Power line intelligent deicing system installed by using unmanned aerial vehicle

Abstract

The invention relates to the technical field of power line deicing, in particular to an intelligent power line deicing system installed by using an unmanned aerial vehicle. The system comprises an image acquisition module, a feature extraction module, a data analysis module, a deicing control module and a data storage module. According to the invention, image sampling is carried out on an ice layer attached to the surface of a power line through the image acquisition module, and corresponding feature extraction is carried out on image information by using an edge detection unit, a texture detection unit and a color analysis unit in the feature extraction module; and through information fusion of the data analysis module and construction of a corresponding ice layer thickness prediction model, a corresponding ultrasonic deicing frequency range is selected according to the ice layer thickness, and efficient removal of the ice layer on the surface of the power line is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of power line deicing, and in particular to an intelligent power line deicing system installed by using an unmanned aerial vehicle. Background Art

[0002] Deicing of power lines is an important measure to ensure the stable operation of the power system, especially in winter under low temperature weather conditions. Ice covering the lines may cause excessive stretching and breakage of the conductors, or even collapse of the towers, thus causing large-scale power outages. At present, the common methods of deicing power lines are: mechanical deicing, AC short-circuit current ice melting, DC short-circuit ice melting, natural ice removal and ultrasonic deicing. Among them, the ultrasonic deicing method can not only effectively remove the ice layer attached to the surface of the power line, but also minimize the loss of the power line.

[0003] At present, when using ultrasound to remove ice attached to the surface of power lines, the frequency range of ultrasound needs to be guaranteed to be within 20Khz to 100Khz. Too high a frequency will cause damage to the power lines, and different frequencies of ultrasound are used for ice layers of different thicknesses. Due to the complex environment in which the power lines are located, the thickness of the ice attached to the surface of the power lines is also different, and the frequencies to be used are different. If the ice layer is thin and high-frequency ultrasound is used, it is easy to cause a waste of resources. If low-frequency ultrasound is used to de-ice the thick ice layer, the work efficiency is low. Therefore, in order to achieve better de-icing effects, we propose an intelligent de-icing system for power lines installed using drones. Summary of the invention

[0004] The purpose of the present invention is to provide an intelligent deicing system for power lines installed by using a drone to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above-mentioned object, the present invention provides an intelligent deicing system for power lines installed by using a drone, comprising an image acquisition module, a feature extraction module, a data analysis module, a deicing control module and a data storage module;

[0006] The image acquisition module uses a high-definition camera carried by a drone to take long-range and close-up photos of the circuit line. The long-range photo captures a wide range of the power line, identifies and marks nodes where the ice layer attached to the surface of the power line is too thick, and takes a high-definition close-up photo of the ice layer attached to the surface of the power line to obtain image information of the ice layer attached to the surface of the power line, and transmits the image information and the marked node information to the feature extraction module;

[0007] The feature extraction module receives the image information transmitted by the image acquisition module, extracts the edge features, texture features and color features of the ice layer attached to the power line in the image, transmits the feature information to the data analysis module, and the feature extraction module transmits the received marked node information to the deicing control module;

[0008] The data analysis module receives the feature information transmitted by the feature extraction module, performs a comprehensive analysis on the edge features, texture features and color features of each section of the power line image, and compares the features with the original data stored in the data storage module to obtain the thickness information of the ice layer attached to the surface of each section of the power line, and calculates the optimal frequency range for ultrasonic deicing of the section of the power line, and transmits the frequency range information to the deicing control module;

[0009] The deicing control module receives the marked node information transmitted by the feature extraction module, controls the drone to perform hot air deicing on the node, adjusts the frequency range of the ultrasonic wave according to the frequency range information transmitted by the data analysis module, and performs ultrasonic deicing on each section of the power line. After the deicing is completed, the deicing result of the power line is photographed by the high-definition camera carried by the drone, and the deicing result of the power line section is transmitted to the data storage module;

[0010] The data storage module is used to store the original comparison data on the one hand, and to receive the feature information corresponding to each section of the power line on the other hand, to make the feature information extracted by the feature extraction module correspond to the deicing method adopted by the deicing control module, to establish a new data comparison group, and to memorize and store it, to expand the original database, so that the frequency range of the ultrasonic wave can be quickly adjusted when the same image feature information is encountered next time.

[0011] As a further improvement of the technical solution, the feature extraction module includes an edge detection unit, a texture detection unit and a color analysis unit. The edge detection unit performs edge feature detection on the image of the ice layer attached to each section of the power line. Two 3×3 convolution kernels are used to calculate the approximate gradient values ​​of the image in the horizontal and vertical directions, respectively. The two convolution kernels are convolved with the original image to obtain a gradient image in the horizontal direction and a gradient image in the vertical direction. The edge information in the horizontal direction is obtained according to the gradient image in the horizontal direction, and the grayscale change of the pixel in the horizontal direction is detected. Similarly, the edge information in the vertical direction is obtained, and the edge feature points of the ice layer image are extracted.

[0012] The texture detection unit performs texture feature detection on the image of the ice layer attached to each section of the power line, describes the texture feature of the image by counting the frequency of occurrence of the grayscale combination of two pixels at a certain distance and direction in the image, and then extracts the corresponding texture feature information;

[0013] The color analysis unit performs ice layer color analysis on the image of ice layer attached to each section of power line, counts each pixel in the image, records the frequency of occurrence of its color value in a specific color space, uses the number of pixels in each color interval obtained by counting as the ordinate of the histogram, and the color interval as the abscissa, to construct a color histogram and obtain corresponding ice layer color data.

[0014] As a further improvement of the technical solution, when detecting the ice layer attached to the power line, the edge detection unit first uses a traditional edge detection algorithm for preliminary detection, and then uses a deep learning method for further optimization and refinement.

[0015] As a further improvement of the technical solution, when detecting the ice layer attached to the power line, the texture detection unit uses a gray level co-occurrence matrix to detect the texture of the ice layer, and multiple texture feature parameters can be extracted from the gray level co-occurrence matrix.

[0016] As a further improvement of the present technical solution, the color analysis unit performs preprocessing of illumination normalization and color correction on the image to reduce the impact of illumination changes and color changes on the color histogram and obtain more accurate color data.

[0017] As a further improvement of the present technical solution, the data analysis module fuses the edge feature information, texture feature information and color feature information of the image transmitted by the feature extraction module, connects each feature vector to form a comprehensive feature vector, and then constructs an ice thickness prediction model, and selects the corresponding ultrasonic de-icing frequency range according to the ice thickness.

[0018] As a further improvement of the present technical solution, the controlled deicing module includes an ultrasonic deicing unit and a hot air deicing unit. The ultrasonic deicing unit controls the vibration frequency of the ultrasonic wave to process the ice layer attached to the power line. The hot air deicing unit heats the air or steam to increase its temperature to a level sufficient to melt the ice layer, and then blows the hot air or steam toward the power line to melt the ice layer and fall off.

[0019] As a further improvement of the technical solution, the hot air deicing unit performs hot air deicing on nodes where the ice layer attached to the power line is too thick, and transmits the position information of the node to the data storage module for memory storage.

[0020] As a further improvement of the present technical solution, the data storage module includes a feature recognition unit, a memory storage unit and a raw data unit. The feature recognition unit recognizes the feature information in the feature extraction module. If the feature information corresponds to the feature information in the memory storage unit, the de-icing module is directly controlled to perform de-icing. The memory storage unit memorizes and stores the de-icing method and effect of the de-icing module to expand the database. The raw data unit records the original comparison data.

[0021] As a further improvement of the present technical solution, the image acquisition module includes an environment detection unit and a parameter adjustment unit. The environment detection unit detects environmental information near the power line and adjusts the parameters of the high-definition camera according to the environment. The parameter adjustment unit adjusts the shooting parameters of the high-definition camera to ensure a clear image of the attached ice layer of the power line.

[0022] Compared with the prior art, the present invention has the following beneficial effects:

[0023] 1. The intelligent deicing system for power lines installed by drones samples the ice layer attached to the surface of the power lines through the image acquisition module, and transmits the image sampling information to the feature extraction module. The edge detection unit, texture detection unit and color analysis unit in the feature extraction module extract the corresponding features of the image information, and compare them with the original data of the data storage module to preliminarily obtain the ice thickness of the corresponding image. The data analysis module then fuses the edge feature information, texture feature information and color feature information of the image, connects each feature vector to form a comprehensive feature vector, and then constructs an ice thickness prediction model. According to the ice thickness, the corresponding frequency range of ultrasonic deicing is selected to achieve efficient removal of ice on the surface of the power lines.

[0024] 2. Use the hot air deicing unit to perform hot air deicing on the nodes where the ice layer attached to the power line is too thick, and deicing is performed on the corresponding nodes of the power line, cooperating with the ultrasonic deicing unit to ensure that the ice layer attached to the surface of the power line is completely removed;

[0025] 3. The de-icing control module transmits the corresponding adjustment information of de-icing and the corresponding de-icing results to the memory storage unit of the data storage module, expanding the database. It is convenient for the next time when the characteristic information corresponds to the characteristic information in the memory storage unit, the de-icing module can be directly controlled to perform de-icing, which greatly saves data processing time and improves de-icing efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is a schematic diagram of the overall process of the present invention;

[0027] Figure 2 It is a specific flow chart of the present invention;

[0028] Figure 3 It is a schematic diagram of the flow of controlling the deicing module of the present invention;

[0029] Figure 4 It is a schematic diagram of the flow of the data storage module of the present invention.

[0030] The meaning of each number in the figure is:

[0031] 100, image acquisition module; 110, environment detection unit; 120, parameter adjustment unit; 200, feature extraction module; 210, edge detection unit; 220, texture detection unit; 230, color analysis unit; 300, data analysis module; 400, de-icing control module; 410, ultrasonic de-icing unit; 420, hot air de-icing unit; 500, data storage module; 510, feature recognition unit; 520, memory storage unit; 530, raw data unit. DETAILED DESCRIPTION

[0032] The following will be combined with the accompanying drawings in the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0033] Deicing of power lines is an important measure to ensure the stable operation of the power system. Especially in winter, when the temperature is low, ice on the lines may cause excessive stretching, breakage, or even collapse of the power towers, thus causing large-scale power outages. At present, the common methods of deicing power lines are: mechanical deicing, AC short-circuit current deicing, DC short-circuit deicing, natural deicing, and ultrasonic deicing. Among them, ultrasonic deicing can not only effectively remove the ice layer attached to the surface of the power line, but also minimize the loss of the power line. Figure 1 As shown, the present invention provides an intelligent deicing system for power lines installed by using a drone, including an image acquisition module 100, a feature extraction module 200, a data analysis module 300, a deicing control module 400 and a data storage module 500;

[0034] The image acquisition module 100 uses a drone carrying a high-definition camera to take long-range and close-up photos of the circuit line. The long-range photo captures a wide range of the power line, identifies and marks nodes with excessively thick ice layers on the surface of the power line, and takes high-definition close-up photos of the ice layers on the surface of the power line to obtain image information of the ice layers on the surface of the power line, and transmits the image information and the marked node information to the feature extraction module 200.

[0035] The feature extraction module 200 receives the image information transmitted by the image acquisition module 100, extracts the edge features, texture features and color features of the ice layer attached to the power line in the image, and transmits the feature information to the data analysis module 300. The feature extraction module 200 transmits the received marked node information to the deicing control module 400;

[0036] The feature information transmitted by the feature extraction module 200 is received through the data analysis module 300, and the edge features, texture features and color features of each section of the power line image are comprehensively analyzed, and compared with the original data stored in the data storage module 500, so as to obtain the thickness information of the ice layer attached to the surface of each section of the power line, and calculate the optimal frequency range of ultrasonic deicing for the section of the power line, and at the same time, transmit the frequency range information to the deicing control module 400;

[0037] By controlling the deicing module 400 to receive the marked node information transmitted by the feature extraction module 200, the drone is controlled to perform hot air deicing on the node, and the frequency range of the ultrasonic wave is adjusted according to the frequency range information transmitted by the data analysis module 300, and ultrasonic deicing is performed on each section of the power line. After the deicing is completed, the deicing result of the power line is photographed by the high-definition camera carried by the drone, and the deicing result of the power line section is transmitted to the data storage module 500;

[0038] The data storage module 500 is used to store the original comparison data on the one hand, and receives the characteristic information corresponding to each section of the power line on the other hand, and makes the characteristic information extracted by the characteristic extraction module 200 correspond to the deicing method adopted by the deicing control module 400, so as to establish a new data comparison group, store it in memory, expand the original database, and be able to quickly adjust the frequency range of the ultrasonic wave when encountering the same image characteristic information next time.

[0039] like Figure 2 As shown, the feature extraction module 200 includes an edge detection unit 210, a texture detection unit 220 and a color analysis unit 230. The edge detection unit 210 performs edge feature detection on the image of the ice layer attached to each section of the power line. Two 3×3 convolution kernels are used to calculate the approximate gradient values ​​of the image in the horizontal and vertical directions respectively. The two convolution kernels are convolved with the original image to obtain a gradient image in the horizontal direction and a gradient image in the vertical direction. The edge information in the horizontal direction is obtained according to the gradient image in the horizontal direction, and the grayscale change of the pixel in the horizontal direction is detected. Similarly, the edge information in the vertical direction is obtained, and then the edge feature points of the ice layer image are extracted;

[0040] The texture detection unit 220 performs texture feature detection on the image of the ice layer attached to each section of the power line, and describes the texture feature of the image by counting the frequency of the grayscale combination of two pixels at a certain distance and direction in the image, thereby extracting the corresponding texture feature information;

[0041] The color analysis unit 230 performs ice layer color analysis on the image of ice layer attached to each section of power line. By counting each pixel point in the image, the frequency of occurrence of its color value in a specific color space is recorded, and the number of pixels in each color interval obtained by counting is used as the ordinate of the histogram, and the color interval is used as the abscissa, a color histogram is constructed, and the corresponding ice layer color data is obtained.

[0042] In order to improve the accuracy of edge features extracted by the edge detection unit 210 for the ice layer image, when detecting the ice layer attached to the power line, the edge detection unit 210 first uses a traditional edge detection algorithm for preliminary detection, and then uses a deep learning method for further optimization and refinement; that is, firstly uses the Sobel operator to perform preliminary detection on the ice layer, and then uses a convolutional neural network to obtain specific edge feature information;

[0043] Among them, the Sobel operator is a commonly used edge detection algorithm, which detects edges by calculating the horizontal and vertical gradients of each pixel in the image. For ice layer images, convolution operations can be performed in the horizontal and vertical directions respectively to obtain a gradient image, and then the edge is determined according to a certain threshold;

[0044] For each pixel in the image, perform element-level multiplication and summation operations on the surrounding 3×3 pixel area and the corresponding convolution kernel. For example, for horizontal calculations, multiply the pixel and its adjacent columns of pixels on the left and right by the corresponding elements of the horizontal convolution kernel and then add them to obtain the horizontal gradient value of the pixel. The same is true for the vertical direction. After the horizontal and vertical gradient values ​​of each pixel in the image are obtained through the above calculations, the gradient amplitude of the pixel can be calculated according to the following formula:

[0045] Among them, G is the edge strength, G X is the horizontal gradient value, G Y is the vertical gradient value.

[0046] Among them, the convolutional neural network first prepares a large number of annotated ice layer image data sets, where the annotations include the edge information of the ice layer, and then trains the convolutional neural network model to learn the characteristics and edge patterns of the ice layer images. In the testing phase, the ice layer image to be detected is input into the trained model, and the edge strength G obtained by the edge detection algorithm is input into the convolutional neural network model to obtain the edge detection result.

[0047] In order to better extract the texture features of the ice layer image, the texture detection unit 220 detects the ice layer texture using a gray level co-occurrence matrix when detecting the ice layer attached to the power line, and can extract multiple texture feature parameters from the gray level co-occurrence matrix;

[0048] Among them, the gray-level co-occurrence matrix describes the texture characteristics of the image by counting the frequency of gray-level combinations of two pixels at a certain distance and direction in the image. For ice layer images, different distances and directions can be selected to calculate the gray-level co-occurrence matrix. Multiple texture feature parameters can be extracted from the gray-level co-occurrence matrix, such as contrast, energy, correlation, entropy, etc. These parameters can reflect the texture roughness, uniformity, directionality and other characteristics of the ice layer;

[0049] For a grayscale image, first determine a distance d and a direction w, then count the number of occurrences of all grayscale combinations of two pixels in the image that satisfy this distance and direction, and form these numbers into a matrix, which is the grayscale co-occurrence matrix.

[0050] First, establish the value of distance d and direction w. The distance d can be selected according to the size of the image and the characteristics of the texture. It is generally a distance of 1 to several pixels. Secondly, for a given distance and direction, traverse each pixel in the image, find another pixel that satisfies the distance and direction relationship with the current pixel, record the gray value combination of these two pixels, and count the number of times this combination appears in the image. The number of times all possible gray value combinations appear is combined into a matrix. This matrix is ​​the grayscale co-occurrence matrix. The number of rows and columns of the matrix is ​​equal to the grayscale level of the image, and the corresponding texture feature information is obtained.

[0051] In order to better extract the color features of the ice layer image, the color analysis unit performs pre-processing of illumination normalization and color correction on the image to reduce the influence of illumination change and color change on the color histogram and obtain more accurate color data. First, illumination normalization is performed to eliminate the influence of illumination change on the image, and then color correction is performed to eliminate color deviation, so that the color of the image is more realistic and accurate.

[0052] The reflection model is used to achieve illumination normalization. It is assumed that the image is composed of illumination component and reflection component. By estimating the illumination component and reflection component, the illumination changes in the image can be removed, thereby achieving illumination normalization. By converting the image from one color space to another, the color deviation can be eliminated, making the color of the image more realistic and accurate.

[0053] In order to more accurately obtain the ice thickness of each section of the line, the data analysis module 300 fuses the edge feature information, texture feature information and color feature information of the image transmitted by the feature extraction module 200, connects each feature vector to form a comprehensive feature vector, and then constructs an ice thickness prediction model, and selects the corresponding ultrasonic deicing frequency range according to the ice thickness;

[0054] A simple feature splicing method is used to connect the feature vectors to form a comprehensive feature vector. The model is trained using an image dataset with labeled ice thickness. The fused feature vector is used as input and the corresponding ice thickness is used as output. The relationship between the feature and the ice thickness is learned by training the model, and then the ice thickness corresponding to the comprehensive feature vector is obtained. The corresponding ultrasonic de-icing frequency range is determined based on the ice thickness.

[0055] like Figure 3 As shown, the control deicing module 400 includes an ultrasonic deicing unit 410 and a hot air deicing unit 420. The ultrasonic deicing unit 410 controls the vibration frequency of the ultrasonic wave to process the ice layer attached to the power line. The hot air deicing unit 420 heats the air or steam to increase its temperature to a level sufficient to melt the ice layer, and then blows the hot air or steam toward the power line to melt the ice layer and fall off.

[0056] For example, if the ice layer attached to the surface of the power line is thin (<10mm), a higher deicing frequency can be selected, such as 80kHz to 100kHz; if the ice layer attached to the surface of the power line is thin (>10mm), a lower deicing frequency can be selected, such as 20kHz to 40kHz. The higher the frequency of the ultrasonic wave, the greater its energy and the smaller the vibration amplitude; conversely, the lower the frequency, the smaller the energy and the greater the vibration amplitude.

[0057] Among them, for the thinner ice layer attached to the surface of the power line, due to its relatively small surface area, a higher energy density is required to effectively break and remove the ice layer. The higher frequency ultrasonic wave has higher energy and can act more concentratedly on a smaller surface area, thereby improving the de-icing efficiency. At the same time, the higher frequency ultrasonic wave has a stronger penetrating ability during the propagation process, and can more easily penetrate the thinner ice layer and directly act on the surface of the power line. This can reduce the loss of energy in the ice layer and improve the de-icing effect. In addition, the higher frequency ultrasonic wave has a smaller vibration amplitude, which can reduce the mechanical damage to the power line and reduce the risk of line damage.

[0058] In order to completely remove the ice layer attached to the surface of the power line, the hot air deicing unit 420 performs hot air deicing on the nodes where the ice layer attached to the power line is too thick, and de-ices the corresponding nodes of the power line, and cooperates with the ultrasonic deicing to ensure that the ice layer attached to the surface of the power line is completely removed, and transmits the location information of the node to the data storage module 500 for memory storage, thereby expanding the database.

[0059] like Figure 4 As shown, the data storage module 500 includes a feature recognition unit 510, a memory storage unit 520 and a raw data unit 530. The feature recognition unit 510 recognizes the feature information in the feature extraction module 200. If the feature information corresponds to the feature information in the memory storage unit 520, the deicing module 400 is directly regulated to perform deicing. The memory storage unit 520 memorizes and stores the deicing method and effect of the deicing module 400 to expand the database. The raw data unit 530 records the raw comparison data.

[0060] The memory storage unit 520 records the de-icing method and effect of the de-icing module 400. When the feature recognition unit 510 recognizes that the feature information of the corresponding ice layer image is the same as the data of the memory storage unit 520, the de-icing module 400 can be directly controlled to perform de-icing, which greatly saves data processing time and improves de-icing efficiency.

[0061] The data storage module 500 constructs a new data model comparison library including:

[0062] The data collection layer forms a data comparison group by collecting the feature information extracted by the feature extraction module 200 and the corresponding deicing measures of the deicing control module 400;

[0063] The data storage layer is used to store the corresponding data comparison groups to ensure the integrity and security of the data;

[0064] The data processing layer cleans the stored data to remove duplicate data, erroneous data, and outliers;

[0065] The data comparison layer compares the feature data groups of the data comparison group with the feature data groups in the original database and excludes duplicate feature data groups;

[0066] By establishing this data model comparison library, the original data has been greatly expanded to facilitate dealing with more complex situations. At the same time, a lot of time has been saved and the processing efficiency of the system has been improved.

[0067] Taking into account that the external environment will affect the shooting effect of the high-definition camera, the image acquisition module 100 includes an environment detection unit 110 and a parameter adjustment unit 120. The environment detection unit 110 detects the environmental information near the power line and adjusts the parameters of the high-definition camera according to the environment. The parameter adjustment unit 120 adjusts the shooting parameters of the high-definition camera to ensure that a clear image of the attached ice layer of the power line is obtained.

[0068] In summary, the intelligent de-icing system for power lines installed by using a drone samples the ice layer attached to the surface of the power line through the image acquisition module 100, and transmits the image sampling information to the feature extraction module 200. The edge detection unit 210, the texture detection unit 220 and the color analysis unit 230 in the feature extraction module 200 extract the corresponding features of the image information, and compare them with the original data of the data storage module 500 to preliminarily obtain the ice thickness of the corresponding image. The data analysis module 300 then fuses the edge feature information, texture feature information and color feature information of the image, connects the feature vectors to form a comprehensive feature vector, and then constructs an ice thickness prediction model. The thickness selects the frequency range of ultrasonic deicing corresponding to achieve efficient removal of ice on the surface of the power line. The hot air deicing unit 420 is used to perform hot air deicing on the nodes where the ice layer attached to the power line is too thick. Deicing is performed on the corresponding nodes of the power line, and the ultrasonic deicing unit 410 cooperates with each other to ensure that the ice layer attached to the surface of the power line is completely removed. The deicing module 400 is controlled to pass the corresponding adjustment information of deicing and the corresponding deicing results to the memory storage unit 520 of the data storage module 500, thereby expanding the database. It is convenient for the next time when the characteristic information corresponds to the characteristic information in the memory storage unit 520, the deicing module 400 can be directly controlled to perform deicing, which greatly saves data processing time and improves deicing efficiency.

[0069] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. An intelligent deicing system for power lines installed using drones, characterized in that: It comprises an image acquisition module (100), a feature extraction module (200), a data analysis module (300), a deicing control module (400) and a data storage module (500); The image acquisition module (100) uses a high-definition camera carried by an unmanned aerial vehicle to take long-range and close-range multi-segment photographs of the circuit line. The long-range photographing takes a wide range of views of the power line and identifies and marks nodes where the ice layer attached to the surface of the power line is too thick. The close-range photographing takes a high-definition photograph of the ice layer attached to the surface of the power line to obtain image information of the ice layer attached to the surface of the power line, and transmits the image information and the marked node information to the feature extraction module (200); The feature extraction module (200) receives the image information transmitted by the image acquisition module (100), extracts the edge features, texture features and color features of the ice layer attached to the power line in the image, and transmits the feature information to the data analysis module (300); the feature extraction module (200) transmits the received marked node information to the deicing control module (400); The data analysis module (300) receives the feature information transmitted by the feature extraction module (200), performs a comprehensive analysis on the edge features, texture features and color features of each section of the power line image, and compares the features with the original data stored in the data storage module (500), obtains the thickness information of the ice layer attached to the surface of each section of the power line, and calculates the most suitable frequency range for ultrasonic deicing of the section of the power line, and transmits the frequency range information to the deicing control module (400); The deicing control module (400) receives the marked node information transmitted by the feature extraction module (200), controls the drone to perform hot air deicing on the node, adjusts the frequency range of the ultrasonic wave according to the frequency range information transmitted by the data analysis module (300), performs ultrasonic deicing on each section of the power line, and after deicing is completed, takes a photo of the deicing result of the power line through a high-definition camera carried by the drone, and transmits the deicing result of the section of the power line to the data storage module (500); The data storage module (500) is used to store original comparison data on the one hand, and to receive characteristic information corresponding to each section of the power line on the other hand, to make the characteristic information extracted by the characteristic extraction module (200) correspond to the deicing method adopted by the deicing control module (400), to establish a new data comparison group, and to memorize and store the data, thereby expanding the original database and being able to quickly adjust the frequency range of the ultrasonic wave when encountering the same image characteristic information next time.

2. The intelligent deicing system for power lines installed by drone according to claim 1 is characterized in that: The feature extraction module (200) comprises an edge detection unit (210), a texture detection unit (220) and a color analysis unit (230). The edge detection unit (210) performs edge feature detection on an image of ice layers attached to each section of a power line, and uses two 3×3 convolution kernels to respectively calculate the approximate gradient values ​​of the image in the horizontal and vertical directions, and performs convolution operations on the two convolution kernels and the original image to obtain a gradient image in the horizontal direction and a gradient image in the vertical direction. Edge information in the horizontal direction is obtained based on the gradient image in the horizontal direction, and grayscale changes of pixels in the horizontal direction are detected. Similarly, edge information in the vertical direction is obtained, and edge feature points of the ice layer image are extracted. The texture detection unit (220) performs texture feature detection on the image of the ice layer attached to each section of the power line, describes the texture feature of the image by counting the frequency of occurrence of the grayscale combination of two pixels at a certain distance and direction in the image, and then extracts the corresponding texture feature information; The color analysis unit (230) performs ice layer color analysis on the image of ice layer attached to each section of power line, performs statistics on each pixel point in the image, records the frequency of occurrence of its color value in a specific color space, uses the number of pixels in each color interval obtained by statistics as the ordinate of the histogram, and the color interval as the abscissa, to construct a color histogram, and obtain corresponding ice layer color data.

3. The intelligent deicing system for power lines installed by drone according to claim 2 is characterized in that: When detecting ice layers attached to power lines, the edge detection unit (210) first uses a traditional edge detection algorithm to perform preliminary detection, and then uses a deep learning method to perform further optimization and refinement.

4. The intelligent deicing system for power lines installed by using a drone according to claim 2 is characterized in that: When detecting the ice layer attached to the power line, the texture detection unit (220) detects the ice layer texture using a gray level co-occurrence matrix, and a plurality of texture feature parameters can be extracted from the gray level co-occurrence matrix.

5. The intelligent deicing system for power lines installed by using drones according to claim 2 is characterized in that: The color analysis unit (230) performs preprocessing of illumination normalization and color correction on the image, thereby reducing the impact of illumination changes and color changes on the color histogram and obtaining more accurate color data.

6. The intelligent deicing system for power lines installed by using a drone according to claim 1 is characterized in that: The data analysis module (300) fuses the edge feature information, texture feature information and color feature information of the image transmitted by the feature extraction module (200), connects the feature vectors to form a comprehensive feature vector, and then constructs an ice layer thickness prediction model, and selects the corresponding ultrasonic deicing frequency range according to the ice layer thickness.

7. The intelligent deicing system for power lines installed by using drones according to claim 1 is characterized in that: The control deicing module (400) comprises an ultrasonic deicing unit (410) and a hot air deicing unit (420). The ultrasonic deicing unit (410) controls the vibration frequency of ultrasonic waves to process the ice layer attached to the power line. The hot air deicing unit (420) heats air or steam to increase its temperature to a level sufficient to melt the ice layer, and then blows the hot air or steam toward the power line, so that the ice layer is melted by the heat and falls off.

8. The intelligent deicing system for power lines installed by using a drone according to claim 7 is characterized in that: The hot air deicing unit (420) performs hot air deicing on nodes where the thickness of ice attached to the power line is too large, and transmits the position information of the node to the data storage module (500) for memory storage.

9. The intelligent deicing system for power lines installed by using drones according to claim 1 is characterized in that: The data storage module (500) comprises a feature recognition unit (510), a memory storage unit (520) and a raw data unit (530); the feature recognition unit (510) recognizes feature information in the feature extraction module (200); if the feature information corresponds to feature information in the memory storage unit (520), the deicing module (400) is directly regulated to perform deicing; the memory storage unit (520) memorizes and stores the deicing method and effect of the deicing module (400) to expand the database; and the raw data unit (530) records raw comparison data.

10. The intelligent deicing system for power lines installed by using drones according to claim 1 is characterized in that: The image acquisition module (100) comprises an environment detection unit (110) and a parameter adjustment unit (120); the environment detection unit (110) detects environmental information near a power line and adjusts parameters of a high-definition camera according to the environment; and the parameter adjustment unit (120) adjusts shooting parameters of the high-definition camera to ensure that a clear image of an ice layer attached to the power line is obtained.

Images