Power transmission line intelligent diagnosis method and system based on image recognition

Abstract

The invention relates to the technical field of power transmission line image recognition, and discloses a power transmission line intelligent diagnosis method and system based on image recognition, and the method comprises the steps: obtaining a topological structure of a power transmission line in a target region, and dividing the grade of the power transmission line according to the topological structure; establishing a first comprehensive model, wherein the first comprehensive model comprises a first model and a plurality of second models; taking the second image as a training set to train a plurality of second models at the same time to obtain a first comprehensive model after training is completed, and recording the first comprehensive model as a second comprehensive model; and performing intelligent real-time diagnosis on the power transmission line according to the second comprehensive model. According to the invention, targeted image acquisition and identification can be carried out according to the topological structure and grade of the power transmission line, and the accuracy and efficiency of diagnosis are effectively improved. And meanwhile, inspection planning is performed by comprehensively considering time and fund constraints, so that the economical efficiency and feasibility of inspection work are ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of transmission line image recognition, and particularly to an intelligent diagnosis method and system for transmission lines based on image recognition. Background Art

[0002] With the continuous development of the power system, the safe operation of transmission lines has become increasingly important. Traditional transmission line diagnosis methods mainly rely on manual inspections, which are not only time-consuming and labor-intensive but also difficult to achieve real-time and comprehensive monitoring. Especially in adverse weather or remote areas, the difficulty and danger of manual inspections will increase significantly. Therefore, it is particularly important to develop an efficient and intelligent transmission line diagnosis method.

[0003] In recent years, with the rapid development of image recognition technology, its application in transmission line monitoring has gradually attracted attention. Image recognition technology can obtain image information of transmission lines through devices such as cameras and use algorithms to analyze and process the images, thereby realizing real-time monitoring and fault diagnosis of the state of transmission lines. This method has the advantages of strong real-time performance, high accuracy, wide application range, etc., providing a new solution for the safe operation of transmission lines.

[0004] However, there are still some problems in the application of existing image recognition technology in transmission line monitoring. For example, how to perform targeted image acquisition and recognition according to the topological structure and level of transmission lines, and how to carry out efficient inspection planning under limited financial and time constraints. Summary of the Invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] Therefore, the present invention provides an intelligent diagnosis method and system for transmission lines based on image recognition, which can solve the problems mentioned in the background art.

[0007] To solve the above technical problems, the present invention provides the following technical solutions: In the first aspect, the present invention provides an intelligent diagnosis method for transmission lines based on image recognition, including: Obtaining the topological structure of transmission lines in the target area and dividing the transmission line levels according to the topological structure; The transmission line topological structure includes a transformer structure, a step-up substation structure, a step-down substation structure, and transmission lines connecting each structure; The transmission line levels include one or several of the following: the first level, the second level, and the third level; Establishing a first comprehensive model, the first comprehensive model including a first model and several second models; The first model is used to obtain a second image; The second model is used to identify several fault states of the transmission line within the target area; All the several second models are connected to the first model; Using the second image as a training set to simultaneously train the several second models, a first integrated model after training is obtained, denoted as the second integrated model; Carry out intelligent real-time diagnosis of the transmission line according to the second integrated model.

[0008] As a preferred solution of the intelligent diagnosis method for transmission lines based on image recognition according to the present invention, wherein: the first model is used to obtain a second image, including: Select a first image acquisition strategy according to the transmission line level, and establish a first model based on the first image acquisition strategy; The first image acquisition strategy includes at least one of the following: a fixed image acquisition strategy and a non-fixed image acquisition strategy; The first image acquisition strategy includes a first objective function, a first period constraint, and a second capital constraint; Using the first objective function as the reward function of the first model, the first period constraint and the second capital constraint as the constraint conditions of the first model, and the output of the first model is the optimal first image acquisition strategy; And perform a first preprocessing on the first image obtained by the first image acquisition strategy to obtain a second image.

[0009] As a preferred solution of the intelligent diagnosis method for transmission lines based on image recognition according to the present invention, wherein: the fixed image acquisition strategy includes: Retrieve the video data, image data, and the positions of the installed image acquisition devices of the transmission line within the target area; Obtain the frame images in the video data and the images in the image data according to a fixed image acquisition frequency; If there is no image acquisition device that can be retrieved for the transmission line within the target area, install image acquisition devices at key nodes, and the key nodes include at least one or more of the following: transformer nodes, step-up substation nodes, step-down substation nodes, middle-section nodes of long-distance transmission lines, and high-risk area nodes; After waiting for the installation to end, retrieve the video data, image data, and the positions of the installed image acquisition devices of the transmission line within the target area; Obtain the frame images in the video data and the images in the image data according to a fixed image acquisition frequency.

[0010] As a preferred solution of the intelligent diagnosis method for transmission lines based on image recognition according to the present invention, wherein: the non-fixed image acquisition strategy includes: Remove the transmission lines within the coverage area of the image acquisition device in the transmission line topology structure in the target area according to the position of the image acquisition device; Perform non-fixed image acquisition inspection planning on the transmission line topology in the remaining area of the transmission line topology structure in the target area; The non-fixed image acquisition inspection planning includes presetting a first objective function and completing the inspection planning in combination with the first cycle constraint and the second capital constraint.

[0011] As a preferred solution of the intelligent diagnosis method for transmission lines based on image recognition according to the present invention, wherein: the first cycle constraint includes: Establish a first mapping table between diagnostic requirements and time; Combined with the user's diagnostic requirements, obtain the time through the first mapping table; According to the transmission line topology in the remaining area and in combination with the time, generate a first cycle constraint; The first cycle constraint includes a first-level line part constraint, a second-level line part constraint, and a third-level line part constraint.

[0012] As a preferred solution of the intelligent diagnosis method for transmission lines based on image recognition according to the present invention, wherein: the second capital constraint includes: Obtain the total first cost of installing image acquisition devices at key nodes; Obtain the total second cost of the non-fixed image acquisition strategy, and the total second cost includes the first cost of the first-level transmission lines, the second cost of the second-level transmission lines, and the third cost of the third-level transmission lines; There is a functional relationship between the second cost and the time; Establish a second capital constraint based on the total first cost and the total second cost.

[0013] As a preferred solution of the intelligent diagnosis method for transmission lines based on image recognition according to the present invention, wherein: the several fault states include at least one or more of the following: broken wire state, insulator damage state, burned state, flashover state, tower tilt or collapse state, foreign object hanging on the wire state, and corrosion and aging state.

[0014] In a second aspect, the present invention provides an intelligent diagnosis system for transmission lines based on image recognition, including: A structure discrimination module, configured to obtain the transmission line topology structure in the target area and divide the transmission line grades according to the topology structure; The transmission line topology structure includes a transformer structure, a step-up substation structure, a step-down substation structure, and transmission lines connecting each structure; The transmission line levels include one or several of the following: the first level, the second level, and the third level; A model establishment module for establishing a first comprehensive model, where the first comprehensive model includes a first model and several second models; The first model is used to obtain a second image; The second models are used to identify several fault states in the transmission lines within the target area; All of the several second models are connected to the first model; A model training module for simultaneously training the several second models using the second image as a training set to obtain the first comprehensive model after training, denoted as the second comprehensive model; A diagnosis module for performing intelligent real-time diagnosis of the transmission lines according to the second comprehensive model.

[0015] In a third aspect, the present invention provides a computer device, including a memory and a processor. The memory stores a computer program, and when the processor executes the computer program, the steps of the method described above are implemented.

[0016] In a fourth aspect, the present invention provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described above are implemented.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention proposes an intelligent diagnosis method for transmission lines based on image recognition, obtains the transmission line topology structure within the target area, and divides the transmission line levels according to the topology structure; establishes a first comprehensive model, where the first comprehensive model includes a first model and several second models; simultaneously trains the several second models using the second image as a training set to obtain the first comprehensive model after training, denoted as the second comprehensive model; performs intelligent real-time diagnosis of the transmission lines according to the second comprehensive model. The present invention can perform targeted image acquisition and recognition according to the topology structure and levels of the transmission lines, effectively improving the accuracy and efficiency of diagnosis. At the same time, by comprehensively considering time and capital constraints for inspection planning, the economy and feasibility of the inspection work are ensured. In addition, the method and system can also monitor various fault states of the transmission lines in real time, providing a strong guarantee for the safe operation of the power system. Description of the Drawings

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0019] Figure 1 It is a method flow chart of an intelligent diagnosis method for transmission lines based on image recognition provided by an embodiment of the present invention; Figure 2 It is a schematic diagram of the topological structure of a transmission line of an intelligent diagnosis method for transmission lines based on image recognition provided by an embodiment of the present invention; Figure 3 It is a schematic diagram of the position of a fixed image acquisition device in the topological structure of a transmission line of an intelligent diagnosis method for transmission lines based on image recognition provided by an embodiment of the present invention; Figure 4 It is a patrol inspection diagram of a non-fixed image acquisition device in the topological structure of a transmission line of an intelligent diagnosis method for transmission lines based on image recognition provided by an embodiment of the present invention; Figure 5 It is a first model framework structure diagram of an intelligent diagnosis method for transmission lines based on image recognition provided by an embodiment of the present invention; Figure 6 It is a schematic diagram of the first comprehensive model structure of an intelligent diagnosis method for transmission lines based on image recognition provided by an embodiment of the present invention; Figure 7 It is an internal structure diagram of a computer device of an intelligent diagnosis method for transmission lines based on image recognition provided by an embodiment of the present invention. Detailed implementation manners

[0020] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will provide a detailed description of the specific implementation manners of the present invention with reference to the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present invention.

[0021] Embodiment 1, referring to Figures 1 - 7 This is the first embodiment of the present invention. This embodiment provides an intelligent diagnosis method for transmission lines based on image recognition, including: In the existing related technologies, although the application of image recognition technology in the monitoring of transmission lines has gradually received attention, it still faces many challenges. For example, how to accurately and efficiently obtain the image data of transmission lines and effectively process and analyze this data to achieve real-time monitoring and fault diagnosis of the status of transmission lines is an urgent problem to be solved currently.

[0022] This application provides a method that can effectively solve the above-mentioned problems. Next, multiple embodiments will be combined to elaborate in detail on how to implement the intelligent diagnosis method of transmission lines based on image recognition. Figure 1 A method flow chart of an intelligent diagnosis method of transmission lines based on image recognition is shown, including: S101, obtain the topological structure of the transmission lines in the target area, and divide the transmission line levels according to the topological structure; In the embodiments of this application, the topological structure of the transmission lines includes a transformer structure, a step-up substation structure, a step-down substation structure, and the transmission lines connecting each structure, as Figure 2 shown.

[0023] It should be noted that if there is no transformer structure, step-up substation structure, or step-down substation structure in the topological structure of the transmission lines, the operation steps of this application do not need to be used for diagnosis, and other existing technologies can be used for diagnosis.

[0024] In the embodiments of this application, the transmission line levels include one or several of the following: the first level, the second level, and the third level.

[0025] It should be noted that the fixed area can be the entire county, a randomly divided area, an urban area, a town, or any other geographically defined area. After obtaining the topological structure of the transmission lines in the target area, the transmission lines can be divided into different levels according to factors such as the importance and voltage level of the transmission lines in the topological structure, such as the first level, the second level, and the third level. This division helps to perform targeted image acquisition and recognition subsequently.

[0026] In an optional embodiment, the first level refers to transmission lines that may be 110 kV or 66 kV at the county level. These lines are mainly used to connect regional substations with county-level substations and supply power to important users.

[0027] The second level generally refers to the voltage range from 35 kV to 10 kV. Such lines are used to distribute power from county-level substations to each township or large industrial users.

[0028] The third level refers to voltages below 1 kilovolt (kV), that is, the commonly mentioned 400 volts (V) / 230 volts (V), which is used to directly supply electricity to residential and small commercial users.

[0029] In the embodiment of this application, as Figure 2 shown, the first-level transmission line is the transmission line between the step-up substation and the primary step-down substation, that is, the position where the red poles are located; the second-level transmission line is the line between the primary step-down substation and the secondary step-down substation, that is, the position where the yellow poles are located; the third-level transmission line is the line between the secondary step-down substation and the transformer, between the transformer and the residence, and between the transformer and the factory, that is, the position where the green poles are located.

[0030] It should be noted that relevant technical personnel can make a more refined level division according to the actual situation. For example, in some specific areas, such as the city center or large industrial areas, there may be higher-level transmission lines. At this time, the transmission line levels can be expanded and refined as needed. This flexible level division method enables the intelligent diagnosis method of the present invention to more accurately adapt to different power network environments and requirements. At the same time, through the detailed division of the transmission line levels, a more accurate basis can also be provided for the subsequent selection of image acquisition strategies and fault diagnosis.

[0031] It should also be noted that the classification of transmission lines is to diagnose different transmission lines and electrical equipment around the transmission lines at different levels. Because the importance and fault impact range of different transmission line levels are different, classifying and diagnosing them can more reasonably allocate resources and improve the diagnosis efficiency. For example, for the first-level transmission line, since it connects large substations and important power users, once a fault occurs, its impact range will be very wide. Therefore, the diagnosis of it needs to be more strict and timely. For the third-level transmission line, since it connects residential and small commercial users, the fault impact range is relatively small. Therefore, the diagnosis frequency can be appropriately reduced to save resources.

[0032] S102. Establish a first comprehensive model, where the first comprehensive model includes a first model and several second models; In the embodiment of this application, the first model is used to obtain a second image; In the embodiment of this application, the first model is used to obtain a second image including: Step 1: Select a first image acquisition strategy according to the transmission line level, and establish a first model based on the first image acquisition strategy; It should be noted that the first image acquisition strategy includes at least one of the following: a fixed image acquisition strategy and a non-fixed image acquisition strategy. The fixed image acquisition strategy is a strategy for images that can be acquired by fixed image acquisition devices. For example, pictures or videos regularly taken by cameras. The non-fixed image acquisition strategy is to temporarily add image acquisition devices according to actual needs, such as drone patrol or shooting with a portable camera, etc.

[0033] It should be noted that when selecting the first image acquisition strategy, the system will comprehensively consider various factors such as the level of the transmission line, historical fault data, weather conditions, geographical location, etc. For example, for transmission lines with a higher level, frequent failures, or in a harsh environment, the system may be more inclined to adopt a non-fixed image acquisition strategy to ensure more comprehensive and accurate image information can be obtained.

[0034] It should be noted that different levels of transmission lines correspond to different first image acquisition strategies. Generally, different levels of transmission lines all include a part of the fixed image acquisition strategy and the non-fixed image acquisition strategy. The specific situation needs to be based on the actual on-site situation. For example, the first-level transmission line may adopt a non-fixed image acquisition strategy with a higher frequency, such as drone patrol, to ensure strict monitoring of key lines. While the third-level transmission line may rely more on the fixed image acquisition strategy, such as regularly taken monitoring videos, to reduce costs while ensuring a certain monitoring accuracy.

[0035] It should also be noted that after determining the first image acquisition strategy, the system will further establish a first model based on this strategy. This model is responsible for automatically or manually triggering the image acquisition device according to the selected strategy to capture the image data of the transmission line. For example, under the fixed image acquisition strategy, the model may set the shooting time and interval of the camera; while under the non-fixed image acquisition strategy, the model may dispatch drones or portable cameras for patrol.

[0036] In the embodiment of the present application, the fixed image acquisition strategy includes: Retrieving the video data, image data, and the positions of the installed image acquisition devices of the transmission line in the target area, as Figure 3 shown; Obtaining the frame images in the video data and the images in the image data according to the fixed image acquisition frequency; If there is no retrievable image acquisition device for the transmission line in the target area, install image acquisition devices at key nodes. The key nodes include at least one or more of the following: transformer nodes, step-up substation nodes, step-down substation nodes, middle-section nodes of long-distance transmission lines, and high-risk area nodes; After waiting for the installation to end, retrieve the video data, image data of the installed image acquisition devices on the transmission lines in the target area, and the positions of the installed image acquisition devices. Obtain the frame images in the video data and the images in the image data according to a fixed image acquisition frequency.

[0037] It should be noted that the mid-section nodes of long-distance transmission lines refer to the middle-position nodes of relatively long transmission lines in this application. According to the experimental data in the specific implementation process of this application, for transmission lines with a distance exceeding 50 kilometers, the probability of faults occurring at their mid-section nodes is relatively high. Therefore, it is necessary to install image acquisition devices at these positions to improve the accuracy and efficiency of monitoring. Since these positions are far from the substations at both ends and are often in relatively remote geographical environments, they are vulnerable to various factors such as weather and external damage, so they are high-risk areas for faults. By installing image acquisition devices at these key nodes, comprehensive monitoring of the transmission lines can be achieved, and the accuracy and timeliness of fault diagnosis can be improved.

[0038] It should be noted that the high-risk area nodes in this application refer to the nodes corresponding to the fault-prone areas or high-risk areas comprehensively evaluated according to various factors such as historical fault data, geological conditions, and meteorological conditions. For example, in areas with complex geological conditions where natural disasters such as landslides or debris flows are likely to occur, or in areas with harsh meteorological conditions and frequent lightning activities, the system may mark the transmission line nodes corresponding to these areas as high-risk area nodes and preferentially install image acquisition devices at these positions to strengthen the monitoring and protection of the transmission lines.

[0039] In the embodiment of this application, the non-fixed image acquisition strategy includes: Remove the transmission lines within the coverage area of the image acquisition device from the transmission line topology in the target area according to the position of the image acquisition device; Perform non-fixed image acquisition inspection planning on the remaining transmission line topology in the target area, as Figure 4 shown.

[0040] It should be noted that in practice, the detection range of image acquisition devices is limited. Therefore, when implementing the non-fixed image acquisition strategy, it is necessary to consider the detection ranges of the existing fixed image acquisition devices or the fixed image acquisition devices installed using the method of this application, as follows: Step ①: Determine whether there are fixed image acquisition devices in the target area; If not, install fixed image acquisition devices. Since they are installed in advance, the parameters of the fixed image acquisition devices are already known, so proceed to step ②; If it exists, obtain the image data or video frame data of all existing fixed image acquisition devices, and determine the effective recognition range of these image data or video frame data. For example, the image data or video frame data can be processed through an image recognition algorithm, and the specific position and range of the transmission line in the image can be recognized according to the functions to be implemented at specific positions, so as to determine the effective recognition range of the fixed image acquisition device, mark this effective recognition range, and enter step ②; Specifically, if there are states such as broken wire state, insulator damage state, burning state, flashover state, tower inclination or collapse state, foreign object hanging on the line state, and corrosion and aging state that need to be recognized in this application, then when processing the image data or video frame data through an image recognition algorithm, it is necessary to recognize according to the image features corresponding to these specific states. For example, the broken wire state may correspond to the interruption of the continuity of the transmission line in the image, so the detection accuracy requirement is slightly lower and the effective recognition range is higher; the insulator damage state may correspond to abnormal image features of the insulator part, so the detection range requirement is slightly higher and the effective recognition range is shorter. The burning state may be manifested as abnormal color or morphological changes of the transmission line or related equipment in the image. At this time, the detection accuracy requirement is relatively high, and the effective recognition range needs to be flexibly adjusted according to the specific degree of burning. The flashover state may correspond to abnormal discharge phenomena around the transmission line in the image, which requires the image recognition algorithm to accurately capture the instantaneous light intensity change. The effective recognition range is usually small and concentrated near the discharge point. The tower inclination or collapse state can be judged by the straightness or position change of the tower in the image. The detection accuracy requirement is moderate, and the effective recognition range needs to cover the entire tower area. The foreign object hanging on the line state may be manifested as an object that should not exist on the transmission line in the image. At this time, the detection accuracy requirement is relatively high, and the effective recognition range needs to be flexibly set according to the size and position of the foreign object. The corrosion and aging state may correspond to the gradual changes in the surface texture, color, etc. of the transmission line or equipment in the image. This requires the image recognition algorithm to be able to track and analyze these changes for a long time. The effective recognition range is usually large to detect potential hidden dangers in time.

[0041] Step ②: Remove the transmission lines within the coverage area of the image acquisition device in the transmission line topology structure of the target area, that is, the effective recognition range, such as Figure 4 the blue pentagon range shown; enter step ③.

[0042] Step ③: Plan the inspection route for non-fixed image acquisition of the transmission line topology in the remaining area within the target area. In this application, to meet the first cycle constraint and the second capital constraint, and also for the rationality and convenience of route planning, when using a drone or other equipment for non-fixed image acquisition inspection, the inspection route can pass through the effective recognition range of the fixed image acquisition device. However, when entering this range, the drone or other inspection equipment will no longer perform other tasks except flying.

[0043] Step ④: Implement non-fixed image acquisition according to the completed route plan.

[0044] In the embodiment of this application, the first image acquisition strategy includes a first objective function, a first cycle constraint, and a second capital constraint. That is to say, both the fixed image acquisition strategy and the non-fixed image acquisition strategy also include their respective first objective functions, first cycle constraints, and second capital constraints.

[0045] Step 2: Use the first objective function as the reward function of the first model, and the first cycle constraint and the second capital constraint as the constraint conditions of the first model. The output of the first model is the optimal first image acquisition strategy. In the embodiment of this application, a first model is established using the reinforcement learning method. Different from the existing reinforcement learning methods, this application uses the first objective functions of two different strategies as the reward function of reinforcement learning, and uses the common first cycle constraint and the second capital constraint as the constraint conditions of the first model in reinforcement learning. The agent in reinforcement learning is the output of the first model. The optimal first model output, that is, the optimal first image acquisition strategy, is obtained through reinforcement learning. Among them, the optimal first image acquisition strategy includes the optimal fixed image acquisition strategy and the optimal non-image acquisition strategy.

[0046] Specifically, the first objective function in the fixed image acquisition strategy in this application is as follows: First, it is necessary to determine the main objective of the fixed image acquisition strategy, that is, to maximize the acquisition efficiency and accuracy of image data. This objective can be achieved through the following sub-objectives: one is to ensure the clarity of the image data for subsequent image recognition and processing; the second is to ensure the timeliness of the image data, that is, to be able to reflect the state of the transmission line in real time or near real time; the third is the influence of the layout of the image acquisition device. To achieve these sub-objectives, this application designs a comprehensive reward function that considers multiple factors such as the clarity, timeliness, and device layout of the image data, and uses a reinforcement learning algorithm to find the optimal image acquisition strategy.

[0047] The specific reward function is expressed as follows: ; Among them, is the first objective function of the fixed image acquisition strategy, represents the clarity score of the fixed image acquisition device, represents clarity, represents the delay of fixed image acquisition, represents the ratio of the coverage area of the actual fixed image acquisition device to the coverage area of the transmission line to be detected in the entire area, represents the ratio of the predicted coverage area of the fixed image acquisition device to the coverage area of the transmission line to be detected in the entire area, represents the coverage area of the transmission line to be detected in the entire area, represents the actual coverage area of the fixed image acquisition device, represents the predicted coverage area of the fixed image acquisition device, , , represent weight coefficients, and their sum is 1.

[0048] The first objective function in the non-fixed image acquisition strategy of this application is as follows: First, it is necessary to determine the main objectives of the non-fixed image acquisition strategy, namely, to maximize the inspection efficiency, accuracy, and comprehensiveness of coverage. This objective can also be achieved through multiple sub-objectives: one is to ensure the comprehensiveness of inspection coverage without missing any potential fault points; the second is delay; the third is to ensure the clarity of inspection images for subsequent analysis. To achieve these sub-objectives, this application has also designed a comprehensive reward function that comprehensively considers multiple factors such as inspection coverage rate, inspection route optimization, image clarity, and cost, and uses reinforcement learning algorithms to find the optimal non-fixed image acquisition strategy.

[0049] The specific reward function is expressed as follows: ; Among them, represents the first objective function of the non-fixed image acquisition strategy, represents the ratio of the actual coverage area of the non-fixed image acquisition device to the coverage area of the transmission line to be detected in the entire area, represents the delay of non-fixed image acquisition, represents the clarity score of the non-fixed image acquisition device, represents the ratio of the predicted coverage area of the non-fixed image acquisition device to the coverage area of the transmission line to be detected in the entire area, Actual coverage area of the non-fixed image acquisition device, represents the predicted coverage area of the non-fixed image acquisition device, , , Denotes the weight coefficient, and the sum is 1.

[0050] In the embodiment of the present application, the first cycle constraint includes: Establish a first mapping table between diagnostic requirements and time; Combined with the user's diagnostic requirements, obtain the time through the first mapping table; According to the transmission line topology in the remaining area, combined with the time, generate the first cycle constraint; In the embodiment of the present application, the first cycle constraint includes the first-level line section constraint, the second-level line section constraint, and the third-level line section constraint.

[0051] In the embodiment of the present application, the first mapping table is shown as follows: Transmission line level Diagnostic requirement description Inspection time Remarks Constraints for the first - level line department High - risk areas, key power transmission paths 7 days Including emergency inspections after severe weather 1 day Conduct more frequent inspections for areas vulnerable to natural disasters Constraints for the second - level line department Medium - risk, connecting important but non - critical nodes 14 days Adjust inspection frequency according to seasonal changes 7 days Increase inspection frequency during thunderstorm seasons or windy seasons Constraints for the third - level line department Low - risk, supplying power to residential or small - business users 30 days Mainly rely on fixed image acquisition devices 15 days If there is no abnormality, maintain the regular inspection frequency; if there are abnormalities, change the inspection frequency It should be noted that the first-level transmission lines usually involve high voltage levels (such as 110 kV or 66 kV), which are the main roads connecting large substations and important users. Due to their importance, more stringent monitoring is required. The second-level transmission lines are generally medium voltage levels (such as 35 kV to 10 kV), which are used to distribute power from county-level substations to townships or large industrial users. The importance of such lines is secondary to that of the first level, but regular maintenance is still required. The third-level transmission lines refer to low voltage levels (such as below 1 kV), which are mainly used to directly supply power to residential areas and small commercial users. Since the impact of faults on the overall power grid is relatively small, the inspection cycle is relatively long.

[0052] The first cycle constraint in the present application is specifically as follows: First-level line section constraint: For high-risk areas and key power transmission paths, it is usually set to be inspected at least once within 7 days; for areas vulnerable to natural disasters, it may be shortened to once a day.

[0053] Second-level line section constraint: For transmission lines with medium risk and connecting important but non-critical nodes, it is conventionally set to be inspected once every 14 days; but in specific seasons such as thunderstorms or windy seasons, it may be increased to once every 7 days.

[0054] Third-level line section constraint: For low-risk power supply lines mainly serving residential or small commercial users, it is generally inspected once every 30 days; if there is no abnormality, maintain the regular frequency, and adjust the inspection cycle in case of abnormality.

[0055] Combined with the user's diagnostic requirements: By establishing a first mapping table between diagnostic requirements and time, combined with the actual needs of the user (such as emergency inspection requirements), the inspection time and frequency can be dynamically adjusted. For example, arrange an emergency inspection immediately after bad weather to ensure the safe operation of power facilities.

[0056] Generate cycle constraints for the transmission line topology in the remaining area in combination with time: Considering the actual situation of the remaining uncovered transmission line area and combining the time factor mentioned above, reasonably arrange the inspection plan to ensure that all key points can be properly monitored.

[0057] In the embodiment of the present application, the second capital constraint includes: Obtain the total first cost of installing image acquisition devices at key nodes; Obtain the total second cost of the non-fixed image acquisition strategy, where the total second cost includes the first cost of the first-level transmission lines, the second cost of the second-level transmission lines, and the third cost of the third-level transmission lines; In the embodiment of the present application, the second cost has a functional relationship with the time; Establish a second capital constraint based on the total first cost and the total second cost.

[0058] The second capital constraint in the present application is specifically as follows: First of all, it should be considered that the first cost, the second cost, and the third cost in the present application all include the later maintenance costs of the fixed image acquisition devices and the non-fixed image acquisition devices, the inspection cost of the non-fixed image acquisition devices, the server usage cost, and the possible labor cost; Specifically expressed as follows: ; Among them, represents the total cost, represents the total first cost, represents the total second cost, represents the installation cost of the i-th fixed image acquisition device, and n represents the number of installed fixed image acquisition devices, represents the maintenance cost of the i-th fixed image acquisition device, represents the maintenance cost of the non-fixed image acquisition devices, m represents the total inspection distance of the non-fixed image acquisition devices, in units of ten meters, represents the inspection cost of the j-th meter, represents the server usage cost, represents the labor cost, represents the total investment in user diagnosis requirements.

[0059] It should be noted that as Figure 5As shown, in the reinforcement learning framework, the agent (i.e., the output of the first model) makes decisions based on the current transmission line status, the status of the image acquisition device, and environmental information (such as weather conditions, geographical locations, etc.), that is, selects the optimal image acquisition strategy. This decision-making process is restricted by the first cycle constraint and the second capital constraint, that is, the image acquisition task needs to be completed within the specified time cycle, and equipment deployment and maintenance need to be carried out within the budget. The agent will gradually find the optimal strategy that meets these constraint conditions through continuous trial and error and learning. The agent selects an action based on the current state, and then obtains a new state and the corresponding reward from the environment. This process is iterated until the optimal strategy is reached. The reward function and the strategy jointly determine the behavior of the agent, while the first cycle constraint and the second capital constraint ensure the feasibility and economy of the system.

[0060] In the embodiment of the present application, the non-fixed image acquisition inspection plan includes presetting a first objective function and completing the inspection plan in combination with the first cycle constraint and the second capital constraint.

[0061] It should be noted that after obtaining the optimal first image acquisition strategy, the second image acquisition is performed. That is, the inspection images are acquired according to the optimal non-fixed image acquisition strategy. At this time, a drone or other inspection equipment can be used to inspect the transmission lines in the target area according to the pre-planned inspection route and obtain the corresponding inspection image data. These inspection image data will be used for subsequent transmission line status recognition and diagnosis.

[0062] Step 3: Perform first preprocessing on the first image obtained by the first image acquisition strategy to obtain a second image.

[0063] Since there are seven different second models in the subsequent part of this application, when performing the first preprocessing, the images obtained by the optimal first image acquisition strategy need to be classified into seven, respectively corresponding to the input formats and requirements required by the seven different second models. The classification basis can be factors such as the fault type of the transmission line in the image, the type of the image acquisition device, the inspection time, etc. The specific classification methods and criteria will be described in detail in the subsequent construction of the second model. After classification, targeted preprocessing operations are performed on each type of image, including denoising, enhancement, cropping, etc., to improve the quality of the image and the efficiency of subsequent processing. After obtaining the preprocessed second image, the subsequent transmission line status recognition and diagnosis process can be entered.

[0064] In the embodiments of the present application, the second model is used to identify several fault states in the transmission line within the target area, and the several fault states include at least one or more of the following: broken wire state, insulator damage state, burned state, flashover state, tower tilt or collapse state, foreign object hanging on the wire state, and corrosion and aging state.

[0065] In the embodiments of the present application, the several second models are all connected to the first model, as Figure 6 shown in the schematic diagram of the first comprehensive model of the present application. After the image data generated by the first model undergoes the first preprocessing, it is sent to each fault state recognition model in the second model. Each second model respectively identifies specific types of fault states and outputs corresponding diagnostic results. These diagnostic results are processed by the normalization module to ensure the consistency and comparability of the output results. Finally, the output module provides a comprehensive diagnostic result.

[0066] Among them, the first model: is responsible for the acquisition and preliminary processing of images.

[0067] The first preprocessing: preprocesses the original image to generate a second image suitable for analysis.

[0068] The second model: includes multiple models specifically for identifying different fault states.

[0069] The normalization module: normalizes the outputs of each second model to ensure the consistency and comparability of the output results.

[0070] Output: provides a comprehensive diagnostic result.

[0071] It should be noted that by establishing the first comprehensive model, which includes a first model and several second models, a comprehensive and intelligent diagnosis of the transmission line status can be achieved. First, through the first model, combined with the reinforcement learning method, multiple factors such as the efficiency, accuracy, cost, and time cycle of image acquisition can be comprehensively considered, and an optimal image acquisition strategy can be automatically generated, including a fixed image acquisition strategy and a non-fixed image acquisition strategy. This not only improves the acquisition efficiency and accuracy of image data, but also ensures the timeliness and comprehensiveness of the inspection, while controlling the cost.

[0072] Then, using a drone or other inspection equipment, according to the pre-planned inspection route, the transmission line within the target area is inspected, and the corresponding inspection image data is acquired according to the optimal non-fixed image acquisition strategy. These image data are preprocessed and then sent to different second models for fault state identification.

[0073] Since the second model has been specifically trained and optimized for specific fault states, it can accurately identify fault states such as broken wires, damaged insulators, burnout, flashover, pole tilt or collapse, foreign objects hanging on the line, and corrosion and aging in the transmission line. This greatly improves the accuracy and efficiency of fault identification.

[0074] Finally, by integrating the output results of the first model and several second models, a comprehensive, accurate, and timely intelligent diagnosis of the transmission line status can be achieved. This not only helps to promptly detect and handle potential safety hazards, improve the operation safety and stability of the power grid, but also provides strong data support for the maintenance and management of the transmission line, reduces the operation and maintenance costs, and improves the operation and maintenance efficiency.

[0075] S103, using the second image as a training set to simultaneously train several second models, obtaining the first integrated model after training, denoted as the second integrated model; Specifically, when training the second integrated model, the present application first needs to input the preprocessed second image as a training set into each second model. These second models respectively identify different transmission line fault states, such as broken wire state, damaged insulator state, burnout state, etc. By simultaneously training these models, the present application can make full use of the information in the training data, improve the generalization ability and recognition accuracy of the models.

[0076] During the training process, the present application adopts the method of supervised learning, that is, using training data with known labels to guide the training of the models. The labeled data can come from historical inspection records, manual annotation, or other reliable sources. By comparing the prediction results of the models with the true labels, the present application can calculate the loss function and update the parameters of the models through the backpropagation algorithm to minimize the loss function.

[0077] After the training is completed, the present application will obtain the first integrated model after training, that is, the second integrated model. This model can simultaneously identify multiple transmission line fault states and has high accuracy and robustness. In practical applications, the present application can input the inspection image data into the second integrated model to obtain the recognition results of the fault states, thereby providing strong support for subsequent repair and maintenance work.

[0078] In addition, it is worth noting that when training the second comprehensive model, the present application can also consider adopting advanced technologies such as transfer learning and ensemble learning to further improve the performance and stability of the model. For example, the present application can use a model that has been trained in other related fields as a pre-trained model, and accelerate the training process of the second comprehensive model and improve the recognition accuracy through transfer learning; or the present application can integrate the recognition results of multiple second models, and improve the reliability and stability of the final recognition result through methods such as voting and weighted averaging. The specific application of these methods will depend on the actual application scenarios and requirements.

[0079] S104, perform intelligent real-time diagnosis of the transmission line according to the second comprehensive model.

[0080] In the embodiment of the present application, the preprocessed inspection image data is input into the trained second comprehensive model, and the model will perform intelligent recognition and diagnosis on the state of the transmission line in the image. The model can automatically detect various possible fault states in the transmission line, such as broken wires, damaged insulators, burnout, flashover, pole tilt or collapse, foreign objects hanging on the line, and corrosion and aging. By real-time analyzing the inspection image data, the method of the present application can quickly locate the fault point and provide timely and accurate information support for subsequent maintenance and repair work.

[0081] In the intelligent diagnosis process, the second comprehensive model will make full use of its powerful feature extraction and classification capabilities to deeply analyze the input image. The model will first extract features from the image and identify key information, such as the shape, color, and texture of the transmission line. Then, the model will use this feature information and combine it with the pre-learned fault state knowledge base to classify and judge the state of the transmission line. Finally, the model will output the recognition result of the fault state, including information such as the fault type, location, and possible severity.

[0082] It is worth noting that the method of the present application also fully considers the balance between real-time and accuracy in the intelligent diagnosis process. By optimizing the image acquisition strategy, preprocessing process, and model training process, the method of the present application can achieve fast response and real-time diagnosis while ensuring the diagnosis accuracy. This is of great significance for ensuring the safe and stable operation of the transmission line. Especially in bad weather or emergency situations, it can quickly detect and handle faults, reducing power outage time and economic losses.

[0083] In summary, the present invention proposes an intelligent diagnosis method for transmission lines based on image recognition, which obtains the topological structure of the transmission lines in the target area and divides the transmission line levels according to the topological structure; establishes a first comprehensive model, which includes a first model and several second models; uses the second image as a training set to train several second models simultaneously to obtain the first comprehensive model after training, denoted as the second comprehensive model; and performs intelligent real-time diagnosis of the transmission lines according to the second comprehensive model. The present invention can perform targeted image acquisition and recognition according to the topological structure and levels of the transmission lines, effectively improving the accuracy and efficiency of diagnosis. At the same time, by comprehensively considering time and capital constraints for inspection planning, the economy and feasibility of the inspection work are ensured. In addition, the method and system can also monitor various fault states of the transmission lines in real time, providing a strong guarantee for the safe operation of the power system.

[0084] Example 2, in a preferred embodiment, when there are no fixed image acquisition devices installed in the area to be measured initially. According to the proposed method, it is first necessary to analyze this specific area (as Figure 2 shown) and determine the key positions to install fixed image acquisition devices. These positions may include transformer nodes, step-up substation nodes, step-down substation nodes, mid-section nodes of long-distance transmission lines, and high-risk area nodes, etc. Once these key positions are determined, install fixed image acquisition devices at these positions ( Figure 3 the blue pentagons in the figure represent the installed devices).

[0085] Next, based on the topological structure of the transmission lines in the remaining uncovered area, plan the non-fixed image acquisition inspection route (as shown by the red arrows in Figure 4 the figure). This step takes into account the first cycle constraint and the second capital constraint, ensuring the economy and feasibility of the inspection work while also ensuring that the inspection route can cover all key transmission lines to achieve a non-missing inspection.

[0086] During the actual inspection process, use non-fixed image acquisition devices (such as cameras carried by drones) to perform tasks according to the pre-set strategy. This not only covers the pre-determined inspection path but also should be flexibly adjusted according to the real-time weather conditions, light conditions, and the specific conditions of the transmission lines to ensure that the quality of the collected image data is high enough to provide reliable data support for subsequent intelligent diagnosis. In this way, the accuracy and efficiency of transmission line fault detection can be effectively improved, thus ensuring the safe and stable operation of the power system.

[0087] It should be noted that according to the actual implementation results, it can be shown that by installing fixed image acquisition devices and planning non-fixed image inspection routes, all key transmission lines are ensured to be covered. This enables the system to timely detect and locate various types of faults (such as broken wires, insulator damage, etc.), thereby improving the accuracy and efficiency of fault detection.

[0088] During the inspection process, by adjusting the image acquisition strategy considering weather conditions, light conditions, and the specific conditions of the transmission lines, the quality of the collected data can be ensured, and high reliability can be maintained even in harsh environments.

[0089] According to the classification of transmission line levels (first, second, and third levels), different inspection frequencies and strategies are adopted to achieve effective allocation of resources. For example, high-risk areas or key power transmission paths are inspected more frequently, while the inspection frequency for low-risk areas can be appropriately reduced.

[0090] Considering time and cost constraints comprehensively for inspection planning ensures the economy and feasibility of the inspection work. By reasonably arranging the use of fixed and non-fixed image acquisition devices, comprehensive monitoring can be achieved while controlling costs.

[0091] Inputting the preprocessed inspection image data into the trained model can analyze the state of the transmission line in real time, quickly locate the fault point, and provide timely and accurate information support for subsequent maintenance work.

[0092] Embodiment 3, in this embodiment, an intelligent diagnosis system for transmission lines based on image recognition is further provided, including: A structure discrimination module, configured to obtain the topological structure of the transmission lines in the target area and divide the transmission line levels according to the topological structure; The topological structure of the transmission lines includes a transformer structure, a step-up substation structure, a step-down substation structure, and the transmission lines connecting each structure; The transmission line levels include one or several of the following: first level, second level, and third level; A model establishment module, configured to establish a first comprehensive model, and the first comprehensive model includes a first model and several second models; The first model is used to obtain a second image; The second models are used to identify several fault states in the transmission lines in the target area; All the several second models are connected to the first model; A model training module, configured to use the second image as a training set to train the several second models simultaneously, and obtain the first comprehensive model after training, denoted as the second comprehensive model; A diagnosis module for performing intelligent real-time diagnosis of transmission lines according to the second comprehensive model.

[0093] Each of the above unit modules can be embedded in a processor in a computer device in hardware form or independent of the processor, or stored in a memory in the computer device in software form, so that the processor can call and execute the operations corresponding to each of the above modules.

[0094] This embodiment also provides a computer device, which can be a terminal, and its internal structure diagram can be as Figure 7 shown. The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. 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 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 carrier network, NFC (Near Field Communication) or other technologies. When the computer program is executed by the processor, it implements an intelligent diagnosis method for transmission lines based on image recognition. The display screen of the computer device 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, a trackball or a touchpad provided on the shell of the computer device, or an external keyboard, a touchpad or a mouse, etc.

[0095] This embodiment also provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented: Obtain the topological structure of the transmission lines in the target area, and divide the transmission line levels according to the topological structure; The topological structure of the transmission lines includes a transformer structure, a step-up substation structure, a step-down substation structure, and transmission lines connecting each structure; The transmission line levels include one or several of the following: the first level, the second level, and the third level; Establish a first comprehensive model, which includes a first model and several second models; The first model is used to obtain a second image; The second model is used to identify several fault states in the transmission lines in the target area; All of the several second models are connected to the first model; Use the second image as a training set to train several second models simultaneously to obtain the first comprehensive model after training, denoted as the second comprehensive model; Perform intelligent real-time diagnosis of the power transmission line according to the second comprehensive model.

[0096] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

[0097] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present application can be implemented in various computer languages.

[0098] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the specified functions in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0099] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory generate a manufactured article including instruction means, and the instruction means implements the specified functions in Figure 1 one process or multiple processes and / or blocks Figure 1 one block or multiple blocks.

[0100] These computer program instructions can also be loaded onto a computer or other programmable data processing device, so that a series of operation steps are performed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide means for implementing the specified functions in Figure 1 one process or multiple processes and / or blocksFigure 1 Steps of the functions specified in one or more boxes.

[0101] Although the preferred embodiments of the present application have been described, those skilled in the art can make additional changes and modifications to these embodiments once they learn the basic creative concepts. Therefore, the appended claims are intended to be construed as including the preferred embodiments and all changes and modifications falling within the scope of the present application.

[0102] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application is also intended to include these modifications and variations.

Claims

1. An intelligent diagnosis method for transmission lines based on image recognition, characterized in that, Including: Obtain the topological structure of the transmission lines within the target area, and divide the transmission line levels according to the topological structure; The topological structure of the transmission lines includes a transformer structure, a step-up substation structure, a step-down substation structure, and the transmission lines connecting each structure; The transmission line levels include one or several of the following: the first level, the second level, and the third level; Establish a first comprehensive model, which includes a first model and several second models; The first model is used to obtain a second image; The second models are used to identify several fault states in the transmission lines within the target area; All of the several second models are connected to the first model; Use the second image as a training set to simultaneously train the several second models, and obtain the first comprehensive model after training, denoted as the second comprehensive model; Conduct intelligent real-time diagnosis of the transmission lines according to the second comprehensive model.

2. The intelligent diagnosis method for transmission lines based on image recognition according to claim 1, wherein The first model is used to obtain a second image, including: Select a first image acquisition strategy according to the transmission line levels, and establish a first model based on the first image acquisition strategy; The first image acquisition strategy includes at least one of the following: a fixed image acquisition strategy and a non-fixed image acquisition strategy; The first image acquisition strategy includes a first objective function, a first period constraint, and a second capital constraint; Use the first objective function as the reward function of the first model, the first period constraint and the second capital constraint as the constraint conditions of the first model, and the output of the first model is the optimal first image acquisition strategy; And perform a first preprocessing on the first image obtained by the first image acquisition strategy to obtain a second image.

3. The intelligent diagnosis method for transmission lines based on image recognition according to claim 2, wherein The fixed image acquisition strategy includes: Retrieve the video data, image data of the image acquisition devices already installed on the transmission lines within the target area, and the locations of the already installed image acquisition devices; Obtain the frame images in the video data and the images in the image data according to a fixed image acquisition frequency; If there are no retrievable image acquisition devices on the transmission lines within the target area, install image acquisition devices at key nodes, and the key nodes include at least one or more of the following: transformer nodes, step-up substation nodes, step-down substation nodes, mid-section nodes of long-distance transmission lines, and high-risk area nodes; After waiting for the installation to end, retrieve the video data, image data of the image acquisition devices already installed on the transmission lines within the target area, and the locations of the already installed image acquisition devices; Obtain the frame images in the video data and the images in the image data according to a fixed image acquisition frequency.

4. The intelligent diagnosis method for transmission lines based on image recognition according to claim 3, characterized in that, The non-fixed image acquisition strategy includes: Remove the transmission lines within the coverage area of the image acquisition devices from the topological structure of the transmission lines within the target area according to the locations of the image acquisition devices; Conduct a non-fixed image acquisition inspection plan for the remaining transmission line topology within the target area; The non-fixed image acquisition inspection plan includes a preset first objective function, and completes the inspection plan in combination with the first period constraint and the second capital constraint.

5. The intelligent diagnosis method for transmission lines based on image recognition according to claim 4, characterized in that, The first period constraint includes: Establish a first mapping table between diagnostic requirements and time; Obtain time through the first mapping table in combination with the user's diagnosis requirements; Generate first-cycle constraints according to the power transmission line topology within the remaining area in combination with the time; The first-cycle constraints include first-level line part constraints, second-level line part constraints, and third-level line part constraints.

6. The intelligent diagnosis method for transmission lines based on image recognition according to claim 5, wherein The second capital constraint includes: Obtain the total first cost of installing image acquisition devices at key nodes; Obtain the total second cost of the non-fixed image acquisition strategy, where the total second cost includes the first cost amount of the first-level power transmission line, the second cost amount of the second-level power transmission line, and the third cost amount of the third-level power transmission line; The second cost amount has a functional relationship with the time; Establish a second capital constraint based on the total first cost and the total second cost.

7. The intelligent diagnosis method for transmission lines based on image recognition according to claim 6, characterized in that, The several fault states include at least one or more of the following: broken wire state, insulator damage state, burned state, flashover state, pole tilt or collapse state, foreign object hanging on the wire state, and corrosion and aging state.

8. An intelligent diagnosis system for transmission lines based on image recognition, which applies the method according to any one of claims 1 to 7, is characterized in that Include: A structure differentiation module, configured to obtain the power transmission line topology structure within the target area and divide the power transmission line levels according to the topology structure; The power transmission line topology structure includes a transformer structure, a step-up substation structure, a step-down substation structure, and power transmission lines connecting the structures; The power transmission line levels include one or several of the following: first level, second level, and third level; A model establishment module, configured to establish a first comprehensive model, where the first comprehensive model includes a first model and several second models; The first model is used to obtain a second image; The second model is used to identify several fault states in the power transmission lines within the target area; The several second models are all connected to the first model; A model training module, configured to use the second image as a training set to simultaneously train several second models to obtain the first comprehensive model after training, denoted as the second comprehensive model; A diagnosis module, configured to perform intelligent real-time diagnosis of power transmission lines according to the second comprehensive model.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method described in any one of claims 1 to 7.

Images