Construction method and system of earth surface contour line model

Abstract

The invention relates to a construction method and system of a surface contour line model. The construction method comprises the following steps: S1, determining at least one ground object of a plant, and dividing the plant according to the number of the selected ground objects to form sub-plants; s2, according to the record of the factory file, screening the geometries belonging to each sub-factory, and leaving the geometries meeting a predetermined elevation condition; s3, simplifying the factory file according to a mode of only retaining the elevation of the sub-factory and the geometry meeting the predetermined elevation condition, and outputting the factory with the geometry meeting the predetermined elevation condition; and S4, carrying out fusion on the generated ContoursNew contour line set and the topographic file. The construction system can execute the construction method and comprises one or more of the following hardware: a data loading module, a space calculation module, a graphic processing unit and a storage array.

Description

Technical Field

[0001] The present invention relates to the technical field of surveying and mapping processing, and in particular to a method and system for constructing a surface contour model. Background Art

[0002] In the field of surveying and mapping, the construction of terrain contour lines based on 3D models is a crucial foundation for site planning, flood simulation, and volume calculations. Traditional surveying and mapping methods use total stations, lidar, and other equipment to acquire surface point cloud data. These data are then generated using Delaunay triangulation or regular grid interpolation to create a continuous terrain surface. Contour lines are then extracted to represent elevation distribution. For complex scenarios like industrial plants, existing technologies typically treat 3D building models and natural terrain models as separate datasets, rendering existing volume calculation methods inapplicable.

[0003] CN107247835A discloses a method for verifying the accuracy of reservoir capacity calculation. The method includes the following steps: simulating reservoir terrain using a concentric frustum model; constructing N concentric circular rings as a model for simulating reservoir terrain contour lines, with each ring representing a contour line, and the elevation values ​​of the contour lines increasing in an arithmetic progression from the inside out; calculating the theoretical volume of the simulated reservoir using the frustum volume formula, and obtaining the theoretical water surface area of ​​the simulated reservoir using the circle area formula; calculating the capacity of the simulated reservoir using the reservoir capacity calculation method with the accuracy to be verified, to obtain the volume and water surface area of ​​the simulated reservoir; and comparing the theoretical volume and theoretical reservoir area of ​​the simulated reservoir with the volume and water surface area calculated using the reservoir capacity calculation method with the accuracy to be verified, to calculate the accuracy value of the reservoir capacity calculation method with the accuracy to be verified.

[0004] However, existing technologies face significant bottlenecks when processing factory terrain transformed through 3D scene reconstruction. First, when artificial structures and natural terrain form a multi-level floor structure, the original terrain model is cut into discrete 3D blocks, resulting in a loss of contour continuity and the inability to directly perform volume integral calculations. Second, factory 3D model files (such as the .rvm format) typically contain tens of thousands of geometric entities, and their node attributes lack a unified elevation annotation standard, resulting in a large amount of ambiguity in floor object recognition and elevation extraction. Furthermore, existing fusion algorithms struggle to accurately represent cliff effects when dealing with the junction of regular building outlines and curved terrain, often resulting in abnormal contour line interruptions or non-physical overlaps. Furthermore, traditional methods rely on spatial relationship detection through full model traversal, which suffers from low computational efficiency and excessive memory usage when dealing with large-scale datasets. Efficiently fusing multi-source heterogeneous models while maintaining the topological integrity of contour lines has become a key challenge restricting the accuracy of engineering analysis.

[0005] In addition, on the one hand, there are differences in understanding among those skilled in the art; on the other hand, the applicant studied a large number of documents and patents when making the present invention, but due to space limitations, not all details and contents are listed in detail. However, this does not mean that the present invention does not have the characteristics of these prior arts. On the contrary, the present invention already has all the characteristics of the prior art, and the applicant reserves the right to add relevant prior art to the background technology. Summary of the Invention

[0006] To address the shortcomings of existing technologies, the present invention provides a method and system for constructing a surface contour model, addressing at least some of the aforementioned technical issues. This method combines key model objects on a 3D model with a ground model to form a local surface contour model. This local surface contour model is then integrated into the original 3D ground model, enabling continued volumetric calculations and providing effective data guidance for subsequent flood and sewage analysis.

[0007] The present invention discloses a method for constructing a surface contour model, which comprises the following steps:

[0008] S1. Determine at least one ground object in the factory area, and divide the factory area into sub-factory areas according to the number of the selected ground objects;

[0009] S2. According to the records in the factory area files, the geometric bodies belonging to each sub-factory area are screened and the geometric bodies that meet the predetermined elevation conditions are retained;

[0010] S3. Simplify the plant area file by retaining only the "elevation of the sub-plant area" and the "geometry that meets the predetermined elevation conditions," and output the plant area with the geometry that meets the predetermined elevation conditions;

[0011] S4. Fuse the generated ContoursNew contour line set with the terrain file.

[0012] This method effectively solves the compatibility problem of multi-source data by generating a standardized .dwg format fusion file. After completing the sub-factory division, geometry screening and model simplification, the output file discards redundant .rvm model data and only retains the contour terrain features. Because the .dwg format is natively supported by engineering software such as ArcGIS and AutoCAD Civil 3D, it can be directly used in scenarios such as flood risk zoning analysis and debris flow path simulation. Contour data is generated through forced elevation assignment and closed loops to form a unified and abstract terrain expression model. Its simplified data structure reduces the complexity of downstream system analysis and improves the efficiency of terrain parameter calling in disaster protection strategy formulation.

[0013] According to a preferred embodiment, step S1 includes: identifying ground objects by parsing the node attributes or node names of the plant model file, hierarchically clustering the ground objects through the top elevation values ​​of the bounding boxes to determine the horizontal plane divisions, each horizontal plane corresponds to a sub-plant area, and its spatial boundary is determined by the two-dimensional projection convex hull of the ground object to which it belongs.

[0014] The present invention realizes automatic horizontal plane division through node attribute recognition and hierarchical clustering algorithm, solving the problem of manual intervention caused by the lack of floor object attributes. The node attribute priority matching mechanism uses industrial software standard attribute tags to quickly locate the floor, and the hierarchical clustering automatically merges similar elevation areas based on the bounding box top elevation, overcoming the grouping deviation caused by the distribution of discrete elevation points. The convex hull boundary calculation accurately describes the projection range of the sub-factory area by extracting the bottom surface vertex set and performing geometric simplification, avoiding the invalid area inclusion caused by the direct use of the bounding box. This technology frees the horizontal plane division process from dependence on manual labeling, while ensuring that the sub-factory area boundary is consistent with the actual terrain topology.

[0015] According to a preferred embodiment, node attribute identification includes: traversing the model tree nodes, detecting nodes containing preset attribute identifiers or name keywords, recursively penetrating the nested node structure until the underlying geometry is obtained, and / or dynamically updating the ground object set based on the user's manual selection results.

[0016] A hybrid recognition strategy combines automated inspection with manual intervention to address the heterogeneity of attribute tags within complex model trees. Recursive penetration through nested node structures ensures complete capture of floor geometry within deep assemblies, and a regular expression matching engine supports multilingual keyword recognition, adapting to the diverse naming requirements of international engineering projects. The user manual selection function precisely maps 3D coordinates to model tree nodes, and its secondary confirmation mechanism effectively mitigates the risk of misuse. This technical solution maintains a high level of automation while providing flexible manual correction channels for specialized scenarios.

[0017] According to a preferred embodiment, step S2 includes: performing volume screening on the geometric bodies of non-ground objects to exclude objects that do not meet the predetermined volume conditions; performing Z-axis projection overlap detection and XY plane inclusion judgment on the remaining geometric bodies, and retaining the geometric bodies that match the spatial boundary and elevation benchmark of the sub-factory area.

[0018] The geometry screening strategy accurately preserves key structures to generate a terrain dataset that meets the needs of disaster modeling. Volumetric screening excludes non-topography-related components such as pipes and supports, focusing the output file on key buildings that influence surface runoff. Z-axis projection overlap detection ensures real-world interaction between retained objects and the ground, providing effective hazard-bearing volume data for landslide potential analysis. The XY plane contains the determined spatial topological relationships, which are directly mapped to the inundation area calculation parameters in flood simulation, preventing invalid geometry from affecting algorithm convergence.

[0019] According to a preferred embodiment, volume screening includes: calculating the axial dimensions of the bounding box of the geometric body, marking it as a candidate object when each axial dimension exceeds a preset threshold and the sum of the dimensions meets a predetermined condition, and enabling convex hull volume calculation for geometric bodies with hollow structures.

[0020] The convex hull volume calculation mode optimizes screening accuracy for hollow structures, addressing the inherent flaws of traditional bounding box sizing. When internal voids are detected in a geometric object, the system automatically switches to a convex hull generation algorithm, calculating the effective volume to avoid misjudgment. This technology is particularly suitable for typical industrial scenarios such as frame structures, ensuring that screening results accurately reflect the spatial characteristics of the component.

[0021] According to a preferred embodiment, step S3 includes: performing bounding box projection inclusion relationship judgment on the ground object set, and eliminating the ground objects that are completely contained; performing Boolean union operation or bounding box fusion on adjacent ground objects with consistent elevations to generate a merged geometric body set.

[0022] Ground object merging and contour simplification operations directly support the efficient use of data in environmental assessments. Boolean operations are used to merge adjacent floor plates into a single geometric body, eliminating the interference of fragmented polygons in GIS overlay analysis. The Douglas-Peucker algorithm simplifies the bottom contours, preserving curvature while reducing the number of vertices, reducing the time required to load terrain data for large-scale industrial sites into ArcGIS. This simplification mechanism ensures that the output data meets the accuracy requirements of agricultural land planning while adapting to the real-time rendering performance of mobile mapping tools.

[0023] According to a preferred embodiment, the Boolean union operation includes: extracting geometric triangular mesh data, constructing a space partitioning tree to accelerate intersection surface calculation, deleting redundant facets and stitching boundary edges, and switching to bounding box union to generate an alternative geometry when the operation fails.

[0024] A fault-tolerant Boolean operation mechanism enhances the robustness of the model simplification process. When encountering non-manifold geometry, the system switches to bounding box fusion mode to generate substitute geometry, preserving key elevation features while avoiding process interruptions. This technology significantly improves the success rate of model simplification, especially when dealing with mixed level of detail models, ensuring continuous execution.

[0025] According to a preferred embodiment, step S4 includes: constructing a spatial index structure of the original terrain contour lines, traversing each contour line in the new contour line set, retrieving adjacent terrain contour lines and determining the inclusion, intersection or separation relationship, and performing line segmentation and contour line replacement for the intersection situation.

[0026] Dynamic spatial indexing and segmentation ensure that the fused contour data meets both accuracy and compatibility requirements. R-Tree indexing accelerates terrain line retrieval, ensuring that the output file retains only valid terrain feature data, significantly reducing storage requirements. The precise intersection coordinates generated by the Bentley-Ottmann algorithm conform to CAD software geometry specifications, avoiding topological errors when imported into third-party systems. Elevation change markers generated during cliff effect processing provide feature markers for terrain rendering in tools like GlobalMapper, enabling vertical boundaries to be accurately identified as critical risk areas in disaster simulations.

[0027] According to a preferred embodiment, determining the inclusion relationship includes: using the ray method to analyze the projection inclusion status of the new contour line and the terrain line in the XY plane, combining the elevation difference to trigger the cliff effect processing, and performing the endpoint extension and closing operation on the non-closed contour line.

[0028] The ray method includes a judgment and closed-loop extension mechanism, enhancing data consistency across platforms. Projecting standard polygons generated by inclusion state analysis ensures that contour lines in Surfer are meshed without distortion due to boundary gaps. The endpoint extension algorithm automatically connects open line segments to the nearest terrain line, meeting the slope calculation requirements for continuous contour lines in forestry planning. The rule for generating duplicate coordinate points triggered by elevation differences enables AutoCAD Civil 3D to accurately interpret the transitions between buildings and natural terrain, providing reliable input for earthwork calculations in infrastructure construction.

[0029] The present invention also discloses a system for constructing a surface contour model. The system is capable of executing the aforementioned construction method and includes one or more of the following hardware:

[0030] Data loading module, used to parse the tree node structure of the factory 3D model file and extract the geometric data of the ground objects;

[0031] Spatial computing module, used to perform bounding box containment relationship determination, elevation interval clustering and spatial projection relationship analysis;

[0032] A graphics processing unit for performing Boolean operations on three-dimensional geometry, bottom contour line generation, and contour line projection transformation;

[0033] The storage array is used to cache the original terrain contour dataset, intermediate processing results, and the fused plant terrain model in layers. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a schematic diagram provided by the present invention for identifying ground objects by node attributes or names;

[0035] Figure 2 is a flowchart of the steps of the construction method provided by the present invention;

[0036] Figure 3 is a schematic diagram of a terrain file (.dwg format) in one embodiment provided by the present invention;

[0037] Figure 4 1 is a top view of a factory file (.rvm format) in an embodiment provided by the present invention;

[0038] Figure 5 1 is an axial schematic diagram of a factory file (.rvm format) in an embodiment provided by the present invention;

[0039] Figure 6 is an extracted and simplified model diagram in an embodiment provided by the present invention;

[0040] Figure 7 is a simplified model contour line diagram in an embodiment provided by the present invention;

[0041] Figure 8 This is a schematic diagram of a plant area + terrain file (.dwg format) after integrating the model contour lines in an embodiment provided by the present invention;

[0042] Figure 9 is a hardware configuration diagram of the construction system provided by the present invention;

[0043] Figure 10 is a schematic diagram of a triangular mesh model of ground objects in an example three-dimensional factory model;

[0044] Figure 11 This is a contour effect display diagram of the example;

[0045] Figure 12 This is a schematic diagram of the boundary stroke effect of the model extraction of the ground object in the example factory 3D model.

[0046] Reference Signs List

[0047] 100: data loading module; 200: spatial computing module; 300: graphics processing unit; 400: storage array. DETAILED DESCRIPTION

[0048] The following is a detailed description with reference to the accompanying drawings.

[0049] Ground object (i.e., floor): The foundation surface of the main structure inside and outside the building, constructed of engineering materials such as concrete, asphalt, or masonry, and processed to achieve a flat, solid, and durable effect. It is a 3D model component with flatness and bearing capacity characteristics. It is identified by the node attribute "TYPE = FRMW / FLOOR" or the name contains "floor", such as Figure 1 As shown in the figure, the text is a software screenshot. The design and construction of ground objects involve multiple aspects, including but not limited to foundation preparation and screed installation. Each type has different performance requirements and technical specifications depending on its intended use. As the foundation of factory construction, ground objects play a vital role in subsequent factory construction. Figure 10 An example of a triangular mesh model of ground objects in a three-dimensional factory model is shown.

[0050] Bounding box: A simplified geometric structure used to represent the boundaries of an object in computer graphics, computational geometry, physical simulation, and video games. It is an abstract geometric body that is usually used to approximate or completely contain an object of a more complex shape. The main purpose of the bounding box is to simplify and accelerate various spatial queries and calculations, such as collision detection, visibility testing, object selection, etc. For a three-dimensional geometric object, its bounding box contains two points: the lower left corner point LB(x,y,z) and the upper right corner point RT(x,y,z), satisfying that the coordinate components of LB are all less than or equal to the corresponding coordinate components of RT (that is, LB.x<=RT.x and LB.y<=RT.y and LB.z<=RT.z), which is used for fast spatial relationship determination.

[0051] Elevation: The spatial coordinate value of a surface point in the vertical direction, which is associated with the plane coordinates (x, y) using the z = f(x, y) function, and constitutes the basic parameters for terrain modeling, where x and y are plane coordinates and z represents elevation information.

[0052] Cliff effect: When the artificial floor is integrated with the natural terrain, the vertical boundary feature is formed by the sudden change in elevation of the regular building outline and the curved terrain, which is manifested as the dense overlap or sudden interruption of the contour lines at the intersection.

[0053] Contour line: A closed curve formed by all continuous points with equal elevation in a 3D terrain model. It is used to visualize the terrain undulation characteristics and elevation distribution patterns. The contour line effect of the example is shown in the figure below. Figure 11 shown.

[0054] Volume intersection operation: A spatial Boolean intersection operation between three-dimensional geometric models. It generates an interface by calculating the overlapping volume between the models, and is used to extract the spatial geometric data of the area where the building bottom surface contacts the terrain.

[0055] Volume calculation: Based on the fused contour model, the volume of the enclosed space between the surface and the reference surface in the specified area is calculated using the spatial integration method.

[0056] Sub-factory area: An independent elevation unit formed by dividing the 3D model of the factory area according to the horizontal plane. Each unit contains a set of ground objects with uniform elevation characteristics, which is used for hierarchical processing of complex model data.

[0057] Node feature: A topological unit that identifies component attributes in a 3D model data structure. It identifies specific types of objects through node name string matching or key-value attributes, such as the "TYPE" attribute filtering mechanism for ground objects.

[0058] Contour line: A closed polygon representing the outer edge of a three-dimensional object on a horizontal projection plane, specifically the two-dimensional boundary where the base of a building meets the terrain, used to simplify geometric expression during three-dimensional fusion calculations. Figure 12 The figure shows the boundary stroke effect of model extraction for ground objects in an example 3D model of a factory area.

[0059] Example 1

[0060] like Figure 2 As shown, the present invention discloses a method for constructing a surface contour model, which includes the following steps:

[0061] S1. Determine at least one ground object in a factory area, and divide the factory area into sub-factory areas according to the number of the selected ground objects.

[0062] The terrain file (.dwg format) and factory file (.rvm format) of the example are as follows: Figure 3 and Figure 4 and Figure 5 As shown. Preferably, in step S1, the plant file (.rvm format) can be loaded first and its data structure can be parsed, the node hierarchy in the model management tree structure can be traversed, and the ground objects can be identified through a matching mechanism, wherein the scope of the ground objects can be clarified through the attributes of the model objects, the node names of the model management tree structure, or manual selection by the user. Furthermore, the specific matching mechanism includes: priority retrieval of the identifier "TYPE=FRMW / FLOOR" in the node attribute set; when the attribute field is missing, scanning can be performed to see if the node name contains keywords such as "floor" or "ground object" (the matching process uses a full-word string matching algorithm to ensure accuracy), and the geometry of the nodes that meet the conditions is marked as the initial ground object set; the ground objects can also be determined by manual selection by the user.

[0063] Exemplarily, when parsing plant files (in.rvm format), the traversal of the model tree structure is implemented using the Depth-First Search algorithm to achieve complete access to node levels. Specifically, during implementation, each child node is recursively accessed starting from the root node. When a key-value pair of "TYPE=FRMW / FLOOR" is detected in the node attribute set, the node is immediately marked as a floor node, and its geometric index number is recorded. For the matching of node names, the system establishes a regular expression matching engine and uses the " / floor / i" pattern to achieve case-insensitive fuzzy matching. Recognition is triggered when the node name contains the Chinese character "floor" or the pinyin "diping" (Note: The multi-language matching ability is a common supplement based on the internationalization requirements of industrial software). When encountering a nested node structure (such as a ground object containing sub-assemblies), the traversal module automatically penetrates the hierarchical relationship to ensure that the geometries of all leaf nodes are detected.

[0064] Exemplarily, in the graphical user interface (GUI), the system can provide a box selection tool and a lasso selection tool for the 3D model. The user can perform multiple selection operations, and the spatial coordinate information of the selected geometries is captured in real time through the OpenGL picking algorithm. The system maps the user selection set to the model tree structure, reversely locates the corresponding node path, and creates a temporary attribute label UserSelected=TRUE in memory. To prevent misoperations, the system sets a secondary confirmation dialog box, and a warning prompt is popped up when the overlap rate between the manually selected object and the automatic recognition result is more than 50%.

[0065] Furthermore, the set of identified ground objects will serve as the basis for horizontal plane division, and each ground object corresponds to an independent horizontal plane. When there are multiple non-continuous floors (for example, there are sunken squares or overhead platforms in the factory area), the corresponding number of horizontal planes will be automatically generated according to the three-dimensional spatial distribution of the ground objects. For the determination of the number of horizontal planes, a spatial elevation clustering algorithm is implemented: the RT.z values ​​of the bounding boxes of each ground object are grouped according to the engineering accuracy threshold, and the ground objects in the same elevation range are merged into a single horizontal plane, and finally a set of horizontal planes covering the entire factory area is formed. For example, for the identified ground object set, the RT.z value of the bounding box of each geometric body is extracted to form an elevation data set {z1, z2, ..., zn}. A hierarchical clustering algorithm (HierarchicalClustering) is used for grouping: first, the Euclidean distance matrix of all elevation point pairs is calculated, and the longest distance method (Complete-linkage) is used for clustering and merging. The elevation error threshold ε allowed by the project can be set to 0.1 meters. When the inter-cluster distance between two clusters is less than ε, they are merged into the same horizontal plane, and the final output cluster center sequence {Zg1, Zg2, ..., Zgk} is used as the division basis. For example, if the elevation of the ground objects in a factory area is [10.05m, 10.12m, 15.30m, 15.28m], the algorithm will generate two horizontal planes: Zg1 = 10.08m (the average of 10.05 and 10.12) and Zg2 = 15.29m.

[0066] Furthermore, during the partitioning process, each horizontal plane corresponds to an independent sub-factory area, and its spatial extent is determined by the two-dimensional projection boundary of the floor within that plane. When a single horizontal plane covers the entire plant area, a complete, unsegmented model is formed. The sub-factory area partitioning results output from this stage provide a spatial topological basis for geometric object screening in the subsequent S2 stage, ensuring that only key structures in the corresponding elevation layer are retained within each sub-area. For example, the two-dimensional projection extent of each horizontal plane corresponding to the sub-factory area is determined by calculating the convex hull of the corresponding ground objects. The specific steps include: 1) extracting the coordinates of the bottom vertices of all ground objects on the horizontal plane; 2) calculating the convex hull boundary of the point set using the Graham sweep algorithm; and 3) performing Ramer-Douglas-Peucker simplification on the convex hull polygon, retaining vertices with a curvature change greater than 5°. The resulting polygon is stored in WKT (Well-Known Text) format, such as POLYGON((x1 y1, x2 y2, ..., xn yn)), which serves as the spatial topological boundary of the sub-factory area.

[0067] Preferably, if no ground objects are detected in the model tree, the system can initiate an emergency response process, which may include: 1) logging an "E001: No ground objects" error in the log file; 2) sending a red warning light signal to the user interface; and 3) automatically switching to the lowest elevation mode, using the LB.z value of the entire plant's 3D model bounding box as the default horizontal plane elevation. The generated sub-plant area now covers the entire plant area, and the AutoGenerated = true metadata tag is added to the output file.

[0068] S2. According to the records in the factory area documents, the geometric bodies belonging to each sub-factory area are screened, and the geometric bodies that meet the predetermined elevation conditions are retained.

[0069] Preferably, the system can perform multi-dimensional geometric object screening based on the results of the sub-factory area division. There are usually many objects in a factory area, but certain objects such as large buildings and large equipment occupy a large amount of ground volume and have relatively regular shapes. Therefore, these specific objects can be screened in step S2.

[0070] Preferably, after completing the sub-factory division, the system can traverse the geometric body dataset in the factory area file (.rvm format), first excluding the identified ground object models through the model tree node attribute filtering mechanism, and then performing a 3D bounding box analysis on the remaining geometric bodies to perform preliminary filtering by judging the model volume, thereby obtaining the objects to be selected. For example, for each geometric body, the absolute value of its axial bounding box in the X / Y / Z axis direction is calculated. If the length, width, and height are all greater than 1 meter, and the sum of the three values ​​exceeds 10 meters, it is marked as a candidate object. The logic for determining the volume filter condition "length, width, and height are all greater than 1 meter, and the sum of the three exceeds 10 meters" is as follows: 1) Calculate the dimensions of the geometry's bounding box in the X / Y / Z axes: Δx = RT.x - LB.x, Δy = RT.y - LB.y, and Δz = RT.z - LB.z; 2) Set Boolean flags flag1 = (Δx > 1.0 && Δy > 1.0 && Δz > 1.0); 3) Calculate the sum of the dimensions: sum = Δx + Δy + Δz; 4) Set Boolean flags flag2 = (sum > 10.0); 5) Final determination: valid = flag1 && flag2. If the geometry contains voids (such as framed buildings), the system automatically uses the convex hull volume calculation mode: the convex hull is generated using the QuickHull algorithm and the valid volume is recalculated.

[0071] Furthermore, for the set of geometric bodies (i.e., objects to be selected) that have passed the volume screening, spatial relationship judgment is further performed: the RT.z value of the bounding box of the corresponding ground object of each sub-factory area is extracted as the elevation reference, and the LB.z and RT.z values ​​of the bounding box of the selected geometric body are compared. If it is completely above the ground elevation (geometric body LB.z>ground RT.z) or below (geometric body RT.z<ground LB.z), a culling operation is performed, such as overhead pipelines. Exemplarily, the steps for determining the spatial relationship in the Z-axis direction include: 1) obtaining the bounding box Ground_AABB of the ground object in the sub-factory area; 2) calculating the overlapping interval [Z_min, Z_max] between the bounding box Obj_AABB of the selected object that passes the volume screening and the projection of Ground_AABB in the Z-axis direction, where Z_min = max(Obj_AABB.LB.z, Ground_AABB.LB.z), Z_max = min(Obj_AABB.RT.z, Ground_AABB.RT.z); 3) when Z_min>Z_max, it is determined that there is no Z-axis overlap, and a culling operation is performed; 4) calculating the effective overlap height Δh = Z_max-Z_min, and when Δh<0.01 meters, it is considered as invalid overlap to prevent misjudgment caused by floating-point operation errors.

[0072] Furthermore, for geometric bodies that overlap with the ground on the Z axis (geometric body LB.z ≤ ground RT.z and geometric body RT.z ≥ ground LB.z), a projection inclusion detection algorithm is used to verify whether it is located within the spatial boundary of the sub-factory area to which it belongs. The XY plane projection detection process is as follows: 1) Sort the vertex sequence of the sub-factory boundary polygon {P1(x1,y1),P2(x2,y2),...,Pn(xn,yn)} in a clockwise order; 2) Calculate the center point C of the bounding box of the candidate geometry = ((Obj_AABB.LB.x + Obj_AABB.RT.x) / 2, (Obj_AABB.LB.y + Obj_AABB.RT.y) / 2); 3) Shoot a horizontal ray from point C to the right and count the number of intersections with each edge of the polygon; 4) Use the Bentley-Ottmann algorithm to optimize intersection calculations and establish a sweep line status tree and event queue, reducing the algorithm complexity to O((n+k)log n); 5) If the number of intersections is odd, the point is considered internal; otherwise, it is considered external. The resulting retained geometry can be used as a retained object in subsequent processes.

[0073] Preferably, when the system detects that a sub-factory area does not have a qualified geometry, it automatically triggers the elevation benchmark verification process and recalculates the bounding box RT.z value of the ground object in the area to prevent mis-filtering. For example, when a sub-factory area has no valid geometry, the verification process is triggered: 1) Reload the ground object set corresponding to the sub-factory area; 2) Perform bounding box integrity verification: Check whether the RT.z value of Ground_AABB is within a reasonable engineering range (such as -50m to +500m), and mark it as an abnormality if it is out of range; 3) Start the ground object re-identification process: a) Expand the node name matching range to include synonyms such as "base" and "platform"; b) Enable the point cloud density analysis method. When the point cloud density of a plane area is >500 points / m 2 When the normal vector verticality is greater than 0.95, it is identified as a floor; 4) After updating the ground object list, recalculate the bounding box RT.z value. If a valid geometry cannot be obtained, add <warning>Label and keep the original model data.

[0074] S3. Simplify the factory area file by retaining only the "elevation of the sub-factory area" and the "geometry that meets the predetermined elevation conditions," and output the factory area with the geometry that meets the predetermined elevation conditions.

[0075] Preferably, after completing the sub-factory division and geometric body screening, the system starts the model simplification process, first traversing the marked ground object set and performing multi-level spatial optimization operations: for any two ground objects e1 and e2, extract the plane projection range of their bounding boxes (LB.x to RT.x, LB.y to RT.y), and use the two-dimensional inclusion relationship judgment algorithm (when the right boundary of e1 ≥ the right boundary of e2 and the upper boundary ≥ the upper boundary of e2, it is determined that e1 contains e2), and the completely contained small floor objects are removed from the set, and the objects with the largest projection range are retained in the temporary list list1. Exemplarily, for two ground objects e1 and e2, the system performs the following calculations:

[0076] Extract the XY plane projection range:

[0077] The projection of e1 is a rectangle R1 = [x1_min, y1_min] → [x1_max, y1_max]

[0078] The projection of e2 is a rectangle R2 = [x2_min, y2_min] → [x2_max, y2_max]

[0079] Necessary and sufficient conditions for determining inclusion relations:

[0080] x1_min≤x2_min∧x1_max≥x2_max∧y1_min≤y2_min∧y1_max≥y2_max

[0081] When this condition is satisfied, e1 completely contains the planar projection of e2.

[0082] Execution logic: Traverse all ground object pairs and remove the contained objects if there is a containment relationship.

[0083] Furthermore, for the objects in list1, the system can perform spatial proximity analysis: calculate the minimum plane distance Δd between objects = √[(x1-x2) 2 +(y1-y2) 2 ], when Δd≤0.5 meters and the elevation difference Δz≤0.01 meters, the CAD Boolean union operation is used to merge the adjacent floors into a single geometry to form an optimized list2 set. Exemplarily, performing CAD Boolean operations on two geometries may include: 1) extracting the triangular mesh data of geometries A and B; 2) constructing a space partitioning BSP tree to accelerate the calculation of intersection surfaces; 3) deleting redundant facets in overlapping areas; 4) stitching boundary edges to generate a fused single geometry. When the Boolean operation fails (such as non-manifold geometry), it can automatically switch to the bounding box fusion mode: take the union of the two bounding boxes to generate a new bounding box; create a new cuboid geometry to replace the original object.

[0084] Furthermore, list3 is generated by arranging the objects in list2 in ascending order of the bounding box top elevation RT.z. This sorting ensures that the lower floor contours are processed first (for example, the floor outside the factory area is processed before the high floor inside the factory). Finally, list3 is traversed to perform three-dimensional boundary line extraction: for each floor geometry, its bottom surface outer contour polyline is extracted (when the bottom surface is a complex curved surface, a two-dimensional polygon is generated using a plane projection method), and the z coordinates of all vertices are forced to be the RT.z value of the floor to form a standardized ContoursNew contour data set. Exemplarily, the bottom contour extraction process may include: 1) extracting the facets whose normal vectors are vertically downward from all triangular facets of the geometry; 2) performing edge classification on the selected facets: free edges (belonging to only one facet) are marked as contour edges; shared edges (belonging to two facets) are marked as internal edges; 3) connecting free edges to form a closed polygon chain. When the bottom surface is curved (such as a curved floor), plane projection + polygon simplification can be used: project the three-dimensional vertices to the XY plane; apply the Douglas-Peucker algorithm (ε = 0.01m) to simplify the broken lines; generate two-dimensional polygons and then back-project them to the original elevation plane.

[0085] Furthermore, the system generates a spatial relationship log in real time during the simplification process, recording the IDs of the removed objects and the reasons for removal (such as "nested floor" or "fragmented block"). When a sub-factory area has no valid ground objects after processing, it automatically inherits the lowest elevation value of step S1 as the benchmark. After this stage of processing, the number of geometric objects in the factory model is reduced by more than three orders of magnitude. Figure 6 This is the simplified model diagram of the example extraction. Figure 7 It is a simplified model contour map of the example, which can lay the data foundation for subsequent contour line fusion.

[0086] S4. Fuse the generated ContoursNew contour line set with the terrain file.

[0087] Preferably, after simplifying the plant model and generating the ContoursNew contour set, the system can activate the terrain fusion processing module. This module first establishes a spatial index structure for the original terrain contours: each contour line in the terrain file (.dwg format) is parsed into a closed polygon and an R-Tree spatial index is constructed. The index range is generated based on the extreme coordinates of the contour bounding box. When traversing each contour line l1 in the ContoursNew set, the spatial index is used to quickly retrieve candidate terrain contour lines that intersect the l1 bounding box.

[0088] Furthermore, for the contour lines in the terrain, its relationship with l1 can be determined, where the relationship can include the following categories:

[0089] 1) If it neither intersects nor contains l1, then skip the original contour line of this terrain;

[0090] 2) If l1 completely contains the original contour line of the terrain, then the original contour line of the terrain is discarded;

[0091] 3) If the original contour line of the terrain completely contains l1, then l1 is added to the original contour line of the terrain;

[0092] 4) If the original contour line of the terrain intersects with l1, forming two intersection points p1 and p2, the original contour line of the terrain is broken into two segments, the internal contour line of l1 is discarded, and then based on l1, a new complete contour line is drawn from p1 along the indented part to p2, thus forming a new contour line.

[0093] Preferably, the above judgment process can be performed only on the original contour lines of the terrain near l1. After completing the judgment of one contour line, the other contour lines in the ContoursNew set can be taken out in turn and the above judgment steps can be repeated.

[0094] Specifically, for each candidate terrain contour line, a multi-level relationship check is performed: first, the projection relationship between l1 and the terrain line in the XY plane is calculated, and the inclusion status is determined using the ray method (when all vertices of the l1 polygon lie within the terrain line, it is considered included; otherwise, it is excluded). If there is a projection intersection, a three-dimensional spatial relationship check is performed, and the corresponding elevation values ​​Z_new and Z_terrain of the two contour lines are extracted. When |Z_new - Z_terrain|>0.01m, duplicate coordinate points are generated according to the "cliff effect" processing principle. For intersections with two intersections, the system uses the Bentley-Ottmann algorithm to calculate the precise intersection coordinates. The original terrain line is split into two sub-segments at points p1 and p2. The sub-segment within the projection area of ​​l1 is deleted, and the corresponding segment of l1 is inserted at the break to form a continuous contour line.

[0095] After each fusion operation, the spatial index of the terrain line is updated, that is, the newly generated contour line data overwrites the original contour line of the terrain to ensure that the subsequent processing can obtain the latest data status. The final output is as follows Figure 8 The plant site + terrain file (.dwg format) shown can be stored in the AutoCAD 2018 drawing database format. Each contour line is treated as a separate POLYLINE entity, with its vertex Z coordinates uniformly assigned to the corresponding elevation value. The entity's extended data records the merge flag ("Original" or "Merged") and source information (plant site number / terrain zone). When processing open contour lines, the system automatically extends the line segment's start and end points to the nearest terrain line to form a closed loop (the maximum allowable distance for this extension algorithm is 5 meters; any distance beyond this is marked as an open contour line).

[0096] Example 2

[0097] This embodiment is a further improvement of embodiment 1, and repeated contents will not be repeated here.

[0098] like Figure 9 As shown, the present invention discloses a system for constructing a surface contour model, which can execute the construction method described in Example 1, wherein the construction system may include one or more of the following hardware:

[0099] The data loading module 100 is used to parse the tree node structure of the factory 3D model file (.rvm format) and extract the geometric data of the ground objects;

[0100] The spatial calculation module 200 is used to perform bounding box containment relationship determination, elevation interval clustering and spatial projection relationship analysis;

[0101] Graphics processing unit 300, used to perform Boolean operations on three-dimensional geometric bodies, generate bottom contour lines, and perform contour line projection transformation;

[0102] The storage array 400 is used to cache the original terrain contour data set, the intermediate processing results and the fused plant terrain model in layers.

[0103] Furthermore, the components that make up the system can work together via a high-speed data bus.

[0104] Preferably, the data loading module 100 is configured with a multi-core processor and a high-speed solid-state storage device, and is equipped with an industrial-grade file parsing chipset, which specializes in processing three-dimensional model data parsing tasks in .rvm format. Its memory channel integrates an ECC error correction mechanism to ensure the stability of the node tree traversal process. The module is connected to the main control unit through a PCIe4.0 interface to realize real-time loading and preprocessing of factory model data.

[0105] Preferably, the spatial computing module 200 is composed of heterogeneous computing units, including an FPGA accelerator card and a multi-threaded CPU cluster. The FPGA has a built-in bounding box relationship judgment logic circuit and adopts a parallel pipeline architecture to synchronously process the spatial inclusion detection of multiple geometric bodies. The CPU cluster runs a hierarchical clustering algorithm to complete the elevation interval grouping. Its three-level cache structure optimizes the access efficiency of large-scale point cloud data.

[0106] Preferably, the graphics processing unit 300 is equipped with a dedicated Boolean operation engine and a ray tracing coprocessor, and its video memory bandwidth meets the high concurrent data throughput requirements when simplifying three-dimensional models. The built-in geometry shader unit performs bottom contour line extraction in real time and accelerates the curvature calculation in the Douglas-Peucker algorithm through the CUDA core.

[0107] Preferably, the storage array 400 adopts a distributed NVMe SSD architecture, dividing the logical storage areas to cache the original terrain data, intermediate processing results and fused contour line sets respectively, wherein the terrain index partition is configured with a non-volatile memory module, and the R-Tree spatial index is quickly constructed and updated through memory computing technology.

[0108] Furthermore, the construction system may also include one or more of the following hardware:

[0109] Human-computer interaction terminal, used to provide a 3D model visualization interface and input response for users to manually select ground objects;

[0110] Network communication unit, used to achieve distributed computing collaboration and remote data synchronization of model data among multiple nodes;

[0111] The power management system is used to dynamically allocate the power supply to each hardware module to maintain continuous and stable operation of the system.

[0112] Preferably, the human-computer interaction terminal is equipped with a high-resolution touch screen and a three-dimensional trackball input device. Its display controller supports the OpenGL 4.6 graphics interface, rendering the visualization effect of the factory model simplification process in real time, and accelerating the dynamic display of .dwg format contour line data through a hardware decoder.

[0113] Preferably, the network communication unit integrates dual 10 Gigabit fiber interfaces and a time-sensitive network protocol stack to achieve task collaboration and status synchronization among multiple nodes, and its hardware encryption engine ensures the security of remote data transmission.

[0114] Preferably, the power management system adopts a redundant power supply design and configures an intelligent power distribution unit to dynamically adjust the voltage for computing-intensive modules to ensure the stability of long-term computing.

[0115] The construction system achieves deep collaboration among modules through customized drivers, uses backplane connection technology to optimize signal integrity at the physical layer, and provides full-process hardware acceleration support for each step of the construction method.

[0116] It should be noted that the above-mentioned specific embodiments are exemplary, and those skilled in the art can come up with various solutions inspired by the disclosure of the present invention, and these solutions also belong to the disclosure scope of the present invention and fall within the protection scope of the present invention. Those skilled in the art should understand that the description of the present invention and its drawings are illustrative and do not constitute a limitation of the claims. The scope of protection of the present invention is defined by the claims and their equivalents. The description of the present invention contains multiple inventive concepts, such as "preferably" or "according to a preferred embodiment", which means that the corresponding paragraph discloses an independent concept, and the applicant reserves the right to file a divisional application based on each inventive concept. Throughout the text, the features guided by "preferably" are only an optional method and should not be understood as having to be set. Therefore, the applicant reserves the right to abandon or delete the relevant preferred features at any time.< / warning>

Claims

1. A method for constructing a surface contour model, characterized in that: It includes the following steps: S1. Determine at least one ground object in the factory area, and divide the factory area into sub-factory areas according to the number of the selected ground objects; S2. According to the records in the factory area files, the geometric bodies belonging to each sub-factory area are screened and the geometric bodies that meet the predetermined elevation conditions are retained; S3. Simplify the factory area file by retaining only the "elevations of the sub-factory areas" and the "geometric bodies that meet the predetermined elevation conditions," and output the factory area with the geometric bodies that meet the predetermined elevation conditions. S4. Fuse the generated ContoursNew contour line set with the terrain file.

2. The construction method according to claim 1, characterized in that Step S1 includes: identifying ground objects by parsing the node attributes or node names of the plant model file, hierarchically clustering the ground objects through the top elevation values ​​of the bounding boxes to determine horizontal plane divisions, each horizontal plane corresponds to a sub-plant area, and its spatial boundary is determined by the two-dimensional projection convex hull of the ground objects to which it belongs.

3. The construction method according to claim 1 or 2, characterized in that The node attribute identification includes: traversing the model tree nodes, detecting nodes containing preset attribute identifiers or name keywords, recursively penetrating the nested node structure until the underlying geometry is obtained, and / or dynamically updating the ground object set based on the user's manual selection results.

4. The construction method according to any one of claims 1 to 3, characterized in that: Step S2 includes: performing volume screening on the geometric bodies of non-ground objects to exclude objects that do not meet the predetermined volume conditions; performing Z-axis projection overlap detection and XY plane inclusion judgment on the remaining geometric bodies, and retaining the geometric bodies that match the spatial boundary and elevation benchmark of the sub-factory area.

5. The construction method according to any one of claims 1 to 4, characterized in that: The volume screening includes: calculating the axial dimensions of the bounding box of the geometric body, marking the geometric body as a candidate for selection when each axial dimension exceeds a preset threshold and the sum of the dimensions meets a predetermined condition, and enabling convex hull volume calculation for the geometric body with a hollow structure.

6. The construction method according to any one of claims 1 to 5, characterized in that: Step S3 includes: performing bounding box projection inclusion relationship judgment on the ground object set to eliminate the ground objects that are completely included; performing Boolean union operation or bounding box fusion on adjacent ground objects with consistent elevation to generate a merged geometric body set.

7. The construction method according to any one of claims 1 to 6, characterized in that: The Boolean union operation includes: extracting geometric triangular mesh data, building a space partitioning tree to accelerate intersection surface calculation, deleting redundant facets and stitching boundary edges, and switching to bounding box union to generate an alternative geometric body when the operation fails.

8. The construction method according to any one of claims 1 to 7, characterized in that: Step S4 includes: constructing a spatial index structure of the original terrain contour lines, traversing each contour line in the new contour line set, retrieving adjacent terrain contour lines and determining the inclusion, intersection or separation relationship, and performing line segmentation and contour line replacement for the intersection situation.

9. The construction method according to any one of claims 1 to 8, characterized in that: The determination of the inclusion relationship includes: using the ray method to analyze the projection inclusion status of the new contour line and the terrain line in the XY plane, combining the elevation difference to trigger the cliff effect processing, and performing the endpoint extension and closing operation on the non-closed contour line.

10. A system for constructing a surface contour model, characterized in that: The construction system is capable of executing the construction method according to any one of claims 1 to 9, and includes one or more of the following hardware: A data loading module (100) is used to parse the tree node structure of the factory area three-dimensional model file and extract the geometric data of the ground objects; A spatial calculation module (200) is used to perform bounding box containment relationship determination, elevation interval clustering and spatial projection relationship analysis; A graphics processing unit (300) for performing three-dimensional geometric body Boolean operations, bottom contour line generation, and contour line projection transformation; The storage array (400) is used for caching the original terrain contour data set, the intermediate processing results and the fused plant terrain model in layers.

Images