Power transmission line defect intelligent identification method and device based on inspection unmanned aerial vehicle

Abstract

This invention discloses a method and device for intelligent identification of power transmission line defects based on inspection drones, belonging to the field of intelligent power inspection technology. The method includes: acquiring inspection images of equipment components on the power transmission line using an inspection drone as inspection equipment images; cleaning the inspection equipment images to obtain preprocessed sample images; and performing intelligent identification on the preprocessed sample images to obtain a defect sample dataset. This invention obtains basic information such as the location of the captured images, the voltage level of the power transmission line, and the type of equipment through location information, providing a basis for image classification; and improving the detection rate and accuracy of defects in equipment at different voltage levels.

Description

Technical Field

[0001] This invention belongs to the field of intelligent power inspection technology, and particularly relates to a method and device for intelligent identification of transmission line defects based on inspection drones. Background Technology

[0002] With the continuous and in-depth advancement of intelligent power system construction, transmission line inspection, as a core link in power grid operation and maintenance, directly affects the overall safe and stable operation of the power grid in terms of its automation and intelligence level. Unmanned aerial vehicle (UAV) inspection has significant advantages such as high operational efficiency, wide coverage, and strong environmental adaptability, and has become the mainstream technical means for transmission line inspection. Autonomous inspection modes have been widely promoted and applied in transmission lines of all voltage levels.

[0003] In autonomous drone inspection operations, intelligent defect recognition of inspection images is a key technological support, and the efficiency and accuracy of recognition directly determine the quality and effectiveness of the inspection work. Currently, the industry has gradually applied intelligent recognition algorithms for defect detection, but existing technologies and models still have significant shortcomings: traditional recognition models mostly adopt a single architecture and have not been specifically adapted and optimized for different voltage levels and different equipment components; inspection images are easily affected by factors such as lighting, weather conditions, and shooting angles, and low-quality images such as blurry, backlit, and overexposed images seriously interfere with the recognition effect; some defect samples, such as hardware wear and insulator damage, are scarce, resulting in insufficient model learning and low recognition accuracy; at the same time, the existing sample library lacks localization adaptation and has not built targeted datasets based on regional equipment operating conditions and geographical climate characteristics, making it difficult to meet the actual operation and maintenance needs of power grids in various regions.

[0004] Disadvantages of existing technology:

[0005] (1) The intelligent identification model for defects in inspection drones adopts a single detection architecture and does not perform pre-screening processing on inspection images. Low-quality images such as blurry, backlit, and overexposed images directly enter the identification process, resulting in a high rate of missed and false reports in defect identification. At the same time, it does not match exclusive model packages according to voltage level and equipment components, resulting in poor identification adaptability in different scenarios and low overall identification accuracy. Before the breakthrough, the average detection rate of similar models in the industry was only 61.7% and the accuracy rate was only 17%, which is far from meeting the requirements for practical engineering.

[0006] (2) The existing identification model has obvious shortcomings in the construction of the sample library. The sample collection lacks comprehensiveness and specificity. The localized sample library is not built in combination with the regional equipment environment. Furthermore, there is a scarcity of field samples for some high-frequency defects (such as wear of hardware and damage to the sheds of composite insulators). The model is unable to fully learn the defect features. At the same time, the sample annotation lacks unified standards and the annotation accuracy is insufficient, which further affects the model training effect.

[0007] (3) The model training is disconnected from engineering applications. Localized training has not been achieved, the model iteration efficiency is low and data security is difficult to guarantee. At the same time, there is a lack of a closed-loop training mechanism of "training and testing at the same time and manual review". The model is deployed directly after training is completed, and the training strategy is not adjusted in time based on the actual test feedback from the production site. As a result, the model performs poorly in actual inspection scenarios and is far behind the level of manual recognition.

[0008] (4) The defect identification work lacks a systematic technical support and long-term management mechanism. No professional multi-team collaborative task force has been established. The connection between algorithm development, sample labeling, power transmission business, and digital support is not smooth. At the same time, there is a lack of a long-term mechanism for sample feedback and dynamic model optimization. After the model is deployed, it cannot be continuously iterated according to equipment updates and scene changes, and the identification ability is difficult to maintain stability in the long term.

[0009] Therefore, there is an urgent need for an intelligent method and device for identifying transmission line defects to solve the problem that the scarcity of the sample database makes it difficult to meet the actual operation and maintenance needs of power grids in various regions. Summary of the Invention

[0010] In view of the shortcomings of the above or existing technologies, this invention proposes a method and device for intelligent identification of transmission line defects based on inspection drones. This method effectively reduces the workload of operation and maintenance personnel in image interpretation and defect verification, lowers the cost of power grid operation and maintenance, improves the timeliness of defect detection and handling efficiency, and provides a solid technical guarantee for the safe and reliable operation of the power grid. It has important engineering practical value and good prospects for promotion and application.

[0011] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0012] In a first aspect, the present invention provides a method for intelligent identification of power transmission line defects based on inspection drones, including:

[0013] Inspection drones are used to obtain inspection images of equipment components on power transmission lines as inspection equipment images;

[0014] Preprocessed sample images are obtained by cleaning the images of the inspection equipment.

[0015] Intelligent identification is performed on preprocessed sample images to obtain a defect sample dataset.

[0016] A preferred embodiment of the present invention is that the inspection images of equipment components on the power transmission line acquired by the inspection drone are used as inspection equipment images; specifically including:

[0017] The inspection drone acquires equipment image information at inspection points along the power transmission line;

[0018] The equipment image information includes equipment images and location information, wherein the location information includes voltage level, transmission line name, equipment name, and shooting location;

[0019] Use the device image information as the device image name.

[0020] A preferred embodiment of the present invention is that the step of cleaning the inspection equipment image to obtain a preprocessed sample image specifically includes:

[0021] Image cleaning algorithms, including image blurring, backlight detection, and overexposure detection, are used to clean images acquired by inspection drones. After removing blurred, overexposed, and backlit images, preprocessed sample images are obtained.

[0022] A preferred embodiment of the present invention is that the step of intelligently identifying preprocessed sample images to obtain a defect sample dataset includes:

[0023] Images are differentiated and categorized using location information to form a device image dataset;

[0024] Extract defect technical features from the equipment image dataset, and generate a defect sample dataset based on the defect technical features.

[0025] Furthermore, the step of extracting defect technical features from the device image dataset and generating a defect sample dataset based on these features specifically includes:

[0026] The equipment image datasets are divided into wear-related equipment image datasets and other equipment image datasets.

[0027] Technical features of wear images are extracted from wear-related equipment image datasets, and wear image defect samples are generated from these technical features.

[0028] Extract technical features of other types of equipment images from other types of equipment image data and generate other types of defect samples.

[0029] Furthermore, the process of dividing the equipment image dataset into wear-related equipment image datasets and other types of equipment image datasets specifically includes:

[0030] The wear-related equipment image dataset is a collection of wear images of connecting hardware, composite insulator skirts, U-shaped hanging rings, right-angle hanging plates, and connecting plates on transmission lines.

[0031] The other types of equipment image dataset is a collection of defect images of insulators with pollution and icing, missing clamps, abnormal spacers, missing bolts, and rusted guy wires on transmission lines.

[0032] Furthermore, wear image technical features are extracted from the wear-related equipment image dataset, and these features are used to generate wear image defect samples; specifically including:

[0033] Obtain the component bounding box coordinates of the wear image;

[0034] Separate the defective component layer from the component frame coordinate information;

[0035] Replace the defect-free component layer with the merged layer to form a merged defect-free image;

[0036] The defective component layer is replaced and merged with the defect-free layer to form a defect sample;

[0037] Defect samples are stored in the defect database.

[0038] Secondly, this invention proposes an intelligent identification device for power transmission line defects based on an inspection drone, comprising:

[0039] The image acquisition unit acquires inspection images of equipment components on the power transmission line using the inspection drone as inspection equipment images;

[0040] The image cleaning unit cleans the images from the inspection equipment to obtain pre-processed sample images;

[0041] The image recognition unit intelligently identifies preprocessed sample images to obtain a defect sample dataset.

[0042] Thirdly, the present invention proposes a computer device, comprising: a memory and a processor, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the processor causes the processor to perform the steps of the intelligent identification method for power line defects based on inspection drones.

[0043] Fourthly, the present invention proposes a computer-readable storage medium storing a computer program thereon, characterized in that: when the computer program is executed by a processor, it implements the steps of the intelligent identification method for power line defects based on inspection drones.

[0044] The beneficial effects of this invention are as follows:

[0045] 1. This invention obtains basic information such as the location of captured images, the voltage level of transmission lines, and the type of equipment through point location information, providing a basis for image classification; and improving the detection rate and accuracy of defects in equipment of different voltage levels.

[0046] 2. This invention removes low-quality images such as blurry, backlit, and overexposed images through image cleaning to ensure the quality of the input image. Then, the image is named according to the location information, associated with the voltage level and equipment component information, and the defect type is accurately matched, enabling targeted defect identification.

[0047] 3. This invention synthesizes scarce wear-related samples, effectively alleviating the problem of insufficient wear-related samples and improving the reliability of defect identification. Attached Figure Description

[0048] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 The present invention provides a structural diagram of a method for intelligent identification of power transmission line defects based on inspection drones;

[0050] Figure 2 A flowchart of a specific embodiment of the present invention;

[0051] Figure 3 A flowchart illustrating an embodiment of the image cleaning method provided by the present invention;

[0052] Figure 4 The intelligent defect identification flowchart provided by this invention;

[0053] Figure 5 A schematic diagram of a work task sheet provided by the present invention;

[0054] Figure 6 The present invention provides a structural diagram of a power transmission line defect intelligent identification device based on an inspection drone. Detailed Implementation

[0055] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0056] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0057] It should be noted that, unless otherwise defined, the technical or scientific terms used in the embodiments of this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this disclosure pertains. The terms "comprising" or "including," and similar terms as used in this disclosure, mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, but do not exclude other elements or objects. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0058] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that mutually excludes other embodiments. It should be noted that the embodiments of the present invention can be applied to any applicable scenario.

[0059] See Figure 1 This invention provides a method for intelligent identification of power transmission line defects based on inspection drones, including:

[0060] Step 101: Obtain inspection images of equipment components on the power transmission line using an inspection drone as inspection equipment images;

[0061] Step 102: Clean the images of the inspection equipment to obtain preprocessed sample images;

[0062] Step 103: Perform intelligent recognition on the preprocessed sample images to obtain a defect sample dataset.

[0063] See Figure 2 This invention utilizes inspection drones to inspect and photograph equipment on power transmission lines, acquiring inspection images. By cleaning and intelligently identifying these images, a defect sample dataset is obtained. This invention comprehensively improves the automation and accuracy of defect identification, effectively reduces the workload of maintenance personnel in image interpretation and defect verification, lowers power grid maintenance costs, and enhances the timeliness and efficiency of defect detection and handling. It provides a solid technical guarantee for the safe and reliable operation of the power grid, possessing significant engineering practical value and promising prospects for widespread application.

[0064] In this embodiment, inspection images of equipment components on the power transmission line are obtained by an inspection drone and used as inspection equipment images; specifically, this includes:

[0065] Inspection drones acquire equipment image information at inspection points along power transmission lines;

[0066] The equipment image information includes equipment images and location information. The location information refers to the inspection points of the inspection drone, which includes the voltage level, transmission line name, equipment name, and shooting angle information. The images taken by the inspection drone at the inspection points are named using the location information, that is, the equipment image information is used as the name of the captured images.

[0067] This workflow takes images from inspection drones as input. When the drones capture and transmit the images, they are named in a standard format: "Voltage Level-Line Name-Tower Number-Shooting Location," such as "750kV-Tuha Line-Tower No. 25-Left Phase Crossarm Hanging Point." First, the location information distinguishes the type of defect of interest in the image: images without defects of interest are directly returned as defect-free; other images are judged by a defect detection model. If a defect is found, defect information including the defect name and image coordinates (defectname; x1, y1, x2, y2) is summarized; otherwise, it is returned as defect-free. Defects of interest include wear-related images (U-shaped hanging ring wear, right-angle hanging plate wear, connecting plate wear). The original defect detection process is abandoned, and a component detection model and a classification model are trained for binary classification of defect presence and absence. First, all components are located and the target component area is cut out. Then, the image classification model determines whether the component has a defect. Finally, all defect information, including category and coordinates, is summarized and sent to the business platform, achieving accurate identification and reporting of defects in power transmission equipment.

[0068] In step 102, the images of the inspection equipment are cleaned to obtain preprocessed sample images, specifically including:

[0069] Image cleaning algorithms, including image blurring, backlight detection, and overexposure detection, are used to clean images acquired by inspection drones. After removing blurred, overexposed, and backlit images, preprocessed sample images are obtained.

[0070] This invention removes low-quality images, such as blurry, backlit, and overexposed images, through image cleaning to ensure the quality of the input image. Then, the image is named according to the location information, associated with the voltage level and equipment component information, and the defect type is accurately matched, enabling targeted defect identification.

[0071] See Figure 3 This invention cleans images from inspection equipment to obtain preprocessed sample images; specifically as follows:

[0072] Fuzz detection:

[0073] Laplacian variance is used to process edge sharpness, Tenengrad gradient is used to process gradient strength, Canny edge density is used to process the number of edges, and grayscale texture entropy is used to obtain texture complexity.

[0074] The criteria for fuzzy judgment are:

[0075] Calculate the area of ​​the entire image and the central region separately. The central region is taken as 60% or 80% of the area. The Laplacian variance of the center is less than 20% of the entire image, the Tenengrad gradient of the center is less than 30% of the entire image, and the density of the center edge is less than 30% of the entire image.

[0076] If the central region satisfies any of the above criteria, it is determined to be fuzzy.

[0077] Overexposure detection:

[0078] HSV spatial overexposure monitoring is adopted, and the overexposure ratio of the entire RO area and the overexposure ratio of the core area are used for detection;

[0079] The criteria for determining overexplosion are:

[0080] Iterate through all pixels in the image with a brightness ≥ 245 and a saturation ≤ 10.

[0081] Calculate the percentage of the above pixels in the entire image;

[0082] If the entire image occupies more than 90%, calculate the central region of the image again (the center area is taken as 60% or 80%).

[0083] When the central area also exceeds 90%, it is considered overexposed.

[0084] Backlight detection:

[0085] Grayscale image brightness and dark area ratio, average brightness of the entire image;

[0086] Criteria for judging backlighting:

[0087] The proportion of pixels with a grayscale value ≤25 exceeds 60%, and the average brightness of the entire image is <70;

[0088] If both of the above conditions are met, it is determined to be backlighting.

[0089] Images identified as blurry, overexposed, or backlit will be deleted.

[0090] This invention integrates image blurring algorithm, backlight detection algorithm, overexposure detection algorithm and low quality detection algorithm to judge the quality of the original image from multiple dimensions. It is used to automatically remove low quality invalid images such as blurred, backlit, and overexposed images, and provide clear and usable image data for subsequent sample processing and defect identification tasks.

[0091] See Figure 4 In step 103, intelligent recognition is performed on the preprocessed sample images to obtain a defect sample dataset, including:

[0092] Images are differentiated and categorized using location information to form a device image dataset;

[0093] Extract defect technical features from the equipment image dataset, and generate a defect sample dataset based on the defect technical features.

[0094] In this embodiment of the invention, images are classified by location information, i.e., image naming. Components of the same equipment at the same voltage level are grouped together. The captured images are learned and filtered to distinguish between normal and defective images. Then, feature extraction is performed on defective images to form defect samples. This invention obtains basic information about the location of captured images, transmission line voltage level, and equipment type through location information, providing a basis for image classification and improving the detection rate and accuracy of defects in equipment at different voltage levels.

[0095] For high-frequency defects at three voltage levels (750 kV, 220 kV, and 110 kV), sample collection and labeling were carried out, and dedicated intelligent recognition models (including detection model YOLOv8-det, classification model YOLOv8-cls, and attitude estimation model YOLOv8-pose) were trained for each voltage level. A differentiated recognition system was constructed to effectively improve the detection rate and accuracy of equipment defects at different voltage levels.

[0096] In this embodiment, defect technical features are extracted from the equipment image dataset, and a defect sample dataset is generated based on these defect technical features; specifically, this includes:

[0097] The equipment image datasets are divided into wear-related equipment image datasets and other equipment image datasets.

[0098] Technical features of wear images are extracted from wear-related equipment image datasets, and wear image defect samples are generated from these technical features.

[0099] Extract technical features of other types of equipment images from other types of equipment image data and generate other types of defect samples.

[0100] Because the identification of wear-related equipment components on transmission lines is difficult and the number of samples obtained is small, they are considered scarce samples. Therefore, the defect types are divided into wear-related and other types. For wear-related defects, more wear-related defect samples are generated by extracting and synthesizing wear-related defect features to supplement the lack of wear-related defect samples and improve the reliability of identification.

[0101] In this embodiment of the invention, the equipment image dataset is divided into wear-related equipment image datasets and other types of equipment image datasets; specifically including:

[0102] The wear-related equipment image dataset is a collection of wear images of connecting hardware, composite insulator skirts, U-shaped hanging rings, right-angle hanging plates, and connecting plates on power transmission lines.

[0103] Other equipment image datasets are collections of defect images of insulators with pollution and icing, missing clamps, abnormal spacers, missing bolts, and rusted guy wires on power transmission lines.

[0104] See Figure 5This involves extracting wear image technical features from a dataset of wear-related equipment images and generating wear image defect samples from these features; specifically, this includes:

[0105] Obtain the component bounding box coordinates of the wear image;

[0106] Separate the defective component layer from the component frame coordinate information;

[0107] Replace the defect-free component layer with the merged layer to form a merged defect-free image;

[0108] The defective component layer is replaced and merged with the defect-free layer to form a defect sample;

[0109] Defect samples are stored in the defect database.

[0110] To address defects such as hardware wear, which are characterized by scarce samples and susceptibility to interference from lighting and shadows, an image segmentation (layered) model (Qwen-Image-Layered) was used to separate defective images into defective component layers and defective background layers, and defect-free images into defect-free component layers and defect-free background layers. The defective component layers were then merged with the defect-free background layers to synthesize new defect sample images. A total of 12,800 defect samples were synthesized, significantly alleviating the problem of insufficient samples, reducing false alarms caused by background interference, and improving the defect detection rate by an average of 12%.

[0111] This invention synthesizes scarce wear-related samples, effectively alleviating the problem of insufficient wear-related samples and improving the reliability of defect identification.

[0112] Example 2

[0113] See Figure 6 This invention proposes an intelligent identification device for power transmission line defects based on inspection drones, comprising:

[0114] The image acquisition unit 201 acquires inspection images of equipment components on the power transmission line using an inspection drone as inspection equipment images;

[0115] Image cleaning unit 202 cleans the images of the inspection equipment to obtain preprocessed sample images;

[0116] The image recognition unit 203 performs intelligent recognition on the preprocessed sample images to obtain a defect sample dataset.

[0117] The system embodiments described above are merely illustrative. The image acquisition unit, image cleaning unit, and image recognition unit may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Furthermore, in the accompanying drawings of the system embodiments provided by this invention, the connection relationships between modules indicate that they have communication connections, which can be specifically implemented as one or more communication buses or signal lines. Those skilled in the art can understand and implement this without any creative effort.

[0118] It should be noted that the system described in Embodiment 2 corresponds to the method described in Embodiment 1. Therefore, the parts of the system described in Embodiment 2 that are not detailed can be obtained by referring to the content of the method described in Embodiment 1, and will not be repeated here.

[0119] Example 3

[0120] The present invention proposes a computer device, comprising: a memory and a processor, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the processor causes the processor to perform the steps of the intelligent identification method for power line defects based on inspection drones.

[0121] Of course, the computer device may also have wired or wireless network interfaces, keyboards, and input / output interfaces for input and output. The computer device may also include other components for implementing device functions, which will not be elaborated here.

[0122] For example, the computer program may be divided into one or more units, which are stored in the memory and executed by the processor to perform the present invention. The one or more units may be a series of computer program instruction segments capable of performing a specific function, which describe the execution process of the computer program in the computer device.

[0123] The processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor. The processor is the control center of the computer device, connecting all parts of the computer device via various interfaces and lines.

[0124] The memory can be used to store the computer programs and / or units. The processor implements various functions of the computer device by running or executing the computer programs and / or units stored in the memory and by calling data stored in the memory. Furthermore, the memory may include high-speed random access memory, and may also include non-volatile memory, such as hard disks, RAM, plug-in hard disks, smart media cards (SMC), secure digital cards (SD), flash cards, at least one disk storage device, flash memory device, or other volatile solid-state storage devices.

[0125] Example 4

[0126] This invention proposes a computer-readable storage medium storing a computer program thereon, characterized in that: when the computer program is executed by a processor, it implements the steps of the intelligent identification method for power line defects based on inspection drones.

[0127] Specifically, the computer-readable storage medium may include: any entity or device capable of carrying the computer program code, recording media, USB flash drive, portable hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0128] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of the invention to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A method for intelligent identification of power transmission line defects based on inspection drones, characterized in that, include: Inspection drones are used to obtain inspection images of equipment components on power transmission lines as inspection equipment images; Preprocessed sample images are obtained by cleaning the images of the inspection equipment. Intelligent identification is performed on preprocessed sample images to obtain a defect sample dataset.

2. The intelligent identification method for power transmission line defects based on inspection drones according to claim 1, characterized in that, Inspection images of equipment components on power transmission lines are obtained using inspection drones; specifically, this includes: The inspection drone acquires equipment image information at inspection points along the power transmission line; The equipment image information includes equipment images and location information, wherein the location information includes voltage level, transmission line name, equipment name, and shooting location; Use the device image information as the device image name.

3. The intelligent identification method for power transmission line defects based on inspection drones according to claim 1, characterized in that, The process of cleaning the images from the inspection equipment to obtain preprocessed sample images specifically includes: Image cleaning algorithms, including image blurring, backlight detection, and overexposure detection, are used to clean images acquired by inspection drones. After removing blurred, overexposed, and backlit images, preprocessed sample images are obtained.

4. The intelligent identification method for power transmission line defects based on inspection drones according to claim 1, characterized in that, The process of intelligently identifying preprocessed sample images to obtain a defect sample dataset includes: Images are differentiated and categorized using location information to form a device image dataset; Extract defect technical features from the equipment image dataset, and generate a defect sample dataset based on the defect technical features.

5. The intelligent identification method for power transmission line defects based on inspection drones according to claim 4, characterized in that, The defect technical features are extracted from the equipment image dataset, and a defect sample dataset is generated based on the defect technical features. Specifically, it includes: The equipment image datasets are divided into wear-related equipment image datasets and other equipment image datasets. Technical features of wear images are extracted from wear-related equipment image datasets, and wear image defect samples are generated from these technical features. Extract technical features of other types of equipment images from other types of equipment image data and generate other types of defect samples.

6. The intelligent identification method for power transmission line defects based on inspection drones according to claim 5, characterized in that, The equipment image datasets are divided into wear-related equipment image datasets and other equipment image datasets; specifically including: The wear-related equipment image dataset is a collection of wear images of connecting hardware, composite insulator skirts, U-shaped hanging rings, right-angle hanging plates, and connecting plates on transmission lines. The other types of equipment image dataset is a collection of defect images of insulators with pollution and icing, missing clamps, abnormal spacers, missing bolts, and rusted guy wires on transmission lines.

7. The intelligent identification method for power transmission line defects based on inspection drones according to claim 5, characterized in that, Technical features of wear images are extracted from a dataset of images of wear-related equipment, and these features are used to generate wear image defect samples; specifically, this includes: Obtain the component bounding box coordinates of the wear image; Separate the defective component layer from the component frame coordinate information; Replace the defect-free component layer with the merged layer to form a merged defect-free image; The defective component layer is replaced and merged with the defect-free layer to form a defect sample; Defect samples are stored in the defect database.

8. A power transmission line defect intelligent identification device based on inspection drones, characterized in that, The method for intelligent identification of transmission line defects based on inspection drones as described in any one of claims 1-7 includes: The image acquisition unit acquires inspection images of equipment components on the power transmission line using the inspection drone as inspection equipment images; The image cleaning unit cleans the images from the inspection equipment to obtain pre-processed sample images; The image recognition unit intelligently identifies preprocessed sample images to obtain a defect sample dataset.

9. A computer device, comprising: The power transmission line equipment defect detection system according to claim 5; Alternatively, a memory and a processor, wherein the memory stores computer-readable instructions that, when executed by the processor, cause the processor to perform the steps of the intelligent identification method for power line defects based on an inspection drone according to any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the intelligent identification method for power line defects based on inspection drones as described in any one of claims 1-7.

Images