Power transmission line mechanical external damage risk identification method and system based on point cloud and RTK fusion

Abstract

The invention provides a power transmission line mechanical external breaking risk identification method and system based on point cloud and RTK fusion, and relates to the field of surveying and mapping. Coordinate conversion processing is carried out, and a structured point cloud voxel set is generated; a point cloud subset corresponding to a power transmission line area is separated from the point cloud region to generate a three-dimensional geometric skeleton line, a mechanical point cloud region belonging to construction machinery is identified, the shortest distance distribution and spatial azimuth angle relation between the mechanical point cloud region and the power transmission line geometric skeleton line is calculated, and a topological proximity relation descriptor is obtained; determining spatial position vectors, calculating penetration direction and penetration depth parameters of each position vector relative to the boundary of the protection area, and constructing spatial intrusion vector field distribution data; and carrying out vector direction consistency analysis, calculating a penetration depth parameter cumulative sum, and generating a mechanical external damage risk identification result when the penetration depth parameter cumulative sum exceeds a preset risk threshold condition. According to the invention, the accuracy of power transmission line mechanical external damage risk identification is improved.

Description

Technical Field

[0001] This invention relates to the field of surveying and data processing, and more specifically, to a method and system for identifying mechanical damage risks of transmission lines based on point cloud and RTK fusion. Background Technology

[0002] Currently, common methods for identifying mechanical damage risks to power transmission lines typically involve using lidar to collect point cloud data of the surrounding environment or RTK positioning devices to obtain the location information of construction machinery. The distance between the machinery and the line is then calculated based on the point cloud data, or the RTK position is used to determine whether the machinery has entered a pre-defined protection zone, thereby identifying any potential damage risk. However, relying solely on lidar point cloud data for distance calculations is susceptible to equipment installation errors and environmental interference, leading to insufficient accuracy in point cloud coordinates. Using RTK positioning alone makes it difficult to accurately obtain the three-dimensional spatial relationship between the machinery and the line. Furthermore, existing technologies lack comprehensive consideration of multi-dimensional features such as intrusion direction and depth when analyzing the spatial relationship between machinery and the line. These factors all affect the accuracy and reliability of mechanical damage risk identification. Therefore, improving the accuracy and comprehensiveness of mechanical damage risk identification for power transmission lines has become a technical problem that needs to be solved. Summary of the Invention

[0003] In view of this, the present invention provides a method and system for identifying mechanical external damage risks of transmission lines based on point cloud and RTK fusion.

[0004] According to one aspect of the present invention, a method for identifying mechanical external damage risks of transmission lines based on point cloud and RTK fusion is provided. The method includes: acquiring the three-dimensional position coordinates of the vehicle collected in real time by an RTK positioning device mounted on a construction vehicle, and local point cloud data of the surrounding environment collected by a lidar with the three-dimensional position coordinates as the center, to obtain a point cloud data sequence with spatiotemporal markers; performing coordinate transformation processing on the point cloud data sequence with spatiotemporal markers, transforming the local point cloud data collected by the lidar from the equipment coordinate system to the geographic coordinate system, and combining the three-dimensional position coordinates provided by the RTK positioning device as rigid constraints to perform registration optimization of multi-frame point cloud data, generating a structured point cloud voxel set with a unified geographic coordinate reference; separating the point cloud subset corresponding to the transmission line area from the structured point cloud voxel set, extracting the skeleton of the point cloud subset, generating the three-dimensional geometric skeleton line of the transmission conductor, and identifying the points cloud voxel set belonging to the transmission line area. For the mechanical point cloud region of construction machinery, the shortest distance distribution and spatial azimuth relationship between the mechanical point cloud region and the geometric skeleton line of the transmission line are calculated to obtain the topological proximity descriptor of the mechanical point cloud and the geometric skeleton line of the transmission line. Based on the topological proximity descriptor of the mechanical point cloud and the geometric skeleton line of the transmission line, the spatial position vector of each point in the mechanical point cloud region relative to the geometric skeleton line of the transmission line is determined. Combined with the preset spatial boundary of the transmission line protection zone, the penetration direction and penetration depth parameters of each position vector relative to the boundary of the protection zone are calculated to construct spatial intrusion vector field distribution data describing the intrusion state of the mechanical point cloud region into the protection zone. Vector direction consistency analysis is performed on the spatial intrusion vector field distribution data to extract vector clusters with the same penetration direction. The cumulative sum of the penetration depth parameters of the vector clusters is calculated. When the cumulative sum of the penetration depth parameters exceeds the preset risk threshold condition, a mechanical external damage risk identification result containing the coordinates of the intrusion area and the risk level is generated.

[0005] According to another aspect of the present invention, a computer system is provided, comprising: a processor; and a memory, wherein the memory stores computer-readable code, which, when executed by the processor, causes the processor to perform the method described above.

[0006] This invention acquires the RTK 3D position coordinates of construction vehicles and local point cloud data from LiDAR to form a spatiotemporally labeled point cloud data sequence. Through coordinate transformation, the point cloud is converted from the equipment coordinate system to the geographic coordinate system, and multi-frame registration optimization is performed using RTK coordinates as rigid constraints. This generates a structured point cloud voxel set with a unified geographic coordinate reference, effectively correcting equipment dynamic errors and multi-frame point cloud drift, and improving the global coordinate consistency and local structural feature preservation of the point cloud data. A subset of the transmission line area point cloud is separated from the structured point cloud voxel set, and a 3D geometric skeleton line is extracted. Simultaneously, the topological proximity relationship descriptor between the mechanical point cloud area and the data is calculated. The skeleton line simplifies complex point cloud data and characterizes spatial relationships from two dimensions: distance and azimuth, reducing subsequent computational complexity while enriching the data. Spatial relationship description dimension: Based on the topological proximity descriptor, the spatial position vector of the mechanical point cloud relative to the conductor skeleton is determined. Combined with the protection zone boundary, the penetration direction and penetration depth parameters are calculated to construct the spatial intrusion vector field distribution data, transforming discrete point cloud information into a continuous vector field description, realizing dynamic quantitative characterization of the mechanical intrusion state; directional consistency analysis is performed on the spatial intrusion vector field to extract vector clusters and calculate the cumulative sum of vector amplitudes. Cluster analysis avoids interference from isolated noise points and aggregates the overall intrusion characteristics, which are compared with risk thresholds to generate risk identification results. Thus, the progressive processing from data fusion to spatial modeling to intrusion quantification can comprehensively consider the spatial position relationship between machinery and transmission lines, dynamic intrusion trends, and overall intrusion scale, effectively improving the accuracy and reliability of identifying external mechanical damage risks to transmission lines. Attached Figure Description

[0007] Figure 1 This is a schematic diagram of an application scenario provided by the present invention; Figure 2 This is a flowchart illustrating a method for identifying mechanical external damage risks of transmission lines based on point cloud and RTK fusion, provided by the present invention. Figure 3 This is a schematic diagram of the structure of a computer system provided in an embodiment of the present invention. Detailed Implementation

[0008] To facilitate a clearer understanding of this invention, we first introduce the application scenarios of the method for identifying mechanical external damage risks of transmission lines based on point cloud and RTK fusion, such as... Figure 1 As shown, the application scenario of this invention includes a computer system 10 and a data acquisition cluster. Specifically, the data acquisition cluster may include an RTK positioning device and a lidar mounted on a construction vehicle. Both the RTK positioning device and the lidar can be network-connected to the computer system 10 to facilitate data interaction via the network.

[0009] It is understood that computer system 10 may refer to a device that executes the method for identifying mechanical external damage risks of transmission lines based on point cloud and RTK fusion provided in the embodiments of the present invention. The computer system 10 may be a background server or remote control device, or a laptop computer, desktop computer, etc., but is not limited thereto.

[0010] Further, please see Figure 2 This is a flowchart illustrating a method for identifying mechanical external damage risks of transmission lines based on point cloud and RTK fusion, provided by an embodiment of the present invention. Figure 2 As shown, this method can be derived from... Figure 1 The method for identifying mechanical external damage risks of transmission lines based on point cloud and RTK fusion is executed by computer system 10, and may include the following steps:

[0011] Step S100: Obtain the three-dimensional position coordinates of the vehicle in real time collected by the RTK positioning device on the construction vehicle, and the local point cloud data of the surrounding environment collected by the lidar with the three-dimensional position coordinates as the center, to obtain a point cloud data sequence with spatiotemporal markers.

[0012] RTK positioning devices, or Real-Time Dynamic Carrier Phase Differential Positioning devices, utilize the Global Navigation Satellite System (GNSS) to acquire the three-dimensional position coordinates (longitude, latitude, and elevation) of construction vehicles in a geographic coordinate system with high precision and real-time through real-time differential calculations between a base station and a rover. LiDAR, a sensor that obtains distance information to a target object by emitting a laser beam and measuring the time of reflection, scans the surrounding environment using the three-dimensional position coordinates provided by the RTK positioning device as the center, thereby collecting local point cloud data of the surrounding environment. Point cloud data is a collection of numerous discrete three-dimensional spatial points, each representing a sampling point on the surface of an object, recording the point's position information in three-dimensional space. Spatiotemporal markers are added to these point cloud data with temporal and spatial identifiers. The temporal identifier records the specific time of point cloud data acquisition, while the spatial identifier is associated with the three-dimensional position coordinates acquired by the RTK positioning device, enabling the point cloud data to correspond to specific time and spatial locations, resulting in a point cloud data sequence with spatiotemporal markers. For example, at a power transmission line construction site, construction vehicles are equipped with RTK positioning devices and lidar. During the vehicle's operation, the RTK positioning device collects the vehicle's three-dimensional position coordinates once per second, while the lidar scans around these coordinates to collect point cloud data of the surrounding environment. The lidar then marks these point cloud data with the collection time and the corresponding three-dimensional position coordinates of the vehicle, resulting in a series of point cloud data sequences with spatiotemporal markers.

[0013] Step S200: Perform coordinate transformation processing on the point cloud data sequence with spatiotemporal markers, transform the local point cloud data collected by the lidar from the device coordinate system to the geographic coordinate system, and combine the three-dimensional position coordinates provided by the RTK positioning device as rigid constraints to perform registration optimization of multi-frame point cloud data, and generate a structured point cloud voxel set with a unified geographic coordinate reference.

[0014] The device coordinate system is a coordinate system defined by the lidar itself. Its origin and coordinate axis directions are usually determined according to the installation and design of the lidar. The geographic coordinate system is a globally unified coordinate system used to describe the position of objects on Earth.

[0015] In one implementation, step S200 may specifically include the following steps S210 to S260: Step S210: Analyze the point cloud data sequence with spatiotemporal markers, extract the time synchronization error between the lidar and the RTK positioning device, calculate the timestamp deviation between adjacent sampling times, and generate a time synchronization error sequence as the time reference correction basis for coordinate transformation.

[0016] The spatiotemporally marked point cloud data sequence contains the time information of the point cloud data acquired by the LiDAR and the time information of the 3D position coordinates acquired by the RTK positioning device. Due to potential differences in the sampling frequency and clock precision of the LiDAR and RTK positioning devices, they are not perfectly synchronized in time, resulting in time synchronization errors. Parsing the spatiotemporally marked point cloud data sequence involves detailed analysis and processing of this data to extract the timestamp information of the LiDAR and RTK positioning devices. By comparing the LiDAR timestamps and RTK timestamps at adjacent sampling times, the deviation between them is calculated, resulting in a series of timestamp deviation values. These deviation values ​​constitute the time synchronization error sequence. This time synchronization error sequence serves as a crucial basis for time reference correction during subsequent coordinate transformation processes, ensuring the temporal consistency between the point cloud data and the RTK 3D position coordinates.

[0017] Step S220: Based on the time synchronization error sequence, the local point cloud data collected by the lidar is resampled with timestamps to ensure that the point cloud data is consistent with the time reference of the RTK three-dimensional position coordinates, and time-aligned point cloud data units are generated.

[0018] Timestamp resampling is the process of adjusting the timestamps of local point cloud data acquired by a LiDAR based on a time synchronization error sequence. By applying the time synchronization error sequence to the timestamps of the point cloud data, the timestamps of each point cloud data are corrected, ensuring that the time reference of the point cloud data is the same as the time reference of the RTK 3D position coordinates. For example, if the time synchronization error sequence shows that the LiDAR timestamp is 0.05 seconds faster than the RTK positioning device's timestamp, then during timestamp resampling, the timestamps of all point cloud data acquired by the LiDAR are subtracted by 0.05 seconds, thus achieving time alignment.

[0019] Step S230: Extract the dynamic error features of the lidar equipment, including attitude angle fluctuations and ranging drift caused by vehicle bumps. By analyzing the error distribution patterns of historical data, generate a set of dynamic error features of the equipment.

[0020] The lidar is mounted on a construction vehicle. During vehicle movement, bumps and vibrations cause fluctuations in the lidar's attitude angles (such as pitch, roll, and yaw), and may also cause ranging drift, meaning a deviation between the distance measured by the lidar and the actual distance. The equipment dynamic error characteristics quantify and describe these attitude angle fluctuations and ranging drift caused by vehicle movement. By analyzing historically collected point cloud data and corresponding vehicle motion information, the error distribution patterns of attitude angle fluctuations and ranging drift are statistically analyzed, such as the amplitude range and frequency distribution of attitude angle fluctuations, and the deviation range and trend of ranging drift. These error distribution patterns are then organized and summarized to generate a set of equipment dynamic error characteristics.

[0021] Step S240: Combine the dynamic error feature set of the equipment to perform error compensation processing on the time-aligned point cloud data units, correct the coordinate tilt caused by attitude angle fluctuation and the distance deviation caused by ranging drift, and obtain the intermediate point cloud data after error compensation.

[0022] The device dynamic error feature set includes error information related to attitude angle fluctuations and ranging drift caused by factors such as vehicle bumps in the lidar. This information is used to perform error compensation processing on time-aligned point cloud data units. Attitude angle fluctuations cause the coordinates of the point cloud data to tilt, resulting in deviations in the spatial position of the point cloud data. Ranging drift, on the other hand, makes the distance values ​​recorded in the point cloud data inaccurate. Error compensation processing corrects these deviations through a series of calculations and adjustments. For example, based on the amplitude and direction of the attitude angle fluctuations, the correction amount for coordinate tilt is calculated, and the coordinates of the time-aligned point cloud data units are rotated and translated to eliminate the influence of coordinate tilt. Based on the deviation value of ranging drift, the distance information of each point cloud data is adjusted to compensate for the distance deviation caused by ranging drift. After error compensation processing, the obtained point cloud data is more accurate in coordinates and distances, becoming the intermediate point cloud data after error compensation.

[0023] In one implementation, step S240 may specifically include the following steps S241 to S246: Step S241: Analyze the attitude angle fluctuations in the dynamic error feature set of the equipment, generate an attitude angle time series, decompose the periodic components in the sequence through Fourier transform, extract the characteristic frequency parameters of the fluctuation components, and determine the characteristic frequency parameters through the frequency values ​​corresponding to the peak values ​​of the spectrum.

[0024] The dynamic error feature set of the equipment contains information related to attitude angle fluctuations. Analyzing this information extracts the changes in attitude angles over time, resulting in an attitude angle time series. This one-dimensional time series records the attitude angle values ​​of the LiDAR at different times. Fourier transform is used to process the attitude angle time series, decomposing it into periodic components of different frequencies. For example, after performing a Fourier transform on the attitude angle time series, a spectrum is obtained. Several distinct peaks are observed in the spectrum; the frequencies corresponding to these peaks are the characteristic frequency parameters of the attitude angle fluctuations, such as 2Hz and 3Hz. These characteristic frequency parameters reflect the main frequency components of the attitude angle fluctuations.

[0025] Step S242: Based on the characteristic frequency parameter and the sampling frequency of the lidar, the attitude angle fluctuation error is divided into two types of components: when the characteristic frequency is greater than or equal to half of the sampling frequency, it is marked as the first type of fluctuation error component; when the characteristic frequency is less than half of the sampling frequency, it is marked as the second type of drift error component.

[0026] According to the Nyquist sampling theorem, the sampling frequency must be greater than twice the highest frequency of the signal to accurately recover the original signal. When the characteristic frequency is greater than or equal to half the sampling frequency of the lidar, this fluctuation error has a high frequency characteristic and is labeled as a Type I fluctuation error component; when the characteristic frequency is less than half the sampling frequency, the fluctuation error has a relatively low frequency and exhibits a certain degree of drift, and is labeled as a Type II drift error component. For example, if the lidar's sampling frequency is 10Hz, and the characteristic frequency of the attitude angle fluctuation is 6Hz, then this fluctuation error is labeled as a Type I fluctuation error component; if the characteristic frequency is 3Hz, it is labeled as a Type II drift error component. This classification method helps to employ different compensation methods for different types of errors.

[0027] Step S243: For the first type of fluctuation error component, the original error is canceled by constructing an inverse fluctuation sequence to generate the first type of error compensation amount. The amplitude of the compensation amount is equal to the peak value of the fluctuation error, and the phase difference is half a cycle.

[0028] For the first type of fluctuation error component, due to its high frequency characteristics, the original error can be offset by constructing an inverse fluctuation sequence. The inverse fluctuation sequence is a sequence with the same amplitude as the original fluctuation error sequence but a phase difference of half a period. When the inverse fluctuation sequence is added to the original error sequence, the fluctuations cancel each other out, thus achieving the purpose of error compensation. For example, if the first type of fluctuation error component is a sinusoidal wave with a peak value of A and a phase of φ, then the constructed inverse fluctuation sequence is a sinusoidal wave sequence with an amplitude of A and a phase of φ+π. Using this inverse fluctuation sequence as the first type of error compensation quantity and superimposing it with the original first type of fluctuation error component can, to a certain extent, eliminate the impact of this error on the time-aligned point cloud data units.

[0029] Step S244: For the second type of drift error component, the error sequence is smoothed by the sliding window averaging method. The window length is determined according to the reciprocal of the characteristic frequency. The second type of error compensation amount is generated, which is the average value of the error within the window.

[0030] For the second type of drift error component, due to its relatively low frequency and drift characteristics, the sliding window averaging method can effectively smooth the error sequence. The window length is determined based on the reciprocal of the characteristic frequency. A fixed-length window is slid across the error sequence, and the average error value within the window is calculated. This average value is used as the compensation value at the center point of the window. By sliding the window sequentially, the compensation values ​​for the entire error sequence are obtained. These compensation values ​​constitute the second type of error compensation. In this way, noise and fluctuations in the error sequence can be reduced, making the error sequence smoother and thus better compensating for the impact of the second type of drift error component on time-aligned point cloud data units.

[0031] Step S245: Integrate the first type of error compensation amount and the second type of error compensation amount to generate a comprehensive error compensation coefficient matrix. Perform coordinate transformation correction on the time-aligned point cloud data units. During the correction process, compensate for the coupling error of attitude angle fluctuation and ranging drift.

[0032] The first and second types of error compensation quantities are integrated and a comprehensive error compensation coefficient matrix is ​​generated through certain mathematical operations. This matrix contains comprehensive compensation information for attitude angle fluctuations and ranging drift errors. When performing coordinate transformation correction on time-aligned point cloud data units, the coordinates of each point cloud data unit are adjusted using the comprehensive error compensation coefficient matrix. Since attitude angle fluctuations and ranging drift may have a coupling effect, meaning their errors may affect each other, the comprehensive error compensation coefficient matrix can simultaneously consider this coupling error during the correction process, resulting in more accurate coordinate correction. For example, by multiplying the comprehensive error compensation coefficient matrix with the coordinate matrix of the time-aligned point cloud data unit using matrix multiplication, the corrected coordinates are obtained, achieving comprehensive compensation for attitude angle fluctuations and ranging drift errors.

[0033] Step S246: Compare the corrected point cloud data with the RTK 3D position coordinates, calculate the root mean square value of the coordinate residual, and determine the intermediate point cloud data after error compensation when the root mean square value is less than the nominal ranging error of the lidar.

[0034] The point cloud data, after coordinate transformation and correction, is spatially compared with the 3D position coordinates provided by the RTK positioning device. The difference between the coordinates of each point cloud data and the corresponding RTK 3D position coordinate is calculated to obtain the coordinate residual. The root mean square value of the coordinate residual is a statistical measure of these residuals, reflecting the overall error level between the corrected point cloud data and the RTK 3D position coordinates. The nominal ranging error of the lidar is the maximum permissible ranging error specified in the design and manufacturing process of the lidar. When the root mean square value of the coordinate residual is less than the nominal ranging error of the lidar, it indicates that the error of the point cloud data is within an acceptable range after error compensation processing. In this case, the corrected point cloud data is determined as the intermediate point cloud data after error compensation.

[0035] Step S250: Using RTK 3D position coordinates as rigid constraints, perform multi-frame registration on the intermediate point cloud data after error compensation. By analyzing the overlapping area characteristics of point clouds in adjacent frames, calculate the registration deviation correction coefficient to minimize the spatial position error of point clouds in multiple frames.

[0036] Multi-frame registration is the process of aligning and fusing multiple frames of point cloud data acquired at different times. Using RTK 3D position coordinates as a rigid constraint means that the 3D position coordinates provided by the RTK positioning device are fixed reference points during the registration process. The overlapping region characteristics of adjacent frame point clouds are analyzed, such as the density, normal vector, and color of the point clouds within the overlapping region. By comparing the differences of these characteristics between adjacent frames, a registration deviation correction coefficient is calculated. This coefficient is used to adjust the position and orientation of the adjacent frame point cloud data, making them more spatially aligned. Through continuous iterative calculation and adjustment, the spatial position error between multiple frame point cloud data is minimized, ensuring that the multiple frame point cloud data form a continuous and accurate whole in the geographic coordinate system. For example, if a deviation in the normal vector direction of the point clouds within the overlapping region is found in two adjacent frame point cloud data, a registration deviation correction coefficient is calculated, and the position and orientation of one frame are adjusted. The overlapping region characteristics are compared again, and this process is repeated until the spatial position error of the multiple frame point cloud data is minimized.

[0037] Step S260: Divide the registered point cloud data into voxels, count the point cloud density and normal vector distribution of each voxel unit, and generate a structured point cloud voxel set with a unified geographic coordinate reference. The spatial resolution of the voxel unit is dynamically adjusted according to the point cloud density.

[0038] Voxelization divides the registered point cloud data into small 3D voxel units, similar to dividing 3D space into small cubes. The point cloud density within each voxel unit (i.e., the number of points within the voxel) and the distribution of point cloud normal vectors (reflecting the orientation of the point cloud surface within the voxel) are statistically analyzed. By statistically analyzing the point cloud density and normal vector distribution of each voxel unit, a structured set of point cloud voxels with a unified geographic coordinate reference is generated. The spatial resolution of the voxel units is dynamically adjusted according to the point cloud density. When the point cloud density is high, the size of the voxel units is reduced to improve the ability to describe the details of the object's surface; when the point cloud density is low, the size of the voxel units is increased to reduce the amount of data and computational complexity.

[0039] Step S300: Separate the point cloud subset corresponding to the transmission line region from the structured point cloud voxel set, extract the skeleton of the point cloud subset, generate the three-dimensional geometric skeleton line of the transmission conductor, identify the mechanical point cloud region belonging to the construction machinery in the structured point cloud voxel set, calculate the shortest distance distribution and spatial azimuth relationship between the mechanical point cloud region and the geometric skeleton line of the transmission conductor, and obtain the topological proximity relationship descriptor between the mechanical point cloud and the geometric skeleton of the transmission conductor.

[0040] The structured point cloud voxel set contains point cloud information of the transmission line and its surrounding environment. It is necessary to separate the point cloud subset corresponding to the transmission line region. This can be achieved by analyzing point cloud characteristics, such as point cloud density, normal vector distribution, and spatial location, combined with the geometric features and distribution patterns of the transmission line. Skeleton extraction processes the separated transmission line point cloud subset. By analyzing the principal direction vector and curvature variation characteristics of the point cloud, a three-dimensional geometric skeleton line of the transmission conductor is fitted and generated. This skeleton line approximately represents the central axis of the transmission conductor. The mechanical point cloud regions belonging to construction machinery within the structured point cloud voxel set are identified. This can be done by analyzing the motion trajectory, velocity vector distribution, and geometric shape of the point cloud, marking point cloud regions that match the characteristics of construction machinery. The shortest distance distribution between the mechanical point cloud region and the geometric skeleton line of the transmission conductor is calculated; that is, the shortest distance from each point within the mechanical point cloud region to the geometric skeleton line of the transmission conductor. The distribution patterns of these distances are statistically analyzed. Simultaneously, the spatial azimuth relationship between the mechanical point cloud region and the geometric skeleton line of the transmission conductor is calculated, reflecting their relative position and direction in space. By combining these shortest distance distributions and spatial azimuth relationships, a topological proximity descriptor for the geometric skeleton of the mechanical point cloud and the transmission line is obtained. This descriptor is used to describe the spatial topological relationship between the mechanical point cloud region and the transmission line.

[0041] In one implementation, step S300 may specifically include the following steps S310 to S360: Step S310: Perform feature enhancement on the structured point cloud voxel set, extract the point cloud density, normal vector direction and time dynamic change rate of each voxel unit, and generate a multimodal voxel feature set.

[0042] Feature enhancement involves processing structured point cloud voxel sets to highlight key features for better subsequent analysis and recognition. The point cloud density of each voxel unit is extracted, representing the number of points within that voxel, reflecting the density of object distribution within that region. The normal vector direction reflects the orientation of the point cloud surface within the voxel, crucial for identifying object shape and surface features. The temporal dynamic change rate considers the changes in point cloud data at different times, such as point cloud movement and deformation. These features are combined to generate a multimodal voxel feature set. This multimodal voxel feature set contains feature information from multiple aspects of the point cloud data, providing richer evidence for region segmentation and target recognition.

[0043] Step S320: Based on the multimodal voxel feature set, the structured point cloud voxel set is initially segmented using a region growing algorithm. Potential transmission line areas are marked by analyzing the density continuity and normal vector consistency of voxel units.

[0044] In one implementation, step S320 may specifically include the following steps S321 to S326: Step S321: Analyze the multimodal voxel feature set, extract the point cloud density value and normal vector direction angle of the voxel unit, and generate the density-normal vector feature set.

[0045] Parsing a multimodal voxel feature set involves extracting and processing the various feature information stored within it. The point cloud density value and normal vector orientation angle of each voxel unit are extracted from the multimodal voxel feature set, and these two features are combined to generate a density-normal vector feature set. The density-normal vector feature set is a two-dimensional feature set, where each element contains the point cloud density value and normal vector orientation angle information of a voxel unit, providing more specific feature basis for region segmentation. For example, in the multimodal voxel feature set corresponding to a structured point cloud voxel set, the point cloud density value and normal vector orientation angle are extracted for each voxel unit, and these values ​​are stored in a two-dimensional array to obtain the density-normal vector feature set.

[0046] Step S322: Calculate the density gradient and normal angle of adjacent voxel units in the density-normal vector feature set. When the density gradient is within a preset range and the normal angle is within a preset range, they are determined to be adjacent voxels with continuity.

[0047] Density gradient refers to the rate of change of point cloud density between adjacent voxel units, obtained by calculating the ratio of the density difference between adjacent voxel units to the distance between them. Normal vector angle refers to the angle between the normal vectors of adjacent voxel units, calculated using the vector dot product formula. The preset range is a threshold pre-set based on the characteristics and actual conditions of the transmission line. When the density gradient and normal vector angle of adjacent voxel units are both within the preset range, it indicates that these two adjacent voxel units have similar density and normal vector characteristics, and they are determined to be continuous adjacent voxel units.

[0048] Step S323: Initialize the region growth seed point. Select voxel units whose point cloud density values ​​conform to the characteristics of the transmission line and whose normal vector direction angles conform to the characteristics of the transmission line as the initial seed point and add them to the region growth queue.

[0049] Transmission lines typically possess certain point cloud density and orientation characteristics. Based on these characteristics, the ranges for point cloud density values ​​and normal vector direction angles are pre-defined. From the structured point cloud voxel set, voxel units with point cloud density values ​​within the preset range and normal vector direction angles conforming to the transmission line orientation characteristics are selected and used as initial seed points. A region growth queue is used to store voxel units to be processed. The initial seed points are added to the region growth queue as the starting point for region growth.

[0050] Step S324: Take the current voxel unit from the region growth queue, traverse its eight neighboring voxel units, and mark the adjacent voxel units that meet the conditions as the same region based on the density continuity and normal vector consistency determination results, and add them to the region growth queue.

[0051] A voxel is selected from the region growth queue as the current voxel. Its eight neighbors refer to the eight voxel units adjacent to the current voxel in 3D space. These eight adjacent voxel units are traversed, and based on the density continuity and normal vector consistency determination results calculated in step S322, it is determined whether each adjacent voxel unit meets the conditions. If the conditions are met, the adjacent voxel unit is marked as belonging to the same region as the current voxel unit and added to the region growth queue for further processing of its adjacent voxel units. For example, a voxel unit is selected from the region growth queue, and its eight neighbors are traversed. It is found that three of the adjacent voxel units meet the density continuity and normal vector consistency conditions. These three adjacent voxel units are marked as belonging to the same region as the current voxel unit and added to the region growth queue.

[0052] Step S325: Repeat the region growth process until all voxel units in the queue have been processed, generating a region point cloud with multiple connected component labels.

[0053] Voxel units are continuously retrieved from the region growth queue and processed according to the method in step S324. Adjacent voxel units that meet the conditions are marked as belonging to the same region and added to the queue, until all voxel units in the region growth queue have been processed. During this process, multiple connected components composed of adjacent voxel units are formed, each of which is marked as a region, ultimately generating a region point cloud with multiple connected component labels. For example, after multiple iterations, the region growth queue gradually becomes empty, at which point multiple connected components with different labels are generated, each corresponding to a region point cloud.

[0054] Step S326: Filter connected regions in the point cloud with volumes within a preset range, retain connected regions whose point cloud density conforms to the characteristics of transmission lines and whose normal vector directions are consistent, and mark them as potential transmission line regions. The connected region markings of potential transmission line regions are consistent with the markings of seed points.

[0055] The preset range is a volume threshold pre-set based on the actual conditions of the transmission line. Connected regions in the region's point cloud with volumes within this preset range are filtered out, removing those that are too small or too large, as these may be noise or other non-transmission line areas. Connected regions containing point cloud densities consistent with transmission line characteristics and with aligned normal vector directions are retained and marked as potential transmission line regions. To maintain consistency in region labeling, the connected region labels for potential transmission line regions are consistent with the labels for the initial seed points. For example, if the preset volume range is 10-100 voxels, connected regions with volumes less than 10 or greater than 100 voxels in the region's point cloud are filtered out, retaining those with point cloud densities consistent with transmission line characteristics and aligned normal vector directions, and these are marked as potential transmission line regions with the same labels as the initial seed points.

[0056] Step S330: Extract tower point cloud features from potential transmission line areas, analyze point cloud distribution features based on the cylindrical geometry of the towers, detect the tower center point coordinates and radius parameters, and generate tower point cloud marking information.

[0057] Towers are an important component of power transmission lines, typically exhibiting a cylindrical geometry. This study extracts point cloud features from potential transmission line areas and analyzes their spatial distribution. Due to the cylindrical geometry of the towers, their point cloud distribution follows certain patterns, such as a circular distribution on a horizontal cross-section. By analyzing these point cloud distribution features, fitting algorithms (such as least squares) are used to fit the center point coordinates and radius parameters of the towers. These center point coordinates, radius parameters, and corresponding point cloud region information are combined to generate tower point cloud labeling information, used to identify the tower's position and geometric features within the point cloud data. For example, in the point cloud data of potential transmission line areas, some point clouds exhibit a circular distribution. The center point coordinates and radius parameters of the towers represented by these point clouds are fitted using least squares, and this information is stored in a data structure to obtain the tower point cloud labeling information.

[0058] Step S340: Based on the tower point cloud marking information, perform secondary segmentation on the potential transmission line area. By analyzing the point cloud connectivity and trend between towers, separate the point cloud subset corresponding to the transmission conductor and remove the non-conductor point clouds of towers and insulators.

[0059] Based on the tower point cloud marking information, the location and geometric features of the towers are known. The connectivity of the point clouds between towers is analyzed, i.e., determining which point clouds connect different towers and their directional trends. Transmission lines are typically connected between towers along a specific direction. Based on this information, point clouds connected between towers and whose directional trends match those of transmission lines are separated, resulting in a subset of point clouds corresponding to the transmission lines. Simultaneously, non-conductor point clouds of towers and insulators are removed, as these may interfere with the analysis and identification of transmission lines. For example, in the point cloud data of potential transmission line areas, based on the tower point cloud marking information, some point clouds are found to connect different towers and whose directional trends are consistent with those of the transmission lines. These point clouds are separated as the subset of point clouds corresponding to the transmission lines, while the point clouds of towers and insulators are removed.

[0060] Step S350: Extract the skeleton from the point cloud subset of the transmission line. By analyzing the principal direction vector and curvature change characteristics of the point cloud, a three-dimensional geometric skeleton line of the transmission line is generated. The node density of the skeleton line is dynamically adjusted according to the curvature of the transmission line.

[0061] Skeleton extraction is the process of simplifying a subset of transmission line point clouds into a three-dimensional geometric skeleton line representing its central axis. The principal direction vector of the point cloud is analyzed; this vector reflects the main distribution direction of the point cloud in space and is calculated using methods such as principal component analysis. Curvature variation characteristics reflect the degree of curvature of the transmission line and are obtained by calculating the curvature of the point cloud. Based on the principal direction vector and curvature variation characteristics, a fitting algorithm (such as spline fitting) is used to generate the three-dimensional geometric skeleton line of the transmission line. The node density of the skeleton line is dynamically adjusted according to the conductor curvature. When the conductor curvature is large, the node density is increased to more accurately represent the bending shape of the conductor; when the conductor curvature is small, the node density is decreased to reduce the amount of data. For example, in a subset of transmission line point clouds, the principal direction vector of the point cloud is calculated using principal component analysis, the curvature of the point cloud is calculated, a spline fitting algorithm is used to generate a three-dimensional geometric skeleton line, and the node density of the skeleton line is adjusted according to the conductor curvature so that the skeleton line can accurately represent the shape of the transmission line.

[0062] Step S360: Identify dynamic point cloud regions in the structured point cloud voxel set, excluding the transmission line region. By analyzing the motion trajectory and velocity vector distribution of the point cloud, mark the mechanical point cloud regions belonging to the construction machinery. Calculate the shortest distance distribution and spatial azimuth relationship between the mechanical point cloud regions and the geometric skeleton line of the transmission conductor. Generate a topological proximity descriptor between the mechanical point cloud and the geometric skeleton of the transmission conductor.

[0063] A dynamic point cloud region refers to an area where the position of the point cloud changes at different times, determined by analyzing the motion trajectory and velocity vector distribution of the point cloud. The motion trajectory refers to the path of position change of the point cloud at different times, and the velocity vector refers to the direction and magnitude of displacement of the point cloud per unit time. The motion trajectory and velocity vector of the point cloud are calculated by comparing and analyzing point cloud data at different times. Dynamic point cloud regions with stable motion trends and geometric shapes conforming to the characteristics of construction machinery are marked as mechanical point cloud regions. The shortest distance from each point within the mechanical point cloud region to the geometric skeleton line of the transmission line is calculated, and the distribution pattern of these distances is statistically analyzed. Simultaneously, the spatial azimuth relationship between the mechanical point cloud region and the geometric skeleton line of the transmission line is calculated. These shortest distance distributions and spatial azimuth relationships are combined to generate a topological proximity descriptor for the mechanical point cloud and the geometric skeleton line of the transmission line.

[0064] In one implementation, step S360 may specifically include the following steps S361 to S366: Step S361: Extract the remaining point cloud data that is not marked as a transmission line area from the structured point cloud voxel set, divide it into continuous time window point cloud subsets according to the timestamp order, and dynamically set the duration of each time window according to the sampling frequency of the lidar.

[0065] Remaining point cloud data not marked as transmission line areas is extracted from the structured point cloud voxel set. This point cloud data may include construction machinery, other dynamic objects, etc. These remaining point cloud data are divided into consecutive time window subsets according to timestamp order. Each time window contains point cloud data collected within a certain time period. The duration of the time window is dynamically set according to the LiDAR sampling frequency. For example, if the LiDAR sampling frequency is 10Hz, i.e., 10 frames of point cloud data are collected per second, the time window can be set to 0.5 seconds, with each time window containing 5 frames of point cloud data. This division facilitates subsequent analysis of the point cloud's motion trajectory and velocity vector distribution.

[0066] Step S362: Calculate the centroid coordinates of the point cloud subset for each time window. By analyzing the changes in the centroid coordinates of adjacent time windows, generate the motion trajectory and velocity vector of the point cloud region. The magnitude of the velocity vector is the distance the centroid moves per unit time, and the direction is the direction of the centroid movement.

[0067] For each time window subset of the point cloud, calculate its centroid coordinates. The centroid coordinates are the average of the coordinates of all points within the subset, obtained by summing the coordinates of each point and dividing by the number of points. Analyze the changes in centroid coordinates between adjacent time windows, and calculate the displacement of the centroid between adjacent time windows. The magnitude and direction of the displacement correspond to the motion trajectory and velocity vector of the point cloud region, respectively. The magnitude of the velocity vector is the distance the centroid moves per unit time, i.e., the magnitude of the displacement of the centroid coordinates between adjacent time windows divided by the duration of the time window; the direction of the velocity vector is the direction of the centroid's movement. For example, in two adjacent 0.5-second time window subsets of the point cloud, if the centroid coordinates of the first time window are (x1, y1, z1) and the centroid coordinates of the second time window are (x2, y2, z2), then the displacement of the centroid is (x2-x1, y2-y1, z2-z1), the magnitude of the velocity vector is the magnitude of the displacement divided by 0.5 seconds, and the direction is the direction of the displacement.

[0068] Step S363: Perform continuity analysis on the motion trajectory and velocity vector. When the velocity vector direction changes within a preset range for multiple consecutive time windows, it is determined to be a dynamic point cloud region with a stable motion trend.

[0069] Continuity analysis evaluates motion trajectories and velocity vectors to determine the stability of motion within a point cloud region. The preset range is a threshold value for velocity vector direction changes set in advance based on actual conditions. When the velocity vector direction changes within the preset range for multiple consecutive time windows, it indicates that the motion direction of the point cloud region is relatively stable, and the region is determined to be a dynamic point cloud region with a stable motion trend.

[0070] Step S364: Extract the geometric features of the dynamic point cloud region, including the size, volume and surface curvature distribution of the bounding rectangle, and match them with the preset construction machinery morphology feature library to select dynamic point cloud regions that conform to the mechanical morphology features.

[0071] The geometric features of the dynamic point cloud region are extracted. The bounding rectangle size refers to the length, width, and height of the smallest rectangle that can enclose the dynamic point cloud region; the volume refers to the three-dimensional spatial volume occupied by the dynamic point cloud region; and the surface curvature distribution reflects the degree of curvature of the surface of the dynamic point cloud region. A pre-set construction machinery morphological feature library is a database storing the geometric features of various construction machinery. The extracted geometric features of the dynamic point cloud region are matched with the features in the construction machinery morphological feature library. By calculating the similarity between features, dynamic point cloud regions that match the mechanical morphological characteristics are selected.

[0072] Step S365: Cluster the selected dynamic point cloud regions, merge point cloud units with spatial distances within a preset range into the same mechanical point cloud region, and generate mechanical point cloud region labeling information.

[0073] Clustering is performed on the selected dynamic point cloud regions that conform to mechanical morphological characteristics. Clustering is the process of merging point cloud units with similar characteristics into a single region. The preset range is a spatial distance threshold pre-set based on actual conditions. The spatial distance between each point cloud unit is calculated, and when the spatial distance between two point cloud units is within the preset range, they are merged into the same mechanical point cloud region. Labeling information is generated for each mechanical point cloud region. The labeling information may include the region's number, location, size, etc., for subsequent analysis and identification. For example, in the selected dynamic point cloud regions, the spatial distance between point cloud units is calculated. If some point cloud units are found to be less than the preset distance of 1 meter, these point cloud units are merged into a single mechanical point cloud region, and labeling information is generated for that region.

[0074] Step S366: For each mechanical point cloud region, calculate the shortest distance and spatial azimuth angle from each point in the region to the geometric skeleton line of the transmission line, statistically analyze the distance distribution pattern and azimuth angle variation trend, and generate a topological proximity descriptor between the mechanical point cloud and the geometric skeleton of the transmission line.

[0075] For each mechanical point cloud region, the shortest distance from each point within the region to the geometric skeleton line of the transmission line is calculated using an algorithm for calculating the shortest distance from a point to a curve. Simultaneously, the spatial azimuth angle of each point relative to the geometric skeleton line of the transmission line is calculated. The spatial azimuth angle reflects the point's position and orientation in space relative to the transmission line. The distribution patterns of these shortest distances are statistically analyzed, including calculating the average and standard deviation of the distances, as well as the trends in the spatial azimuth angle, such as the range and frequency of change. These shortest distance distribution patterns and spatial azimuth angle trends are combined to generate a topological proximity descriptor for the mechanical point cloud and the geometric skeleton line of the transmission line. This descriptor describes the spatial topological relationship between the mechanical point cloud region and the transmission line, providing a basis for subsequent assessment of whether the mechanical point cloud region poses a threat to the transmission line. For example, within a mechanical point cloud region, the shortest distances from each point within the region to the geometric skeleton line of the transmission line are calculated, the distribution patterns of these distances are statistically analyzed, and the spatial azimuth angle of each point is calculated and its trends analyzed. This information is then combined to form a topological proximity descriptor.

[0076] Step S400: Based on the topological proximity descriptor of the mechanical point cloud and the geometric skeleton of the transmission line, determine the spatial position vector of each point in the mechanical point cloud region relative to the geometric skeleton line of the transmission line. Combined with the preset spatial range boundary of the transmission line protection zone, calculate the penetration direction and penetration depth parameters of each position vector relative to the boundary of the protection zone, and construct spatial intrusion vector field distribution data describing the intrusion state of the mechanical point cloud region into the protection zone.

[0077] In one implementation, step S400 may specifically include the following steps S410 to S460: Step S410: parse the topological proximity descriptor, extract the shortest distance value and spatial azimuth parameter between each point in the mechanical point cloud region and the geometric skeleton line of the transmission line, and generate a distance-azimuth feature set.

[0078] Parsing the topological proximity descriptor is a process of detailed analysis of its internally stored information. The topological proximity descriptor records rich spatial relationship information between the mechanical point cloud region and the geometric skeleton line of the transmission line. Specifically, the shortest distance value between each point in the mechanical point cloud region and the geometric skeleton line of the transmission line reflects the nearest spatial interval between that point and the transmission line, while the spatial azimuth parameter indicates the point's directional position relative to the geometric skeleton line of the transmission line. Through parsing the topological proximity descriptor, this key information is extracted and integrated into a set, resulting in a distance-azimuth feature set. This set stores the distance and azimuth information of each point in a structured manner, providing a direct and accurate data foundation for subsequent determination of spatial location vectors.

[0079] Step S420: Based on the distance-azimuth feature set, determine the spatial position vector of each point in the mechanical point cloud region relative to the geometric skeleton line of the transmission line. The starting point of the vector is the projection point of the point on the skeleton line, the ending point is the mechanical point cloud coordinates, the vector direction is from the projection point to the mechanical point cloud coordinates, and the magnitude is equal to the shortest distance value.

[0080] The distance-azimuth feature set provides the necessary parameters for determining the spatial position vector. For each point in the mechanical point cloud region, its projection point on the geometric skeleton line of the transmission line needs to be found first. The projection point is the point on the skeleton line closest to the mechanical point cloud coordinates, which can be calculated through spatial geometric relationships and related algorithms. The vector connecting this projection point as the starting point and the mechanical point cloud coordinates as the ending point is the spatial position vector. Its direction is clearly from the projection point to the mechanical point cloud coordinates, and its magnitude is exactly equal to the shortest distance between this point and the geometric skeleton line of the transmission line, which is obtained from the distance-azimuth feature set. In this way, the spatial position vector relative to the geometric skeleton line of the transmission line is determined for each point in the mechanical point cloud region, thus accurately describing the spatial positional relationship of each point relative to the transmission line.

[0081] In one implementation, step S420 may specifically include the following steps S421 to S426: Step S421: Analyze the three-dimensional parameter description of the geometric skeleton line of the transmission conductor. For each point cloud coordinate in the mechanical point cloud region, calculate its projection point coordinates on the skeleton line. The projection point is the point on the skeleton line that is closest to the mechanical point cloud coordinates.

[0082] The three-dimensional parameter description of the geometric skeleton line of a transmission conductor contains specific information about the skeleton line in three-dimensional space, such as the curve equation and key point coordinates. Analyzing these parameter descriptions is crucial for accurately finding the projection point on the skeleton line that is closest to each point in the mechanical point cloud region. For each point cloud coordinate in the mechanical point cloud region, spatial geometric calculations and optimization algorithms are used to continuously search for points on the skeleton line that minimize the distance between that point and the mechanical point cloud coordinates; this point is the projection point. For example, an iterative search method can be used, starting from the key points of the skeleton line and gradually expanding the search outwards, continuously comparing distances to ultimately determine the coordinates of the projection point.

[0083] Step S422: Based on the projection point coordinates and the mechanical point cloud coordinates, calculate the unit direction vector of the skeleton line pointing to the mechanical point cloud coordinates. The direction of the unit direction vector is from the projection point to the mechanical point cloud coordinates, and the module length is standardized to a unified benchmark.

[0084] After obtaining the coordinates of the projection point and the mechanical point cloud, the vector pointing from the projection point to the mechanical point cloud coordinates is calculated. To facilitate subsequent calculations and comparisons, this vector is converted into a unit direction vector. The direction of the unit direction vector remains consistent with the original vector, i.e., from the projection point to the mechanical point cloud coordinates, while its magnitude is standardized to a uniform reference value, typically 1. This process can be achieved by dividing the original vector by its magnitude. The unit direction vector accurately represents the directional information of the mechanical point cloud coordinates relative to the projection point, providing a directional basis for the subsequent generation of preliminary spatial position vectors. For example, in transmission line monitoring, for a point in the mechanical point cloud region, a unit direction vector is calculated based on its projection point coordinates and its own coordinates. This vector clearly indicates the direction of that point relative to the geometric skeleton line of the transmission conductor.

[0085] Step S423: Extract the shortest distance value of the point from the distance-azimuth feature set, and combine the unit direction vector with the shortest distance value to generate a preliminary spatial position vector. The starting point of the preliminary vector is the coordinates of the projection point, and the ending point is the coordinates of the mechanical point cloud.

[0086] The shortest distance between the point and the geometric skeleton line of the transmission line is obtained from the distance-azimuth feature set. This shortest distance value is then combined with the previously calculated unit direction vector, specifically by multiplying the unit direction vector by the shortest distance value, to obtain a vector with actual distance meaning. This vector starts at the coordinates of the projection point and ends at the coordinates of the mechanical point cloud, thus forming the preliminary spatial position vector. By integrating direction and distance information into a single vector, the spatial positional relationship of each point in the mechanical point cloud region relative to the geometric skeleton line of the transmission line can be more intuitively represented.

[0087] Step S424: Analyze the local curvature of the geometric skeleton line of the transmission conductor. When the curvature is within the preset range, correct the direction of the initial spatial position vector so that the vector direction is always perpendicular to the tangent direction of the skeleton line at the projection point, and compensate for the direction deviation caused by the bending of the conductor.

[0088] The geometric skeleton line of a transmission line may exhibit different curvatures at different locations, with local curvature reflecting the degree of bending of the skeleton line in a specific local area. Analyzing the local curvature of the skeleton line can be achieved by calculating the rate of change of the curve near the projection point. The preset range is a curvature threshold pre-set based on the actual conditions and accuracy requirements of the transmission line. When the local curvature is within the preset range, it indicates that the bending of the conductor has a significant impact on the direction of the initial spatial position vector, requiring direction correction. The goal of the correction is to ensure that the direction of the initial spatial position vector is always perpendicular to the tangent direction of the skeleton line at the projection point. This can be achieved through vector operations and geometric transformations, such as calculating the tangent direction vector and then adjusting the direction of the initial spatial position vector based on the perpendicular relationship. This correction compensates for the directional deviation caused by conductor bending, allowing the spatial position vector to more accurately reflect the actual positional relationship of each point in the mechanical point cloud region relative to the transmission line. For example, in areas where the transmission line has a certain degree of curvature, analyzing the local curvature of the skeleton line and performing direction correction can more accurately determine the spatial position vector of points in the mechanical point cloud region.

[0089] Step S425: Calculate the overall motion trend vector of the mechanical point cloud region, and perform a weighted average of the spatial position vectors of each point in the region. The weight values ​​are dynamically adjusted according to the distance from the point cloud to the center of gravity of the region to enhance the consistency of the region's motion direction.

[0090] The overall motion trend vector of the mechanical point cloud region reflects the direction and velocity of motion of the region as a whole. First, the centroid of the mechanical point cloud region is calculated; it is the weighted average of the coordinates of all point clouds within the region. Then, for each point within the region, its weight is dynamically adjusted based on its distance from the centroid—points closer to the centroid have a higher weight, and points farther away have a lower weight. A weighted average is then calculated by multiplying each spatial position vector by its corresponding weight, summing the results, and dividing by the total weight to obtain the overall motion trend vector of the mechanical point cloud region. This weighted averaging method strengthens the consistency of the region's motion direction and reduces the impact of abnormal movements at individual points on the overall motion trend.

[0091] Step S426: Standardize the corrected spatial position vector to ensure that the error between the vector magnitude and the shortest distance value is within a preset range, and generate standardized spatial position vectors for each point in the mechanical point cloud region.

[0092] The corrected spatial position vector may have some error in its modulus compared to the shortest distance value. To ensure the accuracy of the vector modulus, the corrected spatial position vector is standardized. The preset range is an error threshold pre-set according to the accuracy requirements. By adjusting the vector modulus, the error between it and the shortest distance value is made within the preset range. For example, a scaling method can be used, multiplying the vector modulus by an adjustment coefficient to make the error meet the requirements. After standardization, standardized spatial position vectors are generated for each point in the mechanical point cloud region. These vectors can more accurately represent the spatial positional relationship of each point relative to the geometric skeleton line of the transmission line, providing an accurate data basis for subsequent calculations of penetration direction and penetration depth parameters. In practical applications, for mechanical point cloud regions generated by construction machinery near transmission lines, standardizing the corrected spatial position vectors can improve the accuracy of the analysis of the relationship between the mechanical point cloud region and the transmission line.

[0093] Step S430: Extract the real-time deformation characteristics of the transmission conductor, including the conductor swing amplitude and bending curvature caused by environmental factors. By analyzing historical deformation data, generate dynamic deformation parameters of the conductor.

[0094] During actual operation, transmission lines are affected by various environmental factors, such as wind and temperature changes, resulting in real-time deformation. Extracting these real-time deformation characteristics requires monitoring and analyzing the conductor's sway amplitude and curvature. Sway amplitude reflects the range of sway in the vertical and horizontal directions, while curvature reflects the degree of bending. Sensors installed on the transmission lines, such as displacement sensors and strain sensors, collect deformation data in real time. Simultaneously, historical deformation data is analyzed. This historical data contains the conductor's deformation under different environmental conditions. Statistical analysis and modeling of this data can reveal patterns and trends in deformation. Based on these analytical results, dynamic deformation parameters of the conductor are generated. These parameters describe the real-time deformation of the conductor under different environmental conditions, providing a basis for subsequent adjustments to the pre-defined spatial boundaries of the transmission line protection zone. For example, long-term transmission line monitoring has accumulated a large amount of historical conductor deformation data. Analyzing this data allows for accurate prediction of the conductor's dynamic deformation parameters under current environmental conditions.

[0095] Step S440: Combine the dynamic deformation parameters of the conductor to dynamically adjust the preset spatial range boundary of the transmission line protection zone, correct the boundary offset of the protection zone caused by conductor deformation, and generate dynamic protection zone boundary parameters.

[0096] The preset spatial boundary of the transmission line protection zone is based on the conductor under normal conditions. However, when the conductor deforms, the protection zone boundary needs to be adjusted accordingly. The preset spatial boundary is dynamically adjusted by incorporating conductor dynamic deformation parameters. These parameters include information such as the conductor's sway amplitude and curvature. Based on this information, the offset of the protection zone boundary caused by conductor deformation is calculated. For circular protection zones, the boundary radius may need adjustment; for non-circular protection zones, the boundary curve equation or key point coordinates need adjustment. During the adjustment process, the protection zone boundary always includes the geometric skeleton of the transmission conductor to ensure the safety of the transmission line. By correcting and updating the boundary parameters, dynamic protection zone boundary parameters are generated. These parameters reflect the adjusted protection zone boundary situation due to conductor deformation in real time, providing accurate boundary information for subsequent computational machine point cloud regions to assess encroachment on the protection zone.

[0097] In one implementation, step S440 may specifically include the following steps S441 to S446: Step S441: Analyze the dynamic deformation parameters of the conductor, extract the swing amplitude and bending curvature of the conductor, and generate a deformation feature vector. The composition of the deformation feature vector is dynamically set according to the number of deformation monitoring points of the conductor.

[0098] The dynamic deformation parameters of a conductor contain various information about the conductor during real-time deformation. Analyzing these parameters is crucial for extracting key deformation features, namely the conductor's swing amplitude and curvature. The swing amplitude reflects the degree of swaying in different directions, while the curvature reflects the conductor's bending condition. The composition of the deformation feature vector is dynamically set based on the number of deformation monitoring points. If there are many monitoring points, the deformation feature vector can contain more information, such as the swing amplitude and curvature of each monitoring point; if there are few monitoring points, the deformation feature vector can be simplified. For example, for a transmission conductor with multiple monitoring points, the deformation feature vector can be a multi-dimensional vector containing the swing amplitude and curvature of each monitoring point. By analyzing the conductor's dynamic deformation parameters and generating the deformation feature vector, specific characteristic information is provided for subsequent calculations of the conductor's deformation offset.

[0099] Step S442: Based on the deformation feature vector, calculate the deformation offset of the conductor at each monitoring point. The offset is the difference between the current coordinates and the initial coordinates of the monitoring point, and generate a set of deformation offsets.

[0100] The deformation feature vector provides information such as the sway amplitude and curvature of the conductor at each monitoring point. Based on this information, the deformation offset of the conductor at each monitoring point is calculated. The deformation offset is the difference between the current coordinates and the initial coordinates of the monitoring point. The initial coordinates are the coordinates of the monitoring point when the conductor is not deformed, and the current coordinates are the coordinates of the monitoring point under real-time deformation. By comparing and calculating the coordinates of each monitoring point, the corresponding deformation offset is obtained. The deformation offsets of all monitoring points are collected to generate a deformation offset set. This set contains the deformation of the conductor at each monitoring point, providing important data support for subsequent correction of the preset spatial boundary of the transmission line protection zone. For example, in real-time monitoring of transmission lines, for each monitoring point, its deformation offset is calculated based on the deformation feature vector, and these offsets are integrated into the deformation offset set.

[0101] Step S443: Substitute the set of deformation offsets into the preset spatial range boundary description of the transmission line protection zone, and correct the center coordinates and radius parameters (for circular protection zones) or other boundary parameters (for non-circular protection zones) in the boundary description. The center coordinate correction is equal to the deformation offset, and the radius parameter correction is dynamically adjusted according to the swing amplitude.

[0102] The pre-defined spatial boundary description of a transmission line protection zone includes the center coordinates, radius parameters (for circular protection zones), or other boundary parameters (for non-circular protection zones). The boundary parameters are corrected by substituting the set of deformation offsets into these boundary descriptions. For the center coordinates, the correction is equal to the deformation offset of the corresponding monitoring point, i.e., moving the center position of the boundary according to the conductor's deformation. For the radius parameter of a circular protection zone, it is dynamically adjusted according to the conductor's sway amplitude; the larger the sway amplitude, the larger the radius parameter correction, ensuring the protection zone completely covers the conductor's deformation range. For non-circular protection zones, other boundary parameters, such as the coefficients of the curve equation and the coordinates of key points, are adjusted based on the deformation offset and the boundary geometry. This correction allows the protection zone boundary to adapt to the real-time deformation of the conductor. For example, when the transmission line sways and bends, substituting the set of deformation offsets into the pre-defined spatial boundary description and correcting the center coordinates and radius parameters allows the protection zone boundary to better protect the transmission line.

[0103] Step S444: Perform a validity check on the revised boundary description to ensure that the boundary of the protected area always includes the geometric skeleton line of the transmission line. If the check fails, readjust the radius parameter correction amount (for circular protected areas) or other boundary parameters (for non-circular protected areas) until the inclusion condition is met.

[0104] The validity verification of the revised boundary description is to ensure that the protected area boundary can truly protect the transmission lines. The verification standard is that the protected area boundary always includes the geometric skeleton line of the transmission lines. Spatial geometric calculations and judgment algorithms are used to check whether the revised boundary meets this condition. If the verification fails, it indicates that the correction of the boundary parameters is not accurate enough, and the radius parameter correction (for circular protected areas) or other boundary parameters (for non-circular protected areas) need to be readjusted. An iterative adjustment method can be used, gradually changing the boundary parameters and verifying again until the inclusion condition is met.

[0105] Step S445: Extract real-time environmental parameters, including wind speed and temperature. Analyze the influence of environmental parameters on conductor deformation, generate environmental influence correction coefficients, and further correct the deformation offset set.

[0106] Real-time environmental parameters such as wind speed and temperature have a significant impact on the deformation of transmission lines. Wind speed causes the conductors to sway, and temperature changes cause thermal expansion and contraction, thus altering the conductor's length and shape. Real-time environmental parameters can be extracted using meteorological sensors installed near transmission lines. Analyzing the influence of environmental parameters on conductor deformation can be achieved through historical data statistical analysis and machine learning modeling. For example, a regression model can be established between wind speed, temperature, and conductor deformation. By training the model with a large amount of historical data, a quantitative relationship between environmental parameters and deformation can be obtained. Based on these influence patterns, an environmental impact correction coefficient is generated. This coefficient is used to further correct the deformation offset set to more accurately reflect the impact of environmental factors on conductor deformation. For example, under conditions of high wind speed and significant temperature changes, the deformation offset set can be adjusted according to the environmental impact correction coefficient, making the correction of the protected area boundary more consistent with the actual situation.

[0107] Step S446: Integrate the corrected boundary description parameters and environmental impact correction coefficients to generate dynamic protection zone boundary parameters. The update frequency of the dynamic protection zone boundary parameters should be consistent with the conductor deformation monitoring frequency.

[0108] The revised boundary description parameters and environmental impact correction coefficients are integrated to obtain a complete set of parameters, namely the dynamic protection zone boundary parameters. These parameters contain accurate boundary information of the transmission line protection zone after considering the real-time deformation of the conductors and the influence of environmental factors. To ensure that the protection zone boundary can reflect the latest conductor deformation in a timely manner, the update frequency of the dynamic protection zone boundary parameters is consistent with the conductor deformation monitoring frequency. The conductor deformation monitoring frequency is preset according to actual needs and the performance of the monitoring equipment. For example, if the conductor deformation is monitored once per second, then the dynamic protection zone boundary parameters are also updated once per second. In this way, the protection zone boundary can be adjusted in real time and accurately, providing strong support for the safety protection of transmission lines. For example, in the real-time monitoring of transmission lines, the dynamic protection zone boundary parameters are continuously updated to adapt to the real-time deformation of the conductors and environmental changes.

[0109] Step S450: Substitute the spatial position vector of the mechanical point cloud region into the dynamic protected area boundary parameters, calculate the penetration direction and penetration depth parameters of the position vector relative to the protected area boundary. The penetration direction is the angle between the vector direction and the normal vector of the protected area boundary. The penetration depth is determined by comparing the vector magnitude with the protected area boundary radius (for circular protected areas) or other boundary parameters (for non-circular protected areas).

[0110] In one implementation, step S450 may specifically include the following steps S451 to S456: Step S451: Analyze the boundary parameters of the dynamic protected area, extract the three-dimensional parameter description of the protected area boundary, and calculate the coordinates of the protected area boundary point and the boundary normal vector corresponding to each point in the mechanical point cloud area based on the parameter description. The direction of the boundary normal vector points from the boundary point to the outside of the protected area.

[0111] The dynamic protected area boundary parameters contain detailed information about the protected area boundary in three-dimensional space. Analyzing these parameters is crucial for obtaining a three-dimensional description of the protected area boundary, such as the boundary curve equation and key point coordinates. Based on these parameter descriptions, for each point in the mechanical point cloud region, the corresponding protected area boundary point coordinates are found through spatial geometric calculations and optimization algorithms. The boundary point is the point on the protected area boundary closest to these mechanical point cloud coordinates. Simultaneously, the normal vector at the boundary point is calculated, with its direction pointing outwards from the boundary point. The normal vector can be obtained by differentiating the boundary curve equation or by utilizing geometric relationships. For example, for a circular protected area, the normal vector is the vector pointing from the center of the circle to the boundary point; for a non-circular protected area, the normal vector is calculated based on the geometry and parameters of the boundary. By analyzing the dynamic protected area boundary parameters, the corresponding protected area boundary point coordinates and normal vector are determined for each point in the mechanical point cloud region, providing a foundation for subsequent calculations of penetration direction and penetration depth parameters.

[0112] Step S452: Extract the direction vector of the spatial position vector of the mechanical point cloud region, calculate the cosine value of the angle between the direction vector and the normal vector of the boundary of the protected area, and when the cosine value of the angle is within the preset range, determine that the position vector direction points into the interior of the protected area and is determined as the penetration direction.

[0113] The direction vector of the spatial position vector of each point in the mechanical point cloud region is extracted. This vector represents the direction information of the spatial position vector. The cosine value of the angle between the direction vector and the normal vector of the protected area boundary is calculated using the vector dot product formula. The preset range is a threshold value of the cosine value set in advance according to the actual situation and the accuracy requirements of the judgment. When the cosine value of the angle is within the preset range, it means that the angle between the direction vector and the normal vector is within a certain range, that is, the direction of the position vector points into the protected area, and this direction is determined to be the penetration direction. For example, if the preset range is that the cosine value of the angle is less than 0, when the calculated cosine value of the angle is less than 0, it means that the angle between the direction vector and the normal vector is greater than 90 degrees, the direction of the position vector points into the protected area, and this is determined to be the penetration direction. In this way, it is possible to accurately determine whether the spatial position vector of each point in the mechanical point cloud region penetrates the boundary of the protected area and the direction of penetration.

[0114] Step S453: Extract the shortest distance value of the point from the distance-azimuth feature set, combine it with the radius parameter of the protected area boundary (for circular protected areas) or other boundary parameters (for non-circular protected areas), and determine the penetration depth parameter by comparing the magnitude relationship between the shortest distance value and the radius parameter or other boundary parameters.

[0115] The shortest distance between each point in the mechanical point cloud region and the geometric skeleton line of the transmission line is obtained from the distance-azimuth feature set. The shortest distance value is compared with the radius parameter of the protected area boundary (for circular protected areas) or other boundary parameters (for non-circular protected areas). If the shortest distance value is greater than the distance corresponding to the boundary radius or other boundary parameters, it indicates that the point has penetrated the protected area boundary, and the penetration depth is the shortest distance value minus the distance corresponding to the boundary parameter; if the shortest distance value is less than or equal to the distance corresponding to the boundary parameter, the penetration depth is 0. This comparison accurately determines the penetration depth parameter of each point in the mechanical point cloud region relative to the protected area boundary, providing a quantitative indicator for assessing the degree of mechanical intrusion into the transmission line protected area. For example, in a circular protected area, comparing the shortest distance value with the radius parameter determines the penetration depth parameter, allowing for the determination of whether a point in the mechanical point cloud region has intruded into the protected area and the depth of intrusion.

[0116] Step S454: Perform statistical analysis on the penetration direction of each point within the same mechanical point cloud region, calculate the mode of the direction, generate the overall penetration direction vector of the region, and determine the direction of the overall penetration direction vector of the region by combining the penetration direction of each point within the region.

[0117] Statistical analysis of the penetration directions of points within the same mechanical point cloud region is conducted to understand the overall intrusion direction of the region. The frequency of each penetration direction is calculated, and the direction with the highest frequency is identified as the mode. Based on the mode, an overall penetration direction vector for the region is generated. This vector combines the penetration directions of each point within the region, reflecting the overall intrusion trend of the mechanical point cloud region. For example, in a mechanical point cloud region, by statistically analyzing the penetration directions of each point and identifying the direction with the highest frequency, this direction can be used as the direction of the overall penetration direction vector for the region.

[0118] Step S455: Combine the overall penetration direction vector and penetration depth parameters of the region to calculate the intrusion risk index of the mechanical point cloud region. The intrusion risk index is dynamically adjusted according to the penetration depth parameters and the number of point clouds in the region.

[0119] The intrusion risk index of a mechanical point cloud region is calculated by combining the overall penetration direction vector and penetration depth parameters. The intrusion risk index is a comprehensive indicator used to assess the degree of intrusion risk posed by a mechanical point cloud region to the transmission line protection zone. This index is dynamically adjusted based on the penetration depth parameter and the number of point clouds in the region. A greater penetration depth indicates a deeper intrusion and a higher risk; a larger number of point clouds in the region indicates a larger intrusion area and also a higher risk. For example, a weighted summation method can be used, multiplying the penetration depth parameter and the number of point clouds in the region by their respective weighting coefficients, and then summing them to obtain the intrusion risk index. The weighting coefficients are pre-set based on the actual situation and the focus of the risk assessment.

[0120] Step S456: Integrate the penetration direction, penetration depth parameters and regional intrusion risk index of each point to generate a set of intrusion features of the mechanical point cloud region on the protected area, which serves as the basis for constructing spatial intrusion vector field distribution data.

[0121] By integrating the penetration direction, penetration depth parameters, and regional intrusion risk index of each point within the mechanical point cloud region, a complete dataset is obtained—the intrusion feature set of the mechanical point cloud region to the protected area. This set contains detailed information on the intrusion of the mechanical point cloud region to the transmission line protected area, including the intrusion direction, intrusion depth, and the intrusion risk level of the entire region for each point. Based on the intrusion feature set, spatial intrusion vector field distribution data describing the intrusion status of the mechanical point cloud region to the protected area is constructed. This data can intuitively display the intrusion situation of the mechanical point cloud region to the protected area, providing strong support for the safety monitoring and risk early warning of transmission lines. For example, by graphically displaying the data in the intrusion feature set, the spatial intrusion vector field distribution data is obtained, clearly showing the intrusion range, direction, and risk level of the mechanical point cloud region to the transmission line protected area.

[0122] Step S460: Integrate the spatial location vectors, penetration direction and penetration depth parameters of all points in the mechanical point cloud area to generate spatial intrusion vector field distribution data describing the intrusion status of the mechanical point cloud area to the protected area. The density of the vector field is dynamically adjusted according to the intrusion risk of the mechanical point cloud area.

[0123] Integrating the spatial location vectors, penetration direction, and penetration depth parameters of all points in a mechanical point cloud region is crucial for comprehensively and accurately describing the intrusion status of the mechanical point cloud region into the transmission line protection zone. This information is organized according to specific rules and formats to generate spatial intrusion vector field distribution data. A vector field is a three-dimensional vector distribution field, where each vector represents the spatial location vector, penetration direction, and penetration depth information of a point. The density of the vector field is dynamically adjusted based on the intrusion risk of the mechanical point cloud region; the higher the intrusion risk, the greater the density of the vector field to highlight high-risk areas; conversely, the lower the intrusion risk, the smaller the density of the vector field. This method provides a visual representation of the intrusion of the mechanical point cloud region into the protection zone, offering a visual basis for transmission line safety monitoring and risk warning. For example, in a mechanical point cloud region generated by construction machinery near a transmission line, by integrating relevant information from each point, spatial intrusion vector field distribution data is generated, and the vector field density is adjusted according to the intrusion risk, clearly showing the severity and extent of the intrusion.

[0124] Step S500: Perform vector direction consistency analysis on the spatial intrusion vector field distribution data, extract vector clusters with the same penetration direction, calculate the cumulative sum of penetration depth parameters of the vector clusters, and when the cumulative sum of penetration depth parameters exceeds the preset risk threshold condition, generate mechanical external damage risk identification results containing the coordinates of the intrusion area and the risk level.

[0125] In one implementation, step S500 may specifically include the following steps S510-S560: Step S510: Analyze the spatial intrusion vector field distribution data, extract the rate of change parameters of all spatial location vectors over time, and generate a dynamic feature set of penetration depth by combining the current penetration depth parameter. The dynamic feature set includes the vector amplitude change trend and directional stability index.

[0126] This study analyzes spatial intrusion vector field distribution data, which includes information such as spatial position vectors, penetration direction, and penetration depth parameters for each point in the mechanical point cloud region. The rate of change parameter of all spatial position vectors over time is extracted by comparing and calculating spatial position vectors at different times. The rate of change parameter reflects the temporal change of the spatial position vector, including changes in vector amplitude and direction. Combined with the current penetration depth parameter, this information is integrated to generate a dynamic feature set for penetration depth. This set includes vector amplitude change trends and directional stability indices. The vector amplitude change trend describes how the magnitude of the spatial position vector changes over time, such as increasing, decreasing, or stabilizing; the directional stability index reflects the degree of change in vector direction over time, with smaller directional changes indicating higher stability. For example, by analyzing spatial intrusion vector field distribution data over a period of time, calculating the rate of change parameter of the spatial position vector, and combining it with the current penetration depth parameter, a dynamic feature set for penetration depth is generated, providing richer feature information for spatiotemporal joint clustering.

[0127] Step S520: Based on the dynamic feature set of penetration depth, perform spatiotemporal joint clustering on spatial intrusion vectors. By analyzing the directional consistency and penetration depth change trend of spatial intrusion vectors in the same spatial region over a continuous time window, spatial intrusion vectors that meet the spatiotemporal correlation conditions are divided into the same dynamic cluster.

[0128] In one implementation, step S520 may specifically include the following steps S521 to S526: Step S521: Analyze the dynamic feature set of penetration depth, extract the timestamp information, spatial coordinates and penetration direction consistency index of each spatial intrusion vector, and generate the spatiotemporal feature matrix of spatial intrusion vector.

[0129] The dynamic feature set of penetration depth is analyzed to extract the timestamp information, spatial coordinates, and penetration direction consistency index for each spatial intrusion vector. The timestamp records the specific moment of vector acquisition, the spatial coordinates represent the vector's position in three-dimensional space, and the penetration direction consistency index reflects the stability of the vector direction over time. This information is then organized into a matrix, namely the spatial intrusion vector spatiotemporal feature matrix. Each row of this matrix represents a spatial intrusion vector, and each column represents a feature. By generating the spatiotemporal feature matrix, the spatiotemporal characteristics of spatial intrusion vectors can be structurally represented, facilitating cluster analysis. For example, in a power transmission line monitoring scenario, analyzing the dynamic feature set of penetration depth and extracting relevant information for each spatial intrusion vector to generate the spatial intrusion vector spatiotemporal feature matrix facilitates subsequent spatial grid partitioning and time series segmentation.

[0130] Step S522: Based on the spatiotemporal feature matrix of the spatial intrusion vector, the spatial intrusion vector is divided into preset spatial grid units, each grid unit corresponds to a local spatial region, and the temporal distribution density of the spatial intrusion vector within the region is statistically analyzed.

[0131] Based on the spatiotemporal characteristic matrix of spatial intrusion vectors, the spatial intrusion vectors are divided according to preset spatial grid units. These preset spatial grid units are three-dimensional spatial grids pre-defined according to the size and accuracy requirements of the actual monitoring area. Each grid unit corresponds to a local spatial region, and the spatial intrusion vectors in the matrix are assigned to the corresponding grid units according to their spatial coordinates. The temporal distribution density of spatial intrusion vectors within each local spatial region is statistically analyzed. Temporal distribution density refers to the number of spatial intrusion vectors within a certain time range. By statistically analyzing the temporal distribution density, the distribution of spatial intrusion vectors in different spatial regions and at different times can be understood, providing a reference for time series segmentation. For example, dividing the transmission line monitoring area into multiple spatial grid units, assigning spatial intrusion vectors to the corresponding grid units, and statistically analyzing the temporal distribution density of spatial intrusion vectors within each grid unit reveals that certain areas have higher temporal distribution densities, potentially indicating a higher risk of intrusion.

[0132] Step S523: Perform time series segmentation on the spatial intrusion vectors within each local spatial region, divide the spatial intrusion vectors with consistent penetration direction within a preset range into the same time subsequence, and calculate the slope of the penetration depth change of the subsequence.

[0133] Time-series segmentation of spatial intrusion vectors within each local spatial region is performed to group vectors with similar penetration directions and intrusion trends into a single subsequence. Within each local spatial region, the spatial intrusion vectors are sorted chronologically, and then the consistency of their penetration directions within a continuous time window is analyzed. The preset range is a threshold for penetration direction variation pre-set based on actual conditions and analytical accuracy requirements. When the consistency of the penetration directions of spatial intrusion vectors within a continuous time window is within the preset range, they are grouped into the same time subsequence. For each time subsequence, the slope of its penetration depth change is calculated; the slope reflects the rate of change of penetration depth over time.

[0134] Step S524: When the slope of the penetration depth change matches the intrusion trend characteristics, mark the spatial intrusion vector of the corresponding time subsequence as a candidate correlation vector and generate a candidate vector set.

[0135] The slope of the penetration depth change reflects the change in the penetration depth of a spatial intrusion vector over time. Intrusion trend characteristics are pre-set slope thresholds and trend conditions based on actual conditions and risk assessment requirements. For example, a penetration depth slope greater than zero and a slope growth rate within a certain range indicate that the intrusion depth is increasing at a certain rate. When the slope of the penetration depth change of a certain time series meets the intrusion trend characteristics, it indicates that the spatial intrusion vectors within that subsequence have a significant intrusion trend, and these vectors are marked as candidate correlation vectors. All candidate correlation vectors are collected to generate a candidate vector set. For example, if the preset intrusion trend characteristics are a penetration depth slope greater than 0.1 and a slope growth rate between 0.01 and 0.05, when the slope of the penetration depth change of a certain time series meets these conditions, the spatial intrusion vectors within that subsequence are marked as candidate correlation vectors, generating a candidate vector set, which provides a basis for determining spatiotemporal correlation.

[0136] Step S525: Calculate the spatial distance and time interval of the spatial intrusion vectors in the candidate vector set. When the spatial distance and time interval are within the preset range, they are determined to have spatiotemporal correlation and are merged into the same dynamic cluster.

[0137] The spatial distance and time interval of spatial intrusion vectors in the candidate vector set are calculated. Spatial distance refers to the three-dimensional spatial distance between the origins of two vectors, and time interval refers to the difference in acquisition time between two vectors. The preset range is a threshold for spatial distance and time interval set in advance based on the actual situation and clustering accuracy requirements. When the spatial distance and time interval of spatial intrusion vectors are both within the preset range, it indicates that they are correlated in both space and time, and these vectors are merged into the same dynamic cluster.

[0138] Step S526: Perform noise filtering on the merged dynamic clusters to remove sub-clusters containing isolated spatial intrusion vectors or spatial intrusion vectors with abrupt changes in penetration direction, and retain dynamic clusters with continuous spatiotemporal evolution characteristics. The identifier of the dynamic clusters includes spatial region number and time series marker.

[0139] Noise filtering of the merged dynamic clusters aims to remove anomalous vectors or sub-clusters, improving the cluster's quality and accuracy. Isolated spatial intrusion vectors are those without significant spatiotemporal correlation with other vectors within the cluster, while vectors with abrupt changes in penetration direction are those whose penetration direction changes drastically within a short period. By analyzing and judging the vectors in the dynamic clusters, sub-clusters containing these anomalous vectors are removed. Dynamic clusters with continuous spatiotemporal evolution characteristics are retained, meaning the vectors within the cluster exhibit continuous spatial and temporal trends without significant abrupt changes. An identifier is generated for each retained dynamic cluster, containing a spatial region number and a time-series marker, for subsequent management and querying. For example, if a sub-cluster containing an isolated spatial intrusion vector is found in a merged dynamic cluster, that sub-cluster is removed, and dynamic clusters with continuous spatiotemporal evolution characteristics are retained, with an identifier containing a spatial region number and a time-series marker generated for them.

[0140] Step S530: Calculate the penetration depth duration parameter and penetration depth growth rate of each dynamic cluster, and filter out dynamic clusters whose duration exceeds the preset duration and whose penetration depth growth rate meets the intrusion acceleration characteristics, and mark them as high-risk intrusion clusters.

[0141] The penetration depth duration parameter is calculated for each dynamic cluster. Duration refers to the length of time the penetration depth of a vector within the cluster remains greater than zero. The penetration depth growth rate refers to the rate of change of penetration depth over time, obtained through time series analysis of the penetration depth of vectors within the cluster. The preset duration is a time threshold pre-set based on actual conditions and risk assessment requirements. The intrusion acceleration characteristic is a threshold and trend condition for the penetration depth growth rate pre-set based on actual conditions and risk warning requirements. When the penetration depth duration of a dynamic cluster exceeds the preset duration and the penetration depth growth rate meets the intrusion acceleration characteristic, it indicates that the intrusion behavior represented by that cluster has a high risk, and it is marked as a high-risk intrusion cluster.

[0142] Step S540: Perform time-varying weighted accumulation on the penetration depth parameter in the high-risk intrusion cluster. The weight value increases over time to strengthen the impact of recent intrusion behavior and generate a time-varying weighted cumulative sum. The calculation of the cumulative sum takes into account the correction effect of the penetration direction consistency index.

[0143] Time-varying weighted accumulation of penetration depth parameters in high-risk intrusion clusters aims to more accurately assess the risk level of intrusion behavior. The weight values ​​increase over time, with more recent intrusions receiving greater weight, thus amplifying the impact of recent intrusions on the accumulated sum. The time-varying weighted accumulated sum is obtained by multiplying the penetration depth parameter of each vector in the high-risk intrusion cluster by its corresponding weight value and then summing the results. The correction effect of the penetration direction consistency index is considered during the calculation of the accumulated sum. The penetration direction consistency index reflects the stability of the vector's penetration direction over time. High penetration direction consistency indicates strong directionality and persistence of the intrusion behavior, resulting in a significant correction effect on the accumulated sum; conversely, low penetration direction consistency indicates more chaotic intrusion behavior, resulting in a smaller correction effect. For example, an exponential weighting method can be used, where the weight values ​​increase exponentially over time, and the weight values ​​are adjusted based on the penetration direction consistency index to generate the time-varying weighted accumulated sum, providing a more accurate basis for risk assessment.

[0144] Step S550: Compare the time-varying weighted cumulative sum with the preset dynamic risk threshold condition. The dynamic risk threshold is adaptively adjusted according to the real-time load status of the transmission line and the environmental risk coefficient. When the cumulative sum exceeds the adjusted risk threshold, it is determined that the corresponding mechanical point cloud area has external damage risk.

[0145] The time-varying weighted cumulative sum is compared with a preset dynamic risk threshold, which is adaptively adjusted based on the real-time load status of the transmission line and the environmental risk coefficient. The real-time load status of the transmission line reflects its operational condition; the higher the load, the lower the risk threshold. The environmental risk coefficient considers the impact of environmental factors on the transmission line, such as wind speed, temperature, and humidity; the higher the environmental risk, the lower the risk threshold. By monitoring the load status and environmental factors of the transmission line in real time, the risk threshold is dynamically adjusted. When the time-varying weighted cumulative sum of high-risk intrusion clusters exceeds the adjusted risk threshold, it indicates that the corresponding mechanical point cloud region poses an external damage risk to the transmission line, and the region is determined to have an external damage risk. For example, when the transmission line load is large and the environmental risk is high, the dynamic risk threshold is lowered. When the time-varying weighted cumulative sum of high-risk intrusion clusters exceeds the adjusted risk threshold, an external damage risk is promptly determined, providing timely early warning for the safety protection of the transmission line.

[0146] Step S560: Extract the spatiotemporal distribution features of the mechanical point cloud region with external damage risk, including the intrusion start time, current intrusion speed and predicted intrusion path. Combine the time-varying weighted cumulative sum with the comparison results of the dynamic risk threshold to generate a mechanical external damage risk identification result containing multi-dimensional risk indicators.

[0147] In one implementation, step S560 may specifically include the following steps S561 to S566: Step S561: Analyze the vector spatiotemporal feature matrix of the high-risk intrusion cluster, extract the vector spatial coordinates and penetration depth parameters corresponding to the earliest timestamp within the cluster, and determine the intrusion start time and initial intrusion location of the mechanical point cloud region.

[0148] This paper analyzes the spatiotemporal feature matrix of a high-risk intrusion cluster, which contains spatiotemporal feature information of spatial intrusion vectors within the cluster. The earliest timestamp in the matrix is ​​extracted, along with its corresponding vector spatial coordinates and penetration depth parameter. The earliest timestamp represents the time when the mechanical point cloud region begins to intrude into the transmission line protection zone. The intrusion start time of the mechanical point cloud region is determined based on the earliest timestamp, and the initial intrusion location is determined based on the corresponding vector spatial coordinates. For example, in the spatiotemporal feature matrix of a high-risk intrusion cluster, the earliest timestamp is found to be t0, with corresponding vector spatial coordinates (x0, y0, z0) and a penetration depth parameter d0. This determines the intrusion start time as t0 and the initial intrusion location as (x0, y0, z0), providing fundamental information for risk assessment and prediction.

[0149] Step S562: Based on the vector amplitude change data of continuous time windows within the cluster, calculate the amplitude increment per unit time, and combine it with the direction parameter of the spatial location vector to generate the current intrusion velocity vector of the mechanical point cloud region. The magnitude of the velocity vector is the ratio of the amplitude increment to the time interval, and the direction is the penetration direction.

[0150] Based on vector amplitude variation data within continuous time windows within a high-risk intrusion cluster, this data reflects the temporal change in the magnitude of the spatial intrusion vector. The amplitude increment per unit time is calculated as the difference in vector amplitude between adjacent time windows divided by the time interval. Combined with the direction parameter of the spatial location vector, which represents the vector's penetration direction, a current intrusion velocity vector for the mechanical point cloud region is generated. The magnitude of this velocity vector is the ratio of the amplitude increment to the time interval, reflecting the magnitude of the intrusion; its direction is the penetration direction, reflecting the direction of the intrusion. For example, in two consecutive time windows t1 and t2, with vector amplitudes A1 and A2 respectively and a time interval Δt, the amplitude increment is calculated as A2-A1, the magnitude of the current intrusion velocity vector is (A2-A1) / Δt, and the direction is the penetration direction, accurately describing the current intrusion velocity and direction of the mechanical point cloud region.

[0151] Step S563: Substitute the current intrusion velocity vector into the preset kinematic prediction model, and combine it with the mass inertia parameters of the mechanical point cloud region to predict the intrusion path coordinate sequence within a preset time period in the future. The path prediction takes into account the rebound effect correction of the protected area boundary.

[0152] The current intrusion velocity vector is substituted into a preset kinematic prediction model. This model is a mathematical model established based on the laws of motion of an object, such as Newton's laws of motion. Combined with the mass inertia parameters of the mechanical point cloud region, which reflect its mass and inertial characteristics and influence its trajectory, the intrusion path coordinate sequence is predicted for a preset time period. This means predicting the positional changes of the mechanical point cloud region over a future period based on the current intrusion velocity and mass inertia parameters. The path prediction considers corrections for the rebound effect at the protected area boundary. When the intrusion path of the mechanical point cloud region encounters the boundary, the path is corrected based on the boundary's elasticity and physical characteristics to simulate the rebound effect. For example, when the intrusion path approaches the protected area boundary, the path is adjusted based on the boundary's elasticity coefficient and collision angle, predicting the intrusion path coordinate sequence for a preset time period, thus providing more accurate prediction information for the safety protection of transmission lines.

[0153] Step S564: Analyze the comparison results between the time-varying weighted cumulative sum and the dynamic risk threshold, and calculate the excess risk coefficient. The excess risk coefficient is the ratio of the portion of the cumulative sum that exceeds the threshold to the threshold, and is used to characterize the severity of the risk.

[0154] The analysis compares the results of the time-varying weighted cumulative sum with the dynamic risk threshold. The time-varying weighted cumulative sum reflects the intrusion risk level of the mechanical point cloud region, while the dynamic risk threshold is adaptively adjusted based on the real-time load status of the transmission line and the environmental risk coefficient. The excess risk coefficient is calculated as the ratio of the portion of the cumulative sum exceeding the threshold to the threshold value, i.e., (time-varying weighted cumulative sum - dynamic risk threshold) / dynamic risk threshold. A larger excess risk coefficient indicates a more severe intrusion risk in the mechanical point cloud region.

[0155] Step S565: Integrate the intrusion start time, current intrusion speed, predicted intrusion path and excess risk coefficient to generate a multi-dimensional risk indicator set. The indicator set includes the risk evolution rate in the time dimension, the risk diffusion range in the spatial dimension and the risk level in the severity dimension.

[0156] By integrating the intrusion start time, current intrusion speed, predicted intrusion path, and excess risk coefficient, a multidimensional risk indicator set is generated. This set includes risk indicators across three dimensions: time, space, and severity. The time dimension, risk evolution rate, reflects how risk changes over time, such as the rate of change in intrusion speed and the rate of increase in penetration depth. The spatial dimension, risk diffusion range, is determined by the predicted intrusion path; the larger the area covered by the predicted intrusion path, the larger the risk diffusion range. The severity dimension, risk level, is classified according to the excess risk coefficient; the larger the excess risk coefficient, the higher the risk level. For example, by integrating the intrusion start time, current intrusion speed, predicted intrusion path, and excess risk coefficient to generate a multidimensional risk indicator set, the risk evolution rate is the rate of increase in intrusion speed, the risk diffusion range is the area covered by the predicted intrusion path, and the risk level is divided into low, medium, and high levels based on the excess risk coefficient, comprehensively and accurately describing the situation of mechanical external damage risk.

[0157] Step S566: Standardize and encode the multidimensional risk indicator set to generate mechanical external damage risk identification results that conform to the data interface specifications of the transmission line monitoring system. The numerical accuracy of each risk indicator in the results is consistent with the sampling resolution of the lidar.

[0158] Standardizing the multidimensional risk indicator set involves converting it into a unified encoding format for easier reception and processing by the transmission line monitoring system. This standardization process includes quantifying and encoding each risk indicator to conform to the data interface specifications of the transmission line monitoring system. The numerical precision of each risk indicator in the result is consistent with the sampling resolution of the lidar, ensuring data accuracy and consistency. For example, indicators such as risk evolution rate, risk diffusion range, and risk level in the multidimensional risk indicator set are quantified and converted into binary codes to generate mechanical external damage risk identification results that conform to the data interface specifications of the transmission line monitoring system, facilitating real-time monitoring and management of transmission lines.

[0159] This invention also provides a computer system, including a memory and a processor. The memory stores a computer program that can run on the processor. When the processor executes the program, it implements the steps in the method for identifying mechanical external damage risks of transmission lines based on point cloud and RTK fusion provided in this invention.

[0160] Please see details. Figure 3 This is a schematic diagram of the structure of a computer system provided in an embodiment of the present invention. Figure 3As shown, the computer system 1000 described above may include: a processor 1001, a network interface 1004, and a memory 1005. Furthermore, the computer system 1000 may also include: a user interface 1003, and at least one communication bus 1002. The communication bus 1002 is used to implement communication between these components. The user interface 1003 may include a standard wired interface or a wireless interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface). The memory 1005 may be high-speed RAM or non-volatile memory, such as at least one disk storage device. Optionally, the memory 1005 may also be at least one storage device located remotely from the aforementioned processor 1001. Figure 3 As shown, the memory 1005, which is a computer-readable storage medium, may include an operating system, a network communication module, a user interface module, and a device control application program. Figure 3 In the computer system 1000 shown, the network interface 1004 provides network communication functions; the user interface 1003 is mainly used to provide an input interface; and the processor 1001 can be used to call the device control application stored in the memory 1005 to implement the methods provided in the above embodiments.

[0161] It should be understood that the computer system 1000 described in the embodiments of the present invention can execute the foregoing text. Figure 2 The implementation principle and beneficial effects of the method for identifying mechanical external damage risks of transmission lines based on point cloud and RTK fusion in the corresponding embodiments will not be elaborated here.

Claims

1. A method for identifying mechanical external damage risks of transmission lines based on point cloud and RTK fusion, characterized in that, The method includes: The three-dimensional position coordinates of the vehicle are collected in real time by the RTK positioning device on the construction vehicle, and the local point cloud data of the surrounding environment collected by the lidar with the three-dimensional position coordinates as the center are obtained to obtain a point cloud data sequence with spatiotemporal markers. The point cloud data sequence with spatiotemporal markers is subjected to coordinate transformation processing to transform the local point cloud data collected by the lidar from the device coordinate system to the geographic coordinate system. Combined with the three-dimensional position coordinates provided by the RTK positioning device as rigid constraints, the registration optimization of multi-frame point cloud data is carried out to generate a structured point cloud voxel set with a unified geographic coordinate reference. The point cloud subset corresponding to the transmission line region is separated from the structured point cloud voxel set. The skeleton of the point cloud subset is extracted to generate the three-dimensional geometric skeleton line of the transmission conductor. The mechanical point cloud region belonging to the construction machinery in the structured point cloud voxel set is identified. The shortest distance distribution and spatial azimuth relationship between the mechanical point cloud region and the geometric skeleton line of the transmission conductor are calculated to obtain the topological proximity relationship descriptor between the mechanical point cloud and the geometric skeleton of the transmission conductor. Based on the topological proximity descriptor of the mechanical point cloud and the geometric skeleton of the transmission line, the spatial position vector of each point in the mechanical point cloud region relative to the geometric skeleton line of the transmission line is determined. Combined with the preset spatial range boundary of the transmission line protection zone, the penetration direction and penetration depth parameters of each position vector relative to the boundary of the protection zone are calculated, and spatial intrusion vector field distribution data describing the intrusion state of the mechanical point cloud region into the protection zone are constructed. A vector direction consistency analysis is performed on the spatial intrusion vector field distribution data to extract vector clusters with the same penetration direction. The cumulative sum of the penetration depth parameters of the vector clusters is calculated. When the cumulative sum of the penetration depth parameters exceeds a preset risk threshold condition, a mechanical external damage risk identification result containing the coordinates of the intrusion area and the risk level is generated.

2. The method according to claim 1, characterized in that, The coordinate transformation process performed on the spatiotemporally marked point cloud data sequence converts the local point cloud data acquired by the lidar from the device coordinate system to the geographic coordinate system. Using the three-dimensional position coordinates provided by the RTK positioning device as rigid constraints, multi-frame point cloud data registration optimization is performed to generate a structured point cloud voxel set with a unified geographic coordinate reference, including: The point cloud data sequence with spatiotemporal markers is analyzed, the time synchronization error between the lidar and the RTK positioning device is extracted, the timestamp deviation between adjacent sampling times is calculated, and a time synchronization error sequence is generated as the time reference correction basis for coordinate transformation. Based on the time synchronization error sequence, the local point cloud data collected by the lidar is timestamped and resampled to keep the time reference of the point cloud data consistent with the RTK three-dimensional position coordinates, thereby generating time-aligned point cloud data units. Extract the dynamic error characteristics of the lidar equipment, and generate a set of dynamic error characteristics by analyzing the error distribution patterns of historical data. By combining the device's dynamic error feature set, error compensation processing is performed on the time-aligned point cloud data units to correct the coordinate tilt caused by attitude angle fluctuations and the distance deviation caused by ranging drift, thus obtaining the error-compensated intermediate point cloud data. Using RTK 3D position coordinates as rigid constraints, multi-frame registration is performed on the intermediate point cloud data after error compensation. By analyzing the overlapping area characteristics of point clouds in adjacent frames, the registration deviation correction coefficient is calculated to minimize the spatial position error of point clouds in multiple frames. The registered point cloud data is divided into voxels, and the point cloud density and normal vector distribution of each voxel are statistically analyzed to generate a structured point cloud voxel set with a unified geographic coordinate reference.

3. The method according to claim 2, characterized in that, The process involves combining the device's dynamic error feature set to perform error compensation processing on time-aligned point cloud data units, correcting coordinate tilt caused by attitude angle fluctuations and distance deviations caused by ranging drift, to obtain error-compensated intermediate point cloud data, including: The attitude angle fluctuations in the dynamic error feature set of the equipment are analyzed to generate an attitude angle time series. The periodic components in the series are decomposed by Fourier transform, and the characteristic frequency parameters of the fluctuation components are extracted. The characteristic frequency parameters are determined by the frequency values ​​corresponding to the peak values ​​of the spectrum. Based on the characteristic frequency parameters and the sampling frequency of the lidar, the attitude angle fluctuation error is divided into two types of components: when the characteristic frequency is greater than or equal to half of the sampling frequency, it is marked as the first type of fluctuation error component; when the characteristic frequency is less than half of the sampling frequency, it is marked as the second type of drift error component. For the first type of fluctuation error component, the original error is canceled by constructing an inverse fluctuation sequence to generate the first type of error compensation amount. The amplitude of the compensation amount is equal to the peak value of the fluctuation error, and the phase difference is half a period. For the second type of drift error component, the sliding window averaging method is used to smooth the error sequence. The window length is determined according to the reciprocal of the characteristic frequency to generate the second type of error compensation amount, which is the average value of the error within the window. The first type of error compensation amount and the second type of error compensation amount are integrated to generate a comprehensive error compensation coefficient matrix. The coordinate transformation correction is performed on the time-aligned point cloud data units. The correction process simultaneously compensates for the coupling error of attitude angle fluctuation and ranging drift. The corrected point cloud data is spatially compared with the RTK 3D position coordinates, and the root mean square value of the coordinate residual is calculated. When the root mean square value is less than the nominal ranging error of the lidar, it is determined as the intermediate point cloud data after error compensation.

4. The method according to claim 1, characterized in that, The process involves separating a subset of point clouds corresponding to the transmission line region from the structured point cloud voxel set, extracting the skeleton from the point cloud subset to generate a three-dimensional geometric skeleton line of the transmission conductor, identifying mechanical point cloud regions belonging to construction machinery in the structured point cloud voxel set, calculating the shortest distance distribution and spatial azimuth relationship between the mechanical point cloud regions and the geometric skeleton line of the transmission conductor, and obtaining a topological proximity descriptor between the mechanical point cloud and the geometric skeleton line of the transmission conductor, including: Feature enhancement is performed on the structured point cloud voxel set to extract the point cloud density, normal vector direction and time dynamic change rate of each voxel unit, and generate a multimodal voxel feature set. Based on the multimodal voxel feature set, the structured point cloud voxel set is initially segmented, and potential transmission line areas are marked by analyzing the density continuity and normal vector consistency of voxel units. Extract tower point cloud features from potential transmission line areas, analyze point cloud distribution features based on the cylindrical geometry of the towers, detect the center point coordinates and radius parameters of the towers, and generate tower point cloud marking information; Based on the tower point cloud marking information, the potential transmission line area is segmented in a secondary manner. By analyzing the point cloud connectivity and directional trend between towers, the point cloud subset corresponding to the transmission conductor is separated, and the non-conductor point clouds of towers and insulators are removed. The skeleton of the transmission line point cloud subset is extracted. By analyzing the main direction vector and curvature change characteristics of the point cloud, a three-dimensional geometric skeleton line of the transmission line is generated. The node density of the skeleton line is dynamically adjusted according to the curvature of the conductor. The system identifies dynamic point cloud regions outside the transmission line region within a structured point cloud voxel set. By analyzing the motion trajectory and velocity vector distribution of the point cloud, it marks the mechanical point cloud regions belonging to construction machinery. It calculates the shortest distance distribution and spatial azimuth relationship between the mechanical point cloud regions and the geometric skeleton line of the transmission conductor, and generates a topological proximity descriptor between the mechanical point cloud and the geometric skeleton of the transmission conductor.

5. The method according to claim 4, characterized in that, Based on the multimodal voxel feature set, a region growing algorithm is used to initially segment the structured point cloud voxel set. Potential transmission line regions are marked by analyzing the density continuity and normal vector consistency of the voxel units, including: The multimodal voxel feature set is analyzed, and the point cloud density value and normal vector direction angle of the voxel unit are extracted to generate a density-normal vector feature set. Calculate the density gradient and normal angle of adjacent voxel units in the density-normal vector feature set. When the density gradient and normal angle are within a preset range, they are determined to be adjacent voxels with continuity. Initialize region growth seed points by selecting voxel units whose point cloud density values ​​conform to the characteristics of transmission lines and whose normal vector direction angles conform to the characteristics of transmission line orientation, and add them to the region growth queue. Take the current voxel unit from the region growth queue, traverse the adjacent voxel units in its eight neighborhoods, and mark the adjacent voxel units that meet the conditions as the same region based on the density continuity and normal vector consistency determination results, and add them to the region growth queue. Repeat the region growth process until all voxel units in the queue have been processed, generating a region point cloud with multiple connected component labels. Filter connected regions in the point cloud of the region whose volume is within a preset range, retain connected regions whose point cloud density conforms to the characteristics of power transmission lines and whose normal vector direction is consistent, and mark them as potential power transmission line regions. The connected region markings of potential power transmission line regions are consistent with the markings of seed points.

6. The method according to claim 5, characterized in that, The identification of dynamic point cloud regions outside the transmission line region within the structured point cloud voxel set involves analyzing the motion trajectory and velocity vector distribution of the point clouds, marking mechanical point cloud regions belonging to construction machinery, calculating the shortest distance distribution and spatial azimuth relationship between the mechanical point cloud regions and the geometric skeleton line of the transmission conductor, and generating a topological proximity descriptor between the mechanical point cloud and the geometric skeleton of the transmission conductor, including: Extract the remaining point cloud data that is not marked as a transmission line area from the structured point cloud voxel set, divide it into continuous time window point cloud subsets according to the timestamp order, and dynamically set the duration of each time window according to the sampling frequency of the lidar. Calculate the centroid coordinates of the point cloud subset for each time window. By analyzing the changes in the centroid coordinates of adjacent time windows, generate the motion trajectory and velocity vector of the point cloud region. The magnitude of the velocity vector is the distance the centroid moves per unit time, and the direction is the direction of the centroid movement. The motion trajectory and velocity vector are continuously analyzed. When the velocity vector direction changes within a preset range for multiple consecutive time windows, it is determined to be a dynamic point cloud region with a stable motion trend. Extract the geometric features of the dynamic point cloud region, including the size, volume and surface curvature distribution of the bounding rectangle, and match them with the preset construction machinery morphology feature library to select dynamic point cloud regions that match the morphology features of the machinery. Cluster the selected dynamic point cloud regions and merge point cloud units with spatial distances within a preset range into the same mechanical point cloud region to generate mechanical point cloud region labeling information. For each mechanical point cloud region, calculate the shortest distance and spatial azimuth angle from each point in the region to the geometric skeleton of the transmission line, statistically analyze the distance distribution pattern and azimuth angle variation trend, and generate a topological proximity descriptor between the mechanical point cloud and the geometric skeleton of the transmission line.

7. The method according to claim 1, characterized in that, The topological proximity descriptor based on the mechanical point cloud and the geometric skeleton of the transmission line determines the spatial position vector of each point in the mechanical point cloud region relative to the geometric skeleton line of the transmission line. Combined with the preset spatial boundary of the transmission line protection zone, the penetration direction and penetration depth parameters of each position vector relative to the boundary of the protection zone are calculated. This constructs spatial intrusion vector field distribution data describing the intrusion state of the mechanical point cloud region into the protection zone, including: The topological proximity descriptor is analyzed to extract the shortest distance and spatial azimuth parameters between each point in the mechanical point cloud region and the geometric skeleton line of the transmission line, and to generate a distance-azimuth feature set. Based on the distance-azimuth feature set, the spatial position vector of each point in the mechanical point cloud region relative to the geometric skeleton line of the transmission line is determined. The starting point of the vector is the projection point of the point on the skeleton line, the ending point is the mechanical point cloud coordinates, the vector direction is from the projection point to the mechanical point cloud coordinates, and the magnitude is equal to the shortest distance value. Extract real-time deformation characteristics of transmission lines, including the sway amplitude and curvature of the lines caused by environmental factors, and generate dynamic deformation parameters of the lines by analyzing historical deformation data. Based on the aforementioned conductor dynamic deformation parameters, the preset spatial range boundary of the transmission line protection zone is dynamically adjusted to correct the boundary offset of the protection zone caused by conductor deformation, thereby generating dynamic protection zone boundary parameters. Substitute the spatial position vector of the mechanical point cloud region into the dynamic protected area boundary parameters to calculate the penetration direction and penetration depth parameters of the position vector relative to the protected area boundary. The penetration direction is the angle between the vector direction and the normal vector of the protected area boundary, and the penetration depth is determined by comparing the vector magnitude with the radius of the protected area boundary. By integrating the spatial location vectors, penetration direction, and penetration depth parameters of all points in the mechanical point cloud region, spatial intrusion vector field distribution data describing the intrusion status of the mechanical point cloud region to the protected area is generated. The density of the vector field is dynamically adjusted according to the intrusion risk of the mechanical point cloud region.

8. The method according to claim 7, characterized in that, Based on the distance-azimuth feature set, the spatial position vector of each point in the mechanical point cloud region relative to the geometric skeleton line of the transmission line is determined. The starting point of the vector is the projection point of that point on the skeleton line, the ending point is the mechanical point cloud coordinates, the vector direction is from the projection point to the mechanical point cloud coordinates, and the magnitude is equal to the shortest distance value, including: The three-dimensional parameter description of the geometric skeleton line of the transmission line is analyzed. For each point cloud coordinate in the mechanical point cloud region, the coordinates of its projection point on the skeleton line are calculated. The projection point is the point on the skeleton line that is closest to the mechanical point cloud coordinate. Based on the projection point coordinates and the mechanical point cloud coordinates, calculate the unit direction vector of the skeleton line pointing to the mechanical point cloud coordinates. The direction of the unit direction vector is from the projection point to the mechanical point cloud coordinates, and the module length is standardized to a unified benchmark. Extract the shortest distance value of the point from the distance-azimuth feature set, and combine the unit direction vector with the shortest distance value to generate a preliminary spatial position vector. The starting point of the preliminary vector is the coordinates of the projection point, and the ending point is the coordinates of the mechanical point cloud. Analyze the local curvature of the geometric skeleton line of the transmission conductor. When the curvature is within the preset range, correct the direction of the initial spatial position vector so that the vector direction is always perpendicular to the tangent direction of the skeleton line at the projection point, and compensate for the directional deviation caused by the bending of the conductor. The overall motion trend vector of the mechanical point cloud region is calculated, and the spatial position vectors of each point in the region are weighted and averaged. The weight values ​​are dynamically adjusted according to the distance of the point cloud to the center of gravity of the region to enhance the consistency of the region's motion direction. The corrected spatial position vector is standardized to ensure that the error between the vector magnitude and the shortest distance value is within a preset range, thereby generating standardized spatial position vectors for each point in the mechanical point cloud region.

9. The method according to claim 8, characterized in that, The method of dynamically adjusting the preset spatial boundary of the transmission line protection zone by combining the conductor dynamic deformation parameters, correcting the boundary offset of the protection zone caused by conductor deformation, and generating dynamic protection zone boundary parameters includes: The dynamic deformation parameters of the conductor are analyzed, the swing amplitude and bending curvature of the conductor are extracted, and a deformation feature vector is generated. The composition of the deformation feature vector is dynamically set according to the number of deformation monitoring points of the conductor. Based on the deformation feature vector, the deformation offset of the conductor at each monitoring point is calculated. The offset is the difference between the current coordinates and the initial coordinates of the monitoring point, and a set of deformation offsets is generated. Substitute the set of deformation offsets into the preset spatial range boundary description of the transmission line protection zone, and correct the center coordinates and radius parameters in the boundary description. The center coordinate correction is equal to the deformation offset, and the radius parameter correction is dynamically adjusted according to the swing amplitude. The validity of the revised boundary description is verified to ensure that the boundary of the protected area always includes the geometric skeleton line of the transmission line. If the verification fails, the radius parameter correction amount is readjusted until the inclusion condition is met. Real-time environmental parameters are extracted, and by analyzing the influence of environmental parameters on conductor deformation, environmental influence correction coefficients are generated to further correct the deformation offset set. By integrating the revised boundary description parameters and environmental impact correction coefficients, dynamic protection zone boundary parameters are generated. The update frequency of the dynamic protection zone boundary parameters is consistent with the conductor deformation monitoring frequency.

10. A computer system, characterized in that, include: processor; And a memory, wherein the memory stores computer-readable code that, when executed by the processor, causes the processor to perform the method as described in any one of claims 1 to 9.

Images