention, an electronic device is provided, comprising a processor and a memory, the memory storing a program or instructions executable on the processor, the program or instructions, when executed by the processor, implementing the steps of a near-field coal seam pressure analysis method as described in the first aspect of the present invention.

[0027] According to a third aspect of the present invention, a readable storage medium is provided, on which a program or instructions are stored, which, when executed by a processor, implement the steps of a near-field coal seam pressure analysis method as described in the first aspect of the present invention.

[0028] The beneficial effects of this invention are as follows: By employing strict spatiotemporal filtering rules, this invention eliminates a large amount of irrelevant background data, allowing the model to focus on signals directly influenced by mining activities, resulting in more accurate analysis results. It cleverly solves the alignment problem between high-frequency sensor data and low-frequency point cloud data, unifying the causal chain between mining behavior and roof deformation in time. By using a self-attention network, it can automatically learn and quantify the contribution weight of specific nearby coal seam mining activities to roof deformation from spatiotemporal sequences, providing a quantitative basis for mine pressure control and safety decisions. Simultaneously, the applicable scope of the method is clearly defined, applicable only to a single nearby coal seam, avoiding model misuse and performance degradation under complex geological conditions, and improving the reliability of industrial applications. Attached Figure Description

[0029] Figure 1 This is a flowchart of a method for near-distance coal seam pressure analysis in an embodiment of the present invention. Detailed Implementation

[0030] Various exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be noted that, unless otherwise specifically stated, the relative arrangement, numerical expressions, and values ​​of the components and steps set forth in these embodiments do not limit the scope of the invention.

[0031] The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the invention or its application or use.

[0032] Techniques, methods, and apparatus known to those skilled in the art may not be discussed in detail, but where appropriate, such techniques, methods, and apparatus should be considered part of the specification. In all examples shown and discussed herein, any specific values ​​should be interpreted as merely exemplary and not as limitations. Therefore, other examples of exemplary embodiments may have different values.

[0033] It should be noted that similar labels and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be discussed further in subsequent figures.

[0034] In the specification of this invention, the terms "first" and "second" may explicitly or implicitly include one or more of the same feature. In the description of this invention, unless otherwise stated, "multiple" means two or more. Furthermore, in the specification, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0035] like Figure 1 As shown in the figure, this embodiment introduces a method for near-distance coal seam pressure analysis, including steps 1100-1700.

[0036] Step 1100: Collect the first time-series data during the current working face mining process. The first time-series data includes vibration data, mine pressure data, and delamination data.

[0037] In this application, the current working face refers to the working face that is being mined and is used as the monitoring object. Various sensors are installed in the stope and roadways of this working face to collect vibration data, mine pressure data, and delamination data.

[0038] Vibration data can be acquired using microseismic sensors. Vibration data reflects the elastic waves released when coal and rock masses fracture, break, or slip under mining stress. Essentially, it is the transient energy release generated by the propagation of cracks, frictional slip, or sudden fracture within the rock mass. When the cutting of a coal mining machine or disturbance from nearby coal seam mining causes stress concentration in the overlying strata or coal pillar on the current working face, exceeding the rock mass strength, the rock mass will develop micro-fractures or even large-scale fractures. These fracturing processes propagate outwards in the form of elastic waves, which are picked up by vibration sensors.

[0039] The parameters of the seismic data include event time, hypocenter location, released energy, and frequency spectrum. The hypocenter location is used to pinpoint the spatial coordinates of the rupture and determine if it occurred within the mining area's influence zone. Released energy reflects the severity of the rupture; higher energy indicates more severe rock mass damage. High-frequency signals correspond to small-scale, near-field rupture, while low-frequency signals correspond to large-scale, far-field, or deep rock strata movement.

[0040] Mine pressure data can be collected through pressure sensors deployed on hydraulic supports. Mine pressure data refers to the working resistance of the hydraulic supports, directly reflecting the vertical load exerted on the supports by the roof of the working face, i.e., the magnitude and variation of roof pressure. Essentially, it is a real-time measurement of the interaction force between the supports and the roof. In fully mechanized mining faces, hydraulic supports support the roof and transmit roof pressure to the base. The working resistance is calculated from the fluid pressure within the support columns. Mine pressure data is the most direct indicator for determining whether the roof is stable and whether a large-scale pressure surge or support collapse is imminent. Under conditions of close proximity to coal seams, it characterizes the additional stress increment borne by the current coal seam roof due to the mining of adjacent coal seams.

[0041] Mine pressure data includes initial support force, end-of-cycle resistance, and time-series variation. Initial support force represents the initial support force actively applied to the roof during support installation, reflecting the level of human control. End-of-cycle resistance represents the pressure borne by the support at the end of a mining cycle (just before support relocation), representing the actual load on the roof during mining. Time-series variation shows that the support resistance fluctuates regularly with the advancement of the mining machine and cyclical pressure. When mining of nearby coal seams leads to stress superposition, the working resistance will abnormally increase or fluctuate more severely.

[0042] Delamination data is primarily obtained through roof delamination instruments, used to monitor the movement and separation within the roof strata. Delamination data reflects the relative displacement along the bedding plane at different depths within the roof strata, i.e., the amount of separation between the overlying and underlying strata. Essentially, it represents the width of open cracks resulting from the failure of inter-strata bonding or anchoring. Delamination data directly describes the degree of loss of structural integrity within the roof. When mining of nearby coal seams causes stress redistribution, leading to roof bending and subsidence at the current working face, strata of different lithologies and at different levels will separate due to differences in stiffness. The greater the delamination, the more prone the roof is to delamination-type collapse, serving as an important early warning sign of roof accidents.

[0043] Top separation meters typically have two anchoring points. One is a shallow base point, 2-3m deep, located within the anchoring zone. The other is a deep base point, 5-10m deep, located within stable rock strata outside the anchoring zone. By measuring the displacement difference between the two base points relative to the borehole opening, the shallow and deep separation amounts can be obtained respectively. The shallow separation amount characterizes the separation within the rock strata within the anchoring zone and is usually directly related to the effectiveness of the anchor support. The deep separation amount characterizes the separation of the rock strata above or further from the anchoring zone from the top plate of the anchoring zone, reflecting the movement of overlying key strata or high-lying rock strata. Separation data also includes the separation rate; a rapidly increasing separation rate indicates that the top plate is undergoing continuous separation and there is a risk of instability.

[0044] Vibration data, mine pressure data, and delamination data are high-frequency data, collected at a frequency of minutes or seconds, reflecting the real-time dynamic response of the roof strata.

[0045] Step 1200: Use lidar to periodically scan the top plate to obtain second time-series data containing lidar point cloud data of the top plate.

[0046] The lidar scans the roof at a cycle of 0.5-1 hour to obtain point cloud data. The low frequency is used to reflect macroscopic deformation of the roof, such as subsidence and bending.

[0047] Step 1300: Based on the current work trajectory and the timestamp corresponding to the second time series data, obtain the data segment within the first time series data that is affected by the current work surface operation and perform feature fusion to obtain the third time series data.

[0048] The process of acquiring the third time series data includes: filtering the first time series data based on the position coordinates of the coal mining machine during the mining process and a first distance threshold, and removing time series data with working positions greater than a preset distance.

[0049] Spatially filtering the first-time-series data is performed based on the current working face's operational trajectory. Specifically, the real-time position of the coal mining machine at the current working face is determined according to the operational trajectory, and an influence area is defined, such as a semi-circular area with a radius of 15m centered on the coal mining machine (in the face advance direction). Only the first-time-series data collected within this area is retained. Data outside this area is considered irrelevant background noise and is discarded directly.

[0050] The first time-series data is also time-aligned using the timestamps corresponding to the second time-series data. The timestamps of the second time-series data represent the scanning time of the lidar. Based on the lidar scanning time, a first time-series data segment is found whose time difference with the lidar scanning time is less than a time threshold and meets the above spatial filtering requirements. The time threshold can be set according to the coal mining machine's advance speed, for example, 5 minutes, or other time intervals.

[0051] The first time-series data segment, obtained through spatial filtering and time alignment, is subjected to feature fusion to obtain the third time-series data.

[0052] Specifically, a one-dimensional convolutional neural network is used to fuse the first time-series data to obtain the third time-series data. The first time-series data fused using the one-dimensional convolutional neural network is the data segment within the first time-series data that is affected by the current work face operation.

[0053] Convolutional kernels can be designed at various scales. Multiple parallel convolutional kernels can be used to extract short-term abrupt changes and long-term trend features, respectively. Small convolutional kernels are used to capture local abrupt changes and transient shock signals, such as microseismic events. Large convolutional kernels are used to capture long-term trends and slow changes, such as the cyclical upward phase of mine pressure.

[0054] The convolutional feature map is compressed into a fixed-length feature vector through a pooling layer, representing the temporal fingerprint of the sensor data within the 5-minute window.

[0055] Vibration data, mine pressure data, and delamination data are processed separately through a one-dimensional convolutional neural network to extract feature vectors. The extracted feature vectors are then concatenated to obtain the comprehensive feature vector of the third time series data.

[0056] Third-time series data is not a simple scalar value, but a vector containing rich information such as short-term mutation energy, long-term change rate, and spectral distribution, which significantly improves the ability to characterize mining response.

[0057] The time-domain features extracted by the one-dimensional convolutional neural network effectively suppress single-point noise and random fluctuations, making the third time-series data more representative of the true intensity of the mining response during that period.

[0058] Step 1400: Based on the timestamp of the third time series data and the collected operation records of the nearby coal seam, obtain the relative distance between the working face operation position and the nearby coal seam operation position corresponding to the data in the third time series data.

[0059] A nearby coal seam refers to another coal seam that is very close to the current working face, as specified in advance based on geological data, for example, less than 50 meters away. It is usually an adjacent coal seam above or below. The proximity of a coal seam is unique. This method is not applicable if multiple coal seams exist that could potentially affect the current working face.

[0060] For nearby coal seams, this application does not require sensor data from the nearby coal seams; only the operation records of the nearby coal seams are needed. The operation records include data such as the operation location and operation time.

[0061] Each sample in the third time-series data has undergone rigorous spatiotemporal filtering. Temporally, it is aligned with the time of a specific lidar scan; spatially, it only includes sensor data from the area near the current working face location. Based on the sample's timestamp, the position of the current coal mining machine at that moment, and the nearby coal seam location at the same moment, are obtained from the current working face's trajectory. The relative distance between these two positions is then calculated.

[0062] Step 1500: Extract features from the second time series data based on the timestamp of the third time series data to obtain the roof change features.

[0063] The extraction process of the roof change features includes steps 1510-1520.

[0064] Step 1510: Perform three-dimensional registration and differential comparison between the second time series data of the current time and the data of the previous period to obtain the point cloud difference clusters on the roof surface caused by coal and rock spalling.

[0065] Because lidar has a low acquisition frequency, such as 0.5-1 hour, the overall morphology of the roof changes only slightly within a short time interval. Therefore, the point cloud acquired in a single scan can be considered a static snapshot of the roof at that moment. By performing three-dimensional registration and differential comparison between the current point cloud and the point cloud from the previous period, areas of roof surface change caused by mining activities can be extracted, such as coal and rock spalling, subsidence, or delamination. Furthermore, the roof change characteristics can be quantitatively characterized from these differential clusters.

[0066] 3D registration: The second time series data acquired at the current time is registered with the point cloud of the previous period using rigid or non-rigid body registration to eliminate system deviations caused by slight changes in the scanner pose.

[0067] Differential comparison: The spatial distance between corresponding points in the two registered point clouds is calculated to obtain point cloud difference clusters. These difference clusters represent the net change areas on the roof surface caused by coal and rock spalling, fracturing, and delamination. Positive values ​​indicate bulging or no change, while negative values ​​indicate spalling or subsidence. This invention focuses on the volume loss caused by spalling.

[0068] Step 1520: Perform spatial clustering analysis on the point cloud difference clusters to obtain information on the exfoliated coal seam, and obtain roof change characteristics based on the information on the exfoliated coal and rock.

[0069] Specifically, obtaining the top plate change characteristics based on the point cloud difference clusters includes steps 1521-1523.

[0070] Step 1521: Perform spatial clustering on the point cloud difference clusters to obtain multiple discrete point clusters.

[0071] Spatial clustering analysis is performed on the points in the differential clusters, and spatially adjacent points with variations exceeding the noise threshold are grouped into the same exfoliated coal and rock. Each discrete point cluster corresponds to an independent coal and rock exfoliation body, including blocky exfoliation, sheety exfoliation, or powdery exfoliation areas.

[0072] Step 1522: Identify each cluster of points as exfoliated coal and rock, and calculate the three-dimensional geometric dimensions of each exfoliated coal and rock based on the shape of each cluster of points.

[0073] For each point cluster (exfoliated coal and rock), calculate its bounding box. In the local coordinate system, calculate the maximum and minimum values ​​of all points within the cluster in the X, Y, and Z directions. The three-dimensional geometric dimensions of the exfoliated coal and rock include length, width, and height. The length is the absolute value of the difference between the maximum and minimum values ​​in the X direction. The width is the absolute value of the difference between the maximum and minimum values ​​in the Y direction. The height is the absolute value of the difference between the maximum and minimum values ​​in the Z direction.

[0074] Step 1523: According to the preset particle size range, obtain the roof change characteristics based on the quantity, volume or projected area distribution of the exfoliated coal and rock.

[0075] The number of exfoliated coal and rock is equivalent to the number of discrete point clusters mentioned above. The volume of the exfoliated coal and rock can be estimated by fitting a surface to the discrete points or by accumulating voxels, and can be approximated as the bounding box volume or calculated using a Delaunay triangulation to determine the volume beneath the surface area. The grid coverage area is calculated as the projected area by projecting the points within the clusters onto the roof plane.

[0076] According to the preset particle size range, such as 1-5mm, 5-10mm, 10-30mm, 30-60mm, >60mm, the number of exfoliated coal and rock in each range, the volume of exfoliated coal and rock in each range, and the projected area of ​​exfoliated coal and rock in each range are counted in the scanning area. The roof change characteristics are obtained by combining them.

[0077] For example, for a 5m x 3m area on the top plate of the current working face, the point cloud density is approximately 100 points / m. 2 A total of approximately 1500 points were identified. The minimum size for determining effective peeling was set to 1 mm.

[0078] 3D registration and differential comparison: The point cloud data Cloud_current at time t and Cloud_prev at time t+1 are registered to align their coordinate systems. Then, the Z-direction difference of each corresponding point is calculated to obtain the difference value. Negative values ​​indicate roof subsidence, while positive values ​​may indicate coal accumulation (not considered in this application). Some point cloud data are shown in Table 1.

[0079] Table 1: Partial Point Cloud Data Table

[0080]

[0081] A portion of the point clouds in Table 1 were filtered, and points with |ΔZ|>1mm were retained. Points 107 and 108 were discarded as noise.

[0082] DBSCAN was used to cluster the point cloud shown in Table 1 above, with a neighborhood radius of 0.05m and a minimum number of points of 3. Clusters A and B were obtained, each representing a single exposed solid. Cluster A contains points 101, 102, 103, 104, 105, and 106. The X-coordinates of these points are in the range of 1.50-1.55, and the Y-coordinates are in the range of 0.80-0.82, with a distance of less than 0.05m between them. Cluster B contains points 201, 202, 203, and 204. The X-coordinates of these points are in the range of 3.2-3.23, and the Y-coordinates are in the range of 1.50-1.52.

[0083] For cluster A, the bounding box is calculated. The minimum X-coordinate of a point in cluster A is 1.50m, and the maximum X-coordinate is 1.55m, yielding a length of 50mm. The minimum Y-coordinate of a point in cluster A is 0.80m, and the maximum Y-coordinate is 0.82m, yielding a width of 20mm. The peeling depth is calculated using ΔZ, where ΔZ is the minimum value in cluster A. min_A The maximum value of ΔZ is -6.3mm. max_A The depth of spalling is -2.7mm, and the maximum spalling depth is taken as |ΔZ. min_A The thickness is 6.3mm. The three-dimensional dimensions are 50mm long, 20mm wide, and 6.3mm deep. The volume of the peeled-off solid is approximately the volume of the bounding box, 6300mm². 3 .

[0084] For cluster B, the bounding box is calculated. The minimum X-coordinate of a point in cluster B is 3.20m, and the maximum X-coordinate is 3.23m, yielding a length of 30mm. The minimum Y-coordinate of a point in cluster B is 1.50m, and the maximum Y-coordinate is 1.52m, yielding a width of 20mm. The peeling depth is calculated using ΔZ, where ΔZ is the minimum value in cluster B. min_B The maximum value of ΔZ is -2.1mm. max_B The depth of spalling is -1.9mm, and the maximum spalling depth is taken as |ΔZ. min_B The thickness is 2.1mm. The three-dimensional dimensions are 30mm long, 20mm wide, and 2.1mm deep. The volume of the peeled-off solid is approximately the volume of the bounding box, 1260mm². 3 .

[0085] The preset particle size ranges include: Range 1: 1-5mm, Range 2: 5-10mm, Range 3: 10-30mm, Range 4: 30-60mm, and Range 5: >60mm. Cluster A has a maximum 3D size of 50mm, belonging to Range 4. Cluster B has a maximum 3D size of 30mm, which is the lower limit of Range 4, and therefore also falls under Range 4.

[0086] Assuming that within the entire 5m*3m area, besides the two locations mentioned above, other small abrasion clusters were obtained through complete point cloud analysis, then: Interval 1 contains 12 clusters, which are tiny particles with a maximum size of 2-4mm. Interval 2 contains 5 clusters, with a size of 6-8mm. Interval 3 contains 3 clusters, with a size of 15-25mm. Interval 4 contains 2 clusters, corresponding to clusters A and B mentioned above. Interval 5 contains none.

[0087] At this scanning time, the total number of detached particles was 22. The total volume of all detached particles in interval 1 was 200 mm. 3 The total volume of all peeled-off soil particles in interval 2 is 800 mm. 3 The total volume of all peeled-off soil particles in interval 3 is 3000 mm². 3The total volume of all peeled-off soil particles in interval 4 is 7560 mm². 3 Therefore, the total volume of all exfoliated particles in all intervals is 11560 mm. 3 .

[0088] Finally, for this scanning moment, the spalling-related roof change eigenvector includes the overall spalling intensity (total number 22, total volume 11560 mm). 3 ), particle size distribution (quantity or volume percentage of each interval), and spatial distribution (dividing a 5m*3m region into 1m*1m grids, independently calculating the above indicators for each grid to form a spatial feature map).

[0089] Step 1600: Construct a vector sequence based on the third time series data, the relative distance, and the roof change characteristics.

[0090] For each lidar scan moment, a vector fused from multiple sources is obtained. This vector includes the relative distance to the nearby coal seam operation location, roof change characteristic vectors, vibration data, mine pressure data, and delamination data. The vectors obtained from all scan moments are arranged in chronological order to form a vector sequence. This vector sequence simultaneously encodes the temporal evolution of spatial distance, roof mining status, and mining response.

[0091] Step 1700: Input the vector sequence into the self-attention neural network to obtain the contribution weight of the nearby coal seam mining to the current working face roof deformation.

[0092] The self-attention network uses the current working face's operational status features as the query vector, the distance to the nearest coal seam and the roof status features as the key vector, and the roof change features as the value vector to obtain the impact of the nearest coal seam operation on the current working face.

[0093] A self-attention network is used to learn the correlation strength between near-shore coal seam mining activity and the current roof deformation characteristics at each time step in the sequence. The model automatically calculates and outputs a contribution weight through an attention weight mechanism, such as a value between 0 and 1 or a weight distribution, quantitatively representing the degree of influence of that specific near-shore coal seam mining activity on the current roof deformation. A higher weight indicates that the near-shore coal seam mining is the main factor causing the current roof deformation.

[0094] This invention specifically addresses the problem of mutual interference between single, clearly defined, and spatially adjacent coal seams. It is applicable to scenarios where upper and lower seams are mined simultaneously but the distance between them is small. It possesses clarity and exclusivity.

[0095] Clarity: Prior to the analysis, the nearest coal seam was identified using geological data.

[0096] Exclusivity: If multiple coal seams that may have an impact exist, or if they are far apart, then the near-field scenario described in this invention does not apply. This is because multi-source interference would undermine the assumption of this invention based on a single perturbation source response, causing the attention mechanism to fail to converge effectively.

[0097] This invention employs strict spatiotemporal filtering rules to eliminate a large amount of irrelevant background data, allowing the model to focus on signals directly influenced by mining activities, resulting in more accurate analysis results. It cleverly solves the alignment problem between high-frequency sensor data and low-frequency point cloud data, unifying the causal chain between mining behavior and roof deformation over time. By using a self-attention network, it can automatically learn and quantify the contribution weight of specific nearby coal seam mining activities to roof deformation from spatiotemporal sequences, providing quantitative basis for mine pressure control and safety decisions. Furthermore, the method's applicability is clearly defined, applicable only to a single nearby coal seam, avoiding model misuse and performance degradation under complex geological conditions, and improving the reliability of industrial applications.

[0098] In this embodiment, the roof plate variation characteristics also include the local deflection change rate, the non-uniform settlement coefficient, and the roof plate surface roughness.

[0099] Deflection refers to the bending deformation of the roof strata in the direction perpendicular to the bedding plane, usually expressed as subsidence. The local deflection rate reflects the degree of bending of the roof surface per unit length, i.e., the rate of change of the slope, essentially the curvature or second derivative. By dividing the roof point cloud into a regular grid, fitting a local plane or quadratic surface within each grid, and calculating the vertical displacement of the grid center point relative to the initial reference plane or the plane of the previous period, the local deflection rate can be calculated based on this vertical displacement.

[0100] A high rate of deflection indicates that the roof is bending violently over a short distance, typically corresponding to stress concentration zones, delamination boundaries, or locations where fracture is imminent. This indicator can provide early warning of local instability, such as the bending deformation zone before the old roof breaks.

[0101] Settlement refers to the vertical displacement of the roof under the influence of mining. The non-uniform settlement coefficient measures the dispersion of settlement at various points within the monitoring area, reflecting the uniformity of roof subsidence. Settlement values ​​are obtained at all points within the monitoring area, and the coefficient of variation of the settlement values ​​is calculated as the non-uniform settlement coefficient.

[0102] When the non-uniform settlement coefficient is close to 0, the roof settles uniformly, the stress is balanced, and the stability is good. If the non-uniform settlement coefficient is large, such as exceeding 0.3, it indicates that the roof settlement is extremely uneven, with local subsidence basin areas, which may be accompanied by the risk of delamination, step displacement, or local collapse. The non-uniform settlement coefficient can be directly used to determine whether mining of nearby coal seams has led to local stress concentration.

[0103] Roof surface roughness describes the degree of undulation of the roof surface within a local area, i.e., the degree of deviation of the actual surface from an ideal plane. Mining-induced spalling, fracture opening, and residual broken rock fragments all increase roughness. Within each grid, a least-squares plane is fitted to the point cloud, and the absolute value of the residual from each point to this fitted plane is calculated. The roof surface roughness is defined as the root mean square of the residual.

[0104] If the surface roughness of the roof is low, it indicates that the roof surface is flat, the rock mass is intact, and there is no obvious spalling or breakage. If the surface roughness of the roof is high, it indicates that the roof is uneven, with spalling pits, protruding rock blocks, or open fractures, indicating that the roof has suffered significant mining damage. Increased roughness is often a direct manifestation of spalling, spalling, and delamination, and is related to vibration energy and peak mine pressure.

[0105] In this embodiment, the first time-series data further includes at least one of the following: ambient temperature and humidity, support stress, or coal mining machine load data.

[0106] Changes in ambient temperature can affect the viscosity of hydraulic support fluids, sensor accuracy, and the thermal stress state of the rock mass. For example, increased temperature may cause drift in support resistance readings or induce minor expansion of the rock strata. High humidity may exacerbate the softening and argillaceization of the roof strata (especially mudstone and shale), reducing rock mass strength and thus affecting roof stability and delamination rate.

[0107] Install mine temperature and humidity sensors, such as explosion-proof temperature and humidity transmitters, in roadways or on supports near the working face. The sampling frequency can be on the order of minutes or synchronized with vibration data.

[0108] Support stress refers to the mechanical stress generated inside the hydraulic support column, top beam, or base. Compared to conventional mine pressure data (support working resistance), it focuses more on the stress state of the support structure itself, reflecting the details of the interaction between the support and the surrounding rock. An abnormally high support stress may indicate roof pressure, uneven load on the support, or column damage.

[0109] Stress sensors are installed in key parts of the support structure (such as the inner cavity of the column and the top beam pin) to monitor stress changes in real time. The sampling frequency is on the order of minutes or seconds, and can be synchronized with mine pressure data.

[0110] Coal mining machine load data refers to the power, current, or torque of the cutting and traction units of the coal mining machine. The load directly reflects the ease with which the coal mining machine cuts coal and rock. When the coal mining machine enters hard rock or stress concentration areas, the load will increase significantly; conversely, the load may decrease when passing through faults or fractured zones. This data can serve as a direct quantification of the coal mining machine's operating status, helping to distinguish between normal cutting and encountering abnormal geological bodies.

[0111] Parameters such as cutting motor current and traction motor power are read from the coal mining machine control system and transmitted via industrial Ethernet or wireless means. The sampling frequency can be on the order of seconds. The coal mining machine control system itself is typically on the order of milliseconds, so it can be downsampled for use.

[0112] This embodiment introduces an electronic device, including a processor and a memory, wherein the memory stores programs or instructions that can run on the processor, and when the programs or instructions are executed by the processor, they implement the steps of a near-field coal seam pressure analysis method as described in any embodiment of the present invention.

[0113] This embodiment introduces a readable storage medium storing a program or instructions that, when executed by a processor, implement the steps of a near-field coal seam pressure analysis method as described in any embodiment of the present invention.

[0114] While specific embodiments of the present invention have been described in detail by way of examples, those skilled in the art should understand that the above examples are for illustrative purposes only and are not intended to limit the scope of the invention. Those skilled in the art should understand that modifications can be made to the above embodiments without departing from the scope and spirit of the invention.

[0115] Those skilled in the art will recognize that the modules and algorithm steps described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0116] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above-described apparatus and equipment can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0117] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0118] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of the embodiments of the present invention, depending on actual needs.

[0119] In addition, the functional modules in the embodiments of the present invention can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0120] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0121] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this application.

[0122] It should be understood that the sequence numbers of the steps in the invention's content and embodiments do not absolutely imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention. The foregoing description of embodiments of this disclosure has been provided for illustrative and descriptive purposes. The foregoing description is not exhaustive and is not intended to limit this disclosure to the exact form disclosed. Various modifications and variations may exist based on the foregoing teachings, or various modifications and variations may be derived from the practice of this disclosure. These embodiments were chosen and described to illustrate the principles of this disclosure and its practical application, so that those skilled in the art can utilize this disclosure in various implementations and modifications suitable for the specific purpose of the concept.

Claims

1. A method for close-range coal seam pressure analysis, characterized in that, Includes the following steps: Collect the first time-series data during the current mining process at the working face. The first time-series data includes vibration data, mine pressure data, and delamination data. The top plate is periodically scanned using lidar to obtain second time-series data containing lidar point cloud data of the top plate; Based on the current work trajectory and the timestamp corresponding to the second time series data, the data segments within the first time series data affected by the current work operation are obtained and their features are fused to form the third time series data. Based on the timestamp of the third time series data and the collected operation records of the nearby coal seam, the relative distance between the working face operation position and the nearby coal seam operation position corresponding to the data in the third time series data is obtained. Based on the timestamp of the third time series data, feature extraction is performed on the second time series data to obtain the roof change features; A vector sequence is constructed based on the third time-series data, the relative distance, and the roof change characteristics; The vector sequence is input into the self-attention neural network to obtain the contribution weight of the nearby coal seam mining to the current working face roof deformation.

2. The method for close-range coal seam pressure analysis as described in claim 1, characterized in that, The extraction process of the roof change features includes: The second time series data of the current time is registered and compared with the data of the previous period in three dimensions to obtain the point cloud difference clusters on the roof surface caused by coal and rock spalling. Spatial clustering analysis of point cloud difference clusters is used to obtain information on the exfoliated coal seam, and roof change characteristics are obtained based on the information on the exfoliated coal and rock.

3. The method for close-range coal seam pressure analysis as described in claim 2, characterized in that, The specific features of roof variation obtained from point cloud difference clusters include: Spatial clustering of point cloud difference clusters yields multiple discrete point clusters; Each point cluster is identified as exfoliated coal and rock, and the three-dimensional geometric dimensions of each exfoliated coal and rock are calculated based on the shape of each point cluster. Based on the preset particle size range, the roof variation characteristics are obtained according to the quantity, volume, or projected area distribution of the exfoliated coal and rock.

4. The method for close-range coal seam pressure analysis as described in claim 2, characterized in that, The roof plate variation characteristics also include local deflection variation rate, non-uniform settlement coefficient and roof plate surface roughness.

5. The method for close-range coal seam pressure analysis as described in claim 1, characterized in that, The self-attention network uses the current working face's operational status features as the query vector, the distance to the nearest coal seam and the roof status features as the key vector, and the roof change features as the value vector to obtain the impact of the nearest coal seam operation on the current working face.

6. The method for close-range coal seam pressure analysis as described in claim 1, characterized in that, A one-dimensional convolutional neural network is used to fuse the first time series data to obtain the third time series data.

7. The method for close-range coal seam pressure analysis as described in claim 1, characterized in that, The process of acquiring the third time-series data includes: Based on the position coordinates of the coal mining machine during the mining process and the first distance threshold, the first time series data is filtered to remove time series data with working positions greater than a preset distance.

8. The method for close-range coal seam pressure analysis as described in claim 1, characterized in that, The first time series data also includes at least one of the following: ambient temperature and humidity, support stress, or coal mining machine load data.

9. An electronic device, characterized in that, It includes a processor and a memory, the memory storing programs or instructions that can run on the processor, the programs or instructions being executed by the processor to implement the steps of a near-field coal seam pressure analysis method as described in any one of claims 1 to 8.

10. A readable storage medium, characterized in that, The readable storage medium stores a program or instructions that, when executed by a processor, implement the steps of a near-field coal seam pressure analysis method as described in any one of claims 1 to 8.

Images