Digital topographic data generation method for digital twinning of overhead transmission line

Abstract

The invention discloses a digital topographic data generation method for digital twinning of an overhead transmission line, and belongs to the field of geographic information. Low-precision SRTM data obtained by a public channel is used as basic data, and high-precision stripe point cloud data of the overhead transmission line are combined to generate digital twinning data of the overhead transmission line; the adjustment fitting of the terrain is realized by performing ground point cloud segmentation extraction, tower point cloud classification, tower foot point cloud extraction, tower foot point cloud range filtering refinement, ground point cloud digital terrain extraction, elevation model conversion, SRTM data fusion adjustment and the like on point cloud data. According to the invention, with the extracted tower foot point cloud data as a reference, local elevation fitting and point cloud refinement processing are carried out on the target ground tower foot range, so that the use cost of the digital terrain is greatly reduced, and the precision of the local terrain data is also improved. On the premise that the precision of the ground within the two-kilometer range of the overhead transmission line channel is guaranteed, it can be guaranteed that the final three-dimensional terrain achieves the attractive and smooth effect.

Description

Technical Field

[0001] The present invention relates to the field of geographic information, and in particular to a method for generating digital terrain data of a digital twin of an overhead transmission line. Background Art

[0002] "Digital twin" is a simulation process that fully utilizes data such as physical models, sensor updates, and operation history to integrate multiple disciplines, multiple physical quantities, multiple scales, and multiple probabilities. In recent years, with the advancement of digital transformation, the development momentum of "digital twin" has been strong, and it has achieved rapid development in both theoretical research and practical application.

[0003] Through digital transformation, power grid companies can build a "digital twin power grid" covering the entire process and production of the power grid, and improve the ability to control complex power grids. In the field of power transmission, due to the large geographical span and complex terrain of the power grid, in order to realize the digital twin of the power grid, digital terrain data, as one of the basic data of the digital twin of the power grid, becomes particularly important.

[0004] At present, the digital terrain data that can be obtained through public channels is generally SRTM data, with a spatial resolution of 90 meters, 30 meters or 12.5 meters, and the accuracy is relatively low. After being fused with the high-precision power grid digital twin data, the accuracy difference between the two is very large, resulting in the inability to accurately fit the combined parts, the effect after fusion is very poor, or even unable to be used normally. Summary of the invention

[0005] The present invention provides a method for generating digital terrain data of a digital twin of an overhead power transmission line, which reduces the use cost of digital terrain data and improves the accuracy of digital terrain data.

[0006] The present invention provides the following technical solutions:

[0007] A method for generating digital terrain data of a digital twin of an overhead transmission line, the method comprising:

[0008] S1: Acquire digital terrain data and strip laser point cloud data of transmission lines;

[0009] S2: performing ground segmentation on the strip laser point cloud data to extract strip ground point cloud data;

[0010] S3: performing pole tower classification on the strip laser point cloud data, extracting pole tower point clouds, and extracting pole tower foot point clouds based on the pole tower point clouds;

[0011] S4: filtering, adjusting and refining the strip ground point cloud data using the extracted tower foot point cloud to obtain strip ground point cloud refined data;

[0012] S5: converting the strip ground point cloud refined data and the elevation model of the digital terrain data into a geodetic ellipsoid elevation model respectively;

[0013] S6: extracting the boundary of the strip ground point cloud refined data, and fusing the strip ground point cloud refined data and the digital terrain data through a boundary weighted loss function to obtain final digital terrain data.

[0014] Further, the S2 includes:

[0015] The strip laser point cloud data is segmented on the ground using a cloth filtering algorithm to extract strip ground point cloud data.

[0016] Further, the S3 includes:

[0017] S31: extracting the tower point cloud from the strip laser point cloud data by using the classification label of the tower;

[0018] S32: filtering out the tower body point cloud using the height percentage for the tower point cloud, and retaining the tower leg point cloud;

[0019] S33: clustering the tower leg point cloud to divide it into tower leg groups, calculating the lowest point for each group of tower legs, and obtaining the tower foot point cloud.

[0020] Further, the S32 includes:

[0021] The height of the tower is calculated, and 80% of the height of the tower is used as a segmentation point to segment the tower into a tower body and a tower foot, and the point cloud data of the tower foot portion is retained as the tower foot point cloud.

[0022] Further, the S4 includes:

[0023] S41: grouping the tower foot point cloud using the Euclidean clustering method, and setting an adjustment radius for each group of tower foot point cloud as a search range;

[0024] S42: for each group of tower foot point clouds, a neighbor search is performed on the strip ground point cloud data within the adjustment radius to obtain a target ground point cloud set;

[0025] S43: Calculating the Euclidean distance between each set point in the target ground point cloud set and the tower foot point cloud;

[0026] d p1 =distance(p1,q)

[0027] Among them, p1 is the collection point of the target ground point cloud set, q is the tower foot point cloud, d p1is the Euclidean distance between the set point p1 and the tower foot point cloud q, and distance() represents the calculation function of the Euclidean distance;

[0028] S44: Calculate the new elevation D of each set point in the target ground point cloud set h1 , and reset the elevation of the rendezvous point to D h1 ;

[0029] D h1 =p h1 ×(1-d p1 / d max )+q h ×(d p1 / d max )

[0030] Among them, p h1 is the original elevation of the collection point p1, D h1 is the new elevation of the collection point p1, q h is the base elevation of the tower foot point cloud q, d max is the maximum search area radius;

[0031] S45: Calculate the difference between the new elevation of each set point in the target ground point cloud set and the reference elevation of the tower foot point cloud. When the difference is greater than a set first threshold or less than a set second threshold, remove the corresponding set point.

[0032] Furthermore, the S4 further includes:

[0033] S46: Counting the number of point clouds of the tower foot point cloud within the search range, and comparing it with a set number threshold. If the number of searched point clouds is less than the number threshold, it is determined that the point cloud is empty;

[0034] S47: Select a reference plane at the hole in the point cloud, mesh the reference plane according to the set resolution, sample the coordinates of the grid center point, calculate the elevation of the point cloud interpolation point using the distance weighted method, and generate a new point cloud interpolation point based on the calculated elevation and the grid center point to obtain strip ground point cloud refined data.

[0035] D h =P h ×(1-distance(P,q edge ) / d max )+q he ×(distance(P,q edge ) / d max )

[0036] Among them, P is the center point of the grid, q edge is the closest point to point P in the target ground point cloud set, Ph is the elevation of point P, q he Q edge The elevation of the point, D h Interpolate the elevation of the points for the computed point cloud.

[0037] Further, the S6 includes:

[0038] S61: Sampling the digital terrain data into digital terrain point cloud data;

[0039] S62: extracting boundary contour point cloud from the strip ground point cloud refinement data, and obtaining contour polygon vector data using the extracted boundary contour point cloud;

[0040] S63: using the outline polygon vector data, removing the point cloud located inside the outline polygon vector data in the digital terrain point cloud data;

[0041] S64: merging the eliminated digital terrain point cloud data and the strip ground point cloud refined data to obtain merged point cloud data and smoothing the boundary points of the merged point cloud data;

[0042] S65: resampling the smoothed merged point cloud data into digital elevation model data according to the resolution to obtain final digital terrain data.

[0043] Further, the S64 includes:

[0044] S641: traverse the boundary point clouds of the strip ground point cloud refined data, and calculate the normal vector of each boundary point cloud;

[0045] S642: Taking each boundary point cloud as the center and the normal vector of the boundary point cloud as the direction constraint, searching for neighboring points of the digital terrain point cloud data after elimination;

[0046] S643: For the searched neighboring points, use the boundary weighted loss function to recalculate the elevation D of the neighboring points h2 , and reset the elevation of the neighboring point to D h2 ;

[0047] D h2 =P h2 ×(1-d p2 / d max )+E h ×(d p2 / d max )

[0048] Among them, p h2 is the original elevation of the neighboring point p2, E h is the base elevation of the boundary point cloud E, d p2is the Euclidean distance between the neighboring point p2 and the boundary point cloud E, D h2 Recalculate the elevation for neighboring point p2.

[0049] Furthermore, the digital terrain data is SRTM data, and the strip laser point cloud data is obtained by scanning with a laser radar using an unmanned aerial vehicle.

[0050] The present invention has the following beneficial effects:

[0051] The present invention uses low-precision SRTM data obtained from public channels as basic data, combined with high-precision strip point cloud data of overhead transmission lines, and realizes terrain adjustment fitting by performing ground point cloud segmentation and extraction, tower point cloud classification, tower foot point cloud extraction, tower foot point cloud range filtering and refinement, ground point cloud digital terrain extraction, elevation model conversion, and SRTM data fusion adjustment on the point cloud data. The present invention uses the extracted tower foot point cloud data as a benchmark to perform local elevation fitting and point cloud refinement processing on the target ground tower foot range, which greatly reduces the cost of using digital terrain and improves the accuracy of local terrain data. The present invention generates local high-precision terrain data with low-cost free terrain data, which is helpful for the promotion and implementation of digital twins of transmission lines with large geographical spans. The present invention can ensure that the final three-dimensional terrain achieves a beautiful and smooth effect under the premise of ensuring the accuracy of the ground within two kilometers of the overhead transmission line channel. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] Figure 1 This is a flow chart of the method for generating digital terrain data for the digital twin of an overhead transmission line of the present invention. DETAILED DESCRIPTION

[0053] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0054] The present invention provides a method for generating digital terrain data of a digital twin of an overhead transmission line, such as Figure 1 As shown, the method includes:

[0055] S1: Acquire digital terrain data and strip laser point cloud data of transmission lines.

[0056] In this step, the digital terrain data obtained is SRTM (Shuttle Radar Topography Mission) data, and the strip laser point cloud data is obtained by using a drone to scan with a laser radar. In recent years, drones have been widely used by power grid companies in the construction and inspection of transmission lines, and laser point cloud data of transmission lines have been gradually accumulated. The laser point cloud data generated by drones using laser radar scanning combined with RTK precise positioning has high accuracy.

[0057] By loading geographic information data and digital terrain data into the rendering engine, three-dimensional geographic information is constructed as the background, and then the transmission line laser point cloud data generated by the drone or the high-precision digital twin built based on the laser point cloud data is loaded and integrated to generate a digital twin scene of the transmission line.

[0058] When constructing a digital twin scene, low-precision free digital terrain data is used to render and generate three-dimensional geographic information as the background, while the digital twin of the transmission line uses high-precision laser point cloud data generated by drones or a high-precision twin model built based on laser point cloud data. Due to the large difference in accuracy between the two, the digital twin floats in the air or is inserted below the ground when rendered in the scene, which is more obvious in hilly and mountainous areas with large terrain undulations.

[0059] S2: Perform ground segmentation on the strip laser point cloud data to extract the strip ground point cloud data.

[0060] For example, the ground segmentation may use a cloth simulation filter (CSF).

[0061] S3: Classify the strip laser point cloud data into towers, extract the tower point cloud, and extract the tower foot point cloud based on the tower point cloud.

[0062] In this step, the strip ground point cloud data is classified into towers and the tower point cloud is extracted. The tower foot point cloud is extracted by filtering the lowest point of the tower. A specific implementation example is as follows:

[0063] S31: extracting tower point clouds from strip laser point cloud data through classification labels of towers.

[0064] S32: Use the height percentage to filter out the tower point cloud and keep the tower leg point cloud.

[0065] For example, calculate the height of the tower, use 80% of the tower height as the segmentation point to segment the tower point cloud, segment the tower into the tower body and tower legs, and retain the point cloud data of the tower leg part as the tower leg point cloud.

[0066] S33: clustering the tower leg point cloud and dividing it into tower leg groups, calculating the lowest point for each group of tower legs, and obtaining the tower foot point cloud.

[0067] S4: Use the extracted tower foot point cloud to filter, adjust and refine the strip ground point cloud data by weighting adjacent points and removing outliers to obtain the strip ground point cloud refined data to ensure the accuracy of the tower foot and foundation foot position terrain.

[0068] As an example, a specific implementation of S4 includes:

[0069] S41: The tower foot point cloud is grouped using the Euclidean clustering method, and an adjustment radius is set for each group of tower foot point clouds as the search range.

[0070] S42: Perform KDTree indexing on the strip ground point cloud data, and for each group of tower foot point clouds, perform a neighbor search on the strip ground point cloud data within the adjustment radius to obtain a target ground point cloud set.

[0071] S43: Calculate the Euclidean distance between each set point in the searched target ground point cloud set and the tower foot point cloud.

[0072] d p1 =distance(p1,q)

[0073] Among them, p1 is the collection point of the target ground point cloud set, q is the tower foot point cloud, d p1 It is the Euclidean distance between the set point p1 and the tower foot point cloud q, and distance() represents the calculation function of the Euclidean distance.

[0074] In this step, the elevation of the tower foot point cloud is used as the reference elevation, and the elevation value of the target ground point cloud is smoothed according to the distance weighted method.

[0075] S44: Calculate the new elevation D of each point in the target ground point cloud set h1 , and reset the elevation of the rendezvous point to D h1 .

[0076] D h1 =p h1 ×(1-d p1 / d max )+q h ×(d p1 / d max )

[0077] Among them, p h1 is the original elevation of the collection point p1, D h1 is the new elevation of the collection point p1, q h is the base elevation of the tower foot point cloud q, d maxis the maximum search area radius.

[0078] S45: Calculate the difference between the new elevation of each set point in the target ground point cloud set and the reference elevation of the tower foot point cloud. When the difference is greater than a set first threshold or less than a set second threshold, remove the corresponding set point.

[0079] S46: Count the target ground point cloud within the adjustment range, traverse the tower foot point cloud, count the number of point clouds of the tower foot point cloud within the search range, and compare it with the set number threshold. If the number of searched point clouds is less than the number threshold, it is judged to be a point cloud hole.

[0080] Whether it is a point cloud hole can be determined according to the function Q(p1,q), where q is the tower foot point cloud, p1 is the collection point of the target ground point cloud set, and d max is the maximum search range, ∑(distance(p1,q)<d max ) is the Euclidean distance less than d max The sum of the number of points, p min is the quantity threshold.

[0081]

[0082] Q(p1,q)=1 indicates a point cloud hole.

[0083] S47: Select a reference plane at the hole in the point cloud, mesh the reference plane according to the set resolution, sample the coordinates of the grid center point, use the distance weighted method to calculate the elevation of the point cloud interpolation point, and generate a new point cloud interpolation point based on the calculated elevation and the grid center point to obtain the strip ground point cloud refined data.

[0084] In this step, the coordinates of the center point of the grid are used as the initial value, and its elevation is reassigned to the elevation of the newly calculated interpolation point, thereby generating point cloud interpolation point data and realizing the refinement of the strip ground point cloud. The reference plane can be any plane near the point cloud hole. In the actual calculation, the plane where the lowest point elevation of the tower foot point cloud is located is selected. The calculation formula for the interpolation point elevation is as follows:

[0085] D h =P h ×(1-distance(P,q edge ) / d max )+q he ×(distance(P,q edge ) / d max )

[0086] Where P is the center point of the grid, q edge is the closest point to point P in the target ground point cloud set, P h is the elevation of point P, qhe Q edge The elevation of the point, D h Interpolate the elevation of the points for the computed point cloud.

[0087] S5: Convert the strip ground point cloud refined data and the elevation model of the digital terrain data into the geodetic ellipsoid elevation model respectively.

[0088] S6: extract the boundary of the strip ground point cloud refined data, fuse the strip ground point cloud refined data and the digital terrain data through the boundary weighted loss function, and obtain the final digital terrain data.

[0089] The specific implementation method is:

[0090] S61: Sampling the ground elevation grid data of the digital terrain data into digital terrain point cloud data.

[0091] S62: extracting boundary contour point cloud from the strip ground point cloud refinement data, and obtaining contour polygon vector data polygon by simplifying the extracted boundary contour point cloud.

[0092] S63: using the outline polygon vector data polygon, filtering and eliminating the point cloud located inside the outline polygon vector data in the digital terrain point cloud data through a geometric method.

[0093] S64: merging the eliminated digital terrain point cloud data and the strip ground point cloud refined data to obtain merged point cloud data and smoothing the boundary points of the merged point cloud data.

[0094] Then, the smoothed digital terrain point cloud data and the strip ground point cloud refined data are merged into a complete point cloud.

[0095] As a specific implementation example, S64 includes:

[0096] S641: traverse the boundary point clouds of the strip ground point cloud refinement data, calculate the normal vector of each boundary point cloud, and perform KDTree indexing on the digital terrain point cloud data after elimination.

[0097] S642: traverse all boundary point clouds of the strip ground point cloud refined data, take each boundary point cloud as the center, use the normal vector of the boundary point cloud as the direction constraint, and search for neighboring points within a set threshold range of the digital terrain point cloud data after elimination.

[0098] S643: For the neighboring points within the searched threshold range, the elevation D of the neighboring points is recalculated using the boundary weighted loss function. h2 , and reset the elevation of the neighboring point to D h2 .

[0099] Dh2 =p h2 ×(1-d p2 / d max )+E h ×(d p2 / d max )

[0100] Among them, P h2 is the original elevation of the neighboring point p2, E h is the base elevation of the boundary point cloud E, d p2 is the Euclidean distance between the neighboring point p2 and the boundary point cloud E, D h2 Recalculate the elevation for neighboring point p2.

[0101] S65: resampling the smoothed merged point cloud data into digital elevation model (DEM) data according to the resolution to obtain final digital terrain data.

[0102] The present invention uses low-precision SRTM data obtained from public channels as basic data, combined with high-precision strip point cloud data of overhead transmission lines, and realizes terrain adjustment fitting by performing ground point cloud segmentation and extraction, tower point cloud classification, tower foot point cloud extraction, tower foot point cloud range filtering and refinement, ground point cloud digital terrain extraction, elevation model conversion, and SRTM data fusion adjustment on the point cloud data. The present invention uses the extracted tower foot point cloud data as a benchmark to perform local elevation fitting and point cloud refinement processing on the target ground tower foot range, which greatly reduces the cost of using digital terrain and improves the accuracy of local terrain data. The present invention generates local high-precision terrain data with low-cost free terrain data, which is helpful for the promotion and implementation of digital twins of transmission lines with large geographical spans. The present invention can ensure that the final three-dimensional terrain achieves a beautiful and smooth effect under the premise of ensuring the accuracy of the ground within two kilometers of the overhead transmission line channel.

[0103] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for generating digital terrain data of a digital twin of an overhead transmission line, characterized in that: The method comprises: S1: Acquire digital terrain data and strip laser point cloud data of transmission lines; S2: performing ground segmentation on the strip laser point cloud data to extract strip ground point cloud data; S3: performing pole tower classification on the strip laser point cloud data, extracting pole tower point clouds, and extracting pole tower foot point clouds based on the pole tower point clouds; S4: filtering, adjusting and refining the strip ground point cloud data using the extracted tower foot point cloud to obtain strip ground point cloud refined data; S5: converting the strip ground point cloud refined data and the elevation model of the digital terrain data into a geodetic ellipsoid elevation model respectively; S6: extracting the boundary of the strip ground point cloud refined data, and fusing the strip ground point cloud refined data and the digital terrain data through a boundary weighted loss function to obtain final digital terrain data.

2. The method for generating digital terrain data of a digital twin of an overhead transmission line according to claim 1, characterized in that: The S2 includes: The strip laser point cloud data is segmented on the ground using a cloth filtering algorithm to extract strip ground point cloud data.

3. The method for generating digital terrain data of a digital twin of an overhead transmission line according to claim 1, characterized in that: The S3 includes: S31: extracting the tower point cloud from the strip laser point cloud data by using the classification label of the tower; S32: filtering out the tower body point cloud using the height percentage for the tower point cloud, and retaining the tower leg point cloud; S33: clustering the tower leg point cloud to divide it into tower leg groups, calculating the lowest point for each group of tower legs, and obtaining the tower foot point cloud.

4. The method for generating digital terrain data of a digital twin of an overhead transmission line according to claim 3, characterized in that: The S32 includes: The height of the tower is calculated, and 80% of the height of the tower is used as a segmentation point to segment the tower into a tower body and a tower foot, and the point cloud data of the tower foot portion is retained as the tower foot point cloud.

5. The method for generating digital terrain data of a digital twin of an overhead transmission line according to any one of claims 1 to 4, characterized in that: The S4 includes: S41: grouping the tower foot point cloud using the Euclidean clustering method, and setting an adjustment radius for each group of tower foot point cloud as a search range; S42: for each group of tower foot point clouds, a neighbor search is performed on the strip ground point cloud data within the adjustment radius to obtain a target ground point cloud set; S43: Calculating the Euclidean distance between each set point in the target ground point cloud set and the tower foot point cloud; d p1 =distance(p1,q) Among them, p1 is the collection point of the target ground point cloud set, q is the tower foot point cloud, d p1 is the Euclidean distance between the set point p1 and the tower foot point cloud q, and distance() represents the calculation function of the Euclidean distance; S44: Calculate the new elevation D of each set point in the target ground point cloud set h1 , and reset the elevation of the rendezvous point to D h1 ; D h1 =p h1 ×(1-d p1 / d max )+q h ×(d p1 / d max ) Among them, p h1 is the original elevation of the collection point p1, D h1 is the new elevation of the collection point p1, q h is the base elevation of the tower foot point cloud q, d max is the maximum search area radius; S45: Calculate the difference between the new elevation of each set point in the target ground point cloud set and the reference elevation of the tower foot point cloud. When the difference is greater than a set first threshold or less than a set second threshold, remove the corresponding set point.

6. The method for generating digital terrain data of a digital twin of an overhead transmission line according to claim 5, characterized in that: The S4 further comprises: S46: Counting the number of point clouds of the tower foot point cloud within the search range, and comparing it with a set number threshold. If the number of searched point clouds is less than the number threshold, it is determined that the point cloud is empty; S47: Select a reference plane at the hole in the point cloud, mesh the reference plane according to the set resolution, sample the coordinates of the grid center point, calculate the elevation of the point cloud interpolation point using the distance weighted method, and generate a new point cloud interpolation point based on the calculated elevation and the grid center point to obtain strip ground point cloud refined data. D h =P h ×(1-distance(P,q edge ) / d max )+q he ×(distance(P,q edge ) / d max ) Among them, P is the center point of the grid, q edge is the closest point to point P in the target ground point cloud set, P h is the elevation of point P, q he for q edge The elevation of the point, D h Interpolate the elevation of the points for the computed point cloud.

7. The method for generating digital terrain data of a digital twin of an overhead transmission line according to claim 6, characterized in that: The S6 comprises: S61: Sampling the digital terrain data into digital terrain point cloud data; S62: extracting boundary contour point cloud from the strip ground point cloud refinement data, and obtaining contour polygon vector data using the extracted boundary contour point cloud; S63: using the outline polygon vector data, removing the point cloud located inside the outline polygon vector data in the digital terrain point cloud data; S64: merging the eliminated digital terrain point cloud data and the strip ground point cloud refined data to obtain merged point cloud data and smoothing the boundary points of the merged point cloud data; S65: resampling the smoothed merged point cloud data into digital elevation model data according to the resolution to obtain final digital terrain data.

8. The method for generating digital terrain data of a digital twin of an overhead transmission line according to claim 7, characterized in that: The S64 includes: S641: traverse the boundary point clouds of the strip ground point cloud refined data, and calculate the normal vector of each boundary point cloud; S642: Taking each boundary point cloud as the center and the normal vector of the boundary point cloud as the direction constraint, searching for neighboring points of the digital terrain point cloud data after elimination; S643: For the searched neighboring points, use the boundary weighted loss function to recalculate the elevation D of the neighboring points h2 , and reset the elevation of the neighboring point to D h2 ; D h2 =p h2 ×(1-d p2 / d max )+E h ×(d p2 / d max ) Among them, p h2 is the original elevation of the neighboring point p2, E h is the base elevation of the boundary point cloud E, d p2 is the Euclidean distance between the neighboring point p2 and the boundary point cloud E, D h2 Recalculate the elevation for neighboring point p2.

9. The method for generating digital terrain data of a digital twin of an overhead transmission line according to claim 1, characterized in that: The digital terrain data is SRTM data, and the strip laser point cloud data is obtained by using a drone to scan using a laser radar.

Images