Power transmission line external damage prevention monitoring and early warning device and method

Abstract

The invention relates to the technical field of power transmission line monitoring, in particular to a power transmission line external damage prevention monitoring and early warning device and method.The method is characterized in that an unmanned aerial vehicle monitors a power transmission line according to a set route, scans the environment of the power transmission line through a laser radar and collects position and distance information to obtain three-dimensional point cloud data; establishing a three-dimensional point cloud model through the three-dimensional point cloud data; an unmanned aerial vehicle is used for routing inspection, pictures are shot, the three-dimensional point cloud model is updated in real time, and real-time monitoring is conducted; the distance in the three-dimensional point cloud model is monitored, a preset value is set, and when the distance is lower than the preset value, an alarm is given; therefore, monitoring and early warning of external damage prevention are performed on the power transmission line based on the three-dimensional point cloud technology, and real-time monitoring is performed, so that the accuracy and timeliness of hidden danger identification are greatly improved, the human input is reduced, and the operation and maintenance inspection efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power transmission line monitoring, and in particular to a power transmission line anti-external damage monitoring and early warning device and method. Background Art

[0002] Transmission lines are the main power facilities responsible for power transmission and distribution between power plants and substations, and between substations. They are called the main arteries of the power grid system, and their safe and stable operation plays a vital role in ensuring the supply of electric energy. High-voltage transmission lines are laid outdoors with bare wires, exposed to the natural environment, and are often affected by external factors, such as foreign objects invading safety limits and external forces damaging transmission lines. Every year, power grid units invest a lot of manpower, material resources and financial resources in preventing transmission lines from breaking. Traditional transmission line monitoring methods mainly rely on manual inspections, which have high labor costs, low inspection efficiency and lack of real-time monitoring functions. In order to overcome these defects, a transmission line anti-breaking monitoring and early warning system came into being. The system has the advantages of automation, real-time, dynamic early warning and data analysis, and can realize all-weather real-time monitoring and automatic inspection of transmission lines. It can accurately and efficiently identify key features, highly reliable alarms, and conduct digitized record analysis, which reduces manpower input and improves work efficiency. It is an important technical means to achieve transmission line status operation, maintenance management, and improve the lean level of production and operation management. It is an important guarantee for the safe and stable operation of the power system, and provides strong support for the sustainable development of the power industry.

[0003] In the existing technology, with the rapid development and construction of cities, the large-scale construction scenes along the overhead transmission lines have also increased accordingly; if the distance between these large machines and the overhead transmission lines is not properly controlled during operation, it is very easy to cause damage to the lines or electric shock accidents. Summary of the invention

[0004] The purpose of the present invention is to provide a transmission line anti-external damage monitoring and early warning device and method to solve the problem that large-scale construction operations along overhead transmission lines in the prior art have also increased accordingly; during the operation of these large machines, if the distance from the overhead transmission lines is not properly controlled, it is very easy to cause damage to the lines or electric shock accidents to people.

[0005] To achieve the above object, the present invention provides a transmission line anti-extrinsic damage monitoring and early warning method, comprising the following steps:

[0006] The drone monitors the power transmission line according to the set route, and relies on the laser radar to scan the power transmission line environment, collect position and distance information, and obtain three-dimensional point cloud data;

[0007] Establishing a three-dimensional point cloud model through the three-dimensional point cloud data;

[0008] Use drones to inspect, take pictures, and update the three-dimensional point cloud model in real time for real-time monitoring;

[0009] The distance in the three-dimensional point cloud model is monitored and a preset value is set. When the distance is lower than the preset value, an alarm is triggered.

[0010] Wherein, in the step of establishing a three-dimensional point cloud model using the three-dimensional point cloud data:

[0011] Preprocessing the three-dimensional point cloud data to obtain preprocessed point cloud data;

[0012] Performing instance segmentation on the preprocessed point cloud data to obtain point cloud segmentation data;

[0013] Performing point cloud feature extraction on the point cloud segmentation data to obtain point cloud feature data;

[0014] Registering the point cloud feature data to obtain point cloud registration data;

[0015] Three-dimensional reconstruction is performed based on the point cloud registration data to obtain a three-dimensional point cloud model.

[0016] Wherein, in the step of preprocessing the three-dimensional point cloud data to obtain preprocessed point cloud data:

[0017] Using a bilateral filtering method, determining a neighborhood range of each point in the three-dimensional point cloud data;

[0018] According to the distance in the spatial domain and the similarity in the value range, the weight of each point in the neighborhood to the current point is calculated;

[0019] Perform weighted averaging on the points in the neighborhood according to the weights to obtain the filtering result of the current point;

[0020] Using a statistical filter method, calculating the mean and standard deviation of points in a neighborhood of the three-dimensional point cloud data;

[0021] According to the set standard deviation multiple threshold, determine whether the current point is an outlier. If the distance from the current point to the neighborhood mean is greater than the set standard deviation multiple threshold, it will be regarded as an outlier and removed.

[0022] Wherein, in the step of performing instance segmentation on the pre-processed point cloud data to obtain point cloud segmentation data:

[0023] The point cloud feature image is recognized through the convolutional neural network to complete the preliminary category classification;

[0024] Perform pixel-level segmentation on the image objects classified by categories to obtain pixel points;

[0025] Each of the pixel points is classified into a specific object instance of the same category, and the precise outline and mask of each object are generated to help the automatic ranging algorithm to obtain the target position more accurately.

[0026] Wherein, in the step of performing point cloud feature extraction on the point cloud segmentation data to obtain point cloud feature data:

[0027] Through the laser radar three-dimensional scanning technology, the depth information of each pixel in the image is obtained;

[0028] The camera captures the light in the scene, which passes through the camera's small hole and is projected onto the imaging plane to form an image;

[0029] After passing through the pinhole, the three-dimensional point is projected onto the image plane of the camera to form a two-dimensional pixel point;

[0030] By means of a plurality of the three-dimensional points, two-dimensional pixel coordinates are formed, and the two-dimensional pixel coordinates represent the measured distance.

[0031] Wherein, in the step of registering the point cloud feature data to obtain point cloud registration data:

[0032] Use the SAC-IA coarse registration algorithm based on FPFH features for fast coarse registration;

[0033] The coarsely registered data is finely registered using the ICP algorithm;

[0034] A feature point matching algorithm is used to align obvious feature points in the transmission line, where the obvious feature points include but are not limited to tower bases and suspension points.

[0035] Among them, in the step of monitoring the distance in the three-dimensional point cloud model and setting a preset value, and issuing an alarm when the distance is lower than the preset value:

[0036] In the step of monitoring the distance in the three-dimensional point cloud model and setting a preset value, and issuing an alarm when the distance is lower than the preset value:

[0037] Divide into multiple levels of distance preset values, the preset values ​​represent the safe distance between the power transmission line and the mechanical object, and each level of the preset value sends a different warning signal;

[0038] When the distance displayed by the two-bit pixel coordinates is lower than the third-level preset value, an audible and visual alarm is issued to alert the staff;

[0039] When the distance displayed by the two-dimensional pixel coordinates is lower than the secondary preset value, the on-site management personnel are notified to check;

[0040] When the distance displayed by the two-bit pixel coordinates is lower than the first-level preset value, the on-site construction is stopped;

[0041] Set up the remote APP to receive alarm information, video playback and remote real-time viewing.

[0042] The present invention also provides a transmission line anti-external damage monitoring and early warning device, which adopts the above-mentioned transmission line anti-external damage monitoring and early warning method, including an unmanned aerial vehicle, a high-performance processor, a dual-channel high-definition camera and a solar power supply module, and the high-performance processor, the dual-channel high-definition camera and the solar power supply module are all arranged on the unmanned aerial vehicle.

[0043] The present invention provides a power transmission line anti-external damage monitoring and early warning device and method. A drone monitors the power transmission line according to a set route, and at the same time relies on a laser radar to scan the power transmission line environment, collects position and distance information, and obtains three-dimensional point cloud data; a three-dimensional point cloud model is established through the three-dimensional point cloud data; a drone is used for inspection, pictures are taken, and the three-dimensional point cloud model is updated in real time for real-time monitoring; the distance in the three-dimensional point cloud model is monitored, and a preset value is set. When the distance is lower than the preset value, an alarm is issued;

[0044] Therefore, based on the three-dimensional point cloud technology, the transmission lines can be monitored and warned to prevent external damage, and real-time monitoring can be carried out, which greatly improves the accuracy and timeliness of hidden danger identification, reduces manpower input, and improves the efficiency of operation and maintenance inspections. Secondly, by monitoring the distance, the passive emergency rescue of external force damage is changed to active prevention in advance, which greatly reduces economic losses and improves the risk resistance of the power grid. In addition, the system can be combined with many advanced technologies to build a comprehensive and three-dimensional power grid operation and maintenance system. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art are briefly introduced below.

[0046] Figure 1 It is a coordinate diagram of instance segmentation of the preprocessed data in the present invention.

[0047] Figure 2 It is a flow chart of the steps of the power transmission line anti-external damage monitoring and early warning method of the present invention.

[0048] Figure 3 It is a flow chart of the steps of establishing a three-dimensional point cloud model through three-dimensional point cloud data of the present invention.

[0049] Figure 4 It is a flow chart of the steps of preprocessing three-dimensional point cloud data of the present invention.

[0050] Figure 5 It is a flow chart of the steps of performing instance segmentation on preprocessed data of the present invention.

[0051] Figure 6 It is a flow chart of the steps of extracting point cloud features from preprocessed data of the present invention.

[0052] Figure 7 It is a flow chart of the steps of registering the preprocessed data of the present invention.

[0053] Figure 8 The present invention is a flow chart of the steps of monitoring the distance in the three-dimensional point cloud model and setting a preset value, and issuing an alarm when the distance is lower than the preset value.

[0054] Fig. 9 It is a structural diagram of the power transmission line anti-external damage monitoring and early warning device of the present invention.

[0055] 1-UAV, 2-High-performance processor, 3-Dual-channel HD camera, 4-Solar power supply module. DETAILED DESCRIPTION

[0056] Embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, but should not be construed as limiting the present invention.

[0057] See also Figures 1 to 8 The present invention provides a transmission line anti-extreme damage monitoring and early warning method, comprising the following steps:

[0058] S101: UAV 1 monitors the power transmission line according to the set route, and at the same time relies on laser radar to scan the power transmission line environment, collects position and distance information, and obtains three-dimensional point cloud data;

[0059] LiDAR acquires the position and distance information of an object by emitting a laser beam and measuring its reflection time. Specifically, LiDAR is composed of a laser transmitter, a receiver, a scanning system, and a control circuit. The laser transmitter of the LiDAR emits a laser beam. The laser beam will be reflected back after encountering an object in the environment. The receiver captures the reflected light signal and converts it into an electrical signal. Through a scanning system, such as a rotating mirror or a mechanical turntable, the laser beam is scanned within a certain angle range to detect the environment in all directions. The control circuit decodes, corrects and integrates the received data, extracts the position and distance information of each laser beam, and generates three-dimensional point cloud data. The higher the accuracy and resolution of the LiDAR, the more accurate the three-dimensional point cloud data provided will be in recording the details of the environment; the three-dimensional point cloud data is a collection of a large number of three-dimensional points, each point contains three-dimensional coordinate information and other related information, forming the three-dimensional point cloud data;

[0060] For large transmission lines, a long-distance, large-field-of-view LiDAR system may be required. The device has excellent strong electric and magnetic interference resistance and can adapt to various complex environments and climatic conditions. Complete the scan in the shortest possible time to improve the efficiency of data collection. Combined with GPS positioning and attitude measurement technology, high-precision spatial positioning is achieved to ensure the accuracy of the three-dimensional point cloud model and the stability and reliability of the device in practical applications.

[0061] S102: Establishing a three-dimensional point cloud model through the three-dimensional point cloud data;

[0062] Using advanced computer algorithms, the 3D point cloud data is processed, including denoising, filtering, registration and other steps, to eliminate errors, optimize data quality, and ultimately generate an accurate and realistic 3D model. This model not only accurately captures all surface details of the object or scene, but also completely preserves its 3D form, spatial structure and size information, providing a solid foundation for subsequent analysis, simulation, design or visualization applications.

[0063] The specific steps include:

[0064] S201: preprocessing the three-dimensional point cloud data to obtain preprocessed point cloud data;

[0065] When processing 3D point cloud data, a preprocessing algorithm is used to remove noise, extract effective information, and divide the 3D point cloud data into different parts for subsequent modeling. After the preprocessing is completed, an efficient 3D modeling algorithm needs to be used to convert the point cloud data into a 3D model; in addition, based on the characteristics of transmission lines and power towers, an algorithm based on surface reconstruction, such as Poisson reconstruction or Delaunay triangulation, can automatically construct a smooth surface model based on the distribution of point cloud data, thereby accurately reflecting the actual form of transmission lines and power towers; after obtaining the preliminary 3D model of the transmission line, we also need to optimize the model and post-process it to improve the accuracy and aesthetics of the model. Specific measures include: smoothing the model to remove burrs and uneven parts on the surface; enhancing the details of the model to highlight the characteristics of the transmission line and power tower; and texture mapping the model to enhance the realism of the model.

[0066] The specific steps include:

[0067] S301: using a bilateral filtering method to determine a neighborhood range of each point in the three-dimensional point cloud data;

[0068] The specific operation purpose of using the bilateral filtering method is to effectively remove noise and smooth the surface while retaining the edge features of the point cloud data. This method determines the neighborhood range of each point in three-dimensional space, comprehensively considers the spatial distance and color (or intensity) similarity, and performs weighted processing on the point cloud data. Its advantage is that it can not only remove high-frequency noise, but also maintain the geometric structure and feature details of the point cloud data, avoiding the problem of edge blur or feature loss that may be caused by traditional filtering methods. The advantage of this is that the pre-processed point cloud data is of higher quality, providing a more accurate and reliable foundation for subsequent applications such as three-dimensional modeling, distance monitoring and real-time alarm, and improving the performance and accuracy of the overall system.

[0069] S302: Calculate the weight of each point in the neighborhood to the current point based on the distance in the spatial domain and the similarity in the value range;

[0070] By calculating the weights, the spatial position and attribute (such as color, intensity, etc.) information of the point cloud data are comprehensively considered during the filtering process to more finely control the filtering effect. By calculating the weight of each point in the neighborhood to the current point, it can be ensured that the edge and feature details are effectively retained while smoothing the point cloud surface. The advantage of this method is that it can better maintain the geometric structure and attribute characteristics of the point cloud data while removing noise, avoiding the problems of over-smoothing or feature loss that may be caused by traditional filtering methods. The benefit of this is that the pre-processed point cloud data is more accurate and reliable, providing a better data foundation for subsequent applications such as 3D modeling, distance monitoring and real-time alarm.

[0071] S303: Perform weighted averaging on the points in the neighborhood according to the weights to obtain a filtering result for the current point;

[0072] By comprehensively considering the influence of each point in the neighborhood on the current point (i.e., weight), the true value of the current point can be estimated more accurately. By weighted averaging, noise can be effectively removed while retaining the details and features of the point cloud data. The advantage of this method is that it can reduce information loss while smoothing the point cloud surface, avoiding edge blur and feature loss. The benefit of this is that the preprocessed point cloud data is of higher quality and has better continuity and smoothness.

[0073] S304: using a statistical filter method to calculate the mean and standard deviation of points in the neighborhood of the three-dimensional point cloud data;

[0074] By statistically analyzing the distribution characteristics of points in the neighborhood, noise points or outliers are identified and removed, thereby enhancing the overall quality of point cloud data. The statistical filter method calculates the mean and standard deviation of points in the neighborhood, sets a reasonable threshold range, and removes points outside this range as noise or outliers. The advantage of this method is that it can not only effectively remove noise from point cloud data, but also retain the geometric features and details of the point cloud to a certain extent, avoiding the problems of over-smoothing or feature loss that may be caused by traditional filtering methods.

[0075] S305: According to the set standard deviation multiple threshold, determine whether the current point is an outlier. If the distance between the current point and the neighborhood mean is greater than the set standard deviation multiple threshold, it is regarded as an outlier and removed.

[0076] By using statistical analysis methods, noise points or isolated points in point cloud data can be automatically identified and removed, thereby optimizing the quality of point cloud data. The purpose of this operation is to improve the accuracy and reliability of point cloud data, laying a solid foundation for subsequent 3D modeling, distance monitoring and other applications. Its advantage is that by setting a reasonable standard deviation multiple threshold, it is possible to effectively remove abnormal points that are significantly different from the distribution of neighboring points while retaining the main features of the point cloud, thereby avoiding modeling errors caused by noise interference. The advantage of this is that the preprocessed point cloud data is purer and more accurate, and can more realistically reflect the 3D morphology of the actual object.

[0077] S202: performing instance segmentation on the pre-processed point cloud data to obtain point cloud segmentation data;

[0078] The instance segmentation algorithm is used to accurately segment each object in the image or video, and different instances are accurately labeled at the pixel level. The multi-dimensional perception data is fused and matched with high precision through the two-dimensional and three-dimensional data registration algorithms. At the same time, the spatial coordinate system mapping and conversion algorithm is used to synchronize the data to a unified coordinate system. Combined with the three-dimensional spatial measurement technology and multi-scale feature extraction technology, accurate distance data between the transmission line and the vehicle machinery can be obtained.

[0079] The specific steps include:

[0080] S401: Recognize the point cloud feature image through a convolutional neural network and complete preliminary category classification;

[0081] The powerful feature extraction and classification capabilities of convolutional neural networks are used to efficiently and accurately identify and classify preprocessed point cloud feature images. Through multi-layer convolution, pooling and other operations, convolutional neural networks can learn the deep features of point cloud data, thereby effectively distinguishing different objects or components; convolutional neural networks are highly sensitive to local features of images (including point cloud feature images) and can automatically extract multi-level and multi-scale feature information. This enables the network to capture subtle differences in point cloud data and improve the accuracy and robustness of classification. In addition, convolutional neural networks also have good generalization capabilities and can maintain stable performance in different scenarios and conditions.

[0082] Through preliminary category classification, it can provide strong support for subsequent instance segmentation. After knowing the approximate category of each point, it is easier to achieve accurate segmentation of the point cloud and improve the accuracy and efficiency of segmentation. Secondly, the application of convolutional neural networks can greatly reduce the cost of manual annotation and intervention and improve the degree of automation of data processing. Finally, with the continuous development of convolutional neural network technology, its application prospects in the field of point cloud processing will become more and more broad, injecting new vitality into the development of the field of three-dimensional vision.

[0083] S402: Segment the image objects classified by categories at the pixel level to obtain pixel points;

[0084] When calculating pixels:

[0085] To transform the point cloud coordinates into the camera coordinate system, one or more transformation matrices are usually required. These matrices include intrinsic matrix, extrinsic matrix, calibration matrix, etc.

[0086] From the world coordinate system to the camera coordinate system, we need to know the position and posture of the camera relative to the world coordinate system, which is described by the camera's external parameter matrix R (rotation matrix) and T (translation vector). Figure 1 .

[0087] 3D point P w =[X w , Y w , Z w , 1] T The coordinates in the world coordinate system can be converted to the camera coordinate system through the following transformation:

[0088] P c =RP w +T

[0089] After conversion to the camera coordinate system, the three-dimensional coordinates need to be further converted into two-dimensional pixel coordinates (pixel points) through the camera's intrinsic parameter matrix (including information such as focal length and principal point coordinates).

[0090] 3D point PC =[X C , Y C , X C ] T Convert to two-dimensional pixel coordinates p = [u, v] T Using the pinhole camera model, the formula is as follows:

[0091] [\begin{align*}

[0092] u&=f x \frac{X C}{Z C}+c x \v&=f y \frac{Y C}{Z C}+c y

[0093] \end{align*}]

[0094] Among them, f x and f y are the focal lengths of the camera in the x-axis and y-axis directions, c x and c y are the coordinates of the principal point of the image.

[0095] S403: Classify each of the pixel points into a specific object instance of the same category, generate an accurate outline and mask of each object, and help the automatic ranging algorithm to obtain the target position more accurately.

[0096] It can accurately distinguish and identify different object instances in point cloud data, so as to provide more accurate target location information for subsequent automatic ranging algorithms.

[0097] S203: performing point cloud feature extraction on the point cloud segmentation data to obtain point cloud feature data;

[0098] The specific steps include:

[0099] S501: Obtaining depth information of each pixel in the image through laser radar three-dimensional scanning technology;

[0100] LiDAR can accurately measure the distance of an object's surface by emitting a laser beam and receiving the reflected signal, thereby generating high-precision 3D point cloud data. This technology can not only provide 3D coordinate information for each pixel, but also obtain richer data information such as the reflection intensity of the point.

[0101] The depth information reflects the distance between the object and the camera. The accuracy of the depth information directly affects the accuracy of the distance measurement. Therefore, point cloud data collection must be strictly in accordance with the collection standards and specifications; combined with the instance segmentation results and depth information, each object instance is positioned in three-dimensional space to calculate its distance relative to the camera; based on the position information of the target, the relative distance between the targets or the absolute distance from the target to the observation point can be further calculated. For example, the distance from a vehicle or large machinery to a transmission line.

[0102] S502: Capturing light in the scene through a camera, the light passes through a small hole in the camera and is projected onto an imaging plane to form an image;

[0103] Utilizing the imaging principle of the camera, the complex light information in the three-dimensional scene can be intuitively presented in the form of a two-dimensional image, making it easier for people to observe and analyze. At the same time, the process of the camera capturing light and forming an image is real-time and continuous, and can record dynamic changes in the scene in real time, making dynamic monitoring and analysis possible.

[0104] S503: After the three-dimensional point passes through the small hole, it is projected onto the image plane of the camera to form a two-dimensional pixel point;

[0105] By projecting three-dimensional points onto a two-dimensional image plane, we can use mature image processing techniques and algorithms, such as edge detection, corner detection, texture analysis, etc., to extract feature information from the image. These feature information are important for subsequent object recognition and classification.

[0106] S504: Form two-dimensional pixel coordinates based on the plurality of three-dimensional points, wherein the two-dimensional pixel coordinates represent the measured distance.

[0107] The point cloud data in the three-dimensional space is converted into pixel coordinates in the two-dimensional image space, and the distance between the original three-dimensional points is indirectly represented by these pixel coordinates, realizing the effective mapping from the three-dimensional space to the two-dimensional image space, while retaining the key information about the shape, structure and distance of the object in the point cloud data. This mapping relationship not only reduces the dimension of data processing and reduces the computational complexity, but also allows us to use mature technologies and algorithms in the field of two-dimensional image processing to analyze and utilize these point cloud data. Through the formed two-dimensional pixel coordinates, we can perform edge detection, texture analysis, shape recognition and other operations to extract feature information from the point cloud data.

[0108] S204: registering the point cloud feature data to obtain point cloud registration data;

[0109] The specific steps include:

[0110] S601: Use the SAC-IA coarse registration algorithm based on FPFH features for fast coarse registration;

[0111] By using the local geometric shape description capability of the FPFH feature descriptor and combining it with the sampling consistency idea of ​​the SAC-IA algorithm, we can achieve fast and preliminary alignment of two point cloud data. Through coarse registration, we can provide a good initial position for the subsequent precise registration algorithm, thereby accelerating the entire registration process and improving the registration accuracy.

[0112] The SAC-IA algorithm performs calculations by sampling corresponding point pairs, which reduces the computational complexity and makes the rough registration process more efficient. At the same time, the calculation of the FPFH feature is relatively simple and can quickly extract local features of the point cloud;

[0113] The SAC-IA algorithm is robust to point cloud noise and density, and can perform stable registration in the presence of noise or uneven point cloud density. In addition, the FPFH feature is also invariant to the rotation of the point cloud, further enhancing the robustness of the algorithm.

[0114] S602: Perform fine registration on the roughly registered data by using an ICP algorithm;

[0115] The ICP algorithm is used to further optimize the results of the rough registration to obtain more accurate registration data. Although the rough registration can provide a good initial position for the subsequent registration, its accuracy is often limited. The ICP algorithm can iteratively find the nearest point pair and calculate the optimal transformation, thereby gradually reducing the error between the two point clouds and achieving more accurate registration; the ICP algorithm can not only process three-dimensional point sets, but also surfaces expressed in various forms such as implicit surfaces and parametric curves, and has a wide range of applicability.

[0116] S603: Use a feature point matching algorithm to align obvious feature points in the transmission line, where the obvious feature points include but are not limited to tower bases and suspension points.

[0117] In the point cloud data of the transmission line, accurately identify and match those points with significant geometric or physical characteristics, such as tower bases and suspension points. These feature points should be consistent in point cloud data from different perspectives or positions, so they can be used as a reliable reference in the registration process. By accurately matching these feature points, a coordinate transformation or rigid body transformation (such as translation, rotation, etc.) can be found so that the corresponding points between multiple point clouds can accurately coincide, thereby achieving accurate registration of the transmission line point cloud data.

[0118] S205: Perform three-dimensional reconstruction based on the point cloud registration data to obtain a three-dimensional point cloud model.

[0119] Convert the precisely registered point cloud data into a model with 3D geometric information; this process integrates the registered point cloud data to construct the 3D shape and structure of the object, providing a basis for subsequent analysis, visualization, simulation and interaction. The 3D point cloud model can accurately reflect the spatial position, shape and size of the object, providing reliable data support for various application scenarios.

[0120] S103: using the drone 1 to inspect, take pictures, and update the three-dimensional point cloud model in real time to perform real-time monitoring;

[0121] During the inspection process, the drone 1 uses a high-precision camera to take real-time pictures, capture three-dimensional information of the ground or target objects, and immediately convert this information into three-dimensional point cloud data for model building; then, a processing algorithm is used to update and build a dynamic three-dimensional point cloud model in real time. The model accurately reflects the current status of the target area. This process realizes real-time monitoring of the target area, can promptly detect and track any changes or abnormal situations, and provides timely and accurate information support for subsequent decisions and actions.

[0122] S104: monitoring the distance in the three-dimensional point cloud model and setting a preset value, and issuing an alarm when the distance is lower than the preset value.

[0123] Closely monitor the key distance indicators in the model and preset safety distance thresholds; once it is detected that any distance value in the model is lower than the preset safety threshold, the system will immediately and automatically trigger the alarm mechanism and send a clear and unambiguous warning signal to the operator so that they can quickly identify risks, assess the situation, and immediately take necessary countermeasures, thereby effectively avoiding potential safety hazards or accidents.

[0124] The specific steps include:

[0125] S701: Divide into multiple levels of distance preset values, the preset values ​​represent the safe distance between the power transmission line and the mechanical object, and each level of the preset value sends a different warning signal;

[0126] By finely dividing the multi-level preset values ​​of the safety distance, accurate monitoring and early warning of the distance between the transmission line and the mechanical object can be achieved. Each level of preset value represents a different safety level. When the actual distance is lower than a certain level of preset value, the corresponding early warning signal will be triggered, thereby promptly reminding the operator or system to take corresponding safety measures.

[0127] S702: When the distance displayed by the two-bit pixel coordinates is lower than the third-level preset value, an audible and visual alarm is issued to alert the staff;

[0128] By combining the display of two-dimensional pixel coordinates with the setting of three-level preset values, accurate monitoring of the distance between the transmission line and surrounding objects can be achieved. When the distance is lower than the safety threshold, the sound and light alarm is used in an intuitive and urgent way to quickly attract the attention of the staff, reminding them to take necessary measures in time to prevent potential safety risks or accidents.

[0129] S703: When the distance displayed by the two-bit pixel coordinate is lower than the secondary preset value, notify the on-site management personnel to check;

[0130] When the actual distance is lower than this value, promptly notify site managers to conduct an on-site inspection so that they can make an accurate assessment of the site situation and take necessary measures to prevent potential safety risks or accidents.

[0131] S704: When the distance displayed by the two-bit pixel coordinates is lower than the first-level preset value, the on-site construction is stopped;

[0132] When the actual distance is lower than this value, timely measures shall be taken to stop on-site construction to prevent unsafe approach or contact between the transmission line and surrounding objects caused by construction activities; ensure the safety of the transmission line and its surrounding environment, and prevent potential damage or accidents caused by construction.

[0133] S705: Set up the remote APP to receive alarm information, video playback and remote real-time viewing.

[0134] Develop a smart management platform to monitor the status of on-site operations in real time, and implement functions such as user management, equipment management, authority management, log management, and alarm record query; special events on site can be displayed through charts, maps, etc.; the platform designs a unified data interface to support access to various front-end devices; develop a video surveillance APP to support real-time monitoring and viewing, receiving alarm information, video playback and downloading, etc.;

[0135] In addition, the 3D point cloud data collection and calculation processing calibration file pre-loading method is adopted to compare the target objects in the real-time video monitoring screen with the pre-generated 3D point cloud models of the transmission lines and towers, and the distance information between the target objects and the transmission lines can be calculated and displayed in real time and accurately. Considering that the 3D space environment will not change after the construction of the transmission lines and towers is completed, this project innovatively adopts the method of only collecting and processing the 3D point cloud data in advance once, and then downloading the calibration file to the device, and no real-time collection is required later. The pre-collected and processed 3D point cloud data reduces the pressure of edge computing and ensures the accuracy and efficiency of measurement.

[0136] See also Fig. 9The present invention also provides a power transmission line anti-external damage monitoring and early warning device, including an unmanned aerial vehicle 1, a high-performance processor 2, a dual-channel high-definition camera 3 and a solar power supply module 4, wherein the high-performance processor 2, the dual-channel high-definition camera 3 and the solar power supply module 4 are all arranged on the unmanned aerial vehicle 1.

[0137] Among them, the high-performance processor 2 adopts a high-performance, low-power AI core control board, integrates a high-efficiency AI neural network processing unit NPU, has a computing power of up to 4.0Tops, and can easily achieve 3 trillion operations per second with extremely low power consumption. It has built-in professional AI algorithms and provides a "hardware + software + algorithm" system solution, which takes into account performance, power consumption, and computing accuracy to the greatest extent. The device supports intelligent recognition functions: the dual-channel high-definition camera 3 adopts an ultra-high-definition 24-megapixel main camera to achieve high-definition acquisition during the day; the ultra-low illumination night vision 4-megapixel secondary camera adopts a special optical design and image sensor technology, with high sensitivity and low noise level, and can capture useful image information in weak light, realizing clear image acquisition at night or in low light conditions. Working 24 hours a day, it can ensure clear and high-quality images regardless of day or night; real-time online capture and quick response;

[0138] The camera uses the global positioning system (GPS) and inertial navigation system (INS), and the 3D lidar device integrates the GPS receiver and INS unit to ensure that the lidar can obtain accurate position and attitude information (including latitude and longitude, altitude, pitch angle, roll angle and yaw angle, etc.) in real time during movement, providing an accurate geographic reference frame for the point cloud data;

[0139] At the same time, distributed cameras can be installed to provide multi-point image coverage, visual full-tower coverage and other functions; the solar power supply module 4 can utilize solar energy to generate electricity and is fully installed on the drone 1, thus becoming an integrated device of software and hardware algorithms without the need to deploy a back-end server, and a higher level of integration can be achieved through technical means such as chip design and integration.

[0140] What is disclosed above is only one or more preferred embodiments of the present application, and cannot be used to limit the scope of rights of the present application. Ordinary technicians in this field can understand that all or part of the processes of implementing the above embodiments and equivalent changes made according to the claims of the present application are still within the scope covered by the present application.

Claims

1. A method for monitoring and early warning of power transmission line damage prevention, characterized in that: The steps include: The drone monitors the power transmission line according to the set route, and relies on the laser radar to scan the power transmission line environment, collect position and distance information, and obtain three-dimensional point cloud data; Establishing a three-dimensional point cloud model through the three-dimensional point cloud data; Use drones to inspect, take pictures, and update the three-dimensional point cloud model in real time for real-time monitoring; The distance in the three-dimensional point cloud model is monitored and a preset value is set. When the distance is lower than the preset value, an alarm is triggered.

2. The power transmission line anti-external damage monitoring and early warning method according to claim 1, characterized in that: In the step of establishing a three-dimensional point cloud model using the three-dimensional point cloud data: Preprocessing the three-dimensional point cloud data to obtain preprocessed point cloud data; Performing instance segmentation on the preprocessed point cloud data to obtain point cloud segmentation data; Performing point cloud feature extraction on the point cloud segmentation data to obtain point cloud feature data; Registering the point cloud feature data to obtain point cloud registration data; Three-dimensional reconstruction is performed based on the point cloud registration data to obtain a three-dimensional point cloud model.

3. The power transmission line anti-external damage monitoring and early warning method according to claim 2, characterized in that: In the step of preprocessing the three-dimensional point cloud data to obtain preprocessed point cloud data: Using a bilateral filtering method, determining a neighborhood range of each point in the three-dimensional point cloud data; According to the distance in the spatial domain and the similarity in the value range, the weight of each point in the neighborhood to the current point is calculated; Perform weighted averaging on the points in the neighborhood according to the weights to obtain the filtering result of the current point; Using a statistical filter method, calculating the mean and standard deviation of points in a neighborhood of the three-dimensional point cloud data; According to the set standard deviation multiple threshold, determine whether the current point is an outlier. If the distance from the current point to the neighborhood mean is greater than the set standard deviation multiple threshold, it will be regarded as an outlier and removed.

4. The power transmission line anti-external damage monitoring and early warning method according to claim 3, characterized in that: In the step of performing instance segmentation on the pre-processed point cloud data to obtain point cloud segmentation data: The point cloud feature image is recognized through the convolutional neural network to complete the preliminary category classification; Perform pixel-level segmentation on the image objects classified by categories to obtain pixel points; Each of the pixel points is classified into a specific object instance of the same category, and the precise outline and mask of each object are generated to help the automatic ranging algorithm to obtain the target position more accurately.

5. The power transmission line anti-external damage monitoring and early warning method according to claim 4, characterized in that: In the step of extracting point cloud features from the point cloud segmentation data to obtain point cloud feature data: Through the laser radar three-dimensional scanning technology, the depth information of each pixel in the image is obtained; The camera captures the light in the scene, which passes through the camera's small hole and is projected onto the imaging plane to form an image; After passing through the pinhole, the three-dimensional point is projected onto the image plane of the camera to form a two-dimensional pixel point; By means of a plurality of the three-dimensional points, two-dimensional pixel coordinates are formed, and the two-dimensional pixel coordinates represent the measured distance.

6. The power transmission line anti-external damage monitoring and early warning method according to claim 5, characterized in that: In the step of registering the point cloud feature data to obtain point cloud registration data: Use the SAC-IA coarse registration algorithm based on FPFH features for fast coarse registration; The coarsely registered data is finely registered using the ICP algorithm; A feature point matching algorithm is used to align obvious feature points in the transmission line, wherein the obvious feature points include tower bases and suspension points.

7. The power transmission line anti-external damage monitoring and early warning method according to claim 6, characterized in that: In the step of monitoring the distance in the three-dimensional point cloud model and setting a preset value, and issuing an alarm when the distance is lower than the preset value: Divide into multiple levels of distance preset values, the preset values ​​represent the safe distance between the power transmission line and the mechanical object, and each level of the preset value sends a different warning signal; When the distance displayed by the two-bit pixel coordinates is lower than the third-level preset value, an audible and visual alarm is issued to alert the staff; When the distance displayed by the two-dimensional pixel coordinates is lower than the secondary preset value, the on-site management personnel are notified to check; When the distance displayed by the two-bit pixel coordinates is lower than the first-level preset value, the on-site construction is stopped; Set up the remote APP to receive alarm information, video playback and remote real-time viewing.

8. A power transmission line anti-external damage monitoring and early warning device, using the power transmission line anti-external damage monitoring and early warning method as claimed in claim 7, characterized in that: The invention comprises an unmanned aerial vehicle, a high-performance processor, a dual-channel high-definition camera and a solar power supply module. The high-performance processor, the dual-channel high-definition camera and the solar power supply module are all arranged on the unmanned aerial vehicle.

Images