Pole stay wire inspection method and system

Abstract

The invention discloses an electric pole stay wire inspection method and system, and relates to the technical field of electric power, and the method comprises the steps: obtaining the visual monitoring data of a to-be-detected stay wire in the flight process of an unmanned aerial vehicle, and enabling the visual monitoring data to be collected through a visual machine carried on the unmanned aerial vehicle; and detecting attribute information of the to-be-detected stay wire based on the visual monitoring data to determine whether the to-be-detected stay wire is in an abnormal state. According to the invention, unmanned aerial vehicle inspection, a visual machine and a data analysis technology are organically fused, intelligent monitoring of the state of the electric pole stay wire can be realized, and one set of equipment can perform monitoring operation on multiple point positions of multiple electric pole stay wires, so that the labor cost and the hardware installation cost are reduced. Multiple attribute information such as the inclination angle, the fracture condition and the tightness of the electric pole stay wire are detected and analyzed cooperatively, the overall state of the electric pole stay wire can be evaluated more comprehensively and accurately, and a complete data support is provided for operation and maintenance management of electric power facilities.

Description

Technical Field

[0001] This invention relates to the field of power technology, and in particular to a method and system for inspecting guy wires on utility poles. Background Technology

[0002] Guy wires are an indispensable part of the power system. By fixing the guy wires, the overhead line towers can be effectively prevented from overturning due to wind or other external forces. When the pole is high or has many crossarms, guy wires are needed to improve the stress on the pole and prevent it from bending or breaking.

[0003] In current utility pole guy wire inspection scenarios, a common monitoring solution is a combination of multiple sensors and a wireless network. This solution installs various sensors on the guy wires, such as displacement sensors, tilt sensors, and tension sensors, to monitor data such as displacement, tilt changes, and elastic deformation. The data is then transmitted to a monitoring center via a wireless network (such as LoRa or WiFi) for anomaly identification. When the monitored data exceeds a preset threshold (such as excessive displacement or abnormal tension), the system automatically triggers an alarm, enabling staff to efficiently detect and address abnormal guy wire failures caused by external forces or aging.

[0004] However, in reality, most sensors installed on guy wires require continuous power, which is prone to instability in solar power supply in the field, resulting in low data acquisition frequency, large processing delays, and inability to update monitoring data in real time. Furthermore, installing multiple sensors on guy wires may affect the structural strength and lifespan of the original guy wires, posing safety hazards. Summary of the Invention

[0005] This invention provides a method and system for inspecting utility pole guy wires, which solves the problems of existing utility pole guy wire inspection schemes where the installed hardware is greatly affected by the environment and may have a certain degree of impact on the utility pole guy wires.

[0006] The first aspect of this invention provides a method for inspecting utility pole guy wires, comprising:

[0007] The visual monitoring data of the test cable is acquired during the flight of the UAV; the visual monitoring data is collected by a vision machine mounted on the UAV.

[0008] The attribute information of the test cable is detected based on the visual monitoring data to determine whether the test cable is in an abnormal state; the attribute information includes at least one of the following: the tilt angle, breakage status, and tension of the test cable.

[0009] Optionally, the step of acquiring visual monitoring data of the guy wire under test during the flight of the UAV includes:

[0010] The Hough transform algorithm based on the attention mechanism is used to map the original image acquired by the vision machine from the pixel space to the parameter space to suppress visual data interference caused by non-pull-wire areas and highlight the pull-wire area in the parameter space, so as to extract the center line parameters corresponding to the pull-wire to be tested.

[0011] Based on the centerline parameters, the aircraft trajectory and flight status parameters of the UAV for the test cable are determined;

[0012] The drone is controlled to fly according to the aircraft trajectory and flight status parameters so that the visual monitoring data can be collected by the vision machine.

[0013] Optionally, if the attribute information includes the tilt angle of the guy wire to be tested, and the visual monitoring data is acquired by a binocular vision acquisition unit serving as the vision machine, then the step of detecting the attribute information of the guy wire to be tested based on the visual monitoring data includes:

[0014] Feature point matching is performed on the left and right images of the test cable in the visual monitoring data to obtain matching point pairs that meet preset conditions;

[0015] Using the coordinate information of the matching point pairs, disparity values ​​corresponding to feature points at different positions on the test line are generated;

[0016] Based on the disparity value and the two-dimensional coordinate information of each feature point in the left or right image, the three-dimensional spatial straight line equation of the wire to be tested is fitted, and the tilt angle of the wire to be tested is calculated through the three-dimensional spatial straight line equation.

[0017] Optionally, if the attribute information includes the breakage status of the guy wire under test, and the visual monitoring data is acquired by a binocular vision acquisition unit serving as the vision machine, then the step of detecting the attribute information of the guy wire under test based on the visual monitoring data includes:

[0018] A trained fracture prediction model is obtained; the fracture prediction model is obtained by iteratively training the training data through convolution and pooling strategies, and the training data includes normal cable images and abnormal cable images of different fracture types collected by the UAV.

[0019] The left and right images of the guy wire to be tested from the visual monitoring data are input into the fracture prediction model to output the fracture prediction probability at different positions on the guy wire to be tested.

[0020] Each fracture prediction probability is compared with a fracture threshold to determine the fracture location and fracture extent that reflect the fracture condition.

[0021] Optionally, if the attribute information includes the tension of the tension of the wire to be tested, and the visual monitoring data is acquired by a binocular vision acquisition unit serving as the vision machine, then the step of detecting the attribute information of the tension of the wire to be tested based on the visual monitoring data includes:

[0022] A three-dimensional point cloud belonging to the main body of the pull wire under test is constructed using the left and right images in the visual monitoring data; the three-dimensional point cloud is a point cloud after removing noise points;

[0023] The centerline equation of the guy wire to be tested is determined based on the coordinate information of multiple points in the three-dimensional point cloud, and the morphological features of the guy wire to be tested are calculated through the centerline equation; wherein, the morphological features include the centerline curvature, the cross-sectional radius of the cross-sectional point cloud, and the ellipticity, and the cross-sectional point cloud refers to the point cloud obtained by the intersection of the plane of the centerline and the three-dimensional point cloud.

[0024] The morphological features are input into the trained tension prediction model to output the tension of the test cable; the tension is compared with a tension threshold to determine whether the test cable is in an abnormal tension state.

[0025] Optionally, if the visual monitoring data is acquired by the lidar, and the attribute information also includes the corrosion status of the guy wire to be tested, then the step of detecting the attribute information of the guy wire to be tested based on the visual monitoring data includes:

[0026] Based on the laser reflection signal belonging to the test wire in the visual monitoring data, a three-dimensional point cloud for at least one period is constructed.

[0027] Based on the reflection intensity information of the three-dimensional point cloud, it is determined whether there is corrosion on the surface of the wire to be tested.

[0028] A second aspect of the present invention provides a pole guy wire inspection system, comprising:

[0029] The acquisition module is used to acquire visual monitoring data of the test cable during the flight of the UAV; the visual monitoring data is acquired by a vision machine mounted on the UAV.

[0030] The detection module is used to detect the attribute information of the test wire based on the visual monitoring data to determine whether the test wire is in an abnormal state; the attribute information includes at least one of the following: the tilt angle, breakage status, and tension of the test wire.

[0031] A third aspect of the present invention provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the pole guy wire inspection method described above.

[0032] The fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed, implements the pole guy wire inspection method as described above.

[0033] The fifth aspect of the present invention provides a computer program product, the computer program product comprising a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions, wherein when the program instructions are executed by a computer, the computer performs the pole guy wire inspection method as described above.

[0034] As can be seen from the above technical solutions, the present invention has the following advantages:

[0035] This invention organically integrates drone inspection, machine vision, and data analysis technologies to achieve intelligent and safe monitoring of the status of utility pole guy wires. Automated drone inspection can replace a large amount of manual inspection, efficiently completing data collection and processing in collaboration with edge computers or data analysis centers. This single system can monitor multiple locations on multiple utility pole guy wires, significantly reducing labor costs and hardware installation expenses. Furthermore, by detecting and collaboratively analyzing multiple attributes of the utility pole guy wires, such as tilt angle, breakage status, and tightness, a more comprehensive and accurate assessment of the overall condition of the guy wires can be achieved. This provides complete data support for the operation and maintenance management of power facilities, facilitating the development of scientific and reasonable maintenance strategies and improving the intelligent management level of the power system. Attached Figure Description

[0036] To more clearly illustrate the technical solutions in the embodiments of the present invention 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 only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0037] Figure 1 A flowchart illustrating the steps of a pole guy wire inspection method provided in this embodiment of the invention;

[0038] Figure 2 A structural block diagram of a pole guy wire inspection system provided in an embodiment of the present invention;

[0039] Figure 3 This is a structural block diagram of a computer device provided in an embodiment of the present invention. Detailed Implementation

[0040] This invention provides a method and system for inspecting utility pole guy wires, which addresses the problems of existing utility pole guy wire inspection schemes where the installed hardware is greatly affected by the environment and may have a certain degree of impact on the utility pole guy wires.

[0041] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0042] It should be noted that, in the optional embodiments of the present invention, the data related to object information, etc., requires the permission or consent of the object when the embodiments of the present invention are applied to specific products or technologies. Furthermore, the collection, use, and processing of the relevant data must comply with the relevant laws, regulations, and standards of the relevant countries and regions. In other words, if the embodiments of the present invention involve data related to an object, it needs to be obtained with the object's authorization and consent, the authorization and consent of relevant departments, and in accordance with the relevant laws, regulations, and standards of the country and region. If the embodiments involve personal information, the acquisition of all personal information requires the individual's consent. If sensitive information is involved, the separate consent of the information subject is required. The embodiments also need to be implemented with the object's authorization and consent.

[0043] Please see Figure 1 , Figure 1 The flowchart illustrates the steps of a pole guy wire inspection method provided in this embodiment of the invention.

[0044] This invention provides a method for inspecting guy wires on utility poles, comprising:

[0045] Step 101: Acquire visual monitoring data of the test cable during the flight of the UAV; the visual monitoring data is collected by a vision machine mounted on the UAV.

[0046] In this embodiment of the invention, the drone, as a mobile platform, can be equipped with a binocular vision acquisition unit or a vision machine such as a lidar to achieve a flexible and efficient data acquisition process. The vision machine may include a binocular vision acquisition unit or a lidar.

[0047] To achieve adaptive line-following flight control for the UAV, step 101 may further include the following sub-steps:

[0048] S11. Using the Hough transform algorithm based on the attention mechanism, the original image acquired by the vision machine is mapped from the pixel space to the parameter space to suppress visual data interference caused by non-pull-wire areas and highlight the pull-wire area in the parameter space, so as to extract the center line parameters corresponding to the pull-wire to be tested.

[0049] In practical applications, drones need to fly along the wire for monitoring. However, when visual machines collect data from the wire, they may be interfered with by background vegetation, buildings, etc. Therefore, an improved Hough transform algorithm based on attention mechanism can be used to suppress interference.

[0050] First, a spatial attention module needs to be established. Spatial attention uses a Convolutional Block Attention Module (CBAM) structure to generate the weight matrix. This is used to suppress the response in the non-stretched region. The formula for its weight matrix is:

[0051]

[0052] In the above formula, I is the input image, AvgPool and MaxPool are average pooling sampling and max pooling sampling, respectively, used to extract global features of the image; MLP is a multilayer perceptron, which performs nonlinear transformation on the pooled features; The Sigmoid activation function maps the output to the 0-1 range, generating an attention weight matrix to highlight the line area.

[0053] Next, the edge feature layer and semantic feature layer of ResNet50 need to be fused through the FPN network to output the feature map. .

[0054] The FPN network fuses features at different scales through upsampling and lateral connections, enabling the network to capture both the detailed edge information of the thread and the overall semantic information.

[0055] Perform a Hough transform on the processed weighted feature map to establish a parameter space accumulation matrix. Its transformation formula is

[0056]

[0057] In the above formula, These are the pixel coordinates in the feature map. The spatial attention weight matrix in coordinates The weight value at the location; θ is the polar radius, representing the perpendicular distance from the origin to the line; θ is the polar angle, representing the angle between the perpendicular direction and the line. For the Dirac function, when hour, The value is 1 if it is 1, otherwise it is 0.

[0058] In summary, the Hough transform algorithm based on the attention mechanism described above changes the image as follows.

[0059] Step 1: Generate a weighted feature map using an attention mechanism—the pixel signal in the line region is enhanced, while the pixel signal in the background interference region is weakened (visually, the line is clearer and the background is "faded"). Step 2: Perform a Hough transform on the weighted feature map, mapping the image from pixel space to a parameter space, such as a polar coordinate parameter space (ρ, θ), where ρ is the distance from the origin to the line, and θ is the angle between the line normal and the x-axis. This ultimately highlights the peak value of the line parameter corresponding to the line in the parameter space, filtering out false peak values ​​generated by background interference.

[0060] S12. Based on the centerline parameters, determine the UAV's trajectory and flight status parameters for the test cable.

[0061] By accumulating the responses of pixels in the parameter space, a model is constructed. The detection threshold can be set to max(H)×0.6 to filter out the centerline parameters of the string, such as the parameters in the centerline equation y=ax+b. and b.

[0062] S13. Control the drone to fly according to the aircraft trajectory and flight status parameters so that visual monitoring data can be collected through the vision machine.

[0063] After extracting the centerline parameters, or the centerline equation, the drone's flight trajectory can be constructed based on this, allowing the drone to inspect along the centerline direction of the cable. For example, this invention can use a model to construct the flight trajectory, whose state-space equation can be:

[0064]

[0065] in, This represents the position vector of the UAV in three-dimensional space (such as parameter space). It is the velocity vector; The acceleration vector is the control input quantity; Let be the process noise vector, and its covariance matrix be... , used to describe system uncertainty; The quaternion attitude is used to represent the rotational attitude of the UAV; This is the angular velocity vector.

[0066] Since deviations often occur during drone flight, it is necessary to construct an objective function to minimize the deviation between the actual position of the drone and the target position, while limiting acceleration and its variation. The objective function can be shown below:

[0067]

[0068] in, For the time domain of prediction, it means predicting the state 10 time steps ahead; For control of the time domain, i.e., the number of currently adjustable control input steps; Let K be the actual position of the UAV at the k-th time step. The target position for fitting the center line of the string; weight matrix The importance of dimensions S=diag(0.1,0.1,0.2) is used to limit the magnitude of acceleration to avoid over-control, and to limit the change in acceleration to ensure smooth flight.

[0069] Step 102: Detect the attribute information of the guy wire under test based on visual monitoring data to determine whether the guy wire under test is in an abnormal state; the attribute information includes at least one of the following: the tilt angle, the breakage status, and the tightness of the guy wire under test.

[0070] Furthermore, if the attribute information includes the tilt angle of the guy wire to be measured, and the visual monitoring data is acquired by a binocular vision acquisition unit acting as a vision machine, step 102 may include the following sub-step (pole guy wire tilt angle detection):

[0071] S21. Perform feature point matching on the left and right images of the cable to be tested in the visual monitoring data to obtain matching point pairs that meet the preset conditions; generate disparity values ​​corresponding to feature points at different positions on the cable to be tested using the coordinate information of the matching point pairs; fit the three-dimensional spatial straight line equation of the cable to be tested based on the disparity values ​​and the two-dimensional coordinate information of each feature point in the left or right image, and calculate the tilt angle of the cable to be tested using the three-dimensional spatial straight line equation.

[0072] In this embodiment of the invention, binocular vision technology is used to reconstruct the three-dimensional structure of the pull line. Specifically, the Scale-Invariant Feature Transform (SIFT) algorithm can be used to extract feature points from the left and right images. The SIFT algorithm simulates different scale spaces by constructing a Difference-of-Gaussian pyramid of the image. At each scale level, extreme points are detected in the image. To ensure the stability of the detected extreme points, non-maximum suppression is performed in both the scale space and the image space, retaining only local extreme points as candidate feature points. Then, a 128-dimensional feature descriptor is calculated for each candidate point. In the feature point matching stage, the similarity is measured by calculating the Euclidean distance between the feature point descriptors of the left and right images, and a distance threshold is set to filter out preliminary matching point pairs.

[0073] In addition, the RANSAC algorithm can be introduced to remove mismatched points. Its core idea is to randomly sample a small number of matching point pairs to calculate the fundamental matrix F. The fundamental matrix F describes the epipolar geometric relationship between the left and right images, satisfying the condition... (in and (These are the homogeneous coordinates of the matching points in the left and right images, respectively). Through multiple random sampling and verification, the calculation result with the largest number of interior points can be selected as the final base matrix. Matching point pairs that satisfy the epipolar constraint are retained, thus obtaining matching point pairs that meet the preset conditions.

[0074] For the filtered matching pairs (Belongs to the left image) and (Belonging to the right image), calculate the disparity according to the definition of disparity. To improve the accuracy and robustness of disparity calculation, a window-based block matching algorithm is adopted. Matching is performed within a local window centered on the matching point. The optimal matching point is determined by calculating the similarity of pixels within the window, thereby obtaining a more accurate disparity value and finally generating a disparity value map of each point on the line.

[0075] Convert two-dimensional coordinates on the image plane to three-dimensional spatial coordinates. Based on the principle of binocular vision triangulation, the intrinsic parameter matrix of the camera is known. (in , Let be the focal length of the camera in the x and y directions. Using the principal point coordinates, baseline distance b (distance between the optical centers of the two cameras), and parallax d, the three-dimensional coordinates (X, Y, Z) of the feature points on the line in the world coordinate system are calculated using the following formula:

[0076]

[0077] Where (x, y) are the two-dimensional image coordinates of the feature points on the line. It is the effective focal length (approximately equal to) or This formula is derived based on the principle of similar triangles. It transforms two-dimensional coordinates on the image plane into three-dimensional spatial coordinates by utilizing the relationship between parallax, baseline distance, and focal length. After establishing the three-dimensional coordinates, the formula is then used to determine the three-dimensional coordinates of feature points at different positions on the drawing line. The equation of the spatial straight line of the drawn wire is fitted using the least squares method. The direction vector of the straight line... Calculated using the vector dot product formula:

[0078]

[0079] The tilt angle is .

[0080] Furthermore, if the attribute information includes the breakage status of the guy wire under test, and the visual monitoring data is acquired by a binocular vision acquisition unit acting as a vision machine, step 102 may include the following sub-step (breakage detection):

[0081] S22. Obtain the trained fracture prediction model. The fracture prediction model is obtained by iteratively training the training data through convolution and pooling strategies. The training data includes normal guy wire images and abnormal guy wire images of different fracture types collected by UAV. Input the left and right images of the guy wire to be tested from the visual monitoring data into the fracture prediction model to output the fracture prediction probability at different positions on the guy wire to be tested. Compare each fracture prediction probability with the fracture threshold to determine the fracture position and fracture range used to reflect the fracture situation.

[0082] For example, this part of the algorithm can adopt an improved DeepLabv3 model + network structure for image semantic segmentation. Specifically, in the encoder part, depthwise separable convolution can be used to replace traditional convolution operations, reducing the number of model parameters and computational complexity. In the decoder part, a spatial pyramid pooling module is introduced. This module performs multi-scale feature extraction on the feature map through pooling operations at different scales, and then merges the feature maps of different scales and connects them with the feature maps of the corresponding layers of the encoder, enhancing the model's ability to extract features from fracture regions of different sizes. At the same time, CBAM (Convolutional Block Attention) is embedded in the network to weight the feature map in both the channel dimension and the spatial dimension, making the model pay more attention to feature regions related to fractures.

[0083] In DeepLabv3's encoder, the kernel size of the depthwise separable convolution is set to 3×3, the number of convolutional layers is set to 3, and the stride is set to 2. Spatial pyramid pooling: This module can contain three dilated convolutional branches of different scales, and their pooling scales can be set to 6, 12, and 18 respectively, to cover fracture features of different sizes.

[0084] Furthermore, a large number of left-right image pairs containing normal guy wires and different fracture types (such as partial fracture and complete fracture) can be collected by drones for training data annotation. During the training of the fracture prediction model, a combination of the cross-entropy loss function and the Dice loss function can be used as the optimization objective. (Cross-entropy loss function) The formula used to measure the difference between the predicted probability distribution and the true label distribution is:

[0085]

[0086] Where C is the total number of categories, It is the true label (0 or 1) of the pixel p belonging to category c. This is the probability that pixel p belongs to class c, as predicted by the model. (Dice loss function) The formula used to measure the degree of overlap between the predicted region and the actual region is:

[0087]

[0088] The total loss function can be set as follows: By adjusting the weight coefficient α to balance the contributions of the two losses, the Adam optimizer can be used to update the network parameters during training. The initial learning rate is set to 0.0001, and a cosine annealing learning rate adjustment strategy is adopted to gradually reduce the learning rate as training progresses.

[0089] In the model application phase, the left and right images of the guy wire to be tested can be input into the trained model to output semantic segmentation results of the normal and broken areas of the guy wire. Then, threshold segmentation can be used to convert the predicted probability map into a binary image, clearly showing the break location and extent. Furthermore, to improve detection efficiency, model pruning techniques can be used to compress the trained model, remove redundant parameters, and deploy a lightweight model on an edge computing terminal to achieve real-time detection.

[0090] Furthermore, if the attribute information includes the tension of the tension of the wire to be tested, and the visual monitoring data is acquired by a binocular vision acquisition unit acting as a vision machine, step 102 may include the following sub-step (tension detection):

[0091] S23. Construct a 3D point cloud of the main body of the guy wire to be tested using the left and right images from the visual monitoring data. The 3D point cloud is the point cloud after removing noise points. Determine the centerline equation of the guy wire to be tested based on the coordinate information of multiple points in the 3D point cloud, and calculate the morphological features of the guy wire to be tested using the centerline equation. The morphological features include the centerline curvature, the cross-sectional radius of the cross-sectional point cloud, and the ellipticity. The cross-sectional point cloud refers to the point cloud obtained by the intersection of the plane of the centerline and the 3D point cloud. Input the morphological features into the trained tension prediction model to output the tension of the guy wire to be tested. The tension is compared with the tension threshold to determine whether the guy wire to be tested is in an abnormal tension state.

[0092] This part of the algorithm obtains a disparity map by performing stereo matching on the left and right images, and then calculates the 3D coordinates of each point on the cable based on the principle of triangulation, thereby constructing the 3D point cloud data of the cable. However, when acquiring the point cloud data, environmental interference and measurement errors may occur, resulting in noisy points and outliers, requiring filtering. Specifically, representative points are automatically selected as seed points based on the density and distribution characteristics of the point cloud. The normal vector angle and distance between points in the current region are calculated and judged, and thresholds for the normal vector angle and distance are set. If the normal vector angle between the point to be judged and points in the region is less than the normal vector angle threshold, and the distance is less than the distance threshold, then the point is considered similar to the current region. Points that meet the similarity conditions are added to the current region. This process of selection and similarity judgment is repeated continuously, expanding the region until no new points can be added. Finally, noisy points that do not belong to the main body of the cable are removed.

[0093] The moving least squares method is used to fit a polynomial function to each point in the point cloud within its local neighborhood, thus calculating the new position of the point and achieving point cloud smoothing. Let a point in the point cloud... Given a local neighborhood of n points, and a fitted polynomial function f(x), the goal of the least squares method is to minimize the following error function E:

[0094]

[0095] in, These are the coordinates of points within the neighborhood. It is the actual value at that point. The weights are determined based on factors such as distance. By solving the above equation, the polynomial coefficients are obtained, and then the smoothed point positions are obtained.

[0096] Centerline fitting is performed on the guy wire. The centerline is fitted using the least squares method. Each point in the 3D point cloud data of the guy wire is projected onto a plane (usually a plane roughly parallel to the guy wire). Then, the parametric equation of the centerline is obtained by minimizing the sum of squared distances from each point to the fitted line. For example, the centerline equation can be set as y = ax + b, and the coordinates of the points in the point cloud are... The goal of the least squares method is to solve for the parameters. And b (which can be considered as a centerline parameter) make the error function :

[0097]

[0098] Solving this system of equations will give the values ​​of a and b, thus determining the equation of the centerline.

[0099] Calculate the curvature of the centerline. At each point on the centerline, the curvature value is obtained by calculating the ratio of the second derivative to the first derivative of the curve near that point. For the parametric curves x=x(t) and y=y(t), the curvature calculation formula is:

[0100]

[0101] Curvature reflects the degree of bending of the centerline; the curvature of the centerline will differ depending on the tension of the tension. At different locations along the centerline, a plane perpendicular to the centerline is selected and intersected with the point cloud of the tensioned wire to obtain a cross-sectional point cloud. We need to extract the cross-sectional features. An ellipse fitting algorithm is used to fit the cross-sectional point cloud. The general equation of an ellipse is:

[0102]

[0103] To simplify calculations, the cross-sectional point cloud is typically rotated and translated first, making its principal axes parallel to the coordinate axes. Then, the least squares method is used to fit the ellipse equation. After obtaining the ellipse equation, morphological features such as the cross-sectional radius and ellipticity are calculated. The cross-sectional radius can be taken as the average of the major and minor semi-axes of the ellipse, and the ellipticity is used to describe the degree to which the cross-section deviates from a circle.

[0104] Finally, a tension prediction model is established between the tension of the drawstring and its shape characteristics. This requires parameter calibration and data analysis using a large amount of experimental data, employing methods such as multiple linear regression or support vector regression to establish the mapping relationship between the two. Taking multiple linear regression as an example, assuming the tension is y and the shape characteristic parameters are... The regression equation is:

[0105]

[0106] in, It is the regression coefficient. This is the error term. The regression coefficients are solved using methods such as the least squares method to derive the tightness prediction model.

[0107] In the model application stage, the extracted morphological features to be tested can be input into the model to calculate the tension value of the tension of the tension cable to be tested, and compare it with the preset threshold to determine whether the tension cable to be tested is in an abnormal tension state.

[0108] In summary, this invention organically integrates UAV inspection technology, binocular vision technology, and edge computing and data analysis technology. The UAV, as a mobile platform, is equipped with a binocular vision acquisition unit to achieve flexible and efficient data acquisition; 5G / satellite communication ensures real-time and stable data transmission; and the edge computing and data analysis center completes data processing and status assessment, forming a complete and automated monitoring system that overcomes the limitations of traditional single technologies. In tilt angle detection, SIFT feature point matching and RANSAC algorithms are combined with the principle of binocular vision parallax. Through stereo correction and precise 3D coordinate calculation, high-precision tilt angle measurement is achieved. Fracture detection employs an improved DeepLabv3+ network structure, incorporating spatial pyramid pooling and CBAM modules, along with a specific loss function, to enhance the ability to identify minute cracks. Tightness detection involves complex processing and morphological feature extraction of the 3D point cloud acquired by binocular vision. Each algorithm is optimized for the characteristics of binocular vision, enabling accurate detection.

[0109] Furthermore, to better cope with complex external environments, the binocular vision acquisition unit in this embodiment of the invention can be equipped with an infrared supplementary light, enabling the vision acquisition unit to operate normally in illumination environments ranging from 0-10000 lx, effectively eliminating the effects of uneven illumination. For example, the inclusion of an infrared supplementary light allows the vision acquisition unit to acquire clear image data under different illumination conditions such as at night, on cloudy days, and under direct strong light, ensuring the normal operation of the detection algorithm and the accuracy of the detection results.

[0110] In practice, the order of execution of the above sub-steps S21, S22, and S23 can be unlimited or performed simultaneously. That is, the order of performing tilt angle, fracture condition, and tightness detection can be determined by the user.

[0111] In terms of effectiveness, drones equipped with binocular vision acquisition units can automatically fly along preset routes to collect data on power pole guy wires. Collaborating with edge computers and data analysis centers, they can efficiently automate data acquisition and processing. Furthermore, by detecting multiple key parameters of the guy wires, such as tilt angle, breakage, and tightness, they overcome the limitations of traditional single-parameter detection. Through multi-parameter collaborative analysis, they can intelligently and comprehensively monitor the overall condition of the guy wires, providing complete data support for the operation and maintenance management of power facilities. This helps in formulating scientific and reasonable maintenance strategies and improving the intelligent management level of the power system.

[0112] Furthermore, if the visual monitoring data is acquired via lidar, and the attribute information also includes the corrosion status of the wire under test, step 102 may include the following sub-steps:

[0113] S24. Based on the laser reflection signal belonging to the guy wire to be tested in the visual monitoring data, construct a three-dimensional point cloud for at least one period; based on the reflection intensity information of the three-dimensional point cloud, determine whether there is rust on the surface of the guy wire to be tested.

[0114] Airborne lidar can quickly acquire high-precision 3D point cloud data of utility pole guy wires by actively emitting laser beams and receiving reflected signals. The point cloud obtained by radar scanning contains the native physical parameter of reflection intensity, which refers to the intensity value of the laser reflection signal. This reflection intensity is directly related to the surface material, roughness, and degree of corrosion of the guy wire. Therefore, by combining the reflection intensity information of the point cloud, it is possible to preliminarily determine whether there are defects such as corrosion on the surface of the guy wire. Generally, healthy metal guy wires have smooth surfaces, high reflectivity to lasers, and high and stable reflection intensity values. Corroded areas are oxide layers (such as rust), with rough surfaces and porous materials, low reflectivity to lasers, and significantly reduced and fluctuating reflection intensity values. Therefore, the judgment method can be: statistically analyze the mean and variance of the reflection intensity of the guy wire point cloud. If the intensity value of a local area is lower than 70% of the mean of the baseline period (the threshold can be adjusted according to the actual scene), and the variance is greater than the threshold, it can be preliminarily determined that there is a corrosion defect in that area. If further confirmation is required, it can be combined with visible light image fusion analysis.

[0115] It should be further explained that, following the sub-step "constructing a three-dimensional point cloud for at least one period based on the laser reflection signal belonging to the guy wire in the visual monitoring data," the sub-step "comparing the three-dimensional point clouds of each period to detect at least one attribute information among the guy wire's tilt angle, breakage condition, and tightness" can also be included. Therefore, as an alternative solution, multi-period point cloud data comparison and reflection intensity analysis from radar can be combined to achieve integrated detection of guy wire morphology changes and surface defects.

[0116] In practice, for the above two sub-steps: "comparing the three-dimensional point clouds of different periods to detect at least one of the attribute information of the tension angle, fracture condition and tightness of the wire to be tested", and "determining whether there is corrosion on the surface of the wire to be tested based on the reflection intensity information of the three-dimensional point cloud", at least one of these two sub-steps can be executed as needed, and the order of execution between these two sub-steps can be unlimited or executed simultaneously.

[0117] Taking a binocular vision acquisition unit as an example, this invention aims to protect the complete data processing flow from UAV data acquisition and image preprocessing (denoising, stereo correction, enhancement), to the application of various algorithms for tilt angle, breakage, and tightness detection, and finally to the data analysis center's evaluation and early warning. This includes the specific technologies and methods used in each stage of the process. It also protects the overall system architecture design based on a UAV equipped with a binocular vision acquisition unit, an edge computing terminal, and a data analysis center, covering the connection relationships, functional configurations, and collaborative working methods of each part. This invention can achieve the following technical effects:

[0118] (1) Achieve non-contact monitoring of the line: No equipment needs to be installed on the guy wire. The detection can be achieved without changing the original structure. The present invention can achieve stable detection of the guy wire, can avoid severe weather conditions for monitoring, and will not put a burden on the guy wire.

[0119] (2) Solve the problem of high installation cost: This invention does not require hardware installation at each guy wire point. Only one drone is needed to conduct regular inspections of multiple guy wire points and analyze and investigate safety hazards and faults in the guy wires through visual algorithms.

[0120] (3) Improve the accuracy of abnormal warnings: The present invention has designed a fast data processing and response process, which can quickly issue early warning signals when the monitored data shows an abnormal trend.

[0121] (4) Provide support structure for further use of algorithm judgment and optimization: Reserve algorithm module interface in system architecture to support the operation of intelligent model based on monitoring data, and lay the platform foundation for subsequent self-learning, adaptive analysis and control strategies.

[0122] (5) Optimize human-computer interaction experience: Adopt a clear and intuitive graphical interface to display various operating statuses and warning information in real time, so that operators can quickly grasp the overall safety situation of the operation.

[0123] Please see Figure 2 , Figure 2 This is a structural block diagram of a pole guy wire inspection system provided in an embodiment of the present invention.

[0124] The present invention provides a pole guy wire inspection system, comprising:

[0125] The acquisition module 201 is used to acquire visual monitoring data of the guy wire to be tested during the flight of the UAV; the visual monitoring data is acquired by a vision machine mounted on the UAV.

[0126] The detection module 202 is used to detect the attribute information of the guy wire under test based on visual monitoring data to determine whether the guy wire under test is in an abnormal state; the attribute information includes at least one of the following: the tilt angle, the breakage status, and the tightness of the guy wire under test.

[0127] Furthermore, the acquisition module 201 may perform the following steps:

[0128] The Hough transform algorithm based on the attention mechanism is used to map the original image acquired by the vision machine from the pixel space to the parameter space to suppress visual data interference caused by non-pull-wire areas and highlight the pull-wire area in the parameter space, so as to extract the center line parameters corresponding to the pull-wire to be tested.

[0129] Based on the centerline parameters, determine the UAV's trajectory and flight status parameters for the test cable;

[0130] The drone is controlled to fly according to the aircraft trajectory and flight status parameters so that visual monitoring data can be collected by the vision machine.

[0131] Furthermore, if the attribute information includes the tilt angle of the wire to be measured, and the visual monitoring data is acquired by a binocular vision acquisition unit acting as a vision machine, the detection module 202 can perform the following steps:

[0132] Feature point matching is performed on the left and right images of the wire to be tested in the visual monitoring data to obtain matching point pairs that meet the preset conditions;

[0133] By matching the coordinate information of point pairs, disparity values ​​corresponding to feature points at different positions on the test line are generated.

[0134] Based on the disparity value and the two-dimensional coordinate information of each feature point in the left or right image, the three-dimensional spatial straight line equation of the guy wire to be tested is fitted, and the tilt angle of the guy wire to be tested is calculated through the three-dimensional spatial straight line equation.

[0135] Furthermore, if the attribute information includes the breakage status of the guy wire under test, the visual monitoring data is acquired by a binocular vision acquisition unit, which acts as a vision machine. The detection module 202 can then perform the following steps:

[0136] Obtain the trained fracture prediction model; the fracture prediction model is obtained by iteratively training the training data through convolution and pooling strategies. The training data includes normal cable images and abnormal cable images of different fracture types collected by UAV.

[0137] Input the left and right images of the guy wire to be tested from the visual monitoring data into the fracture prediction model to output the fracture prediction probability at different positions on the guy wire to be tested.

[0138] Each predicted fracture probability is compared with a fracture threshold to determine the fracture location and fracture extent used to reflect the fracture situation.

[0139] Furthermore, if the attribute information includes the tension of the tension of the wire to be tested, and the visual monitoring data is acquired by a binocular vision acquisition unit acting as a vision machine, the detection module 202 can perform the following steps:

[0140] A 3D point cloud of the main body of the guy wire under test is constructed using the left and right images from the visual monitoring data; the 3D point cloud is the point cloud after removing noise points.

[0141] The centerline equation of the cable to be tested is determined based on the coordinate information of multiple points in the 3D point cloud, and the morphological features of the cable to be tested are calculated through the centerline equation. The morphological features include the centerline curvature, the cross-sectional radius of the cross-sectional point cloud, and the ellipticity. The cross-sectional point cloud refers to the point cloud obtained by the intersection of the plane of the centerline and the 3D point cloud.

[0142] The morphological features are input into the trained tension prediction model to output the tension of the tension of the wire to be tested; the tension is compared with the tension threshold to determine whether the wire to be tested is in an abnormal tension state.

[0143] Furthermore, if the visual monitoring data is acquired by a lidar device acting as a visual machine, and the attribute information also includes the corrosion status of the wire under test, the detection module 202 can perform the following steps:

[0144] Based on the laser reflection signal belonging to the test wire in the visual monitoring data, construct a three-dimensional point cloud for at least one period.

[0145] Based on the reflection intensity information of the three-dimensional point cloud, determine whether there is corrosion on the surface of the wire to be tested.

[0146] Please see Figure 3 , Figure 3 This is a structural block diagram of an electronic device provided in an embodiment of the present invention.

[0147] An electronic device according to an embodiment of the present invention includes: a memory 301 and a processor 302. The memory 301 stores a computer program. When the computer program is executed by the processor 302, the processor 302 performs the pole guy wire inspection method as described in the above embodiment.

[0148] Memory 301 may be an electronic memory such as flash memory, EEPROM (Electrically Erasable Programmable Read-Only Memory), EPROM, hard disk, or ROM. Memory 301 has storage space 303 for program code 313 for performing any of the method steps described above. For example, storage space 303 for program code may include various program codes 313 for implementing the various steps in the methods described above. These program codes may be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, CDs, memory cards, or floppy disks. The program code may be compressed, for example, in a suitable form. When run by a computing processing device, this code causes the computing processing device to perform the various steps in the methods described above. These program codes may be read from or written to one or more computer program products. These computer program products include program code carriers such as hard disks, CDs, memory cards, or floppy disks. The program code may be compressed, for example, in a suitable form. When this code is run by a computing device, it causes the device to perform the various steps in the pole guy wire inspection method described above.

[0149] This invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the pole guy wire inspection method as described above.

[0150] This invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions, wherein when the program instructions are executed by a computer, the computer performs the pole guy wire inspection method as described above.

[0151] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0152] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces, or indirect coupling or communication connection between apparatuses or units, and may be electrical, mechanical, or other forms.

[0153] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0154] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0155] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0156] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for inspecting guy wires on utility poles, characterized in that, include: Acquire visual monitoring data of the guy wire under test during the flight of the drone; The visual monitoring data is collected by a vision machine mounted on the drone. The attribute information of the test cable is detected based on the visual monitoring data to determine whether the test cable is in an abnormal state; the attribute information includes at least one of the following: the tilt angle, breakage status, and tension of the test cable.

2. The pole guy wire inspection method according to claim 1, characterized in that, The steps for acquiring visual monitoring data of the guy wire under test during the flight of the UAV include: The Hough transform algorithm based on the attention mechanism is used to map the original image acquired by the vision machine from the pixel space to the parameter space to suppress visual data interference caused by non-pull-wire areas and highlight the pull-wire area in the parameter space, so as to extract the center line parameters corresponding to the pull-wire to be tested. Based on the centerline parameters, the aircraft trajectory and flight status parameters of the UAV for the test cable are determined; The drone is controlled to fly according to the aircraft trajectory and flight status parameters so that the visual monitoring data can be collected by the vision machine.

3. The pole guy wire inspection method according to claim 1, characterized in that, If the attribute information includes the tilt angle of the guy wire to be tested, and the visual monitoring data is acquired by a binocular vision acquisition unit serving as the vision machine, then the step of detecting the attribute information of the guy wire to be tested based on the visual monitoring data includes: Feature point matching is performed on the left and right images of the test cable in the visual monitoring data to obtain matching point pairs that meet preset conditions; Using the coordinate information of the matching point pairs, disparity values ​​corresponding to feature points at different positions on the test line are generated; Based on the disparity value and the two-dimensional coordinate information of each feature point in the left or right image, the three-dimensional spatial straight line equation of the wire to be tested is fitted, and the tilt angle of the wire to be tested is calculated through the three-dimensional spatial straight line equation.

4. The pole guy wire inspection method according to claim 1, characterized in that, If the attribute information includes the breakage status of the guy wire under test, and the visual monitoring data is acquired by a binocular vision acquisition unit serving as the vision machine, then the step of detecting the attribute information of the guy wire under test based on the visual monitoring data includes: A trained fracture prediction model is obtained; the fracture prediction model is obtained by iteratively training the training data through convolution and pooling strategies, and the training data includes normal cable images and abnormal cable images of different fracture types collected by the UAV. The left and right images of the guy wire to be tested from the visual monitoring data are input into the fracture prediction model to output the fracture prediction probability at different positions on the guy wire to be tested. Each fracture prediction probability is compared with a fracture threshold to determine the fracture location and fracture extent that reflect the fracture condition.

5. The pole guy wire inspection method according to claim 1, characterized in that, If the attribute information includes the tension of the tension of the tension cable to be tested, and the visual monitoring data is acquired by a binocular vision acquisition unit serving as the vision machine, then the step of detecting the attribute information of the tension cable to be tested based on the visual monitoring data includes: A three-dimensional point cloud belonging to the main body of the pull wire under test is constructed using the left and right images in the visual monitoring data; the three-dimensional point cloud is a point cloud after removing noise points; The centerline equation of the guy wire to be tested is determined based on the coordinate information of multiple points in the three-dimensional point cloud, and the morphological features of the guy wire to be tested are calculated through the centerline equation; wherein, the morphological features include the centerline curvature, the cross-sectional radius of the cross-sectional point cloud, and the ellipticity, and the cross-sectional point cloud refers to the point cloud obtained by the intersection of the plane of the centerline and the three-dimensional point cloud. The morphological features are input into the trained tension prediction model to output the tension of the test cable; the tension is compared with a tension threshold to determine whether the test cable is in an abnormal tension state.

6. The pole guy wire inspection method according to any one of claims 1-5, characterized in that, If the visual monitoring data is acquired by a lidar device acting as the visual machine, and the attribute information also includes the corrosion status of the guy wire under test, then the step of detecting the attribute information of the guy wire under test based on the visual monitoring data includes: Based on the laser reflection signal belonging to the test wire in the visual monitoring data, a three-dimensional point cloud for at least one period is constructed. Based on the reflection intensity information of the three-dimensional point cloud, it is determined whether there is corrosion on the surface of the wire to be tested.

7. A pole guy wire inspection system, characterized in that, include: The acquisition module is used to acquire visual monitoring data of the guy wire under test during the flight of the UAV; The visual monitoring data is collected by a vision machine mounted on the drone. The detection module is used to detect the attribute information of the guy wire under test based on the visual monitoring data, so as to determine whether the guy wire under test is in an abnormal state; The attribute information includes at least one of the following: the tilt angle, breakage condition, and tension of the wire to be tested.

8. An electronic device, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor causes the processor to perform the steps of the pole guy wire inspection method as described in any one of claims 1-6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed, it implements the pole guy wire inspection method as described in any one of claims 1-6.

10. A computer program product, characterized in that, The computer program product includes a computer program stored on a non-transitory computer-readable storage medium, the computer program including program instructions, wherein when the program instructions are executed by a computer, the computer performs the pole guy wire inspection method as described in any one of claims 1-6.

Images