Substation equipment alarm electronic fence construction method based on three-dimensional modeling

Abstract

The invention provides a transformer substation equipment alarm electronic fence construction method based on three-dimensional modeling, and relates to the technical field of three-dimensional modeling, which comprises the following steps: extracting three-dimensional point cloud data of transformer substation equipment, generating two-dimensional contour projection points, and constructing a reference fence line by adopting B-spline interpolation; after an electronic fence boundary is generated through external expansion, anchor points are uniformly sampled and positioned, and finally grid representation of the electronic fence is constructed based on the anchor points. Thus, the problem that a fence coverage area has a dead angle due to the fact that a common rectangular electronic fence on the periphery of the substation equipment cannot adapt to a complex equipment contour can be solved to a certain extent, precise modeling of the complex three-dimensional contour of the substation equipment is achieved, monitoring blind areas are avoided, and the protection precision and adaptability of the electronic fence are improved. Therefore, the safety protection capability of the substation is effectively improved.

Description

Technical Field

[0001] This invention relates to the field of 3D modeling technology, and more specifically, to a method for constructing an electronic fence for substation equipment alarms based on 3D modeling. Background Technology

[0002] With the development of power systems, substations, as the core link in power transmission and distribution, undertake important functions such as voltage transformation and energy distribution. Their safety is directly related to the stable operation of the entire power system. In recent years, with the expansion of substation scale and the improvement of intelligence levels, the complexity and density of equipment have increased rapidly, placing higher demands on substation security. Traditional substation security measures mainly include physical fences and electronic fences. Physical fences effectively prevent unauthorized personnel from entering through physical barriers, but their construction costs are high and they are difficult to adapt to complex terrains, while also lacking flexibility and real-time capabilities. In contrast, electronic fences, due to their low cost, high flexibility, and intelligence, have gradually become an important supplementary means of substation security. Research on electronic fence technology has evolved from simple two-dimensional linear detection to high-precision perception and monitoring based on three-dimensional modeling. However, existing electronic fence construction methods still have many shortcomings in adapting to complex equipment environments, rapid deployment, and refined protection.

[0003] Existing electronic fence technologies largely rely on two-dimensional planar modeling, typically generating fences by manually defining equipment boundaries or through simple geometric expansion. However, this approach exhibits significant limitations in the complex three-dimensional environment of substation equipment: First, two-dimensional modeling can only describe the horizontal projection of the equipment, failing to effectively reflect its height information, leading to a mismatch between the fence and the actual equipment shape, potentially creating blind spots in monitoring. Second, traditional boundary expansion methods are usually based on simple fixed-distance expansion, lacking precise descriptions of the equipment's geometric features and struggling to adapt to uneven or locally complex equipment contours. Third, existing electronic fence construction methods often employ regular structures when generating alarm zones, failing to flexibly adjust fence precision according to equipment shape, easily resulting in resource waste or insufficient protection. Especially in scenarios like substations with dense equipment layouts and diverse shapes, existing technologies struggle to achieve rapid generation and dynamic optimization of high-precision three-dimensional fences. These technical issues not only affect the protection efficiency of electronic fence systems but also limit their widespread application in complex power scenarios. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention is proposed. This invention provides a method for constructing an electronic alarm fence for substation equipment based on 3D modeling, which can, to some extent, solve the problem that ordinary rectangular electronic fences around substation equipment cannot adapt to complex equipment contours, resulting in blind spots in the fence coverage area.

[0005] According to one aspect of the present invention, a method for constructing an electronic fence for substation equipment alarms based on three-dimensional modeling is provided, comprising:

[0006] Acquire three-dimensional point cloud data of substation equipment, extract equipment contour feature points based on the three-dimensional point cloud data, and project the feature points onto a horizontal plane to form two-dimensional contour projection points;

[0007] Based on the two-dimensional contour projection points, a reference fence line for the device is constructed using B-spline curve interpolation, and an electronic fence boundary is generated by extending a preset distance outside the reference fence line.

[0008] Multiple positioning anchor points are generated by uniformly sampling along the boundary of the electronic fence, and a triangular mesh is constructed based on the positioning anchor points to form a meshed representation of the electronic fence;

[0009] Based on the height information of the substation equipment, the gridded representation is stretched in the Z-axis direction to form a three-dimensional electronic fence.

[0010] Furthermore, based on the contour feature points extracted from the 3D point cloud data, a corner detection algorithm is employed, specifically including:

[0011] The 3D point cloud data is divided into cubic grids of equal size, and the spatial distribution covariance of points within each cubic grid is calculated.

[0012] Determine the main direction, project the point cloud data along the main direction, and obtain the projection distribution of the point cloud data in each direction;

[0013] Based on the degree of dispersion of the projection distribution, the edge portions of the point cloud data are identified and marked as edge feature regions;

[0014] Within the edge feature region, a curvature variation matrix is ​​constructed based on the normal vector direction of each point;

[0015] The contour feature points are determined based on the curvature variation matrix.

[0016] Furthermore, the edge feature regions are used to establish a density distribution map through point cloud projection, and the density gradient and direction changes are analyzed to identify the edge feature regions.

[0017] Furthermore, the density gradient analysis generates a density distribution by projecting the point cloud, analyzes the density change trend using a local spatial distribution matrix, identifies locations of abrupt density changes, and extracts edge contour lines through spatial connectivity analysis.

[0018] Furthermore, the curvature variation matrix is ​​constructed by building a polar coordinate system in the neighborhood of the point cloud points, analyzing the change angle of the normal vector direction within the sector, and generating the curvature variation matrix.

[0019] Furthermore, the electronic fence boundary is generated by extending a preset distance outside the reference fence line. The sampling points of the reference fence line are obtained by sampling with equal arc length, the normal vector is calculated and extended outward along its direction to form contour points, and finally the extended points are smoothly connected to generate a closed extended contour line.

[0020] Furthermore, constructing a triangular mesh based on the positioning anchor points includes:

[0021] Obtain uniformly distributed boundary sampling points and use all boundary sampling points as initial positioning anchor points;

[0022] Within the electronic fence, a quadtree partitioning method is used to divide the space, and an internal positioning anchor point is added at the center of the divided sub-region.

[0023] Based on all the positioning anchor points, an initial triangular mesh is constructed using a triangulation algorithm;

[0024] The initial triangular mesh is optimized, and the geometric features of each triangle in the triangular mesh are calculated;

[0025] Record the connection relationships between adjacent triangles at the vertices of the triangular mesh to establish the mesh's topology.

[0026] Furthermore, the quadtree partitioning method is based on the minimum safe distance. The space is partitioned recursively by quadtree, the leaf nodes are classified and anchor points are set, and the adjacent nodes are optimized by level balance to finally construct a hierarchical quadtree structure.

[0027] Furthermore, a super triangle containing all the positioning anchor points is constructed as the initial triangle;

[0028] Insert anchor points sequentially according to their x-coordinates;

[0029] After the anchor point is inserted, the circumcircle criterion of the affected triangle is checked. If other anchor points fall within the circumcircle, the relevant triangle is subjected to edge flipping to optimize the local triangle.

[0030] For each anchor point to be inserted, perform the following operations:

[0031] Find an existing triangle containing the anchor point; if the anchor point is inside a triangle, split that triangle into three new triangles; if the anchor point is on one side of a triangle, split the two adjacent triangles into four new triangles.

[0032] Compared with existing technologies, the substation equipment alarm electronic fence construction method based on 3D modeling provided by this invention extracts 3D point cloud data of substation equipment and generates 2D contour projection points. It then uses B-spline curve interpolation to construct a baseline fence line, expands the fence boundary outwards, and uniformly samples and positions anchor points. Finally, it constructs a meshed representation of the electronic fence based on these anchor points. This approach can, to some extent, solve the problem that ordinary rectangular electronic fences around substation equipment cannot adapt to complex equipment contours, resulting in blind spots in the fence coverage area. It achieves accurate modeling of the complex 3D contours of substation equipment, avoids monitoring blind spots, improves the protection accuracy and adaptability of the electronic fence, and thus effectively enhances the safety protection capabilities of the substation. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings:

[0034] Figure 1 This is a flowchart of a method for constructing an electronic fence for substation equipment alarms based on three-dimensional modeling, according to an embodiment of the present invention. Detailed Implementation

[0035] Hereinafter, exemplary embodiments according to the present invention will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of the present invention, and not all embodiments of the present invention; it should be understood that the present invention is not limited to the exemplary embodiments described herein.

[0036] Figure 1 This is a flowchart illustrating a method for constructing an electronic fence for substation equipment alarms based on 3D modeling, according to an embodiment of the present invention. Figure 1 As shown, the method for constructing an electronic fence for substation equipment alarms based on 3D modeling includes:

[0037] S1: Obtain three-dimensional point cloud data of substation equipment, extract equipment contour feature points based on the three-dimensional point cloud data, and project the feature points onto the horizontal plane to form two-dimensional contour projection points;

[0038] Acquiring 3D point cloud data of substation equipment specifically includes: setting up N scanning stations around the substation equipment using a LiDAR scanner, where N is an integer greater than or equal to 4, and the spacing between each scanning station is 8-12 meters; collecting a set of 3D point cloud data at each scanning station, and registering and fusing the N sets of 3D point cloud data to obtain complete 3D point cloud data of the equipment; performing noise reduction and thinning processing on the 3D point cloud data to remove outliers and redundant points to obtain optimized 3D point cloud data; extracting the contour feature points of the equipment using the Harris corner detection algorithm based on the optimized 3D point cloud data, the contour feature points including edge points, corner points, and curved surface contour points of the equipment; projecting the contour feature points onto the horizontal XOY plane using an orthogonal projection method to generate a 2D contour projection point set of the equipment, the 2D contour projection point set being used as the reference boundary for subsequently constructing an electronic fence.

[0039] More specifically, extracting the device's contour feature points using the Harris corner detection algorithm includes:

[0040] First, the 3D point cloud data is divided into equal-sized cubic grids, and the spatial distribution covariance of points within each cubic grid is calculated. A local coordinate system is established within each grid, with the direction of highest point density as the principal direction. The point cloud data is projected along the principal direction to obtain the projection distribution of the point cloud data in each direction. Based on the dispersion of the projection distribution, the edge portions of the point cloud data are identified and marked as edge feature regions. Within the edge feature regions, a curvature variation matrix is ​​constructed based on the normal vector direction of each point, and the curvature variation trend is analyzed. Locations with drastic curvature changes are selected as contour feature points, and these contour feature points are connected to form the contour feature lines of the device. The contour feature lines are spatially smoothed, and a uniformly distributed set of contour feature points is obtained using dense sampling.

[0041] More specifically, identifying the edge portions of the point cloud data based on the dispersion of the projection distribution includes:

[0042] When projecting point cloud data along the main direction, a uniform grid is established on the projection plane, and the number of projected points in each grid is counted to form a point cloud density distribution map. The density gradient between adjacent grids in the point cloud density distribution map is calculated. When a region of abrupt change in density gradient occurs, the region is marked as an edge candidate region. Within the edge candidate region, the directional consistency of the point cloud projection is calculated. If the point cloud projection exhibits a regular directional change in a local area, the region is determined as an edge feature region. Within the edge feature region, the continuous set of points with the most significant density gradient change is selected as the edge contour line, and edge feature points are obtained by sampling at a fixed interval on the edge contour line. When the edge contour line forks or intersects, the main edge direction is selected as the feature edge, and secondary branches are discarded. Spatial connectivity analysis is performed on the obtained edge feature points to eliminate isolated edge feature points and retain the set of continuous edge feature points.

[0043] It should be noted that the most significant change in density gradient is defined as follows:

[0044] When projecting point cloud data along the main direction, a uniform grid is established on the projection plane, and the number of projected points in each grid is counted to form a point cloud density distribution. At each grid location of the point cloud density distribution, a spatial distribution matrix of local density is calculated within a 5×5 grid range centered on that grid. The spatial distribution matrix is ​​scanned in both row and column directions to record the density variation trend from high to low or from low to high. Using an eight-neighborhood scanning method, when the distance from the highest to the lowest density value in a certain direction is less than three grids, that location is recorded as a density abrupt change location. Spatial connectivity analysis is performed on the density abrupt change locations, and adjacent density abrupt change locations are connected to form edge contour lines.

[0045] On the other hand, a curvature variation matrix is ​​constructed based on the normal vector direction of each point. Specifically, this includes: firstly, selecting a spherical region with radius r in the neighborhood of each point as the search range, and taking all points within this spherical region as the reference point set; centralizing the spatial coordinates of the reference point set so that the target point is located at the origin; calculating the principal direction of the reference point set and projecting the point cloud data onto the plane defined by the principal direction; constructing a polar coordinate system with the target point as the center in the plane, and re-representing the reference point set in the polar coordinate system; dividing the reference point set into multiple sectors along the radial direction, with each sector having an equal angle increment; calculating the spatial distribution characteristics of the reference point set relative to the center point within each sector to obtain the normal vector direction of that sector; comparing the normal vector directions of adjacent sectors to calculate the change angle of the normal vector direction; constructing a curvature variation matrix with the sector index as the row and the change angle of the normal vector as the column, where each element in the matrix represents the curvature variation of the corresponding sector; and recording the location as a region of significant curvature variation when a continuous large change value appears in the curvature variation matrix. More specifically, each element in the matrix can be represented by the following formula:

[0046]

[0047] Where k is the number of points within the sector, M ij The elements of the curvature variation matrix are s and n, where s is the total number of sectors. i Let d be the normal vector of the i-th sector. i Let be the average distance from a point in the i-th sector to the center point.

[0048] It is worth noting that although the above description outlines the general process for extracting equipment contour feature points from 3D point cloud data, the specific implementation details may vary depending on the type of equipment and its point cloud characteristics. For example, for regular prismatic equipment, its point cloud projection distribution will exhibit obvious linear characteristics, and edges can be directly identified by the abrupt changes in density gradient. However, for equipment with complex curved surfaces, such as storage tanks or pipeline systems, more detailed analysis of curvature variation characteristics is required. In practical applications, when dealing with large mechanical equipment, it may be necessary to increase the search radius to obtain more reference points, thereby improving the stability of feature extraction; while for precision instruments, it may be necessary to decrease the search radius to ensure the accuracy of feature extraction. For example, when extracting the contour features of a vertical storage tank, analyzing its projection density distribution can clearly identify the intersection edge between the cylindrical surface and the ground; while when extracting the contour features of a bend, it is necessary to focus on the areas of drastic change in the curvature variation matrix, which usually correspond to the key turning points of the bend. Regardless of the specific implementation scheme adopted, the ultimate goal is to accurately extract the set of key points that can characterize the geometric features of the equipment, providing a reliable spatial reference for subsequent electronic fence construction. These feature points should reflect both the overall outline of the equipment and accurately depict the local geometric details.

[0049] S2: Based on the two-dimensional contour projection points, B-spline curve interpolation is used to construct the equipment reference fence line, and the electronic fence boundary is generated by extending a preset distance outside the reference fence line;

[0050] First, the two-dimensional contour projection points are sorted according to their spatial position to ensure reasonable connections between adjacent points. A cubic B-spline curve is used to interpolate and fit the sorted contour points, with node vectors introduced during the fitting process to control the smoothness of the curve. Based on the generated B-spline curve, the normal vector direction on the curve is calculated, and uniform sampling is performed along the normal vector direction to obtain a densely distributed set of reference fence points. The reference fence point set is connected to form a closed reference fence line, which is then smoothed to eliminate any sharp corners or abrupt changes. On the smoothed reference fence line, the line is extended outwards by a preset distance along the normal vector direction to generate an extended contour line. When self-intersection or intersection occurs during the extension process, local adjustments are made to maintain the topological relationship of the contour line. The extended contour line is sampled at equal intervals to obtain a uniformly distributed set of electronic fence boundary points. Curve fitting is performed on the electronic fence boundary point set to generate the final electronic fence boundary curve. Dynamic densification is applied to the electronic fence boundary curve to ensure accurate boundary representation.

[0051] Extend the outline outward by a preset distance along the direction of the normal vector to generate an extended contour line, specifically including:

[0052] First, dense sampling points are obtained along the baseline fence line using equal arc length intervals. For each sampling point, the tangent vector is calculated, and a local coordinate system is constructed based on the tangent vector. In the local coordinate system, the perpendicular direction of the tangent vector is the normal vector direction of the point. When the sampling point is located at a corner of the baseline fence line, the weighted average of the normal vectors of the adjacent segments is used as the normal vector direction of the point. Along the calculated normal vector direction, each sampling point is translated outward by a preset distance to obtain the expanded contour points. If the spacing between adjacent expanded contour points is found to be too large during the expansion process, the sampling point density in that area is increased to ensure the continuity of the expanded contour. When the expanded contour points intersect, the normal vector direction of the local area is adjusted to keep the expanded contour monotonic. In concave areas, the expansion distance is dynamically adjusted according to the degree of concavity to avoid self-intersection. All expanded contour points are smoothed to eliminate local sharp corners caused by abrupt changes in the normal vector direction. The smoothed expanded contour points are connected in spatial order to form a closed expanded contour line.

[0053] The method for calculating the weighted average normal vector at a corner includes: first, obtaining two adjacent curve segments before and after the corner point; selecting multiple sampling points within a fixed arc length range from the corner point on each curve segment; calculating the normal vector of the sampling points on each curve segment, normalizing the normal vectors on the same curve segment, and calculating the average value to obtain the representative normal vector of that curve segment; calculating the weight coefficient based on the angle between the two curve segments. When the angle is small, the weight coefficients of the two curve segments are approximately equal; when the angle increases, the weight coefficient of the side with the larger angle is decreased, and the weight coefficient of the side with the smaller angle is increased; the weight coefficient is calculated using a function transformation of the cosine value of the angle to ensure that the change of the weight coefficient is smooth and monotonic; the representative normal vectors of the two curve segments are weighted and summed according to the calculated weight coefficients to obtain the initial weighted average normal vector; the initial weighted average normal vector is normalized to obtain the final normal vector direction at the corner; if the curvature change at the corner exceeds a predetermined range, the sampling point density is increased, and more local sampling points are used to participate in the normal vector calculation.

[0054] S3: Generate multiple positioning anchor points by uniformly sampling along the boundary of the electronic fence, and construct a triangular mesh based on the positioning anchor points to form a meshed representation of the electronic fence;

[0055] First, sampling is performed along the boundary curve of the electronic fence at fixed arc length intervals to obtain uniformly distributed boundary sampling points. Based on the size and shape characteristics of the electronic fence, the sampling interval is adaptively adjusted, and all boundary sampling points are used as initial positioning anchor points, each assigned a unique identifier. Within the internal area of ​​the electronic fence, a quadtree partitioning method is used to divide the space, and internal positioning anchor points are added at the center of each sub-region. When the distance between adjacent anchor points exceeds a preset threshold, new anchor points are added using interpolation to ensure uniform anchor point distribution. Based on all positioning anchor points, an initial triangular mesh is constructed using the Delaunay triangulation algorithm. The initial triangular mesh is then optimized by deleting... For triangles that do not meet the minimum angle constraint, the shape quality of the triangles is improved by adjusting the anchor point positions or inserting new anchor points; at the grid boundary, ensure that the sides of the triangles fit the boundary curve of the electronic fence, and perform local re-subdivision of the triangles at the boundary if necessary; for each triangle in the triangular mesh, calculate its area, perimeter, interior angles and other geometric features, and store this feature information; record the connection relationship of adjacent triangles at the vertices of the triangular mesh to establish the topology of the mesh; when the electronic fence boundary changes, only the triangular mesh of the affected area needs to be updated, while the mesh structure of other areas remains unchanged; the completed triangular mesh is encoded and stored to facilitate subsequent fast retrieval and update operations.

[0056] The space is partitioned using a quadtree partitioning method, specifically including:

[0057] First, based on equipment safety requirements, a minimum safety distance is determined as the baseline extension distance for the electronic fence. The circumscribed rectangle of the equipment outline point cloud data is obtained, and this rectangle is expanded by 1.2 times as the initial division area for the root node. The initial area is divided into four equal sub-regions, and each sub-region is recursively divided. During the recursive division process: when a sub-region contains equipment outline points, it is further divided into four equal parts; when a sub-region is completely outside the equipment outline and exceeds the safety distance from the outline, it is marked as an external region and the division stops; when the size of a sub-region is smaller than the preset minimum allowable size, it is marked as a leaf node and the division stops.

[0058] All leaf node regions are classified and processed as follows:

[0059] When a leaf node intersects with the device outline, a boundary anchor point is added at the intersection, and transition anchor points are added along the outline in fixed steps.

[0060] When a leaf node is completely within the device outline, add an internal anchor point at the center of the area.

[0061] When a leaf node is located in a concave area, add dense anchor points along the concave contour, with the anchor point spacing not exceeding 1 / 4 of the minimum safe distance.

[0062] When the leaf node is located around the protruding component, the annular anchor points are evenly distributed around the protruding point according to the polar coordinates.

[0063] After spatial partitioning, a hierarchical balancing process is performed on adjacent leaf nodes: the hierarchical difference between all adjacent leaf nodes is checked, and if the difference is greater than 1, the node with the lower level is forcibly split; anchor points are added to the new leaf nodes after the split to ensure the continuity of anchor point distribution; this process is repeated until the hierarchical difference between all adjacent nodes does not exceed 1. Finally, a quadtree hierarchical structure is constructed. This method, through adaptive spatial partitioning and targeted anchor point placement strategies, achieves precise coverage of complex equipment outlines and effectively avoids monitoring blind spots. The hierarchical and local characteristics of the quadtree structure also facilitate the dynamic adjustment of the electronic fence.

[0064] The Delaunay triangulation algorithm is used to construct the initial triangular mesh, specifically including: first, constructing a super triangle containing all positioning anchor points as the initial triangle; sorting the positioning anchor points according to their x-coordinates and inserting each anchor point sequentially; for the anchor point to be inserted, searching for an existing triangle containing that point; if the point is inside a triangle, splitting that triangle into three new triangles; if the point is on the edge of a triangle, splitting the two adjacent triangles into four new triangles; after inserting a new anchor point, checking the circumcircle criterion of the affected triangles, if other points fall within the circumcircle of the triangle, performing an edge flipping operation to optimize the local triangles; in areas with dense equipment or complex equipment contours, increasing the sampling point density improves mesh accuracy; for equipment recessed areas, additional anchor points are inserted at the recesses to ensure the mesh accurately describes the equipment contour; when the area ratio of adjacent triangles is too large, inserting new anchor points reduces the abrupt change in mesh size.

[0065] The specific steps for inserting additional anchor points in the recess are as follows:

[0066] First, along the equipment outline, take three adjacent vertices P1, P2, and P3, and calculate vectors V1 = P2 - P1 and V2 = P3 - P2. Calculate the angle θ between these two vectors. When the angle is less than a preset threshold α (usually 120 degrees), P2 is considered a concave point. Draw a circle with P2 as the center and radius R equal to 1 / 3 of the smaller of the lengths of vectors V1 and V2. Within this circle, first, along the angle bisector, place three anchor points A1, A2, and A3 inwards from P2 with a step size of R / 4. Then, using the angle bisector as the axis of symmetry, place two sets of anchor points on both sides at 15-degree intervals. Specifically, arrange four directions on each side of the angle bisector. Two anchor points are placed in each direction with a step size of R / 4; this adds a total of 19 anchor points at each recessed point. When the distance between a newly added anchor point and an existing anchor point is less than the minimum allowable distance (R / 8), the original anchor point is retained and the newly added anchor point is deleted. For each newly added anchor point, it is verified whether it falls within the equipment area; if it is outside the equipment area, it is deleted. If the distance between two adjacent recessed points is less than 2R, the midpoint of these two recessed points is taken, and a set of anchor points is added with the midpoint as the center and R as the radius, arranged in the same way as above. Finally, the triangles formed by all the newly added anchor points and the original contour points are checked, and the anchor points corresponding to those triangles with a circumscribed circle radius greater than 2R are deleted.

[0067] Establishing the mesh topology involves: first, assigning a unique identifier to each triangular cell and recording the numbers of the three vertices constituting the triangle; numbering the three edges of each triangle and recording the starting and ending vertex information of each edge; establishing a vertex-triangle adjacency list to record all triangular cells connected to each vertex; establishing a triangle-triangle adjacency list to record the numbers of the three triangles adjacent to each triangular cell, and marking the corresponding position as a boundary edge if an edge has no adjacent triangles; for each vertex, storing the numbers of its adjacent vertices in counter-clockwise order to form vertex-vertex adjacency relationships; and specifically marking the vertex when it is located on a boundary. The system implements boundary attributes and records the preceding and following boundary points. It establishes an edge-triangle adjacency list, recording one or two triangular cells to which each edge belongs. If an edge belongs to only one triangle, it is marked as a boundary edge, and its position within the boundary is recorded. A local coordinate system is established for each triangle, recording the relative positions of its three vertices. Geometric features of each triangle are stored, including area, perimeter, interior angles, circumcircle, and incircle. When a local mesh update occurs, only the topological relationships of the affected region need to be modified, while the topological structure of other regions remains unchanged. A fast retrieval mechanism is established to enable traversal of adjacent cells starting from any vertex.

[0068] S4: Based on the height information of the substation equipment, the gridded representation is stretched in the Z-axis direction to form a three-dimensional electronic fence, wherein the size of the grid cell increases with the height.

[0069] First, the three-dimensional spatial information of each piece of equipment in the substation is acquired, including the bottom and top elevations. Multiple height layers are then divided along the Z-axis at preset intervals, with smaller intervals for lower layers and progressively larger intervals as height increases. For each height layer, a corresponding horizontal profile mesh is created based on a two-dimensional mesh. When a height layer intersects with the top or bottom of an equipment, the mesh is densified at that layer to ensure accurate representation of the equipment boundary. Adjacent height layers are connected using triangular prism elements to form a complete three-dimensional mesh system. As height increases, the horizontal mesh size gradually increases according to a preset ratio, improving efficiency while maintaining computational accuracy. In sensitive areas around the equipment, regardless of height, a denser mesh is maintained to ensure accurate representation of the equipment boundary. This system enables precise monitoring of the area; establishes vertical connections between different grid layers, recording the correspondence between adjacent grid cells; when the size of a grid layer changes, a smooth connection with adjacent layers is achieved through transition cells; for inclined or irregular equipment surfaces, deformed triangular prism cells are used for transition to ensure grid continuity; each grid cell records its spatial region type, such as equipment area, passageway area, or work area; based on the safety level of different areas, grid cells are assigned corresponding attribute tags; when the equipment configuration changes, only the 3D grid of the affected area needs to be updated, while the grid structure of other areas remains unchanged; the geometric information and topological relationships of each cell are stored in the grid data structure, supporting fast spatial query and update operations.

[0070] In summary, the method for constructing an alarm electronic fence for substation equipment based on 3D modeling, as described in this invention, is explained. It extracts 3D point cloud data of the substation equipment and generates 2D contour projection points. A baseline fence line is constructed using B-spline curve interpolation. After expanding the fence boundary, anchor points are uniformly sampled and positioned. Finally, a meshed representation of the electronic fence is constructed based on these anchor points. This method can, to some extent, solve the problem that ordinary rectangular electronic fences around substation equipment cannot adapt to complex equipment contours, resulting in blind spots in the fence coverage area. It achieves accurate modeling of the complex 3D contours of substation equipment, avoids monitoring blind spots, improves the protection accuracy and adaptability of the electronic fence, and thus effectively enhances the safety protection capabilities of the substation.

[0071] Here, those skilled in the art will understand that the specific operations of each step in the above-described method for constructing an electronic alarm fence for substation equipment based on 3D modeling have been referenced above. Figure 1 The method for constructing an electronic fence for substation equipment alarms based on 3D modeling has been described in detail, and therefore, its repeated description will be omitted.

Claims

1. A method for constructing an electronic fence for substation equipment alarms based on 3D modeling, characterized in that, include: Acquire three-dimensional point cloud data of substation equipment, extract equipment contour feature points based on the three-dimensional point cloud data, and project the feature points onto a horizontal plane to form two-dimensional contour projection points; Based on the two-dimensional contour projection points, a reference fence line for the device is constructed using B-spline curve interpolation, and an electronic fence boundary is generated by extending a preset distance outside the reference fence line. Multiple positioning anchor points are generated by uniformly sampling along the boundary of the electronic fence, and a triangular mesh is constructed based on the positioning anchor points to form a meshed representation of the electronic fence; Based on the height information of the substation equipment, the gridded representation is stretched in the Z-axis direction to form a three-dimensional electronic fence.

2. The method for constructing an electronic fence for substation equipment alarms based on three-dimensional modeling according to claim 1, characterized in that, The corner detection algorithm is used to extract the contour feature points of the 3D point cloud data extraction device, specifically including: The 3D point cloud data is divided into cubic grids of equal size, and the spatial distribution covariance of points within each cubic grid is calculated. Determine the main direction, project the point cloud data along the main direction, and obtain the projection distribution of the point cloud data in each direction; Based on the degree of dispersion of the projection distribution, the edge portions of the point cloud data are identified and marked as edge feature regions; Within the edge feature region, a curvature variation matrix is ​​constructed based on the normal vector direction of each point; The contour feature points are determined based on the curvature variation matrix.

3. The method for constructing an electronic fence for substation equipment alarms based on three-dimensional modeling according to claim 2, characterized in that, The edge feature regions are identified by establishing a density distribution map through point cloud projection and analyzing density gradient and direction changes.

4. The method for constructing an electronic fence for substation equipment alarms based on three-dimensional modeling according to claim 3, characterized in that, The density gradient analysis is performed by generating a density distribution through point cloud projection, analyzing the density change trend using a local spatial distribution matrix, identifying locations of abrupt density changes, and extracting edge contour lines through spatial connectivity analysis.

5. The method for constructing an electronic fence for substation equipment alarms based on three-dimensional modeling according to claim 2, characterized in that, The curvature variation matrix is ​​constructed by building a polar coordinate system in the neighborhood of the point cloud points, analyzing the change angle of the normal vector direction within the sector, and generating the curvature variation matrix.

6. The method for constructing an electronic fence for substation equipment alarms based on three-dimensional modeling according to claim 1, characterized in that, The electronic fence boundary is generated by extending a preset distance outside the baseline fence line. The sampling points of the baseline fence line are obtained by sampling with equal arc length. The normal vector is calculated and extended outward along its direction to form contour points. Finally, the extended points are smoothly connected to generate a closed extended contour line.

7. The method for constructing an electronic fence for substation equipment alarms based on three-dimensional modeling according to claim 1, characterized in that, Constructing a triangular mesh based on the positioning anchor points includes: Obtain uniformly distributed boundary sampling points and use all boundary sampling points as initial positioning anchor points; Within the electronic fence, a quadtree partitioning method is used to divide the space, and an internal positioning anchor point is added at the center of the divided sub-region. Based on all the positioning anchor points, an initial triangular mesh is constructed using a triangulation algorithm; The initial triangular mesh is optimized, and the geometric features of each triangle in the triangular mesh are calculated; Record the connection relationships between adjacent triangles at the vertices of the triangular mesh to establish the mesh's topology.

8. The method for constructing an electronic fence for substation equipment alarms based on three-dimensional modeling according to claim 7, characterized in that, The method of partitioning space using a quadtree is based on the minimum safe distance. The space is partitioned recursively using a quadtree, and the leaf nodes are classified and anchored. The adjacent nodes are optimized by combining hierarchical balance, and finally a hierarchical quadtree structure is constructed.

9. The method for constructing an electronic fence for substation equipment alarms based on three-dimensional modeling according to claim 7, characterized in that, The triangulation algorithm for constructing the initial triangular mesh includes: Construct a super triangle containing all the positioning anchor points as the initial triangle; Insert anchor points sequentially according to their x-coordinates; After the anchor point is inserted, the circumcircle criterion of the affected triangle is checked. If other anchor points fall within the circumcircle, the relevant triangle is subjected to edge flipping to optimize the local triangle. For each anchor point to be inserted, perform the following operations: Find an existing triangle containing the anchor point; if the anchor point is inside a triangle, split that triangle into three new triangles; if the anchor point is on one side of a triangle, split the two adjacent triangles into four new triangles.

10. The method for constructing an electronic fence for substation equipment alarms based on three-dimensional modeling according to claim 9, characterized in that, The triangulation algorithm also includes inserting additional anchor points at the recesses.

Images