Accurate positioning method and system based on geological mineral three-dimensional modeling

Abstract

The invention relates to the technical field of geological information, and discloses a precise positioning method and system based on geological mineral product three-dimensional modeling, and the method comprises the steps: deploying a laser ranging array, a micro-seismic array and a fiber grating sensing array, and synchronously collecting multi-source deformation and stress data; after space-time alignment and filtering, an improved ICP algorithm is adopted to reconstruct a millimeter-level precision three-dimensional point cloud model; a safety coefficient cloud picture is calculated by combining a geomechanical constitutive equation, and a risk area is automatically delineated according to a three-level threshold value; and closed-loop control instructions such as voice alarm, thermodynamic diagram pushing, grouting coordinates and avoiding paths are generated. The system is composed of a sensing deployment unit, a data preprocessing unit, a point cloud reconstruction unit, a mechanical analysis unit, a risk delineation unit and an instruction distribution and execution unit. Millimeter-level positioning, sub-hour-level early warning and full-automatic intervention of goaf deformation are achieved, the early warning accuracy rate exceeds 95%, the emergency response is shortened to the minute level, and mine safety and mining efficiency are improved.

Description

Technical Field

[0001] This invention belongs to the field of geological information technology, specifically relating to a method and system for precise positioning based on three-dimensional geological and mineral resource modeling. Background Technology

[0002] With the continuous expansion of mining scale and the ongoing development of deep resources, goaf stability monitoring has become a core aspect of ensuring safe production in underground mines. In traditional mining techniques such as room-and-pillar mining and ore-holding mining, the goaf, as a key area for stress redistribution in the surrounding rock, has its morphological evolution and roof settlement directly related to the probability of major disasters such as roof falls and spalling.

[0003] Current mainstream monitoring methods still rely on manual inspections and point-based sensor deployment. The former is limited by the hazardous nature of the working environment and the discreteness of data acquisition, while the latter is difficult to construct a complete three-dimensional spatial situational awareness system due to insufficient spatial coverage density and response lag. Especially in areas with complex geological structures, where rock mass structural surfaces are well-developed and mechanical parameters are significantly heterogeneous, existing technologies cannot achieve spatiotemporal coupling analysis of dynamic changes in goaf morphology and precursor signals of micro-fractures, resulting in a lack of foresight and spatial positioning accuracy in risk warning.

[0004] Among these, the fusion sensing technology based on 3D laser scanning and microseismic monitoring is gradually becoming an important development direction for goaf safety assessment. This technology aims to reconstruct the real-time deformation process and energy release trajectory of the surrounding rock in the goaf through high-precision spatial modeling and continuous microseismic event capture. Its basic principle is to align spatial geometric data and rock fracture signals in a unified coordinate system to identify the stress concentration evolution path and the spatial distribution of potentially unstable blocks. However, existing systems mostly remain at the level of data acquisition and visualization, lacking a deep coupling mechanism with geomechanical models. They cannot correlate the spatial clustering characteristics and energy release patterns of microseismic events with the attitude and strength parameters of rock mass structural surfaces, causing early warning models to remain at the stage of empirical threshold judgment, making it difficult to support dynamic risk evolution prediction and proactive prevention and control decisions.

[0005] Existing technologies for constructing stability assessment systems for goaf areas generally suffer from three major defects: fragmented three-dimensional morphological perception, isolated microseismic signal interpretation, and static geological models. Long update cycles for scanning data and the inability to synchronize with microseismic events lead to a disconnect between morphological changes and the fracturing process; microseismic location results lack mechanical interpretation based on rock mass structural characteristics, resulting in high false alarm rates and an inability to identify key hazardous blocks; geological models are mostly static profiles or simplified blocks, lacking time dimensions and real-time monitoring feedback, thus failing to provide evolutionary extrapolation capabilities for stability prediction. Under complex conditions such as deep, high-stress ground conditions and strong tectonic disturbances, these defects directly prevent the system from forming an effective early warning window before roof collapses occur, and also prevent it from providing spatially accurate decision-making basis for support scheme optimization and personnel evacuation route planning, constituting a technical bottleneck that urgently needs to be overcome in intelligent mine safety management. Summary of the Invention

[0006] This invention provides a method and system for precise positioning based on three-dimensional geological and mineral modeling. By deploying a distributed multimodal sensor array in the goaf area and combining a high-precision time-series point cloud reconstruction algorithm with a dynamic coupling model of geomechanical parameters, a three-dimensional digital twin of the goaf area's full life cycle deformation evolution is constructed. This enables millimeter-level spatial positioning and sub-hour-level time warning of roof settlement rate, surrounding rock stress redistribution path, and local collapse precursor characteristics, thereby establishing an active perception and precise intervention capability for goaf stability risks across the entire three-dimensional spatial domain.

[0007] This invention provides a method for precise positioning based on three-dimensional geological and mineral resource modeling, comprising: Laser ranging sensor arrays, microseismic sensor arrays, and fiber optic strain sensor arrays are deployed at a preset spatial density on the surface of the roof, sidewalls, and key pillars in the goaf. The laser ranging sensor array is used to collect normal displacement data of points on the roof and sidewall surfaces. The microseismic sensor array is used to collect spatial coordinates and energy level data of rock mass fracture events. The fiber optic strain sensor array is used to collect axial and circumferential strain distribution data inside the pillars. The normal displacement data, spatial coordinate and energy level data, and axial and circumferential strain distribution data are synchronously transmitted to the central processing unit through an industrial ring network. The central processing unit performs timestamp alignment, unified spatial coordinate transformation, noise filtering and outlier removal operations on the raw sensor data to form a multi-source heterogeneous monitoring data stream with a unified spatiotemporal reference. Based on the multi-source heterogeneous monitoring data stream, the point cloud dynamic reconstruction module is invoked to reconstruct the three-dimensional point cloud of the internal spatial morphology of the goaf at a preset time interval. The point cloud dynamic reconstruction module adopts an improved iterative nearest point algorithm, introduces geological constraint terms and deformation rate penalty terms, so that the registration error of adjacent period point clouds is controlled within 0.5mm, and outputs a high-fidelity three-dimensional morphological model containing the roof curvature change gradient and the sidewall displacement vector field. The high-fidelity three-dimensional morphological model is input into the geomechanical coupling analysis module. The geomechanical coupling analysis module embeds the rock mass mechanical constitutive equation and damage evolution criteria. Combined with the ore body occurrence conditions, surrounding rock classification parameters, and historical mining disturbance records, it calculates the current three-dimensional stress field distribution, plastic zone expansion boundary, and safety factor spatial cloud map of the goaf. The safety factor spatial cloud map uses any spatial point in the goaf as the calculation unit and outputs the stability margin value of that point under the current load conditions. Based on the stability margin value, an automatic risk area delineation program is initiated. The automatic risk area delineation program sets three-level warning thresholds. When the stability margin value of a spatial point is lower than the first threshold, it is marked as a yellow warning area; when it is lower than the second threshold, it is marked as an orange warning area; and when it is lower than the third threshold, it is marked as a red warning area. The three-level warning thresholds are obtained by calibrating on-site measured data based on the ore type, burial depth, and service life. The spatial coordinate range and deformation rate data of the yellow, orange, and red warning areas are input into the warning instruction generation module. The warning instruction generation module automatically generates a corresponding set of control instructions based on the warning level. The set of control instructions includes sending a voice alarm instruction to the on-site broadcasting system, pushing a three-dimensional risk heat map instruction to the dispatch center, issuing a local grouting reinforcement coordinate instruction to the support operation unit, and sending an avoidance path replanning instruction to the mining equipment. The control command set is distributed to the corresponding execution terminal through a wireless communication link to achieve a closed-loop response to the risk of goaf.

[0008] This invention provides a precise positioning system based on three-dimensional geological and mineral modeling, comprising: A distributed multimodal sensor array deployment unit is used to deploy laser ranging sensor arrays, microseismic sensor arrays, and fiber optic strain sensor arrays on the surface of the roof, sidewalls, and key pillars in the goaf at a preset spatial density. The laser ranging sensor array is used to collect normal displacement data of points on the roof and sidewall surfaces, the microseismic sensor array is used to collect spatial coordinates and energy level data of rock mass fracture events, and the fiber optic strain sensor array is used to collect axial and circumferential strain distribution data inside the pillars. The multi-source data synchronous transmission and preprocessing unit is used to synchronously transmit the normal displacement data, spatial coordinate and energy level data, and axial and circumferential strain distribution data to the central processing unit through the industrial ring network, and perform timestamp alignment, unified spatial coordinate transformation, noise filtering and outlier removal operations on the raw sensor data to form a multi-source heterogeneous monitoring data stream with a unified spatiotemporal reference. The point cloud dynamic reconstruction unit is used to reconstruct the three-dimensional point cloud of the internal spatial morphology of the goaf based on the multi-source heterogeneous monitoring data stream at a preset time interval. The point cloud dynamic reconstruction unit adopts an improved iterative nearest point algorithm, introduces geological constraint terms and deformation rate penalty terms, so that the registration error of adjacent period point clouds is controlled within 0.5mm, and outputs a high-fidelity three-dimensional morphological model including the roof curvature change gradient and the sidewall displacement vector field. The geomechanical coupling analysis unit is used to input the high-fidelity three-dimensional morphological model into the calculation module embedded with the rock mass mechanical constitutive equation and damage evolution criterion. Combined with the ore body occurrence conditions, surrounding rock classification parameters, and historical mining disturbance records, it calculates the current three-dimensional stress field distribution, plastic zone expansion boundary, and safety factor spatial cloud map of the goaf. The safety factor spatial cloud map uses any spatial point in the goaf as the calculation unit and outputs the stability margin value of that point under the current load conditions. The automatic risk area delineation unit is used to set three-level early warning thresholds based on the stability margin value. When the stability margin value of a spatial point is lower than the first threshold, it is marked as a yellow early warning area; when it is lower than the second threshold, it is marked as an orange early warning area; and when it is lower than the third threshold, it is marked as a red early warning area. The three-level early warning thresholds are obtained by calibrating the data from field measurements based on the type of ore, burial depth, and service life. The early warning instruction generation and distribution unit is used to input the spatial coordinate range and deformation rate data of the yellow, orange, and red early warning areas into the instruction generation module, and automatically generate corresponding control instruction sets according to the early warning level. The control instruction sets include sending voice alarm instructions to the on-site broadcasting system, pushing three-dimensional risk heat map instructions to the dispatch center, issuing local grouting reinforcement coordinate instructions to the support operation unit, and sending avoidance path replanning instructions to the mining equipment. The control instruction sets are then distributed to the corresponding execution terminals through a wireless communication link. The closed-loop response execution unit is used to receive and execute the control command set to realize physical intervention and operation process adjustment for goaf risks.

[0009] Furthermore, the measurement range of a single sensor in the laser ranging sensor array is 0.5m to 50m, the repeatability is ±0.1mm, the sampling frequency is 10 times per second, the installation tilt error is less than 0.5°, the sensor housing protection rating is higher than IP67, and the operating temperature range is -20℃ to +70℃.

[0010] Furthermore, the microseismic sensor array employs a three-component velocity-type sensing element with a frequency response range of 2Hz to 1000Hz, a dynamic range greater than 120dB, an event location accuracy better than 1m in three-dimensional space, and an energy resolution capable of distinguishing micro-fracture events from 0.1J to 10J.

[0011] Furthermore, the fiber optic strain sensor array has a grating length of 10 mm, a strain measurement range of -5000 με to +5000 με, a resolution of 0.1 με, a temperature compensation accuracy of ±0.5℃, and achieves more than 64 sensing points connected in series on a single optical fiber through wavelength division multiplexing technology.

[0012] Furthermore, the improved iterative nearest point algorithm in the point cloud dynamic reconstruction module has an objective function that includes three terms: the first term is the sum of squares of traditional point-to-point distances, the second term is the negative logarithm of the cosine of the angle between the normals of corresponding points in adjacent point clouds, and the third term is the square of the displacement vector magnitude of the corresponding point. The weighting coefficients of the three terms are dynamically adjusted according to the current average settlement rate of the goaf area, and the weight allocation strategy is stored in the coefficient adjustment table, which is obtained by offline training based on historical settlement data.

[0013] Furthermore, the rock mass mechanical constitutive equation in the geomechanical coupling analysis module adopts the modified Hoek-Brown criterion, and its expression is as follows: ; in For the maximum principal stress, For the minimum principal stress, denoted as uniaxial compressive strength of the rock block, and denoted as m, the rock mass integrity coefficient, reflecting the influence of rock mass structural integrity on strength. For intact rock masses, the m value is close to the rock block test value (e.g., m≈25 for granite), while for rock masses with well-developed joints, the m value is significantly reduced (e.g., m≈5 for fractured rock masses). Specific values ​​are obtained by referring to tables based on on-site rock mass quality grading (e.g., RMR, Q system) or correcting through indoor rock block tests. denoted as s, the rock mass disturbance coefficient, characterizes the degree to which external factors such as mining disturbance and groundwater soaking weaken the rock mass strength. For undisturbed fresh rock masses, s=1.0; for strongly disturbed or water-saturated rock masses, s≤0.1. Values ​​need to be determined on-site in conjunction with mining methods (e.g., blasting, mechanical excavation) and hydrogeological conditions. The damage evolution criterion adopts a cumulative damage model based on strain energy density, where the damage variable D is defined as the ratio of the current cumulative plastic strain energy to the strain energy corresponding to the peak rock mass strength.

[0014] Furthermore, the three-level early warning thresholds in the automatic risk area delineation program are set as follows: the first threshold is 1.5, the second threshold is 1.2, and the third threshold is 0.9. These thresholds are applicable to goaf areas with medium-hard rock masses, burial depths of less than 500m, and service life of less than 3 years. For soft rock masses or deep mining conditions, the thresholds are calibrated by a combination of on-site fracturing tests and numerical inversion.

[0015] Furthermore, the spatial positioning accuracy requirements of the local grouting reinforcement coordinate command in the early warning command generation module are: horizontal error less than 0.3m, vertical error less than 0.2m, grouting pressure setting range of 0.5MPa to 3MPa, initial setting time of grouting material controlled within 30 minutes, and final setting strength higher than 20MPa.

[0016] Furthermore, the mining equipment avoidance path replanning instruction in the closed-loop response execution unit uses an improved A* algorithm for path generation. The cost function includes three terms: the first term is the path length, the second term is the integral risk value of crossing the risk area, and the third term is the total change in path curvature. The weight coefficients of the three terms are manually set by the scheduler according to the current production urgency.

[0017] Furthermore, the system also includes a historical data backtracking analysis unit, which stores three-dimensional morphological models, stress field distribution maps, early warning records and intervention measure execution logs for all periods. It supports sliding comparison along the time axis, extrapolation prediction of deformation trends, and evaluation of the effectiveness of intervention measures. The data storage period of the historical data backtracking analysis unit is more than 10 years, the data compression ratio is higher than 5:1, and the retrieval response time is less than 3 seconds.

[0018] Compared with the prior art, the advantages and positive effects of the present invention are as follows: This invention, by constructing a collaborative architecture of a distributed multimodal sensor array and a high-precision point cloud dynamic reconstruction algorithm, has for the first time achieved millimeter-level spatial resolution and minute-level temporal resolution monitoring of three-dimensional morphological changes in goaf areas, solving the fundamental deficiency of traditional manual inspection and single-point sensor monitoring in failing to form a full-domain three-dimensional situational awareness.

[0019] By introducing a geomechanical coupling analysis module, a physical correlation is established between apparent deformation data and the internal stress state of the rock mass, upgrading risk warning from experience-based threshold judgment to quantitative prediction based on mechanical mechanisms, and improving the warning accuracy rate to over 95%.

[0020] By linking the three-level early warning threshold with the closed-loop response execution unit, a fully automated process from risk identification to physical intervention is achieved, reducing emergency response time from hours to minutes and effectively preventing roof collapse and sidewall accidents.

[0021] After deployment, in typical metal mine goaf applications, the roof settlement monitoring error is less than 0.8mm, the risk area positioning deviation is less than 0.4m, the average early warning time reaches 4 hours, the consumption of support materials is reduced by 30%, and the mining operation interruption time is reduced by 70%, significantly improving the mine's safety production level and resource extraction efficiency. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the overall technical solution architecture of a precise positioning method and system based on three-dimensional geological and mineral modeling proposed in this invention; Figure 2 This is a schematic diagram of the core principle framework of the collaborative architecture of distributed multimodal sensing array and high-precision temporal point cloud reconstruction algorithm in this invention; Figure 3 This is a flowchart illustrating the preprocessing and spatiotemporal benchmark unification process for multi-source heterogeneous monitoring data streams in this invention. Figure 4 This is a schematic diagram of the stress-deformation joint evolution analysis framework driven by the dynamic coupling model of geomechanical parameters in this invention; Figure 5 This is a diagram of the logic framework for automatic risk area delineation and closed-loop response control driven by a three-level early warning threshold in this invention. Figure 6 This is a schematic diagram of the multi-level interaction relationship and data flow between the digital twin of the goaf area and the execution terminal in this invention; Detailed Implementation

[0023] Please refer to Figure 1-6 Laser ranging sensor arrays, microseismic sensor arrays, and fiber optic strain sensor arrays are deployed at a predetermined spatial density on the roof, sidewalls, and key pillars of the goaf. The laser ranging sensor array is used to collect normal displacement data of points on the roof and sidewall surfaces. The microseismic sensor array is used to collect spatial coordinates and energy level data of rock mass fracturing events. The fiber optic strain sensor array is used to collect axial and circumferential strain distribution data inside the pillars. The measurement range of a single sensor in the laser ranging sensor array is 0.5m to 50m, the repeatability is ±0.1mm, the sampling frequency is 10 times per second, the installation tilt error is less than 0.5°, the sensor housing protection rating is higher than IP67, and the operating temperature range is -20℃ to +70℃.

[0024] The microseismic sensor array employs a three-component velocity sensing element with a frequency response range of 2Hz to 1000Hz, a dynamic range greater than 120dB, an event localization accuracy better than 1m in three-dimensional space, and an energy resolution capable of distinguishing micro-fracture events from 0.1J to 10J. The fiber optic strain sensor array has a grating length of 10mm, a strain measurement range of -5000με to +5000με, a resolution of 0.1με, and a temperature compensation accuracy of ±0.5℃. Wavelength division multiplexing (WDM) technology enables more than 64 sensing points to be connected in series on a single fiber. During deployment, based on the geometry of the goaf and historical collapse records, laser rangefinders are arranged in a grid pattern at 3m intervals along the roof direction and in a staggered, quincunx pattern at 5m intervals in the sidewall area.

[0025] Microseismic sensors are radially deployed within a 20m radius of the mine pillar, ensuring that any rupture event is simultaneously captured by at least three sensors. Fiber optic strain sensors are buried every 2m along the mine pillar axis, with each group containing four circumferentially distributed sensing points to detect uneven stress distribution across the cross-section. All sensors undergo zero-point calibration and temperature drift compensation testing before installation, and are then calibrated using a laser total station to ensure strict alignment of the measurement origin with the mine's global coordinate system.

[0026] The normal displacement data, spatial coordinates and energy level data, and axial and circumferential strain distribution data are synchronously transmitted to the central processing unit via an industrial ring network. The central processing unit performs timestamp alignment, unified spatial coordinate transformation, noise filtering, and outlier removal on the raw sensor data, forming a multi-source heterogeneous monitoring data stream with a unified spatiotemporal reference. Timestamp alignment uses a network time protocol synchronization mechanism, and all sensors have built-in high-stability crystal oscillators, with a timing error of less than 1ms. Unified spatial coordinate transformation is achieved through a pre-stored sensor installation coordinate table and a transformation matrix between the sensor and the mine geodetic coordinate system. The transformation process incorporates a 7-parameter Bursa model to eliminate installation deviations and coordinate system rotation errors.

[0027] Noise filtering is performed in two stages: the first stage is hardware-level filtering, using a combination of moving average and median filtering at the sensor end, with a window length of 10 sampling points; the second stage is software-level filtering, using wavelet thresholding in the central processing unit, selecting the db4 wavelet basis, decomposing at 5 levels, and using a soft thresholding function. Outlier removal employs a dual verification method of the 3-sigma criterion and the isolated forest algorithm. The former removes outliers deviating from the mean by three standard deviations, while the latter identifies sparse outlier clusters in high-dimensional space by constructing a random tree forest. The preprocessed data stream is divided into time slices, each 1 minute long, containing valid measurements from all sensors within that time segment. The data packet header includes a timestamp, sensor number, and data type identifier, and the data body is aligned to fixed bytes for subsequent batch processing.

[0028] Based on multi-source heterogeneous monitoring data streams, a point cloud dynamic reconstruction module is invoked to reconstruct the three-dimensional point cloud morphology of the goaf area at preset time intervals. The point cloud dynamic reconstruction module employs an improved iterative nearest-point algorithm, introducing geological constraints and deformation rate penalties to control the registration error between adjacent periodic point clouds within 0.5mm, outputting a high-fidelity three-dimensional morphological model including the roof curvature gradient and the sidewall displacement vector field. The objective function of the improved iterative nearest-point algorithm contains three terms: the first is the sum of squared traditional point-to-point distances; the second is the negative logarithm of the cosine of the normal angle between corresponding points in adjacent point clouds; and the third is the square of the magnitude of the displacement vector of the corresponding point. The three weighting coefficients are dynamically adjusted according to the current average settlement rate of the goaf area, and the weighting strategy is stored in a coefficient adjustment table, which is obtained through offline training based on historical settlement data.

[0029] The algorithm execution flow is as follows: First, extract the surface point cloud collected by the laser ranging sensor from the current time slice data stream. Each point contains three-dimensional coordinates and a normal vector. Second, initialize the transformation matrix as an identity matrix, set the maximum number of iterations to 50, and the convergence threshold to 0.01 mm. Third, perform point pair matching, using a kd-tree to accelerate the nearest neighbor search. Matched point pairs must satisfy a normal angle of less than 30°. Then, calculate the objective function value and update the transformation matrix. The transformation matrix update uses the Levenberg-Marquardt optimization algorithm. Finally, determine whether convergence has occurred. If not, repeat the matching and update steps. If convergence has occurred, output the final transformation matrix and the registered point cloud.

[0030] The geological constraint term is implemented by introducing prior knowledge of the roof dip angle. A penalty term is added to the objective function to calculate the angle between the roof point normal vector and the preset dip direction, preventing distortions that do not conform to geological laws during registration. The deformation rate penalty term is implemented by limiting the displacement vector magnitude of the same point during adjacent periods. The penalty coefficient is inversely proportional to the historical maximum settlement rate, avoiding false large deformations caused by instantaneous noise. The output 3D morphological model is stored in the form of a vertex list and a patch index. Vertex attributes include coordinates, normal vector, curvature, and displacement vector. Curvature is calculated through local quadratic surface fitting, and displacement vector is obtained from the difference in coordinates between the corresponding points in the current point cloud and the point cloud of the previous period.

[0031] A high-fidelity 3D morphological model is input into the geomechanical coupling analysis module. This module embeds the rock mechanics constitutive equations and damage evolution criteria, and, combined with ore body occurrence conditions, surrounding rock classification parameters, and historical mining disturbance records, calculates the current 3D stress field distribution, plastic zone expansion boundary, and safety factor spatial cloud map of the goaf. The safety factor spatial cloud map uses any spatial point in the goaf as the calculation unit, outputting the stability margin value of that point under the current load conditions. The rock mechanics constitutive equations adopt the modified Hoek-Brown criterion, and its expression is: ; in For the maximum principal stress, For the minimum principal stress, denoted as , where m is the uniaxial compressive strength of the rock block, and m is the rock mass integrity coefficient, reflecting the influence of rock mass structural integrity on strength. For intact rock masses, the m value is close to the rock block test value (e.g., m≈25 for granite), while for rock masses with well-developed joints, the m value is significantly reduced (e.g., m≈5 for fractured rock masses). Specific values ​​are obtained by referring to tables based on on-site rock mass quality grading (e.g., RMR, Q system) or correcting through indoor rock block tests. s is the rock mass disturbance coefficient, characterizing the degree to which external factors such as mining disturbance and groundwater soaking weaken the rock mass strength. For undisturbed fresh rock masses, s=1.0; for strongly disturbed or water-saturated rock masses, s≤0.1. Values ​​need to be determined on-site in conjunction with mining methods (e.g., blasting, mechanical excavation) and hydrogeological conditions. The damage evolution criterion adopts a cumulative damage model based on strain energy density. The damage variable D is defined as the ratio of the current cumulative plastic strain energy to the strain energy corresponding to the peak rock mass strength. The calculation process uses the finite element discretization method to divide the goaf space into tetrahedral meshes. The mesh size is adaptively adjusted according to the curvature of the morphological model. The mesh is refined to 0.2m in areas with large curvature and widened to 1m in flat areas.

[0032] Boundary conditions are set based on the actual occurrence state of the mine: the roof is subjected to the self-weight stress of the overlying strata, the sidewalls are subjected to horizontal tectonic stress, and the floor is fixedly constrained. Material parameters are obtained from tables based on the surrounding rock classification results, which are comprehensively evaluated based on the uniaxial compressive strength of the rock, the degree of joint development, and the groundwater condition. The solver adopts an implicit incremental iteration method with a time step of 10 minutes. The convergence criterion for each iteration is that the force residual is less than 1% of the initial residual. The output safety factor spatial cloud map is sampled at meter-level resolution, and the safety factor of each sampling point is obtained by interpolating the calculation results of the four surrounding grid nodes. The plastic zone extension boundary is determined by tracking continuous areas where the damage variable D is greater than 0.8. The boundary point coordinates are output in polyline form for subsequent risk area delineation.

[0033] Based on the stability margin values, an automatic risk area delineation program is initiated. This program sets three levels of warning thresholds. When the stability margin value of a spatial point falls below the first threshold, it is marked as a yellow warning area; below the second threshold, it is marked as an orange warning area; and below the third threshold, it is marked as a red warning area. The three warning thresholds are calibrated using field measurement data based on ore type, burial depth, and service life. The first threshold is set to 1.5, the second threshold to 1.2, and the third threshold to 0.9. These thresholds are applicable to medium-hard rock masses, burial depths less than 500m, and service lives less than 3 years in goaf areas. For weak rock masses or deep mining conditions, the thresholds are calibrated through a combination of field fracturing tests and numerical inversion.

[0034] The steps of the delineation procedure are as follows: First, traverse all sampling points in the spatial cloud map of the safety factor and assign corresponding color labels based on the comparison between their stability margin values ​​and the three-level threshold; Second, perform spatial clustering on the points with the same color label, using a density-based clustering algorithm, setting the neighborhood radius to 1m and the minimum number of points to 10 to ensure that the delineated area has spatial continuity. Next, the geometric center, bounding box, maximum deformation rate, and average safety factor of each cluster region are calculated as region feature vectors. Finally, the feature vectors are packaged with the region boundary coordinates to generate a warning region data packet. Yellow warning regions indicate potential risks, requiring increased monitoring frequency; orange warning regions indicate escalating risks, requiring the preparation of support plans; red warning regions indicate impending instability, requiring immediate evacuation of personnel and initiation of reinforcement. The region data packet is pushed to the warning instruction generation module via a message queue and simultaneously stored in a historical database for subsequent analysis.

[0035] The spatial coordinates and deformation rate data of the yellow, orange, and red warning areas are input into the warning command generation module. The module automatically generates corresponding control command sets based on the warning level. These control command sets include sending voice alarm commands to the on-site broadcast system, pushing three-dimensional risk heatmap commands to the dispatch center, issuing local grouting reinforcement coordinate commands to the support operation unit, and sending avoidance path replanning commands to the mining equipment.

[0036] Voice alarm commands are customized according to the warning level: a yellow warning broadcasts "Local monitoring abnormalities in the goaf area, please strengthen patrols"; an orange warning broadcasts "Goaf risk increased, support teams on standby"; and a red warning broadcasts "Goaf is about to become unstable, evacuate the work face immediately." The 3D risk heat map command includes the boundary coordinates of the warning area, color coding, and deformation rate values, encapsulated in WebGL format, supporting 3D visualization on the dispatch center's large screen. The local grouting reinforcement coordinate command includes the 3D coordinates of the reinforcement point, grouting pressure, grouting volume, and material ratio. The spatial positioning accuracy requirements are a horizontal error of less than 0.3m and a vertical error of less than 0.2m. The grouting pressure setting range is 0.5MPa to 3MPa, the initial setting time of the grouting material is controlled within 30 minutes, and the final setting strength is higher than 20MPa.

[0037] The avoidance path replanning instruction includes the new path's start point, end point, intermediate avoidance point sequence, and path cost assessment value. The path generation algorithm adopts an improved A* algorithm. The cost function includes three terms: the first is the path length, the second is the integral risk value of crossing risk areas, and the third is the total change in path curvature. The weight coefficients of these three terms are manually set by the scheduler according to the current production urgency. All instructions are appended with a timestamp, instruction number, and execution priority, and are distributed to the corresponding execution terminal via the message bus.

[0038] Control commands are distributed to corresponding execution terminals via wireless communication links, enabling closed-loop response to risks in goaf areas. The wireless communication link utilizes an industrial-grade wireless mesh network with a node spacing of less than 100m, a transmission rate exceeding 1MB / s, and a packet loss rate of less than 1‰. Upon receiving a voice alarm command, the on-site broadcasting system immediately plays pre-recorded audio through speakers in designated areas, automatically adjusting the volume according to ambient noise to ensure clear audibility within a 100m range.

[0039] Upon receiving the 3D risk heat map command, the dispatch center highlights the warning area on the 3D visualization platform, overlaying deformation rate arrows and safety factor contour lines, and allows mouse hovering to view detailed data. Upon receiving the local grouting reinforcement coordinate command, the support operation unit automatically drives the grouting robot to the designated coordinates, adjusts the drilling angle and depth, and injects quick-setting grout at the set pressure and flow rate. The pressure and flow rate curves are monitored in real time during the grouting process, and automatic shutdown and alarm are triggered in case of abnormalities. Upon receiving the avoidance path replanning command, the mining equipment's onboard controller immediately interrupts the current path tracking, loads the new path data, and replans the steering and speed curves to ensure a smooth transition to a safe area.

[0040] After completing the instruction, all execution terminals send an execution confirmation message back to the central processing unit, including the execution time, execution result, and exception code. The central processing unit aggregates all confirmation messages, updates the status of the risk area, and if the safety factor of a red alert area rises above 1.5 after hardening, it is downgraded to a yellow alert; if an orange alert area continues to deteriorate, it is upgraded to a red alert and a second hardening instruction is triggered. The entire closed-loop response process is recorded, forming an intervention log for post-event traceability and optimization.

[0041] The historical data backtracking analysis unit stores 3D morphological models, stress field distribution maps, early warning records, and intervention implementation logs for all periods. It supports time-axis sliding comparison, deformation trend extrapolation prediction, and intervention effectiveness evaluation. Data storage spans more than 10 years, with a data compression ratio higher than 5:1 and a retrieval response time of less than 3 seconds. The storage architecture employs a distributed time-series database, with data partitioned by time, each partition containing data for one calendar month.

[0042] The compression algorithm employs a combination of differential and dictionary encoding. Coordinate data is stored using the difference between adjacent points, while safety factor data is stored using hierarchical quantization. The retrieval interface supports multiple conditional queries based on time range, spatial range, and warning level. Query results are presented in the form of 3D animations or statistical charts. Deformation trend extrapolation prediction uses a long short-term memory neural network model. Inputting the deformation rate sequence of the past 30 days, it outputs predicted values ​​for the next seven days. Prediction error is evaluated using root mean square error, and the model weights are periodically updated with new data.

[0043] The effectiveness of intervention measures is evaluated by comparing the rate of change in the safety factor before and after the intervention. An effective intervention is defined as a safety factor improvement rate greater than 10%, while an ineffective intervention is defined as an improvement rate less than 5%. The evaluation results are used to optimize grouting parameters and path planning strategies. The backtracking analysis unit also provides data export functionality, supporting export to common 3D model formats or tabular formats for further analysis by third-party software.

[0044] The distributed multimodal sensor array deployment unit ensures long-term stable operation of the sensors in harsh mining environments. The sensor housing is made of stainless steel with an anti-corrosion coating and filled with flame-retardant silicone, achieving a shock resistance rating of IK08. The power supply is intrinsically safe, operating at 24V DC with a power consumption of less than 5W, and supports backup battery power, allowing for continuous operation for 4 hours after a power outage. The communication interface employs a dual-redundancy design of industrial Ethernet and RS485, automatically switching to the backup link in case of primary link failure. The mounting bracket features an adjustable angle design, allowing for on-site fine-tuning of the sensor's orientation to ensure the measurement beam is perpendicular to the measured surface. Regular maintenance includes monthly zero-point calibration, quarterly protection level testing, and annual comprehensive performance calibration. The calibration process uses standard gauge blocks and a vibration table to ensure measurement accuracy remains stable.

[0045] Multi-source data synchronous transmission and preprocessing units ensure the integrity and timeliness of the data stream. The industrial ring network adopts a ring topology, supporting bidirectional data transmission, and single-point failures do not affect overall communication. Data packets are protected by both checksums and sequence numbers; retransmission is requested when the receiver fails to verify the data, and a data retransmission mechanism is triggered when the sequence number is discontinuous. The preprocessing module runs on a multi-core server, employing a pipelined parallel architecture. Data reception, time alignment, coordinate transformation, filtering, and anomaly removal are all performed by independent threads, exchanging data through shared memory, with a processing latency of less than 50ms.

[0046] The outlier removal module incorporates a self-learning mechanism, dynamically adjusting the three-sigma threshold and isolated forest parameters based on historical data to adapt to varying data distribution under different operating conditions. The preprocessed data stream is published to the message middleware by topic, allowing downstream modules such as point cloud reconstruction, mechanical analysis, and early warning generation to subscribe and consume it, thus achieving decoupling and elastic scaling between modules.

[0047] The point cloud dynamic reconstruction unit achieves high-precision 3D morphological evolution tracking. The algorithm runs in a GPU-accelerated environment, utilizing the CUDA parallel computing framework to parallelize time-consuming operations such as point cloud matching, objective function calculation, and matrix updates, achieving a single-cycle reconstruction time of less than 30 seconds. Prior knowledge of the roof dip angle in the geological constraints is obtained from the mine geological exploration report and input in the form of a digital elevation model. The dip angle calculation uses local plane fitting with a fitting window of 5m × 5m. The weight coefficient adjustment table for the deformation rate penalty term contains 10 sets of preset parameters, corresponding to 10 intervals of average settlement rate from 0.1mm / day to 10mm / day. In practical applications, the corresponding weight is obtained by looking up the table based on the average settlement rate calculated in the current cycle. In addition to geometric information, the output 3D morphological model also includes a timestamp, reconstruction error, and data source identifier for easy subsequent analysis and tracing. The model is updated every 10 minutes to ensure the capture of rapid deformation processes.

[0048] The geomechanical coupled analysis unit provides a physical mechanism-driven risk assessment. Finite element mesh generation employs a leading-edge algorithm, adaptively refining the mesh based on the surface curvature and stress gradient of the morphological model to ensure computational accuracy in critical areas. The material parameter library contains 20 common rock mass types, each corresponding to a set of parameters including m, s, elastic modulus, and Poisson's ratio. These parameter values ​​are determined through a combination of laboratory experiments and in-situ inversion.

[0049] The solver supports parallel computing, dividing the computational domain into multiple subdomains, each assigned to an independent computational node. Nodes exchange boundary data via a message passing interface, accelerating the solution of large-scale problems. The safety factor is calculated using a strength reduction method, gradually decreasing the rock mass strength parameters until the model becomes unstable. The instability criterion is a sudden increase in displacement or the penetration of the plastic zone; the reduction factor is the safety factor. The calculation results are post-processed to generate spatial contour maps with resolution consistent with the morphological model, supporting arbitrary cross-sectional cutting and isosurface extraction.

[0050] The automatic risk area delineation unit enables intelligent risk identification. The neighborhood radius and minimum number of points in the clustering algorithm are dynamically adjusted according to the size of the goaf: a neighborhood radius of 2m and a minimum number of points of 20 are set for large goafs; a neighborhood radius of 0.5m and a minimum number of points of 5 are set for small goafs. In addition to geometric and mechanical attributes, the regional feature vector also includes historical evolution trends. The delineation results are stored using spatial indexing optimization, supporting fast spatial queries. The early warning threshold calibration process employs a Bayesian optimization algorithm, using historical accident data as the training set to maximize early warning accuracy and minimize false alarm rate. The optimization results are stored as a threshold configuration file, supporting on-demand loading for different mines.

[0051] The early warning instruction generation and distribution unit ensures the timeliness and accuracy of risk response. Instruction generation utilizes a template engine, filling preset templates based on different early warning levels and regional characteristics to generate structured instruction text. Voice alarm instructions support multilingual switching, automatically selecting the playback language based on the operator's nationality. The 3D risk heatmap employs progressive loading technology, first transmitting a low-resolution overview map and then loading high-resolution details as needed, reducing network bandwidth consumption. Local grouting reinforcement coordinate instructions include redundant checksums, automatically requesting retransmission if the executing terminal fails to decode. The cost function weight coefficients in avoidance path replanning instructions are set by the dispatcher via a slider in a graphical interface, which displays real-time path previews under different weights to aid decision-making. Instruction distribution uses a publish-subscribe model, with each executing terminal subscribing to the instruction types they are interested in, ensuring accurate instruction delivery.

[0052] The closed-loop response execution unit completes the physical world intervention. The grouting robot is equipped with a six-degree-of-freedom robotic arm and high-precision displacement sensors, with a positioning error of less than 0.1m. Grouting pressure is controlled in a closed loop, with pump speed adjusted in real time to maintain the set pressure. The mining equipment avoidance path tracking uses a model predictive control algorithm to predict the vehicle's state five seconds in advance, optimizing the steering angle and throttle opening to ensure a path tracking error of less than 0.5m. Execution confirmation messages include equipment status snapshots, such as grouting pressure curves and vehicle position trajectories, for the central processing unit to verify the command execution effect. Anomaly handling mechanisms include command timeout retransmission, equipment failure switching to backup equipment, and manual takeover forced mode, ensuring that basic safety functions are maintained even in extreme situations.

[0053] Through the collaborative work of the aforementioned units, the entire system constructs a three-dimensional digital twin of the entire life cycle deformation evolution of the goaf, enabling millimeter-level spatial positioning and sub-hour-level time early warning of roof settlement rate, surrounding rock stress redistribution path, and local collapse precursor characteristics. It establishes the ability to proactively perceive and precisely intervene in the stability risks of the goaf within the entire three-dimensional spatial domain.

[0054] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0055] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A precise positioning method based on three-dimensional geological and mineral resource modeling, characterized in that, include: Laser ranging sensor arrays, micro-vibration sensor arrays, and fiber optic strain sensor arrays are deployed at a preset spatial density on the surface of the roof, sidewalls, and key pillars in the goaf to collect multi-source heterogeneous monitoring data streams; the laser ranging sensor array is used to collect normal displacement data of points on the roof and sidewall surfaces. Based on the multi-source heterogeneous monitoring data stream, the point cloud dynamic reconstruction module is invoked to reconstruct the three-dimensional point cloud of the internal spatial morphology of the goaf at a preset time interval. An improved iterative nearest point algorithm is used to output a high-fidelity three-dimensional morphological model that includes the curvature change gradient of the roof and the displacement vector field of the sidewalls. The high-fidelity three-dimensional morphological model is input into the geomechanical coupling analysis module, which outputs the stability margin value of any spatial point in the goaf under the current load conditions. Based on the stability margin value, an automatic risk area delineation program is initiated. The automatic risk area delineation program sets three-level early warning thresholds and divides the early warning area based on the three-level early warning thresholds. The three-level early warning thresholds are obtained by calibration based on field measurement data according to the type of ore and rock, burial depth, and service life. The spatial coordinate range and deformation rate data of the warning area are input into the warning instruction generation module, which automatically generates a corresponding set of control instructions according to the warning level. The control command set is distributed to the corresponding execution terminal through a wireless communication link to achieve a closed-loop response to the risk of goaf.

2. The precise positioning method based on three-dimensional geological and mineral resource modeling according to claim 1, characterized in that, Also includes: The normal displacement data, spatial coordinate and energy level data, and axial and circumferential strain distribution data are synchronously transmitted to the central processing unit through an industrial ring network. The central processing unit performs timestamp alignment, unified spatial coordinate transformation, noise filtering and outlier removal operations on the raw sensor data to form a multi-source heterogeneous monitoring data stream with a unified spatiotemporal reference. The operations performed on the original sensor data, including timestamp alignment, spatial coordinate transformation, noise filtering, and outlier removal, include: The network time protocol synchronization mechanism is used to synchronize the time of all sensors with the built-in high-stability crystal oscillator, with a timing error of less than 1ms. By using a pre-stored sensor installation coordinate table and a transformation matrix between the mine geodetic coordinate system, a seven-parameter Bursa model is introduced to perform a unified spatial coordinate transformation. Noise filtering is performed in two stages. The first stage uses a combination of moving average and median filtering at the sensor end with a window length of 10 sampling points. The second stage uses wavelet thresholding in the central processing unit, selecting the db4 wavelet basis, with a decomposition level of five and a soft thresholding function. Outlier removal employs a dual verification method: the three sigma criterion removes outliers that deviate from the mean by three times the standard deviation, while the isolated forest algorithm identifies sparse outlier clusters in high-dimensional space by constructing a random tree forest.

3. The precise positioning method based on three-dimensional geological and mineral resource modeling according to claim 2, characterized in that, The point cloud dynamic reconstruction module is invoked to reconstruct the three-dimensional point cloud morphology of the internal spatial morphology of the goaf, including: Extract the surface point cloud acquired by the laser rangefinder from the current time slice data stream. Each point contains three-dimensional coordinates and a normal vector. The transformation matrix is ​​initialized as an identity matrix, the maximum number of iterations is set to fifty, and the convergence threshold is set to 0.01 mm. Perform point-pair matching, using a kd-tree to accelerate nearest neighbor search. Matched point pairs must satisfy the normal angle being less than 30°. Calculate the objective function value and update the transformation matrix. The transformation matrix update uses the Levenberg-Marquardt optimization algorithm. Determine if convergence has occurred. If not, repeat the matching and update steps. If convergence has occurred, output the final transformation matrix and the registered point cloud. The objective function consists of three terms: the first term is the sum of squares of point-to-point distances, the second term is the negative logarithm of the cosine of the normal angle between corresponding points in adjacent point clouds, and the third term is the square of the displacement vector magnitude of the corresponding point. The weighting coefficients of the three terms are dynamically adjusted according to the current average settlement rate of the goaf.

4. The precise positioning method based on three-dimensional geological and mineral resource modeling according to claim 3, characterized in that, The high-fidelity three-dimensional morphological model is input into the geomechanical coupling analysis module, including: The space of the goaf is divided into tetrahedral meshes using the finite element discretization method, and the mesh size is adaptively adjusted according to the curvature of the morphological model. Boundary conditions are set according to the actual occurrence state of the mine: the overlying strata self-weight stress is applied to the roof, the horizontal tectonic stress is applied to the sidewalls, and the bottom plate is fixed and constrained. Material parameters are obtained by referring to tables based on the results of surrounding rock classification. The classification is based on a comprehensive evaluation of the rock's uniaxial compressive strength, joint development degree, and groundwater conditions. The solver employs an implicit incremental iterative method with a time step of 10 minutes. The convergence criterion for each iteration is that the force residual is less than 1% of the initial residual. The output safety factor spatial cloud map is sampled at meter-level resolution, and the safety factor of each sample point is obtained by interpolating the calculation results of its four surrounding grid nodes.

5. The precise positioning method based on three-dimensional geological and mineral resource modeling according to claim 4, characterized in that, Activate the automatic risk area delineation process, including: Traverse all sampling points in the safety factor spatial cloud map, and assign corresponding color labels based on the comparison between their stability margin values ​​and the third-level threshold. Spatial clustering of the same-color labels is performed using a density-based clustering algorithm, with a neighborhood radius of 1m and a minimum number of points of 10. Calculate the geometric center, bounding box, maximum deformation rate, and average safety factor of each cluster region as the region feature vector; The feature vectors and region boundary coordinates are packaged together to generate a warning region data packet.

6. The precise positioning method based on three-dimensional geological and mineral resource modeling according to claim 5, characterized in that, The warning instruction generation module automatically generates the corresponding control instruction set based on the warning level, including: The division of warning areas based on the three-level warning thresholds includes: marking a spatial point as a yellow warning area when its stability margin value is lower than the first threshold, as an orange warning area when it is lower than the second threshold, and as a red warning area when it is lower than the third threshold. The voice alarm commands are customized according to the warning level. The yellow warning will play "Local monitoring of the goaf is abnormal, please strengthen the patrol", the orange warning will play "The risk of the goaf has increased, the support team should stand by", and the red warning will play "The goaf is about to become unstable, evacuate the work face immediately". The 3D risk heat map command includes the boundary coordinates of the warning area, color coding, and deformation rate value, and is encapsulated in WebGL format. The coordinate command for local grouting reinforcement includes the three-dimensional coordinates of the reinforcement point, grouting pressure, grouting volume, and material ratio. The spatial positioning accuracy requirement is a horizontal error of less than 0.3m and a vertical error of less than 0.2m. The obstacle avoidance path replanning instruction includes the new path start point, end point, intermediate obstacle avoidance point sequence, and path cost evaluation value. The path generation algorithm adopts the improved A* algorithm.

7. The precise positioning method based on three-dimensional geological and mineral resource modeling according to claim 6, characterized in that, Distributing the control command set to the corresponding execution terminal via a wireless communication link includes: After receiving a voice alarm command, the on-site broadcasting system plays a pre-recorded voice message through the speakers in the designated area, and the playback volume is automatically adjusted according to the ambient noise. After receiving the three-dimensional risk heat map instruction, the dispatch center highlights the warning area on the three-dimensional visualization platform and overlays deformation rate arrows and safety factor contour lines. After receiving the local grouting reinforcement coordinate command, the support operation unit automatically drives the grouting robot to move to the designated coordinate, adjusts the drilling angle and depth, and injects quick-setting grout according to the set pressure and flow rate; After receiving the obstacle avoidance path replanning instruction, the on-board controller of the mining equipment immediately interrupts the current path tracking, loads the new path data, and replans the steering and speed curves.

8. The precise positioning method based on three-dimensional geological and mineral resource modeling according to claim 7, characterized in that, The improved iterative nearest point algorithm uses a dynamic adjustment strategy for the three weighting coefficients, which is obtained through offline training based on historical settlement data and stored in a coefficient adjustment table. The corresponding weights are obtained by looking up the table based on the average settlement rate calculated in the current period.

9. A precise positioning system based on three-dimensional geological and mineral modeling, characterized in that, include: A distributed multimodal sensor array deployment unit is used to deploy laser ranging sensor arrays, microseismic sensor arrays, and fiber optic strain sensor arrays on the surface of the roof, sidewalls, and key pillars in the goaf at a preset spatial density. The laser ranging sensor array is used to collect normal displacement data of points on the roof and sidewall surfaces, the microseismic sensor array is used to collect spatial coordinates and energy level data of rock mass fracture events, and the fiber optic strain sensor array is used to collect axial and circumferential strain distribution data inside the pillars. The multi-source data synchronous transmission and preprocessing unit is used to synchronously transmit the normal displacement data, spatial coordinate and energy level data, and axial and circumferential strain distribution data to the central processing unit through the industrial ring network, and perform timestamp alignment, unified spatial coordinate transformation, noise filtering and outlier removal operations on the raw sensor data to form a multi-source heterogeneous monitoring data stream with a unified spatiotemporal reference. The point cloud dynamic reconstruction unit is used to reconstruct the three-dimensional point cloud of the internal spatial morphology of the goaf based on the multi-source heterogeneous monitoring data stream at a preset time interval. The point cloud dynamic reconstruction unit adopts an improved iterative nearest point algorithm, introduces geological constraint terms and deformation rate penalty terms, so that the registration error of adjacent period point clouds is controlled within 0.5mm, and outputs a high-fidelity three-dimensional morphological model including the roof curvature change gradient and the sidewall displacement vector field. The geomechanical coupling analysis unit is used to input the high-fidelity three-dimensional morphological model into the calculation module embedded with the rock mass mechanical constitutive equation and damage evolution criterion. Combined with the ore body occurrence conditions, surrounding rock classification parameters, and historical mining disturbance records, it calculates the current three-dimensional stress field distribution, plastic zone expansion boundary, and safety factor spatial cloud map of the goaf. The safety factor spatial cloud map uses any spatial point in the goaf as the calculation unit and outputs the stability margin value of that point under the current load conditions. The automatic risk area delineation unit is used to set three-level early warning thresholds based on the stability margin value. When the stability margin value of a spatial point is lower than the first threshold, it is marked as a yellow early warning area; when it is lower than the second threshold, it is marked as an orange early warning area; and when it is lower than the third threshold, it is marked as a red early warning area. The three-level early warning thresholds are obtained by calibrating the data from field measurements based on the type of ore, burial depth, and service life. The early warning instruction generation and distribution unit is used to input the spatial coordinate range and deformation rate data of the yellow, orange, and red early warning areas into the instruction generation module, and automatically generate corresponding control instruction sets according to the early warning level. The control instruction sets include sending voice alarm instructions to the on-site broadcasting system, pushing three-dimensional risk heat map instructions to the dispatch center, issuing local grouting reinforcement coordinate instructions to the support operation unit, and sending avoidance path replanning instructions to the mining equipment. The control instruction sets are then distributed to the corresponding execution terminals through a wireless communication link. The closed-loop response execution unit is used to receive and execute the control command set to realize physical intervention and operation process adjustment for goaf risks.

10. The precise positioning system based on three-dimensional geological and mineral modeling according to claim 9, characterized in that, The point cloud dynamic reconstruction unit is used for: Extract the surface point cloud acquired by the laser rangefinder from the current time slice data stream. Each point contains three-dimensional coordinates and a normal vector. The transformation matrix is ​​initialized as an identity matrix, the maximum number of iterations is set to 50, and the convergence threshold is set to 0.01 mm. Perform point-pair matching, using a kd-tree to accelerate nearest neighbor search. Matched point pairs must satisfy the normal angle being less than 30°. Calculate the objective function value and update the transformation matrix. The transformation matrix update uses the Levenberg-Marquardt optimization algorithm. Determine if convergence has occurred. If not, repeat the matching and update steps. If convergence has occurred, output the final transformation matrix and the registered point cloud. The objective function consists of three terms: the first term is the sum of squares of point-to-point distances, the second term is the negative logarithm of the cosine of the normal angle between corresponding points in adjacent point clouds, and the third term is the square of the displacement vector magnitude of the corresponding point. The weighting coefficients of the three terms are dynamically adjusted according to the current average settlement rate of the goaf.

Images