Power channel operation and maintenance environment hidden danger detection method based on hidden danger prediction model

Abstract

The invention relates to a power channel operation and maintenance environment hidden danger detection method based on a hidden danger prediction model. The method comprises the following steps: collecting regional image data of a power channel region at each time point, and identifying each environment image and a power channel image of the regional image data; environment types and image features corresponding to the environment images are identified through an image feature classification model, and the actual distance between each environment image and the power channel image is identified; analyzing each hidden danger type of the power channel region based on each environment type, the image feature and the actual spacing, and identifying the hidden danger probability of the power channel region in each hidden danger type based on the actual spacing of each environment image with the same image content in the image data of each region, and obtaining operation and maintenance environment hidden danger information of the power channel region. By adopting the method, the accuracy of detecting the environmental hidden danger of power channel operation and maintenance can be improved.

Description

Technical Field

[0001] The present application relates to the technical field of power channel operation and maintenance, and in particular to a method for detecting hidden dangers in the power channel operation and maintenance environment based on a hidden danger prediction model. Background Art

[0002] Power channel operation and maintenance refers to the effective management, monitoring, maintenance, repair and upgrading of power system equipment and facilities to ensure the reliable operation and stable power supply of the power system. This includes the maintenance of power lines, power operation, and power emergency repairs. The main contents include line inspection, troubleshooting, emergency repair and restoration, etc., to ensure the safe and stable supply of electricity. Among them, during line inspection and troubleshooting, the cost of power channel operation and maintenance caused by environmental factors increases year by year, which is one of the main factors causing power channel failures. Therefore, how to improve the efficiency and accuracy of environmental hazard detection in power channel operation and maintenance is the current research focus.

[0003] The traditional method of detecting environmental hazards in power channel operation and maintenance is to conduct regular manual inspections. During the manual inspections, each location of the power channel is manually observed and the environmental hazards that may cause power channel failures are manually analyzed. However, due to the poor comprehensiveness of manual observations and the influence of manual subjective factors on the efficiency and accuracy of manual analysis, the accuracy of environmental hazard detection in power channel operation and maintenance is low. Summary of the invention

[0004] Based on this, it is necessary to provide a method, device, computer equipment, computer-readable storage medium and computer program product for detecting hidden dangers in the operation and maintenance environment of a power channel based on a hidden danger prediction model in order to solve the above-mentioned technical problems.

[0005] In a first aspect, the present application provides a method for detecting hidden dangers in an electric power channel operation and maintenance environment based on a hidden danger prediction model, comprising:

[0006] Using satellite image acquisition software, collect regional image data of the power channel area at various time points, and for each regional image data, based on the regional image data, identify each environmental image of the regional image data and the power channel image of the regional image data through an image recognition network;

[0007] Identify the environment type corresponding to each of the environment images and the image features of each of the environment images through an image feature classification model, and identify the actual distance between each environment image and the power channel image;

[0008] Analyze the types of hidden dangers in the power channel area based on the environment types, the image features, and the actual distances, and identify the probability of hidden dangers in the power channel area in each hidden danger type based on the actual distances of the environment images with the same image content in the image data of each area;

[0009] Each hidden danger type of the power channel area and the hidden danger probability of the power channel area in each hidden danger type are used as the hidden danger information of the operation and maintenance environment of the power channel area.

[0010] Optionally, the identifying, based on the regional image data, each environment image of the regional image data and the power channel image of the regional image data through an image recognition network includes:

[0011] Using an image edge recognition algorithm, the image edge information in the regional image data is respectively identified, and the image content surrounded by each image edge information is respectively screened;

[0012] Input each image content into the image recognition network to identify the image type corresponding to the image content and the image information corresponding to the image content; the image type includes an environment image type and a power channel image type;

[0013] The image content of the environment image type and the image information corresponding to the image content of the environment image type are used as the environment image, and the image content of the power channel image type and the image information corresponding to the image content of the power channel image type are used as the power channel image.

[0014] Optionally, the image feature classification model includes an image classification network, an image feature extraction network for each environment type, and a text feature extraction network. Identifying the environment type corresponding to each environment image and the image features of each environment image through the image feature classification model includes:

[0015] Input each environment image into the image classification network to identify the environment type corresponding to each environment image;

[0016] Based on the image feature extraction network of each environment type, first image features are extracted from the image content of each environment image respectively, and based on the text feature extraction network, second image features are extracted from the image information of each environment image respectively;

[0017] The first image feature of each of the environment images and the second image feature of each of the environment images are used as image features of each of the environment images.

[0018] Optionally, identifying an actual distance between each environment image and the power channel image includes:

[0019] For each environment image, identifying the image edge information of the environment image at the environment edge position point of the regional image data, and the image edge information of the power channel image at each channel edge position point of the regional image data;

[0020] Calculate the position distance between each environment edge position point and each channel edge position point respectively, and select the adjacent channel edge position point corresponding to the minimum position distance of each environment edge position point from the position distances between each environment edge position point and each channel edge position point;

[0021] The position distance between all the environment edge position points of the environment image and the adjacent channel edge position point corresponding to each environment edge position point is used as the actual distance between the environment image and the power channel image.

[0022] Optionally, analyzing the types of hidden dangers in the power channel area based on the environment types, the image features, and the actual distances includes:

[0023] For each environment image, based on the environment type of the environment image, query each initial hidden danger type corresponding to the environment type in the hidden danger database, and extract the sample image feature of each sample image in the sample images corresponding to each initial hidden danger type through the image feature extraction network;

[0024] Respectively calculating the actual spacing corresponding to the environment image and the spacing deviation value between each sample image and the actual spacing of the sample of the power channel image, and selecting the sample image corresponding to the spacing deviation value lower than the deviation value threshold as the target sample image corresponding to the environment image;

[0025] The average similarities between the image features of the environment image and the sample image features of each target sample image are calculated respectively by using a similarity recognition algorithm, and the initial hidden danger type with the maximum average similarity is selected as the hidden danger type corresponding to the environment image.

[0026] Optionally, the identifying the hidden danger probability of each hidden danger type in the power channel area based on the actual distance between each environment image with the same image content in each area image data includes:

[0027] The actual distances between the environment images with the same image content in each area image data are sorted according to the time sequence corresponding to the time point to which each area image data belongs, so as to obtain the actual distance change information of the same image content;

[0028] Based on the hidden danger types corresponding to the environmental images of each identical image content, the mapping relationship between the spacing change information corresponding to the hidden danger type of each identical image content and the hidden danger probability is queried in the hidden danger database, and based on the actual spacing change information of each identical image content and the mapping relationship corresponding to each identical image content, the hidden danger probability corresponding to each identical image content is identified;

[0029] The hidden danger probability of each identical image content is used as the initial hidden danger probability of each hidden danger type of the identical image content, and for each hidden danger type, among the initial hidden danger probabilities of the same hidden danger type, the initial hidden danger probability corresponding to the highest hidden danger probability is selected as the hidden danger probability of the hidden danger type.

[0030] In the second aspect, the present application also provides a power channel operation and maintenance environment hidden danger detection device based on a hidden danger prediction model, comprising:

[0031] The acquisition module is used to acquire regional image data of the power channel area at various time points through satellite image acquisition software, and for each regional image data, based on the regional image data, identify various environmental images of the regional image data and the power channel image of the regional image data through an image recognition network;

[0032] An identification module, used to identify the environment type corresponding to each of the environment images and the image features of each of the environment images through an image feature classification model, and to identify the actual distance between each environment image and the power channel image;

[0033] an analysis module, configured to analyze the types of hidden dangers in the power channel area based on the types of environment, the features of the image, and the actual distances, and to identify the probability of hidden dangers in the power channel area in each hidden danger type based on the actual distances of the environment images with the same image content in the image data of each area;

[0034] The determination module is used to use each hidden danger type of the power channel area and the hidden danger probability of the power channel area in each hidden danger type as the operation and maintenance environment hidden danger information of the power channel area.

[0035] Optionally, the acquisition module is specifically used to:

[0036] Using an image edge recognition algorithm, the image edge information in the regional image data is respectively identified, and the image content surrounded by each image edge information is respectively screened;

[0037] Input each image content into the image recognition network to identify the image type corresponding to the image content and the image information corresponding to the image content; the image type includes an environment image type and a power channel image type;

[0038] The image content of the environment image type and the image information corresponding to the image content of the environment image type are used as the environment image, and the image content of the power channel image type and the image information corresponding to the image content of the power channel image type are used as the power channel image.

[0039] Optionally, the identification module is specifically used to:

[0040] Input each environment image into the image classification network to identify the environment type corresponding to each environment image;

[0041] Based on the image feature extraction network of each environment type, first image features are extracted from the image content of each environment image respectively, and based on the text feature extraction network, second image features are extracted from the image information of each environment image respectively;

[0042] The first image feature of each of the environment images and the second image feature of each of the environment images are used as image features of each of the environment images.

[0043] Optionally, the identification module is specifically used to:

[0044] For each environment image, identifying the image edge information of the environment image at the environment edge position point of the regional image data, and the image edge information of the power channel image at each channel edge position point of the regional image data;

[0045] Calculate the position distance between each environment edge position point and each channel edge position point respectively, and select the adjacent channel edge position point corresponding to the minimum position distance of each environment edge position point from the position distances between each environment edge position point and each channel edge position point;

[0046] The position distance between all the environment edge position points of the environment image and the adjacent channel edge position point corresponding to each environment edge position point is used as the actual distance between the environment image and the power channel image.

[0047] Optionally, the analysis module is specifically used to:

[0048] For each environment image, based on the environment type of the environment image, query each initial hidden danger type corresponding to the environment type in the hidden danger database, and extract the sample image feature of each sample image in the sample images corresponding to each initial hidden danger type through the image feature extraction network;

[0049] Respectively calculating the actual spacing corresponding to the environment image and the spacing deviation value between each sample image and the actual spacing of the sample of the power channel image, and selecting the sample image corresponding to the spacing deviation value lower than the deviation value threshold as the target sample image corresponding to the environment image;

[0050] The average similarities between the image features of the environment image and the sample image features of each target sample image are calculated respectively by using a similarity recognition algorithm, and the initial hidden danger type with the maximum average similarity is selected as the hidden danger type corresponding to the environment image.

[0051] Optionally, the analysis module is specifically used to:

[0052] The actual distances between the environment images with the same image content in each area image data are sorted according to the time sequence corresponding to the time point to which each area image data belongs, so as to obtain the actual distance change information of the same image content;

[0053] Based on the hidden danger types corresponding to the environmental images of each identical image content, the mapping relationship between the spacing change information corresponding to the hidden danger type of each identical image content and the hidden danger probability is queried in the hidden danger database, and based on the actual spacing change information of each identical image content and the mapping relationship corresponding to each identical image content, the hidden danger probability corresponding to each identical image content is identified;

[0054] The hidden danger probability of each identical image content is used as the initial hidden danger probability of each hidden danger type of the identical image content, and for each hidden danger type, among the initial hidden danger probabilities of the same hidden danger type, the initial hidden danger probability corresponding to the highest hidden danger probability is selected as the hidden danger probability of the hidden danger type.

[0055] In a third aspect, the present application provides a computer device, wherein the computer device comprises a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program, the steps of any one of the methods in the first aspect are implemented.

[0056] In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the steps of any one of the methods in the first aspect are implemented.

[0057] In a fifth aspect, the present application provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the steps of any one of the methods in the first aspect are implemented.

[0058] The above-mentioned method for detecting hidden dangers in the operation and maintenance environment of the power channel based on the hidden danger prediction model collects regional image data of the power channel area at various time points through satellite image acquisition software, and for each regional image data, based on the regional image data, identifies each environmental image of the regional image data and the power channel image of the regional image data through an image recognition network; identifies the environment type corresponding to each of the environmental images and the image features of each of the environmental images through an image feature classification model, and identifies the actual distance between each environmental image and the power channel image; analyzes each hidden danger type of the power channel area based on each of the environmental types, each of the image features, and each of the actual distances, and identifies the hidden danger probability of the power channel area in each hidden danger type based on the actual distance of each of the environmental images with the same image content in each of the regional image data; and uses each hidden danger type of the power channel area and the hidden danger probability of the power channel area in each hidden danger type as the operation and maintenance environment hidden danger information of the power channel area. This scheme first collects regional images of the power channel area at various time points through satellite image acquisition software, which not only ensures the comprehensiveness of the collected images, but also the all-round, multi-angle, and global images of the power channel area that can be obtained from the perspective of satellite image acquisition, avoiding the one-sidedness, limitations, and dislocation of manual observation, thereby affecting the accuracy of analyzing the hidden dangers of the power channel operation and maintenance environment. Then, this scheme uses an image recognition network to identify the various environmental images and power channel images of the power channel area, thereby analyzing the hidden dangers of interference from each environmental image to the power channel image, and improving the comprehensiveness and accuracy of identifying the hidden dangers of the power channel area. After that, this sub-scheme identifies the various hidden danger types of the power channel area by considering the various environmental types, image features, and actual spacing of each environmental image, and identifies the hidden danger probability of the power channel area in each hidden danger type through the actual spacing of each environmental image with the same image content in each regional image data. That is, the hidden danger types in the power channel area and the hidden danger probabilities of each hidden danger type are identified from different angles and in an all-round manner, avoiding the high labor costs of manual analysis and the influence of human subjective factors, thereby comprehensively improving the accuracy of environmental hidden danger detection in power channel operation and maintenance. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related technologies, the drawings required for use in the embodiments or the related technical descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0060] Figure 1It is a flowchart of a method for detecting hidden dangers in an operation and maintenance environment of a power channel based on a hidden danger prediction model in one embodiment;

[0061] Figure 2 A schematic diagram of a process of an example of hidden danger detection in an electric power channel operation and maintenance environment based on a hidden danger prediction model in one embodiment;

[0062] Figure 3 It is a structural block diagram of a hidden danger detection device for an electric power channel operation and maintenance environment based on a hidden danger prediction model in one embodiment;

[0063] Figure 4 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0064] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0065] The method for detecting hidden dangers in the operation and maintenance environment of a power channel based on a hidden danger prediction model provided by an embodiment of the present application can be applied to the application environment of hidden danger detection in the operation and maintenance environment of a power channel. Among them, the method can be applied to a terminal, can be applied to a server, can be applied to a system including a terminal and a server, and is implemented through the interaction between the terminal and the server. Among them, the terminal can be, but is not limited to, various personal computers, laptops, etc. Among them, the terminal first collects regional images of the power channel area at various time points through satellite image acquisition software, which not only ensures the comprehensiveness of the collected images, but also the all-round, multi-angle, and global images of the power channel area that can be obtained from the perspective of satellite image acquisition, avoiding the one-sidedness, limitations, and dislocation of manual observation, thereby affecting the accuracy of analyzing the hidden dangers of the operation and maintenance environment of the power channel. Then, this scheme uses an image recognition network to respectively identify the various environmental images and the power channel images of the power channel area, thereby analyzing the hidden dangers of interference of each environmental image on the power channel image, and improving the comprehensiveness and accuracy of identifying the hidden dangers of the power channel area. Then, this sub-scheme identifies various hidden danger types in the power channel area by considering various environmental types, image features, and actual spacings of each environmental image, and identifies the hidden danger probability of the power channel area in various hidden danger types through the actual spacings of environmental images with the same image content in each area image data. That is, the hidden danger types and hidden danger probabilities of various hidden danger types in the power channel area are identified from different angles and in all directions, avoiding the high labor cost of manual analysis and the problem of being affected by manual subjective factors, thereby comprehensively improving the accuracy of environmental hidden danger detection for power channel operation and maintenance.

[0066] In an exemplary embodiment, Figure 1 As shown, a method for detecting hidden dangers in the operation and maintenance environment of a power channel based on a hidden danger prediction model is provided, and the method is applied to a terminal as an example for explanation, including the following steps S101 to S104. Among them:

[0067] Step S101, using satellite image acquisition software, collects regional image data of the power channel area at various time points, and for each regional image data, based on the regional image data, identifies various environmental images of the regional image data and the power channel image of the regional image data through an image recognition network.

[0068] In this embodiment, the terminal collects image data of the power channel area under the condition of unit time interval in real time through satellite image acquisition software based on satellite remote sensing technology, and obtains regional image data of the power channel area at each time point. Among them, the time interval between each time point is the unit time interval, and the unit time interval is the time period preset by the staff in the terminal, and the time period can be 1 day, 10 days, 15 days, 20 days, etc. Among them, the power channel area is the area through which the power channel passes, and the power channel includes transmission lines and substation projects. Among them, each power channel area is an area range manually demarcated by the staff. The power channel area includes a small substation area, a medium-sized substation area, a large substation area, or a current transmission center area where multiple cables intersect. Then, the terminal identifies each environmental image of the regional image data and the power channel image of the regional image data through an image recognition network. Among them, each environmental image is a range image of the environment such as facilities and buildings of each non-power equipment in the power channel image, and the power channel image is an image containing all power equipment. The non-power equipment includes, for example, plastic greenhouse buildings, temporary construction buildings, carport buildings, and buildings such as movable buildings (board houses, iron houses, RVs, containers), and non-movable buildings (buildings, factories, roads, and structures fixedly connected to the ground). The power equipment includes, for example, wires, transformer boxes, meter boxes, electric poles, electric pole brackets, power construction equipment, etc. The image recognition network is a convolutional neural network.

[0069] Step S102, identifying the environment type corresponding to each environment image and the image features of each environment image through an image feature classification model, and identifying the actual distance between each environment image and the power channel image.

[0070] In this embodiment, the terminal identifies the environment type corresponding to each environment image and the image features of each environment image through an image feature classification model, and identifies the actual distance between each environment image and the power channel image. Wherein, the image feature classification model includes an image classification network, an image feature extraction network for each environment type, and a text feature extraction network. Wherein, the image classification network is an Alexnet network based on deep learning, the text feature extraction network is a text feature extractor (bag of words model) based on natural language processing technology (NLP), and the image feature extraction network is a CNN convolutional neural network obtained by network training through sample data of each environment type. Wherein, the environment type is the environment type corresponding to each environment image, and the environment type includes but is not limited to equipment type, facility type, and building type. Wherein, the actual distance between each environment image and the power channel image is the distance range of the closest interval distance between each edge position point in the image edge information of the environment image and each channel edge position point of the power channel image. The specific recognition process will be described in detail later.

[0071] Step S103, based on each environment type, each image feature, and each actual distance, analyze each hidden danger type in the power channel area, and identify the hidden danger probability of the power channel area in each hidden danger type based on the actual distance of each environment image with the same image content in the image data of each area.

[0072] In this embodiment, the terminal analyzes the various hidden danger types in the power channel area based on the various environment types, the image features, and the actual spacing, and identifies the hidden danger probabilities of the power channel area in various hidden danger types based on the actual spacing of the environment images with the same image content in the image data of each area. Among them, the hidden danger types include but are not limited to covering / contact hidden danger types, destruction hidden danger types, and short circuit hidden danger types. The hidden danger probability of each hidden danger type ranges from 1% to 100%. The higher the hidden danger probability, the more likely it is that the hidden danger type will occur in the power channel area, and the lower the hidden danger probability, the less likely it is that the hidden danger type will occur in the power channel area. The specific identification process will be described in detail later.

[0073] Step S104: taking each hidden danger type of the power channel area and the hidden danger probability of the power channel area in each hidden danger type as the hidden danger information of the operation and maintenance environment of the power channel area.

[0074] In this embodiment, the terminal uses various hidden danger types of the power channel area and the hidden danger probabilities of the power channel area in each hidden danger type as the hidden danger information of the operation and maintenance environment of the power channel area.

[0075] Based on the above scheme, firstly, the regional images of the power channel area at each time point are collected through the satellite image acquisition software, which not only ensures the comprehensiveness of the collected images, but also the all-round, multi-angle, and global images of the power channel area that can be obtained from the perspective of satellite image acquisition, avoiding the one-sidedness, limitations, and dislocation of manual observation, thereby affecting the accuracy of analyzing the hidden dangers of the power channel operation and maintenance environment. Then, this scheme uses the image recognition network to identify the various environmental images and power channel images of the power channel area respectively, so as to analyze the hidden dangers of the interference of each environmental image to the power channel image respectively, thereby improving the comprehensiveness and accuracy of identifying the hidden dangers of the power channel area. After that, this sub-scheme identifies the various hidden danger types of the power channel area by considering the various environmental types, image features, and actual spacings of each environmental image, and identifies the hidden danger probability of the power channel area in each hidden danger type through the actual spacing of each environmental image with the same image content in each regional image data. That is, the hidden danger types in the power channel area and the hidden danger probabilities of each hidden danger type are identified from different angles and in an all-round manner, avoiding the high labor costs of manual analysis and the influence of human subjective factors, thereby comprehensively improving the accuracy of environmental hidden danger detection in power channel operation and maintenance.

[0076] Optionally, based on the regional image data, each environmental image of the regional image data and the power channel image of the regional image data are identified through an image recognition network, including: using an image edge recognition algorithm to respectively identify each image edge information in the regional image data, and respectively filter the image content surrounded by each image edge information; input each image content into the image recognition network to respectively identify the image type corresponding to the image content and the image information corresponding to the image content; the image type includes an environmental image type and a power channel image type; the image content of the environmental image type and the image information corresponding to the image content of the environmental image type are used as the environmental image, and the image content of the power channel image type and the image information corresponding to the image content of the power channel image type are used as the power channel image.

[0077] In this embodiment, the terminal uses an image edge recognition algorithm to respectively identify the image edge information in the regional image data, and respectively screens the image content surrounded by each image edge information. The image edge recognition algorithm is a Canny multi-level edge detection algorithm. The algorithm can identify the image edge information of each image content in the regional image data.

[0078] Then, the terminal inputs each image content into the image recognition network, and identifies the image type corresponding to the image content and the image information corresponding to the image content. Among them, the image information is the physical information corresponding to the image content. The physical information is the physical information of the equipment / facility / building corresponding to the image content, and the physical information includes but is not limited to stability information, range change mode information, conductivity information, volume information, etc. The identification process of the physical information is that after the terminal identifies the equipment / facility / building corresponding to each image content based on the image recognition network, it measures the image range of the image content in proportion (i.e., the size ratio of the satellite image to the actual object), identifies the volume information of the equipment / facility / building corresponding to the graphic content, and then, the terminal queries the conductivity information, hardness information, density information, quality information, burn resistance information, and hardness information corresponding to each equipment / facility / building corresponding to the graphic content in the environmental database. Among them, the physical information is text information.

[0079] The image type includes an environment image type and a power channel image type. Then, the terminal uses the image content of the environment image type and the image information corresponding to the image content of the environment image type as the environment image, and uses the image content of the power channel image type and the image information corresponding to the image content of the power channel image type as the power channel image.

[0080] Based on the above scheme, after performing image recognition on the image content surrounded by each image edge information identified by the image edge recognition algorithm through image recognition network decomposition, the image type of each image content is determined, thereby improving the comprehensiveness and accuracy of recognition.

[0081] Optionally, the image feature classification model includes an image classification network, an image feature extraction network for each environment type, and a text feature extraction network. The image feature classification model is used to identify the environment type corresponding to each environment image and the image features of each environment image, including: inputting each environment image into the image classification network to identify the environment type corresponding to each environment image; extracting the first image features in the image content of each environment image based on the image feature extraction network for each environment type, and extracting the second image features in the image information of each environment image based on the text feature extraction network; and using the first image features of each environment image and the second image features of each environment image as the image features of each environment image.

[0082] In this embodiment, the terminal inputs each environment image into the image classification network to identify the environment type corresponding to each environment image.

[0083] Then, the terminal extracts the first image feature in the image content of each environment image based on the image feature extraction network of each environment type, and extracts the second image feature in the image information of each environment image based on the text feature extraction network. The first image feature is the content feature of the image content of the environment image, and the image information is the text feature of the physical information corresponding to the image information.

[0084] Finally, the terminal uses the first image feature of each environmental image and the second image feature of each environmental image as the image feature of each environmental image.

[0085] Based on the above solution, after identifying the environment type corresponding to each environment image, the image features are extracted through the image feature extraction network of each environment type, thereby improving the comprehensiveness and accuracy of image feature extraction.

[0086] Optionally, identifying the actual distance between each environment image and the power channel image includes: for each environment image, identifying the environment edge position points of the image edge information of the environment image in the regional image data, and the image edge information of the power channel image in each channel edge position points of the regional image data; respectively calculating the position distance between each environment edge position point and each channel edge position point, and screening the adjacent channel edge position points corresponding to the minimum position distance of each environment edge position point from the position distance between each environment edge position point and each channel edge position point; and taking the position distance between all environment edge position points of the environment image and the adjacent channel edge position points corresponding to each environment edge position point as the actual distance between the environment image and the power channel image.

[0087] In this embodiment, the terminal identifies the image edge information of the environment image at the environment edge position point of the regional image data, and the image edge information of the power channel image at the edge position points of each channel of the regional image data for each environment image. The spacing between each edge position point (i.e., environment edge position point or channel edge position point) is a unit spacing preset by the staff in the terminal, such as 1 pixel, 1.5 pixels, 2 pixels, etc. Each edge position point is a two-dimensional position information with each unit spacing as a distance scale after a two-dimensional coordinate system is constructed with the edge of the regional image data.

[0088] Then, the terminal calculates the position spacing between each environment edge position point and each channel edge position point respectively, and selects the adjacent channel edge position point corresponding to the minimum position spacing of each environment edge position point from the position spacing between each environment edge position point and each channel edge position point, wherein the position spacing is the image position spacing.

[0089] Finally, the terminal performs proportional scaling processing on the positional spacing between all environmental edge position points of the environmental image and the adjacent channel edge position points corresponding to each environmental edge position point, and obtains the actual positional spacing value corresponding to the positional spacing between all environmental edge position points and the adjacent channel edge position points corresponding to each environmental edge position point, as the actual spacing between the environmental image and the power channel image. The proportional scaling processing is a distance ratio value of the actual distance corresponding to a single pixel point of the image preset in the terminal.

[0090] Based on the above scheme, by screening the position distance between each environmental edge position point and the adjacent channel edge position point corresponding to each environmental edge position point as the actual distance between the environmental image and the power channel image, the comprehensiveness and fault tolerance of the recognition of the actual distance are improved, and the problem of selecting the position distance between single position points as the actual distance is avoided, thereby affecting the subsequent identification of hidden danger types and hidden danger probabilities.

[0091] Optionally, based on each environment type, each image feature, and each actual spacing, each hidden danger type in the power channel area is analyzed, including: for each environment image, based on the environment type of the environment image, querying each initial hidden danger type corresponding to the environment type in the hidden danger database, and extracting the sample image feature of each sample image in the sample image corresponding to each initial hidden danger type through the image feature extraction network; respectively calculating the actual spacing corresponding to the environment image and the spacing deviation value between each sample image and the sample actual spacing of the power channel image, and screening the sample images corresponding to the spacing deviation value lower than the deviation value threshold as the target sample images corresponding to the environment image; respectively calculating the average similarity between each image feature of the environment image and the sample image feature of each target sample image through a similarity recognition algorithm, and screening the initial hidden danger type with the maximum average similarity as the hidden danger type corresponding to the environment image.

[0092] In this embodiment, for each environmental image, based on the environmental type of the environmental image, the terminal queries the initial hidden danger types corresponding to the environmental type in the hidden danger database, and extracts the sample image features of each sample image in the sample images corresponding to the initial hidden danger types through the image feature extraction network. Among them, each hidden danger database stores various hidden danger types corresponding to different environmental types. The corresponding relationship is a database obtained by the staff through data statistics, expert advice, and Internet data query after summary and improvement.

[0093] Afterwards, the terminal calculates the actual spacing corresponding to the environmental image and the spacing deviation value between each sample image and the actual spacing of the power channel image, and selects the sample images corresponding to the spacing deviation value lower than the deviation value threshold preset in the terminal as the target sample images corresponding to the environmental image. This avoids the problem that the hidden danger types corresponding to the sample images with larger actual spacing deviation values ​​are different, thereby affecting the accuracy of identifying the hidden danger types of the environmental image.

[0094] The terminal calculates the similarity between each image feature of the environment image and the sample image feature of each target sample image through a similarity recognition algorithm, and then calculates the average value of the similarity between all image features of the environment image and all sample image features of each target sample image to obtain the average similarity between the environment image and each target sample image. The similarity recognition algorithm is a cosine similarity recognition algorithm.

[0095] Finally, the terminal selects the initial hidden danger type with the maximum average similarity as the hidden danger type corresponding to the environmental image.

[0096] Based on the above scheme, by considering the environmental types, image features, and actual spacings of each environmental image, the various hidden danger types in the power channel area are identified, and the hidden danger types in the power channel area are identified from different angles and in an all-round manner, avoiding the high labor cost of manual analysis and the influence of human subjective factors, thereby comprehensively improving the accuracy of environmental hidden danger detection in power channel operation and maintenance.

[0097] Optionally, based on the actual spacing of each environmental image with the same image content in each area image data, the hidden danger probability of each hidden danger type in the power channel area is identified, including: sorting the actual spacing of each environmental image with the same image content in each area image data according to the time sequence corresponding to the time point to which each area image data belongs, to obtain the actual spacing change information of the same image content; based on the hidden danger type corresponding to each environmental image with the same image content, querying the mapping relationship between the spacing change information corresponding to each hidden danger type with the same image content and the hidden danger probability in the hidden danger database, and identifying the hidden danger probability corresponding to each same image content based on the actual spacing change information of each same image content and the mapping relationship corresponding to each same image content; taking the hidden danger probability of each same image content as the initial hidden danger probability of each hidden danger type with the same image content, and for each hidden danger type, screening the initial hidden danger probability corresponding to the highest hidden danger probability among the initial hidden danger probabilities of the same hidden danger type as the hidden danger probability of the hidden danger type.

[0098] In this embodiment, the terminal sorts the actual spacing of each environmental image of the same image content in each regional image data according to the time sequence corresponding to the time point to which each regional image data belongs, and obtains the actual spacing change information of the same image content. The actual spacing change information includes spacing changes such as the actual spacing increases, the actual spacing decreases, the actual spacing increases first and then decreases, the actual spacing decreases first and then increases, and the actual spacing fluctuates within a certain range, and the actual spacing change information includes a negative value, which indicates that the spacing distance between the environmental image and the power channel image is different from the initial spacing distance direction. For example, image A is located 5m to the left of the power channel image. After 2 days, image A is located 3m to the right of the power channel image. If the left side is a positive value and the right side is a negative value, the image spacing change information of image A is from +5m to -3m.

[0099] Then, based on the hidden danger types corresponding to each environmental image of each identical image content, the terminal queries the mapping relationship between the spacing change information corresponding to each hidden danger type of the identical image content and the hidden danger probability in the hidden danger database, and identifies the hidden danger probability corresponding to each identical image content based on the actual spacing change information of each identical image content and the mapping relationship corresponding to each identical image content. Among them, the mapping relationship between the spacing change information corresponding to different hidden danger types and the hidden danger probability includes the hidden danger probability corresponding to each spacing change value range of different spacing change situations. The terminal identifies the target spacing change value range to which each identical image content belongs by identifying the spacing change situation corresponding to the actual spacing change information of each identical image content and the spacing change value corresponding to the actual spacing change information of each identical image content, and uses the hidden danger probability corresponding to each target spacing change value range as the hidden danger probability corresponding to each identical image content.

[0100] Then, the hidden danger probability of each identical image content is finally used as the initial hidden danger probability of each hidden danger type of the identical image content, and for each hidden danger type, the initial hidden danger probability corresponding to the highest hidden danger probability is selected from the initial hidden danger probabilities of the same hidden danger type as the hidden danger probability of the hidden danger type.

[0101] Based on the above scheme, by identifying the actual spacing change information between environmental images with the same image content, the hidden danger probability corresponding to each identical image content is comprehensively analyzed, and then the hidden danger probability of each hidden danger type is judged, thereby improving the accuracy of identifying the hidden danger probability of each hidden danger type.

[0102] This application also provides an example of potential danger detection in the power channel operation and maintenance environment based on a potential danger prediction model, such as Figure 2 As shown, the specific processing process includes the following steps:

[0103] Step S201, collecting regional image data of the power channel area at various time points through satellite image acquisition software.

[0104] Step S202: using an image edge recognition algorithm, respectively identifying each image edge information in the regional image data, and respectively screening the image content surrounded by each image edge information.

[0105] Step S203, input each image content into the image recognition network to identify the image type corresponding to the image content and the image information corresponding to the image content; the image type includes an environment image type and a power channel image type.

[0106] Step S204, taking the image content of the environment image type and the image information corresponding to the image content of the environment image type as the environment image, and taking the image content of the power channel image type and the image information corresponding to the image content of the power channel image type as the power channel image.

[0107] Step S205: input each environment image into an image classification network to identify the environment type corresponding to each environment image.

[0108] Step S206, based on the image feature extraction network of each environment type, respectively extract the first image feature in the image content of each environment image, and based on the text feature extraction network, respectively extract the second image feature in the image information of each environment image.

[0109] Step S207: Using the first image feature of each environment image and the second image feature of each environment image as the image feature of each environment image.

[0110] Step S208 , for each environment image, identifying the image edge information of the environment image at the environment edge position point of the regional image data, and the image edge information of the power channel image at the edge position point of each channel of the regional image data.

[0111] Step S209, respectively calculate the position distance between each environment edge position point and each channel edge position point, and select the adjacent channel edge position point corresponding to the minimum position distance of each environment edge position point from the position distance between each environment edge position point and each channel edge position point.

[0112] Step S210: taking the position distance between all the environment edge position points of the environment image and the adjacent channel edge position points corresponding to each environment edge position point as the actual distance between the environment image and the power channel image.

[0113] Step S211, for each environmental image, based on the environmental type of the environmental image, query the initial hidden danger types corresponding to the environmental type in the hidden danger database, and extract the sample image features of each sample image in the sample images corresponding to the initial hidden danger types through the image feature extraction network.

[0114] Step S212, respectively calculating the actual spacing corresponding to the environment image and the spacing deviation value between each sample image and the sample actual spacing of the power channel image, and selecting the sample image corresponding to the spacing deviation value lower than the deviation value threshold as the target sample image corresponding to the environment image.

[0115] Step S213, using a similarity recognition algorithm, respectively calculate the average similarity between each image feature of the environment image and the sample image features of each target sample image, and select the initial hidden danger type with the maximum average similarity as the hidden danger type corresponding to the environment image.

[0116] Step S214 , sorting the actual distances of the environment images with the same image content in each area image data according to the time sequence corresponding to the time points to which the area image data belongs, to obtain actual distance change information of the same image content.

[0117] Step S215, based on the hidden danger types corresponding to each environmental image of the same image content, query the mapping relationship between the spacing change information corresponding to the hidden danger type of each same image content and the hidden danger probability in the hidden danger database, and identify the hidden danger probability corresponding to each same image content based on the actual spacing change information of each same image content and the mapping relationship corresponding to each same image content.

[0118] Step S216, taking the hidden danger probability of each identical image content as the initial hidden danger probability of each hidden danger type of the identical image content, and for each hidden danger type, selecting the initial hidden danger probability corresponding to the highest hidden danger probability from the initial hidden danger probabilities of the identical hidden danger type as the hidden danger probability of the hidden danger type.

[0119] Step S217, taking each hidden danger type of the power channel area and the hidden danger probability of the power channel area in each hidden danger type as the hidden danger information of the operation and maintenance environment of the power channel area.

[0120] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps does not have a strict order restriction, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0121] Based on the same inventive concept, the embodiment of the present application also provides a device for detecting hidden dangers in the operation and maintenance environment of a power channel based on a hidden danger prediction model, which is used to implement the method for detecting hidden dangers in the operation and maintenance environment of a power channel based on a hidden danger prediction model. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in the embodiments of one or more hidden danger prediction models for detecting hidden dangers in the operation and maintenance environment of a power channel provided below can refer to the limitations of the method for detecting hidden dangers in the operation and maintenance environment of a power channel based on a hidden danger prediction model, and will not be repeated here.

[0122] In an exemplary embodiment, Figure 3 As shown, a power channel operation and maintenance environment hidden danger detection device based on a hidden danger prediction model is provided, including: a collection module 310, an identification module 320, an analysis module 330 and a determination module 340, wherein:

[0123] The acquisition module 310 is used to acquire regional image data of the power channel area at each time point through satellite image acquisition software, and for each regional image data, based on the regional image data, identify each environmental image of the regional image data and the power channel image of the regional image data through an image recognition network;

[0124] The identification module 320 is used to identify the environment type corresponding to each of the environment images and the image features of each of the environment images through an image feature classification model, and to identify the actual distance between each environment image and the power channel image;

[0125] An analysis module 330 is used to analyze the types of hidden dangers in the power channel area based on the types of environment, the image features, and the actual distances, and identify the probability of hidden dangers in the power channel area in each hidden danger type based on the actual distances of the environment images with the same image content in the image data of each area;

[0126] The determination module 340 is used to use each hidden danger type of the power channel area and the hidden danger probability of the power channel area in each hidden danger type as the hidden danger information of the operation and maintenance environment of the power channel area.

[0127] Optionally, the acquisition module 310 is specifically used for:

[0128] Using an image edge recognition algorithm, the image edge information in the regional image data is respectively identified, and the image content surrounded by each image edge information is respectively screened;

[0129] Input each image content into the image recognition network to identify the image type corresponding to the image content and the image information corresponding to the image content; the image type includes an environment image type and a power channel image type;

[0130] The image content of the environment image type and the image information corresponding to the image content of the environment image type are used as the environment image, and the image content of the power channel image type and the image information corresponding to the image content of the power channel image type are used as the power channel image.

[0131] Optionally, the identification module 320 is specifically configured to:

[0132] Input each environment image into the image classification network to identify the environment type corresponding to each environment image;

[0133] Based on the image feature extraction network of each environment type, first image features are extracted from the image content of each environment image respectively, and based on the text feature extraction network, second image features are extracted from the image information of each environment image respectively;

[0134] The first image feature of each of the environment images and the second image feature of each of the environment images are used as image features of each of the environment images.

[0135] Optionally, the identification module 320 is specifically configured to:

[0136] For each environment image, identifying the image edge information of the environment image at the environment edge position point of the regional image data, and the image edge information of the power channel image at each channel edge position point of the regional image data;

[0137] Calculate the position distance between each environment edge position point and each channel edge position point respectively, and select the adjacent channel edge position point corresponding to the minimum position distance of each environment edge position point from the position distances between each environment edge position point and each channel edge position point;

[0138] The position distance between all the environment edge position points of the environment image and the adjacent channel edge position point corresponding to each environment edge position point is used as the actual distance between the environment image and the power channel image.

[0139] Optionally, the analysis module 330 is specifically used to:

[0140] For each environment image, based on the environment type of the environment image, query each initial hidden danger type corresponding to the environment type in the hidden danger database, and extract the sample image feature of each sample image in the sample images corresponding to each initial hidden danger type through the image feature extraction network;

[0141] Respectively calculating the actual spacing corresponding to the environment image and the spacing deviation value between each sample image and the actual spacing of the sample of the power channel image, and selecting the sample image corresponding to the spacing deviation value lower than the deviation value threshold as the target sample image corresponding to the environment image;

[0142] The average similarities between the image features of the environment image and the sample image features of each target sample image are calculated respectively by using a similarity recognition algorithm, and the initial hidden danger type with the maximum average similarity is selected as the hidden danger type corresponding to the environment image.

[0143] Optionally, the analysis module 330 is specifically used to:

[0144] The actual distances between the environment images with the same image content in each area image data are sorted according to the time sequence corresponding to the time point to which each area image data belongs, so as to obtain the actual distance change information of the same image content;

[0145] Based on the hidden danger types corresponding to the environmental images of each identical image content, the mapping relationship between the spacing change information corresponding to the hidden danger type of each identical image content and the hidden danger probability is queried in the hidden danger database, and based on the actual spacing change information of each identical image content and the mapping relationship corresponding to each identical image content, the hidden danger probability corresponding to each identical image content is identified;

[0146] The hidden danger probability of each identical image content is used as the initial hidden danger probability of each hidden danger type of the identical image content, and for each hidden danger type, among the initial hidden danger probabilities of the same hidden danger type, the initial hidden danger probability corresponding to the highest hidden danger probability is selected as the hidden danger probability of the hidden danger type.

[0147] Each module in the above-mentioned power channel operation and maintenance environment hidden danger detection device based on hidden danger prediction model can be implemented in whole or in part by software, hardware and their combination. Each of the above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to each of the above modules.

[0148] In an exemplary embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as shown in FIG. Figure 4 As shown. The computer device includes a processor, a memory, an input / output interface, a communication interface, a display unit and an input device. Among them, the processor, the memory and the input / output interface are connected through a system bus, and the communication interface, the display unit and the input device are connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The input / output interface of the computer device is used to exchange information between the processor and an external device. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be implemented through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a method for detecting hidden dangers in the operation and maintenance environment of a power channel based on a hidden danger prediction model is implemented. The display unit of the computer device is used to form a visually visible picture, which can be a display screen, a projection device or a virtual reality imaging device. The display screen can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covering the display screen, or a button, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse.

[0149] Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0150] In an exemplary embodiment, a computer device is provided, including a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program, the steps of any one of the methods in the first aspect are implemented.

[0151] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps of any one of the methods in the first aspect are implemented.

[0152] In one embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the steps of any one of the methods in the first aspect.

[0153] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant regulations.

[0154] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., but are not limited to this.

[0155] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0156] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for detecting hidden dangers in the operation and maintenance environment of a power channel based on a hidden danger prediction model, characterized in that: The method comprises: Using satellite image acquisition software, collect regional image data of the power channel area at various time points, and for each regional image data, based on the regional image data, identify each environmental image of the regional image data and the power channel image of the regional image data through an image recognition network; Identify the environment type corresponding to each of the environment images and the image features of each of the environment images through an image feature classification model, and identify the actual distance between each environment image and the power channel image; Analyze the types of hidden dangers in the power channel area based on the environment types, the image features, and the actual distances, and identify the probability of hidden dangers in the power channel area in each hidden danger type based on the actual distances of the environment images with the same image content in the image data of each area; Each hidden danger type of the power channel area and the hidden danger probability of the power channel area in each hidden danger type are used as the hidden danger information of the operation and maintenance environment of the power channel area.

2. The method according to claim 1, characterized in that The identifying, based on the regional image data, each environment image of the regional image data and the power channel image of the regional image data through an image recognition network includes: Using an image edge recognition algorithm, the image edge information in the regional image data is respectively identified, and the image content surrounded by each image edge information is respectively screened; Input each image content into the image recognition network to identify the image type corresponding to the image content and the image information corresponding to the image content; the image type includes an environment image type and a power channel image type; The image content of the environment image type and the image information corresponding to the image content of the environment image type are used as the environment image, and the image content of the power channel image type and the image information corresponding to the image content of the power channel image type are used as the power channel image.

3. The method according to claim 2, characterized in that The image feature classification model includes an image classification network, an image feature extraction network for each environment type, and a text feature extraction network. The image feature classification model is used to identify the environment type corresponding to each environment image and the image features of each environment image, including: Input each environment image into the image classification network to identify the environment type corresponding to each environment image; Based on the image feature extraction network of each environment type, first image features are extracted from the image content of each environment image respectively, and based on the text feature extraction network, second image features are extracted from the image information of each environment image respectively; The first image feature of each of the environment images and the second image feature of each of the environment images are used as image features of each of the environment images.

4. The method according to claim 2, characterized in that: The identifying the actual distance between each environment image and the power channel image includes: For each environment image, identifying the image edge information of the environment image at the environment edge position point of the regional image data, and the image edge information of the power channel image at each channel edge position point of the regional image data; Calculate the position distance between each environment edge position point and each channel edge position point respectively, and select the adjacent channel edge position point corresponding to the minimum position distance of each environment edge position point from the position distances between each environment edge position point and each channel edge position point; The position distance between all the environment edge position points of the environment image and the adjacent channel edge position point corresponding to each environment edge position point is used as the actual distance between the environment image and the power channel image.

5. The method according to claim 1, characterized in that: The analyzing each type of hidden danger in the power channel area based on each type of environment, each image feature, and each actual distance includes: For each environment image, based on the environment type of the environment image, query each initial hidden danger type corresponding to the environment type in the hidden danger database, and extract the sample image feature of each sample image in the sample images corresponding to each initial hidden danger type through the image feature extraction network; Respectively calculating the actual spacing corresponding to the environment image and the spacing deviation value between each sample image and the actual spacing of the sample of the power channel image, and selecting the sample image corresponding to the spacing deviation value lower than the deviation value threshold as the target sample image corresponding to the environment image; The average similarities between the image features of the environment image and the sample image features of each target sample image are calculated respectively by using a similarity recognition algorithm, and the initial hidden danger type with the maximum average similarity is selected as the hidden danger type corresponding to the environment image.

6. The method according to claim 5, characterized in that The identifying the hidden danger probability of each hidden danger type in the power channel area based on the actual distance between each environment image with the same image content in each area image data includes: The actual distances between the environment images with the same image content in each area image data are sorted according to the time sequence corresponding to the time point to which each area image data belongs, so as to obtain the actual distance change information of the same image content; Based on the hidden danger types corresponding to the environmental images of each identical image content, the mapping relationship between the spacing change information corresponding to the hidden danger type of each identical image content and the hidden danger probability is queried in the hidden danger database, and based on the actual spacing change information of each identical image content and the mapping relationship corresponding to each identical image content, the hidden danger probability corresponding to each identical image content is identified; The hidden danger probability of each identical image content is used as the initial hidden danger probability of each hidden danger type of the identical image content, and for each hidden danger type, among the initial hidden danger probabilities of the same hidden danger type, the initial hidden danger probability corresponding to the highest hidden danger probability is selected as the hidden danger probability of the hidden danger type.

7. A device for detecting hidden dangers in the operation and maintenance environment of a power channel based on a hidden danger prediction model, characterized in that: The device comprises: The acquisition module is used to acquire regional image data of the power channel area at various time points through satellite image acquisition software, and for each regional image data, based on the regional image data, identify various environmental images of the regional image data and the power channel image of the regional image data through an image recognition network; An identification module, used to identify the environment type corresponding to each of the environment images and the image features of each of the environment images through an image feature classification model, and to identify the actual distance between each environment image and the power channel image; an analysis module, configured to analyze the types of hidden dangers in the power channel area based on the types of environment, the features of the image, and the actual distances, and to identify the probability of hidden dangers in the power channel area in each hidden danger type based on the actual distances of the environment images with the same image content in the image data of each area; The determination module is used to use each hidden danger type of the power channel area and the hidden danger probability of the power channel area in each hidden danger type as the operation and maintenance environment hidden danger information of the power channel area.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 6 are implemented.

Images