Method for monitoring and evaluating blind area of coal mining face

Abstract

The application relates to the technical field of data processing, and discloses a blind area monitoring and evaluation method for a coal mining face, which comprises the following steps: coupling the coordinate sequence of a cutter head in a coal mining machine and the support state of a hydraulic support into the motion state of the coal mining machine; predicting the blind area probability density distribution of a mine based on the motion state and the dynamic blind area volume change rate of the mine; outputting the monitoring point coordinate sequence in the mine according to the blind area probability density distribution, and planning the scanning track of a mobile monitoring device in the coal mining machine based on the monitoring point coordinate sequence; performing projection transformation on the multi-source sensing data under the scanning track to obtain a three-dimensional topological working face of the mine; extracting the connected domain cavity features in the three-dimensional topological working face, and determining the blind area volume proportion and the spatial distribution entropy value of the mine according to the connected domain cavity features; and the application can improve the accuracy of blind area evaluation of the coal mining face.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method for monitoring and evaluating blind spots in coal mining faces. Background Technology

[0002] In the field of blind spot monitoring and assessment in coal mining faces, existing technologies often struggle to accurately capture the dynamic coordination relationship between the coal mining machine and surrounding equipment. The description of the coal mining machine's motion state is relatively simplistic, failing to effectively integrate key information such as cutterhead coordinate changes and hydraulic support status. This results in a lag in the perception of dynamic changes in blind spots, making it difficult to reflect the actual distribution of blind spots in the mine in real time, and thus leading to low prediction accuracy of blind spot probability density distribution.

[0003] Meanwhile, existing technologies have significant shortcomings in monitoring point planning and data processing. The selection of monitoring points often lacks specificity, and scanning trajectories are not optimized in conjunction with blind zone probability characteristics, resulting in limited coverage and accuracy of multi-source sensing data. Furthermore, the spatiotemporal registration and feature extraction processes during data conversion and 3D modeling are not sufficiently refined, making it difficult to accurately identify key features such as connected domain voids. Ultimately, this leads to significant discrepancies between the assessed blind zone volume ratio and spatial distribution entropy and the actual situation, failing to provide reliable decision-making support for coal mining operations. Summary of the Invention

[0004] This invention provides a method for monitoring and evaluating blind spots in coal mining faces, the main purpose of which is to solve the problem of low accuracy in evaluating blind spots in coal mining faces.

[0005] To achieve the above objectives, the present invention provides a method for monitoring and evaluating blind spots in coal mining faces, comprising:

[0006] S1. Couple the coordinate sequence of the cutterhead in the coal mining machine with the support state of the hydraulic support to form the motion state of the coal mining machine;

[0007] S2. Based on the motion state and the dynamic blind zone volume change rate of the coal mining machine, predict the blind zone probability density distribution of the coal mining machine;

[0008] S3. Output the coordinate sequence of monitoring points in the mine according to the probability density distribution of the blind zone, and plan the scanning trajectory of the mobile monitoring equipment in the coal mining machine based on the coordinate sequence of the monitoring points;

[0009] S4. Perform projection transformation on the multi-source sensing data under the scanning trajectory to obtain the three-dimensional topological working face of the mine;

[0010] S5. Extract the connected domain void features in the three-dimensional topological working face, and determine the blind zone volume ratio and spatial distribution entropy value of the mine based on the connected domain void features.

[0011] In a preferred embodiment, coupling the coordinate sequence of the cutterhead in the coal mining machine with the support state of the hydraulic support to form the motion state of the coal mining machine includes:

[0012] The real-time coordinate sequence of the tool head is collected by a positioning sensor;

[0013] Simultaneously acquire the support status of the hydraulic support;

[0014] The real-time coordinate sequence is spatiotemporally aligned with the support state, and the spatiotemporally aligned real-time coordinate sequence is weighted and fused with the support state to obtain the motion state of the coal mining machine.

[0015] In a preferred embodiment, predicting the blind zone probability density distribution of the coal mining machine based on the motion state and the dynamic blind zone volume change rate of the coal mining machine includes:

[0016] Based on the three-dimensional position coordinates of the motion state, point cloud data of the area to be mined in front of the coal mining machine is acquired in real time;

[0017] The blind zone volume of the coal mining machine is obtained based on the point cloud data, and the dynamic blind zone volume change rate of the coal mining machine is generated based on the change of the blind zone volume over a continuous period of time.

[0018] Predict the initial probability density distribution of the blind zone in the coal mining machine based on the motion state and the dynamic blind zone volume change rate;

[0019] Spatially smooth the initial probability density distribution to generate the blind zone probability density distribution map of the coal mining machine.

[0020] In a preferred embodiment, predicting the initial probability density distribution of the blind zone in the coal mining machine based on the motion state and the dynamic blind zone volume change rate includes:

[0021] The dynamic geological deformation field of the area in front of the coal mining machine is generated based on the travel velocity vector and three-dimensional coordinates in the motion state.

[0022] Residual compensation is performed on the dynamic geological deformation field and the dynamic blind zone volume change rate to obtain the risk feature vector of the blind zone;

[0023] Spatial probability mapping is performed on the risk feature vector to obtain the initial probability density distribution of the blind zone;

[0024] The initial probability density distribution is dynamically meshed to eliminate discontinuous regions at the boundaries of the initial probability density distribution after dynamic mesh optimization, thereby obtaining the initial probability density distribution of the blind zone.

[0025] In a preferred embodiment, the step of outputting the coordinate sequence of monitoring points in the mine according to the probability density distribution of the blind zone, and planning the scanning trajectory of the mobile monitoring equipment in the coal mining machine based on the monitoring point coordinate sequence, includes:

[0026] Extract the spatial boundary coordinates of the high-probability blind zone regions from the blind zone probability density distribution;

[0027] Initial monitoring points are randomly generated within the spatial boundary coordinates;

[0028] Based on the real-time motion status of the coal mining machine and the roof stress distribution data, the initial monitoring point is dynamically optimized to obtain the coordinate sequence of the initial monitoring point after dynamic path optimization.

[0029] The scanning trajectory of the mobile monitoring device is generated based on the coordinate sequence fitting.

[0030] In a preferred embodiment, the dynamic path optimization of the initial monitoring point based on the real-time motion state of the coal mining machine and the roof stress distribution data includes:

[0031] Based on the pose parameters of the coal mining machine, the equipment obstacle avoidance constraints of the coal mining machine are generated;

[0032] Areas exceeding a preset threshold in the roof stress distribution diagram of the coal mining machine are marked as high-risk restricted areas;

[0033] A three-dimensional path optimization objective function is constructed for the initial monitoring point based on monitoring coverage, total path length, and restricted area avoidance rate.

[0034] In a preferred embodiment, the step of performing multi-objective path optimization on the initial monitoring points with the optimization objectives of maximizing monitoring coverage, minimizing total path length, and avoiding restricted areas, to obtain the coordinate sequence of the initial monitoring points after dynamic path optimization, includes:

[0035] A three-dimensional path optimization objective function is constructed for the initial monitoring point based on monitoring coverage, total path length, and restricted area avoidance rate.

[0036] Based on the kinematic constraints of the coal mining machine, the three-dimensional path optimization objective function is solved by joint optimization to obtain the Pareto optimal solution set of the initial monitoring point;

[0037] Discretize the path scheme with the highest comprehensive fitness in the Pareto optimal solution set to obtain the coordinate sequence of monitoring points after dynamic path optimization.

[0038] In a preferred embodiment, the three-dimensional path optimization objective function is as follows:

[0039]

[0040] In the formula, For the path scheme, The weight of the monitoring coverage, The number of blocks in the high-probability blind zone region. For the first One monitoring point, For the first Blind zone probability density weights for each monitoring point For the first Effective coverage coefficient of monitoring points The weight is the total length of the path. The total number of the initial monitoring points. For the first One monitoring point, For the first The three-dimensional spatial coordinates of each monitoring point For the first The three-dimensional spatial coordinates of each monitoring point The weight of the restricted area avoidance rate, The number of restricted blocks, For the trajectory in the restricted area The cumulative time spent inside, Forbidden Zone Spatial conflict penalty items, This is the formula for calculating Euclidean distance. This represents the total number of restricted blocks.

[0041] In a preferred embodiment, the step of projecting and transforming the multi-source sensing data under the scanning trajectory to obtain the three-dimensional topological working face of the mine includes:

[0042] Spatiotemporal registration of lidar scanning data and camera image data;

[0043] The registered lidar scanning data and camera image data are used to generate a 3D point cloud through perspective projection transformation.

[0044] The three-dimensional point cloud is reconstructed to obtain the three-dimensional topological working surface of the mine.

[0045] In a preferred embodiment, the step of extracting the connected domain void features in the three-dimensional topological working face and determining the blind zone volume ratio and spatial distribution entropy value of the mine based on the connected domain void features includes:

[0046] The three-dimensional topological working surface is divided into sub-regions, and the connected domain holes in the sub-regions are identified;

[0047] The total mineable volume of the three-dimensional topological working face is obtained. Based on the ratio of the total volume of the connected domain cavities to the total mineable volume, the proportion of the blind zone volume of the mine is obtained.

[0048] The blind zone probability density value of the sub-region is determined based on the characteristics of the connected domain voids;

[0049] The spatial distribution entropy value of the mine is obtained by integrating the probability density value of the blind zone based on the number of the sub-regions.

[0050] Compared with the prior art, the present invention has the following beneficial effects:

[0051] 1. This invention achieves precise characterization of the coal mining machine's motion state by coupling the coordinate sequence of the coal mining machine's cutterhead with the support state of the hydraulic support. This coupling process not only achieves spatiotemporal alignment and weighted fusion, but also provides more comprehensive and accurate dynamic basic data for subsequent prediction of blind zone probability density distribution, thereby enhancing the ability to perceive dynamic changes in mine blind zones and laying a solid foundation for improving the accuracy of blind zone assessment.

[0052] 2. This invention plans the scanning trajectory of a mobile monitoring device based on the probability density distribution of blind zones, constructs a three-dimensional topological working face by combining projection transformation of multi-source sensing data, and extracts the void features of connected domains to determine the volume ratio and spatial distribution entropy value of blind zones. This series of processing steps, from monitoring point selection and data acquisition to feature analysis, forms a complete technical closed loop, effectively improving the accuracy of capturing the spatial features of blind zones, thereby significantly improving the accuracy of blind zone assessment in coal mining faces and providing strong support for the safe and efficient operation of coal mining. Attached Figure Description

[0053] Figure 1 A flowchart illustrating a method for monitoring and evaluating blind spots in a coal mining face according to an embodiment of the present invention;

[0054] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0055] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0056] This application provides a method for monitoring and evaluating blind spots in coal mining faces. The execution entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the method for monitoring and evaluating blind spots in coal mining faces can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0057] Reference Figure 1 The diagram shown is a flowchart illustrating a method for monitoring and evaluating blind spots in a coal mining face according to an embodiment of the present invention. In this embodiment, the method for monitoring and evaluating blind spots in a coal mining face includes:

[0058] S1. Couple the coordinate sequence of the cutterhead in the coal mining machine with the support state of the hydraulic support to form the motion state of the coal mining machine;

[0059] In this embodiment of the invention, coupling the coordinate sequence of the cutterhead in the coal mining machine with the support state of the hydraulic support to form the motion state of the coal mining machine includes:

[0060] The real-time coordinate sequence of the tool head is collected by a positioning sensor;

[0061] Simultaneously acquire the support status of the hydraulic support;

[0062] The real-time coordinate sequence is spatiotemporally aligned with the support state, and the spatiotemporally aligned real-time coordinate sequence is weighted and fused with the support state to obtain the motion state of the coal mining machine.

[0063] Specifically, a laser positioning sensor is installed at the center of the cutterhead. The sensor's detection direction is aligned with a fixed reference point on the coal face, which is pre-set on the top or bottom plate of the face and remains in a fixed position. The laser positioning sensor continuously detects the position of the cutterhead relative to the reference point at a frequency of 10 times per second. After each detection, it automatically records the current time (accurate to milliseconds) and the corresponding three-dimensional coordinates (including coordinate values ​​in the X, Y, and Z axes). These timestamped three-dimensional coordinates are arranged in chronological order of detection time to form a real-time coordinate sequence of the cutterhead.

[0064] Furthermore, pressure sensors and displacement sensors are installed on the hydraulic cylinders of the hydraulic support. The pressure sensors detect the hydraulic pressure within the cylinders, while the displacement sensors detect the extension / retraction length of the cylinders. Both sensors are connected to the same clock module as the laser positioning sensor, ensuring that all sensors use the same time reference. The support status of the hydraulic support is acquired in real time via the pressure and displacement sensors. Each acquisition synchronously records the current time (accurate to milliseconds), and the hydraulic pressure and extension / retraction length values ​​are used as core parameters of the support status, continuously collecting and forming support status data.

[0065] Furthermore, the timestamp corresponding to each coordinate is extracted from the real-time coordinate sequence, and the timestamp corresponding to each support state parameter is extracted from the support state data. The timestamps of the two sequences are then compared. For a given coordinate in the real-time coordinate sequence, a support state parameter with the same timestamp in the support state data is found. If no two timestamps are exactly the same, the support state parameter with the closest timestamp (the time difference between the two is no more than 0.1 seconds) is selected for matching, thus completing the spatiotemporal alignment. Based on the operating characteristics of the coal mining machine, the weights of the real-time coordinate sequence and the support state are determined: the real-time coordinate sequence reflects the positional changes of the cutterhead and has a significant impact on the position and trajectory during motion, so its weight is set to 0.6; the support state reflects the stress stability of the coal mining machine and has a significant impact on the operational smoothness during motion, so its weight is set to 0.4.

[0066] Furthermore, the real-time coordinates and support status parameters at the same time point after alignment are integrated according to the above weights. The coordinate data is converted into motion position related indicators after weight calculation, and the support status parameters are converted into operation stability related indicators after weight calculation. The two are combined to form the motion state of the coal mining machine, which includes position, direction of movement, and operational stability.

[0067] In summary, coupling the coordinate sequence of the cutterhead in a coal mining machine with the support state of the hydraulic support to obtain the machine's motion state enables a precise depiction of its motion. The cutterhead coordinate sequence directly reflects the real-time position and trajectory of the coal mining machine, while the hydraulic support support state is related to the spatial conditions and support stability of the working face. Through spatiotemporal alignment and weighted fusion, these two sets of data can integrate the dual key information of position dynamics and environmental support, providing a more comprehensive and three-dimensional description of the coal mining machine's motion state.

[0068] In summary, this coupled processing provides reliable foundational data for the entire process of subsequent blind zone monitoring and assessment. Precise motion status ensures that the point cloud data and dynamic blind zone volume change rate generated based on it are more in line with actual working conditions, thereby improving the accuracy of blind zone probability density distribution prediction and guaranteeing the accuracy of the entire blind zone monitoring and assessment process from the source.

[0069] S2. Based on the motion state and the dynamic blind zone volume change rate of the coal mining machine, predict the blind zone probability density distribution of the coal mining machine;

[0070] In this embodiment of the invention, predicting the blind zone probability density distribution of the coal mining machine based on the motion state and the dynamic blind zone volume change rate of the coal mining machine includes:

[0071] Based on the three-dimensional position coordinates of the motion state, point cloud data of the area to be mined in front of the coal mining machine is acquired in real time;

[0072] The blind zone volume of the coal mining machine is obtained based on the point cloud data, and the dynamic blind zone volume change rate of the coal mining machine is generated based on the change of the blind zone volume over a continuous period of time.

[0073] Predict the initial probability density distribution of the blind zone in the coal mining machine based on the motion state and the dynamic blind zone volume change rate;

[0074] Spatially smooth the initial probability density distribution to generate the blind zone probability density distribution map of the coal mining machine.

[0075] The prediction of the initial probability density distribution of the blind zone in the coal mining machine based on the motion state and the dynamic blind zone volume change rate includes:

[0076] The dynamic geological deformation field of the area in front of the coal mining machine is generated based on the travel velocity vector and three-dimensional coordinates in the motion state.

[0077] Residual compensation is performed on the dynamic geological deformation field and the dynamic blind zone volume change rate to obtain the risk feature vector of the blind zone;

[0078] Spatial probability mapping is performed on the risk feature vector to obtain the initial probability density distribution of the blind zone;

[0079] The initial probability density distribution is dynamically meshed to eliminate discontinuous regions at the boundaries of the initial probability density distribution after dynamic mesh optimization, thereby obtaining the initial probability density distribution of the blind zone.

[0080] Specifically, a 3D lidar is installed in front of the coal mining machine's cutterhead, with its scanning range covering an area within 5 meters in front of the machine's direction of movement. The lidar is connected to the coal mining machine's motion monitoring system, which reads the 3D position coordinates in real time and uses these coordinates to determine the lidar's current position reference.

[0081] Furthermore, the lidar scans the area to be mined ahead at a frequency of 8 times per second. Each scan emits a laser beam and receives the reflected signal. The three-dimensional coordinates of the scanned point relative to the lidar are calculated based on the time difference of the reflected signal. These coordinates are then combined with the current three-dimensional position coordinates of the coal mining machine to convert the scanned point coordinates into coordinates in the global coordinate system of the working face. The set of coordinates of all scanned points forms the point cloud data of the area to be mined. Each generated point cloud data set includes a timestamp synchronized with the movement status.

[0082] Furthermore, the coordinate range of all scanning points is extracted from the point cloud data to determine the spatial boundary of the area to be mined (including the roof and floor of the coal seam and its side boundaries). Based on the structural dimensions of the coal mining machine and the installation location of the lidar, the definition of the blind zone is determined: the spatial area in the point cloud data that is not covered but is within the operating contact range of the coal mining machine is the blind zone.

[0083] Furthermore, using spatial modeling tools, the spatial range corresponding to the blind zone is divided into several cubic grids with a side length of 0.1 meters. The number of all grids that completely belong to the blind zone is counted, and the volume of each grid is 0.001 cubic meters. The volume of the blind zone is obtained by multiplying the number of grids by the volume of a single grid. The blind zone volume is continuously recorded every 10 seconds, and the difference in blind zone volume between two adjacent time points is calculated. This difference is divided by the interval between the two time points (10 seconds) to obtain the change in blind zone volume over a continuous period of time. These values ​​are arranged in chronological order to form the dynamic rate of change of blind zone volume.

[0084] Furthermore, the current three-dimensional position, direction of movement, and speed are extracted from the movement status of the coal mining machine, and the main spatial orientation of the blind zone is determined based on the direction of movement (for example, when moving in the southeast direction, the blind zone is mostly concentrated in the unscanned area on the southeast side).

[0085] Furthermore, by combining the rate of change of the dynamic blind zone volume with the movement speed, if the rate of change is positive and the movement speed is fast, it indicates that the blind zone is expanding and the expansion speed is related to the movement, and the initial probability density of the corresponding region increases; if the rate of change is negative and the movement speed is slow, it indicates that the blind zone is shrinking, and the initial probability density of the corresponding region decreases.

[0086] Furthermore, each spatial grid where a blind zone may exist is assigned an initial probability value (the probability value ranges from 0 to 1, where 0 indicates that it is impossible to be a blind zone and 1 indicates that it is definitely a blind zone), and the probability value distribution of all grids forms the initial probability density distribution.

[0087] Specifically, inertial navigation sensors are installed on the coal mining machine. By combining the traveling velocity vector (including the direction and magnitude of the velocity) with three-dimensional coordinates, the current spatial location of the coal seam where the coal mining machine is located is determined. Based on the three-dimensional coordinates, the original geological data of the coal seam (including coal seam hardness and rock strata distribution) of this location over the past 10 minutes is traced. Combined with the traveling velocity vector, the direction of compression and disturbance of the coal seam in front by the coal mining machine is determined—if it travels along the positive X-axis, the coal seam in the positive X-axis direction in front is more significantly compressed.

[0088] Furthermore, the coal seam within a 5-meter radius ahead is divided into a 1-meter × 1-meter × 1-meter spatial grid. Based on velocity vectors, the potential deformation degree of each grid within the next minute is calculated (e.g., the greater the velocity, the greater the estimated deformation degree of the grid in the same direction ahead). Simultaneously, the correspondence between the grid positions and the actual coal seam positions is corrected using three-dimensional coordinates. The deformation degree of each grid (represented by a value of 0-1, with 1 indicating maximum deformation) is arranged spatially to form a dynamic geological deformation field, which is updated in real time as the coal mining machine moves.

[0089] Furthermore, the deformation degree values ​​for each spatial grid are extracted from the dynamic geological deformation field, and the corresponding time point change values ​​are extracted from the dynamic blind zone volume change rate. These two values ​​are then compared within the same spatial grid and at the same time point. The difference (residual) between the deformation degree value and the blind zone volume change value for the same grid at the same time point is calculated. If the residual is positive, it indicates that the deformation predicted by the deformation field is greater than the actual change in the blind zone, and the deformation weight of that grid needs to be reduced. If the residual is negative, the deformation weight is increased. By adjusting the weights, the deviation between the two values ​​is controlled within 0.05, thus completing the residual compensation.

[0090] Furthermore, the deformation trend (expansion or shrinkage), deformation rate, and corresponding blind zone volume change correlation of each grid are extracted from the compensated results, and this information is integrated into a risk feature vector containing three-dimensional parameters.

[0091] Furthermore, a correspondence is established between each parameter in the risk feature vector and the spatial grid of the area in front of the coal mining machine, with each spatial grid corresponding to a risk feature vector. Based on the deformation trend parameter in the vector, if the trend is expanding, the basic probability of the grid becoming a risk area increases by 0.3; based on the deformation rate parameter, the faster the rate, the higher the basic probability increases by 0.2-0.4 (0.4 when the rate is fastest); based on the correlation parameter of blind zone volume change, the higher the correlation (i.e., the more correlated the deformation and blind zone change), the higher the basic probability increases by 0.1-0.3.

[0092] Furthermore, the adjusted probability values ​​(totaling no more than 1) are assigned to the corresponding spatial grids, and the probability values ​​of all grids are distributed according to their spatial location to form an initial probability density distribution.

[0093] Furthermore, the spatial grid of the initial probability density distribution is examined, and the probability value difference between adjacent grids (referring to grids sharing a face) is calculated. If the difference exceeds 0.2, it is determined to be a boundary discontinuity region.

[0094] Furthermore, for discontinuous areas, the original grid is divided into smaller 0.5m × 0.5m × 0.5m subgrids. Within the subgrids, a transitional allocation is performed based on the probability values ​​of adjacent grids—for example, the probability of the left grid is 0.8, the probability of the right grid is 0.5, and the probability values ​​of the middle subgrids are allocated in a gradient of 0.7 and 0.6.

[0095] Furthermore, after refining and transitioning the meshes of all discontinuous boundary regions as described above, the original large meshes are deleted, and the refined sub-meshes and their corresponding probability values ​​are retained. At this point, the probability value difference between adjacent sub-meshes is controlled within 0.1, thus obtaining the initial probability density distribution of the blind zone to eliminate discontinuous boundary regions.

[0096] Furthermore, taking each spatial grid in the initial probability density distribution as the center, we select neighboring grids within a 3x3x3 radius around it (a total of 26 neighboring grids). We calculate the sum of the probability values ​​of the center grid and its neighboring grids, and then divide this sum by the total number of grids (27) to obtain the smoothed probability value of the center grid. We perform the above operation sequentially on all grids in the initial probability density distribution. For grids located at the boundary and with fewer than 26 neighboring grids, we only select the actual neighboring grids to calculate the average probability value.

[0097] Furthermore, the regions where the probability values ​​originally changed abruptly were adjusted to a gradual state. In the final blind zone probability density distribution map, the probability value of each grid can smoothly reflect the possibility of that location becoming a blind zone.

[0098] In summary, predicting the probability density distribution of blind zones in a mine based on the motion state and the dynamic blind zone volume change rate of the coal mining machine can significantly improve the accuracy and dynamic fit of blind zone distribution prediction.

[0099] In summary, the motion status integrates key information from the cutterhead coordinates and the hydraulic support status, providing real-time three-dimensional position of the coal mining machine and associated working face environmental characteristics, thus offering a precise spatial reference for blind zone prediction. The dynamic blind zone volume change rate, on the other hand, captures the dynamic evolution of the blind zone as mining operations progress by observing the changes in blind zone volume over a continuous period. The combination of these two metrics ensures that the initial probability density distribution both closely matches the actual working conditions of the coal mining machine and reflects the real-time changing trend of the blind zone.

[0100] In summary, spatial smoothing of the initial probability density distribution can eliminate interference such as boundary discontinuities, generating a blind zone probability density distribution map that more closely reflects the actual spatial distribution characteristics. This distribution map can provide a reliable spatial basis for the subsequent output of monitoring point coordinate sequences and the planning of scanning trajectories for mobile monitoring equipment, ensuring that monitoring resources are precisely tilted towards high-probability blind zone areas, thereby improving the overall effectiveness of blind zone monitoring and assessment from a core perspective.

[0101] S3. Output the coordinate sequence of monitoring points in the mine according to the probability density distribution of the blind zone, and plan the scanning trajectory of the mobile monitoring equipment in the coal mining machine based on the coordinate sequence of the monitoring points;

[0102] In this embodiment of the invention, the step of outputting the coordinate sequence of monitoring points in the mine according to the probability density distribution of the blind zone, and planning the scanning trajectory of the mobile monitoring equipment in the coal mining machine based on the coordinate sequence of the monitoring points, includes:

[0103] Extract the spatial boundary coordinates of the high-probability blind zone regions from the blind zone probability density distribution;

[0104] Initial monitoring points are randomly generated within the spatial boundary coordinates;

[0105] Based on the real-time motion status of the coal mining machine and the roof stress distribution data, the initial monitoring point is dynamically optimized to obtain the coordinate sequence of the initial monitoring point after dynamic path optimization.

[0106] The scanning trajectory of the mobile monitoring device is generated based on the coordinate sequence fitting.

[0107] Specifically, in the probability density distribution map of the blind zone, spatial grids with a probability value greater than 0.7 are defined as high-probability blind zone areas. These grids are the core objects for subsequent extraction of boundary coordinates.

[0108] Furthermore, the spatial grid of all high-probability blind zone regions is traversed, and the three-dimensional coordinate range of each grid (i.e., the minimum and maximum coordinate values ​​of each grid in the X, Y, and Z axes) is recorded. From these records, the minimum and maximum coordinates in the X-axis, Y-axis, and Z-axis directions are extracted. These six coordinate values ​​together constitute the spatial boundary coordinates of the high-probability blind zone regions. The minimum and maximum coordinates in the X-axis determine the left and right boundaries of the region, the minimum and maximum coordinates in the Y-axis determine the front and rear boundaries of the region, and the minimum and maximum coordinates in the Z-axis determine the upper and lower boundaries of the region.

[0109] Furthermore, using the spatial boundary coordinates of the high-probability blind zone as a range, randomly generated intervals are defined between the minimum and maximum coordinates in the X-axis direction, the Y-axis direction, and the Z-axis direction. Three-dimensional coordinate points are generated within these intervals using a random number generation tool, ensuring that each coordinate point falls within the high-probability blind zone (i.e., the coordinate values ​​do not exceed the determined spatial boundary coordinate range).

[0110] Furthermore, the number of initial monitoring points is determined based on the volume of the high-probability blind zone area. Five monitoring points are generated per cubic meter of area. After generation, the three-dimensional coordinates of each monitoring point are recorded, and these coordinates are the initial monitoring points.

[0111] Furthermore, the real-time travel direction, travel speed, and current three-dimensional coordinates are extracted from the coal mining machine's motion status to determine the expected travel path of the coal mining machine in the next 10 seconds (a straight path extending along the travel direction and according to the travel speed). Stress distribution data of the roof is obtained by stress sensors installed on the roof, and high-stress areas where the stress value exceeds the safety threshold (80% of the maximum stress that the roof can withstand in a pre-set manner) are marked.

[0112] Furthermore, the coordinates of the initial monitoring point are compared with the expected travel path. If the monitoring point is too close to the expected path (less than 0.5 meters), it is adjusted 0.3 meters away from the path. The coordinates of the monitoring point are compared with the high stress area. If the monitoring point is in the high stress area, it is moved 0.4 meters to the low stress area (stress value is less than 50% of the safety threshold).

[0113] Furthermore, the distribution density of monitoring points is adjusted according to the travel speed. The faster the speed, the smaller the spacing between monitoring points along the travel direction is by 0.2 meters to ensure timely tracking of blind spot changes. The coordinates of the adjusted monitoring points are arranged in chronological order to form the coordinate sequence of the initial monitoring points after dynamic path optimization.

[0114] Furthermore, the coordinate sequence of the initial monitoring points after dynamic path optimization is sorted according to the generation time, and straight line segments are drawn with the coordinates of two adjacent monitoring points as endpoints. For corners formed by three or more consecutive monitoring points (where the angle between adjacent straight line segments is less than 90 degrees), a transition point is added at the corner. The coordinates of the transition point are the midpoint between the coordinates of the two monitoring points before and after the corner, so that the trajectory at the corner changes from a broken line to a smooth curve.

[0115] Furthermore, all straight lines and transition curves are connected to ensure that there are no obvious breaks between adjacent trajectory segments. The resulting continuous trajectory is the scanning trajectory of the mobile monitoring device. This trajectory covers all optimized monitoring points and can be adjusted in real time to follow changes in high-probability blind spots.

[0116] In summary, outputting the coordinate sequence of monitoring points based on the probability density distribution of blind zones and planning the scanning trajectory of mobile monitoring equipment can make the selection of monitoring points more targeted.

[0117] In summary, by extracting the spatial boundary coordinates of high-probability blind zone areas to generate initial monitoring points, it is possible to ensure that monitoring resources are prioritized for areas with higher blind zone risk, avoid ineffective investment of monitoring resources in low-probability areas, and improve the accuracy of monitoring.

[0118] In summary, dynamic path optimization by combining the real-time motion status of the coal mining machine with the stress distribution data of the roof can generate scanning trajectories with monitoring coverage and total path length as optimization objectives, while meeting the equipment's obstacle avoidance constraints and avoiding high-risk restricted areas.

[0119] In summary, this process ensures the comprehensiveness of the monitoring range while reducing unnecessary path consumption, enabling mobile monitoring equipment to efficiently acquire perception data of key areas. This provides high-quality data support for subsequent 3D topology construction and blind spot feature extraction, thereby improving the reliability of blind spot assessment.

[0120] S4. Perform projection transformation on the multi-source sensing data under the scanning trajectory to obtain the three-dimensional topological working face of the mine;

[0121] In this embodiment of the invention, the dynamic path optimization of the initial monitoring point based on the real-time motion state of the coal mining machine and the roof stress distribution data includes:

[0122] Based on the pose parameters of the coal mining machine, the equipment obstacle avoidance constraints of the coal mining machine are generated;

[0123] Areas exceeding a preset threshold in the roof stress distribution diagram of the coal mining machine are marked as high-risk restricted areas;

[0124] A three-dimensional path optimization objective function is constructed for the initial monitoring point based on monitoring coverage, total path length, and restricted area avoidance rate.

[0125] The optimization objectives are to maximize monitoring coverage, minimize total path length, and minimize restricted area avoidance rate. Multi-objective path optimization is performed on the initial monitoring points to obtain a dynamically optimized coordinate sequence of the initial monitoring points, including:

[0126] A three-dimensional path optimization objective function is constructed for the initial monitoring point based on monitoring coverage, total path length, and restricted area avoidance rate.

[0127] Based on the kinematic constraints of the coal mining machine, the three-dimensional path optimization objective function is solved by joint optimization to obtain the Pareto optimal solution set of the initial monitoring point;

[0128] Discretize the path scheme with the highest comprehensive fitness in the Pareto optimal solution set to obtain the coordinate sequence of monitoring points after dynamic path optimization.

[0129] The objective function for the three-dimensional path optimization is as follows:

[0130]

[0131] In the formula, For the path scheme, The weight of the monitoring coverage, The number of blocks in the high-probability blind zone region. For the first One monitoring point, For the first Blind zone probability density weights for each monitoring point For the first Effective coverage coefficient of monitoring points The weight is the total length of the path. The total number of the initial monitoring points. For the first One monitoring point, For the first The three-dimensional spatial coordinates of each monitoring point For the first The three-dimensional spatial coordinates of each monitoring point The weight of the restricted area avoidance rate, The number of restricted blocks, For the trajectory in the restricted area The cumulative time spent inside, Forbidden Zone Spatial conflict penalty items, This is the formula for calculating Euclidean distance. This represents the total number of restricted blocks.

[0132] The projection transformation of the multi-source sensing data under the scanning trajectory to obtain the three-dimensional topological working face of the mine includes:

[0133] Spatiotemporal registration of lidar scanning data and camera image data;

[0134] The registered lidar scanning data and camera image data are used to generate a 3D point cloud through perspective projection transformation.

[0135] The three-dimensional point cloud is reconstructed to obtain the three-dimensional topological working surface of the mine.

[0136] Specifically, geometric dimensions such as the length, width, and height of the coal mining machine are extracted from its pose parameters. These dimensions, combined with the machine's tilt and rotation angles, determine the actual area occupied by the coal mining machine in three-dimensional space. Using the geometric center of the coal mining machine as a reference point, a safety protection boundary is formed by extending 0.5 meters forward and backward along the length direction, 0.3 meters left and right along the width direction, and 0.2 meters up and down along the height direction. It is stipulated that if any object enters this protection boundary, the coal mining machine must stop or change direction; this protection boundary serves as the obstacle avoidance constraint.

[0137] Furthermore, the stress value of each monitoring point is extracted from the stress distribution map of the roof slab, and the stress value is compared one by one with a preset threshold (a pre-set upper limit of safe stress). For monitoring points whose stress values ​​exceed the preset threshold, their three-dimensional coordinates are marked and the corresponding stress values ​​are recorded. A cubic region is formed by expanding outwards by 0.5 meters from these marked points. All expanded cubic regions are merged to form a continuous high-risk restricted area.

[0138] Furthermore, the total volume of high-probability blind spots and the volume covered by the three-dimensional path of the initial monitoring point are statistically analyzed, and the monitoring coverage rate is obtained by dividing the coverage volume by the total volume.

[0139] Furthermore, the total path length is obtained by summing the straight-line distances between adjacent monitoring points in the three-dimensional path of the initial monitoring points. The path length outside the high-risk restricted area in the three-dimensional path of the initial monitoring points is then calculated and divided by the total path length to obtain the restricted area avoidance rate.

[0140] Furthermore, by combining the three indicators of monitoring coverage, total path length, and restricted area avoidance rate, and requiring the monitoring coverage to be as high as possible, the total path length to be as short as possible, and the restricted area avoidance rate to be as high as possible, the resulting comprehensive evaluation standard becomes the three-dimensional path optimization objective function for the initial monitoring points.

[0141] Specifically, the monitoring coverage rate is calculated as the ratio of the volume of the high-probability blind zone covered by the 3D path of the initial monitoring point to the total volume of the high-probability blind zone; a higher ratio indicates better coverage. The total path length is obtained by summing the straight-line distances between all adjacent monitoring points in the 3D path; a shorter distance indicates a more efficient path. The restricted area avoidance rate is calculated as the ratio of the path length that does not enter the high-risk restricted area to the total path length; a higher ratio indicates better avoidance of the restricted area. A 3D path optimization objective function is constructed based on these three indicators. This function requires improving the monitoring coverage rate (target value not less than 90%) while controlling the total path length (target value not exceeding 1.5 times the longest diagonal of the high-probability blind zone) and ensuring the restricted area avoidance rate (target value not less than 95%). These three indicators work together to form a comprehensive optimization standard for the 3D path of the initial monitoring point.

[0142] Furthermore, the kinematic constraints of the coal mining machine include the maximum turning angle (not exceeding 30 degrees, i.e., the included angle between adjacent line segments in the path shall not exceed 30 degrees), the path curvature corresponding to the maximum moving speed (the turning angle change per meter of path shall not exceed 10 degrees), and the minimum turning radius (not less than 0.8 meters, i.e., the radius of the arc at the bend of the path shall not be less than 0.8 meters).

[0143] Furthermore, these constraints are substituted into the three-dimensional path optimization objective function, and each possible path scheme is checked to see if it meets the constraints: if there are line segments with an angle greater than 30 degrees in the path, the scheme is directly excluded; if the path curvature exceeds 10 degrees per meter, the path turning angle is adjusted until it meets the requirements; if the turning radius is less than 0.8 meters, a transition line segment is added at the turning point to make the radius meet the standard. For the path schemes that meet the constraints, their monitoring coverage, total path length, and restricted area avoidance rate are compared, and schemes that cannot improve one indicator without decreasing another are selected. These schemes together constitute the Pareto optimal solution set for the initial monitoring points.

[0144] Furthermore, for each path scheme in the Pareto optimal solution set, scores are assigned based on monitoring coverage (weight 40%), total path length (weight 30%), and restricted area avoidance rate (weight 30%): 100% monitoring coverage earns 40 points, with 0.4 points deducted for each 1% decrease; 80% of the target total path length earns 30 points, with 3 points deducted for each 10% increase; and 100% restricted area avoidance rate earns 30 points, with 0.3 points deducted for each 1% decrease.

[0145] Furthermore, calculate the total score for each solution (the sum of the three scores), and the solution with the highest total score is the path solution with the highest overall fitness.

[0146] Furthermore, the continuous path of the scheme is segmented at intervals of 0.5 meters, and the three-dimensional coordinates of each segment point are the discretized coordinates of the monitoring point. These coordinates are arranged in the order of the path travel to form a sequence of coordinates of the monitoring points after dynamic path optimization.

[0147] Specifically, the lidar and camera are mounted on the same fixed support of the coal mining machine, with both facing the same direction and at a fixed distance (the distance value is measured and recorded in advance).

[0148] Furthermore, both are connected to the same high-precision clock module to ensure that all data carries a timestamp accurate to the millisecond.

[0149] Furthermore, three fixed markers within the mine (such as roof anchor bolts and floor reference piles) were selected as common observation points. The lidar recorded the three-dimensional coordinates and corresponding timestamps of these markers, and the camera captured images containing these markers and recorded the capture timestamps.

[0150] Furthermore, by matching the lidar data and camera image data at the same time using timestamps, and then using the common observation point as a reference, the pixel coordinates of the camera image are converted into coordinates in the lidar coordinate system according to the installation distance between the two (for example, if the camera is 0.3 meters to the right of the lidar, the X coordinate of the marker in the image needs to be subtracted by 0.3 meters to match the lidar coordinates), thus completing the spatiotemporal registration.

[0151] Furthermore, the three-dimensional coordinates (X, Y, Z axis coordinates) of each scanning point are extracted from the registered LiDAR scanning data, and the pixel color information (RGB value) of the corresponding position is extracted from the registered camera image data.

[0152] Furthermore, based on the principle of perspective projection, the two-dimensional pixel coordinates of the camera image are transformed into three-dimensional space through calibrated intrinsic parameters (camera focal length, pixel size) and extrinsic parameters (relative position with the LiDAR), so that each pixel in the image corresponds one-to-one with the three-dimensional coordinates of the LiDAR scanning point.

[0153] Furthermore, the RGB color information of the corresponding pixel is assigned to the three-dimensional coordinates of the LiDAR scanning point, and the set of all three-dimensional coordinates with color information forms a three-dimensional point cloud.

[0154] Furthermore, the 3D point cloud is preprocessed to remove outlier points whose coordinates exceed the actual range of the mine (such as isolated points more than 50 meters away from the coal mining machine), and noise points are removed through neighborhood checks (if the average distance between a point and the 10 nearest surrounding points exceeds 0.2 meters, it is identified as a noise point and deleted).

[0155] Furthermore, a triangulation method is used to connect the points in the point cloud: taking each point as the center, select the three closest surrounding points. If the area of ​​the triangle formed by these four points is within the range of 0.01-0.1 square meters (which conforms to the structural dimensions of the mine working face), then connect these three points to form a triangular element.

[0156] Furthermore, the above operation is performed sequentially on all points to make the edges of adjacent triangular facets coincide, forming a continuous curved surface.

[0157] Furthermore, the connection relationships of key structures such as the roof, floor, and coal wall are extracted from the curved surface (such as the intersection line between the roof and the coal wall, and the connection boundary between the floor and the roadways on both sides), the position of each structure and their interconnection relationship are clarified, and finally the three-dimensional topological working face of the mine containing the structural connection relationships is obtained.

[0158] Specifically, in the three-dimensional path optimization objective function, , , The values ​​are determined based on the needs of mine operations. If the current mining phase requires prioritizing comprehensive monitoring, Set it to 0.5. Set it to 0.2. Set it to 0.3; if path efficiency needs to be prioritized, Set it to 0.5. and Each is set to 0.25; if risk avoidance needs to be prioritized, Set it to 0.5. Set it to 0.3. Set it to 0.2. Determined by the three-dimensional structure of the high-probability blind zone, the area is divided into cubes of 0.5m × 0.5m × 0.5m. The total volume is divided by the volume of a single block (0.125 cubic meters) to obtain the number of blocks. Taken from the probability density distribution map of the blind zone The probability value of each block (range 0-1), the higher the probability, The larger.

[0159] Furthermore, This is determined by whether the monitoring point falls within the corresponding segment. If the coordinates of the monitoring point are within the three-dimensional coordinate range of the segment, It is 1 if it is true, otherwise it is 0.

[0160] Furthermore, Based on the total volume of the high-probability blind zone, and with one monitoring point set up for every 0.125 cubic meters, the total number of integral blocks is [number missing]. .

[0161] Furthermore, and The monitoring point coordinate sequence derived from dynamic path optimization consists of three-dimensional coordinate values ​​after obstacle avoidance and efficiency optimization.

[0162] Furthermore, The high-risk restricted area is divided into three-dimensional blocks, with the block size matching the high-probability blind zone. The total number of blocks is [number missing]. .

[0163] Furthermore, The dwell time is obtained by calculating the overlap length between the path and the restricted area block and dividing the overlap length by the scanning speed of the mobile monitoring device (fixed at 0.2 m / s).

[0164] Furthermore, Based on the stress value setting of the restricted area, R_k is set to 2 when the stress value exceeds 1.5 times the safety threshold, set to 1.5 when it exceeds 1-1.5 times, and set to 1 when it does not exceed 1 time.

[0165] Furthermore, by superimposing the values ​​from the three parts, the overall performance of the path scheme is comprehensively evaluated. The smaller the value, the better the path.

[0166] Furthermore, in the first part, 1 minus The results reflect the proportion of blind spots that are not effectively covered, multiplied by Then, the importance of monitoring coverage is emphasized—if a certain high-probability segment ( Large areas were not covered by monitoring points. =0), this part of the value will increase significantly, forcing the path to cover key areas. In the second part, the sum of the Euclidean distances between adjacent monitoring points directly corresponds to the actual path length, multiplied by Then, ensure that the path does not detour excessively in pursuit of coverage; the shorter the distance, the less this part "drags down" the overall result.

[0167] Furthermore, in Part Three, The risk cost of staying in the restricted area is reflected in the longer the stay, the higher the risk of staying in the restricted area. The larger the value of this component (e.g., higher), the more the path is encouraged to avoid high-risk areas. The sum of these three factors simultaneously reflects the path's performance in terms of coverage of key areas, actual walking distance, and risk avoidance.

[0168] Furthermore, when monitoring points cover more high-probability blocks ( The number of =1 increases, and the coverage... (larger) The sum will increase; subtracting the sum from 1 will decrease the result, the first part of the value will decrease, and F_total will decrease accordingly. This is because the 3D coordinates of adjacent monitoring points are closer (e.g., distributed along a straight line rather than a broken line). The sum will decrease, and the value of the second part will decrease. Decrease.

[0169] Furthermore, when the path completely avoids high-risk restricted areas ( =0), or only briefly passing through the low-risk restricted area ( Small and (small), the third part's value is close to 0. Decrease.

[0170] Furthermore, if a monitoring point misses multiple high-probability segments, the first part of the value will increase significantly; if the path frequently turns back and forth, causing the total distance between adjacent points to increase, the second part of the value will increase; if the path delves deep into high-risk restricted areas and stays there for a long time, the third part of the value will increase significantly. All three situations will lead to... Increase.

[0171] In summary, projecting and transforming multi-source sensing data under the scanning trajectory to obtain the three-dimensional topological working face of the mine can effectively improve the spatial consistency and effectiveness of the data.

[0172] In summary, by performing spatiotemporal registration of LiDAR scanning data and camera image data, the deviations of multi-source data in the time and space dimensions can be eliminated, ensuring the synergy of different types of sensing data. Then, by generating a three-dimensional point cloud through perspective projection transformation and performing surface reconstruction, the planar and scattered sensing data can be transformed into a three-dimensional spatial structure model that fits the actual working conditions, accurately restoring the three-dimensional shape of the coal mining face.

[0173] In summary, the constructed 3D topological working surface provides a precise spatial carrier for subsequent blind zone feature extraction.

[0174] In summary, its complete and realistic spatial structure ensures more accurate extraction of the void features of the connected domain, avoiding misjudgment of features due to data distortion or insufficient spatial morphology restoration. This provides a reliable basis for subsequent determination of the blind zone volume ratio and spatial distribution entropy value, further ensuring the accuracy of the blind zone assessment results.

[0175] S5. Extract the connected domain void features in the three-dimensional topological working face, and determine the blind zone volume ratio and spatial distribution entropy value of the mine based on the connected domain void features.

[0176] In this embodiment of the invention, the step of extracting the connected domain cavity features in the three-dimensional topological working face and determining the blind zone volume ratio and spatial distribution entropy value of the mine based on the connected domain cavity features includes:

[0177] The three-dimensional topological working surface is divided into sub-regions, and the connected domain holes in the sub-regions are identified;

[0178] The total mineable volume of the three-dimensional topological working face is obtained. Based on the ratio of the total volume of the connected domain cavities to the total mineable volume, the proportion of the blind zone volume of the mine is obtained.

[0179] The blind zone probability density value of the sub-region is determined based on the characteristics of the connected domain voids;

[0180] The spatial distribution entropy value of the mine is obtained by integrating the probability density value of the blind zone based on the number of the sub-regions.

[0181] Specifically, in the three-dimensional topology working surface, sub-regions are divided according to a cube size of 2 meters × 2 meters × 2 meters. The boundaries of each sub-region are clearly defined by three-dimensional coordinates (e.g., X-axis 0-2 meters, Y-axis 0-2 meters, Z-axis 0-2 meters is the first sub-region, and subsequent regions are extended in sequence according to this size).

[0182] Furthermore, each sub-region is traversed to check if there is a continuous space not covered by the 3D point cloud. If there is no point cloud data in a certain space region and the region is connected to the surrounding uncovered regions (connectivity is determined by checking whether the adjacent cubes are also uncovered regions), it is determined to be a connected domain hole, and the 3D coordinate range and volume of each connected domain hole are recorded.

[0183] Furthermore, the total recoverable volume of the three-dimensional topology working face is determined by calculating the volume of the coal seam region in the working face: the three-dimensional boundary of the coal seam is extracted from the three-dimensional topology working face (excluding non-coal seam structures such as roof and floor), the number of cubes in which the coal seam is located is counted according to the size of the sub-region, and the volume of each cube is 8 cubic meters. The total recoverable volume is obtained by multiplying the number of cubes by the volume of each individual cube.

[0184] Furthermore, the volumes of all connected domain cavities are counted and summed to obtain the total volume of connected domain cavities. The total volume of connected domain cavities is then divided by the total mineable volume to obtain the percentage of the mine's blind zone volume (the result is rounded to two decimal places).

[0185] Furthermore, three key indicators were extracted from the characteristics of the cavity in the connected domain: cavity volume (the larger the cavity, the more obvious the blind zone), distance between the cavity and the current position of the coal mining machine (the closer the cavity, the greater the impact on mining), and the connectivity range of the cavity (the more sub-regions connected, the more concentrated the blind zone).

[0186] Furthermore, scoring criteria are set for each indicator: 3 points for volume exceeding 5 cubic meters, 2 points for 3-5 cubic meters, and 1 point for less than 3 cubic meters; 3 points for distance less than 10 meters, 2 points for 10-20 meters, and 1 point for greater than 20 meters; 3 points for more than 5 connected sub-regions, 2 points for 3-5, and 1 point for less than 3. The scores of the three indicators are added together (total score 3-9 points), and the score is divided by 9 to obtain the blind zone probability density value of that sub-region (range 0-1, the larger the value, the higher the probability that the sub-region is a blind zone).

[0187] Furthermore, the blind zone probability density values ​​of each sub-region are arranged sequentially according to their position in three-dimensional space, covering all sub-regions in the order of left to right, front to back, and top to bottom. The integral operation is achieved by calculating the sum of the blind zone probability density values ​​of all sub-regions: first, the probability density value of each sub-region is multiplied by the volume (8 cubic meters) of that sub-region to obtain the probability contribution value of each sub-region; then, the probability contribution values ​​of all sub-regions are added together, and the sum is the spatial distribution entropy value of the mine, which reflects the uniformity of the blind zone probability distribution throughout the working face.

[0188] In summary, extracting the void features of connected domains in the three-dimensional topological working face and determining the volume ratio and spatial distribution entropy of the blind zone in the mine can achieve accurate quantification of the blind zone and effective characterization of its spatial distribution features.

[0189] In summary, the three-dimensional topological working face is generated by projection transformation of multi-source sensing data, which can accurately restore the three-dimensional shape of the coal mining face. The cavity features of the connected domain directly correspond to the spatial shape of the actual blind area, and its extraction can accurately capture key information such as the location, size and outline of the blind area.

[0190] In summary, determining the proportion of blind zone volume by the ratio of the total volume of voids in the connected domain to the total mineable volume can intuitively quantify the scale of blind zone proportion in the working face; the spatial distribution entropy value obtained by integrating the probability density value of blind zone in the sub-region can effectively reflect the spatial dispersion or aggregation state of blind zone.

[0191] In summary, the combination of these two methods provides a comprehensive and accurate assessment of blind spots from both quantitative proportion and spatial distribution dimensions, thereby enhancing the reliability and practicality of blind spot assessment in coal mining faces.

[0192] In the several embodiments provided by this invention, it should be understood that the disclosed method can be implemented in other ways.

[0193] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0194] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, and technology that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0195] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for monitoring and evaluating blind spots in coal mining faces, characterized in that, The method includes: S1. Couple the coordinate sequence of the cutterhead in the coal mining machine with the support state of the hydraulic support to form the motion state of the coal mining machine; S2. Predicting the blind zone probability density distribution of the coal mining machine based on the motion state and the dynamic blind zone volume change rate of the coal mining machine, including: Based on the three-dimensional position coordinates of the motion state, point cloud data of the area to be mined in front of the coal mining machine is acquired in real time; The blind zone volume of the coal mining machine is obtained based on the point cloud data, and the dynamic blind zone volume change rate of the coal mining machine is generated based on the change of the blind zone volume over a continuous period of time. Predict the initial probability density distribution of the blind zone in the coal mining machine based on the motion state and the dynamic blind zone volume change rate; The initial probability density distribution is spatially smoothed to generate the blind zone probability density distribution map of the coal mining machine; S3. Output the coordinate sequence of monitoring points in the mine based on the probability density distribution of the blind zone, and plan the scanning trajectory of the moving monitoring equipment in the coal mining machine based on the monitoring point coordinate sequence, including: Extract the spatial boundary coordinates of the high-probability blind zone regions from the blind zone probability density distribution; Initial monitoring points are randomly generated within the spatial boundary coordinates; Based on the real-time motion status of the coal mining machine and the roof stress distribution data, the initial monitoring point is dynamically optimized to obtain the coordinate sequence of the initial monitoring point after dynamic path optimization. The scanning trajectory of the mobile monitoring device is generated based on the coordinate sequence fitting. S4. Perform projection transformation on the multi-source sensing data under the scanning trajectory to obtain the three-dimensional topological working face of the mine; S5. Extract the connected domain void features in the three-dimensional topological working face, and determine the blind zone volume ratio and spatial distribution entropy value of the mine based on the connected domain void features.

2. The blind zone monitoring and assessment method for coal mining faces as described in claim 1, characterized in that, The coupling of the coordinate sequence of the cutterhead in the coal mining machine with the support state of the hydraulic support to form the motion state of the coal mining machine includes: The real-time coordinate sequence of the tool head is collected by a positioning sensor; Simultaneously acquire the support status of the hydraulic support; The real-time coordinate sequence is spatiotemporally aligned with the support state, and the spatiotemporally aligned real-time coordinate sequence is weighted and fused with the support state to obtain the motion state of the coal mining machine.

3. The blind zone monitoring and assessment method for coal mining faces as described in claim 1, characterized in that, The prediction of the initial probability density distribution of the blind zone in the coal mining machine based on the motion state and the dynamic blind zone volume change rate includes: The dynamic geological deformation field of the area in front of the coal mining machine is generated based on the travel velocity vector and three-dimensional coordinates in the motion state. Residual compensation is performed on the dynamic geological deformation field and the dynamic blind zone volume change rate to obtain the risk feature vector of the blind zone; Spatial probability mapping is performed on the risk feature vector to obtain the initial probability density distribution of the blind zone; The initial probability density distribution is dynamically meshed to eliminate discontinuous regions at the boundaries of the initial probability density distribution after dynamic mesh optimization, thereby obtaining the initial probability density distribution of the blind zone.

4. The method for monitoring and evaluating blind spots in coal mining faces as described in claim 1, characterized in that, The dynamic path optimization of the initial monitoring point based on the real-time motion state of the coal mining machine and the roof stress distribution data includes: Based on the pose parameters of the coal mining machine, the equipment obstacle avoidance constraints of the coal mining machine are generated; Areas exceeding a preset threshold in the roof stress distribution diagram of the coal mining machine are marked as high-risk restricted areas; With the optimization objectives of maximizing monitoring coverage, minimizing total path length, and avoiding restricted areas, multi-objective path optimization is performed on the initial monitoring points to obtain the coordinate sequence of the initial monitoring points after dynamic path optimization.

5. The blind zone monitoring and evaluation method for coal mining faces as described in claim 4, characterized in that, The optimization objectives are to maximize monitoring coverage, minimize total path length, and minimize restricted area avoidance rate. Multi-objective path optimization is performed on the initial monitoring points to obtain a dynamically optimized coordinate sequence of the initial monitoring points, including: A three-dimensional path optimization objective function for the initial monitoring points is constructed based on monitoring coverage, total path length, and restricted area avoidance rate. Based on the kinematic constraints of the coal mining machine, the three-dimensional path optimization objective function is solved by joint optimization to obtain the Pareto optimal solution set of the initial monitoring point; Discretize the path scheme with the highest comprehensive fitness in the Pareto optimal solution set to obtain the coordinate sequence of monitoring points after dynamic path optimization.

6. The blind zone monitoring and assessment method for coal mining faces as described in claim 5, characterized in that, The objective function for the three-dimensional path optimization is as follows: ； In the formula, For the path scheme, The weight of the monitoring coverage, The number of blocks in the high-probability blind zone region. For the first One monitoring point, For the first Blind zone probability density weights for each monitoring point For the first Effective coverage coefficient of monitoring points The weight is the total length of the path. The total number of the initial monitoring points. For the first One monitoring point, For the first The three-dimensional spatial coordinates of each monitoring point For the first The three-dimensional spatial coordinates of each monitoring point The weight of the restricted area avoidance rate, The number of restricted blocks, For the trajectory in the restricted area The cumulative time spent inside, Forbidden Zone Spatial conflict penalty items, This is the formula for calculating Euclidean distance. This represents the total number of restricted blocks.

7. The blind zone monitoring and assessment method for coal mining faces as described in claim 1, characterized in that, The projection transformation of the multi-source sensing data under the scanning trajectory to obtain the three-dimensional topological working face of the mine includes: Spatiotemporal registration of lidar scanning data and camera image data; The registered lidar scanning data and camera image data are used to generate a 3D point cloud through perspective projection transformation. The three-dimensional point cloud is reconstructed to obtain the three-dimensional topological working surface of the mine.

8. The method for monitoring and evaluating blind spots in coal mining faces as described in claim 1, characterized in that, The step of extracting the connected domain void features in the three-dimensional topological working face and determining the blind zone volume ratio and spatial distribution entropy value of the mine based on the connected domain void features includes: The three-dimensional topological working surface is divided into sub-regions, and the connected domain holes in the sub-regions are identified; Obtain the total mineable volume of the three-dimensional topological working face, and obtain the blind zone volume ratio of the mine based on the ratio of the total volume of the connected domain cavities to the total mineable volume; The blind zone probability density value of the sub-region is determined based on the characteristics of the connected domain voids; The spatial distribution entropy value of the mine is obtained by integrating the probability density value of the blind zone based on the number of the sub-regions.

Images