Power transmission line windage yaw galloping alarm method and system based on multi-view three-dimensional monitoring

Abstract

The invention provides a multi-view three-dimensional monitoring power transmission line wind deviation galloping alarm method and system, and relates to the technical field of power transmission line early warning, and the method comprises the steps: obtaining a real-time monitoring image of a target wire, and carrying out the distortion error calibration analysis, and obtaining a calibration image; inputting real-time point cloud data of the calibration image into the registration fusion model to obtain output information including a model parameter estimation result; constructing a real-time three-dimensional model, and comparing and analyzing the real-time three-dimensional model and a predetermined three-dimensional model to obtain a target windage yaw coefficient; judging whether the target windage yaw coefficient accords with a preset windage yaw threshold value or not; if not, an early warning signal is sent out for early warning. According to the method, the technical problems of inaccurate identification of the windage yaw and poor early warning reliability caused by insufficient point cloud processing precision and lack of a refined comparison strategy in the prior art are solved, the accuracy and reliability of monitoring of the windage yaw are improved, and safe and stable operation of a power transmission line is guaranteed.

Description

Technical Field

[0001] This application relates to the field of transmission line early warning technology, specifically to a method and system for alarming transmission line wind deflection and galloping using multi-view three-dimensional monitoring. Background Technology

[0002] Wind-induced galloping of transmission lines refers to the periodic vibration or deviation of transmission conductors under wind force. This can lead to excessive conductor vibration, line contact, wear, or breakage, seriously threatening the safe and stable operation of the power system. Therefore, monitoring and early warning of wind-induced galloping of transmission lines has become a crucial technical requirement in the power industry. Traditional methods often employ sensor deployment for monitoring, but these methods are complex, costly, and cannot comprehensively cover the entire line, limiting monitoring accuracy and timeliness. To overcome these shortcomings, in recent years, multi-view 3D monitoring technology has been applied to wind-induced galloping alarms for transmission lines. This technology utilizes multi-view monitoring equipment to acquire line image data and generate 3D models for wind deviation analysis.

[0003] However, existing methods have significant shortcomings in point cloud data comparison. On the one hand, traditional point cloud data fusion relies heavily on simple image registration methods, which struggle to effectively handle the accumulation of errors from multi-view data, leading to a decrease in the accuracy of the 3D model. On the other hand, existing comparison strategies are mostly limited to overall model matching, lacking detailed analysis of local areas and failing to accurately capture minute wind deflection changes in transmission lines, especially when the wind deflection amplitude is small, resulting in more prominent issues of missed and false alarms. Furthermore, they delay the triggering of early warning signals, affecting the rapid response to wind deflection galloping. Summary of the Invention

[0004] This application provides a method and system for multi-view three-dimensional monitoring of transmission line wind deflection galloping alarm, which solves the technical problems in the prior art, such as inaccurate identification of transmission line wind deflection galloping and poor reliability of early warning due to insufficient point cloud processing accuracy and lack of refined comparison strategies. It achieves the technical effect of improving the accuracy and reliability of transmission line wind deflection galloping monitoring and ensuring the safe and stable operation of transmission lines.

[0005] In view of the above problems, this application provides a method for alarming wind-induced galloping of transmission lines using multi-view three-dimensional monitoring. The method includes: obtaining a real-time monitoring image of a target conductor through dynamic monitoring by a multi-view monitoring device, wherein the real-time monitoring image includes a first image monitored by a first view, and the first view refers to any one view in the multi-view monitoring device; introducing a multi-view measurement distortion error calibration function to perform distortion error calibration analysis on the first image to obtain a first calibration image, and constructing a real-time calibration image set of the multi-view monitoring device; using the real-time point cloud dataset of the real-time calibration image set as input information of a registration and fusion model, and obtaining output information through the registration and fusion model, wherein the output information includes model parameter estimation results; constructing a real-time three-dimensional model of the target conductor based on the model parameter estimation results; comparing and analyzing the real-time three-dimensional model with a predetermined three-dimensional model of the target conductor to obtain a target wind deflection coefficient; obtaining a predetermined wind deflection threshold and determining whether the target wind deflection coefficient meets the predetermined wind deflection threshold; if the target wind deflection coefficient does not meet the predetermined wind deflection threshold, issuing an early warning signal to provide an early warning of wind-induced galloping of the target conductor.

[0006] On the other hand, this application also provides a multi-view three-dimensional monitoring transmission line wind deflection and galloping alarm system, the system comprising: an image monitoring module, used to dynamically monitor and obtain real-time monitoring images of the target conductor through multi-view monitoring equipment, the real-time monitoring images including a first image monitored by a first view, the first view referring to any one of the multi-view monitoring equipment; an image calibration module, used to introduce a multi-view measurement distortion error calibration function to perform distortion error calibration analysis on the first image to obtain a first calibration image, and to construct a real-time calibration image set of the multi-view monitoring equipment; and a registration and fusion module, used to use the real-time point cloud dataset of the real-time calibration image set as a registration and fusion module. The system takes input information and obtains output information through the registration and fusion model, wherein the output information includes model parameter estimation results; a 3D model construction module is used to construct a real-time 3D model of the target conductor based on the model parameter estimation results; a comparison and analysis module is used to compare and analyze the real-time 3D model with a predetermined 3D model of the target conductor to obtain the target wind deflection coefficient; a judgment module is used to obtain a predetermined wind deflection threshold and determine whether the target wind deflection coefficient meets the predetermined wind deflection threshold; and an early warning module is used to issue an early warning signal if the target wind deflection coefficient does not meet the predetermined wind deflection threshold, thereby providing an early warning of wind deflection galloping of the target conductor.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages: Real-time monitoring images of the target conductor are obtained through dynamic monitoring using multi-view monitoring equipment, providing raw image data for subsequent analysis. A multi-view measurement distortion error calibration function is introduced to perform distortion error calibration analysis on the first image, eliminating or reducing distortion errors caused by factors such as the lens characteristics of the multi-view monitoring equipment, resulting in a first calibration image. A real-time calibration image set of the multi-view monitoring equipment is then constructed, making the subsequent image-based 3D model more accurate and improving data reliability. The real-time point cloud dataset of the real-time calibration image set is input into the registration and fusion model to accurately fuse data from different perspectives, obtaining model parameter estimation results. This provides comprehensive and accurate data support for constructing a 3D model of the transmission line, improving the accuracy and reliability of the 3D model. Based on the model parameter estimation results, a real-time 3D model of the target conductor is constructed, intuitively presenting information such as the spatial morphology of the transmission line, providing a visual basis for subsequent comparative analysis and early warning. The real-time 3D model is compared and analyzed with a predetermined 3D model of the target conductor, accurately analyzing the state changes of the transmission line from different angles, comprehensively evaluating the wind deflection galloping state of the transmission line, and obtaining the target wind deflection coefficient. A predetermined wind deflection threshold is obtained, and it is determined whether the target wind deflection coefficient meets the predetermined wind deflection threshold. If the target wind deflection coefficient does not meet the predetermined wind deflection threshold, an early warning signal is issued to provide an early warning of wind deflection galloping of the target conductor, so that relevant personnel can take measures to maintain or adjust the transmission line and avoid the occurrence of faults.

[0008] In summary, this application comprehensively acquires real-time images of transmission lines using multi-view monitoring equipment. After distortion error calibration and registration fusion model processing, a high-precision real-time 3D model is constructed. By comparing and analyzing the real-time 3D model with the predetermined 3D model, the target wind deflection coefficient is calculated, the wind deflection galloping state is accurately determined, and a timely warning signal is issued. This achieves efficient monitoring and accurate early warning of wind deflection galloping of transmission lines, significantly improving the accuracy of monitoring data and the timeliness of early warning, effectively preventing accidents caused by wind deflection galloping, and ensuring the safe and stable operation of transmission lines.

[0009] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0010] Figure 1 This is a flowchart illustrating a multi-view three-dimensional monitoring method for alarming wind-induced galloping of transmission lines, provided in an embodiment of this application.

[0011] Figure 2This is a flowchart illustrating the process of constructing a real-time three-dimensional model of the target conductor in a multi-view three-dimensional monitoring method for alarming wind-induced deviation and galloping of transmission lines, as provided in an embodiment of this application.

[0012] Figure 3 This is a flowchart illustrating the process of obtaining the target wind deflection coefficient in a multi-view three-dimensional monitoring method for alarming wind deflection and galloping of transmission lines, as provided in an embodiment of this application.

[0013] Figure 4 This is a schematic diagram of the structure of a multi-view three-dimensional monitoring transmission line wind deflection and galloping alarm system provided in an embodiment of this application.

[0014] Figure labeling: Image monitoring module 10, Image calibration module 20, Registration and fusion module 30, 3D model construction module 40, Comparison and analysis module 50, Judgment module 60, Early warning module 70. Detailed Implementation

[0015] This application provides a method and system for multi-view three-dimensional monitoring of transmission line wind deflection and galloping alarms. It solves the technical problems in the prior art where insufficient point cloud processing accuracy and lack of refined comparison strategies lead to inaccurate identification and poor early warning reliability of transmission line wind deflection and galloping. It achieves the technical effect of improving the accuracy and reliability of transmission line wind deflection and galloping monitoring and ensuring the safe and stable operation of transmission lines.

[0016] Example 1, as Figure 1 As shown in the figure, this application provides a method for alarming wind-induced galloping of transmission lines using multi-view three-dimensional monitoring. The method includes: Step S1: Obtain a real-time monitoring image of the target conductor through dynamic monitoring using a multi-view monitoring device. The real-time monitoring image includes a first image monitored by a first view, where the first view refers to any one view in the multi-view monitoring device.

[0017] Specifically, a multi-view monitoring device is a device composed of multiple monitoring cameras, each called a "view," capable of simultaneously acquiring image data of a target object from different angles. For example, a multi-view monitoring device with three lenses can capture images of a power transmission line from the left, center, and right angles. Real-time monitoring images refer to images obtained at the current moment through monitoring the target power line using the multi-view monitoring device. Real-time monitoring images are dynamically changing and can reflect the state of the target power line at different times. The first view is any one of the monitoring lenses in the multi-view monitoring device; the first image monitored by the first view refers to the image data acquired by any one of the cameras in the multi-view monitoring device during the monitoring process.

[0018] In power transmission line wind-induced galloping monitoring scenarios, multi-camera monitoring equipment is deployed. Once activated, each camera collects image data of the transmission lines in real time. For example, in monitoring a power transmission line in a mountainous area, the multi-camera monitoring equipment is installed on the transmission tower, with each lens pointing towards a different section of the transmission line. By acquiring images of the transmission lines from different angles through the multi-camera monitoring equipment, comprehensive raw data is provided for subsequent analysis of the transmission lines.

[0019] Step S2: Introduce a multi-view measurement distortion error calibration function to perform distortion error calibration analysis on the first image to obtain a first calibration image, and construct a real-time calibration image set for the multi-view monitoring device.

[0020] Specifically, the lenses of multi-view monitoring devices may cause image distortion during the imaging process due to factors such as optical principles, such as barrel distortion or pincushion distortion. Therefore, it is necessary to correct the monitoring images. The correction process is accomplished by introducing a multi-view measurement distortion error calibration function, which is a mathematical function used to correct image distortion errors in multi-view measurements. The distortion parameters of the first target (such as radial distortion coefficient, tangential distortion coefficient, etc.) are input into the multi-view measurement distortion error calibration function, and the correction algorithm is executed to obtain the first calibration image. For example, if the first image has barrel distortion, the function will adjust the stretched or compressed pixels at the image edges to the correct positions according to the characteristic parameters of the distortion. Similarly, all images captured by the multi-view cameras are calibrated, and the calibrated images are assembled into a real-time calibration image set.

[0021] Distortion error calibration analysis can effectively eliminate distortion errors in images, improve image accuracy, provide a high-quality image foundation for subsequent point cloud data extraction and model construction, and enhance the reliability of monitoring data.

[0022] Step S3: Use the real-time point cloud dataset of the real-time calibration image set as the input information of the registration fusion model, and obtain the output information through the registration fusion model, wherein the output information includes the model parameter estimation results.

[0023] Specifically, the real-time point cloud dataset refers to the 3D point cloud data extracted from the real-time calibration image set, containing the 3D spatial information of the transmission line at the monitoring time. The registration and fusion model is a model used to register and fuse point cloud data from different perspectives, integrating scattered point cloud data into a complete 3D model. The model parameter estimation result refers to the parameter estimates obtained by the registration and fusion model during processing, used to describe the registration and fusion status of the point cloud data.

[0024] The real-time calibration image set has completed unified correction of distortion errors and possesses a unified camera intrinsic parameter matrix. Based on any pair of adjacent viewpoint images in the image set, a stereo matching operation is performed. The semi-global block matching (SGBM) method is preferred to obtain a disparity map. Then, using the disparity formula Z=f*B / d, the depth value of each pixel is derived from the disparity map, thus forming a corresponding depth map. Here, Z represents the spatial depth value corresponding to the pixel, f represents the camera focal length, B represents the baseline length between adjacent cameras, and d represents the disparity value between pixels. Using this depth map and known camera intrinsic parameter information, the two-dimensional coordinates (u, v) of each image pixel are mapped to the corresponding three-dimensional spatial coordinates (X, Y, Z) using an image reprojection method, forming a local point cloud. The specific formula is as follows: X=(uc x )*Z / f x Y = (vc) y )*Z / f y Among them, c x and c y f represents the pixel coordinates of the camera's principal point. x and f y These are the focal length components of the camera in the horizontal and vertical directions. Furthermore, using the camera extrinsic parameters (including the rotation matrix R and the translation vector T), the local point clouds acquired from different image perspectives are uniformly transformed into the same coordinate system to obtain a real-time point cloud dataset, providing a complete and consistent spatial data foundation for subsequent input registration and fusion models.

[0025] The real-time point cloud dataset from the real-time calibration image set is used as input to the registration fusion model. The model utilizes image processing algorithms, such as the Iterative Closest Point (ICP) algorithm, feature-based registration algorithms (such as FPFH and SHOT), and deep learning-based registration algorithms, to align different point cloud datasets. Then, it fuses these point cloud datasets using methods such as weighted averaging to obtain the model parameter estimation results. For example, when processing point cloud data of transmission lines captured from two different angles, the registration fusion model first finds the common transmission tower feature points in both datasets, then aligns and fuses the two datasets according to these feature points to obtain output information. The output information includes model parameter estimation results, such as the rotation matrix and translation vector of the point cloud data; these model parameters describe the fusion state of the point cloud data.

[0026] By using registration and fusion model processing, point cloud data from different perspectives can be accurately fused to construct a more accurate and complete 3D model of the transmission line, improving the accuracy and reliability of the model and providing accurate point cloud data for subsequent comparative analysis and early warning.

[0027] Step S4: Construct a real-time three-dimensional model of the target traverse based on the model parameter estimation results.

[0028] Specifically, a real-time 3D model refers to a 3D model of a transmission line constructed based on the model parameter estimation results, which can reflect the state and position of the transmission line at the monitoring time in real time.

[0029] Based on the model parameter estimation results (rotation matrix, translation vector, etc.) obtained in step S3, the point cloud data is converted into a 3D model using 3D modeling software such as AutoCAD or SolidWorks, constructing a real-time 3D model of the target conductor. This real-time 3D model can intuitively display information such as the direction, curvature, and spatial location of the transmission conductor, providing a precise object for subsequent comparative analysis and helping to accurately determine whether the transmission conductor has experienced wind-induced galloping.

[0030] Step S5: Compare and analyze the real-time three-dimensional model with the predetermined three-dimensional model of the target conductor to obtain the target wind deflection coefficient.

[0031] Specifically, the predetermined 3D model is a pre-defined 3D model of the transmission line, representing its geometric and spatial characteristics under normal or expected conditions. The target wind deflection coefficient is a quantitative indicator obtained through comparative analysis to measure the degree of wind-induced conductor galloping. The real-time 3D model is compared with the predetermined 3D model of the target conductor, analyzing the differences in conductor shape, position, etc., between the two models to obtain the target wind deflection coefficient. For example, a uniform sampling algorithm is used to select representative sets of points or line segments from both models, matching each real-time sample point with its corresponding predetermined sample point, calculating the distance difference between the two sample points, obtaining multiple point-to-point distances, and calculating statistical indicators such as the mean, maximum, and variance of all point-to-point distances, or weighted comprehensive indicators, to obtain the target wind deflection coefficient.

[0032] Step S6: Obtain the predetermined wind deflection threshold and determine whether the target wind deflection coefficient meets the predetermined wind deflection threshold.

[0033] Specifically, the predetermined wind deflection threshold is a pre-set threshold for the wind deflection coefficient, used to determine whether wind-induced galloping reaches the warning standard. The predetermined wind deflection threshold is obtained, and it is then determined whether the target wind deflection coefficient meets the predetermined threshold. For example, if the predetermined wind deflection threshold is set to 0.8, and the target wind deflection coefficient is less than 0.8, the wind-induced galloping state is considered normal; if the target wind deflection coefficient is greater than 0.8, a risk of wind-induced galloping is considered to exist. By comparing the target wind deflection coefficient and the predetermined wind deflection threshold, it is possible to determine whether the wind-induced galloping state of the transmission line is within the normal range, providing a basis for issuing a warning.

[0034] Step S7: If the target wind deflection coefficient does not meet the predetermined wind deflection threshold, issue a warning signal to provide a warning of wind deflection galloping of the target conductor.

[0035] Specifically, if the target wind deflection coefficient does not meet the predetermined wind deflection threshold, it indicates that the wind-induced galloping of the transmission line exceeds the normal range. In this case, early warning equipment (such as audible and visual alarms or remote communication devices) is used to send a warning signal to the monitoring center. For example, in a transmission line monitoring system, when the target wind deflection coefficient exceeds the predetermined wind deflection threshold, the audible and visual alarm on the monitoring tower will light up and sound an alarm, while simultaneously sending the warning information to the remote monitoring center to remind maintenance personnel to perform inspections and maintenance. By issuing warning signals, relevant personnel can be promptly reminded to take appropriate measures, such as adjusting or maintaining the transmission line, to ensure the safe and stable operation of the transmission line and reduce the risk of accidents caused by wind-induced galloping.

[0036] Furthermore, step S2 includes: Step S21: Construct a distortion factor set and perform distortion analysis on the distortion factor set to obtain a distortion factor coefficient set.

[0037] Step S22: Based on the multi-view measurement distortion error calibration function and the distortion factor coefficient set, perform distortion error calibration analysis on the first image to obtain the first calibrated image. The expression for the multi-view measurement distortion error calibration function is as follows: ;in, The first image refers to With the first calibration image The multi-view measurement distortion error calibration function between them, It refers to the i-th distortion factor coefficient in the distortion factor coefficient set, and the distortion factor set includes n distortion factor coefficients.

[0038] Specifically, the distortion factor set is a collection of factors related to image distortion. These factors can be various elements affecting image distortion, such as lens optical parameters (e.g., focal length, aperture), shooting angle, and lighting conditions. These factors work together during image formation, leading to image distortion. The distortion factor coefficient set is a set of coefficients obtained by analyzing each distortion factor in the distortion factor set. Each coefficient corresponds to a distortion factor and reflects the degree of influence of that distortion factor on image distortion.

[0039] First, through shooting experiments with a calibration board (such as a checkerboard pattern), all possible factors affecting image distortion, such as lens distortion, sensor noise, and changes in ambient lighting, are identified and collected, forming a distortion factor set. Then, by analyzing the image offset of the calibration board, distortion analysis is performed on each factor in the distortion factor set, calculating the specific degree of influence of these distortion factors on image distortion, and assigning a distortion factor coefficient to each distortion factor, forming a distortion factor coefficient set. By constructing the distortion factor set and performing distortion analysis, the factors affecting image distortion can be comprehensively identified and quantified, providing an accurate basis for subsequent correction.

[0040] Based on the distortion factor coefficient set, a multi-view measurement distortion error calibration function as described above is constructed. This function corrects the distortion error in the first image to a first calibrated image by calculating the correction value for each pixel. For example, the first image... Given an m×n pixel matrix, for each pixel (x, y), the calibrated position (x, y) of that pixel is calculated using the multi-view measurement distortion error calibration function. ′ y ′ This process yields the first calibration image p. Using a multi-view measurement distortion error calibration function to calibrate and analyze the first image effectively eliminates distortion errors, improves image accuracy and clarity, provides a high-quality image foundation for subsequent point cloud data extraction and model construction, and enhances the accuracy and reliability of wind deflection galloping monitoring and early warning.

[0041] Furthermore, step S21 includes: Step S211: Construct the circuit factor set.

[0042] Step S212: Construct the equipment factor set.

[0043] Step S213: Construct a set of environmental factors.

[0044] Step S214: Based on the line factor set, the equipment factor set, and the environmental factor set, form the distortion factor set; wherein, the line factor set includes reflection characteristic coefficient and motion blur coefficient, the equipment factor set includes lens coefficient, sensor noise coefficient, and equipment aging coefficient, and the environmental factor set includes illumination intensity coefficient, illumination non-uniformity coefficient, temperature coefficient, and atmospheric disturbance coefficient.

[0045] Specifically, during the monitoring process of multi-view monitoring equipment, the characteristics of the transmission line itself, the monitoring equipment, and the surrounding environment can all cause image distortion. All factors related to the characteristics of the transmission line itself that affect image distortion are collected to form a line factor set. Based on the line's reflection characteristics and motion blur, coefficients in the line factor set are determined, including reflection characteristic coefficients and motion blur coefficients. The reflection characteristic coefficient reflects the ability of the transmission line surface to reflect light; different reflection characteristics can lead to varying degrees of uneven brightness or reflected light spots in the image of the line portion. The motion blur coefficient is related to the movement of the transmission line during the monitoring process; slight swaying of the transmission line in the wind can blur the line portion in the image, thus causing distortion. Reflection characteristic data at different angles can be obtained by conducting light reflection tests on transmission line samples in the laboratory, and this data can be converted into reflection characteristic coefficients. The motion blur coefficient can be determined by analyzing the degree of blurring of the line edges in images acquired at different wind speeds and establishing a mathematical model.

[0046] Factors related to the multi-view monitoring equipment that affect image distortion are collected, including lens optical performance, sensor noise, and equipment aging, to form a set of equipment factors. Based on the specific lens performance, sensor noise, and aging status of the monitoring equipment, coefficients are determined within the set of equipment factors, including lens coefficients, sensor noise coefficients, and equipment aging coefficients. Lens coefficients are related to the lens's optical performance; for example, the lens's focal length and aberrations affect image imaging. Sensor noise coefficients reflect the noise level generated by the monitoring equipment's sensors during image acquisition; noise can cause random brightness fluctuations and other distortions in the image. Equipment aging coefficients reflect the impact of component wear and performance degradation on image quality during long-term use. Lens coefficients can be determined using optical design software or by conducting optical performance tests on the lens, based on parameters such as the lens model and specifications. For example, specialized lens aberration testing equipment can be used to measure various lens aberrations to obtain the lens coefficient. Sensor noise coefficients can be determined by acquiring multiple sets of images in a darkroom environment, analyzing the noise distribution patterns in the images, and using statistical methods. The equipment aging factor needs to be determined by regularly testing the equipment's performance, comparing the image quality acquired at different stages of use, and establishing an aging model.

[0047] Factors related to the monitoring environment that affect image distortion are collected, including light intensity, light uniformity, temperature, and atmospheric disturbance, to form an environmental factor set. Based on the actual monitoring environment's light intensity, light uniformity, temperature, and atmospheric disturbance conditions, coefficients within the environmental factor set are determined, including the light intensity coefficient, light uniformity coefficient, temperature coefficient, and atmospheric disturbance coefficient. The light intensity coefficient reflects the impact of ambient light intensity on the image; changes in light intensity affect image brightness and contrast. The light uniformity coefficient reflects the uneven distribution of light within the monitoring area; uneven light distribution can cause some areas of the image to be too bright or too dark. The temperature coefficient represents the impact of ambient temperature on the monitoring equipment and the image; changes in temperature affect equipment performance and image stability. The atmospheric disturbance coefficient reflects the degree of interference from atmospheric factors such as airflow and dust on light propagation; atmospheric disturbance can lead to image blurring and distortion. The light intensity coefficient can be determined by installing light intensity sensors in the monitoring area to collect light intensity data in real time, and then determining the coefficient based on the relationship between this data and image quality. The illumination non-uniformity coefficient can be determined by scanning the illumination distribution of the monitoring area, analyzing the differences in illumination between different areas, and establishing a mathematical model. The temperature coefficient requires testing the monitoring equipment under different temperature environments, observing changes in the equipment's imaging, and obtaining the coefficient through fitting experimental data. The atmospheric disturbance coefficient can be determined by obtaining relevant atmospheric parameters (such as airflow velocity and dust concentration) from meteorological monitoring equipment and analyzing their impact on image quality.

[0048] The aforementioned set of line factors, equipment factors, and environmental factors are combined to form a distortion factor set. This distortion factor set comprehensively covers various factors affecting image distortion, providing a complete set of parameters for subsequent distortion analysis and correction, making image distortion correction more comprehensive and accurate.

[0049] Furthermore, such as Figure 2 As shown, step S4 in this embodiment includes: Step S41: The registration and fusion model randomly samples a first sample set from the real-time point cloud dataset, and obtains the first parameter estimation result of the first model based on the first sample set.

[0050] Step S42: Remove the first sample set from the real-time point cloud dataset to obtain a first non-sample set, wherein the first non-sample set includes multiple non-sample point cloud data.

[0051] Step S43: Calculate the distance from the multiple non-sample point cloud data to the first model in sequence, and filter the multiple non-sample point cloud data in combination with a preset distance threshold to obtain the first inner point set.

[0052] Step S44: Calculate the first data volume of the first interior point set, and determine whether the first data volume meets the preset quantity threshold.

[0053] Step S45: If the first data volume meets the preset quantity threshold, the first model corresponding to the first parameter estimation result is recorded as the real-time three-dimensional model.

[0054] Specifically, the registration and fusion model randomly selects a portion of point cloud data from the real-time point cloud dataset as the first sample set. Based on the first sample set, statistical or optimization algorithms (such as the least squares method) are used to preliminarily estimate the parameters of the first model, obtaining the first parameter estimation results. For example, in a target traverse model represented by 3D point cloud data, the real-time point cloud dataset contains 1000 point cloud data points. The registration and fusion model randomly selects 200 of these point cloud data points as the first sample set. Then, based on the coordinates, normal vectors, and other information of these 200 point cloud data points, algorithms such as the least squares method are used to calculate the relevant parameter estimation results of the first model, such as the center position coordinates and attitude angles of the model.

[0055] The first sample set is removed from the real-time point cloud dataset to obtain the first non-sample set, which includes all remaining point cloud data. For example, in the real-time point cloud dataset described above, after removing the 200 point cloud data points contained in the first sample set, the remaining 800 point cloud data points constitute the first non-sample set. By removing the first sample set from the real-time point cloud dataset, the data in the first sample set is avoided from being reused in subsequent screening processes, ensuring the independence and reliability of the screening results.

[0056] The preset distance threshold is a pre-defined value used to measure whether the distance from non-sample point cloud data to the first model meets the requirements. If the distance is less than this threshold, the non-sample point cloud data is considered a qualified inlier. The distance between each point cloud data in the first non-sample set and the first model is calculated sequentially, for example, using the Euclidean distance formula to calculate the perpendicular distance from each point cloud data to the model surface. Combined with the preset distance threshold, the non-sample point cloud data is filtered, and point cloud data with distances less than the threshold are included in the first inlier set. This first inlier set is a set of point cloud data that meets the model fitting accuracy requirements; the point cloud data in this set has a high degree of fit with the first model.

[0057] Calculate the number of point cloud data points in the first inlier set, i.e., the first data volume. Determine whether the first data volume meets the preset quantity threshold. If it does, it indicates that the first model has a good fitting effect and high reliability; if it does not, further optimization of model parameters or resampling is required. The preset quantity threshold is the minimum requirement for the number of inliers, used to ensure the reliability of the model.

[0058] If the first data volume meets the preset threshold, the first model corresponding to the first parameter estimation result is recorded as the real-time 3D model, meaning that the model is considered to be able to reflect the real-time state and position of the target conductor well. For example, in a transmission line monitoring project, the first inner point set obtained through the above steps has 300 point cloud data points, which meets the preset threshold of 250. Therefore, the first model is used as the real-time 3D model for subsequent monitoring and analysis.

[0059] The above steps, through random sampling and parameter estimation, initially construct a three-dimensional model of the target conductor. By eliminating sample data and calculating the distance to non-sample point cloud data, point cloud data with high model fit is effectively selected, improving the model's accuracy and reliability. The effectiveness of the three-dimensional model is evaluated using a preset quantity threshold to ensure that the constructed real-time three-dimensional model accurately reflects the state of the transmission conductor, providing a reliable foundation for subsequent monitoring and early warning, and improving the accuracy and stability of monitoring and early warning.

[0060] Furthermore, step S44 also includes: Step S441: If the first data volume does not meet the preset quantity threshold, obtain the first re-collection instruction.

[0061] Step S442: According to the first resampling instruction, the registration and fusion model performs random sampling iterations from the real-time point cloud dataset.

[0062] Specifically, the first resampling instruction is a control command used to instruct the registration and fusion model to resample the point cloud data to improve the sampling results. If the first data volume does not meet the preset threshold, the first resampling instruction is generated, instructing the registration and fusion model to resample randomly. According to the first resampling instruction, the registration and fusion model again randomly samples from the real-time point cloud dataset to obtain a new sample set. For example, another set of 200 point cloud data points is randomly selected from the remaining point cloud data as a new sample set. The model parameters are re-estimated using the new sample set, and the subsequent screening and evaluation process is repeated to improve the model's fitting effect and reliability.

[0063] By generating resampling instructions and randomly sampling iteratively, the sampling results are dynamically adjusted, which significantly improves the effective utilization rate of point cloud data, makes model construction more robust, avoids model generation interruption caused by a single sampling failure, and improves the success rate of model fitting.

[0064] Furthermore, such as Figure 3 As shown, step S5 includes: Step S51: Based on the principle of uniform sampling, sequentially obtain the real-time model sample point set of the real-time 3D model and the predetermined model sample point set of the predetermined 3D model.

[0065] Step S52: Obtain the real-time curve and the predetermined curve of the target traverse based on the real-time model sample set and the predetermined model sample set, respectively.

[0066] Step S53: Obtain a predetermined sampling frequency, and sample the real-time curve sequentially based on the predetermined sampling frequency to obtain a real-time sample point set, and sample the predetermined curve to obtain a predetermined sample point set.

[0067] Step S54: Introduce a curve distance judgment strategy to perform point-to-point distance judgment between the real-time sample set and the predetermined sample set to obtain the real-time target distance.

[0068] Step S55: Use the real-time target distance as the target wind deflection coefficient.

[0069] Specifically, based on the principle of uniform sampling, points are uniformly selected on a spatial geometric curve at equal intervals. Sample point sets are then extracted sequentially from both the real-time 3D model and the predetermined 3D model to obtain the real-time model sample point set and the predetermined model sample point set. This avoids analytical bias caused by inconsistent density and improves the accuracy and representativeness of the analysis. The sampling process can employ voxel grid sampling or curve reconstruction uniform sampling algorithms. Taking voxel grid sampling as an example, only one representative point is retained in every 5cm spatial cube, thus forming a uniformly distributed sample point set.

[0070] After constructing the model sample point set, curve fitting or interpolation algorithms are used to transform these discrete spatial points into continuous conductor geometry curves, obtaining the real-time curve and predetermined curve of the target conductor, thereby improving the data's resolvability and the stability of comparison. Taking cubic spline interpolation as an example, the x, y, and z dimensions of the sample point set are fitted simultaneously to generate a smooth parametric three-dimensional curve.

[0071] After fitting, a predetermined sampling frequency is obtained, which is the frequency of equal-interval sampling. This frequency determines the accuracy of point sampling on the curve. Equal-interval sampling is performed on two continuous curves using the same rules and frequency. The real-time curve is sampled to obtain a real-time sample point set, and the predetermined curve is sampled to obtain a predetermined sample point set, so as to establish a consistent point-to-point comparison relationship.

[0072] The curve distance assessment strategy is a rule-based system used to compare the spatial distance between two curve sample points point by point. Multiple point pairs are obtained by mapping the real-time sample point set to a predetermined sample point set according to the sampling distance. The curve distance assessment strategy is then executed, comparing the spatial distance (e.g., Euclidean distance) between the real-time sample point and the predetermined sample point in each point pair. The maximum spatial distance between the two curves is determined as the real-time target distance, and this real-time target distance is output as the target wind deflection coefficient, representing the actual degree of current conductor offset.

[0073] Furthermore, step S54 includes: Step S541: Obtain the grid at the predetermined distance.

[0074] Step S542: Obtain any point pair, wherein the arbitrary point pair includes the first real-time sample point in the real-time sample point set and the first predetermined sample point in the predetermined sample point set.

[0075] Step S543: According to the initial filling scheme in the curve distance judgment strategy, fill the first spatial distance between the first real-time sample point and the first predetermined sample point into the predetermined distance grid to obtain the initial filling grid.

[0076] Step S544: Extract the advanced filling scheme from the curve distance judgment strategy.

[0077] Step S545: Perform advanced filling on the initial filling grid according to the advanced filling scheme to obtain the target filling grid.

[0078] Step S546: Take the fill value of the predetermined grid in the target filled grid as the real-time target distance.

[0079] Specifically, to achieve high-precision detection of conductor wind deflection, a curve distance assessment strategy is introduced. This involves constructing a predetermined distance grid and implementing two-stage filling schemes: initial filling and advanced filling, to more accurately evaluate the point-to-point distances between curves. The initial filling scheme calculates the spatial distance of each point pair and writes it into the predetermined distance grid, using a three-dimensional Euclidean distance formula. The resulting two-dimensional distance matrix, denoted as the initial filling grid, reflects the spatial offset relationship between the original point pairs. Since the original point cloud may contain outliers or sampling perturbations, directly using the initial filling grid to determine the final distance will result in path jitter. The advanced filling scheme performs secondary optimization filling based on the surrounding grid data, using steps such as maximum value substitution to eliminate anomalies or deviations. The final two-dimensional distance grid after advanced filling optimization is denoted as the target filling grid, used to determine the real-time target distance.

[0080] Obtain a predetermined distance grid, which is represented as a two-dimensional matrix, where each grid cell D ij For real-time sample point P i With the predetermined sample Q jThe spatial distance between them. The first real-time sample point is any point in the real-time sample point set, and the first predetermined sample point is the point in the predetermined sample point set corresponding to the first real-time sample point. The two form a point pair. Select a pair of sample points to form a point pair. Calculate the spatial distance between the first real-time sample point and the first predetermined sample point in the point pair based on the initial filling scheme (3D Euclidean distance formula), and write this distance into the corresponding position of the predetermined distance grid to generate an initial filling grid, which can be regarded as the offset distribution map of each corresponding position of the traverse. Next, according to the advanced filling scheme, the predetermined distance grid is divided into partitions to determine advanced blocks. For each grid cell in the advanced block, the grid is optimized and filled according to the values ​​of its adjacent grids to form the target filling grid. Finally, select the filling value of the predetermined grid (i.e., the lower right corner grid) from the target filling grid as the real-time target distance for that area.

[0081] The above steps effectively enhance the ability to accurately judge the wind deflection amplitude of transmission lines in complex environments by constructing distance grids in a hierarchical manner and optimizing the point-to-point matching process. Compared with traditional Euclidean fitting or ICP methods, it has higher robustness and stability.

[0082] Furthermore, step S545 includes: Step S545-1: Divide the initial filling grid into partitions according to the advanced filling scheme to obtain partitioning results; Step S545-2: Obtain the advanced blocks in the partitioning result and extract any grid in the advanced blocks, wherein the arbitrary grid corresponds to any initial fill value; Step S545-3: Construct an arbitrary reference mesh set for the arbitrary mesh, and filter the arbitrary reference mesh set to obtain the arbitrary maximum mesh fill value; Step S545-4: Replace the arbitrary initial fill value with the sum of the arbitrary maximum grid fill value and the arbitrary initial fill value, and use it as the arbitrary advanced fill value of the arbitrary grid; Step S545-5: Based on the arbitrary advanced fill value, form the target fill mesh.

[0083] Furthermore, the arbitrary reference grid set includes the left neighbor grid, the top neighbor grid, and the top-left neighbor grid of the arbitrary grid.

[0084] Specifically, the initial fill grid is partitioned according to the advanced fill scheme. For example, the first row and first column of grid cells are divided into fixed blocks as a reference for subsequent adjustments, and the remaining grid cells are used as advanced blocks, resulting in a partitioning result. Advanced blocks are extracted from the partitioning result, and a grid point is selected sequentially within each block, denoted as an arbitrary grid, which corresponds to an arbitrary initial fill value. The adjacent grid cells of this arbitrary grid are extracted as a reference grid set, including its left neighbor, top neighbor, and top-left neighbor. Then, the maximum value among all arbitrary initial fill values ​​corresponding to this reference grid set is selected and output as the arbitrary maximum grid fill value. The arbitrary maximum grid fill value is added to the arbitrary initial fill value of the arbitrary grid, and the sum is used to replace the arbitrary initial fill value as the arbitrary advanced fill value of the arbitrary grid. The above method is used to progressively fill all grid cells in the advanced blocks, ultimately forming a target fill grid containing all optimized fill values.

[0085] For example, the initial filled mesh obtained through the initial filling scheme is as follows:

[0086] According to the advanced filling scheme, P2Q2, P3Q2, P2Q3, and P3Q3 are designated as advanced blocks, and advanced filling is performed on each grid in the advanced blocks.

[0087] For P2Q2, extract its left neighbor P2Q1, top neighbor P1Q2, and top left neighbor P1Q1 to construct a reference grid set, and filter out any maximum grid fill value of 0.4. Then, any advanced fill value of P2Q2 = 0.4 + 0.5 = 0.9, and use this value to replace the original fill value (0.5).

[0088] For P3Q2, extract its left neighbor P3Q1, top neighbor P2Q2, and top left neighbor P2Q1 to construct a reference mesh set, and filter out any maximum mesh fill value of 0.9. Then, any advanced fill value of P3Q3 = 0.6 + 0.9 = 1.5, and use this value to replace the original fill value (0.6).

[0089] For P2Q3, extract its left neighbor P2Q2, top neighbor P1Q3, and top left neighbor P2Q2 to construct a reference mesh set, and filter out any maximum mesh fill value of 0.9. Then, any advanced fill value of P3Q3 = 0.7 + 0.9 = 1.6, and use this value to replace the original fill value (0.7).

[0090] For P3Q3, extract its left neighbor P3Q2, top neighbor P2Q3, and top left neighbor P2Q2 to construct a reference mesh set, and filter out any maximum mesh fill value of 1.6. Then, any advanced fill value of P3Q3 = 0.9 + 1.6 = 2.5, and use this value to replace the original fill value (0.9).

[0091] The target filled mesh is obtained based on any incrementing fill values ​​of P2Q2, P3Q2, P2Q3, and P3Q3 as follows:

[0092] Take any advanced fill value of the bottom right grid P3Q3 as the target distance.

[0093] Through the above steps, the distance between each point pair is not only based on the original calculation, but also takes into account the spatial trend of its local environment, which enhances the overall smoothness and anomaly suppression ability of the grid, and more accurately reflects the actual distance between curves, providing reliable data support for subsequent wind deflection coefficient calculation.

[0094] In summary, the multi-view three-dimensional monitoring method for alarming wind-induced galloping of transmission lines provided in this application has the following technical effects: Real-time monitoring images of the target transmission line are obtained through dynamic monitoring using multi-view monitoring equipment, providing raw image data for subsequent analysis. A multi-view measurement distortion error calibration function is introduced to perform distortion error calibration analysis on the first image, eliminating or reducing distortion errors caused by factors such as the lens characteristics of the multi-view monitoring equipment, resulting in a first calibration image. A real-time calibration image set of the multi-view monitoring equipment is then constructed, making the subsequent image-based 3D model more accurate and improving data reliability. The real-time point cloud dataset of the real-time calibration image set is input into the registration and fusion model to accurately fuse data from different perspectives, obtaining model parameter estimation results. This provides comprehensive and accurate data support for constructing the 3D model of the transmission line, improving the accuracy and reliability of the 3D model. Based on the model parameter estimation results, a real-time 3D model of the target transmission line is constructed, intuitively presenting information such as the spatial morphology of the transmission line, providing a visual basis for subsequent comparative analysis and early warning. The real-time 3D model is compared and analyzed with the predetermined 3D model of the target conductor to obtain the target wind deflection coefficient. This allows for precise analysis of the transmission line's state changes, a comprehensive assessment of the transmission line's wind-induced galloping state, and the degree of deviation between the actual and expected states of the line is determined by quantifying the target wind deflection coefficient. The system then analyzes whether the target wind deflection coefficient meets a predetermined wind deflection threshold. If it does not, an early warning signal is issued, and based on this signal, a wind-induced galloping warning for the target transmission line is issued, enabling relevant personnel to take measures to maintain or adjust the transmission line and prevent faults.

[0095] Overall, this application embodiment comprehensively acquires real-time images of transmission lines through multi-view monitoring equipment. After distortion error calibration and registration fusion model processing, a high-precision real-time three-dimensional model is constructed. By comparing and analyzing the real-time three-dimensional model with the predetermined three-dimensional model, the target wind deflection coefficient is calculated, the wind deflection galloping state is accurately judged, and a warning signal is issued in a timely manner. This achieves efficient monitoring and accurate early warning of wind deflection galloping of transmission lines, significantly improves the accuracy of monitoring data and the timeliness of early warning, effectively prevents accidents caused by wind deflection galloping, and ensures the safe and stable operation of transmission lines.

[0096] Example 2, as Figure 4 As shown in the figure, this application provides a multi-view three-dimensional monitoring system for transmission line wind-induced galloping alarm, the system comprising: The image monitoring module 10 is used to obtain real-time monitoring images of the target conductor through dynamic monitoring by a multi-view monitoring device. The real-time monitoring images include a first image monitored by a first view, where the first view refers to any one of the views in the multi-view monitoring device.

[0097] The image calibration module 20 is used to introduce a multi-view measurement distortion error calibration function to perform distortion error calibration analysis on the first image to obtain a first calibration image, and to build a real-time calibration image set of the multi-view monitoring device.

[0098] The registration and fusion module 30 is used to take the real-time point cloud dataset of the real-time calibration image set as input information of the registration and fusion model, and obtain output information through the registration and fusion model, wherein the output information includes model parameter estimation results.

[0099] The three-dimensional model construction module 40 is used to construct a real-time three-dimensional model of the target traverse based on the model parameter estimation results.

[0100] The comparison analysis module 50 is used to compare and analyze the real-time three-dimensional model with the predetermined three-dimensional model of the target conductor to obtain the target wind deflection coefficient.

[0101] The judgment module 60 is used to obtain a predetermined wind deflection threshold and determine whether the target wind deflection coefficient meets the predetermined wind deflection threshold.

[0102] The early warning module 70 is used to issue an early warning signal if the target wind deflection coefficient does not meet the predetermined wind deflection threshold, thereby providing an early warning of wind deflection galloping of the target conductor.

[0103] Furthermore, in this embodiment of the application, the image calibration module 20 is also used to perform the following steps: A distortion factor set is constructed, and distortion analysis is performed on the distortion factor set to obtain a distortion factor coefficient set. Based on the multi-view measurement distortion error calibration function and the distortion factor coefficient set, distortion error calibration analysis is performed on the first image to obtain the first calibrated image. The expression for the multi-view measurement distortion error calibration function is as follows: ;in, The first image refers to With the first calibration image The multi-view measurement distortion error calibration function between them, It refers to the i-th distortion factor coefficient in the distortion factor coefficient set, and the distortion factor set includes n distortion factor coefficients.

[0104] Furthermore, in this embodiment of the application, the image calibration module 20 is also used to perform the following steps: Construct a line factor set; construct a device factor set; construct an environmental factor set; based on the line factor set, the device factor set, and the environmental factor set, construct the distortion factor set; wherein, the line factor set includes reflection characteristic coefficient and motion blur coefficient, the device factor set includes lens coefficient, sensor noise coefficient, and device aging coefficient, and the environmental factor set includes illumination intensity coefficient, illumination non-uniformity coefficient, temperature coefficient, and atmospheric disturbance coefficient.

[0105] Furthermore, in this embodiment of the application, the three-dimensional model construction module 40 is also used to perform the following steps: The registration and fusion model randomly samples a first sample set from the real-time point cloud dataset and obtains a first parameter estimation result for the first model based on the first sample set. The first sample set is then removed from the real-time point cloud dataset to obtain a first non-sample set, which includes multiple non-sample point cloud data. The distances from the multiple non-sample point cloud data to the first model are calculated sequentially, and the multiple non-sample point cloud data are filtered in combination with a preset distance threshold to obtain a first interior point set. The first data volume of the first interior point set is calculated, and it is determined whether the first data volume meets a preset quantity threshold. If the first data volume meets the preset quantity threshold, the first model corresponding to the first parameter estimation result is recorded as the real-time 3D model.

[0106] Furthermore, in this embodiment of the application, the three-dimensional model construction module 40 is also used to perform the following steps: If the first data volume does not meet the preset quantity threshold, a first re-sampling instruction is obtained; according to the first re-sampling instruction, the registration and fusion model performs random sampling iterations from the real-time point cloud dataset.

[0107] Furthermore, in this application embodiment, the comparison and analysis module 50 is also used to perform the following steps: Based on the principle of uniform sampling, the real-time model sample point set of the real-time 3D model and the predetermined model sample point set of the predetermined 3D model are obtained sequentially; the real-time curve and the predetermined curve of the target conductor are obtained based on the real-time model sample point set and the predetermined model sample point set, respectively; a predetermined sampling frequency is obtained, and the real-time curve is sampled sequentially based on the predetermined sampling frequency to obtain the real-time sample point set, and the predetermined curve is sampled to obtain the predetermined sample point set; a curve distance judgment strategy is introduced to perform point-to-point distance judgment on the real-time sample point set and the predetermined sample point set to obtain the real-time target distance; the real-time target distance is used as the target wind deflection coefficient.

[0108] Furthermore, in this application embodiment, the comparison and analysis module 50 is also used to perform the following steps: Obtain a predetermined distance grid; obtain any pair of points, the pair of points including a first real-time point in the real-time sample set and a first predetermined point in the predetermined sample set; according to the initial filling scheme in the curve distance judgment strategy, fill the predetermined distance grid with the first spatial distance between the first real-time point and the first predetermined point to obtain an initial filled grid; extract the advanced filling scheme in the curve distance judgment strategy; perform advanced filling on the initial filled grid according to the advanced filling scheme to obtain a target filled grid; take the filling value of the predetermined grid in the target filled grid as the real-time target distance.

[0109] Furthermore, in this application embodiment, the comparison and analysis module 50 is also used to perform the following steps: The initial fill grid is partitioned according to the advanced fill scheme to obtain a partitioning result; advanced blocks are obtained from the partitioning result, and arbitrary grids in the advanced blocks are extracted, wherein the arbitrary grids correspond to arbitrary initial fill values; an arbitrary reference grid set of the arbitrary grids is constructed, and the arbitrary reference grid set is filtered to obtain an arbitrary maximum grid fill value; the sum of the arbitrary maximum grid fill value and the arbitrary initial fill value is used to replace the arbitrary initial fill value as the arbitrary advanced fill value of the arbitrary grid; the target fill grid is formed based on the arbitrary advanced fill value.

[0110] Furthermore, the arbitrary reference grid set includes the left neighbor grid, the top neighbor grid, and the top-left neighbor grid of the arbitrary grid.

[0111] Through the foregoing detailed description of a multi-view three-dimensional monitoring method for alarming wind-induced galloping of transmission lines, those skilled in the art can clearly understand that the multi-view three-dimensional monitoring system for alarming wind-induced galloping of transmission lines in this embodiment corresponds to the system disclosed in Embodiment 2. As it is similar to the method disclosed in Embodiment 1, it has corresponding functional modules and beneficial effects. For relevant details, please refer to the method section.

[0112] 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.

Claims

1. A method for alarming wind-induced galloping of transmission lines using multi-view three-dimensional monitoring, characterized in that, include: Real-time monitoring images of the target conductor are obtained through dynamic monitoring by multi-view monitoring equipment. The real-time monitoring images include a first image monitored by a first view, where the first view refers to any one view in the multi-view monitoring equipment. A multi-view measurement distortion error calibration function is introduced to perform distortion error calibration analysis on the first image to obtain a first calibration image, and a real-time calibration image set of the multi-view monitoring device is constructed. The real-time point cloud dataset of the real-time calibration image set is used as the input information of the registration fusion model, and the output information is obtained through the registration fusion model, wherein the output information includes the model parameter estimation results; A real-time three-dimensional model of the target traverse is constructed based on the model parameter estimation results; The target wind deflection coefficient is obtained by comparing and analyzing the real-time three-dimensional model with the predetermined three-dimensional model of the target conductor. Obtain a predetermined wind deflection threshold and determine whether the target wind deflection coefficient meets the predetermined wind deflection threshold; If the target wind deflection coefficient does not meet the predetermined wind deflection threshold, an early warning signal is issued to provide an early warning of wind deflection galloping of the target conductor.

2. The method for multi-view three-dimensional monitoring and alarm of transmission line wind deflection galloping according to claim 1, characterized in that, A multi-view measurement distortion error calibration function is introduced to perform distortion error calibration analysis on the first image to obtain a first calibration image, and a real-time calibration image set of the multi-view monitoring device is constructed, including: A set of distortion factors is constructed, and distortion analysis is performed on the set of distortion factors to obtain a set of distortion factor coefficients; Based on the multi-view measurement distortion error calibration function and the distortion factor coefficient set, distortion error calibration analysis is performed on the first image to obtain the first calibrated image. The expression for the multi-view measurement distortion error calibration function is as follows: ； in, The first image refers to With the first calibration image The multi-view measurement distortion error calibration function between them, It refers to the i-th distortion factor coefficient in the distortion factor coefficient set, and the distortion factor set includes n distortion factor coefficients.

3. The method for multi-view three-dimensional monitoring and alarm of transmission line wind deflection galloping according to claim 2, characterized in that, A distortion factor set is constructed, and distortion analysis is performed on the distortion factor set to obtain a distortion factor coefficient set, including: Construct the circuit factor set; Construct the equipment factor set; Construct a set of environmental factors; The distortion factor set is formed based on the line factor set, the equipment factor set, and the environmental factor set; The line factor set includes reflection characteristic coefficient and motion blur coefficient; the equipment factor set includes lens coefficient, sensor noise coefficient and equipment aging coefficient; and the environmental factor set includes light intensity coefficient, light unevenness coefficient, temperature coefficient and atmospheric disturbance coefficient.

4. The method for multi-view three-dimensional monitoring and alarm of transmission line wind deflection galloping according to claim 1, characterized in that, Based on the model parameter estimation results, a real-time three-dimensional model of the target traverse is constructed, including: The registration and fusion model randomly samples a first sample set from the real-time point cloud dataset and obtains the first parameter estimation result of the first model based on the first sample set; The first sample set is removed from the real-time point cloud dataset to obtain the first non-sample set, wherein the first non-sample set includes multiple non-sample point cloud data. The distances from the multiple non-sample point cloud data to the first model are calculated sequentially, and the multiple non-sample point cloud data are filtered in combination with a preset distance threshold to obtain a first set of interior points; Calculate the first data volume of the first interior point set, and determine whether the first data volume meets the preset quantity threshold. If the first data volume meets the preset quantity threshold, the first model corresponding to the first parameter estimation result is recorded as the real-time three-dimensional model.

5. The method for multi-view three-dimensional monitoring and alarm of transmission line wind deflection galloping according to claim 4, characterized in that, Calculating the first data volume of the first interior point set and determining whether the first data volume meets a preset quantity threshold further includes: If the first data volume does not meet the preset quantity threshold, a first re-collection instruction is obtained; According to the first resampling instruction, the registration and fusion model performs random sampling iterations from the real-time point cloud dataset.

6. The method for multi-view three-dimensional monitoring and alarm of transmission line wind deflection galloping according to claim 1, characterized in that, The target wind deflection coefficient is obtained by comparing and analyzing the real-time 3D model with the predetermined 3D model of the target conductor, including: Based on the principle of uniform sampling, the real-time model sample point set of the real-time 3D model and the predetermined model sample point set of the predetermined 3D model are obtained sequentially. The real-time curve and the predetermined curve of the target traverse are obtained based on the real-time model sample set and the predetermined model sample set, respectively. A predetermined sampling frequency is obtained, and the real-time curve is sampled sequentially based on the predetermined sampling frequency to obtain a real-time sample point set; the predetermined curve is sampled to obtain a predetermined sample point set. A curve distance judgment strategy is introduced to perform point-to-point distance judgment between the real-time sample set and the predetermined sample set to obtain the real-time target distance; The real-time target distance is used as the target wind deflection coefficient.

7. The method for multi-view three-dimensional monitoring and alarm of transmission line wind deflection galloping according to claim 6, characterized in that, A curve distance assessment strategy is introduced to perform point-to-point distance assessment between the real-time sample set and the predetermined sample set to obtain the real-time target distance, including: Obtain the grid at the predetermined distance; Obtain any pair of points, wherein the pair of points includes a first real-time sample point in the real-time sample point set and a first predetermined sample point in the predetermined sample point set; According to the initial filling scheme in the curve distance judgment strategy, the first spatial distance between the first real-time sample point and the first predetermined sample point is filled into the predetermined distance grid to obtain the initial filling grid; Extract the advanced filling scheme from the curve distance judgment strategy; The initial filling grid is further filled according to the advanced filling scheme to obtain the target filling grid; The fill value of a predetermined grid in the target filled grid is taken as the real-time target distance.

8. The method for multi-view three-dimensional monitoring and alarm of transmission line wind deflection galloping according to claim 7, characterized in that, The initial filled mesh is further filled according to the advanced filling scheme to obtain the target filled mesh, including: The initial filling grid is partitioned according to the advanced filling scheme to obtain the partitioning results; Obtain the advanced blocks from the partitioning results, and extract any grid from the advanced blocks, wherein the arbitrary grid corresponds to any initial fill value; Construct an arbitrary reference mesh set for the arbitrary mesh, and filter the arbitrary reference mesh set to obtain an arbitrary maximum mesh fill value; The sum of the arbitrary maximum grid fill value and the arbitrary initial fill value is used to replace the arbitrary initial fill value, and is used as the arbitrary advanced fill value of the arbitrary grid. The target filling grid is formed based on the arbitrary advanced fill value.

9. The method for multi-view three-dimensional monitoring and alarm of transmission line wind deflection galloping according to claim 8, characterized in that, The arbitrary reference grid set includes the left neighbor grid, the top neighbor grid, and the top-left neighbor grid of the arbitrary grid.

10. A multi-view three-dimensional monitoring system for transmission line wind-induced galloping alarm, characterized in that, The system is used to execute a multi-view three-dimensional monitoring method for alarming wind-induced galloping of transmission lines according to any one of claims 1-9, comprising: The image monitoring module is used to obtain real-time monitoring images of the target conductor through dynamic monitoring by a multi-view monitoring device. The real-time monitoring image includes a first image monitored by a first view, where the first view refers to any one view in the multi-view monitoring device. The image calibration module is used to introduce a multi-view measurement distortion error calibration function to perform distortion error calibration analysis on the first image to obtain a first calibration image, and to build a real-time calibration image set of the multi-view monitoring device; The registration and fusion module is used to take the real-time point cloud dataset of the real-time calibration image set as the input information of the registration and fusion model, and obtain the output information through the registration and fusion model, wherein the output information includes model parameter estimation results; A 3D model building module is used to build a real-time 3D model of the target traverse based on the model parameter estimation results; The comparison analysis module is used to compare and analyze the real-time three-dimensional model with the predetermined three-dimensional model of the target conductor to obtain the target wind deflection coefficient; The judgment module is used to obtain a predetermined wind deflection threshold and determine whether the target wind deflection coefficient meets the predetermined wind deflection threshold. The early warning module is used to issue an early warning signal if the target wind deflection coefficient does not meet the predetermined wind deflection threshold, thereby providing an early warning of wind deflection galloping of the target conductor.

Images