A Smart Method for Determining Conductor Sag of Transmission Lines Based on UAV Inspection Images

Abstract

This invention relates to the fields of computer vision and power equipment inspection technology, and particularly to an intelligent method for identifying conductor sag in transmission lines based on UAV inspection images. The method includes: S1: Identifying conductor regions based on UAV visible light images and extracting continuous conductor edges; S2: Locating insulator strings within the conductor regions and accurately locating conductor splicing points through gradient direction aggregation analysis; S3: Generating a spatial baseline based on a digital elevation model and the line alignment; S4: Performing physical constraint-based conductor morphology reconstruction, connecting broken edges using the physical characteristics of the conductor's catenary as constraints, and projecting the reconstructed conductor onto a plane perpendicular to the baseline; S5: Calculating the maximum vertical distance from the projection point to the baseline as the sag value, and dynamically diagnosing anomalies based on ambient temperature. By combining image processing, deep learning, physical models, and terrain adaptation techniques, the accuracy and reliability of conductor sag detection can be effectively improved.

Description

Technical Field

[0001] This invention relates to the fields of computer vision and power equipment inspection technology, and in particular to an intelligent method for judging the sag of transmission line conductors based on UAV inspection images. Background Technology

[0002] Conductor sag in power transmission lines is a key parameter for assessing the safe operation of the line; an abnormally large sag can lead to accidents such as flashover and line breakage. Currently, the power industry widely uses visible light imagery from drones for sag detection, but the following technical bottlenecks have long remained unresolved:

[0003] 1. Distortion of feature point positioning caused by insulator string obstruction;

[0004] Existing methods directly identify the connection points between conductors and towers through template matching. However, in real-world inspection scenarios:

[0005] The insulator string is composed of multiple disc-shaped insulators connected in series, and its complex texture forms a high-frequency interference pattern in the image.

[0006] The connection point is blocked by insulator fittings at a rate as high as 60%-80%;

[0007] The occlusion caused the positioning deviation of the attachment point to generally exceed 10 pixels (equivalent to an actual position deviation of >30cm). After the error was amplified by the sag calculation formula, the final result deviated from the true value by more than 15%.

[0008] 2. Errors in baseline setting caused by terrain elevation differences;

[0009] Current standards use the horizontal baseline method to calculate sag, but in mountainous transmission lines:

[0010] The height difference between adjacent towers can reach 50-200 meters (such as the UHV transmission line in the Yunnan-Guizhou Plateau);

[0011] There is an angular deviation between the static horizontal baseline and the actual spatial projection plane of the traverse (typical value 3°-8°);

[0012] This causes the calculated sag value to systematically deviate from the actual physical value, and the amount of deviation is positively correlated with the terrain slope (experimental data shows that for every 1° increase in slope, the error increases by 2.1%).

[0013] 3. The conductor edge breaks under complex lighting conditions;

[0014] Mainstream edge detection algorithms (such as Canny and Sobel) are widely used for wire extraction, but in practical applications:

[0015] Highly reflective areas on the surface of the conductor form oversaturated pixel blocks (occupying more than 40% of the conductor area);

[0016] The shadow cast by the mountain causes the local contrast to drop below 10 (the ideal detection threshold needs to be >30);

[0017] The generated conductors exhibit severe edge fragmentation, with break gaps reaching 5-10 times the conductor diameter, making it impossible to reconstruct a continuous conductor morphology.

[0018] Therefore, there is an urgent need for an intelligent method for judging the sag of power transmission line conductors based on UAV inspection images to solve the above problems. Summary of the Invention

[0019] To achieve the above objectives, this invention provides an intelligent method for determining conductor sag in power transmission lines based on UAV inspection images, comprising:

[0020] S1: Identify the wire region based on UAV visible light images and extract the edges of continuous wires;

[0021] S2: Locate the insulator string within the conductor area and accurately locate the conductor splicing point through gradient direction aggregation analysis;

[0022] S3: Construct a terrain-adaptive dynamic baseline, obtain the three-dimensional geographic coordinates of adjacent tower connection points, and generate a spatial baseline based on the digital elevation model and route alignment;

[0023] S4: Perform physical constraints to reconstruct the shape of the conductor, connect the broken edges with the physical properties of the catenary as constraints, and project the reconstructed conductor onto the vertical plane of the baseline.

[0024] S5: Calculate the maximum vertical distance from the projection point to the baseline as the sag value, and combine it with the ambient temperature to dynamically diagnose anomalies.

[0025] Preferably, the gradient direction aggregation analysis in S2 includes:

[0026] A rectangular search area is defined at the end of the insulator string. This area extends a predetermined length along the direction of conductor extension and its width is an integer multiple of the width of the insulator string.

[0027] For each pixel within the search region, calculate the consistency of its neighborhood gradient direction:

[0028] Extract all edge points in the neighborhood centered at the pixel, whose size is related to the diameter of the wire;

[0029] Calculate the absolute value of the angle deviation between the gradient direction and the vertical direction at each edge point;

[0030] The arithmetic mean of the absolute values ​​of all included angle deviations is used as the consistency score for that pixel;

[0031] The pixel with the lowest consistency score is identified as the splice point. The physical basis for this is that the edge gradient direction at the splice point of the conductor is vertically distributed.

[0032] Preferably, the generation of the dynamic baseline in S3 includes:

[0033] 3D coordinate calculation of the attachment point:

[0034] Obtain the latitude, longitude, and altitude of the pole base point recorded by the drone positioning system;

[0035] The relative elevation of the attachment point is calculated based on the pixel offset between the attachment point and the tower base point in the image, combined with the drone's lens pitch angle, flight altitude and focal length.

[0036] The pixel coordinates are mapped to the geographic coordinate system through a coordinate transformation model;

[0037] Spatial baseline construction:

[0038] Determine the azimuth angle of the route based on the direction of the line connecting the two connection points.

[0039] Topographic profile lines are generated by interpolation in the digital elevation model along the direction of the connecting lines;

[0040] Using the two connection points as endpoints, generate a three-dimensional straight line in the plane containing the terrain profile line.

[0041] Preferably, the wire shape reconstruction of physical constraints in S4 includes:

[0042] Rules for connecting fracture edges:

[0043] For pairs of break points whose distance is less than the maximum suspension span of the conductor, candidate connection paths are generated;

[0044] The rate of curvature change of the candidate path is calculated, which is obtained by the second-order difference approximation of the path point set;

[0045] A connection is established when the rate of change of curvature is within the allowable elastic deformation range of the conductor material;

[0046] Catenary property verification:

[0047] Fit the catenary equation to the reconstructed continuous traverse point set;

[0048] Remove outliers that deviate from the fitted curve by more than the diameter of the conductor.

[0049] Preferably, the calculation of the sag value in S5 includes:

[0050] Maximum vertical distance positioning:

[0051] Establish a local coordinate system for the baseline on the projection plane;

[0052] Calculate the normal distance from each traverse point to the baseline;

[0053] Distance sequences are processed using a sliding window smoothing method, with the window length proportional to the conductor span; temperature compensation mechanism:

[0054] Obtain the coefficient of thermal expansion of the material from the conductor type database;

[0055] The sag correction factor is calculated based on the difference between the real-time ambient temperature and the standard temperature.

[0056] The dynamic threshold is obtained by multiplying the standard sag threshold by the correction factor.

[0057] Preferably, in the calculation of the neighborhood gradient direction consistency:

[0058] The method for determining the neighborhood size is as follows:

[0059] Multiple regions are randomly sampled within the conductor area, and the average distribution density of the conductor edge points is calculated. The minimum neighborhood size required to cover a single conductor is then calculated based on the distribution density.

[0060] Statistical optimization of included angle deviation:

[0061] The conductor is cut off if the deviation exceeds the bending limit angle.

[0062] A weighted average strategy is adopted, and the weights are positively correlated with the gradient magnitude of the edge points.

[0063] Preferably, the establishment of the coordinate transformation model includes:

[0064] Camera imaging geometry correction:

[0065] The pixel coordinates of the mounting point are corrected using the distortion parameters of the drone lens;

[0066] Spatial location calculation:

[0067] Establish a rotation matrix that includes the pitch, roll, and yaw attitude angles of the UAV;

[0068] The corrected pixel coordinates are transformed to the ground coordinate system using collinearity equations.

[0069] Combined multi-view images to optimize the 3D coordinates of the attachment point using bundle adjustment.

[0070] Preferably, the allowable range of the rate of change of curvature is determined by:

[0071] Obtain the elastic modulus and design tension corresponding to the conductor type;

[0072] Derivation of the curvature change rate threshold based on the catenary differential equation:

[0073]

[0074] Where T0 is the horizontal tension, w is the weight per unit length, E is the elastic modulus, and A is the cross-sectional area; in practical applications, a preset percentage of the theoretical threshold is taken as the safety margin.

[0075] Preferably, the parameter settings for the sliding window smoothing method are as follows:

[0076] Mapping relationship between window length and margin:

[0077] Establish a statistical model for the frequency of sag variation under different gauge lengths;

[0078] Query the optimal window length coefficient in the statistical model based on the gap value;

[0079] Smooth weight distribution:

[0080] A Gaussian weight distribution is used, with the peak weight position corresponding to the center of the window;

[0081] The standard deviation is adaptively adjusted based on the amplitude of conductor vibration.

[0082] Preferably, it also includes a verification mechanism for abnormal diagnosis:

[0083] Spatial consistency verification:

[0084] Detect the sag value of the same span in the overlapping area of ​​adjacent routes;

[0085] A recalculation is initiated when the difference exceeds the measurement error tolerance.

[0086] Time continuity verification:

[0087] Compare the current sag value with historical values ​​under the same temperature conditions;

[0088] If the mutation rate exceeds the material creep limit, a manual review will be triggered.

[0089] The beneficial effects of this invention are:

[0090] 1. This invention, through an improved feature point detection algorithm, can effectively identify and eliminate the interference of complex textures of insulator strings on images. By introducing multi-scale template matching and adaptive filtering techniques, this invention can accurately detect the connection points between conductors and towers, ensuring that the positioning error of feature points is controlled within an acceptable range even when insulator fittings are obstructed. The error in sag calculation results can be greatly reduced, and the accuracy improved.

[0091] 2. This invention can automatically adjust the baseline tilt angle according to different terrain changes, eliminating calculation errors caused by terrain elevation differences. By utilizing aerial elevation data from UAVs, it corrects the elevation difference and tilt angle between adjacent towers in real time, and the calculated sag result accurately reflects the actual conductor condition. Experiments show that this method can effectively reduce errors caused by changes in terrain slope and avoid the limitations of the traditional horizontal baseline method in mountainous lines.

[0092] 3. This invention introduces an improved edge detection algorithm, combining adaptive threshold segmentation with a deep learning model. This automatically optimizes the threshold under different lighting conditions, accurately extracting conductor edges. Through multi-channel fusion processing, this method effectively identifies areas with uneven lighting and overlapping shadows, repairs broken sections of conductor edges, and reconstructs continuous conductor shapes. This method can accurately extract conductor edges even in complex environments with strong lighting or shadow obstruction, ensuring accurate sag calculation. Attached Figure Description

[0093] To more clearly illustrate the technical solutions in this invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, those skilled in the art can obtain other drawings based on these drawings without creative effort.

[0094] Figure 1 This is a flowchart of the steps of the method of the present invention;

[0095] Figure 2 This is a flowchart illustrating the steps involved in establishing the coordinate transformation model of the method of the present invention.

[0096] Figure 3 This is a flowchart illustrating the parameter setting steps for the sliding window smoothing method of the present invention. Detailed Implementation

[0097] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. It should also be noted that, to make the embodiments more comprehensive, the following embodiments are the best and preferred embodiments, and those skilled in the art can use other alternative methods to implement some well-known technologies; moreover, the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0098] Please see Figures 1-3This invention provides an intelligent method for identifying conductor sag in power transmission lines based on images from UAV inspections. In step S1, during UAV inspections, image data of the power transmission line is acquired using high-resolution visible light images. Advanced image processing algorithms (such as Canny edge detection and deep learning models) are used to segment the conductor region and automatically identify conductor edges. A continuous edge detection algorithm removes noise and irrelevant objects, ensuring that clear and continuous conductor edges are extracted. The beneficial effect of this step is that it enables stable extraction of conductor edges under complex lighting conditions, such as strong reflections or shadows, improving the accuracy of conductor identification and solving the problem of severe fragmentation in traditional edge detection algorithms.

[0099] In S2, insulator strings are located within the conductor region using deep learning and image segmentation techniques. Insulator strings consist of multiple disc insulators connected in series, and their complex textures often interfere with image analysis. This technique employs a gradient direction aggregation analysis method, using segmentation and feature extraction to accurately locate the conductor-to-tower connection point. By eliminating the influence of insulator hardware obstruction, high-precision connection point positioning is achieved. Its advantages include avoiding positioning deviations caused by insulator obstruction in traditional methods, significantly improving the accuracy of connection point positioning, and ensuring the accuracy of sag calculation.

[0100] In S3, a baseline is constructed using a digital elevation model (DEM) and 3D images captured by drones, combined with terrain data. In complex terrains, such as mountains or slopes, this baseline adaptively adjusts according to the elevation differences in different areas. By acquiring the 3D geographic coordinates of adjacent towers, a spatial baseline is generated based on the route alignment, eliminating errors introduced by traditional horizontal baseline methods. This technology can accurately adapt to elevation changes in complex terrain, improving the accuracy of sag calculations and avoiding systematic errors caused by terrain variations.

[0101] In S4, the physical properties of the catenary are used as constraints to reconstruct the conductor's shape. By combining the physical model of the broken edge and the conductor, the edges at the break are connected to generate a complete conductor shape. The reconstructed conductor shape is projected onto the vertical plane of the baseline to ensure accurate reconstruction of the conductor's true shape. The advantage of this technique is that it can solve the shape distortion problem caused by changes in lighting or edge detection breaks in traditional methods, accurately restoring the true shape of the conductor.

[0102] In step S5, the maximum vertical distance of the conductor on the vertical plane of the baseline is calculated, i.e., the sag value. Simultaneously, considering changes in ambient temperature, the sag value judgment criteria are dynamically adjusted based on temperature data for anomaly diagnosis. By monitoring and analyzing sag data in real time, this invention can accurately identify sag anomalies, provide early warnings of potential line faults, and take preventative maintenance measures. Its beneficial effect is the ability to dynamically and accurately diagnose conductor conditions in conjunction with environmental changes, promptly detect anomalies, and improve the safety and reliability of transmission lines.

[0103] This invention, through an innovative combination of image processing, deep learning, physical models, and terrain adaptation technology, can effectively improve the accuracy and reliability of conductor sag detection, providing the power industry with a more efficient and intelligent transmission line inspection solution.

[0104] In one possible implementation, firstly, the end of the insulator string is identified in the image, and a rectangular search area is defined at its location. This search area extends a predetermined length along the direction of conductor extension, while its width is set to an integer multiple of the width of the insulator string. The purpose of selecting this area is to ensure that it includes the area near the connection point between the insulator string and the conductor, and is large enough to accommodate different shapes of conductors and connection points.

[0105] Within the defined search area, for each pixel, extract all edge points within the neighborhood centered on that pixel, whose size is related to the diameter of the wire. Then, calculate the absolute value of the angle deviation between the gradient direction and the vertical direction of these edge points.

[0106] The arithmetic mean of the absolute values ​​of the included angle deviations of all edge points within the neighborhood is calculated and used as the gradient consistency score for that pixel. This score reflects the perpendicularity of the edge direction within the neighborhood of that pixel.

[0107] Finally, the pixel with the lowest consistency score is determined as the wire splice point. According to physical principles, the edge gradient direction of the wire splice point is usually vertically distributed, that is, the edge at the splice point should be almost perpendicular to the tower direction. Therefore, the pixel with the lowest consistency score represents the location of that splice point.

[0108] The method for locating splices based on gradient direction aggregation analysis solves the problem of feature point location distortion caused by insulator string obstruction. It can accurately identify conductor splices in complex environments and provide accurate data for subsequent intelligent sag identification.

[0109] In one possible implementation, the latitude, longitude, and altitude of the pole's base point are first recorded using the drone's positioning system (such as a GPS module). This data provides a geographical reference for subsequent calculations of the relative position of the anchoring point.

[0110] Based on the pixel offset between the splice point and the tower base point in the image, combined with the drone's camera's pitch angle, flight altitude, and focal length, the elevation difference between the splice point and the tower base point is calculated. Specifically, using the pixel offset and the drone's flight parameters, the relative elevation of the splice point is calculated based on the principle of similar triangles. This method can accurately calculate the vertical position of the splice point, ensuring the precise determination of conductor sag.

[0111] A coordinate transformation model is used to map the coordinates of the hook point in the pixel coordinate system to the geographic coordinate system. By considering factors such as the curvature of the Earth and map projection, it is ensured that the transformed coordinates are consistent with the real-world geographic coordinate system (such as WGS-84). This step ensures an accurate correspondence between image coordinates and actual geographical locations.

[0112] Based on the direction of the line connecting the two connection points, the azimuth angle of that line is calculated. The azimuth angle, calculated clockwise from north, determines the direction of the transmission line. This directional information lays the foundation for generating a spatial baseline.

[0113] Interpolation is performed in the Digital Elevation Model (DEM) along the direction of the line connecting the two anchor points to generate a terrain profile line parallel to the direction of the line. The DEM contains elevation data at different locations along the Earth's surface; through interpolation, the accurate elevation at any location along the direction of the line connecting the two anchor points can be obtained. This step helps to account for the impact of terrain undulations on the traverse suspension, improving calculation accuracy.

[0114] Using the two anchor points as endpoints, a three-dimensional straight line is generated within the plane containing the generated terrain profile, utilizing the two anchor points and terrain data. This straight line not only considers the coordinates of the two anchor points but also incorporates terrain variations, thus generating a precise spatial baseline. This baseline will serve as a reference for subsequent traverse sag calculations.

[0115] Through efficient three-dimensional coordinate calculation and spatial baseline construction, this invention has significant advantages in dynamically adjusting the baseline and adapting to complex terrain, effectively improving the accuracy and reliability of intelligent judgment of conductor sag in transmission lines.

[0116] In one possible implementation, for break points less than the maximum suspension span of the conductor, a candidate connection path is first generated. The maximum suspension span refers to the maximum horizontal distance the conductor can suspend without external force. This distance ensures that a reasonable connection path exists between the two ends of the break. The candidate path is generated by analyzing the spatial relationship between the two break points to preliminarily predict how the conductor should be connected.

[0117] Furthermore, the rate of change of curvature for the candidate paths is calculated. Curvature, a measure of the degree of bending of a curve, is approximated by the second difference of the path point set. The purpose of this calculation is to assess whether the candidate paths conform to the actual bending characteristics of the conductor. The second difference of curvature reflects the degree of change in path bending; if the rate of change is too large, it may indicate that the path is unreasonable or does not conform to physical laws. This method can filter out unsuitable connection paths.

[0118] A connection is established only when the calculated rate of curvature change is within the allowable elastic deformation range of the conductor material. The conductor material has a certain range of elastic deformation; within this range, the conductor can bend freely without permanent deformation or breakage. This step ensures that the connection path conforms to the physical properties of the conductor, avoiding incorrect connections under conditions that do not conform to the material properties.

[0119] After establishing the connections between the breakpoints, the next task is to fit the reconstructed set of conductor points. The catenary is a standard curve describing the bending shape of an ideal conductor, suitable for suspended conductors in static equilibrium. By fitting the catenary equation, the reconstructed conductor shape can be made more consistent with actual physical conditions. The catenary equation accurately reflects the natural shape of the conductor under gravity.

[0120] After fitting, some reconstructed points may deviate significantly from the catenary curve. These points could be due to computational errors, image noise, or other external interference. By setting a deviation tolerance, points that deviate from the fitted curve by more than the diameter of the conductor are considered outliers and discarded. The purpose of this step is to remove abnormal data that does not conform to physical laws, ensuring that the final reconstructed conductor shape is accurate.

[0121] By introducing fracture edge connection rules and catenary characteristic verification, this invention provides a physically constrained conductor morphology reconstruction method, which not only improves reconstruction accuracy but also ensures the consistency between conductor morphology and actual physical properties, effectively solving the problem of conductor fracture detection in UAV inspection images.

[0122] In one possible implementation, firstly, a baseline needs to be determined on the projection plane. This baseline is typically based on the ideal location or trajectory of the transmission line. For ease of calculation, a local coordinate system for the baseline is constructed based on its starting point or other calibration points. This coordinate system defines the positions of all conductor points in the plane.

[0123] For each traverse point, calculate its normal distance to the baseline. This distance represents the vertical distance the traverse has deviated from its ideal position (the baseline). These normal distances reflect the magnitude of the traverse sag; the larger the sag, the farther the traverse deviates from the baseline.

[0124] Because the structure and sag of a conductor are affected by various factors, the normal distance between points may fluctuate. Therefore, a sliding window smoothing method is used to process these distance data. The sliding window method smooths the distance sequence by setting a window length (usually proportional to the span of the conductor), removing noise and abrupt changes. The choice of window length is closely related to the physical characteristics of the conductor; an appropriate window length can effectively smooth the data, thus more accurately reflecting the overall sag variation of the conductor.

[0125] Based on the conductor type database, obtain the coefficient of thermal expansion of the conductor material. The coefficient of thermal expansion is a physical quantity that describes the expansion characteristics of a material under temperature changes. Different conductor materials have different coefficients of thermal expansion, and understanding this coefficient is crucial for accurately calculating sag.

[0126] Sag will change with ambient temperature, especially at higher temperatures where the conductor will sag due to thermal expansion. A correction factor can be calculated based on the difference between the real-time ambient temperature and the standard temperature (usually 25°C).

[0127] The formula for calculating the correction factor is:

[0128] Correction factor = 1 + α(T - T0);

[0129] Where α is the coefficient of thermal expansion, T is the real-time temperature, and T0 is the standard temperature.

[0130] A dynamic sag threshold is obtained by multiplying the standard sag threshold (i.e., the sag limit defined at standard temperature) by a correction factor. This dynamic threshold takes into account the effect of temperature changes on sag, thus more accurately determining whether the conductor exceeds the normal sag range.

[0131] By combining maximum vertical distance positioning, sliding window smoothing method and temperature compensation mechanism, this invention can effectively improve the calculation accuracy and adaptability of sag value, thereby providing an intelligent and accurate method for judging the sag of transmission lines, which helps to improve the inspection efficiency and safety of transmission lines.

[0132] In one possible implementation, the image is first sampled within the conductor region, and multiple different regions are randomly selected for analysis. Each selected region should cover different parts of the conductor, ensuring that no detail of the conductor is missed. In this way, the distribution of conductor edge points in the image can be obtained.

[0133] Within each randomly selected region, the number of points on the conductor edge is counted, and their average distribution density is calculated. Regions with higher distribution density generally indicate more defined features on the conductor edges, while regions with lower distribution density may require higher resolution to extract effective features.

[0134] Based on the distribution density of points at the conductor's edge, the minimum neighborhood size required to cover a single conductor can be deduced. This minimum neighborhood size is the area in the image that can effectively capture the changes in conductor sag. Higher distribution density requires a smaller neighborhood size, while lower distribution density may require a larger neighborhood size to obtain more comprehensive image information.

[0135] This method can automatically and dynamically adjust the neighborhood size based on the density of points on the conductor edge, ensuring that features in different regions are processed appropriately. Through such adaptive adjustment, the processing accuracy of the image can be effectively improved without distorting the conductor information, especially in complex environments (such as images with high noise or background interference).

[0136] When detecting the curvature of a conductor, if the deviation exceeds the set limit angle for conductor curvature, the conductor is cut off. This means that if the calculated curvature angle exceeds the physically possible maximum value, it is limited to a reasonable range. This process effectively avoids abnormal calculation results caused by algorithm errors or image quality issues.

[0137] When calculating the consistency of gradient directions in the neighborhood, a weighted average strategy is adopted, which uses the gradient magnitude of edge points as the basis for weighting. Edge points with larger gradient magnitudes have a greater impact on the final calculation result, while edge points with smaller gradient magnitudes have a lower weight. This approach can better highlight the influence of significant features of the conductor edges and improve the accuracy of sag detection.

[0138] By truncating the deviation values, erroneous judgments caused by extreme bending angles can be avoided. Simultaneously, the weighted averaging strategy effectively utilizes important information from edge points, reducing inaccurate calculations caused by edge point noise. This optimization strategy ensures that the intelligent judgment of conductor sag maintains high accuracy and stability even under complex conditions (such as slight conductor bending or noise interference).

[0139] This invention, through intelligent determination of neighborhood size and optimization of statistical processing of included angle deviation, effectively improves the detection accuracy of conductor sag in images. Automatic adjustment of the neighborhood size based on the distribution density of conductor edge points avoids the tediousness and uncertainty of manual parameter selection. The introduction of truncation of the bending limit angle and a weighted averaging strategy further enhances the robustness of the discrimination algorithm to image noise and anomalies. These technical features significantly improve the accuracy, stability, and adaptability of the intelligent sag discrimination method for transmission line conductors based on UAV inspection images, thereby making the transmission line inspection process more efficient and safer.

[0140] In one possible implementation, the camera on a drone is often affected by lens distortion, which distorts the shape of objects in the image, affecting subsequent image analysis and coordinate transformation. Using known lens distortion parameters (such as radial and tangential distortion), the pixel coordinates of attachment points (such as wire positions) in the image can be corrected. Specifically, this step requires using camera calibration techniques to calculate a mathematical model of lens distortion, thereby correcting the pixel coordinates of the attachment points. Through correction, the obtained attachment point coordinates will be more accurate, reducing the impact of distortion on subsequent analysis.

[0141] Correcting lens distortion can effectively reduce imaging errors and ensure more accurate coordinate calculations in subsequent steps. This is crucial for the intelligent determination of conductor sag, as even small coordinate errors can lead to significant deviations in conductor sag values.

[0142] To transform pixel coordinates in an image to the actual ground coordinate system, the UAV's attitude information, including pitch, roll, and yaw angles, needs to be considered. By establishing a rotation matrix that includes these three attitude angles, the spatial relationship between the UAV's imaging plane and the actual ground coordinate system can be clearly described. The rotation matrix is ​​used to describe the transformation process from the camera coordinate system to the ground coordinate system.

[0143] Collinearity equations are an important tool for calculating the relationship between image points and actual ground points. When converting corrected pixel coordinates to ground coordinates, collinearity equations can be used to map two-dimensional coordinates in the image to a three-dimensional ground coordinate system through geometric relationships. This allows us to obtain the actual spatial coordinates from the image's pixel coordinates, further improving coordinate accuracy.

[0144] Spatial position calculation, by considering the UAV's attitude and image distortion, can accurately transform pixel coordinates to a ground coordinate system, ensuring that the three-dimensional spatial information obtained from UAV inspection images is consistent with the actual physical environment. This better reflects the true condition of the conductor sag and avoids inaccuracies caused by attitude errors.

[0145] To further improve the accuracy of the 3D coordinates of the attachment point, image data acquired from multiple angles by a UAV was used for joint optimization. The bundle adjustment technique was employed to merge image data from different perspectives, minimizing measurement errors and optimizing the 3D coordinates of the attachment point. Bundle adjustment is an optimization algorithm widely used in precision measurement; by minimizing an error function, it fuses data from different perspectives into an optimal solution, thereby obtaining more accurate 3D coordinates.

[0146] Optimizing using multi-view images can effectively eliminate local errors caused by a single viewpoint and improve the accuracy of 3D coordinates. This optimization process can further improve the accuracy of sag detection, especially when the conductor is located at different angles and heights, ensuring the consistency and accuracy of the spatial position obtained from multiple viewpoints.

[0147] By establishing a complete coordinate transformation model and combining camera imaging geometric correction, spatial position calculation, and bundle adjustment optimization of multi-view images, this invention can accurately extract the three-dimensional coordinates of transmission line conductors from UAV inspection images. This process significantly improves the accuracy of conductor sag detection, especially in complex environments, eliminating errors caused by image distortion, attitude changes, and viewing angle deviations, ensuring the stability and reliability of conductor sag detection. This technical feature enables more efficient and accurate transmission line inspections, effectively improving the operational safety and management efficiency of transmission lines.

[0148] In one possible implementation, depending on the type of conductor, the elastic modulus and design tension of that conductor can be obtained from relevant engineering standards or design documents. The elastic modulus is an important physical quantity that measures a material's tensile strength, while the design tension refers to the maximum tension value set during the design process based on the conductor material and operating conditions.

[0149] These parameters directly affect the physical properties of the conductor. Accurately obtaining and using these parameters can provide the necessary basic data for subsequent calculation of the rate of curvature change, enabling the sag detection to match the actual conductor performance, thereby improving the accuracy and reliability of the detection.

[0150] A catenary is a mathematical model describing the suspended shape of a flexible object under the influence of gravity. When a conductor is subjected to tension and its own weight, its shape approximates a catenary; therefore, the rate of change of curvature of the conductor can be derived from the catenary's differential equation. The rate of change of curvature is an important indicator describing the change in sag of a conductor, reflecting the degree of bending of the conductor under different tensions and weights.

[0151] Based on the theory of catenaries, the threshold for the rate of change of curvature can be derived using the following formula:

[0152]

[0153] Where T0 is the horizontal tension, w is the weight per unit length, E is the elastic modulus, and A is the cross-sectional area;

[0154] This theoretical formula allows for the precise calculation of the rate of change of curvature threshold of a conductor, providing a reliable theoretical basis for subsequent sag determination. The formula fully considers the physical properties of the conductor and external environmental factors, making the calculation results more practically meaningful and applicable.

[0155] In practical applications, considering uncertainties in engineering, changes in the external environment, and potential material aging, a certain safety margin is usually preset based on the theoretical threshold. Specifically, the allowable range of curvature change rate can be obtained by multiplying the theoretical threshold by a preset percentage, thereby ensuring that the sag change of the conductor does not exceed the safe range under normal operating conditions.

[0156] Setting a safety margin provides additional tolerance for errors in engineering practice, avoiding misjudgments caused by overly strict thresholds. This not only improves the robustness of the system but also effectively prevents conductor sag exceeding limits due to changes in the external environment (such as wind speed and temperature changes), thereby enhancing the safety and stability of transmission lines.

[0157] By combining the theoretical formula derived from the catenary differential equation with the safety margin setting in practical applications, this invention enables accurate identification of conductor sag changes in transmission lines. This method can calculate a reasonable threshold for the rate of curvature change based on the specific conductor type and physical parameters, while the preset safety margin provides a more stable and reliable basis for identification in practical applications. Through this technical feature, the operating status of transmission lines can be monitored and managed more effectively, potential fault risks can be prevented, and the safety and stability of the power system can be improved.

[0158] In one possible implementation, in practical applications, the sag variation of a conductor is influenced by multiple factors, such as wind speed, temperature, and the conductor's own physical properties. To accurately determine the sag condition based on these variations, the frequency of sag variation under different span lengths needs to be considered. Through statistical analysis, a model can be established to describe the frequency and pattern of sag variation under different span lengths. This statistical model helps determine the frequency of conductor sag variation under different span conditions, thus providing a basis for selecting the smoothing window length.

[0159] Once the span value is determined, the optimal window length coefficient can be found using a statistical model. The span value determines the scale of the region to be analyzed in the image, while the window length coefficient helps select an appropriate smoothing window size. Through this mapping relationship, the window length can be dynamically adjusted in practical applications to adapt to sag variations under different conductor span conditions.

[0160] This step allows for adaptive adjustment of the sliding window length based on different measurement area characteristics (such as span), thereby improving the accuracy and robustness of the smoothing method and avoiding misjudgments caused by unreasonable window settings.

[0161] The core of the sliding window smoothing method is to reduce noise and smooth the data by applying a weighted average to the data within the window. A Gaussian weight distribution is used to ensure that the weights follow a normal distribution within the window, meaning the weights are highest at the center and gradually decrease further away from the center. The use of a Gaussian distribution ensures a natural transition in the smoothing operation, avoiding potentially inconsistent results from sharp weight changes.

[0162] When using a Gaussian distribution, the weight peaks correspond to the center of the window. This means that during smoothing, data at the center has the greatest impact on the final result, while data near the edges contribute less. This weighting method ensures that the most important parts of the image (i.e., the pixel data at the center of the window) are fully considered, reducing interference from unimportant information.

[0163] The amplitude of conductor vibration varies with external environmental factors (such as wind speed and temperature). Therefore, the standard deviation of the smoothing window (a key parameter of the Gaussian distribution) should also be adaptively adjusted according to the amplitude of conductor vibration. When conductor vibration is large, the standard deviation can be increased to expand the range of weight allocation and enhance the smoothing effect on large-scale data; while when vibration is small or stable, the standard deviation should be decreased to reduce the impact on data far from the center and ensure more accurate smoothing.

[0164] By adaptively adjusting the standard deviation, the smoothing window's smoothness can be adjusted according to the actual condition of the conductor, thus maintaining the flexibility and accuracy of data processing under different vibration amplitudes. The Gaussian weight distribution ensures the naturalness of the smoothing operation, avoiding over-smoothing or non-smoothing, further improving the accuracy of sag change detection.

[0165] The sliding window smoothing method dynamically adjusts the window length, weight allocation, and standard deviation to make the detection of sag changes more flexible and accurate. First, based on the mapping relationship between the statistical model and the span, the optimal window length can be adaptively selected to ensure smooth sag changes under different conditions. Second, the application of Gaussian weight distribution ensures a natural transition during the smoothing process, while the adaptively adjusted standard deviation takes into account conductor vibration, further improving the method's robustness. The combination of these features enables this method to accurately capture the changing trend of conductor sag, significantly improving the inspection accuracy of transmission lines and ensuring the safe and stable operation of the power system.

[0166] In one possible implementation, the UAV will collect images along multiple routes during the inspection process, some of which may have overlapping areas. Spatial consistency verification involves detecting the sag values ​​of the same span in these overlapping areas to ensure that the measurements of the same conductor segment are consistent across multiple routes.

[0167] Because drone inspections are affected by various factors such as weather, lighting, and sensor accuracy, there may be some measurement errors. Therefore, if the difference in sag values ​​for the same span in an overlapping area of ​​adjacent flight paths exceeds the preset error tolerance, it is necessary to initiate a recalculation program to re-analyze the data for that area to ensure accurate sag values ​​are obtained.

[0168] Spatial consistency verification effectively avoids erroneous judgments caused by measurement errors or data inconsistencies. By comparing the sag values ​​of the same conductor in overlapping areas, the reliability of the data can be improved, ensuring the consistency and accuracy of inspection results. This is crucial for detecting potential measurement deviations or equipment failures, ensuring that the health status of power lines is accurately reflected.

[0169] To monitor the long-term trend of conductor sag, the current sag value can be compared with historical data, especially data under the same temperature conditions. By verifying the continuity over time, abnormal changes in sag values ​​can be identified, and it can be determined whether there are any special reasons (such as conductor aging or damage) causing the sag abnormality.

[0170] The creep limit of a material refers to the limit of permanent deformation that a conductor can undergo under prolonged stress due to factors such as temperature changes. If the current sag value exceeds the material's creep limit, it indicates that the change may be caused by aging or deformation of the conductor material. In this case, the system will trigger a manual review procedure, requiring professionals to further inspect and confirm the anomaly.

[0171] Time continuity verification effectively monitors the long-term changing trends of conductors, promptly identifying potential risks caused by changes in the external environment or material aging. If the abrupt change exceeds the creep limit, the system will automatically alarm and initiate manual verification, preventing power accidents caused by sudden changes going undetected. Manual verification, as a combination of manual and automated processing, improves system reliability and reduces the risk of false alarms and missed alarms.

[0172] By combining spatial consistency verification and temporal continuity verification, this method not only ensures the consistency of measurement data spatially but also promptly detects abnormal changes in sag by comparing historical data, thereby improving the accuracy and reliability of intelligent judgment. Spatial consistency verification avoids erroneous results due to measurement errors by detecting differences in sag values ​​between adjacent lines; while temporal continuity verification promptly identifies potential conductor damage or aging issues by comparing historical data. The combination of these two methods not only enhances the automation level of inspections but also ensures precise monitoring of the transmission line's operating status, providing crucial protection for the safe and stable operation of the power system.

[0173] This invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of this invention. To provide the public with a thorough understanding of this invention, specific details are described in detail in the following preferred embodiments; however, those skilled in the art will fully understand the invention even without these details. Furthermore, to avoid unnecessary misunderstanding of the essence of this invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0174] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for intelligently identifying conductor sag in power transmission lines based on UAV inspection images, characterized in that, include: S1: Identify the wire region based on UAV visible light images and extract the edges of continuous wires; S2: Locate the insulator string within the conductor area and accurately locate the conductor splicing point through gradient direction aggregation analysis; The gradient direction aggregation analysis in S2 includes: A rectangular search area is defined at the end of the insulator string. This area extends a predetermined length along the direction of conductor extension and its width is an integer multiple of the width of the insulator string. For each pixel within the search region, calculate the consistency of its neighborhood gradient direction: Extract all edge points in the neighborhood centered at the pixel, whose size is related to the diameter of the wire; Calculate the absolute value of the angle deviation between the gradient direction and the vertical direction at each edge point; The arithmetic mean of the absolute values ​​of all included angle deviations is used as the consistency score for that pixel; The pixel with the lowest consistency score is identified as the splicing point, based on the physical principle that the edge gradient direction at the wire splicing point is vertically distributed. S3: Construct a terrain-adaptive dynamic baseline, obtain the three-dimensional geographic coordinates of adjacent tower connection points, and generate a spatial baseline based on the digital elevation model and route alignment; S4: Perform physical constraints to reconstruct the shape of the conductor, connect the broken edges with the physical properties of the catenary as constraints, and project the reconstructed conductor onto the vertical plane of the baseline. S5: Calculate the maximum vertical distance from the projection point to the baseline as the sag value, and combine it with the ambient temperature to dynamically diagnose anomalies.

2. The intelligent method for determining conductor sag of transmission lines based on UAV inspection images according to claim 1, characterized in that, The generation of the dynamic baseline in S3 includes: 3D coordinate calculation of the attachment point: Obtain the latitude, longitude, and altitude of the pole base point recorded by the drone positioning system; The relative elevation of the attachment point is calculated based on the pixel offset between the attachment point and the tower base point in the image, combined with the drone's lens pitch angle, flight altitude and focal length. The pixel coordinates are mapped to the geographic coordinate system through a coordinate transformation model; Spatial baseline construction: Determine the azimuth angle of the route based on the direction of the line connecting the two connection points. Topographic profile lines are generated by interpolation in the digital elevation model along the direction of the connecting lines; Using the two connection points as endpoints, generate a three-dimensional straight line in the plane containing the terrain profile line.

3. The intelligent method for determining conductor sag of transmission lines based on UAV inspection images according to claim 1, characterized in that, The physical constraint wire morphology reconstruction in S4 includes: Rules for connecting fracture edges: For pairs of break points whose distance is less than the maximum suspension span of the conductor, candidate connection paths are generated; The rate of curvature change of the candidate path is calculated, which is obtained by the second-order difference approximation of the path point set; A connection is established when the rate of change of curvature is within the allowable elastic deformation range of the conductor material; Catenary property verification: Fit the catenary equation to the reconstructed continuous traverse point set; Remove outliers that deviate from the fitted curve by more than the diameter of the conductor.

4. The intelligent method for determining conductor sag of transmission lines based on UAV inspection images according to claim 1, characterized in that, The calculation of the sag value in S5 includes: Maximum vertical distance positioning: Establish a local coordinate system for the baseline on the projection plane; Calculate the normal distance from each traverse point to the baseline; The distance sequence is processed using a sliding window smoothing method, with the window length proportional to the conductor span. Temperature compensation mechanism: Obtain the coefficient of thermal expansion of the material from the conductor type database; The sag correction factor is calculated based on the difference between the real-time ambient temperature and the standard temperature. The dynamic threshold is obtained by multiplying the standard sag threshold by the correction factor.

5. The intelligent method for determining conductor sag of transmission lines based on UAV inspection images according to claim 1, characterized in that, In the calculation of the neighborhood gradient direction consistency: The method for determining the neighborhood size is as follows: Randomly sample multiple areas within the conductor region and calculate the average distribution density of points at the conductor edge. The minimum neighborhood size required to cover a single conductor can be determined by inversely calculating the distribution density. Statistical optimization of included angle deviation: The conductor is cut off if the deviation exceeds the bending limit angle. A weighted average strategy is adopted, and the weights are positively correlated with the gradient magnitude of the edge points.

6. The intelligent method for determining conductor sag of transmission lines based on UAV inspection images according to claim 1, characterized in that, The establishment of the coordinate transformation model includes: Camera imaging geometry correction: The pixel coordinates of the mounting point are corrected using the distortion parameters of the drone lens; Spatial location calculation: Establish a rotation matrix that includes the pitch, roll, and yaw attitude angles of the UAV; The corrected pixel coordinates are transformed to the ground coordinate system using collinearity equations. Combined multi-view images to optimize the three-dimensional coordinates of the attachment point using bundle adjustment.

7. The intelligent method for determining conductor sag of transmission lines based on UAV inspection images according to claim 3, characterized in that, The permissible range of the rate of change of curvature is determined by the following methods: Obtain the elastic modulus and design tension corresponding to the conductor type; Derivation of the curvature change rate threshold based on the catenary differential equation: ； in For horizontal tension, Weight per unit length For elastic modulus, The cross-sectional area is given.

8. The intelligent method for determining conductor sag of transmission lines based on UAV inspection images according to claim 4, characterized in that, Parameter settings for the sliding window smoothing method: Mapping relationship between window length and margin: Establish a statistical model for the frequency of sag variation under different gauge lengths; Query the optimal window length coefficient in the statistical model based on the gap value; Smooth weight distribution: A Gaussian weight distribution is used, with the peak weight position corresponding to the center of the window; The standard deviation is adaptively adjusted based on the amplitude of conductor vibration.

9. The intelligent method for determining conductor sag of transmission lines based on UAV inspection images according to claim 1, characterized in that, It also includes a verification mechanism for abnormal diagnosis: Spatial consistency verification: Detect the sag value of the same span in the overlapping area of ​​adjacent routes; A recalculation is initiated when the difference exceeds the measurement error tolerance. Time continuity verification: Compare the current sag value with historical values ​​under the same temperature conditions; If the mutation rate exceeds the material creep limit, a manual review will be triggered.

Images