Three-dimensional laser scanning precision detection system and method for goaf

Abstract

The application discloses a goaf three-dimensional laser scanning accurate detection system and method, and particularly relates to the field of goaf three-dimensional detection and treatment data, and comprises the following steps: through an intelligent sensing system, station scanning, unmanned aerial vehicle (UAV) airborne scanning and equal-interval cross-section scanning are performed on a dry area of a goaf, three-dimensional digital sonar depth measurement is performed on a water-filled area of the goaf, and a goaf boundary patch set is generated by an edge computing node according to scanning poses, echo coordinates, source marks and verification states; the application converts multi-source scanning data into the goaf boundary patch set with the source marks and the verification states, generates a sensitive boundary table based on the influence of unverified boundary patches on filling volume, blasting boundary and stable boundary, and then uses a particle swarm optimization algorithm to drive supplementary measurement and iterative calculation, so that the problem that in the prior art, the unverified boundary is difficult to determine whether it influences treatment parameters, and a detection model is difficult to directly support filling volume, blasting parameters and stability calculation is solved.

Description

Technical Field

[0001] This invention relates to the field of three-dimensional detection and treatment data technology for goaf areas, and more specifically, to a three-dimensional laser scanning precision detection system and method for goaf areas. Background Technology

[0002] In the design of three-dimensional detection and treatment of goaf, existing processing methods mostly rely on station-type three-dimensional laser scanning, UAV-borne scanning, equal-step scanning or three-dimensional digital sonar to acquire point cloud, underwater contour and spatial cross-section data of goaf. Then, point cloud stitching, three-dimensional modeling, volume calculation and stability analysis are completed through edge computing nodes or field workstations for filling treatment, blasting treatment and subsequent re-measurement. In the treatment of goaf areas in coal mines, metal mines, and underground engineering projects, the goaf areas often contain residual pillars, collapsed accumulations, water-filled boundaries, narrow-mouthed hidden cavities, and local undulations in the roof. Although intelligent sensing systems can collect millimeter-level spatial data, they are affected by the access range, operation time, communication bandwidth, and safety control, making it difficult to repeatedly and intensively scan all suspected boundaries on site. The resulting problem is that unconfirmed boundaries in the 3D model can often only be retained as scanning blind spots or objects for manual verification. It is impossible to determine on site whether they will change the filling volume, blasting hole location, charge zoning, or stability calculation results. This means that although the model can show the goaf morphology, the treatment design still requires manual delineation of the filling space, reduction of collapsed accumulations, verification of residual pillar avoidance range, and additional retesting. Even if the particle swarm algorithm participates in the site or parameter optimization, it lacks supplementary scanning criteria for changes in treatment parameters. Therefore, the technical problem to be solved by this application is: how to identify sensitive boundaries affecting governance parameters based on edge computing and intelligent sensing systems during the three-dimensional laser scanning of goaf areas, and use particle swarm optimization algorithm to drive directional supplementary scanning or sonar supplementary measurement so that the detection model can directly support the calculation of filling volume, blasting parameters and stability. Summary of the Invention

[0003] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a three-dimensional laser scanning precision detection system and method for goaf areas. This system converts multi-source scanning data into a goaf area boundary patch set with source markers and verification status. It also generates a sensitive boundary table based on the influence of unverified boundary patches on filling volume, blasting boundary, and stable boundary. Finally, it uses a particle swarm optimization algorithm to drive supplementary measurement and iterative recalculation to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for precise detection of goaf areas using three-dimensional laser scanning, comprising: S1. The intelligent sensing system performs station scanning, UAV airborne scanning and equal-step cross-section scanning on the dry exposed area of ​​the goaf, and performs three-dimensional digital sonar depth sounding on the water-filled area of ​​the goaf. The edge computing node generates the goaf boundary patch set according to the scanning pose, echo coordinates, source marker and verification status. S2. Substitute the cavity boundary, collapse boundary, residual column boundary, underwater boundary, and roof boundary of the goaf boundary into the treatment parameter calculation rules, deduct the collapse, residual column, and underwater occupancy from the cavity enclosure volume to obtain the filling volume, obtain the blasting boundary according to the adjacency relationship between the residual column boundary and the goaf boundary, obtain the stable boundary according to the span connection relationship of the roof boundary, and generate the parameter base table. S3. For uncertified boundary patches in the goaf boundary patch concentration, the sensitive recalculation algorithm of treatment parameters is used to delete and recalculate each patch. The filling volume, blasting boundary and stability boundary after deletion of each uncertified boundary patch are compared with the parameter base table item by item, and a sensitive boundary table is generated based on the uncertified boundary patches whose comparison results have changed. S4. Construct supplementary test particles using the sensitive boundary patches in the sensitive boundary table. Use the particle swarm greedy supplementary test algorithm to calculate the number of sensitive boundary patches and the parameter stabilization number corresponding to each supplementary test particle. Select supplementary test particles in descending order of parameter stabilization number and descending order of sensitive boundary patch number to generate a supplementary test task table.

[0005] In a preferred embodiment, it further includes: S5. Collect supplementary measurement data according to the supplementary measurement task table, and write the supplementary measurement data back to the goaf boundary piece set. Use a Newton-type iterative recalculation algorithm to incrementally recalculate the filling volume, blasting boundary and stable boundary until the filling volume value, blasting boundary piece number set and stable boundary piece number set in the recalculation results of two adjacent rounds are the same, and generate the goaf governance decision model.

[0006] In a preferred embodiment, S1 includes: S1-1. In the dry exposed area of ​​the goaf, the laser echo coordinates formed by the station scanning are configured with station markers, the laser echo coordinates formed by the UAV airborne scanning are configured with airborne markers, and the laser echo coordinates formed by the equal step cross section scanning are configured with cross section markers. The station scanning pose, the UAV airborne scanning pose, and the equal step cross section scanning pose are used as the scanning poses of the corresponding laser echo coordinates. S1-2. For water-filled areas in goaf areas, the sounding distance formed by three-dimensional digital sonar sounding is converted into sonar echo coordinates along the sounding direction, sonar echo coordinates are configured with sonar markers, and the three-dimensional digital sonar sounding pose is used as the scanning pose of the sonar echo coordinates. S1-3. The laser echo coordinates and sonar echo coordinates are converted to the same coordinate system according to their respective scanning poses through the edge computing nodes. Source markers are generated according to the station markers, airborne markers, cross-section markers and sonar markers. Verification status is generated according to the number of types of source markers in the same cross-section. The goaf boundary patch set is generated with the echo coordinates, source markers and verification status in the same cross-section.

[0007] In a preferred embodiment, S2 includes: S2-1. Using the goaf boundary piece set as input, the source classification algorithm is adopted to define the boundary pieces with station markers, airborne markers and cross-section markers that enclose the inner surface of the cavity as cavity boundary pieces, the boundary pieces located inside the cavity boundary pieces that enclose the outer surface of the accumulation as collapse boundary pieces, the boundary pieces located inside the cavity boundary pieces that penetrate the upper and lower boundaries of the cavity as residual column boundary pieces, the boundary pieces with sonar markers that are located in the water-filled area as underwater boundary pieces, and the boundary pieces located above the cavity boundary pieces that connect the two side walls as roof boundary pieces, thus generating a classification boundary piece set; S2-2. Input the cavity boundary pieces in the fractal boundary piece set into the common edge enclosure algorithm, and splice them according to the common edge line to generate the cavity. Input the cavitation boundary piece, residual column boundary piece and underwater boundary piece into the occupancy deduction algorithm, and enclose them to generate cavitation occupancy, residual column occupancy and underwater occupancy respectively. Subtract the volume of cavitation occupancy, residual column occupancy and underwater occupancy from the volume of the cavity to generate the filling volume. S2-3. Input the set of boundary segments into the common edge avoidance algorithm, take the cavity boundary segments that do not share the edge with the residual column boundary segments and splice them to generate the blasting boundary, and input the roof boundary segments into the span concatenation algorithm. Generate the roof span according to the coordinate difference between the two ends and then concatenate them according to the boundary segment number to generate the stable boundary. Then, establish the binding relationship between the filling volume, blasting boundary and stable boundary according to the goaf number, boundary segment number and calculation time to generate the parameter base table.

[0008] In a preferred embodiment, S3 includes: S3-1. Using the set of boundary patches of the goaf as input, the single-source patch retrieval algorithm is used to read the number of source marker types in the verification status. Boundary patches with one source marker type are defined as uncertified boundary patches. The coordinates of the two endpoints of the uncertified boundary patches are sorted and concatenated to generate a common edge line number. Boundary patch numbers with the same common edge line number are used to generate a set of adjacent boundary patch numbers. An uncertified patch column is generated using the boundary patch number, source marker, verification status, common edge line number, and adjacent boundary patch number set. S3-2. Input the uncertified boundary piece list into the counterfactual deletion algorithm, remove the uncertified boundary pieces one by one according to the boundary piece number, and recalculate the filling volume, blasting boundary and stable boundary according to the calculation rules of the parameter base table. The volume difference is generated by subtracting the filling volume of the base table from the recalculated filling volume. The blasting difference set is generated by XORing the recalculated blasting boundary piece number set with the base table blasting boundary piece number set. The stable difference set is generated by XORing the recalculated stable boundary piece number set with the base table stable boundary piece number set. A single piece difference table is generated.

[0009] In a preferred embodiment, S3 further includes: S3-3. Based on the uncertified sheet list, the adjacent sheet cluster algorithm is executed. Uncertified boundary sheets with intersections in the adjacent boundary sheet number sets are merged into uncertified sheet clusters. After the uncertified sheet clusters are removed as a whole, the filling volume, blasting boundary and stability boundary are recalculated. The volume coupling difference is generated by subtracting the sum of volume differences within the cluster from the volume difference of the sheet cluster. The blasting difference set within the cluster is removed from the blasting difference set of the sheet cluster and the union is used to generate the blasting coupling difference set. The stability difference set within the cluster is removed from the stability difference set of the sheet cluster and the union is used to generate the stability coupling difference set. The sheet cluster difference table is generated. S3-4. Input the single-piece difference table and the piece-cluster difference table into the three-parameter cross-determination algorithm. For the same boundary piece number, read the volume difference, blasting difference set, stability difference set, volume coupling difference, blasting coupling difference set, and stability coupling difference set. If any item satisfies that the volume difference is not zero, the blasting difference set is not empty, the stability difference set is not empty, the volume coupling difference is not zero, the blasting coupling difference set is not empty, or the stability coupling difference set is not empty, then generate a sensitive candidate piece list for the boundary piece number; otherwise, generate a non-sensitive piece list for the boundary piece number.

[0010] In a preferred embodiment, S3 further includes: S3-5. Input the sensitive candidate piece list into the back-insertion closure algorithm. Insert the unverified boundary pieces in the sensitive candidate piece list back into the removed goaf boundary piece set according to the boundary piece number, and recalculate the filling volume, blasting boundary, and stable boundary. If the volume difference is zero, the blasting difference set is empty, the stable difference set is empty, the volume coupling difference is zero, the blasting coupling difference set is empty, and the stable coupling difference set is empty after back-insertion, then generate a sensitive boundary table with the boundary piece number, source marker, verification status, volume difference, blasting difference set, stable difference set, volume coupling difference, blasting coupling difference set, and stable coupling difference set. Otherwise, merge the boundary piece number into the non-sensitive piece list.

[0011] In a preferred embodiment, S4 includes: S4-1. Using the sensitive boundary table as input, read the boundary patch number, source marker, volume difference, blasting difference set, stability difference set, volume coupling difference, blasting coupling difference set, and stability coupling difference set. Take the markers that have not entered the source marker among the station marker, airborne marker, cross-section marker, and sonar marker as supplementary measurement markers, and generate station supplementary scanning particles, airborne supplementary scanning particles, cross-section supplementary scanning particles, and sonar supplementary measurement particles according to the supplementary measurement markers. S4-2. Input the station-based supplementary scanning particles, airborne supplementary scanning particles, cross-sectional supplementary scanning particles, and sonar supplementary measurement particles into the particle swarm greedy supplementary measurement algorithm. After merging the supplementary measurement markers into the source markers of the same boundary piece number, recalculate and verify the state. Recalculate the volume difference, blasting difference set, stable difference set, volume coupling difference, blasting coupling difference set, and stable coupling difference set using the counterfactual piece deletion algorithm and the adjacent piece cluster algorithm. Generate the parameter stabilization number based on the number of valid items when the volume difference is zeroed, the blasting difference set is cleared, the stable difference set is cleared, the volume coupling difference is zeroed, the blasting coupling difference set is cleared, and the stable coupling difference set is cleared. S4-3. Delete the supplementary measurement particles with a parameter stabilization number of zero, delete the supplementary measurement particles with a parameter stabilization number less than the other supplementary measurement particle under the same boundary patch number, and generate a supplementary measurement task table according to the rules of decreasing parameter stabilization number, ascending boundary patch number, and supplementary measurement mark arranged according to the station mark, airborne mark, cross section mark and sonar mark.

[0012] In a preferred embodiment, S5 includes: S5-1. Using the supplementary survey task table as input, read the boundary patch number and supplementary survey mark. When the supplementary survey mark is a station mark, perform station scanning; when the supplementary survey mark is an airborne mark, perform UAV airborne scanning; when the supplementary survey mark is a cross-section mark, perform equal-step cross-section scanning; when the supplementary survey mark is a sonar mark, perform three-dimensional digital sonar depth sounding. Generate supplementary survey data by collecting the echo coordinates, scanning pose, and supplementary survey mark according to the boundary patch number. S5-2. Convert the echo coordinates in the supplementary measurement data to the coordinates of the goaf boundary patch set according to the scanning pose. Merge the converted echo coordinates into the boundary patch with the same boundary patch number. After merging the supplementary measurement mark into the source mark, recalculate and verify the status according to the number of source mark types, and generate an updated boundary patch set. S5-3. Input the updated boundary piece set into the Newton-type iterative recalculation algorithm. Generate the current round's filling volume, blasting boundary, and stable boundary according to the calculation rules of the parameter base table. Subtract the previous round's filling volume from the current round's filling volume to generate the volume residual. Generate the blasting residual set by XORing the current round's blasting boundary piece number set with the previous round's blasting boundary piece number set. Generate the stable residual set by XORing the current round's stable boundary piece number set with the previous round's stable boundary piece number set. Use the boundary piece numbers that generate the volume residual, blasting residual set, and stable residual set as the recalculation objects for the next round, until the volume residual is zero, the blasting residual set is empty, and the stable residual set is empty, thus generating the goaf governance decision model.

[0013] In a preferred embodiment, the three-dimensional laser scanning precision detection system for goaf areas includes: The boundary acquisition module performs station scanning, UAV-borne scanning, and equal-step cross-section scanning on the dry exposed area of ​​the goaf through an intelligent sensing system, performs three-dimensional digital sonar depth sounding on the water-filled area of ​​the goaf, and generates a goaf boundary patch set by the edge computing node according to the scanning pose, echo coordinates, source markers, and verification status. The parameter generation module is used to substitute the cavity boundary plates, collapse boundary plates, residual column boundary plates, underwater boundary plates, and roof boundary plates of the goaf boundary plate set into the treatment parameter calculation rules, subtract the collapse, residual column, and underwater occupancy from the cavity enclosure volume to obtain the filling volume, obtain the blasting boundary according to the adjacency relationship between the residual column boundary plate and the goaf boundary plate, obtain the stable boundary according to the span connection relationship of the roof boundary plate, and generate the parameter base table. The sensitive recalculation module uses the governance parameter sensitive recalculation algorithm to delete and recalculate the uncertified boundary patches in the goaf boundary patch set. It compares the filling volume, blasting boundary and stability boundary of each uncertified boundary patch after deletion with the parameter base table item by item, and generates a sensitive boundary table based on the uncertified boundary patches whose comparison results have changed. The greedy supplementary testing module constructs supplementary testing particles using sensitive boundary patches from the sensitive boundary table. It then uses a particle swarm greedy supplementary testing algorithm to calculate the number of sensitive boundary patches and parameter stabilization number corresponding to each supplementary testing particle. Finally, it selects supplementary testing particles in descending order of parameter stabilization number and descending order of sensitive boundary patch number to generate a supplementary testing task table. The iterative decision-making module collects supplementary measurement data according to the supplementary measurement task table and writes the supplementary measurement data back to the goaf boundary piece set. It uses a Newton-type iterative recalculation algorithm to incrementally recalculate the filling volume, blasting boundary, and stability boundary until the filling volume value, blasting boundary piece number set, and stability boundary piece number set are the same in the recalculation results of two adjacent rounds, thus generating a goaf governance decision model.

[0014] The technical effects and advantages of this invention are as follows: 1. By using station scanning, UAV airborne scanning, equal-step cross-section scanning and three-dimensional digital sonar depth measurement, a set of goaf boundary patches with source markers and verification status is generated, so that dry exposed area, underwater area and cross-section boundary are connected in the same model, which relatively improves the expression of the real morphology and spatial relationship of goaf. 2. By calculating the boundary values ​​of cavity, collapse, residual column, underwater, and top plate, the filling volume is obtained by subtracting the occupancy of collapse, residual column, and underwater from the cavity, thus relatively reducing the error of manual estimation of filling volume. 3. By avoiding the shared edges of the residual pillar boundary and the free boundary, the blasting boundary is generated, so that the blasting range avoids the occupation of the residual pillar and the underwater blocking range, providing a calculation basis for the blasting hole location, charge zoning and blasting design of water-filled goaf areas; 4. By deleting and recalculating uncertified boundary patches, coupling patch clusters, and back-interpolation closure, sensitive boundaries that may change the filling volume, blasting boundary, and stability boundary are identified, so that the supplementary measurement objects are transformed from ordinary blind areas into points affected by governance parameters, thereby relatively reducing invalid supplementary scanning; 5. By using the particle swarm greedy supplementary measurement algorithm, the station-based supplementary scanning, airborne supplementary scanning, cross-section supplementary scanning, or sonar supplementary measurement tasks are selected according to the parameter stabilization number, so that the on-site supplementary measurement revolves around filling, blasting, and stability calculation, and the re-measurement path under restricted working conditions is relatively shortened. 6. By writing back the supplementary measurement data and recalculating and updating the filling volume, blasting boundary and stability boundary using Newton-type iterative methods, the goaf governance decision model is corrected according to the supplementary measurement results, providing a data basis for the review and deformation monitoring after the governance is implemented. Attached Figure Description

[0015] Figure 1 This is a roadmap for the sensitive supplementary measurement technology of the goaf boundary area of ​​the present invention.

[0016] Figure 2 This is a diagram illustrating the sensitive boundary identification mechanism driven by the governance parameters of this invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Refer to the instruction manual appendix Figure 1-2 The present invention provides a method for precise detection of goaf areas using three-dimensional laser scanning, comprising: S1. The intelligent sensing system performs station scanning, UAV airborne scanning and equal-step cross-section scanning on the dry exposed area of ​​the goaf, and performs three-dimensional digital sonar depth sounding on the water-filled area of ​​the goaf. The edge computing node generates the goaf boundary patch set according to the scanning pose, echo coordinates, source marker and verification status. In this embodiment, the dry exposed area and the water-filled area of ​​the goaf are first divided according to the on-site water level, the effective area of ​​laser echo, and the sonar sounding area. The area above the water level that can form laser echo coordinates is recorded as the dry exposed area of ​​the goaf, and the area below the water level that forms sonar echo coordinates by three-dimensional digital sonar sounding is recorded as the water-filled area of ​​the goaf. The scanning pose consists of the station coordinates, azimuth angle, pitch angle, roll angle, and acquisition time. The echo coordinates are calculated from the distance, scanning angle, and scanning pose. The edge calculation node uses the goaf construction coordinate system as the coordinate reference and converts the echo data from different scanning sources into boundary patch data with source markings and verification status, so that the subsequent cavity boundary, collapse boundary, residual column boundary, underwater boundary, and roof boundary are all read from the goaf boundary patch. This implementation process includes the following steps: In S1-1, within the dry exposed area of ​​the goaf, the station-based scanning equipment collects laser ranging values, horizontal scanning angles, and vertical scanning angles at the designated station. Combined with the station's scanning pose, the laser echo coordinates generated by the station scan are calculated, and these laser echo coordinates are marked with station markers. The UAV-borne scanning equipment collects airborne laser ranging values, airborne attitude, and collection time. Combined with the UAV's airborne scanning pose, the laser echo coordinates generated by the UAV's airborne scan are calculated, and these laser echo coordinates are marked with airborne markers. The equal-step cross-section scanning proceeds according to the pre-defined... The cross-sections are numbered along the extension direction of the goaf, and laser ranging values ​​and cross-section normals are collected in each cross-section. The laser echo coordinates formed by the equal-step cross-section scanning pose are calculated by combining the equal-step cross-section scanning pose. The laser echo coordinates are marked with cross-section markers. The scanning poses of the station, the UAV-borne scanning poses, and the equal-step cross-section scanning poses are respectively bound to the laser echo coordinates formed by them. The bound fields include scanning source, acquisition time, station coordinates, and attitude angle. The bound fields are directly read and the transformation is performed during subsequent coordinate conversion. In S1-2, for the water-filled area of ​​the goaf, the 3D digital sonar collects the sounding distance, sounding direction, and sounding time at the sonar installation point. The edge computing node takes the coordinates of the installation point in the 3D digital sonar sounding pose as the starting point, takes the sounding distance along the sounding direction to form the sonar measurement point, and then converts the sonar measurement point into sonar echo coordinates. The 3D digital sonar sounding pose includes the coordinates of the sonar installation point, the sounding direction, the sonar attitude, and the sounding time. The sonar echo coordinates are configured with sonar markers and bound to the 3D digital sonar sounding pose. When there are silt, underwater walls, or underwater remnants in the water-filled area, the sonar echo coordinates retain the point arrangement relationship according to the sounding order and sounding direction. When generating the underwater boundary patch later, the edge computing node uses the sonar markers to identify the underwater source and avoids mixing the underwater contour with the laser contour of the dry exposed area. In S1-3, edge computing nodes read laser echo coordinates, sonar echo coordinates, and their bound scanning poses, and convert each echo coordinate to the goaf construction coordinate system according to the station coordinates and attitude angles. The cross-section number formed by the equal-step cross-section scanning is used as the cross-section reference. The echo coordinates formed by station scanning, UAV airborne scanning, and 3D digital sonar depth sounding are assigned to the same cross-section according to their projection position on the cross-section normal. Edge computing nodes connect adjacent echo coordinates in the same cross-section according to the scanning order to form edge patches, and connect the edge patches of adjacent cross-sections to form boundary patches. When the number of ends or points of the cross-section is not equal, triangular patches are used to supplement the patch. When the boundary patch contains one of the station marker, airborne marker, cross-section marker, and sonar marker, the verification status is recorded as single-source status. When the boundary patch contains two or more source markers, the verification status is recorded as multi-source status. The boundary patch number, endpoint coordinates, source marker, verification status, and the corresponding cross-section number together constitute the goaf boundary patch set. Through the above processing, the data generated by station scanning, UAV airborne scanning, equal-step cross-section scanning, and three-dimensional digital sonar bathymetry have completed source differentiation, coordinate conversion, cross-section attribution, and boundary patch organization before entering the subsequent governance parameter calculation; boundary patches in single-source state can enter the uncertified boundary patch extraction process, and boundary patches in multi-source state can participate in the parameter base table calculation as boundary patches after multi-source verification, thereby reducing the misjudgment of cavities, residual columns, collapses, and underwater boundaries caused by mixed scanning sources; In practical applications: the dry exposed space above the goaf is first scanned by a station to obtain the near-distance point cloud of the roof and walls, and then the point cloud of the back side of the residual pillar and the high-level concave cavity is obtained by airborne scanning by a UAV, and equal-step cross-section scanning is performed along the extension direction of the roadway; the area below the water level is obtained by three-dimensional digital sonar sounding to obtain the outline of the underwater walls and silt bodies; the edge computing node transfers the four types of data into the goaf construction coordinate system to generate a goaf boundary patch set with station markers, airborne markers, cross-section markers and sonar markers, which can be directly read for subsequent calculation of filling volume, blasting boundary and stability boundary.

[0019] S2. Substitute the cavity boundary, collapse boundary, residual column boundary, underwater boundary, and roof boundary of the goaf boundary into the treatment parameter calculation rules, deduct the collapse, residual column, and underwater occupancy from the cavity enclosure volume to obtain the filling volume, obtain the blasting boundary according to the adjacency relationship between the residual column boundary and the goaf boundary, obtain the stable boundary according to the span connection relationship of the roof boundary, and generate the parameter base table. In this embodiment, the parameter base table is calculated from the goaf boundary piece set and is used to handle subsequent tasks such as deleting and recalculating uncertified boundary pieces, generating sensitive boundary tables, and supplementary measurement. The edge computing node first classifies the boundary pieces according to the source marker, spatial location, and inter-piece connection relationship, and then converts different types of boundary pieces into cavities, occupants, blasting boundaries, and stable boundaries. Cavities are used to calculate the fillable space, while cavitation occupants, residual pillar occupants, and underwater occupants are used to deduct the occupied parts that cannot be used as fillable space from the cavities. Blasting boundaries are used to define the blasting action area, and stable boundaries are used to define the calculation range of roof stability. This implementation process includes the following steps: In S2-1, the edge computing node reads the boundary piece number, endpoint coordinates, section number, source marker, verification status, and adjacent boundary piece number set of each boundary piece in the goaf boundary piece set, and executes the source classification algorithm according to the source marker and spatial relationship. Boundary pieces with station marker, airborne marker, and section marker, whose surface normal points to the inside of the cavity, and whose surface closes with adjacent boundary pieces through a shared edge to form the inner surface of the goaf, are designated as cavity boundary pieces. Boundary pieces located inside the cavity boundary pieces, whose lower end connects to the bottom boundary, whose upper end is lower than the roof boundary pieces, and whose surface is surrounded by multiple pieces to form the outer contour of the accumulation, are designated as collapse boundary pieces. A boundary piece located inside the cavity boundary piece, with its outer periphery enclosed by closed surfaces forming a columnar outline, and vertically connecting the upper and bottom boundaries of the cavity, is designated as a residual column boundary piece; a boundary piece with sonar markers and endpoint coordinates located within the water-filled area is designated as an underwater boundary piece; a boundary piece located above the cavity boundary piece, with its two ends connected to the side wall boundary pieces via a shared edge line, is designated as a top plate boundary piece; the edge calculation node combines the cavity boundary piece, the collapse boundary piece, the residual column boundary piece, the underwater boundary piece, and the top plate boundary piece, along with the boundary piece number, source marker, verification status, endpoint coordinates, and adjacent boundary piece number set, to generate a fractal boundary piece set; In S2-2, the edge computing node inputs the cavity boundary pieces from the fractal boundary piece set into the common-edge enclosure algorithm. It establishes an inter-piece adjacency graph according to the shared edge numbering, and traverses adjacent cavity boundary pieces along the shared edge starting from any cavity boundary piece until the traversal path returns to the starting boundary piece, forming a closed cavity piece group. Triangulation is performed based on the endpoint coordinates of the closed cavity piece group, and the cavity volume is obtained by summing the directed tetrahedral volumes. After the collapse boundary piece, residual column boundary piece, and underwater boundary piece are input into the occupancy deduction algorithm, they are traversed along the same shared edge. The method forms a caving closed section group, a residual column closed section group, and an underwater closed section group, and calculates the caving occupied volume, residual column occupied volume, and underwater occupied volume respectively. If there is spatial overlap between the caving occupied volume, residual column occupied volume, and underwater occupied volume, the edge calculation node first retains the residual column occupied volume, then deletes the section group that overlaps with the residual column occupied volume from the caving occupied volume, and then deletes the section group that overlaps with the residual column occupied volume and caving occupied volume from the underwater occupied volume. Then, the filling volume is generated by subtracting the residual column occupied volume, the processed caving occupied volume, and the processed underwater occupied volume from the cavity volume. In S2-3, the edge computing node inputs the set of fractal boundary pieces into the common edge avoidance algorithm. First, it reads the shared edge line numbers of the cavity boundary pieces and the residual pillar boundary pieces. The cavity boundary pieces sharing the same edge line with the residual pillar boundary pieces are removed from the blasting candidate pieces. Then, the remaining cavity boundary pieces are spliced ​​together according to the shared edge line numbers to generate the blasting boundary. The data items of the blasting boundary include the goaf number, the blasting boundary piece number set, the endpoint coordinates of the blasting boundary pieces, and the shared edge line number set. The edge computing node inputs the roof boundary pieces into the span concatenation algorithm, reads the coordinates of the two endpoints connecting each roof boundary piece to the two sidewall boundary pieces, and calculates the two... The spatial distance between endpoint coordinates is used as the roof span. The roof boundary pieces are concatenated according to the boundary piece number and the adjacent boundary piece number set to generate a stable boundary. The data items of the stable boundary include the goaf number, the stable boundary piece number set, the roof span value, and the concatenation order. The edge calculation node establishes a binding relationship between the backfill volume, the blasting boundary, and the stable boundary according to the goaf number, the boundary piece number involved in the calculation, and the calculation time to generate a parameter base table. The parameter base table includes at least the goaf number, the calculation time, the backfill volume value, the blasting boundary piece number set, the stable boundary piece number set, and the boundary piece number set involved in the calculation. Through the above processing, the geometric surfaces of the goaf boundary patches are classified into boundary data that can be directly used in the remediation calculation. The deduction relationship between the cavity, the collapsed occupants, the residual column occupants, and the underwater occupants forms the source of the filling volume. The common edge avoidance relationship between the open space and the residual column forms the source of the blasting boundary. The span relationship between the roof boundary patch and the sidewall forms the source of the stable boundary. The parameter base table provides a calculation benchmark for subsequent deletion and recalculation, sensitive boundary determination, and supplementary measurement tasks, reducing the volume deviation and boundary misuse caused by manually delineating the remediation area based solely on the appearance of the model. In practical applications: After completing station scanning, UAV-borne scanning, equal-step cross-section scanning, and 3D digital sonar depth sounding in a certain goaf area, the edge computing node first divides the dry exposed boundary plate of the inner surface of the cavity into cavity boundary plates, the outer contour of the bottom rubble pile into caving boundary plates, the outer wall of the retained pillar that runs through the cavity into residual pillar boundary plates, the surface below the waterline formed by sonar into underwater boundary plates, and the upper surface connecting the two side walls into roof boundary plates. Then, the volume of the cavity is calculated, and the filling volume is obtained by deducting the caving occupancy, residual pillar occupancy, and underwater occupancy. The cavity boundary plates connected to the residual pillars are then removed to form blasting boundaries. Finally, stable boundaries are generated by connecting them according to the roof span, forming a parameter base table for remediation design.

[0020] S3. For uncertified boundary patches in the goaf boundary patch concentration, the sensitive recalculation algorithm of treatment parameters is used to delete and recalculate each patch. The filling volume, blasting boundary and stability boundary after deletion of each uncertified boundary patch are compared with the parameter base table item by item, and a sensitive boundary table is generated based on the uncertified boundary patches whose comparison results have changed. In this embodiment, the sensitive boundary table is used to identify uncertified boundary patches that may alter the backfill volume, blasting boundary, and stable boundary. The edge computing node first extracts boundary patches of single-source status from the goaf boundary patch set, then forms counterfactual calculation copies by removing each patch individually, observing the impact of single boundary patches and adjacent boundary patch clusters on governance parameters. Subsequently, single-patch differences and cluster coupling differences are cross-judged, and misjudgments caused by calculation gaps or cluster splitting are eliminated through back-interpolation closure, ensuring that the sensitive boundary table retains only boundary patches that have a substantial impact on governance parameters. This implementation process includes the following steps: In S3-1, the edge computing node takes the goaf boundary patch set as input, reads the boundary patch number, source marker, verification status, endpoint coordinates, and corresponding section number for each boundary patch, and uses a single-source patch extraction algorithm to read the number of source marker types in the verification status. When the number of source marker types is one, the boundary patch is formed by only one data source: station marker, airborne marker, section marker, or sonar marker, and the edge computing node defines this boundary patch as an uncertified boundary patch. When the number of source marker types is greater than one, the boundary patch is not included in the uncertified patch list. For each uncertified boundary patch, the edge computing node reads the coordinates of the two endpoints of each side of the patch and sets the coordinates of the two endpoints... The endpoint coordinates are sorted in the order of x-coordinate, y-coordinate, and elevation value. Then, the sorted endpoint coordinates are concatenated with the goaf number to form a shared edge line number. When two boundary patches have the same shared edge line number, the edge computing node determines that they have a shared edge line and forms an adjacent boundary patch number set with the other boundary patch numbers that share an edge line with the uncertified boundary patch. Thus, the edge computing node generates an uncertified patch list using the boundary patch number, source marker, verification status, shared edge line number, and adjacent boundary patch number set. When the adjacent boundary patch number set is empty, the uncertified boundary patch is still retained in the uncertified patch list and is subsequently processed as an isolated patch. In S3-2, the edge computing node inputs the uncertified boundary piece list into the counterfactual deletion algorithm, and uses a copy of the goaf boundary piece set as the calculation object; the original goaf boundary piece set is not deleted. The counterfactual deletion algorithm sequentially extracts an uncertified boundary piece according to the boundary piece number, temporarily removes the uncertified boundary piece from the calculation copy, and re-executes the cavity enclosure, occupancy deduction, common edge avoidance, and span connection according to the calculation rules used when the parameter base table was formed, to obtain the recalculated filling volume, recalculated blasting boundary, and recalculated stable boundary. The edge computing node generates the volume difference by subtracting the base table filling volume from the recalculated filling volume; and uses the recalculated blasting boundary... The XOR result between the boundary piece number set and the base table blasting boundary piece number set generates a blasting difference set, where the XOR result is a set of numbers that exist in only one of the two sets of numbers; the XOR result between the recalculated stable boundary piece number set and the base table stable boundary piece number set generates a stable difference set; if the removal of an unproven boundary piece causes the cavity or occupant to be unable to be enclosed, the edge computing node records the volume difference corresponding to this recalculation as a non-zero difference, and includes the affected blasting boundary piece number or stable boundary piece number into the corresponding difference set; after the piece-by-piece processing is completed, the edge computing node generates a single-piece difference table with the boundary piece number, volume difference, blasting difference set, and stable difference set; In S3-3, the edge computing node executes the adjacent cluster algorithm based on the uncertified boundary piece list. Each uncertified boundary piece is treated as a cluster node, and two uncertified boundary pieces with intersecting adjacent boundary piece number sets are considered as the connection relationship between nodes. Uncertified clusters are generated through connected component search. A single uncertified boundary piece that does not form a connection relationship with other uncertified boundary pieces is considered a single uncertified cluster. For each uncertified cluster, the edge computing node removes all uncertified boundary pieces within that cluster from the calculated copy of the goaf boundary piece set, and recalculates the filling volume, blasting boundary, and stability boundary according to the calculation rules of the parameter base table, obtaining the cluster volume difference, cluster blasting difference set, and cluster stability difference set. The volume difference is the recalculated filling volume after the uncertified area is removed minus the filling volume in the base table; the sum of volume differences within an area is the sum of the volume differences of each uncertified boundary area within the uncertified area in the single-area difference table; the edge calculation node generates the volume coupling difference by subtracting the sum of volume differences within the area from the volume difference of the area; the blasting difference set of the area is obtained by removing the blasting difference sets of each uncertified boundary area within the uncertified area from the blasting difference set of the area; the stability difference set of the area is obtained by removing the stability difference sets of each uncertified boundary area within the uncertified area from the stability difference set of the area; this process is used to distinguish between the situation where a single uncertified boundary area affects the governance parameters individually, and the situation where parameter changes only occur after multiple adjacent uncertified boundary areas are removed together, and finally generates the area difference table; In S3-4, the edge computing node inputs the single-piece difference table and the patch / cluster difference table into the three-parameter cross-determination algorithm, and reads the volume difference, blasting difference set, stability difference set, volume coupling difference, blasting coupling difference set, and stability coupling difference set corresponding to the same boundary patch number as the index. Volume difference and volume coupling difference are numerical fields, judged by whether they are zero; blasting difference set, stability difference set, blasting coupling difference set, and stability coupling difference set are numbered set fields, judged by whether they are empty sets. If the volume difference is not zero, the blasting difference set is not empty. If any one of the following conditions is met: empty set, non-empty stable difference set, non-zero volume coupling difference, non-empty blasting coupling difference set, or non-empty stable coupling difference set, the edge computing node adds the boundary piece number to the sensitive candidate piece list. If none of the six conditions are met, the edge computing node adds the boundary piece number to the non-sensitive piece list. This cross-determination simultaneously covers three governance purposes: filling volume, blasting boundary, and stable boundary, and reads both single-piece differences and piece-cluster coupling differences, avoiding the identification of sensitive boundary pieces based solely on single volume changes or single boundary changes. In S3-5, the edge computing node inputs the sensitive candidate piece list into the back-insertion closure algorithm and reads the corresponding uncertified boundary piece according to the boundary piece number in the sensitive candidate piece list. For sensitive candidate boundary pieces from the single-piece difference table, the edge computing node back-inserts the uncertified boundary piece into the single-piece removal copy formed in S3-2 and recalculates the filling volume, blasting boundary, and stable boundary. For sensitive candidate boundary pieces from the patch cluster difference table, the edge computing node back-inserts the uncertified boundary piece into the patch cluster removal copy formed in S3-3 and recalculates the filling volume, blasting boundary, and stable boundary. After recalculation, the volume difference, blasting difference set, stable difference set, and volume coupling are generated again. The edge computing nodes generate a sensitive boundary table using boundary patch number, source marker, verification status, volume difference, blasting difference, stable difference, volume coupling difference, blasting coupling difference, and stable coupling difference. Otherwise, the boundary patch number is merged into the non-sensitive patch column. Through the back-insertion closure process, the edge computing nodes can eliminate non-target differences caused by the fracture of the enclosing structure due to temporary removal, and retain unverified boundary patches that have indeed changed the governance parameters. Through the above processing, uncertified boundary patches are no longer simply retained as boundary patches with insufficient scanning sources, but are further examined for their actual impact on filling volume, blasting boundaries, and stable boundaries; single-patch counterfactual deletion is used to identify independent effects, adjacent patch clusters are used to identify the combined effects of adjacent boundary patches, three-parameter cross-judgment is used to distinguish the sources of change of the three types of governance parameters, and back-interpolation closure is used to confirm whether the change is caused by the corresponding uncertified boundary patch; after the sensitive boundary table is generated, subsequent supplementary measurement tasks do not need to be carried out for all single-source boundary patches, but instead focus on processing boundary patches that change governance parameters; In practical applications: The boundary patches on the back side of a certain goaf pillar were formed only by airborne scanning from a UAV, and the verification status was a single-source state. The edge computing node included them in the uncertified patch list. After deleting the counterfactual patch, the set of blasting boundary patch numbers changed, indicating that the boundary patch affected the blasting range. After multiple adjacent back side boundary patches formed an uncertified patch cluster and were removed as a whole, the set of stable boundary patch numbers changed simultaneously, indicating that these boundary patches also affected the roof span connection. After re-inserting this group of boundary patches, both the blasting difference set and the stable difference set returned to empty sets. This group of boundary patches entered the sensitive boundary table and was given priority in generating airborne or station-based supplementary scanning tasks in subsequent supplementary measurement tasks.

[0021] S4. Construct supplementary test particles using the sensitive boundary patches in the sensitive boundary table. Use the particle swarm greedy supplementary test algorithm to calculate the number of sensitive boundary patches and the parameter stabilization number corresponding to each supplementary test particle. Select supplementary test particles in the order of decreasing parameter stabilization number and decreasing number of sensitive boundary patches to generate a supplementary test task table. In this embodiment, the supplementary measurement task table is used to convert sensitive boundary patches in the sensitive boundary table into on-site supplementary measurement actions. The edge computing node first reads the source marker and six difference fields in the sensitive boundary table to determine which type of scan source is missing for each sensitive boundary patch. Then, supplementary measurement particles are constructed using the missing scan sources, and the stabilization effect on the filling volume, blasting boundary, and stable boundary after supplementary measurement is pre-calculated through virtual supplementary measurement. Finally, invalid particles and inefficient particles in the same patch are deleted, and a supplementary measurement task table is generated according to the parameter stabilization number and the boundary patch number. This implementation process includes the following steps: In S4-1, the edge computing node takes the sensitive boundary table as input and reads the boundary patch number, source marker, volume difference, blasting difference set, stability difference set, volume coupling difference, blasting coupling difference set, and stability coupling difference set one by one. The source marker is represented by a set of markers consisting of station markers, airborne markers, cross-section markers, and sonar markers. The edge computing node uses markers from the four types that have not entered the current source marker as supplementary markers. For example, if the source markers of a sensitive boundary patch only include airborne markers and cross-section markers, then the station markers... Two supplementary measurement markers are formed, one for the sonar marker and one for the ground station marker. When the supplementary measurement marker is a ground station marker, the edge computing node generates ground station supplementary scanning particles using the boundary patch number, the ground station marker, and the six-term difference field. When the supplementary measurement marker is an airborne marker, airborne supplementary scanning particles are generated. When the supplementary measurement marker is a cross-section marker, cross-section supplementary scanning particles are generated. When the supplementary measurement marker is a sonar marker, sonar supplementary measurement particles are generated. Each supplementary measurement particle retains the original boundary patch number and the six-term difference field in the sensitive boundary table. The subsequent parameter stabilization number is calculated based on this six-term difference field. In S4-2, the edge computing node inputs the station-based supplementary scanning particles, airborne supplementary scanning particles, cross-sectional supplementary scanning particles, and sonar supplementary measurement particles into the particle swarm greedy supplementary measurement algorithm. For each supplementary measurement particle, the edge computing node first merges the supplementary measurement marker into the source marker of the same boundary patch number, and then recalculates the verification state according to the number of source marker types. If the number of source marker types increases from one to two or more after merging the supplementary measurement marker, the boundary patch changes from a single-source state to a multi-source state. Subsequently, the edge computing node uses the counterfactual shard deletion algorithm to re-execute the shard removal and filling process with the supplemented source markers and verification states. The fill volume, blasting boundary, and stability boundary are recalculated to obtain the recalculated volume difference, blasting difference set, and stability difference set. Then, the adjacent cluster algorithm is used again to re-execute the overall cluster removal based on the re-measured adjacent cluster relationship, to obtain the recalculated volume coupling difference, blasting coupling difference set, and stability coupling difference set. The volume difference is zeroed out, the blasting difference set is cleared out, the stability difference set is cleared out, the volume coupling difference is zeroed out, the blasting coupling difference set is cleared out, and the stability coupling difference set is cleared out. The six counts are added to generate the parameter stabilization number, which takes a value from 0 to 6. In S4-3, the edge computing node performs particle screening with supplementary measurement particles and parameter stabilization numbers as input. Supplementary measurement particles with zero parameter stabilization numbers indicate that the addition of the supplementary measurement markers has not stabilized any field in the volume difference, blasting difference set, stability difference set, volume coupling difference, blasting coupling difference set, and stability coupling difference set, and the edge computing node deletes them. For multiple supplementary measurement particles under the same boundary patch number, the edge computing node compares the parameter stabilization numbers and deletes the supplementary measurement particles with parameter stabilization numbers less than the other supplementary measurement particle. Supplementary measurement particles with the same parameter stabilization numbers are retained until the sorting stage. During sorting, the edge computing node first sorts them in descending order of parameter stabilization numbers, then in ascending order of boundary patch numbers. When both parameter stabilization numbers and boundary patch numbers are the same, the supplementary measurement markers are arranged in the order of station markers, airborne markers, cross-section markers, and sonar markers. The sorted supplementary measurement particles are converted into a supplementary measurement task table, which includes task number, boundary patch number, supplementary measurement marker, source marker, parameter stabilization number, and task order. Through the above processing, the supplementary measurement tasks are no longer generated according to the scanning blind area or the sparseness of the point cloud, but are instead generated by inverse reasoning based on the influence of sensitive boundary patches on the filling volume, blasting boundary, and stable boundary. The station-based supplementary scanning, airborne supplementary scanning, cross-section supplementary scanning, and sonar supplementary measurement correspond to close-range verification in dry and exposed areas, high-level or back-side verification, cross-section densification verification, and underwater verification in water-filled areas, respectively. The parameter stabilization number is used to indicate the degree of elimination of the six difference fields by a single supplementary measurement, and the supplementary measurement task table is formed according to this to form the task order. In practical applications: A sensitive boundary patch has airborne markers but lacks station markers and cross-section markers. After deletion and recalculation, both blast difference sets and stable difference sets are generated. Edge computing nodes construct station supplementary scanning particles and cross-section supplementary scanning particles respectively. After virtually merging the station markers, the blast difference set is cleared, but the stable difference set still exists, and the parameter stabilization number is one. After virtually merging the cross-section markers, both the blast difference set and the stable difference set are cleared, and the parameter stabilization number is two. After screening, the cross-section supplementary scanning particles are placed before the station supplementary scanning particles, and a supplementary measurement task for the cross-section where the boundary patch is located is generated.

[0022] S5. Collect supplementary measurement data according to the supplementary measurement task table, and write the supplementary measurement data back to the goaf boundary piece set. Use Newton-type iterative recalculation algorithm to incrementally recalculate the filling volume, blasting boundary and stable boundary until the filling volume value, blasting boundary piece number set and stable boundary piece number set in the recalculation results of two adjacent rounds are the same, and generate the goaf governance decision model. In this implementation, the supplementary measurement task table is used to drive on-site supplementary measurements and write the results back to the goaf boundary patch set, so that the sensitive boundary patch changes from a single-source state to a multi-source state or adds underwater depth sounding basis; the edge computing node selects station scanning, UAV airborne scanning, equal-step cross-section scanning or three-dimensional digital sonar depth sounding according to the supplementary measurement mark to form supplementary measurement data; after the supplementary measurement data is written back, the verification state is regenerated, and then incremental recalculation is performed with the parameter base table as the first round of benchmark until the filling volume, blasting boundary and stable boundary no longer change, generating a goaf governance decision model; this implementation process includes the following steps: In S5-1, edge computing nodes take the supplementary measurement task table as input, read the task number, boundary patch number, and supplementary measurement marker in task order, and determine the supplementary measurement action based on the supplementary measurement marker. When the supplementary measurement marker is a station marker, a station is set up near the section where the boundary patch number is located, and laser ranging values, horizontal scanning angles, vertical scanning angles, and station scanning poses are collected and converted into echo coordinates for station supplementary measurement. When the supplementary measurement marker is an airborne marker, a UAV flies along the normal or side-rear direction of the section where the boundary patch number is located, collects airborne laser ranging values, airborne attitude, and UAV airborne scanning poses, and converts them into airborne supplementary measurement coordinates. The echo coordinates are measured; when the supplementary measurement mark is a cross-section mark, a densified cross-section is inserted between the cross-section where the boundary patch number is located and the adjacent cross-section, and the cross-section laser ranging value, cross-section normal and equal step cross-section scanning pose are collected, and converted into the echo coordinates of the cross-section supplementary measurement; when the supplementary measurement mark is a sonar mark, a three-dimensional digital sonar is deployed in the water-filled area where the boundary patch number is located, and the depth measuring distance, depth measuring direction and three-dimensional digital sonar depth measuring pose are collected, and converted into the echo coordinates of the sonar supplementary measurement; the edge computing node combines the echo coordinates, scanning pose, supplementary measurement mark, task number and acquisition time according to the boundary patch number to generate supplementary measurement data; In S5-2, the edge computing node reads the echo coordinates, scan pose, supplementary measurement markers, and boundary patch numbers from the supplementary measurement data. It then converts the echo coordinates in the supplementary measurement data to the coordinates of the goaf boundary patch set according to the scan pose. When the boundary patch number exists in the goaf boundary patch set, the edge computing node inserts the converted echo coordinates into the point list of that boundary patch number and reconstructs the endpoint coordinates, shared edge line numbers, and adjacent boundary patch number sets of that boundary patch according to the cross-sectional order. When a new section is discovered during supplementary measurement, the edge computing node adds the original boundary patch number to the supplementary measurement data. The service number generates an extended boundary patch number, and the newly added patches are incorporated into the goaf boundary patch set. After the supplementary measurement markers are incorporated into the source markers of the same boundary patch number, the edge computing node recounts the number of source marker types. When the number of source marker types is one, the verification status remains a single-source status. When the number of source marker types is greater than one, the verification status changes to a multi-source status. After all supplementary measurement data is written back, the edge computing node generates an updated boundary patch set with the updated boundary patch number, endpoint coordinates, shared edge line number, source marker, verification status, and adjacent boundary patch number set. In S5-3, the edge computing node updates the boundary patch set input to the Newton-type iterative recalculation algorithm, and uses the filling volume, blasting boundary patch number set, and stable boundary patch number set from the parameter base table as the results of the previous round. During the first round of calculation, the edge computing node uses the boundary patch number involved in the supplementary measurement data and its adjacent boundary patch number sets as the objects of this round of recalculation. It re-executes cavity enclosure, occupancy reduction, common edge avoidance, and span connection on the objects of this round of recalculation to generate the filling volume, blasting boundary, and stable boundary for this round. Boundary patches not included in this round of recalculation retain the results of the previous round. The edge computing node generates the volume residual by subtracting the filling volume of the previous round from the filling volume of this round, and generates the blasting residual set by XORing the blasting boundary patch number set of the current round with the blasting boundary patch number set of the previous round. The XOR result of the boundary piece number set and the previous round's stable boundary piece number set generates a stable residual set; the boundary piece numbers involved in the volume residual, blasting residual set, and stable residual set are incorporated into the next round's recalculation object; after the second round, the results of the previous round are taken from the previous round's iteration results, and incremental recalculation is performed on the next round's recalculation object and its adjacent boundary piece number sets; when the volume residual is zero, the blasting residual set is empty, and the stable residual set is empty, the edge computing node will update the boundary piece set, filling volume, blasting boundary piece number set, stable boundary piece number set, supplementary measurement task table execution status, and calculation time combination to generate a goaf governance decision model; if the recalculation object in this round is the same as the recalculation object in the previous round and the three residuals remain unchanged, the edge computing node will return the relevant boundary piece numbers to the sensitive boundary table and generate a record to be remeasured; Through the above processing, the on-site actions in the supplementary measurement task table can be written back to the original boundary piece data, so that the source markers and verification status of the boundary pieces are updated with the supplementary measurement results; the Newton-type iterative recalculation algorithm only performs incremental recalculation around the boundary pieces affected by the supplementary measurement and their adjacent boundary pieces, avoiding recalculation of the entire goaf model after each supplementary measurement, while retaining the source of change of the filling volume, blasting boundary and stability boundary; the goaf governance decision model is composed of the updated boundary piece set and the recalculated governance parameters, which can be directly used as the data basis for filling design, blasting design and stability verification; In practical applications: A boundary segment on the back side of a residual pillar in a goaf is entered into the supplementary measurement task list. The supplementary measurement is marked as a cross-section mark. On-site, a densified cross-section is inserted between the cross-section where the boundary segment is located and the adjacent cross-section, and laser scanning is completed. The edge computing node writes back the echo coordinates formed by the densified cross-section to the boundary segment number. After merging the cross-section mark into the source mark, the verification status changes from single-source status to multi-source status. Subsequently, the edge computing node performs incremental recalculation only on the boundary segment, the adjacent residual pillar boundary segment, and the adjacent roof boundary segment to obtain a new filling volume, blasting boundary segment number set, and stable boundary segment number set. After the residual is zeroed, a goaf governance decision model is generated.

[0023] exist Figure 1 It should be noted that: This diagram is a technical roadmap for sensitive supplementary measurement of goaf boundary patches. The overall process unfolds according to the execution chain of this plan. First, multi-source detection and acquisition are completed through station scanning, UAV-borne scanning, equal-step cross-section scanning, and 3D digital sonar depth sounding. Then, the scanning pose, echo coordinates, source markers, and verification status are uniformly organized into a goaf boundary patch set. The goaf boundary patch set is used for both the calculation of treatment parameters to form a parameter base table containing backfill volume, blasting boundaries, and stable boundaries. On the other hand, unverified boundary patches with only one source marker type are extracted and deleted. The sensitive boundary table is generated through recalculation, patch coupling, and back-insertion closure. Subsequently, based on the sensitive boundary table, station-based supplementary scanning, airborne supplementary scanning, cross-section supplementary scanning, and sonar supplementary measurement particles are constructed. A supplementary measurement task table is generated according to the parameter stabilization number decreasing rule. Finally, a goaf governance decision model is generated through supplementary measurement back-writing and incremental recalculation. The solid arrows in the figure represent the data flow and processing connection between the steps of the scheme. The dashed arrow "benchmark comparison" indicates that the parameter base table is not a sequential processing node, but serves as a reference during sensitive recalculation to determine whether the governance parameters have changed after deleting or back-inserting boundary patches. exist Figure 2It should be noted that: This diagram illustrates the mechanism of sensitive boundary identification driven by governance parameters, focusing on how uncertified boundary patches are identified as sensitive boundary patches. Single-source state boundary patches first enter the counterfactual deletion process. After deletion, the filling volume, blasting boundary, and stable boundary are recalculated and compared with the parameter base table. Simultaneously, the coupling comparison of patch clusters determines whether volume coupling difference, blasting coupling difference set, and stable coupling difference set are generated under the combined effect of adjacent uncertified boundary patches. When any of the six fields changes, the back-insertion closure process is initiated. If the difference can be recovered after back-insertion, it indicates that the uncertified boundary patch does indeed affect the governance parameters, and a sensitive boundary table is generated. If the six fields do not change or the back-insertion closure is not valid, it enters the non-sensitive patch list. The solid arrows in the diagram represent the normal judgment process of sensitive identification, the dashed arrows "comparison" indicate that the parameter base table is used as the reference source for difference comparison, and the dashed arrows "closure not valid" indicate that the candidate boundary patch cannot prove that the difference is caused by the boundary patch after back-insertion verification, and therefore it is transferred to the non-sensitive patch list.

[0024] Furthermore, the three-dimensional laser scanning precision detection system for goaf areas includes: The boundary acquisition module performs station scanning, UAV-borne scanning, and equal-step cross-section scanning on the dry exposed area of ​​the goaf through an intelligent sensing system, performs three-dimensional digital sonar depth sounding on the water-filled area of ​​the goaf, and generates a goaf boundary patch set by the edge computing node according to the scanning pose, echo coordinates, source markers, and verification status. The parameter generation module is used to substitute the cavity boundary plates, collapse boundary plates, residual column boundary plates, underwater boundary plates, and roof boundary plates of the goaf boundary plate set into the treatment parameter calculation rules, subtract the collapse, residual column, and underwater occupancy from the cavity enclosure volume to obtain the filling volume, obtain the blasting boundary according to the adjacency relationship between the residual column boundary plate and the goaf boundary plate, obtain the stable boundary according to the span connection relationship of the roof boundary plate, and generate the parameter base table. The sensitive recalculation module uses the governance parameter sensitive recalculation algorithm to delete and recalculate the uncertified boundary patches in the goaf boundary patch set. It compares the filling volume, blasting boundary and stability boundary of each uncertified boundary patch after deletion with the parameter base table item by item, and generates a sensitive boundary table based on the uncertified boundary patches whose comparison results have changed. The greedy supplementary testing module constructs supplementary testing particles using sensitive boundary patches from the sensitive boundary table. It then uses a particle swarm greedy supplementary testing algorithm to calculate the number of sensitive boundary patches and parameter stabilization number corresponding to each supplementary testing particle. Finally, it selects supplementary testing particles in descending order of parameter stabilization number and descending order of sensitive boundary patch number to generate a supplementary testing task table. The iterative decision-making module collects supplementary measurement data according to the supplementary measurement task table and writes the supplementary measurement data back to the goaf boundary piece set. It uses a Newton-type iterative recalculation algorithm to incrementally recalculate the filling volume, blasting boundary, and stability boundary until the filling volume value, blasting boundary piece number set, and stability boundary piece number set are the same in the recalculation results of two adjacent rounds, thus generating a goaf governance decision model.

[0025] Working principle: First, the intelligent sensing system performs station-based scanning, UAV-borne scanning, and equal-step cross-sectional scanning of the dry exposed area of ​​the goaf. Three-dimensional digital sonar depth measurement is performed on the water-filled area. Edge computing nodes convert data from different sources to the same coordinate system, generating a goaf boundary patch set with source markers and verification status. Then, the boundary patches are divided into cavity boundary patches, collapse boundary patches, residual pillar boundary patches, underwater boundary patches, and roof boundary patches. The backfill volume, blasting boundary, and stability boundary are calculated to form a parameter base table. Subsequently, unverified boundary patches formed from a single source are deleted and recalculated, patch cluster coupling recalculated, and back-inserted for verification to determine which boundary patches will change the backfill volume, blasting boundary, or stability boundary, generating a sensitive boundary table. Finally, the particle swarm optimization algorithm selects station-based, air-borne, cross-sectional, or sonar back-scanning tasks with high re-measurement value from the sensitive boundary table. The re-measurement data is written back to the model, and the backfill volume, blasting boundary, and stability boundary are updated through Newton-class iterative recalculation to form a goaf governance decision model. In practical applications, for example, in a goaf area with residual pillars obstructing the view, bottom collapse accumulation, and localized water-filled areas, conventional 3D models can display the cavity morphology, but it is difficult to determine whether the boundary behind the residual pillar will affect the blasting range, or whether underwater siltation areas should be deducted from the filling volume. This solution first establishes a boundary patch model through multi-type scanning and sonar depth sounding, and then uses edge computing nodes to determine which boundary patches come from only a single data source, and verifies their impact on governance parameters by deleting and recalculating. If the deletion of a boundary patch behind a residual pillar causes a change in the blasting boundary, that boundary patch is added to the sensitive boundary table, and the particle swarm optimization algorithm will prioritize generating cross-section rescanning or station rescanning tasks for that location. After the rescanning is completed, the system writes the new echo coordinates back to the boundary patch set, recalculates the filling volume, blasting boundary, and stability boundary, so that governance personnel can directly determine how much to fill, where to blast, and which roof areas need stability verification based on the updated model.

[0026] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for precise detection of goaf areas using three-dimensional laser scanning, characterized in that, include: S1. The intelligent sensing system performs station scanning, UAV airborne scanning and equal-step cross-section scanning on the dry exposed area of ​​the goaf, and performs three-dimensional digital sonar depth sounding on the water-filled area of ​​the goaf. The edge computing node generates the goaf boundary patch set according to the scanning pose, echo coordinates, source marker and verification status. S2. Substitute the cavity boundary, collapse boundary, residual column boundary, underwater boundary, and roof boundary of the goaf boundary into the treatment parameter calculation rules, deduct the collapse, residual column, and underwater occupancy from the cavity enclosure volume to obtain the filling volume, obtain the blasting boundary according to the adjacency relationship between the residual column boundary and the goaf boundary, obtain the stable boundary according to the span connection relationship of the roof boundary, and generate the parameter base table. S3. For uncertified boundary patches in the goaf boundary patch concentration, the sensitive recalculation algorithm of treatment parameters is used to delete and recalculate each patch. The filling volume, blasting boundary and stability boundary after deletion of each uncertified boundary patch are compared with the parameter base table item by item, and a sensitive boundary table is generated based on the uncertified boundary patches whose comparison results have changed. S4. Construct supplementary test particles using the sensitive boundary patches in the sensitive boundary table. Use the particle swarm greedy supplementary test algorithm to calculate the number of sensitive boundary patches and the parameter stabilization number corresponding to each supplementary test particle. Select supplementary test particles in descending order of parameter stabilization number and descending order of sensitive boundary patch number to generate a supplementary test task table.

2. The method for precise detection of goaf areas using three-dimensional laser scanning according to claim 1, characterized in that: Also includes: S5. Collect supplementary measurement data according to the supplementary measurement task table, and write the supplementary measurement data back to the goaf boundary piece set. Use a Newton-type iterative recalculation algorithm to incrementally recalculate the filling volume, blasting boundary and stable boundary until the filling volume value, blasting boundary piece number set and stable boundary piece number set in the recalculation results of two adjacent rounds are the same, and generate the goaf governance decision model.

3. The method for precise detection of goaf areas using three-dimensional laser scanning according to claim 2, characterized in that: S1 includes: S1-1. In the dry exposed area of ​​the goaf, the laser echo coordinates formed by the station scanning are configured with station markers, the laser echo coordinates formed by the UAV airborne scanning are configured with airborne markers, and the laser echo coordinates formed by the equal step cross section scanning are configured with cross section markers. The station scanning pose, the UAV airborne scanning pose, and the equal step cross section scanning pose are used as the scanning poses of the corresponding laser echo coordinates. S1-2. For water-filled areas in goaf areas, the sounding distance formed by three-dimensional digital sonar sounding is converted into sonar echo coordinates along the sounding direction, sonar echo coordinates are configured with sonar markers, and the three-dimensional digital sonar sounding pose is used as the scanning pose of the sonar echo coordinates. S1-3. The laser echo coordinates and sonar echo coordinates are converted to the same coordinate system according to their respective scanning poses through the edge computing nodes. Source markers are generated according to the station markers, airborne markers, cross-section markers and sonar markers. Verification status is generated according to the number of types of source markers in the same cross-section. The goaf boundary patch set is generated with the echo coordinates, source markers and verification status in the same cross-section.

4. The method for precise detection of goaf areas using three-dimensional laser scanning according to claim 3, characterized in that: S2 includes: S2-1. Using the goaf boundary piece set as input, the source classification algorithm is adopted to define the boundary pieces with station markers, airborne markers and cross-section markers that enclose the inner surface of the cavity as cavity boundary pieces, the boundary pieces located inside the cavity boundary pieces that enclose the outer surface of the accumulation as collapse boundary pieces, the boundary pieces located inside the cavity boundary pieces that penetrate the upper and lower boundaries of the cavity as residual column boundary pieces, the boundary pieces with sonar markers that are located in the water-filled area as underwater boundary pieces, and the boundary pieces located above the cavity boundary pieces that connect the two side walls as roof boundary pieces, thus generating a classification boundary piece set; S2-2. Input the cavity boundary pieces in the fractal boundary piece set into the common edge enclosure algorithm, and splice them according to the common edge line to generate the cavity. Input the cavitation boundary piece, residual column boundary piece and underwater boundary piece into the occupancy deduction algorithm, and enclose them to generate cavitation occupancy, residual column occupancy and underwater occupancy respectively. Subtract the volume of cavitation occupancy, residual column occupancy and underwater occupancy from the volume of the cavity to generate the filling volume. S2-3. Input the set of boundary segments into the common edge avoidance algorithm, take the cavity boundary segments that do not share the edge with the residual column boundary segments and splice them to generate the blasting boundary, and input the roof boundary segments into the span concatenation algorithm. Generate the roof span according to the coordinate difference between the two ends and then concatenate them according to the boundary segment number to generate the stable boundary. Then, establish the binding relationship between the filling volume, blasting boundary and stable boundary according to the goaf number, boundary segment number and calculation time to generate the parameter base table.

5. The method for precise detection of goaf areas using three-dimensional laser scanning according to claim 4, characterized in that: S3 includes: S3-1. Using the set of boundary patches of the goaf as input, the single-source patch retrieval algorithm is used to read the number of source marker types in the verification status. Boundary patches with one source marker type are defined as uncertified boundary patches. The coordinates of the two endpoints of the uncertified boundary patches are sorted and concatenated to generate a common edge line number. Boundary patch numbers with the same common edge line number are used to generate a set of adjacent boundary patch numbers. An uncertified patch column is generated using the boundary patch number, source marker, verification status, common edge line number, and adjacent boundary patch number set. S3-2. Input the uncertified boundary piece list into the counterfactual deletion algorithm, remove the uncertified boundary pieces one by one according to the boundary piece number, and recalculate the filling volume, blasting boundary and stable boundary according to the calculation rules of the parameter base table. The volume difference is generated by subtracting the filling volume of the base table from the recalculated filling volume. The blasting difference set is generated by XORing the recalculated blasting boundary piece number set with the base table blasting boundary piece number set. The stable difference set is generated by XORing the recalculated stable boundary piece number set with the base table stable boundary piece number set. A single piece difference table is generated.

6. The method for precise detection of goaf areas using three-dimensional laser scanning according to claim 5, characterized in that: S3 further includes: S3-3. Based on the uncertified sheet list, the adjacent sheet cluster algorithm is executed. Uncertified boundary sheets with intersections in the adjacent boundary sheet number sets are merged into uncertified sheet clusters. After the uncertified sheet clusters are removed as a whole, the filling volume, blasting boundary and stability boundary are recalculated. The volume coupling difference is generated by subtracting the sum of volume differences within the cluster from the volume difference of the sheet cluster. The blasting difference set within the cluster is removed from the blasting difference set of the sheet cluster and the union is used to generate the blasting coupling difference set. The stability difference set within the cluster is removed from the stability difference set of the sheet cluster and the union is used to generate the stability coupling difference set. The sheet cluster difference table is generated. S3-4. Input the single-piece difference table and the piece-cluster difference table into the three-parameter cross-determination algorithm. For the same boundary piece number, read the volume difference, blasting difference set, stability difference set, volume coupling difference, blasting coupling difference set, and stability coupling difference set. If any item satisfies that the volume difference is not zero, the blasting difference set is not empty, the stability difference set is not empty, the volume coupling difference is not zero, the blasting coupling difference set is not empty, or the stability coupling difference set is not empty, then generate a sensitive candidate piece list for the boundary piece number; otherwise, generate a non-sensitive piece list for the boundary piece number.

7. The method for precise detection of goaf areas using three-dimensional laser scanning according to claim 6, characterized in that: S3 further includes: S3-5. Input the sensitive candidate piece list into the back-insertion closure algorithm. Insert the unverified boundary pieces in the sensitive candidate piece list back into the removed goaf boundary piece set according to the boundary piece number, and recalculate the filling volume, blasting boundary, and stable boundary. If the volume difference is zero, the blasting difference set is empty, the stable difference set is empty, the volume coupling difference is zero, the blasting coupling difference set is empty, and the stable coupling difference set is empty after back-insertion, then generate a sensitive boundary table with the boundary piece number, source marker, verification status, volume difference, blasting difference set, stable difference set, volume coupling difference, blasting coupling difference set, and stable coupling difference set. Otherwise, merge the boundary piece number into the non-sensitive piece list.

8. The method for precise detection of goaf areas using three-dimensional laser scanning according to claim 7, characterized in that: S4 includes: S4-1. Using the sensitive boundary table as input, read the boundary patch number, source marker, volume difference, blasting difference set, stability difference set, volume coupling difference, blasting coupling difference set, and stability coupling difference set. Take the markers that have not entered the source marker among the station marker, airborne marker, cross-section marker, and sonar marker as supplementary measurement markers, and generate station supplementary scanning particles, airborne supplementary scanning particles, cross-section supplementary scanning particles, and sonar supplementary measurement particles according to the supplementary measurement markers. S4-2. Input the station-based supplementary scanning particles, airborne supplementary scanning particles, cross-sectional supplementary scanning particles, and sonar supplementary measurement particles into the particle swarm greedy supplementary measurement algorithm. After merging the supplementary measurement markers into the source markers of the same boundary piece number, recalculate and verify the state. Recalculate the volume difference, blasting difference set, stable difference set, volume coupling difference, blasting coupling difference set, and stable coupling difference set using the counterfactual piece deletion algorithm and the adjacent piece cluster algorithm. Generate the parameter stabilization number based on the number of valid items when the volume difference is zeroed, the blasting difference set is cleared, the stable difference set is cleared, the volume coupling difference is zeroed, the blasting coupling difference set is cleared, and the stable coupling difference set is cleared. S4-3. Delete the supplementary measurement particles with a parameter stabilization number of zero, delete the supplementary measurement particles with a parameter stabilization number less than the other supplementary measurement particle under the same boundary patch number, and generate a supplementary measurement task table according to the rules of decreasing parameter stabilization number, ascending boundary patch number, and supplementary measurement mark arranged according to the station mark, airborne mark, cross section mark and sonar mark.

9. The method for precise detection of goaf areas using three-dimensional laser scanning according to claim 8, characterized in that: S5 includes: S5-1. Using the supplementary survey task table as input, read the boundary patch number and supplementary survey mark. When the supplementary survey mark is a station mark, perform station scanning; when the supplementary survey mark is an airborne mark, perform UAV airborne scanning; when the supplementary survey mark is a cross-section mark, perform equal-step cross-section scanning; when the supplementary survey mark is a sonar mark, perform three-dimensional digital sonar depth sounding. Generate supplementary survey data by collecting the echo coordinates, scanning pose, and supplementary survey mark according to the boundary patch number. S5-2. Convert the echo coordinates in the supplementary measurement data to the coordinates of the goaf boundary patch set according to the scanning pose. Merge the converted echo coordinates into the boundary patch with the same boundary patch number. After merging the supplementary measurement mark into the source mark, recalculate and verify the status according to the number of source mark types, and generate an updated boundary patch set. S5-3. Input the updated boundary piece set into the Newton-type iterative recalculation algorithm. Generate the current round's filling volume, blasting boundary, and stable boundary according to the calculation rules of the parameter base table. Subtract the previous round's filling volume from the current round's filling volume to generate the volume residual. Generate the blasting residual set by XORing the current round's blasting boundary piece number set with the previous round's blasting boundary piece number set. Generate the stable residual set by XORing the current round's stable boundary piece number set with the previous round's stable boundary piece number set. Use the boundary piece numbers that generate the volume residual, blasting residual set, and stable residual set as the recalculation objects for the next round, until the volume residual is zero, the blasting residual set is empty, and the stable residual set is empty, thus generating the goaf governance decision model.

10. A three-dimensional laser scanning precision detection system for goaf areas, characterized in that, include: The boundary acquisition module performs station scanning, UAV-borne scanning, and equal-step cross-section scanning on the dry exposed area of ​​the goaf through an intelligent sensing system, performs three-dimensional digital sonar depth sounding on the water-filled area of ​​the goaf, and generates a goaf boundary patch set by the edge computing node according to the scanning pose, echo coordinates, source markers, and verification status. The parameter generation module is used to substitute the cavity boundary plates, collapse boundary plates, residual column boundary plates, underwater boundary plates, and roof boundary plates of the goaf boundary plate set into the treatment parameter calculation rules, subtract the collapse, residual column, and underwater occupancy from the cavity enclosure volume to obtain the filling volume, obtain the blasting boundary according to the adjacency relationship between the residual column boundary plate and the goaf boundary plate, obtain the stable boundary according to the span connection relationship of the roof boundary plate, and generate the parameter base table. The sensitive recalculation module uses the governance parameter sensitive recalculation algorithm to delete and recalculate the uncertified boundary patches in the goaf boundary patch set. It compares the filling volume, blasting boundary and stability boundary of each uncertified boundary patch after deletion with the parameter base table item by item, and generates a sensitive boundary table based on the uncertified boundary patches whose comparison results have changed. The greedy supplementary testing module constructs supplementary testing particles using sensitive boundary patches from the sensitive boundary table. It then uses a particle swarm greedy supplementary testing algorithm to calculate the number of sensitive boundary patches and parameter stabilization number corresponding to each supplementary testing particle. Finally, it selects supplementary testing particles in descending order of parameter stabilization number and descending order of sensitive boundary patch number to generate a supplementary testing task table. The iterative decision-making module collects supplementary measurement data according to the supplementary measurement task table and writes the supplementary measurement data back to the goaf boundary piece set. It uses a Newton-type iterative recalculation algorithm to incrementally recalculate the filling volume, blasting boundary, and stability boundary until the filling volume value, blasting boundary piece number set, and stability boundary piece number set are the same in the recalculation results of two adjacent rounds, thus generating a goaf governance decision model.

Images