A method and system for reverse modeling of power transmission line corridor based on three-dimensional laser point cloud

Abstract

The application discloses a kind of transmission line channel reverse modeling methods based on three-dimensional laser point cloud, by being segmented to transmission channel scene, the point cloud data of tower, insulator, wire are obtained, and then the reverse modeling of transmission line channel is completed, avoid the time-consuming and labor-consuming problem of artificial manual modeling, for the tower in the built model library, fine model can be quickly built, and compared with traditional artificial tower modeling, the model built in the application not only contains the physical information such as height, distance, size of tower, and has the characteristics of light weight, finally the tower line direction in visual transmission channel can be seen.The application also establishes a three-dimensional tower model library, extracts the position of insulator using parameterized modeling method, finally completes the connection of ground wire and wire in transmission channel, can adapt to the modeling of different towers in various scenes, improve the modeling efficiency of transmission line.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power transmission line modeling, and in particular to a power transmission line channel reverse modeling method and device based on three-dimensional laser point cloud, an electronic device, and a computer readable storage medium. BACKGROUND

[0002] The power transmission line tower is an important part of the power system, mainly used to support and suspend the power transmission line to transmit electric energy from the power plant to the power consumption site. The topography of different regions in China is different. In order to ensure safe and stable power consumption in different regions, the power transmission line usually needs to cross complex terrains such as mountains and rivers, therefore, the design and construction of the power transmission line tower need to consider the characteristics of different terrains. The types of the currently constructed power transmission line towers include wine cup tower, cat head tower, ganzi type, and goat horn tower. The three-dimensional reconstruction of the power transmission channel helps to intuitively visualize the physical conditions of the tower, insulator, etc., and facilitates the understanding of the layout of the line, which can fully play its value in the process of power inspection.

[0003] At present, the fine modeling of the power transmission line tower and the insulator mainly adopts the manual modeling method, and the engineers use computer aided design (CAD) software to model the power transmission. This method allows engineers to accurately create a three-dimensional geometric model of the line tower, taking into account details such as structure, connection and shape. However, manual modeling usually requires a lot of time and labor cost, and the layout and structure of the power transmission line tower may change over time, and maintaining the latest model may require manual modeling engineers to update the modeling data regularly, consuming a lot of manpower and resources.

[0004] To solve the above problems, the present application changes the previous manual modeling method and proposes a power transmission line channel reverse modeling method based on three-dimensional laser point cloud. By segmenting the power transmission channel scene, the point cloud data of the tower, insulator and conductor are obtained, and then the reverse modeling of the power transmission line channel is completed. A three-dimensional tower model library is established, the insulator position is extracted by using the parameterized modeling method, and finally the connection of the ground wire and the conductor in the power transmission channel is completed. The insulator hanging point position is calculated according to the tower insulator suspension coefficient, and the tower insulator suspension experience value is recorded, thereby improving the modeling efficiency. The present application avoids the time-consuming and labor-consuming problem of manual modeling. For the tower in the built model library, a fine model can be quickly built, and compared with the traditional manual tower modeling, the model built in the present application not only contains the physical information of the height, distance, size, etc. of the tower, but also has the characteristics of lightweight, and finally the tower line trend in the visual power transmission channel. SUMMARY

[0005] The purpose of the present application is to provide a three-dimensional laser point cloud-based reverse modeling method for a power transmission line channel, which acquires tower, insulator and wire point cloud data through the collection and segmentation of power transmission channel point cloud data, establishes a three-dimensional tower reverse modeling model library, classifies the tower model using two-dimensional projection, matches the existing insulator three-dimensional refined model with the tower model according to the point cloud data of the insulator in the power transmission channel, completes the reconstruction of the adjacent tower and insulator models in the power transmission channel, records the tower and insulator line hanging point coordinates, completes the connection of the wires, and further completes the reverse modeling of the power transmission line channel.

[0006] To achieve the above purpose, the present application adopts the following technical solutions:

[0007] The first aspect of the present application provides a three-dimensional laser point cloud-based reverse modeling method for a power transmission line channel, which specifically includes the following steps:

[0008] S1, three-dimensional power transmission channel point cloud data acquisition and segmentation;

[0009] S2, three-dimensional point cloud acquisition of different entities and coordinate system correction;

[0010] S3, establishing a three-dimensional tower reverse modeling model library;

[0011] S4, insulator hanging point and wire hanging point position acquisition;

[0012] S5, connecting the wires of the power transmission channel.

[0013] Further, the step S1 specifically includes the following steps:

[0014] S101, power transmission channel point cloud data preprocessing, which performs thinning processing on the original power transmission line channel point cloud data, that is, performs down-sampling on the point cloud data according to a voxel ratio of 0.1, thereby obtaining uniformly distributed power transmission channel point cloud data;

[0015] S102, power transmission channel point cloud data labeling, which manually labels the data in the point cloud using the CloudCompare tool, mainly including tower, wire, jumper, insulator, ground and other categories;

[0016] S103, power transmission channel point cloud data training, which uses the above labeled data as the input data of PointNet++ and trains it;

[0017] S104, power transmission channel point cloud data segmentation, which uses the PointNet++ algorithm to perform semantic segmentation on the power transmission channel point cloud data, and assigns different point cloud colors to different point cloud objects.

[0018] Further, the step S2 specifically includes the following steps:

[0019] After obtaining the single tower point cloud file, the tower can be corrected in the coordinate system, a bounding box is generated by obtaining the tower head part point cloud, and the rotation angle is calculated so that the tower cross arm length horizontal direction is parallel to the positive direction of the x axis.

[0020] Further, the step S3 specifically includes the following steps:

[0021] The model database stores different types of model names, tower head, tower body, tower foot types and parameters, and directly usable tower models, which are as follows:

[0022]

[0023] Wherein, i is the label of the type, name is the name of the tower type, Respectively represent the name of the tower head, tower foot and tower body, Indicates the model of the tower of this type.

[0024] Further, the step S3 specifically includes the following steps:

[0025] Different tower types are classified using Yolov8, and for the classified point cloud tower data, the tower head, tower body and tower foot are segmented respectively using two-dimensional and three-dimensional projection conversion, and are respectively fine modeled by finding key points, so as to construct a three-dimensional tower reverse modeling model library.

[0026] Further, the step S4 specifically includes the following steps:

[0027] When modeling the tower, the point cloud coordinate position of the tower cross arm, i.e. the insulator suspension coordinate, is obtained and recorded according to the experience value, so as to provide hanging point coordinate support for subsequent accurate insulator suspension position.

[0028] Further, the step S4 specifically includes the following steps:

[0029] The tower height is H, the cross arm height is h, the distance between the tower insulator hanging point and the intersection point of the cross arm and the tower is x, and the tower insulator suspension coefficient alpha determines the point position of the tower insulator hanging point;

[0030] Further, the tower insulator suspension coefficient alpha can be adjusted according to the wire length, line type, tower, etc.

[0031] In a second aspect, the application also provides a power transmission line channel reverse modeling system based on three-dimensional laser point cloud, which specifically includes:

[0032] The point cloud acquisition and segmentation module is used for acquiring and segmenting three-dimensional power transmission channel point cloud data;

[0033] ​The correction module is used for point cloud acquisition of different entities in three dimensions and coordinate system correction.

[0034] The model library establishment module is used for establishing a three-dimensional tower reverse modeling model library.

[0035] The insulator hanging point and conductor hanging point position acquisition module is used for acquiring insulator hanging point and conductor hanging point positions.

[0036] The conductor connection module is used for connecting the conductor of the power transmission channel.

[0037] Further, different kinds of model names, tower head, tower body, tower foot types and parameters, and directly usable tower models are stored in the model database, and the details are as follows:

[0038]

[0039] Wherein, i is the label of the kind, name is the name of the tower type, Respectively represent the name of the tower head, tower foot and tower body, Indicate the model of the tower.

[0040] Further, the model library establishment module classifies different tower types by using Yolov8, for the classified point cloud tower data, the tower head, tower body and tower foot are segmented by using two-dimensional and three-dimensional projection conversion, and the key points are found to finely model them, so as to construct a three-dimensional tower reverse modeling model library.

[0041] In a third aspect, the present application further provides an electronic device, comprising:

[0042] At least one processor, and a memory connected in communication with the at least one processor;

[0043] Wherein, the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the above method.

[0044] In a fourth aspect, the present application provides a computer readable storage medium, and the medium stores computer instructions, and the computer instructions are used to make the computer execute the above method.

[0045] The beneficial effects of the present application are as follows:

[0046] ​The application discloses a power transmission line channel reverse modeling method based on three-dimensional laser point clouds, and the point cloud data of a tower, an insulator and a conductor is acquired by segmenting a power transmission channel scene, and then the reverse modeling of the power transmission line channel is completed, so that the time-consuming and labor-consuming problem of manual modeling is avoided, a fine model can be quickly established for a tower in a built model library, and compared with traditional manual tower modeling, the model established in the application not only contains physical information such as the height, distance and size of the tower, but also has the characteristics of light weight, and finally the tower line direction in the power transmission channel can be visually observed. The application also establishes a three-dimensional tower model library, extracts the insulator position by using a parameterized modeling method, and finally completes the connection of the ground wire and the conductor in the power transmission channel, can adapt to the modeling of different towers in various scenes, and improves the modeling efficiency of the power transmission line. BRIEF DESCRIPTION OF DRAWINGS

[0047] In order to more clearly illustrate the specific embodiments of the application or the technical solutions in the prior art, the following will briefly introduce the drawings needed to be used in the specific embodiments or the prior art description. Obviously, the drawings in the following description are some embodiments of the application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0048] Figure 1 Part of the true color point cloud image of the power transmission channel scanned by the unmanned aerial vehicle in the embodiment of the application;

[0049] Figure 2 The PointNet++ semantic segmentation effect diagram in the embodiment of the application;

[0050] Figure 3 The original coordinate system tower point cloud in the embodiment of the application;

[0051] Figure 4 The tower point cloud after correction of the coordinate system in the embodiment of the application;

[0052] Figure 5 The tower linear model in the embodiment of the application;

[0053] Figure 6 The three-dimensional tower model in the embodiment of the application. DETAILED DESCRIPTION

[0054] The exemplary embodiments of the present application are described below with reference to the accompanying drawings, which include various details of the embodiments of the present application to assist in understanding, and should be considered only as exemplary. Therefore, those skilled in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present application. Similarly, in order to be clear and concise, the description in the following description omits the description of known functions and structures.

[0055] The core idea of the present application is a reverse modeling method of a power transmission line channel based on three-dimensional laser point cloud. By collecting and segmenting the point cloud data of the power transmission channel, the point cloud data of the tower, insulator and conductor is obtained. A three-dimensional tower reverse modeling model library is established, the tower model is classified by two-dimensional projection, the existing three-dimensional fine model of the insulator is matched with the tower model according to the point cloud data of the insulator in the power transmission channel, the reconstruction of the adjacent tower and insulator model in the power transmission channel is completed, the coordinates of the tower and insulator line hanging points are recorded, the connection of the conductor is completed, and the reverse modeling of the power transmission line channel is completed. The model established in the present application not only contains physical information such as height, distance and size, but also has the characteristics of light weight. The specific implementation is as follows:

[0056] Step 1: three-dimensional power transmission channel point cloud data acquisition and segmentation

[0057] The unmanned aerial vehicle is equipped with a laser scanning device, and the acquisition of the power transmission corridor point cloud data is completed according to the specified requirements. The PointNet++ algorithm is used to perform semantic segmentation on the power transmission channel point cloud data, and different point cloud colors are assigned to different point cloud objects. PointNet++ is a deep learning model for processing point cloud data. By introducing multi-level feature learning, sampling and aggregation mechanism, and the combination of local and global features, the effective modeling ability of local and global information in point cloud is improved. Its dense feature learning mechanism is similar to convolutional neural network, which further enhances the performance of the model in point cloud tasks. The specific operation is as follows:

[0058] 1) Power transmission channel point cloud data preprocessing: Since the quality of the true color point cloud image of the unmanned aerial vehicle scanning power transmission line channel (such as Figure 1 ) is uneven, and in order to prevent the problem of too large point cloud data causing too slow model running time, the original power transmission line channel point cloud data is thinned, that is, the point cloud data is down-sampled according to the voxel ratio of 0.1, so as to obtain uniformly distributed power transmission channel point cloud data.

[0059] 2) Power transmission channel point cloud data labeling: the labeling personnel manually labels the data in the point cloud by using the CloudCompare tool, which is mainly divided into tower, conductor, jumper, insulator, ground and other categories.

[0060] 3) Transmission channel point cloud data training: the above labeled data is used as the input data of PointNet++ and is trained, PointNet++ introduces a multi-level feature learning mechanism, which gradually captures local and global features in point cloud data by using a series of neural network layers, each layer is responsible for feature extraction at different levels, which can more comprehensively understand the transmission channel point cloud data. In order to effectively process point cloud data, PointNet++ uses the FarthestPoint Sampling (FPS) algorithm, which selects the most representative point subset to reduce computational complexity, and these selected points are used as input for the subsequent feature learning stage. The local feature learning stage of PointNet++ uses the PointNet structure to learn the features of each local region, uses set pooling (Set Pooling) and other operations to integrate local features to obtain global context information, and the last level generates the transmission line channel segmentation result.

[0061] 4) Transmission channel point cloud data segmentation: the transmission line channel segmentation result generated by PointNet++ is shown in the effect diagram as Figure 2

[0062] Step2: Obtain point cloud of different entities in three dimensions and correct coordinate system

[0063] The point cloud segmentation algorithm in Step 1 assigns different colors to different object categories, and the acquisition of towers, insulators and conductors can be completed according to the color information of the objects. Since it is difficult to model the towers and insulators and extract the conductor hanging points based on the coordinate data in the existing coordinate system, the present application corrects the coordinate system by rotating the tower around the tower center point. The specific operation is as follows:

[0064] For double towers in the transmission channel, since the positions are far apart, the tower classification can be completed by using DBSCAN clustering in machine learning. After obtaining the point cloud file of a single tower, the coordinate system of the tower can be corrected, and the rotation angle is calculated to make the tower cross arm length horizontal direction parallel to the x-axis positive direction, and the correction of the tower point cloud coordinate system is completed. Record the rotation angle of the tower so as to return the subsequent model.

[0065] An example of the original coordinate system tower point cloud is shown in Figure 3 , and an example of the tower point cloud after correction of the coordinate system is shown in Figure 4 .

[0066] Step3: Establish a three-dimensional tower reverse modeling model library

[0067] ​For the corrected tower point cloud data, the tower point cloud front view is obtained by projection mapping, and Yolov8 is used to classify different tower types. For the classified point cloud tower data, the tower head, body and foot are segmented by two-dimensional and three-dimensional projection conversion, and the key points are found to fine model them respectively, so as to build a three-dimensional tower reverse modeling model library. The specific implementation is as follows:

[0068] 1) Tower point cloud front view acquisition.

[0069] For the corrected tower point cloud data, the axis of the coordinate system when the tower front view is analyzed, and the point cloud front view can be obtained by mapping projection.

[0070] 2) Yolov8 algorithm image classification.

[0071] When Yolov8 classifies images, the image size is To avoid the tower shape deformation problem caused by filling the tower point cloud front view in the algorithm internal framework, the tower point cloud front view image is first preprocessed, which is expanded to an image with a size of The surrounding area of the tower is filled with white, and the overall shape of the tower is not changed. After image preprocessing, the image is sent to the Yolov8 image classification framework for training to obtain the tower classification name.

[0072] 3) Tower point cloud segmentation and modeling.

[0073] For different tower point cloud models, the mapping image classification is performed by Yolov8 algorithm model, the tower head, body and foot are segmented by two-dimensional and three-dimensional projection conversion method, and the key points are found to fine model them respectively.

[0074] The tower projection front view is converted to a gray image and binarized, and by traversing each row of data, the number of row points and vertical histogram are obtained. The vertical histogram can directly display the data quantity information of all coordinate points of each row of pixels in the tower front projection view, so as to segment the tower head, body and foot parts.

[0075] For the tower head and foot parts, the tower structure points are obtained according to the key points of the components in different tower types, and the tower head and foot modeling models are obtained by connecting the tower structure points.

[0076] For the tower body part, for the intersection position, first use the index data to take out the point cloud of the middle part intersection position part in the point cloud, use the DBSCAN algorithm to cluster the intersection part point cloud, obtain the number of intersection points, and the height of each intersection point, according to the tower body architecture, connect the obtained intersection structure points, obtain the tower body model. The tower head, tower body and tower foot are merged according to the tower architecture, thereby obtaining the linear model of the tower, and the linear model is meshed, that is, the whole three-dimensional modeling model of the tower is obtained. The tower linear model and the three-dimensional model are shown in Figure 5 Figure 6 .

[0077] 4) Construction of three-dimensional tower reverse modeling model library

[0078] For the shape of the tower in different areas, the tower is classified, three-dimensional reverse modeling is performed on different types of tower shapes, and a model database is constructed, and the construction of the multi-tower tower model library is gradually completed.

[0079] Specifically, the model database stores different types of model names, tower head, tower body, tower foot types and parameters, and directly usable tower models, as follows:

[0080] ;

[0081] Wherein, i is the label of the type, name is the name of the tower type, respectively represent the name of the tower head, the tower foot and the tower body, represent the model of the tower of this type. Different types of tower heads, tower feet and tower bodies record their specific parameters, such as height, shape, etc.

[0082] The three-dimensional model of the tower drawn by the above process not only contains physical information such as the height, distance and size of the tower, but also supports multiple formats for export, such as.ply format,.gltf format,.glb format, etc. It also supports loading on the map.

[0083] Step 4: Insulator hanging point and conductor hanging point position acquisition

[0084] 1) Insulator hanging point position acquisition

[0085] Because even the same tower type, its insulator hanging point position is not necessarily the same, such as for the sheep horn tower, part of the sheep horn tower belongs to the straight tower, and part of the sheep horn tower belongs to the strain tower, the insulator suspension mode and suspension position of the two are not the same. Therefore, the present application simultaneously combines the insulator point cloud data and the tower insulator suspension experience value to complete the acquisition of the insulator hanging point position.

[0086] A coordinate system is established with the tower cross arm and the intersection point of the tower as the origin, the tower height as H, the cross arm height as h, the distance from the tower insulator hanging point to the intersection point of the cross arm and the tower as x, the cross arm midpoint intersecting with the tower, and the cross arm length as 2d.

[0087] The following formula is constructed:

[0088]

[0089] wherein, is the tower insulator suspension coefficient, which can be adjusted according to actual conditions such as wire length, line type, tower, etc. Through value, x is calculated to determine the point of the tower insulator hanging point. Further, the cross arm length d is obtained, wherein d>x.

[0090] Further, a cross arm threshold T is set, and the calculated 2d cannot be greater than the cross arm threshold. If 2d>T, then d= , and x is then at most d.

[0091] After the final insulator hanging point x is calculated, it is stored in the tower model data as a tower insulator suspension experience value.

[0092] When modeling the tower, the point cloud coordinate position of the tower cross arm is obtained and recorded according to the experience value, i.e., the insulator suspension coordinate, to provide hanging point coordinate support for subsequent accurate insulator suspension position.

[0093] The insulator point cloud data is obtained, and the insulator is corrected synchronously using the tower correction angle, so that the positive direction of the corrected insulator coincides with the positive direction of the tower. According to the recorded position coordinates of all insulators that can be hung on the tower, the insulator point cloud data is indexed using an index condition. In the cross arm position at different heights of the tower, if there is insulator point cloud data within the cross arm width, it is recorded that there is a vertical insulator hanging point at this position. If there is insulator point cloud data on both sides of the cross arm width, it is recorded that there are horizontal insulator hanging points at the corresponding two positions.

[0094] The insulator hanging point position obtained by the above method is used to suspend the insulator mesh model at the corresponding position, and the insulator mesh model and the tower mesh model are combined to obtain a three-dimensional reconstruction model of the tower carrying the insulator. Similarly, the combined model contains physical information such as height, distance, and size, and supports export in multiple formats and can be loaded in a map.

[0095] 2) Wire hanging point position acquisition

[0096] The ground wire hanging point is located at the uppermost ends of the tower, and the ground wire hanging point position data can be recorded when the tower is modeled.

[0097] For the conductor hanging point at the insulator, in order to ensure accurate suspension, the modeled insulator mesh model is converted into point cloud data, the position coordinates of the conductor hanging point in the insulator are selected and recorded according to the insulator structure, so as to obtain the conductor hanging point position.

[0098] Step 5: Conductor connection of power transmission channel

[0099] The two towers and the insulator in the power transmission channel have been modeled, and since the conductor hanging point positions in the two towers are recorded respectively, it is necessary to know the matching mode of the two groups of conductor hanging point positions, complete the connection of the conductors in the power transmission channel, and thus complete the reverse modeling of the whole power transmission channel.

[0100] The mesh data of the tower and the insulator modeled in the above steps are inversely rotated according to the rotation angle, so as to rotate back to the original tower and insulator point cloud positions. In the present application, in order to obtain the matching mode of the conductor hanging point, the conductor point cloud data is obtained, and the DBSCAN algorithm is used to classify the conductors, so as to obtain single conductor point cloud data. The conductor hanging point position matching is completed by using the conductor information, and the specific implementation is as follows.

[0101] 1) Conductor point cloud rotation. Since the conductor point cloud data in the original coordinate system has no regularity, the desired line point cloud coordinate information cannot be correctly obtained, so the conductor point cloud data is rotated in the coordinate system, the conductor length is parallel to the x-axis of the coordinate system, and the rotation angle is recorded.

[0102] 2) Conductor point cloud data acquisition. For the corrected conductor point cloud data, the DBSCAN algorithm is used to classify the conductors to obtain single conductor point cloud data. For each conductor point cloud data, the data of the two ends of the conductor are obtained respectively, the mean coordinates of the two end data are calculated respectively, and are recorded.

[0103] 3) Position point information conversion back. The mean coordinate information (position point information) recorded is inversely rotated according to the rotation angle, so as to convert the corrected coordinate information back to the coordinate information in the original power transmission channel coordinate system, and is recorded.

[0104] 4) Position matching. The position information of the two ends of the single conductor converted back is respectively calculated with the conductor hanging point information in Step 4, the two hanging points closest to the two ends are matched respectively, and the matching data of different conductor hanging points is obtained.

[0105] 5) Conductor connection. The conductors are connected according to the matching data, and the reverse modeling of the whole power transmission channel is obtained.

[0106] The whole reverse modeling of the power transmission line is carried out according to the above steps in the application, the complexity and time and labor consuming problems of manual modeling are avoided, and the multiple format export modes are also supported, and the power transmission channel can be accurately loaded on the system map, so that the monitoring personnel can conveniently monitor the power transmission channel.

[0107] In another embodiment of the application, a three-dimensional laser point cloud-based power transmission line channel reverse modeling system is provided, which specifically comprises:

[0108] The point cloud acquisition and segmentation module is used for acquiring and segmenting three-dimensional power transmission channel point cloud data.

[0109] The correction module is used for three-dimensional different entity point cloud acquisition and coordinate system correction.

[0110] The model library establishment module is used for establishing a three-dimensional tower reverse modeling model library.

[0111] The insulator hanging point and conductor hanging point position acquisition module is used for acquiring insulator hanging points and conductor hanging point positions.

[0112] The conductor connection module is used for connecting the power transmission channel conductor.

[0113] The specific implementation of the system is the same as the three-dimensional laser point cloud-based power transmission line channel reverse modeling method, and will not be repeated here.

[0114] Further, the application further provides an electronic device comprising a processor and a memory, the storage device can be configured to store a program code of a support multi-mode wireless communication system Internet gateway connection method for executing the above-mentioned method embodiments, and the processor can be configured to execute the program code in the memory, which includes but is not limited to the program code of the support multi-mode wireless communication system Internet gateway connection method for executing the above-mentioned method embodiments. For the convenience of description, only the parts related to the embodiments of the application are shown, and the specific technical details not disclosed are referred to the method part of the embodiments of the application.

[0115] Further, the application further provides a computer readable storage medium. In a computer readable storage medium embodiment according to the application, the computer readable storage medium can be configured to store a program of a support multi-mode wireless communication system Internet gateway connection method for executing the above-mentioned method embodiments, which can be loaded and run by the processor to realize the above-mentioned support multi-mode wireless communication system Internet gateway connection method.

[0116] The medium in the present application can adopt any combination of one or more computer readable media. The medium can be a computer readable signal medium or a computer readable storage medium. The medium may, for example, be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or apparatus, or any suitable combination of the above. More specific examples (non-exhaustive list) of the medium include an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this document, the medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus or device.

[0117] The above detailed description does not constitute a limitation on the protection scope of the present application. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent replacement and improvement made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for reverse modeling of power transmission line channels based on three-dimensional laser point clouds, characterized in that, Specifically, the steps include the following: S1. Acquisition and segmentation of point cloud data for three-dimensional power transmission channels; Step S1 specifically includes the following steps: S101. Preprocessing of point cloud data of transmission channels: The original point cloud data of transmission channels is thinned out and the point cloud data is downsampled at a voxel ratio of 0.1 to obtain uniformly distributed point cloud data of transmission channels. S102. Point cloud data annotation for power transmission channels: Data annotation is performed manually in the point cloud using the CloudCompare tool, and is divided into categories such as towers, conductors, jumpers, insulators, and ground. S103, Training of point cloud data for power transmission channels: The above-labeled data is used as input data for PointNet++ and trained accordingly. S104. Point cloud data segmentation of power transmission channels: The PointNet++ algorithm is used to perform semantic segmentation on the point cloud data of power transmission channels, and different point cloud colors are assigned to different point cloud objects. S2, point cloud acquisition and coordinate system correction for different 3D entities; step S2 specifically includes the following steps: After obtaining the point cloud file of a single tower, the tower is calibrated in the coordinate system. A bounding box is generated by obtaining the point cloud of the tower head part, and the rotation angle is calculated so that the horizontal direction of the tower crossarm length is parallel to the positive x-axis. For corrected tower point cloud data, analyze the axis of the coordinate system in which the tower is located when viewed from the front, and then perform a mapping projection to obtain the front view of the point cloud. Since YOLOv8 processes images of size 640*640 for image classification, to avoid distortion of the tower shape caused by filling the tower point cloud front view with white within the algorithm's internal framework, the tower point cloud front view image is preprocessed and expanded to a size of 640*640. The area around the tower is filled with white without changing the overall shape of the tower. The preprocessed image is then fed into the YOLOv8 image classification framework for training to obtain the tower classification name. S3. Establish a 3D tower reverse modeling model library; S4. Obtaining the locations of insulator hanging points and conductor hanging points; Step S4 specifically includes the following steps: Establish a coordinate system with the intersection of the crossarm and the tower as the origin, the tower height is H, the crossarm height is h, the distance from the insulator hanging point to the intersection of the crossarm and the tower is x, and the midpoint of the crossarm intersects the tower. Construct the following formula: Wherein, α is the suspension coefficient of the pole insulator, which is adjusted according to the wire length, line type, and pole length; x is calculated from the value of α, thereby determining the location of the pole insulator suspension point; After calculating the final insulator attachment point x, it is stored in the tower model data M. i In the middle, it serves as an empirical value for the suspension of pole insulators; S5, connects the transmission line conductors.

2. The method according to claim 1, characterized in that, Step S3 specifically includes the following steps: Yolov8 is used to classify different tower types. For the classified point cloud tower data, the tower head, tower body and tower foot are segmented by two-dimensional and three-dimensional projection transformation. The tower head, tower body and tower foot are then segmented by finding key points and modeling them in detail. This is used to build a three-dimensional tower reverse modeling model library.

3. A reverse modeling system for power transmission line channels based on three-dimensional laser point clouds, characterized in that, Specifically, it includes: The point cloud acquisition and segmentation module is used to acquire and segment three-dimensional power transmission channel point cloud data. This module also performs preprocessing of the power transmission channel point cloud data, thinning the original point cloud data by downsampling it at a voxel ratio of 0.1 to obtain uniformly distributed point cloud data. Point cloud data annotation is performed manually using the CloudCompare tool, categorizing the data into towers, conductors, jumpers, insulators, and ground types. Point cloud data training uses the annotated data as input for PointNet++ training. Finally, point cloud data segmentation utilizes the PointNet++ algorithm for semantic segmentation, assigning different point cloud colors to different point cloud objects. The correction module is used for point cloud acquisition and coordinate system correction of different three-dimensional entities. After acquiring a single tower point cloud file, the correction module performs coordinate system correction on the tower. It generates a bounding box by acquiring the point cloud of the tower head and calculates the rotation angle so that the horizontal direction of the tower crossarm length is parallel to the positive x-axis. For corrected tower point cloud data, analyze the axis of the coordinate system in which the tower is located when viewed from the front, and then perform a mapping projection to obtain the front view of the point cloud. Since YOLOv8 processes images of size 640*640 for image classification, to avoid distortion of the tower shape caused by filling the tower point cloud front view with white within the algorithm's internal framework, the tower point cloud front view image is preprocessed and expanded to a size of 640*640. The area around the tower is filled with white without changing the overall shape of the tower. The preprocessed image is then fed into the YOLOv8 image classification framework for training to obtain the tower classification name. The model library creation module is used to create a 3D tower reverse modeling model library; The module for obtaining the locations of insulator attachment points and conductor attachment points is used to obtain the locations of insulator attachment points and conductor attachment points. A coordinate system is established with the intersection of the tower crossarm and the tower as the origin, the tower height is H, the crossarm height is h, the distance from the tower insulator attachment point to the intersection of the crossarm and the tower is x, and the midpoint of the crossarm intersects with the tower. Construct the following formula: Wherein, α is the suspension coefficient of the pole insulator, which is adjusted according to the wire length, line type, and pole length; x is calculated from the value of α, thereby determining the location of the pole insulator suspension point; After calculating the final insulator attachment point x, it is stored in the tower model data M. i In this context, it serves as an empirical value for the suspension of pole insulators; The conductor connection module is used to connect the conductors of the power transmission channel.

4. In the system according to claim 3, the model library building module uses Yolov8 to classify different tower types. For the classified point cloud tower data, the tower head, tower body, and tower foot are segmented by two-dimensional and three-dimensional projection transformation, and the tower head, tower body, and tower foot are refined by finding key points, thereby constructing a three-dimensional tower reverse modeling model library.

5. An electronic device, comprising: At least one processor, and a memory communicatively connected to at least one of the processors; The memory stores instructions executable by at least one of the processors, which are executed by at least one of the processors to enable the at least one of the processors to perform the method according to any one of claims 1-2.

6. A computer-readable storage medium, characterized in that, The medium stores computer instructions for causing the computer to perform the method described in any one of claims 1-2.

Images