Unmanned aerial vehicle high-voltage line hooking anti-interference control method based on visual feedback

Abstract

The invention provides an unmanned aerial vehicle high-voltage line hooking anti-interference control method based on visual feedback, and relates to the technical field of unmanned aerial vehicle control. The method comprises the steps of obtaining an environment image obtained by shooting the surrounding environment of the unmanned aerial vehicle; based on a preset morphological image recognition strategy, performing feature extraction on the environment image to obtain a difference image; based on a preset post-processing strategy, performing interference elimination on the differential image to obtain an output image; determining position information of the target high-voltage line according to the output image; according to the position information, a control strategy is predicted through a preset model, and the expected speed of the unmanned aerial vehicle is generated; and controlling the unmanned aerial vehicle to fly towards the target high-voltage line at an expected speed, generating a dynamic control quantity based on a preset linear active disturbance rejection control strategy, and correcting the expected speed through the dynamic control quantity until the unmanned aerial vehicle arrives at the target high-voltage line to complete hooking. Therefore, the problems of high calculation and hardware resource requirements and insufficient control precision of a traditional unmanned aerial vehicle autonomous line hanging inhabitation mode can be solved.

Description

Technical Field

[0001] The present invention relates to the field of unmanned aerial vehicle (UAV) control technology, and in particular to an anti-interference control method for high-voltage line hooking of an UAV based on visual feedback. Background Art

[0002] With the rapid development of science and technology, drone technology has emerged as a key player in power inspection. Its exceptional adaptability allows it to penetrate complex areas difficult for humans to reach, bringing new opportunities for inspection. However, drone inspections also have significant limitations, with insufficient battery life being a key factor restricting their widespread adoption. Typically, a single flight lasts no more than 40 minutes, and the operating radius is typically within 5 kilometers. This means that long-distance and extended inspection missions require frequent round-trip flights, significantly reducing operational efficiency. Consequently, autonomous tethered charging technology for drones has emerged.

[0003] The implementation of autonomous drone charging technology relies on accurate identification of high-voltage power lines and reliable autonomous drone posture control. Regarding the specific implementation methods of this technology, current high-voltage power line detection technologies are primarily divided into traditional computer vision algorithms and deep learning-based algorithms. Deep learning technology excels at identifying high-voltage power lines, but its reliance on large-scale training data and high-performance hardware leads to high training costs, making it less than optimal for resource-constrained drone platforms. Traditional computer vision algorithms are less computationally intensive than the aforementioned deep learning algorithms, but they lack the ability to precisely control the drone and remove the power lines after identifying them.

[0004] In summary, the traditional drone autonomous line perching method has the problems of high computing and hardware resource requirements and insufficient drone control accuracy. Summary of the Invention

[0005] In view of this, the purpose of the embodiments of the present application is to provide an anti-interference control method for high-voltage line hanging of drones based on visual feedback, which can improve the problems of high computing and hardware resource requirements and insufficient drone control accuracy in traditional drone autonomous line hanging and perching methods.

[0006] To achieve the above technical objectives, the technical solutions adopted in this application are as follows:

[0007] The present application provides a method for controlling the anti-interference of high-voltage line connection of a UAV based on visual feedback, the method comprising:

[0008] Acquire an environmental image obtained by photographing the surrounding environment of the drone;

[0009] Based on a preset morphological image recognition strategy, feature extraction is performed on the environmental image to extract high-voltage line features in the environmental image, and a differential image containing target high-voltage line structural features is obtained;

[0010] Based on a preset post-processing strategy, interference elimination is performed on the differential image to eliminate interference line segments in the differential image, thereby obtaining an output image representing the complete structural features of the target high-voltage line;

[0011] determining the position information of the target high-voltage line according to the output image;

[0012] According to the position information, a desired speed of the UAV is generated by a preset model predictive control strategy;

[0013] The UAV is controlled to fly toward the target high-voltage line at the desired speed, and during the flight of the UAV, a dynamic control variable is generated based on a preset linear active disturbance rejection control strategy to correct the desired speed through the dynamic control variable until the UAV reaches the target high-voltage line and completes the high-voltage line attachment.

[0014] In some optional embodiments, between acquiring an environmental image obtained by photographing the surrounding environment of the drone and performing feature extraction on the environmental image based on a preset morphological image recognition strategy to extract high-voltage line features in the environmental image and obtain a differential image containing target high-voltage line structural features, the method further includes:

[0015] Preprocessing the environment image to obtain a preprocessed environment image;

[0016] The method of performing feature extraction on the environment image based on a preset morphological image recognition strategy to extract high-voltage line features in the environment image and obtain a differential image containing target high-voltage line structural features includes:

[0017] Based on the preset morphological image recognition strategy, feature extraction is performed on the preprocessed environmental image to extract the high-voltage line features in the preprocessed environmental image to obtain the differential image.

[0018] In some optional implementations, preprocessing the environment image to obtain a preprocessed environment image includes:

[0019] Performing grayscale conversion on the environment image to convert the environment image into a grayscale image:

[0020] I gray =0.299·R+0.587·G+0.114·B

[0021] Where R, G, and B represent the pixel values of the red, green, and blue channels of the environment image, respectively. gray Represents the pixel value of a grayscale image;

[0022] Perform a Gaussian filter operation on the grayscale image to remove noise in the grayscale image to obtain an intermediate image:

[0023]

[0024] Where I(x,y) represents the pixel value of the pixel (x,y) in the grayscale image, G(c,d) represents the Gaussian function, Where σ represents the standard deviation of the Gaussian kernel, and k is half the size of the Gaussian window;

[0025] Binarize the intermediate image to enhance the contrast of the intermediate image, and obtain a binary image as the preprocessed environment image:

[0026]

[0027] Where, I gaussian (x,y) represents the pixel value of the intermediate image, I binary (x,y) represents the pixel value of the binary image, and Y represents the preset pixel threshold.

[0028] In some optional embodiments, the feature extraction of the environment image based on a preset morphological image recognition strategy to extract high-voltage line features in the environment image and obtain a differential image containing target high-voltage line structural features includes:

[0029] Performing an erosion operation on the environment image to remove the target high-voltage line portion in the environment image to obtain an eroded image:

[0030]

[0031] Where, I eroded Represents the image after corrosion, I represents the environment image, B represents the corrosion structure element, Indicates corrosion operation;

[0032] Performing a dilation operation on the eroded image to restore the background debris portion in the eroded image to obtain a dilated image:

[0033] I dilated =I eroded ⊕B

[0034] Where, I dilated Represents the expanded image, I eroded represents the image after corrosion, and ⊕ represents the expansion operation;

[0035] Performing a difference operation between the environment image and the expanded image to extract the high-voltage line features in the environment image to obtain the difference image:

[0036] I wire =I original -I dilated

[0037] Where, I wire represents the difference image, I original represents the original image, I dilated Represents the dilated image.

[0038] In some optional embodiments, the step of performing interference removal on the differential image based on a preset post-processing strategy to remove interfering line segments in the differential image and obtain an output image representing the complete structural features of the target high-voltage line includes:

[0039] Based on a preset background interference removal strategy, a first line segment filtering operation is performed on the difference image to remove interference line segments with direction offset or insufficient length in the difference image, thereby obtaining a first post-processed image;

[0040] Based on a preset line continuity analysis strategy, a second line segment filtering operation is performed on the first post-processed image to remove discontinuous interfering line segments in the first post-processed image, thereby obtaining a second post-processed image as the output image.

[0041] In some optional embodiments, the first line segment filtering operation is performed on the difference image based on a preset background interference removal strategy to remove interference line segments with direction offset or insufficient length in the difference image to obtain a first post-processed image, including:

[0042] According to the azimuth angle difference between any two line segments in the differential image, target line segments in the differential image are eliminated, wherein the target line segments represent line segments whose azimuth angle difference is greater than a first preset threshold value, and the azimuth angle difference is calculated as follows:

[0043] θ diff =|θ i -θ j |

[0044] Where θ diff represents the direction angle difference, θ i represents the direction angle of the i-th line segment in the differential image, θ j represents the direction angle of the jth line segment in the difference image;

[0045] According to the length of each line segment in the differential image, line segments whose length is less than a second preset threshold are eliminated to obtain the first post-processed image. The length is calculated as follows:

[0046]

[0047] Where L represents the length of the line segment, (x1, y1) and (x2, y2) represent the coordinates of the two endpoints of the line segment.

[0048] In some optional embodiments, the performing of a second line segment filtering operation on the first post-processed image based on a preset line continuity analysis strategy to remove discontinuous interfering line segments in the first post-processed image to obtain a second post-processed image as the output image includes:

[0049] According to the connection distance between any two line segments in the first post-processed image, line segments in the first post-processed image whose connection distance is greater than or equal to a third preset threshold are eliminated to obtain the output image. The connection distance is calculated as follows:

[0050]

[0051] Where D connect Indicates the connection distance, (x 1,j ,y 1,j )、(x 2,i ,y 2,i ) represents the coordinates of the endpoints of the two line segments.

[0052] In some optional embodiments, generating the desired speed of the drone based on the position information using a preset model predictive control strategy includes:

[0053] Based on the position information, obtaining the initial path of the UAV flying toward the target high-voltage line, the current position of the UAV, and the yaw angle;

[0054] Determine error information between the UAV and the target high-voltage line according to the initial path, the current position, and the yaw angle:

[0055] ξ e =R(-ψ)(ξ r -ξ)

[0056] Where, ξ e Indicates error information. represents the initial path, represents the current position and yaw angle, R(-ψ) represents the rotation matrix;

[0057] The objective function is solved using the error information and preset constraints to determine the acceleration of the drone. The objective function is as follows:

[0058]

[0059] Where J represents acceleration, Q, R and P are positive definite weight matrices, and T p represents the prediction time window;

[0060] Based on the acceleration, a desired speed of the drone is determined.

[0061] In some optional embodiments, controlling the UAV to fly toward the target high-voltage power line at the desired speed, and generating a dynamic control variable based on a preset linear active disturbance rejection control strategy during the flight of the UAV to correct the desired speed using the dynamic control variable until the UAV reaches the target high-voltage power line and the high-voltage power line is attached, includes:

[0062] Obtaining an extended state estimate and current speed of the UAV;

[0063] Determine the dynamic control amount according to the desired speed, the extended state estimation value and the current speed:

[0064]

[0065] Where u represents the dynamic control quantity, represents the expected speed, Indicates the current speed, K u , K v is the control gain matrix, K d , K p is the differential gain matrix, x2 represents the extended state estimate;

[0066] The extended state estimation value and the current speed of the UAV are cyclically obtained, the dynamic control amount is determined according to the expected speed, the extended state estimation value and the current speed, and the UAV is controlled to fly toward the target high-voltage line with the dynamic control amount until the UAV reaches the target high-voltage line and completes the high-voltage line attachment.

[0067] The invention adopting the above technical solution has the following advantages:

[0068] In the technical solution provided by the present application, an environmental image is first acquired, and then morphological features are extracted from the environmental image to obtain a differential image. The differential image is then post-processed to obtain an output image, and the position information of the target high-voltage line is determined based on the output image. Then, the desired speed of the drone is generated based on the position information of the target high-voltage line, and in the process of controlling the drone to fly toward the target high-voltage line at the desired speed, the desired speed is corrected by the dynamic control amount until the drone reaches the target high-voltage line and the high-voltage line is attached. In this way, by post-processing the differential image, the interfering line segments in the image are eliminated with a smaller amount of calculation than the traditional deep learning algorithm, thereby improving the recognition accuracy of the target high-voltage line. Then, during the flight of the drone, the desired speed is continuously corrected by the dynamic control amount, thereby improving the flight control accuracy and stability of the drone, and improving the problems of the traditional drone autonomous line-hanging and perching method, which have high computing and hardware resource requirements and insufficient drone control accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] The present application may be further illustrated by the non-limiting embodiments provided in the accompanying drawings. It should be understood that the following drawings illustrate only certain embodiments of the present application and are therefore not to be construed as limiting the scope of the present application. It is understood that a person skilled in the art can derive other relevant drawings from these drawings without inventive effort.

[0070] Figure 1 A flow chart of the anti-interference control method for high-voltage line attachment of a UAV based on visual feedback provided in an embodiment of the present application.

[0071] Figure 2 Schematic diagram of the drone platform provided in an embodiment of the present application.

[0072] Figure 3 Schematic diagram of the drone trajectory tracking experiment results provided in the embodiments of the present application.

[0073] Figure 4 Schematic diagram of the image preprocessing experimental results provided in the embodiments of this application.

[0074] Figure 5 Schematic diagram of the experimental results of image morphological feature extraction provided in the embodiments of this application.

[0075] Figure 6 Schematic diagram of the image post-processing experimental results provided in the embodiments of this application.

[0076] Figure 7 This is a schematic diagram of the target high-voltage line position information output result provided in an embodiment of the present application.

[0077] Figure 8 A real-life picture of the drone high-voltage line attachment mission provided in an embodiment of this application. DETAILED DESCRIPTION

[0078] The present application will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that similar or identical parts in the drawings or descriptions are numbered the same. Implementations not shown or described in the drawings are known to those of ordinary skill in the art. In the description of this application, the terms "first," "second," etc. are used solely to distinguish descriptions and are not to be construed as indicating or implying relative importance.

[0079] Please refer to Figure 1 The present application also provides a method for controlling the anti-interference of high-voltage line connection of a UAV based on visual feedback. The method for controlling the anti-interference of high-voltage line connection of a UAV based on visual feedback may include the following steps:

[0080] Step 110, obtaining an environmental image obtained by photographing the surrounding environment of the drone;

[0081] Step 120: performing feature extraction on the environment image based on a preset morphological image recognition strategy to extract high-voltage line features in the environment image and obtain a differential image containing target high-voltage line structural features;

[0082] Step 130: performing interference elimination on the differential image based on a preset post-processing strategy to eliminate interference line segments in the differential image, thereby obtaining an output image representing the complete structural features of the target high-voltage power line;

[0083] Step 140, determining the location information of the target high-voltage line according to the output image;

[0084] Step 150: generating a desired speed of the UAV based on the position information using a preset model prediction control strategy;

[0085] Step 160: Control the UAV to fly toward the target high-voltage line at the desired speed, and generate a dynamic control variable based on a preset linear anti-disturbance control strategy during the flight of the UAV, so as to correct the desired speed through the dynamic control variable until the UAV reaches the target high-voltage line and completes the high-voltage line attachment.

[0086] In the above-mentioned embodiment, an environmental image is first acquired, and then morphological features are extracted from the environmental image to obtain a differential image. The differential image is then post-processed to obtain an output image, and the position information of the target high-voltage line is determined based on the output image. Then, the desired speed of the drone is generated based on the position information of the target high-voltage line, and in the process of controlling the drone to fly toward the target high-voltage line at the desired speed, the desired speed is corrected by the dynamic control amount until the drone reaches the target high-voltage line and the high-voltage line is attached. In this way, by post-processing the differential image, the interfering line segments in the image are eliminated with a smaller amount of computation than the traditional deep learning algorithm, thereby improving the recognition accuracy of the target high-voltage line. Then, during the flight of the drone, the desired speed is continuously corrected by the dynamic control amount, thereby improving the flight control accuracy and stability of the drone, and improving the problems of the traditional drone autonomous line-hanging and perching method, which have high computing and hardware resource requirements and insufficient drone control accuracy.

[0087] To facilitate the implementation of this technical solution, please refer to Figure 2 The present application also provides a drone platform equipped with development boards for various front-end sensors and control operations. Specifically, the underlying implementation of the drone's intelligent perception system relies heavily on accurate three-dimensional perception. While stereoscopic depth cameras are commonly used in high-voltage power line inspection, this technical solution utilizes a more lightweight, standard USB camera to reduce the load. This camera collects color image data and, combined with back-end algorithms, performs estimations to meet the requirements of power line identification. To ensure real-time image processing, the NVIDIA Jetson NX development board is used. The Jetson NX offers an excellent balance between computing performance and weight, making it suitable for the mission requirements of small drones. However, its lightness makes the NX an ideal choice for this task. This technical solution utilizes an 880-inch drone as the platform. This drone is lightweight, rugged, has a large payload capacity, and offers long endurance. It can carry devices such as USB cameras and the Jetson NX. This platform offers excellent scalability, a stable and reliable flight control module, and precise position and velocity control, providing an ideal testing and application foundation for system development.

[0088] The following is a detailed description of the steps of the anti-interference control method for high-voltage line attachment of UAVs based on visual feedback, as follows:

[0089] In step 110, the environmental image can be acquired during the testing phase by pre-storing an RGB image containing the high-voltage power lines in the background, and then accessing it based on user-preset instructions during the subsequent drone control process. Alternatively, the environmental image can be acquired by using a high-resolution camera (i.e., the aforementioned USB camera) onboard the drone to capture real-time images of the high-voltage power lines and their surroundings, and then transmitting these images to the development board in real time for control operations. The acquisition of the environmental image is not specifically limited herein.

[0090] Between step 110 and step 120, the method may further include:

[0091] Preprocessing the environment image to obtain a preprocessed environment image;

[0092] The feature extraction of the environment image based on the preset morphological image recognition strategy to extract the high-voltage line features in the environment image and obtain a differential image containing the target high-voltage line structural features may include:

[0093] Based on the preset morphological image recognition strategy, feature extraction is performed on the preprocessed environmental image to extract the high-voltage line features in the preprocessed environmental image to obtain the differential image.

[0094] In this embodiment, preprocessing the environment image to obtain a preprocessed environment image may include:

[0095] Performing grayscale conversion on the environment image to convert the environment image into a grayscale image:

[0096] I gray =0.299·R+0.587·G+0.114·B (1)

[0097] Where R, G, and B represent the pixel values of the red, green, and blue channels of the environment image, respectively. gray Represents the pixel value of a grayscale image;

[0098] Perform a Gaussian filter operation on the grayscale image to remove noise in the grayscale image to obtain an intermediate image:

[0099]

[0100] Where I(x,y) represents the pixel value of the pixel (x,y) in the grayscale image, G(c,d) represents the Gaussian function, Where σ represents the standard deviation of the Gaussian kernel, and k is half the size of the Gaussian window;

[0101] Binarize the intermediate image to enhance the contrast of the intermediate image, and obtain a binary image as the preprocessed environment image:

[0102]

[0103] Where, I gaussian (x,y) represents the pixel value of the intermediate image, I binary (x, y) represents the pixel value of the binary image, and T represents the preset pixel threshold.

[0104] In this embodiment, a color image (RGB image) is converted into a grayscale image through grayscale conversion, which reduces computational complexity and reduces background interference caused by color information. Among them, the grayscale image is a single channel, and its pixel value is obtained by weighting the RGB channel, as shown in the above formula (1). This embodiment smoothes the image through Gaussian filtering, thereby reducing noise interference and providing a cleaner input image for subsequent wire recognition. In order to adapt to different environmental lighting conditions, the standard deviation σ of the Gaussian filter can be automatically adjusted according to the brightness of the real-time image during the Gaussian filtering process to adapt to environmental changes, as shown in the above formula (2). This dynamic adjustment method helps to improve the robustness of the present technical solution under uneven lighting conditions. After grayscale and denoising processing, this embodiment also performs binarization processing on the image, dividing the pixels into foreground (wires) and background. Binarization enhances the contrast between the wires and the background, making the wires clearer in complex environments, as shown in the above formula (3). The preset pixel threshold can be flexibly set according to the actual situation. Formula (3) sets the pixel value higher than the threshold T to 255 (white) and the pixel value lower than or equal to T to 0 (black), thereby obtaining a binary image, which is beneficial to subsequent feature extraction and high-voltage line recognition.

[0105] In step 120, based on a preset morphological image recognition strategy, feature extraction is performed on the environment image to extract high-voltage line features in the environment image to obtain a differential image containing target high-voltage line structural features, which may include:

[0106] Performing an erosion operation on the environment image to remove the target high-voltage line portion in the environment image to obtain an eroded image:

[0107]

[0108] Where i eroded represents the image after corrosion, i represents the environment image, B represents the corrosion structure element, Indicates the erosion operation;

[0109] Performing a dilation operation on the eroded image to restore the background debris portion in the eroded image to obtain a dilated image:

[0110] I dilated =I eroded ⊕B (5)

[0111] Where, I dilated represents the expanded image, i eroded Represents the image after corrosion, ⊕ represents the dilation operation;

[0112] Performing a difference operation between the environment image and the expanded image to extract the high-voltage line features in the environment image to obtain the difference image:

[0113] i wire =i original -i dilated (6)

[0114] Where, I wire represents the difference image, I original represents the original image (i.e., the environment image), I dilated Represents the dilated image.

[0115] In this embodiment, a differential image characterizing the structural features of the target high-voltage line is obtained by performing morphological feature extraction on the environmental image. The morphological feature extraction includes completely corroding the high-voltage line, restoring and expanding the debris, and differentially extracting the high-voltage line. First, the high-voltage line is corroded and removed by formula (4) (the structural element in formula (4) can be flexibly set according to user needs, and a larger value can be selected to completely corrode the target high-voltage line), and then the background debris in the image is expanded and restored to its original size by formula (5). Finally, the original image and the expanded image are differentiated by formula (6) to remove the background interference in the image and obtain a differential image. It can be understood that in practical applications, in order to improve the effect of morphological feature extraction, the aforementioned pre-processed environmental image can be used. The image is a denoised binary image. In morphological feature extraction, compared with directly processing the original image, it has a better target high-voltage line feature extraction effect.

[0116] In step 130, based on a preset post-processing strategy, interference removal is performed on the difference image to remove interference line segments in the difference image to obtain an output image representing the complete structural features of the target high-voltage line, which may include:

[0117] Based on a preset background interference removal strategy, a first line segment filtering operation is performed on the difference image to remove interference line segments with direction offset or insufficient length in the difference image, thereby obtaining a first post-processed image;

[0118] Based on a preset line continuity analysis strategy, a second line segment filtering operation is performed on the first post-processed image to remove discontinuous interfering line segments in the first post-processed image, thereby obtaining a second post-processed image as the output image.

[0119] In this embodiment, based on a preset background interference removal strategy, a first line segment filtering operation is performed on the difference image to remove interference line segments with direction offset or insufficient length in the difference image to obtain a first post-processed image, which may include:

[0120] According to the azimuth angle difference between any two line segments in the differential image, target line segments in the differential image are eliminated, wherein the target line segments represent line segments whose azimuth angle difference is greater than a first preset threshold value, and the azimuth angle difference is calculated as follows:

[0121] θ diff =|θ i -θ j | (7)

[0122] Where θ diff represents the direction angle difference, θ i represents the direction angle of the i-th line segment in the differential image, θ j represents the direction angle of the jth line segment in the difference image;

[0123] According to the length of each line segment in the differential image, line segments whose length is less than a second preset threshold are eliminated to obtain the first post-processed image. The length is calculated as follows:

[0124]

[0125] Where L represents the length of the line segment, (x1, y1) and (x2, y2) represent the coordinates of the two endpoints of the line segment.

[0126] Understandably, in order to further reduce background interference, this embodiment filters out line segments that meet the typical characteristics of high-voltage power lines by analyzing the directional consistency and length characteristics of the lines. For different scene backgrounds (such as mountains, forests, buildings, etc.), this embodiment adjusts the thresholds of directional consistency and length filtering to adapt to the interference characteristics of different backgrounds. For example, in forest areas, the filtering of short and irregular objects such as branches is enhanced, while in buildings, the filtering of short lines and edges is enhanced. This can more effectively eliminate typical interference objects in specific environments.

[0127] Specifically, in this embodiment, the first preset threshold and the second preset threshold can be flexibly set according to user needs and can be flexibly calibrated according to preliminary experimental data. In this embodiment, since high-voltage lines are usually presented as parallel or similar-direction line segments in the image, by analyzing the directional consistency of adjacent line segments, non-high-voltage line segments with excessive direction deviation can be effectively excluded. Among them, this embodiment calculates the direction angle difference between two line segments through formula (7). If the direction angle difference is greater than the above-mentioned first preset threshold, it is confirmed that there is a direction deviation in the line segment, and the line segment is eliminated. In addition, since high-voltage lines are usually presented as long lines, this embodiment can calculate the length of the line segment through formula (8) to eliminate shorter line segments that may be debris or noise (i.e., line segments less than the second preset threshold).

[0128] In this embodiment, based on a preset line continuity analysis strategy, a second line segment filtering operation is performed on the first post-processed image to remove discontinuous interfering line segments in the first post-processed image to obtain a second post-processed image as the output image, which may include:

[0129] According to the connection distance between any two line segments in the first post-processed image, line segments in the first post-processed image whose connection distance is greater than or equal to a third preset threshold are eliminated to obtain the output image. The connection distance is calculated as follows:

[0130]

[0131] Where D connect Indicates the connection distance, (x 1,j ,y 1,j )、(x 2,i ,y 2,i ) represents the coordinates of the endpoints of the two line segments.

[0132] Based on the continuity and ductility characteristics of the wire, this embodiment analyzes the line segments that have passed the background removal and retains the continuous lines, thereby obtaining a complete high-voltage wire structure. In order to adapt to different background complexities, the line continuity analysis strategy of this embodiment sets the line segment connection distance threshold D connect This is the third preset threshold, which can be adjusted flexibly based on the complexity of the inspection environment. In areas with less background interference, a smaller connection distance threshold is used; in scenes with more complex backgrounds, the connection distance threshold is appropriately increased to ensure continuity identification of wires.

[0133] It is understandable that in practical applications, in order to determine whether two line segments in an image belong to the same target high-voltage line, it is necessary to calculate the minimum connection distance between them. i and L jThey are composed of several pixel points respectively, record the coordinates of their respective end points, and then calculate the two line segments L by formula (9) i 、L j The connection distance (which can be expressed as the Euclidean distance shown in formula (9)) of all endpoint pairs between the two segments (one endpoint is randomly selected for pairing between the two segments as an endpoint pair) is calculated, and the minimum value of the connection distance of all endpoint pairs is selected as the target connection distance. When the target connection distance is less than the third preset threshold and meets the aforementioned directional consistency analysis, the two segments are considered to constitute a continuous high-voltage line portion. If the target connection distance between the two segments is greater than or equal to the third preset threshold or does not meet the directional consistency analysis, the shorter of the two segments is deleted to remove the pseudo connection, thereby optimizing the target recognition and image post-processing quality. It is understandable that in actual applications, in order to improve the recognition accuracy, after determining that the segment is a continuous high-voltage line portion, the length calculation shown in formula (8) can be performed on the continuous segment again to ensure that its ductility meets the characteristics of the high-voltage line.

[0134] Understandably, to further improve recognition accuracy, a small number of high-voltage line feature templates can be pre-set. By comparing the high-voltage line feature templates with the lines detected in the differential image, the lines with the highest similarity can be selected as target high-voltage lines. The similarity can be the cosine similarity between the high-voltage line feature templates and the lines in the differential image.

[0135] In step 140, after determining the output image, the position information of the target high-voltage line contained in the output image is extracted. The position information may include the position coordinates of the target high-voltage line (that is, the midpoint coordinates of the target high-voltage line), the direction angle (that is, the main extension direction of the target high-voltage line, expressed in angle) and the total length information.

[0136] In step 150, generating the desired speed of the UAV based on the position information by using a preset model predictive control strategy may include:

[0137] Based on the position information, obtaining the initial path of the UAV flying toward the target high-voltage line, the current position of the UAV, and the yaw angle;

[0138] Determine error information between the UAV and the target high-voltage line according to the initial path, the current position, and the yaw angle:

[0139] ξ e =R(-ψ)(ξ r -ξ) (10)

[0140] Where, ξ e Indicates error information. represents the initial path, represents the current position and yaw angle, R(-ψ) represents the rotation matrix, which is used to transform the error from the global coordinate system to the body coordinate system;

[0141] The dynamic model of the error information can be expressed as follows:

[0142]

[0143] Where, f(ξ e ,v B ) represents the kinematic model of error information evolution, d ξ Indicates external interference of the drone;

[0144] The objective function is solved using the error information and preset constraints to determine the acceleration of the drone. The objective function is as follows:

[0145]

[0146] Where J represents acceleration, Q, R and P are positive definite weight matrices, and T p represents the prediction time window;

[0147] Based on the acceleration, a desired speed of the drone is determined.

[0148] In this embodiment, the preset constraint condition may include a state constraint, that is, ξ e ∈X e , the error must be within the error range X set flexibly by the user e Inside, as well as the input constraints, that is, v B ∈U, current speed v B The physical limit U must be met (i.e., less than the maximum speed that the drone can reach).

[0149] In this embodiment, by solving the objective function, the minimized trajectory error and the control energy consumption J are obtained and regarded as acceleration. Then, the desired speed is obtained by using the acceleration and the preset time frequency.

[0150] It is understandable that in practical applications, this embodiment uses MPC (Model Predictive Control) as the outer-loop controller to generate the desired speed of the drone flight based on the obtained initial path (which can be generated by a console connected to the drone or other processors carried by the drone. This embodiment mainly acts on error control during the flight of the drone to achieve precise control of the drone. The generation process of the initial path is not described in detail here). It serves as the input of the subsequent inner-loop controller.

[0151] In step 160, controlling the UAV to fly toward the target high-voltage power line at the desired speed, and generating a dynamic control variable based on a preset linear active disturbance rejection control strategy during the flight of the UAV to correct the desired speed using the dynamic control variable until the UAV reaches the target high-voltage power line and the high-voltage power line is attached, may include:

[0152] Obtaining an extended state estimate and current speed of the UAV;

[0153] Determine the dynamic control amount according to the desired speed, the extended state estimation value and the current speed:

[0154]

[0155] Where u represents the dynamic control quantity, represents the expected speed, Indicates the current speed, K u , K v is the control gain matrix, K d , K p is the differential gain matrix, x2 represents the extended state estimate;

[0156] The extended state estimation value and the current speed of the UAV are cyclically obtained, the dynamic control amount is determined according to the expected speed, the extended state estimation value and the current speed, and the UAV is controlled to fly toward the target high-voltage line with the dynamic control amount until the UAV reaches the target high-voltage line and completes the high-voltage line attachment.

[0157] In this embodiment, the reasoning process of the above dynamic control amount is as follows:

[0158] First, build the UAV dynamics model:

[0159]

[0160] Where, v B represents the velocity vector representing the current velocity of the drone, d v Indicates external interference (this is the same as the external interference d ξ The meaning is the same, but they are in different calculation processes. Different formula parameters are used here to distinguish them). u is the control input, which is the actual flight speed of the drone (it can also represent the flight status of the drone, such as speed, direction, etc.).

[0161] Then, an extended state observer (ESO) is constructed to transform the external disturbance d v Modeled as the extended state x2, the above dynamic model can be expressed as:

[0162]

[0163] Where x1 = v B represents the velocity vector, x2 is the estimated value of the extended state, and g(t) is the disturbance change rate;

[0164] The ESO model design is expressed as:

[0165]

[0166] Where x1 and x2 represent the estimated values of the state and extended state, and β1 and β2 represent the feedback gains of the ESO.

[0167] In order to simplify parameter design and computational complexity, the nonlinear function in ESO is replaced with a linear form, and the above ESO model changes to:

[0168]

[0169] Where z1, z2, and z3 represent the state estimates, which represent the position, velocity, and total disturbance of the drone (including external disturbances and modeling errors), respectively. v1 represents the desired position input. α1, α2, and α3 are gain parameters that determine the ESO convergence speed. The corresponding drone control rate is:

[0170]

[0171] Where e1 represents the position error, which is the difference between the desired position input v1 and the position state estimate z1; e2 represents the speed error, which is the difference between the desired speed input v2 and the speed state estimate z2; b represents the control gain parameter; and u0 represents the virtual control input calculated by the PD (proportional derivative) controller.

[0172] Based on the above reasoning process, the final dynamic control rate of the UAV, that is, the dynamic control amount, is expressed as:

[0173]

[0174] Where, Represents the velocity error of the drone.

[0175] It is understandable that in practical applications, this embodiment uses LADRC (Linear Active Disturbances Rejection Controller) as the inner-loop controller to perform the calculation of the above-mentioned dynamic control quantity, and realizes high-precision tracking of the initial path by constructing an ESO observer to compensate for external interference in real time.

[0176] This technical solution uses a dual closed-loop control algorithm combining model predictive control (MPC) and a linear active disturbance rejection controller (LADRC) to enable precise drone flight to a target point and maintain stability in a disturbed environment. The algorithm uses the outer MPC loop to generate the desired trajectory (i.e., the initial path and desired velocity), while the inner LADRC loop accurately tracks the desired trajectory, effectively compensating for drone system uncertainties and external disturbances.

[0177] In order to intuitively illustrate the effectiveness of this technical solution, this application conducts the following experimental verification in combination with the above technical solution:

[0178] Reference Figure 3 , this application is approved by Figure 3 The circular trajectory and the ∞-shaped trajectory shown in the figure are tested, where UAVTrajectory represents the UAV trajectory, CircleTrajectory represents the circular trajectory, and ReferenceTrajectory represents the reference trajectory. It can be seen that the technical solution proposed in this application has a better performance in reference trajectory tracking.

[0179] Regarding the specific implementation process of each step of this technical solution, the experimental data are shown as follows (it is worth noting that the original image is a color image, but it is represented here in grayscale for the convenience of format requirements).

[0180] Image preprocessing is as follows:

[0181] The experiment aims to verify the effectiveness of the proposed technical solution in complex environments. Specific objectives include evaluating the solution's wire detection accuracy, anti-interference capabilities, real-time performance, and coordinate output accuracy in a scene with tall towers and debris as a background.

[0182] The experimental environment was set up in a background containing high towers and other debris, with an image resolution of 640x480 to simulate the complex scene of actual power inspection. The entire experiment was run on an embedded platform with medium computing power (NVIDIA Jetson series), mainly used to evaluate the performance of the algorithm under resource-constrained conditions. Each step / module was tested separately to evaluate its effectiveness in actual application. The experimental results are shown in the figure below. Figure 4 As shown in the figure, from left to right, the original image, the image after grayscale processing, and the image after Gaussian filtering and binarization processing are shown.

[0183] The morphological feature extraction is as follows:

[0184] After image preprocessing is completed, Figure 5As shown, a morphological erosion operation is first used to remove thin wire structures from the image while preserving coarser background objects (such as pylons and trees). The erosion operation uses a larger structuring element (such as a 3x3 or 5x5 square kernel) to completely eliminate thin wires while preserving coarse structures like pylons. The erosion results show that the thin wire structures of the wires are almost completely removed, providing a clean background image for the subsequent dilation and differencing operations.

[0185] Next, a dilation operation is performed on the eroded image to restore background objects (such as trees and power lines) to a near-original form. Through multiple iterations of the dilation operation, the outlines of the background objects are restored more completely, while the power lines remain missing. The dilation result shows a complete background image without power lines, laying a good foundation for the subsequent subtraction process.

[0186] Finally, by performing a differential operation between the original image and the expanded background image, the slender structural features of the wires were successfully extracted. The resulting image retains only the wire's linear structure, while other background artifacts are almost completely removed. This step significantly enhances the saliency of the wires, making them the only recognizable structure in the image. This effectively improves recognition accuracy and provides reliable data support for further wire identification and location.

[0187] The results of background interference removal and wire continuity analysis are as follows:

[0188] After extracting the slender structure of the wires, this technical solution further removes background interference by performing background interference removal and wire continuity analysis on the difference image. First, the background interference removal module uses methods such as directional consistency, color features, and length filtering to remove non-wire debris.

[0189] Directional consistency analysis: High-voltage power lines often appear as parallel or similarly oriented line segments in images. By analyzing the directional consistency of adjacent line segments, the system can remove those with significant directional deviations, retaining only those that match the directional characteristics of the power lines. The first output image shows the results of this directional consistency analysis, which only retains line segments with similar orientations, significantly reducing background noise.

[0190] Length filtering: High-voltage power lines typically appear as long lines in images, while background debris (such as tree branches and short lines) is relatively short. By setting a minimum length threshold, shorter noise segments are removed, ensuring that the remaining segments are more consistent with the characteristics of high-voltage power lines. The results of directional consistency and length filtering show a purer image of the power line structure, effectively eliminating the interference of short debris.

[0191] After eliminating background interference through directional consistency and length filtering, this technical solution further performs wire continuity analysis to ensure the integrity of the identified wire structure. Based on the continuity and ductility characteristics of high-voltage wires, the wire continuity analysis module connects and filters the remaining wire segments to ensure that the output results constitute a complete high-voltage wire.

[0192] Segment Connectivity Analysis: This technology checks whether the distance and direction between the start and end points of adjacent segments are consistent. Segments that meet the connectivity criteria are considered continuous sections of the same wire and, when connected, form a complete wire structure.

[0193] Length and ductility calculation: The total length of the connected segments is calculated to ensure that the ductility of the segments meets the characteristics of high-voltage power lines. The final result image (the second output image) shows the complete power line structure, in which the continuity and length characteristics of the wires are well restored, and the interference of isolated non-wire lines is eliminated. The final output image is as follows Figure 6 shown.

[0194] Regarding the output results of the final contour and high-voltage line position information:

[0195] Reference Figure 7 Finally, the position, direction, angle and length information of the high-voltage power lines are output for use by the UAV inspection system. The output data format is:

[0196] Position coordinates: The coordinates of the start and end points of the identified wires, which clarify the spatial position of the wires in the image;

[0197] Direction angle: The main extension direction of the wire, expressed as an angle, to facilitate the UAV navigation system to adjust the inspection route;

[0198] Length information: The total length of the wire, providing data support for further distance calculation and inspection task planning.

[0199] About drone control to grab high-voltage wires:

[0200] After obtaining the position coordinates, direction angle and length information of the high-voltage power line, these key data are transmitted to the flight control system of the UAV in real time. Given that the UAV is now directly below or near the high-voltage line, the system only needs to carry out fine movement and precise attitude adjustment in the current area, without the need for long-distance flight approach operations. To achieve this goal, this application uses a dual closed-loop control algorithm that combines an outer loop kinematic controller based on model predictive control (MPC) with an inner loop dynamic controller based on linear active disturbance rejection control (LADRC) to ensure that even in a complex interference environment, the UAV can accurately and stably perform grasping operations on the high-voltage line.

[0201] Among them, regarding the outer loop MPC planning posture and fine-tuning:

[0202] The drone's precise current position, real-time attitude, and detailed spatial coordinates of the power line are rapidly fed into the outer-loop MPC. Based on a kinematic error model and taking into account state and input constraints, the MPC calculates the most appropriate fine-tuning trajectory in real time at exceptionally high computational speed, accurately generating the desired velocity. Fine-tuning operations primarily involve appropriate lateral displacement to ensure the drone and power line are in the optimal horizontal relative position; appropriate altitude adjustment to maintain the drone and power line in the same operational plane; and precise yaw angle changes to ensure directional alignment. These fine-tuning operations ensure that the drone and power line maintain a stable and appropriate distance, either at the same level or at the optimal operational angle. The drone avoids extensive flight maneuvers and instead utilizes the MPC's rolling optimization mechanism to minimize motion errors, ensuring that the drone and payload remain within the effective operational range of the power line.

[0203] Among them, regarding the inner loop LADRC dynamic control:

[0204] The inner loop LADRC plays a vital role in the entire process of fine movement and attitude correction of the drone. Specifically, the inner loop LADRC can accurately estimate external interference factors and the uncertainty factors of the system itself in real time by constructing an extended state observer (ESO). Subsequently, combined with the PD (proportional differential) control strategy, these interferences are dynamically compensated. The precise control signal output by the inner loop LADRC can ensure that the speed and attitude of the drone closely follow the desired trajectory given by the outer loop MPC, effectively avoiding flight deviation or attitude shaking caused by external interference. During the entire operation process, the drone's visual system continues to closely monitor the actual position of the high-voltage line. Once the position of the high-voltage line is found to change, the relevant information will be promptly fed back to the outer loop MPC to make online corrections to the desired speed, further ensuring the accuracy and stability of the drone's flight.

[0205] Reference Figure 8 ,The actual effect of this experiment shows that the anti-interference control method for ,UAV high-voltage wire hanging based on visual feedback proposed in ,this technical solution can realize precise control of the UAV and ,complete the high-voltage wire hanging task in practical applications.

[0206] In summary, the experimental process proposed in this application successfully removes background interference and highlights the slender structure of power lines through image preprocessing operations such as morphological erosion, dilation, and difference. Subsequently, directional consistency analysis, length filtering, and continuity analysis are used to accurately remove non-wire debris and restore the complete outline of the high-voltage power lines. This processing flow has excellent real-time performance on the NVIDIA Jetson platform, and the generated high-voltage power line coordinates, direction angles, and length information provide reliable support for efficient drone inspections.

[0207] In terms of drone control, a dual-loop control algorithm is implemented using an outer-loop kinematic controller based on model predictive control (MPC) and an inner-loop dynamic controller based on linear active disturbance rejection control (LADRC). The outer-loop MPC plans the optimal flight trajectory in real time, consistent with state and input constraints. The inner-loop LADRC uses an extended state observer (ESO) to estimate and compensate for external disturbances and model uncertainties, ensuring that the drone executes the hooking and catching action strictly according to the planned trajectory. Experiments have demonstrated that the drone can stably perform lateral movement, attitude correction, and grasping operations beneath power lines, with a high grasping success rate and strong robustness to interference, providing comprehensive data and control support for power inspection tasks.

[0208] Through the description of the above implementation methods, those skilled in the art can clearly understand that the present application can be implemented through hardware or by means of software plus a necessary general hardware platform. Based on this understanding, the technical solution of the present application can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash drive, a mobile hard disk, etc.), including a number of instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each implementation scenario of the present application.

[0209] In summary, the embodiment of the present application provides an anti-interference control method for high-voltage line attachment of a drone based on visual feedback. In this technical solution, an environmental image is first acquired, and then morphological features are extracted from the environmental image to obtain a differential image. The differential image is then post-processed to obtain an output image, and the position information of the target high-voltage line is determined based on the output image. Then, the desired speed of the drone is generated based on the position information of the target high-voltage line, and in the process of controlling the drone to fly toward the target high-voltage line at the desired speed, the desired speed is corrected by the dynamic control amount until the drone reaches the target high-voltage line and the high-voltage line attachment is completed. In this way, by post-processing the differential image, the interference line segments in the image are eliminated with a smaller amount of computation than the traditional deep learning algorithm, thereby improving the recognition accuracy of the target high-voltage line. Then, during the flight of the drone, the desired speed is continuously corrected by the dynamic control amount, thereby improving the flight control accuracy and stability of the drone, and improving the problems of the traditional drone autonomous line-hanging and perching method, which have high computing and hardware resource requirements and insufficient drone control accuracy.

[0210] In the embodiments provided in the present application, it should be understood that the disclosed method can also be implemented in other ways. The method embodiments described above are merely schematic. For example, the flow charts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the methods and computer program products according to the multiple embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, a program segment or a part of code, and a part of the module, program segment or code includes one or more executable instructions for implementing the specified logical function. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented by a dedicated hardware-based system that performs the specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions. In addition, the functional modules in the various embodiments of the present application can be integrated together to form an independent part, or each module can exist alone, or two or more modules can be integrated to form an independent part.

[0211] The above description is merely an embodiment of the present application and is not intended to limit the scope of protection of the present application. For those skilled in the art, various modifications and variations of the present application are possible. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A method for controlling the high-voltage line connection of an unmanned aerial vehicle (UAV) based on visual feedback, characterized in that: The method comprises: Acquire an environmental image obtained by photographing the surrounding environment of the drone; Based on a preset morphological image recognition strategy, feature extraction is performed on the environmental image to extract high-voltage line features in the environmental image, and a differential image containing target high-voltage line structural features is obtained; Based on a preset post-processing strategy, interference elimination is performed on the differential image to eliminate interference line segments in the differential image, thereby obtaining an output image representing the complete structural features of the target high-voltage line; determining the position information of the target high-voltage line according to the output image; According to the position information, a desired speed of the UAV is generated by a preset model predictive control strategy; The UAV is controlled to fly toward the target high-voltage line at the desired speed, and during the flight of the UAV, a dynamic control variable is generated based on a preset linear active disturbance rejection control strategy to correct the desired speed through the dynamic control variable until the UAV reaches the target high-voltage line and completes the high-voltage line attachment.

2. The method according to claim 1, characterized in that Between the environmental image obtained by photographing the surrounding environment of the drone and the feature extraction of the environmental image based on the preset morphological image recognition strategy to extract the high-voltage line features in the environmental image and obtain a differential image containing the target high-voltage line structural features, the method further includes: Preprocessing the environment image to obtain a preprocessed environment image; The method of performing feature extraction on the environment image based on a preset morphological image recognition strategy to extract high-voltage line features in the environment image and obtain a differential image containing target high-voltage line structural features includes: Based on the preset morphological image recognition strategy, feature extraction is performed on the preprocessed environmental image to extract the high-voltage line features in the preprocessed environmental image to obtain the differential image.

3. The method according to claim 2, characterized in that The preprocessing of the environment image to obtain a preprocessed environment image includes: Performing grayscale conversion on the environment image to convert the environment image into a grayscale image: I gray =0.299·R+0.587·G+0.114·B Where R, G, and B represent the pixel values of the red, green, and blue channels of the environment image, respectively. gray Represents the pixel value of a grayscale image; Perform a Gaussian filter operation on the grayscale image to remove noise in the grayscale image to obtain an intermediate image: Where I(x,y) represents the pixel value of the pixel (x,y) in the grayscale image, G(c,d) represents the Gaussian function, Where σ represents the standard deviation of the Gaussian kernel, and k is half the size of the Gaussian window; Binarize the intermediate image to enhance the contrast of the intermediate image, and obtain a binary image as the preprocessed environment image: Where, I gaussian (x,y) represents the pixel value of the intermediate image, I binary (x, y) represents the pixel value of the binary image, and T represents the preset pixel threshold.

4. The method according to claim 1, wherein The method of performing feature extraction on the environment image based on a preset morphological image recognition strategy to extract high-voltage line features in the environment image and obtain a differential image containing target high-voltage line structural features includes: Performing an erosion operation on the environment image to remove the target high-voltage line portion in the environment image to obtain an eroded image: Where, I eroded Represents the image after corrosion, I represents the environment image, B represents the corrosion structure element, Indicates corrosion operation; Performing a dilation operation on the eroded image to restore the background debris portion in the eroded image to obtain a dilated image: Where, I dilated Represents the expanded image, I eroded represents the image after corrosion, represents the expansion operation; Performing a difference operation between the environment image and the expanded image to extract the high-voltage line features in the environment image to obtain the difference image: I wire =I original -I dilated Where, I wire represents the difference image, I original represents the original image, I dilated Represents the image after dilation.

5. The method according to claim 1, wherein The method of performing interference elimination on the differential image based on a preset post-processing strategy to eliminate interference line segments in the differential image and obtain an output image representing the complete structural features of the target high-voltage line includes: Based on a preset background interference removal strategy, a first line segment filtering operation is performed on the difference image to remove interference line segments with direction offset or insufficient length in the difference image, thereby obtaining a first post-processed image; Based on a preset line continuity analysis strategy, a second line segment filtering operation is performed on the first post-processed image to remove discontinuous interfering line segments in the first post-processed image, thereby obtaining a second post-processed image as the output image.

6. The method according to claim 5, characterized in that The first line segment filtering operation is performed on the difference image based on a preset background interference elimination strategy to eliminate interference line segments with direction offset or insufficient length in the difference image to obtain a first post-processed image, including: According to the azimuth angle difference between any two line segments in the differential image, target line segments in the differential image are eliminated, wherein the target line segments represent line segments whose azimuth angle difference is greater than a first preset threshold value, and the azimuth angle difference is calculated as follows: i diff =|θ i -θ j | Where θ diff represents the direction angle difference, θ i represents the direction angle of the i-th line segment in the differential image, θ j represents the direction angle of the jth line segment in the difference image; According to the length of each line segment in the differential image, line segments whose length is less than a second preset threshold are eliminated to obtain the first post-processed image. The length is calculated as follows: Where L represents the length of the line segment, (x1, y1) and (x2, y2) represent the coordinates of the two endpoints of the line segment.

7. The method according to claim 5, characterized in that The step of performing a second line segment filtering operation on the first post-processed image based on a preset line continuity analysis strategy to remove discontinuous interfering line segments in the first post-processed image to obtain a second post-processed image as the output image includes: According to the connection distance between any two line segments in the first post-processed image, line segments in the first post-processed image whose connection distance is greater than or equal to a third preset threshold are eliminated to obtain the output image. The connection distance is calculated as follows: Where D connect Indicates the connection distance, (x 1,j ,y 1,j )、(x 2,i ,y 2,i ) represents the coordinates of the endpoints of the two line segments.

8. The method according to claim 1, characterized in that Generating the desired speed of the UAV according to the position information by using a preset model prediction control strategy includes: Based on the position information, obtaining the initial path of the UAV flying toward the target high-voltage line, the current position of the UAV, and the yaw angle; Determine error information between the UAV and the target high-voltage line according to the initial path, the current position, and the yaw angle: x e =R(-ψ)(ξ r -ξ) Where, ξ e Indicates error information. represents the initial path, represents the current position and yaw angle, R(-ψ) represents the rotation matrix; The objective function is solved using the error information and preset constraints to determine the acceleration of the drone. The objective function is as follows: Where J represents acceleration, Q, R and P are positive definite weight matrices, and T p represents the prediction time window; Based on the acceleration, a desired speed of the drone is determined.

9. The method according to claim 1, characterized in that The controlling the UAV to fly toward the target high-voltage line at the desired speed, and generating a dynamic control variable based on a preset linear active disturbance rejection control strategy during the flight of the UAV, so as to correct the desired speed by the dynamic control variable until the UAV reaches the target high-voltage line and the high-voltage line is attached, including: Obtaining an extended state estimate and current speed of the UAV; Determine the dynamic control amount according to the desired speed, the extended state estimation value and the current speed: Where u represents the dynamic control quantity, represents the expected speed, Indicates the current speed, K u , K v is the control gain matrix, K d , K p is the differential gain matrix, x2 represents the extended state estimate; The extended state estimation value and the current speed of the UAV are cyclically obtained, the dynamic control amount is determined according to the expected speed, the extended state estimation value and the current speed, and the UAV is controlled to fly toward the target high-voltage line with the dynamic control amount until the UAV reaches the target high-voltage line and completes the high-voltage line attachment.

Images