A High-Precision Monitoring Method and System for Transmission Lines Based on BeiDou

Abstract

This application provides a method and system for high-precision transmission line monitoring based on BeiDou, relating to the field of data processing technology. The method includes: deploying a BeiDou dual-frequency RTK module on the top of the tower and activating the lidar; establishing a three-dimensional laser point cloud; performing point cloud alignment based on principal component analysis using the three-dimensional laser point cloud and historical point cloud data to complete coarse matching positioning; establishing the tower position coordinates and constructing a connection line between tower points for trust verification; based on the coarse matching positioning, performing authentication matching under joint authentication according to the point cloud trust level and the nearest point cloud; updating the transmission line point cloud based on the authentication matching results and generating an early warning. This application solves the technical problem of potential security risks caused by the poor timeliness of historical laser point cloud data and the lag in real-time point cloud data updates, avoiding inaccurate updates of historical point cloud data and improving the response speed and security of transmission line monitoring.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method and system for monitoring power transmission lines based on BeiDou high precision. Background Technology

[0002] The development of online monitoring technology for power transmission lines has been a continuous process of progress and improvement, from initial basic theoretical research to current intelligent and automated monitoring. In existing power transmission line monitoring systems, historical laser point cloud data suffers from poor timeliness, and real-time point cloud data updates are delayed. This is particularly problematic in complex environments, where efficient synchronization between historical and real-time point cloud data is difficult. Delayed point cloud data updates prevent the power transmission line monitoring system from accurately reflecting the current situation, thus affecting the system's accuracy and reliability. This is especially true in applications requiring high precision and rapid response (such as power transmission line monitoring), where historical point cloud data often fails to accurately reflect the current environmental state. Furthermore, the time required for lidar data acquisition and processing, particularly in large areas or complex environments, leads to delayed real-time data updates, hindering timely responses to emergencies and increasing safety risks. In conclusion, poor data timeliness and delayed updates directly impact the power transmission line monitoring system's ability to quickly identify and respond to potential hazards. This is especially critical in modern monitoring systems that rely on automated monitoring and intelligent decision-making, where real-time updates are paramount.

[0003] In summary, existing technologies suffer from technical problems such as poor timeliness of historical laser point cloud data and delayed updates of real-time point cloud data, which lead to potential safety risks and affect the accuracy and safety of power transmission line monitoring. Summary of the Invention

[0004] The purpose of this application is to provide a method and system for high-precision power transmission line monitoring based on BeiDou, in order to solve the technical problem in the prior art that the poor timeliness of historical laser point cloud data and the lag in real-time point cloud data updates lead to potential safety risks, thereby affecting the accuracy and safety of power transmission line monitoring.

[0005] In view of the above problems, this application provides a method and system for high-precision monitoring of power transmission lines based on BeiDou.

[0006] Firstly, this application provides a BeiDou-based high-precision transmission line monitoring method, which is implemented through a BeiDou-based high-precision transmission line monitoring system. The method includes: deploying a BeiDou dual-frequency RTK module on the top of the tower and activating a lidar that works in conjunction with the BeiDou dual-frequency RTK module; after positioning the lidar using the BeiDou dual-frequency RTK module, establishing a three-dimensional point cloud based on the lidar's measurement data, wherein the three-dimensional point cloud is fused using the lidar's positioning coordinates; performing point cloud alignment based on principal component analysis using the three-dimensional point cloud and historical point cloud data to complete coarse matching positioning; establishing tower position coordinates using the BeiDou dual-frequency RTK module and establishing a connection line between tower points based on the position coordinates; using the connection line to verify the trust of the three-dimensional point cloud and establish a point cloud trust level; based on the coarse matching positioning, performing authentication matching under joint authentication based on the point cloud trust level and the nearest point cloud; updating the transmission line point cloud based on the authentication matching result and generating an early warning.

[0007] Optionally, the centroids of the laser 3D point cloud and historical point cloud data are calculated separately, and the corresponding centroids are subtracted from the laser 3D point cloud and historical point cloud data to establish a centralized point cloud; the covariance matrix of the centralized point cloud is calculated, and eigenvalues ​​and eigenvectors are established after eigenvalue decomposition; the eigenvector corresponding to the largest eigenvalue is obtained, and the eigenvector is used as the principal direction to complete the coarse matching and positioning of the laser 3D point cloud and the historical point cloud data.

[0008] Optionally, all points in the laser 3D point cloud are parsed and obtained, and any two points are traversed and combined to establish a combination set; the bias verification of all combinations in the combination set is performed using the tower point connection line to establish a bias verification result; and the trust backtracking of each point is performed based on the bias verification result to complete the trust verification and establish the point cloud trust level.

[0009] Optionally, a minimum threshold verification of point cloud trust is performed to obtain anomaly trust point cloud identifiers; location correlation analysis of the aomaly point cloud is performed using the aomaly trust point cloud identifiers, and regional anomalous attention is constructed using the location correlation analysis results and the aomaly trust point cloud identifiers; additional observation point resampling is performed for regional anomalous attention to establish additional observation point resampling results; the laser 3D point cloud is updated using the additional observation point resampling results, and the point cloud trust is updated synchronously.

[0010] Optionally, each point in the coarsely matched laser 3D point cloud is used as the matching center, and a neighbor matching search is performed on the historical point cloud data through nearest neighbor search to establish a neighbor matching search result; a distance error index is calculated based on the neighbor matching search result, and a joint authentication score is performed using the distance error index and the point cloud trust level, and the authentication matching is completed using the joint authentication score.

[0011] Optionally, the historical point cloud data is used to perform feature recognition of local areas, and adaptive segmentation of local areas is generated based on the feature recognition results; the adaptive segmentation of local areas is used to extract local features of the region and establish matching features; the matching features are used to perform similar feature matching of laser 3D point clouds and establish multi-scale assisted matching results; and authentication matching compensation is performed based on the multi-scale assisted matching results.

[0012] Optionally, the spatial location loss of historical point cloud data and updated transmission line point cloud is calculated based on the authentication matching result; the spatial location loss is used to locate transmission line anomalies, and the early warning is established based on the anomaly value and the anomaly location.

[0013] Optionally, based on the updated transmission line point cloud, transmission line morphology anomaly identification is performed to establish an independent anomaly identification system; and transmission line anomaly location compensation is performed based on the independently identified anomaly.

[0014] Optionally, a matching database of authentication matching results is established, and the incremental learning matching model is updated with the matching database; the incremental learning matching model is used for feedback iteration to generate an optimized matching strategy; and subsequent laser 3D point cloud matching management is performed according to the optimized matching strategy.

[0015] Secondly, this application also provides a BeiDou-based high-precision transmission line monitoring system for executing the BeiDou-based high-precision transmission line monitoring method described in the first aspect. The BeiDou-based high-precision transmission line monitoring system includes: a BeiDou deployment module for deploying a BeiDou dual-frequency RTK module on the top of the tower and activating a lidar that works in conjunction with the BeiDou dual-frequency RTK module; and a coordinate fusion module for using the BeiDou dual-frequency RTK module to locate the lidar and then establishing a three-dimensional laser point cloud based on the lidar's measurement data. The three-dimensional laser point cloud is obtained by using the lidar's positioning coordinates. The system includes: a line coordinate fusion module; a coarse matching positioning module, used to perform point cloud alignment based on principal component analysis according to the laser 3D point cloud and historical point cloud data to complete coarse matching positioning; a trust establishment module, used to establish tower position coordinates using a dual-frequency RTK module, and establish tower point connections based on the position coordinates, and use the tower point connections to verify the trust of the laser 3D point cloud to establish point cloud trust; and an authentication matching module, used to perform authentication matching under joint authentication based on the point cloud trust and the nearest point cloud, based on the coarse matching positioning, update the transmission line point cloud according to the authentication matching results, and generate an early warning.

[0016] One or more technical solutions provided in this application have at least the following beneficial effects:

[0017] A BeiDou dual-frequency RTK module is deployed on the top of the tower, and a lidar working in conjunction with the BeiDou dual-frequency RTK module is activated. After the lidar is located using the BeiDou dual-frequency RTK module, a three-dimensional point cloud is built based on the lidar's measurement data. The three-dimensional point cloud is fused using the lidar's positioning coordinates. Point cloud alignment based on principal component analysis is performed using the three-dimensional point cloud and historical point cloud data to complete coarse matching positioning. The tower position coordinates are established using the dual-frequency RTK module, and a line connecting the tower points is established based on the position coordinates. The trust verification of the three-dimensional point cloud is performed using the line connecting the tower points to establish the point cloud trust level. Based on the coarse matching positioning, authentication matching is performed under joint authentication based on the point cloud trust level and the nearest point cloud. The transmission line point cloud is updated based on the authentication matching results, and an early warning is generated. In other words, by deploying a BeiDou dual-frequency RTK module on the top of the tower and positioning it with lidar, a three-dimensional point cloud of the lidar is established. Principal component analysis is used to align the three-dimensional point cloud of the lidar with historical point cloud data to complete coarse matching positioning. Joint authentication and authentication matching are performed using point cloud trust and the nearest point cloud to update the transmission line point cloud and generate early warnings. This avoids inaccurate updates of historical point cloud data and improves the response speed and security of transmission line monitoring.

[0018] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating the BeiDou-based high-precision transmission line monitoring method of this application.

[0021] Figure 2 This is a schematic diagram of the structure of the BeiDou-based high-precision power transmission line monitoring system of this application.

[0022] Figure labeling: BeiDou deployment module 11, coordinate fusion module 12, coarse matching positioning module 13, trust establishment module 14, authentication matching module 15. Detailed Implementation

[0023] This application provides a BeiDou-based high-precision transmission line monitoring method and system, solving the technical problem in existing technologies where the poor timeliness of historical laser point cloud data and the lag in real-time point cloud data updates lead to potential security risks, thus affecting the accuracy and security of transmission line monitoring. By deploying a BeiDou dual-frequency RTK module on the top of the tower and using lidar for positioning, a three-dimensional laser point cloud is established. Principal component analysis is used to align the three-dimensional laser point cloud with historical point cloud data, achieving coarse matching positioning. Joint authentication and authentication matching are performed using point cloud trust and the nearest point cloud, updating the transmission line point cloud and generating early warnings. This avoids inaccurate updates of historical point cloud data, improving the response speed and security of transmission line monitoring.

[0024] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. It should be understood that this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. It should also be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all of them.

[0025] Example 1, please refer to the appendix. Figure 1 This application provides a BeiDou-based high-precision transmission line monitoring method, wherein the BeiDou-based high-precision transmission line monitoring method is executed by a BeiDou-based high-precision transmission line monitoring system, and the BeiDou-based high-precision transmission line monitoring method specifically includes the following steps:

[0026] S100: Deploy a BeiDou dual-frequency RTK module at the top of the tower and activate the lidar that works in conjunction with the BeiDou dual-frequency RTK module.

[0027] Specifically, a BeiDou dual-frequency RTK module is deployed at the top of the tower to receive signals from BeiDou satellites and determine the tower's precise position coordinates. Simultaneously, a lidar system working in conjunction with the BeiDou dual-frequency RTK module is activated. The lidar generates 3D point cloud data of the surrounding environment by measuring the return time of laser pulses. The lidar operates based on optical reflection and ranging. A laser pulse emitted by the transmitter illuminates the target object and is then reflected back to the receiver. By measuring the time it takes for the laser to travel from emission to return, the distance to the target can be calculated, since the speed of light is known. The lidar works in conjunction with the BeiDou dual-frequency RTK module, using RTK positioning to provide precise position coordinates for the lidar, ensuring that every point cloud data point acquired by the lidar corresponds to a precise location.

[0028] The BeiDou dual-frequency RTK module is a positioning module that supports dual-frequency signal reception. It can receive satellite signals from two different frequency bands, eliminating ionospheric delay through dual-frequency observation and improving positioning accuracy. Combined with real-time dynamic differential (RTK) technology, it achieves centimeter-level or even higher positioning accuracy by receiving signals from BeiDou satellites and other satellite systems. For example, after being deployed on the top of a power transmission line tower, the BeiDou dual-frequency RTK module provided location data with an accuracy of 1 to 3 centimeters in a complex mountainous environment, significantly improving the accuracy compared to the 2 to 5-meter accuracy of the traditional GPS system.

[0029] The BeiDou dual-frequency RTK module provides high-precision location information to determine the device's geographical location. LiDAR generates high-precision 3D environmental data to perceive the surrounding environment of the tower, including terrain, obstacles, and equipment status. Through the collaborative work of LiDAR and the RTK module, high-precision point cloud data is acquired in real time, with each point cloud data corresponding to a precise geographic coordinate. By fusing the location data from the RTK module with the 3D point cloud data from the LiDAR, high-precision spatial positioning and environmental modeling are achieved.

[0030] S200: After using the Beidou dual-frequency RTK module to locate the lidar, a three-dimensional point cloud of the lidar is established based on the measurement data of the lidar. The three-dimensional point cloud of the lidar is fused with the positioning coordinates of the lidar.

[0031] Specifically, the BeiDou dual-frequency RTK module receives satellite signals in real time and calculates the precise location of the lidar. In other words, the BeiDou dual-frequency RTK module achieves centimeter-level high-precision positioning by receiving signals from two different frequency bands of BeiDou satellites, thus determining the lidar's specific location coordinates. The lidar scans the environment to acquire distance data from different points, thereby creating a three-dimensional laser point cloud. Each laser reflection point is assigned a three-dimensional coordinate, typically represented by X, Y, and Z. The point cloud is large and dense, accurately depicting the three-dimensional shape of objects.

[0032] A laser 3D point cloud is a data set generated when a lidar scanner scans the surrounding environment. Each point represents the spatial coordinates of a measurement location, accurately reflecting its shape, size, and position. Each point in the laser 3D point cloud corresponds to its three-dimensional spatial coordinates in the real world and the lidar reflection intensity at that point.

[0033] Point cloud data obtained by lidar scanning is typically based on measurements relative to a reference coordinate system. Coordinate fusion combines the lidar-measured point cloud data with precise positioning coordinates provided by the RTK module, ensuring that each point in the point cloud corresponds to an accurate geographic coordinate. This makes the lidar measurements consistent with the actual location in the environment, essentially mapping the local coordinates acquired by the lidar to a global coordinate system. With the centimeter-level positioning accuracy provided by the BeiDou dual-frequency RTK module, each point cloud data point acquired by the lidar has precise geographic coordinates, guaranteeing high accuracy of the monitoring data and avoiding the inaccuracies caused by signal interference and errors in traditional lidar positioning.

[0034] S300: Perform point cloud alignment based on principal component analysis according to the laser 3D point cloud and historical point cloud data to complete coarse matching and positioning.

[0035] Furthermore, this application S300 includes:

[0036] The centroids of the laser 3D point cloud and historical point cloud data are calculated separately. The corresponding centroids of the laser 3D point cloud and historical point cloud data are subtracted to establish a centralized point cloud. The covariance matrix of the centralized point cloud is calculated, and eigenvalues ​​and eigenvectors are established after eigenvalue decomposition. The eigenvector corresponding to the largest eigenvalue is obtained, and the eigenvector is used as the principal direction to complete the coarse matching and positioning of the laser 3D point cloud and the historical point cloud data.

[0037] Specifically, a 3D laser point cloud is acquired from a LiDAR scanner and coarsely matched with historical point cloud data to ensure both point clouds are in the same coordinate system and their positions are roughly correct. For both real-time and historical point clouds, principal component analysis (PCA) is first used to calculate their principal directions. Specifically, the covariance matrix is ​​calculated and eigenvalue decomposition is performed to extract the eigenvector corresponding to the largest eigenvalue, representing the principal direction of change of the point cloud. The centroids of both the real-time and historical point clouds are calculated and centered. The covariance matrix is ​​then calculated using the centered data, and eigenvalue and eigenvector decomposition is performed to obtain eigenvalues ​​and eigenvectors. The eigenvector corresponding to the largest eigenvalue is extracted as the principal direction. Based on the principal direction, the real-time and historical point clouds are rotated and translated to achieve preliminary coarse matching and positioning.

[0038] Specifically, the centroids of the laser-generated 3D point cloud and historical point cloud data are calculated separately. This is done by summing the coordinates of each point in both datasets and then dividing by the total number of points to obtain the centroid coordinates. The centroid of a point cloud is the average position of all points in the dataset, similar to the centroid of an object in geometry, representing the center of the point cloud data. The calculation process is C... x = C y = C z = , where C x C y C z These are the values ​​of the centroid coordinates on the x, y, and z axes, respectively. i It is the value of the i-th point cloud on the x-axis, y i It is the value of the i-th point cloud on the y-axis, z i Let be the value of the i-th point cloud on the z-axis, and n be the total number of points. By subtracting the corresponding centroid of the point cloud from the laser-generated 3D point cloud and historical point cloud data, a centered point cloud is obtained, which is (x... i -C x y i -C y , z i -C zCentered point cloud refers to the process of shifting the geometric center of the point cloud to the origin of the coordinate system by subtracting the coordinates of the centroid of the point cloud from the coordinates of each point, thereby eliminating the influence of coordinate offset. For example, if the coordinates of a point are (100, 200, 300) and its centroid is (50, 75, 120), then the centered coordinates of the point are (100-50, 200-75, 300-120) = (50, 125, 180).

[0039] The covariance matrix of the point cloud is calculated using a centralized point cloud. The covariance matrix, obtained by calculating the deviations and relationships between data points, describes the variations of the point cloud across its various dimensions (X, Y, Z), reflecting the distribution characteristics of the point cloud data. Using existing covariance calculation formulas, a covariance matrix is ​​generated, where each element represents the covariance between different coordinate axes. Eigenvalue decomposition is then performed on the covariance matrix, decomposing it into eigenvalues ​​and eigenvectors. Eigenvalues ​​represent the degree of variation of the data along the direction of the eigenvector, while the eigenvectors indicate the main direction of data variation.

[0040] The eigenvector corresponding to the largest eigenvalue is selected as the principal direction of the point cloud data from all eigenvalues ​​and eigenvectors. This determined principal direction represents the direction with the greatest variability in the data, indicating the strongest changes in the data, and is crucial for point cloud alignment and dimensionality reduction. Based on the principal direction, the real-time point cloud and historical point clouds are rotated and translated to achieve preliminary coarse matching and localization, eliminating angular and positional deviations between the two sets of point clouds and making their spatial relationships more consistent. In other words, based on the calculated principal direction, the two sets of point clouds are rotated respectively to make their principal directions consistent, aligning the main directions of change in the two sets of point clouds. This operation is typically achieved using a rotation matrix. After rotation and translation, the main features of the real-time point cloud and historical point cloud are aligned, achieving coarse matching and localization, i.e., approximate alignment of the two sets of point clouds.

[0041] For example, assuming the eigenvalues ​​obtained from the eigenvalue decomposition of the covariance matrix are λ1=6.3, λ2=3.5, and λ3=1.2, the corresponding eigenvectors are: v1=(0.8,0.6,0.1), v2=(0.3,-0.7,0.6), and v3=(0.5,0.2,-0.8). The eigenvector corresponding to the largest eigenvalue is selected as the principal direction of the point cloud data, i.e., v1=(0.8,0.6,0.1) is the principal direction. λ1, λ2, and λ3 reflect the magnitude of the data variance of the point cloud along the corresponding eigenvector direction, i.e., the degree of dispersion of the point cloud along that direction. The larger the value, the more drastic the change in the point cloud along that direction. v1, v2, and v3 represent the unit vectors of the direction of change of the point cloud data, which are orthogonal basis vectors in three-dimensional space, corresponding one-to-one with λ1, λ2, and λ3. The eigenvalue λ i Represents the point cloud along the feature vector v iVariance of direction; eigenvector v i This represents the spatial direction. By comparing the magnitudes of the eigenvalues, the largest λ is selected. i The corresponding v i This serves as the primary direction for the point cloud. Then, during point cloud registration, aligning the primary directions of the real-time and historical point clouds achieves coarse matching and positioning. Through coarse matching and positioning, the real-time point cloud has been successfully aligned with the historical point cloud, resolving the deviation caused by translation and achieving preliminary point cloud matching.

[0042] Principal component analysis (PCA) is a commonly used statistical method to reduce the dimensionality of data while preserving its main information. In point cloud alignment, PCA is used to identify the main directions of change in point cloud data and calculates the principal directions of the data by performing eigenvalue decomposition on the covariance matrix.

[0043] After coarse matching and localization, Multiple Novelty Kalman Filtering (MI-EKF) is used to filter the two point cloud data, thereby reducing the impact of noise data on the overall data. Kalman filtering optimizes state estimation by combining a weighted average of predictions and new observations, filtering out noise and improving the accuracy of point cloud data. In other words, it updates the state estimate based on the previous time-step estimate and new measurements. MI-EKF is an extension of traditional Kalman filtering, used to handle multi-dimensional state estimation, accurately filtering out noise data and improving estimation accuracy. MI-EKF improves state estimation by incorporating observation information from multiple time steps, using observation data from the current time step and several past time steps to update the state estimate, fully utilizing historical observation information to improve estimation accuracy and robustness.

[0044] After coarse matching and filtering, multiple nearest-point point cloud iterations are performed to further optimize the point cloud alignment results. In each iteration, the ICP algorithm finds the nearest point pair between two point clouds, calculates translation and rotation, and gradually improves the accuracy of point cloud registration. ICP is a commonly used point cloud registration algorithm that iteratively finds the nearest point pair between two point clouds and performs translation and rotation, gradually increasing the overlap between the two point clouds until it converges to the optimal registration result. For example, it calculates the nearest point pair between each point in the real-time point cloud and the historical point cloud, and then calculates the translation and rotation matrices by minimizing the distance between them, gradually improving the matching degree of the point clouds. Through repeated iterations, the error between the point clouds is minimized, and the registration result reaches its optimal state.

[0045] S400: The tower position coordinates are established using the Beidou dual-frequency RTK module, and the tower point connection is established based on the position coordinates. The trust verification of the laser three-dimensional point cloud is performed using the tower point connection to establish the point cloud trust level.

[0046] Furthermore, this application S400 includes:

[0047] All points in the laser 3D point cloud are parsed and obtained, and any two points are traversed and combined to establish a combination set; the bias verification of all combinations in the combination set is performed using the connection of the tower points to establish a bias verification result; the trust backtracking of each point is performed based on the bias verification result to complete the trust verification and establish the point cloud trust level.

[0048] Specifically, the precise position coordinates of each tower in three-dimensional space are established using a BeiDou dual-frequency RTK module. For example, assuming the position information of a certain tower is: longitude 120.123456°, latitude 30.654321°, and elevation 100.5 meters, precise positioning using the dual-frequency RTK module can obtain the specific coordinates of this tower in three-dimensional space. Based on the specific position coordinates of each tower, multiple tower positions are connected by lines in space to form a global reference structure. The lines connecting the tower points form a linear structure connecting the position coordinates of multiple towers, serving as the skeleton of the entire transmission line.

[0049] Software analysis is used to extract the coordinates of all points in a laser-generated 3D point cloud, typically obtaining the (x, y, z) coordinates of each point to determine its position in 3D space. Then, all points are iterated through and combined using any two points; that is, for each point in the point cloud, it is paired with all other points to form point pairs, creating a combination set. A combination set refers to the set formed by every pair of points in the point cloud data.

[0050] The bias verification process verifies the skew consistency of each combination within the tower point connection set, confirming the reasonableness of the relative positions of each pair of points and determining whether they deviate from the normal range, thus forming the bias verification result. Bias verification checks whether the point cloud combination conforms to a predetermined structure or pattern, thereby determining whether its relative position or orientation is correct. For example, the relative position between two points in the combination (such as point A and point B) is calculated and compared with the direction of the tower connection line. If the relative position deviation between points A and B is too large, it indicates that the spatial relationship between them does not meet expectations, therefore the point cloud confidence level is low. Bias consistency can be quantified by calculating the angular difference between the two points and the tower connection line.

[0051] Based on the biased validation results, trust backtracking is performed on each point. Trust backtracking is a process that judges the accuracy by checking the consistency of the point across different combinations. If a point's relative position is consistent with other points in multiple combinations with small deviations, the point cloud trust score for that point is high, i.e., closer to 1. If a point's relative position deviates significantly in some combinations, it indicates that the point may have a large error, and the point cloud trust score is low, closer to 0. Based on the trust backtracking results, a trust score value is assigned to each point. The trust score value is a numerical value between 0 and 1, representing the point's credibility. Through the biased validation and trust backtracking processes, point data with large errors can be effectively eliminated, improving the overall accuracy of the point cloud data. Verification by connecting to tower points ensures the consistency and reliability of the point cloud data within the global structure.

[0052] Furthermore, this application also includes the following steps:

[0053] Perform minimum threshold verification of point cloud trust to obtain anomaly trust point cloud identifiers; use the anomaly trust point cloud identifiers to perform location correlation analysis of the anomaly point cloud, and use the location correlation analysis results and anomaly trust point cloud identifiers to construct regional anomaly concern; perform additional observation point resampling for regional anomaly concern to establish additional observation point resampling results; use the additional observation point resampling results to update the laser 3D point cloud, and simultaneously update the point cloud trust.

[0054] Specifically, a minimum trust threshold is set to filter out point cloud data with a trust level below the minimum threshold, thus identifying and excluding unreliable point cloud data, i.e., distinguishing between trustworthy and abnormal point clouds. The minimum trust threshold verification is performed, which involves comparing the trust level of each point cloud with the set minimum trust threshold. Point clouds with a trust level below the minimum trust threshold are marked as abnormal trust point clouds, and their identifiers are recorded.

[0055] For each point marked as an anomaly, location correlation analysis is performed. By analyzing the location of the anomaly point and the spatial relationship between it and adjacent point clouds, it is determined whether these anomaly point clouds exhibit any regularity or relationship with the surrounding environment, thus identifying potential anomaly regions. In other words, if multiple anomaly points are found to be concentrated in one area, then this area is an anomaly region requiring feature focus. Regional anomaly focus refers to identifying potentially anomaly regions based on the results of location correlation analysis and anomaly confidence level point cloud identifiers, and then focusing on the point cloud data within these regions.

[0056] For identified anomalous areas, additional observation points are resampled. This resampling of the point cloud data within these areas adds more observation points, resulting in more accurate and stable data. Specifically, within the identified anomalous areas, the location and number of additional observation points are planned. LiDAR is used to resample these additional observation points, acquiring more detailed point cloud data. The resampled point cloud data is recorded and organized to obtain the additional observation point resampling results. Based on these results, the LiDAR 3D point cloud data is updated, replacing or supplementing existing anomalous points, and the point cloud confidence level is updated simultaneously. For example, in the resampled point cloud, points with initially low confidence levels will have their confidence level increased, such as from 0.6 to 0.8, due to the addition of more accurate observation data. Resampling increases the number of observation points, improving the accuracy of point cloud data in anomalous areas, effectively identifying and focusing on potential anomalous areas, thereby reducing the possibility of false alarms and missed alarms. Furthermore, targeted resampling avoids a complete rescan, saving resources and time.

[0057] S500: Based on coarse matching and positioning, authentication matching is performed under joint authentication according to the point cloud trust level and the nearest point cloud. The transmission line point cloud is updated according to the authentication matching result, and an early warning is generated.

[0058] Specifically, after registering historical point clouds and real-time point clouds, the system acquires the registered real-time point cloud, the historical point cloud to be updated, and the initial position information of the historical point cloud to be updated. Based on the reliability of the initial pose information of the historical point cloud data and / or the reliability of the pose information after registration, the system evaluates the usability of the registration result of the point cloud data. The registered point cloud replaces and covers the historical point cloud in the corresponding area, thus constructing a real-time update mechanism for the digital channel point cloud of the transmission line and completing the update of the point cloud.

[0059] Specifically, the coarse-matched real-time point cloud refers to real-time point cloud data that has undergone preliminary registration and is roughly aligned to the coordinate system of the historical point cloud, but some deviation may still exist. Using each point in the coarse-matched real-time point cloud as the center, nearest neighbor search is used to find adjacent matching points in the historical point cloud, and the matching distance error is calculated. Matching points are then found in the historical point cloud, and the spatial distance between the two point clouds is calculated, typically using Euclidean distance or other distance metrics. The matching distance error is then calculated. By comparing the distances between matching points, the quality of the point cloud matching can be evaluated. A small error indicates a good match; a large error may indicate an inaccurate match.

[0060] For each matching point pair, its distance error is calculated. Based on the distance error and point cloud confidence level, a joint authentication scoring model is constructed to select authenticated matching pairs that meet the scoring criteria. The joint authentication scoring model is a model that evaluates and scores the matching results by comprehensively considering the distance error and point cloud confidence level. It includes a joint authentication scoring function, which helps to select point cloud pairs with high matching quality and strong reliability, thereby improving the overall accuracy and reliability of the matching. The score is calculated by weighted averaging, combining the error and point cloud confidence level. The joint authentication scoring function is as follows: In this context, S represents the joint authentication score, ranging from [0,1]; T represents the point cloud trust score, ranging from [0,1]; d represents the normalized distance error; and α is a weighting coefficient used to balance the weights of trust score and distance error in the overall score, ranging from [0,1]. For example, assuming a point pair has a distance error d = 0.1m, a point cloud trust score T = 0.8, and a maximum trust score of 1, if the weighting coefficient α = 0.7, the joint authentication score would be 0.7 * 0.1 + (1 - 0.7) * 0.8 = 0.31.

[0061] To further improve matching accuracy, local adaptive segmentation is introduced to extract local multi-scale features and compensate for and correct the authentication matching results, eliminating deviations caused by errors or noise, thus further improving matching accuracy. Based on the final authentication matching results, high-quality real-time point cloud data is used to update the historical point cloud data of the corresponding area, forming a digital update of the transmission line. Updating historical point cloud data replaces the historical point cloud data of the corresponding area, ensuring that the point cloud data is always synchronized with the real-time situation. At the same time, statistical analysis is performed on areas with abnormal matching errors and point cloud confidence, generating early warning information, timely feedback on abnormal states of the transmission line, and guiding subsequent on-site inspections or resampling of additional observation points. By combining distance error, point cloud confidence, and multi-scale feature extraction, matching accuracy can be improved and the quality of point cloud data can be ensured.

[0062] Furthermore, this application S500 includes:

[0063] Using each point in the coarsely matched laser 3D point cloud as the matching center, a neighbor matching search is performed on the historical point cloud data through nearest neighbor search to establish a neighbor matching search result; a distance error index is calculated based on the neighbor matching search result, and a joint authentication score is performed using the distance error index and the point cloud trust level, and the authentication matching is completed using the joint authentication score.

[0064] Specifically, each point in the coarsely matched laser 3D point cloud is used as a matching center, that is, each point is used as a reference and compared with other adjacent point cloud data. Nearest neighbor search is used to find adjacent matching points in the historical point cloud. Each matching center is compared with historical point cloud data to find the closest point. Nearest neighbor search refers to finding the point with the closest spatial distance to each query point (matching center) in a point cloud. This can be done by calculating the Euclidean distance between each point and other points. The adjacent matching search result refers to the result of finding the point closest to the matching center after performing the nearest neighbor search.

[0065] For each pair of matching center points and adjacent matching points, the distance error index between them is calculated, typically measured using Euclidean distance. A smaller distance error indicates a better matching result. A joint certification score is then performed, combining distance error and point cloud trust level. This involves integrating multiple certification factors (such as distance error and point cloud trust level) and comprehensively evaluating the quality of point cloud matching through weighted averages or other methods. Based on the certification score, certified matching pairs that meet the scoring criteria are selected, thus determining which point cloud matches are reliable and which may be incorrect.

[0066] For example, assume that the real-time point cloud after coarse matching includes point A (5.2, 3.1, 10.5) with a point cloud confidence level of 0.85, point B (5.5, 3.4, 10.7) with a point cloud confidence level of 0.90, point C (5.8, 3.7, 10.9) with a point cloud confidence level of 0.88, and historical point cloud data includes point D (5.1, 3.2, 10.6) with a point cloud confidence level of 0.80, point E (5.6, 3.3, 10.8) with a confidence level of 0.87, and point F (6.0, 3.6, 11.0) with a point cloud confidence level of 0.82. Each point in the real-time point cloud is selected as the matching center. For point A, a nearest neighbor search is performed to find the point in the historical point cloud data that is closest to point A. The distance between point A and point D is 0.173, the distance between point A and point E is 0.374, and the distance between point A and point F is 0.931. The nearest neighbor is point D. The same steps are performed for point B, and the nearest neighbor of point B is point E. The nearest neighbor of point C is point F. The distance error is calculated for each pair of matching points: 0.173 for points A and F, 0.374 for points B and E, and 0.931 for points C and F. A joint authentication score is performed using the distance error index and point cloud trust level. A weighted formula is used to calculate the authentication score for each pair of point clouds. Assuming the weight of real-time point cloud trust level is 0.4, the weight of historical point cloud trust level is 0.4, and the weight of distance error is 0.2, the calculated joint authentication score for points A and D is 0.6254, the joint authentication score for points B and E is 0.6332, and the joint authentication score for points C and F is 0.4938. A threshold of 0.6 is set for the authentication score standard. If the authentication score is greater than or equal to 0.6, the point cloud match is considered valid; otherwise, the match is considered invalid. The valid matches determined after matching include points A and D, and points B and E.

[0067] By combining distance error and point cloud confidence, the matching quality of point clouds can be evaluated more accurately, thereby reducing false matches; by using joint authentication scoring, unreliable point cloud matches can be identified and eliminated, thereby improving overall stability.

[0068] Furthermore, this application also includes the following steps:

[0069] The historical point cloud data is used to perform feature recognition of local areas, and adaptive segmentation of local areas is generated based on the feature recognition results; local features of the adaptive segmented local areas are extracted to establish matching features; similar features of the laser 3D point cloud are matched using the matching features to establish multi-scale assisted matching results; authentication matching compensation is performed based on the multi-scale assisted matching results.

[0070] Specifically, to further improve matching accuracy, local region adaptive segmentation is introduced. Historical point cloud data is used for local region feature recognition, identifying regions with specific geometric or spatial features, including surface normals, curvature, planes, and corners. These features are used to identify structural elements or shapes within the point cloud. Features are extracted from historical point cloud data, including surface normal estimation, curvature calculation, and corner detection, for local region feature recognition. Normals are local geometric features of each point in the point cloud, describing the orientation of the surface near that point. Normal estimation is used to identify regular structures such as planes and straight lines. Curvature describes the degree of curvature of the surface surrounding each point in the point cloud. Regions with high curvature are typically corners or curved surfaces, such as curved pipes or curved iron frames. Corners are extremely significant local features in point cloud data, usually appearing at the boundaries of objects. For example, in historical point cloud data of a transmission line, normal estimation reveals significant changes in the normal direction in certain areas, as these areas include transmission towers and intersection structures. Next, curvature calculations revealed that the support surface of the transmission tower has a high curvature, meaning that the support surface is relatively curved.

[0071] Based on feature recognition results, local regions are divided into different sub-regions of the point cloud data. The core of adaptive segmentation is to determine the segmentation strategy based on the feature density, curvature, or normal direction of each local region. Unlike fixed segmentation, adaptive segmentation can be flexibly adjusted according to different scenarios and data characteristics. Different segmentation methods are selected based on the features of the point cloud, and even multiple methods can be applied to the same dataset. For example, normal direction segmentation can separate the support structure from the surrounding open area, while curvature segmentation can separate the support structure from the line itself (usually a straight or relatively flat structure), resulting in two different sub-regions: the tower region and the line region. By analyzing and adjusting the segmented local regions, it may be found that some regions are not segmented accurately enough, or some regions are over-segmented. The segmentation results can be further optimized through boundary smoothing, region merging, etc.

[0072] In adaptive segmentation of local regions, local feature extraction is performed to extract representative local features, such as the principal direction and principal components of each region, to establish matching features. These matching features are then used to perform similarity feature matching on the laser 3D point cloud, identifying the region in the laser point cloud most similar to historical point cloud features. Feature matching is performed at different scales to obtain more comprehensive matching results. Specifically, matching begins with large-scale features to find approximate matching regions. Then, the scale is gradually reduced to match smaller-scale features, improving matching accuracy. Multi-scale auxiliary matching results are used to compensate and correct previous authentication matching results; that is, matching results are optimized based on matching information at different scales, improving the accuracy and reliability of the matching.

[0073] Based on the results of multi-scale assisted matching, the aforementioned authentication matching results are compensated to offset errors in the matching process. For example, if the multi-scale assisted matching results indicate that certain points have high matching similarity across multiple scales, the authentication matching scores for these points can be increased to improve the reliability of the matching. Through adaptive segmentation and multi-scale matching, point clouds in different regions can be accurately matched in complex environments, reducing the impact of noise and external interference on the matching results.

[0074] Furthermore, this application also includes the following steps:

[0075] The spatial location loss of historical point cloud data and updated transmission line point cloud is calculated based on the authentication matching results; the spatial location loss is used to locate transmission line anomalies, and the early warning is established based on the anomaly value and the anomaly location.

[0076] Specifically, based on the final authentication matching results, the geometric difference between the two sets of point clouds (historical point cloud and updated point cloud) is calculated to obtain the spatial position loss. Spatial position loss refers to the spatial displacement or deformation error between the historical point cloud and the updated transmission line point cloud, which helps to assess the accuracy of point cloud matching and determine whether there are any abnormal deviations. For each pair of matched points (historical point cloud data and updated transmission line point cloud), their spatial position difference is calculated using the Euclidean distance formula. The spatial position loss of all point pairs is statistically analyzed, and the overall loss value or loss distribution is calculated.

[0077] Anomaly localization is performed based on determined spatial location loss. A predetermined threshold is set; spatial location loss exceeding this threshold is considered anomaly. Anomaly locations may be caused by factors such as equipment malfunction, environmental changes, or point cloud matching errors. A pre-defined location is established; if the spatial location loss of a matched point pair exceeds this threshold, the region corresponding to that point pair is considered anomaly. All point pairs are checked; if the loss exceeds the threshold, the location of that point pair is recorded as an anomaly location.

[0078] Based on the location of an anomaly and its spatial loss value, early warning information is generated, including the location of the anomaly, its severity, and suggested inspection or maintenance measures, prompting operators to take timely action. The purpose of the early warning is to remind operators to handle anomalies promptly and prevent potential safety risks. Early warning information can be sent to operators via SMS, email, and app push notifications to ensure timely response. For example, if an anomaly is detected near coordinates (10,20,30) with a spatial loss of 1.73, immediate inspection and maintenance are recommended. By calculating the spatial loss and locating the anomaly, abnormal changes in point cloud data, such as deformation of transmission towers or supports, can be detected in real time, allowing for timely early warnings and preventing power outages or safety accidents caused by equipment failure.

[0079] Furthermore, this application also includes the following steps:

[0080] Based on the updated transmission line point cloud, anomaly identification of transmission line morphology is performed, and an independent anomaly identification system is established; based on the independently identified anomaly, anomaly location compensation is performed.

[0081] Specifically, based on the updated transmission line point cloud, anomaly identification is performed. Point cloud data is used to analyze geometric changes in the transmission line, identifying sections that differ from the normal state. Independent anomaly identification refers to identifying abnormal changes unrelated to other areas or points during point cloud data analysis. This is typically achieved through morphological anomaly identification, independent of anomalies identified by other methods (such as spatial location loss). Geometric analysis is performed on the data to calculate indicators such as the curvature and twist of the transmission line, and these are compared with indicators under normal conditions. For example, if the curvature of a certain section of the line exceeds the normal range, then that section of the line can be considered to have a morphological anomaly.

[0082] For each point or small region in the updated transmission line point cloud, an independent anomaly analysis is performed. An anomaly score is assigned to each point or region; the higher the score, the more likely the point or region is to be anomaly. An anomaly threshold is set; if the anomaly score of a point or region exceeds the threshold, that region is considered an anomaly region, thus achieving independent anomaly identification.

[0083] Anomaly location compensation for transmission lines is performed based on independently identified anomalies, correcting or supplementing the anomaly location results identified earlier based on spatial location loss. For example, if morphological anomaly identification reveals severe distortion in a section of the line, and previous anomaly location results did not include this, then the anomaly location results need to be updated to include this new anomaly information. The compensated anomaly location also needs to be alerted, and relevant personnel should be notified promptly to prevent potential faults from escalating.

[0084] Furthermore, this application also includes the following steps:

[0085] A matching database of authentication matching results is established, and the incremental learning matching model is updated with the matching database; the incremental learning matching model is used for feedback iteration to generate an optimized matching strategy; and subsequent laser 3D point cloud matching management is performed according to the optimized matching strategy.

[0086] Specifically, based on the final authentication matching results, a matching database is established, containing the location information, matching quality (such as error and point cloud confidence level) of each matching pair, as well as other related feature data. In other words, the matching feature vector of each successfully matched point pair is extracted from the authentication matching results, a matching quality label is assigned to each matching feature vector, and the matching feature vector and matching quality label are stored one-to-one in the matching database. The matching database is updated regularly; as more point cloud data is collected and matched, the matching database will contain more and more valid matching information.

[0087] Training data, including positive samples (correctly matched point pairs) and negative samples (incorrectly matched point pairs), is extracted from the matching database. The extracted training data is then used to train the incremental learning matching model. The matching accuracy of the incremental learning matching model is improved by iteratively optimizing its parameters. The incremental learning matching model is a model that continuously optimizes itself during point cloud matching, gradually improving matching accuracy and efficiency by continuously receiving new matching pair data.

[0088] For example, multiple matching point pairs are obtained from the authentication matching results. Each matching point pair is labeled, with samples having a joint authentication score greater than or equal to 0.8 designated as positive samples and samples with a joint authentication score less than 0.4 designated as negative samples, ensuring that the ratio of positive to negative samples is close to 1:1, thus obtaining the training set. A random forest model is initialized, with the following forest parameters set: number of trees 100, maximum tree depth 15, minimum number of split samples per node 20, and feature subset size 3. Samples with replacement are performed from the training set to obtain a training subset. A decision tree is constructed based on the training subset, recursively splitting nodes starting from the root node. At each node, the optimal split point is selected from three random features, with the splitting criterion being minimization of Gini impurity (i.e., selecting the split point with the lowest Gini impurity). When a stopping condition is met, such as when the maximum tree depth exceeds 15 layers or the minimum number of split samples per node is less than 20, the current node stops splitting, generating a leaf node and storing its class distribution. The decision trees corresponding to all training subsets are combined to form the initial incremental learning matching model. The performance of the incremental learning matching model is evaluated using 5-fold cross-validation, requiring an initial model accuracy of ≥85%. If this target is not met, the hyperparameters are adjusted and the model is retrained. Simultaneously, authentication matching results are collected in real-time to construct a matching database, which is then used to incrementally update the incremental learning matching model. Specifically, based on the matching database, an incremental dataset is determined, and 50 samples are randomly selected from this dataset as update data. Starting from the root node, each new sample is passed down the tree to the corresponding leaf node. For each leaf node, the statistical information of the leaf node is updated, including sample counts, category distribution, and feature statistics. Based on the updated leaf node statistics, the matching probability prediction is adjusted, and a new optimized matching strategy is output, including a matching error threshold, feature weights, and confidence weights. The updated incremental learning matching model is used to predict subsequent point cloud matching, and the matching results are stored back into the matching database. This incremental update process is repeated cyclically to achieve adaptive optimization.

[0089] A feedback mechanism is established to feed new matching results back into the matching database. The database is enriched by continuously accumulating new matching data. Each time new matching data enters the database, it is analyzed using an incremental learning matching model to identify which matching features significantly impact the results, and the model is updated based on these features. The incremental learning matching model continuously optimizes the matching process based on historical data, authentication results, and feedback information, enabling it to adapt to new point cloud characteristics and improve matching accuracy. Based on the feedback results from the incremental learning matching model, the effectiveness of the current matching strategy is analyzed, and potential areas for improvement are identified. Based on new data and matching results, an optimized matching strategy is generated, including adjustments to the matching threshold, optimization of feature selection, and improvements to the matching algorithm. This optimized strategy guides subsequent matching processes, making matching more efficient and accurate. Optimizing the matching strategy refers to adjusting the matching algorithm and parameters during the matching process based on the optimization results of the incremental learning matching model to improve matching efficiency and accuracy. By continuously optimizing the matching strategy through incremental learning and feedback iteration, the accuracy of point cloud matching is continuously improved, significantly enhancing the efficiency of updating and managing transmission line point cloud data, and solving the problems of the lag in the timeliness of historical laser point cloud data and the difficulty in timely updating and synchronizing real-time point cloud data.

[0090] In summary, the BeiDou-based high-precision transmission line monitoring method provided in this application has at least the following beneficial effects:

[0091] A BeiDou dual-frequency RTK module is deployed on the top of the tower, and a lidar working in conjunction with the BeiDou dual-frequency RTK module is activated. After the lidar is located using the BeiDou dual-frequency RTK module, a three-dimensional point cloud is established based on the lidar's measurement data. The three-dimensional point cloud is fused using the lidar's positioning coordinates. Point cloud alignment based on principal component analysis is performed using the three-dimensional point cloud and historical point cloud data to complete coarse matching positioning. The tower position coordinates are established using the BeiDou dual-frequency RTK module, and a line connecting the tower points is established based on the position coordinates. The line connecting the tower points is used to verify the trust of the three-dimensional point cloud and establish the point cloud trust level. Based on the coarse matching positioning, authentication matching is performed under joint authentication based on the point cloud trust level and the nearest point cloud. The transmission line point cloud is updated based on the authentication matching results, and an early warning is generated. In other words, by deploying a BeiDou dual-frequency RTK module on the top of the tower and positioning it with lidar, a three-dimensional point cloud of the lidar is established. Principal component analysis is used to align the three-dimensional point cloud of the lidar with historical point cloud data to complete coarse matching positioning. Joint authentication and authentication matching are performed using point cloud trust and the nearest point cloud to update the transmission line point cloud and generate early warnings. This avoids inaccurate updates of historical point cloud data and improves the response speed and security of transmission line monitoring.

[0092] Example 2: Based on the same inventive concept as the BeiDou-based high-precision transmission line monitoring method in Example 1, this application also provides a BeiDou-based high-precision transmission line monitoring system. Please refer to the appendix. Figure 2 The BeiDou-based high-precision transmission line monitoring system includes:

[0093] The system comprises the following modules: a BeiDou deployment module 11, used to deploy a BeiDou dual-frequency RTK module on the top of the tower and activate a lidar that works in conjunction with the BeiDou dual-frequency RTK module; a coordinate fusion module 12, used to establish a three-dimensional laser point cloud based on the lidar measurement data after positioning the lidar using the BeiDou dual-frequency RTK module, wherein the three-dimensional laser point cloud is fused using the lidar's positioning coordinates; a coarse matching positioning module 13, used to perform point cloud alignment based on principal component analysis using the three-dimensional laser point cloud and historical point cloud data to complete coarse matching positioning; a trust establishment module 14, used to establish the tower position coordinates using the BeiDou dual-frequency RTK module, establish a line connecting the tower points based on the position coordinates, and use the line connecting the tower points to verify the trust of the three-dimensional laser point cloud and establish the point cloud trust level; and an authentication matching module 15, used to perform authentication matching under joint authentication based on the point cloud trust level and the nearest point cloud, update the transmission line point cloud based on the authentication matching result, and generate an early warning.

[0094] Furthermore, the coarse matching positioning module 13 in the BeiDou-based high-precision transmission line monitoring system is also used for:

[0095] The centroids of the laser 3D point cloud and historical point cloud data are calculated separately. The corresponding centroids of the laser 3D point cloud and historical point cloud data are subtracted to establish a centralized point cloud. The covariance matrix of the centralized point cloud is calculated, and eigenvalues ​​and eigenvectors are established after eigenvalue decomposition. The eigenvector corresponding to the largest eigenvalue is obtained, and the eigenvector is used as the principal direction to complete the coarse matching and positioning of the laser 3D point cloud and the historical point cloud data.

[0096] Furthermore, the trust establishment module 14 in the BeiDou-based high-precision transmission line monitoring system is also used for:

[0097] All points in the laser 3D point cloud are parsed and obtained, and any two points are traversed and combined to establish a combination set; the bias verification of all combinations in the combination set is performed using the connection of the tower points to establish a bias verification result; the trust backtracking of each point is performed based on the bias verification result to complete the trust verification and establish the point cloud trust level.

[0098] Furthermore, the trust establishment module 14 in the BeiDou-based high-precision transmission line monitoring system is also used for:

[0099] Perform minimum threshold verification of point cloud trust to obtain anomaly trust point cloud identifiers; use the anomaly trust point cloud identifiers to perform location correlation analysis of the anomaly point cloud, and use the location correlation analysis results and anomaly trust point cloud identifiers to construct regional anomaly concern; perform additional observation point resampling for regional anomaly concern to establish additional observation point resampling results; use the additional observation point resampling results to update the laser 3D point cloud, and simultaneously update the point cloud trust.

[0100] Furthermore, the authentication and matching module 15 in the BeiDou-based high-precision transmission line monitoring system is also used for:

[0101] Using each point in the coarsely matched laser 3D point cloud as the matching center, a neighbor matching search is performed on the historical point cloud data through nearest neighbor search to establish a neighbor matching search result; a distance error index is calculated based on the neighbor matching search result, and a joint authentication score is performed using the distance error index and the point cloud trust level, and the authentication matching is completed using the joint authentication score.

[0102] Furthermore, the authentication and matching module 15 in the BeiDou-based high-precision transmission line monitoring system is also used for:

[0103] The historical point cloud data is used to perform feature recognition of local areas, and adaptive segmentation of local areas is generated based on the feature recognition results; local features of the adaptive segmented local areas are extracted to establish matching features; similar features of the laser 3D point cloud are matched using the matching features to establish multi-scale assisted matching results; authentication matching compensation is performed based on the multi-scale assisted matching results.

[0104] Furthermore, the authentication and matching module 15 in the BeiDou-based high-precision transmission line monitoring system is also used for:

[0105] The spatial location loss of historical point cloud data and updated transmission line point cloud is calculated based on the authentication matching results; the spatial location loss is used to locate transmission line anomalies, and the early warning is established based on the anomaly value and the anomaly location.

[0106] Furthermore, the authentication and matching module 15 in the BeiDou-based high-precision transmission line monitoring system is also used for:

[0107] Based on the updated transmission line point cloud, anomaly identification of transmission line morphology is performed, and an independent anomaly identification system is established; based on the independently identified anomaly, anomaly location compensation is performed.

[0108] Furthermore, the authentication and matching module 15 in the BeiDou-based high-precision transmission line monitoring system is also used for:

[0109] A matching database of authentication matching results is established, and the incremental learning matching model is updated with the matching database; the incremental learning matching model is used for feedback iteration to generate an optimized matching strategy; and subsequent laser 3D point cloud matching management is performed according to the optimized matching strategy.

[0110] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Figure 1 The BeiDou-based high-precision transmission line monitoring method and specific examples in Example 1 are also applicable to the BeiDou-based high-precision transmission line monitoring system in this example. Through the foregoing detailed description of the BeiDou-based high-precision transmission line monitoring method, those skilled in the art can clearly understand the BeiDou-based high-precision transmission line monitoring system in this example. Therefore, for the sake of brevity, it will not be described in detail here.

[0111] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0112] Obviously, those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A power transmission line monitoring method based on high-precision Beidou, characterized in that, The application relates to a method for establishing a power transmission line point cloud based on a tower pole, and belongs to the technical field of power transmission line point cloud establishment. The method comprises the following steps: a Beidou double-frequency RTK module is arranged at the top of a tower pole, and a laser radar working in cooperation with the Beidou double-frequency RTK module is activated; after the laser radar is positioned by the Beidou double-frequency RTK module, a laser three-dimensional point cloud is established according to the measurement data of the laser radar, wherein the laser three-dimensional point cloud is fused by the positioning coordinates of the laser radar; principal component analysis-based point cloud alignment is carried out according to the laser three-dimensional point cloud and historical point cloud data, and coarse matching positioning is completed, which comprises the following steps: the point cloud centroids of the laser three-dimensional point cloud and the historical point cloud data are respectively calculated, and the laser three-dimensional point cloud and the historical point cloud data are subtracted by the corresponding point cloud centroids, so that the centerized point cloud is established; the covariance matrix is calculated by using the centerized point cloud, and the eigenvalues and eigenvectors are established after the characteristic decomposition; the eigenvector corresponding to the maximum eigenvalue is obtained, and the coarse matching positioning of the laser three-dimensional point cloud and the historical point cloud data is completed by taking the eigenvector as the principal direction; the position coordinates of the tower pole are established by using the Beidou double-frequency RTK module, the tower pole point connecting line is established according to the position coordinates, the trust verification of the laser three-dimensional point cloud is carried out by using the tower pole point connecting line, and the point cloud trust degree is established, which comprises the following steps: all points in the laser three-dimensional point cloud are analyzed and obtained, and the traversal combination of any two points is carried out, so that a combination set is established; the bias verification result of all combinations in the combination set is established by using the tower pole point connecting line; 2. The power transmission line monitoring method based on high-precision Beidou according to claim 1, characterized in that, the trust of each point is traced back according to the bias verification result, so that the trust verification is completed, and the point cloud trust degree is established; on the basis of the coarse matching positioning, the authentication matching under the joint authentication of the point cloud trust degree and the nearest point point cloud is carried out, the power transmission line point cloud is updated according to the authentication matching result, and a pre-warning is generated. after the point cloud trust degree is established, the following steps are further included: the minimum threshold verification of the point cloud trust degree is carried out, and the abnormal trust degree point cloud identifier is obtained; the position correlation analysis of the abnormal point cloud is carried out by using the abnormal trust degree point cloud identifier, the regional abnormal attention is constructed by using the position correlation analysis result and the abnormal trust degree point cloud identifier; 3.The power transmission line monitoring method based on high-precision Beidou according to claim 1, wherein, the additional observation point resampling of the regional abnormal attention is carried out, and the additional observation point resampling result is established; the laser three-dimensional point cloud is updated by using the additional observation point resampling result, and the point cloud trust degree is synchronously updated. on the basis of the coarse matching positioning, the authentication matching under the joint authentication of the point cloud trust degree and the nearest point point cloud comprises the following steps:

4. The power transmission line monitoring method based on high-precision Beidou according to claim 1, characterized in that, each point in the laser three-dimensional point cloud after the coarse matching is taken as a matching center, the adjacent matching search of the historical point cloud data is carried out by using the nearest neighbor search, and the adjacent matching search result is established; the distance error index is calculated according to the adjacent matching search result, the joint authentication score is carried out by using the distance error index and the point cloud trust degree, and the authentication matching is completed by using the joint authentication score. on the basis of the coarse matching positioning, the authentication matching under the joint authentication of the point cloud trust degree and the nearest point point cloud further comprises the following steps: the feature recognition of a local region is carried out by using the historical point cloud data, the self-adaptive segmented local region is generated according to the feature recognition result; the regional local feature extraction is carried out by using the self-adaptive segmented local region, and the matching feature is established. The matching features are used for similar feature matching of laser three-dimensional point clouds, and a multi-scale auxiliary matching result is established. Authentication matching compensation is performed based on the multi-scale auxiliary matching result.

5. The power transmission line monitoring method based on high-precision Beidou according to claim 1, characterized in that, The power line point cloud is updated according to the authentication matching result, and a warning is generated, including: The spatial position loss of the historical point cloud data and the updated power line point cloud is calculated according to the authentication matching result; The spatial position loss is used for power line anomaly positioning, and the warning is established according to the abnormal value and the abnormal position.

6. The power transmission line monitoring method based on high-precision Beidou according to claim 5, characterized in that, The spatial position loss is used for power line anomaly positioning, including: The power line form anomaly is identified based on the updated power line point cloud, and an independent identified anomaly is established; The power line anomaly positioning compensation is performed according to the independent identified anomaly.

7. The power transmission line monitoring method based on high-precision Beidou according to claim 1, characterized in that, After the power line point cloud is updated according to the authentication matching result, including: A matching database of authentication matching results is established, and the matching database is used to update an incremental learning matching model; The feedback iteration is performed using the incremental learning matching model to generate an optimized matching strategy; The subsequent laser three-dimensional point cloud matching management is performed according to the optimized matching strategy.

8. The power transmission line monitoring system based on high-precision Beidou, characterized in that, The steps of the power line monitoring method based on the Beidou high-precision in any one of claims 1 to 7 are implemented, and the power line monitoring system based on the Beidou high-precision includes: A Beidou deployment module is used to deploy a Beidou dual-frequency RTK module on the top of the tower pole, and activate a laser radar that works cooperatively with the Beidou dual-frequency RTK module; A coordinate fusion module is used to establish a laser three-dimensional point cloud according to the measurement data of the laser radar after the laser radar is positioned by the Beidou dual-frequency RTK module, wherein the laser three-dimensional point cloud is fused by the positioning coordinates of the laser radar; A coarse matching positioning module is used to perform point cloud alignment based on principal component analysis according to the laser three-dimensional point cloud and the historical point cloud data to complete coarse matching positioning; A trust degree establishment module is used to establish a tower pole position coordinate by the dual-frequency RTK module, and establish a tower pole point connecting line according to the position coordinate, and perform trust verification of the laser three-dimensional point cloud by the tower pole point connecting line to establish a point cloud trust degree; An authentication matching module is used to perform authentication matching under joint authentication according to the point cloud trust degree and the nearest point point cloud based on the coarse matching positioning, update the power line point cloud according to the authentication matching result, and generate a warning.

Images