Unmanned hot-line work control system based on autonomous decision algorithm

Abstract

An unmanned hot-line work control system based on an autonomous decision algorithm performs flight inspection on a power transmission line through an unmanned aerial vehicle according to a predetermined path, and analyzes the state of the power transmission line in real time by using an image processing technology; and then, a multi-stage feature extraction method is adopted, that is, shallow features are extracted from an original image, deep semantic features are further mined, and finally, the information is integrated through a feature joint perception module to generate a final decision result. Besides, a feedback mechanism is further designed, when suspected abnormity is detected, the unmanned aerial vehicle can automatically adjust the position and hover above a target position, so that more detailed high-definition images are collected, and the data are transmitted back to a ground server for further analysis. Therefore, the fault diagnosis speed is improved, the possibility of false alarm is reduced, and the reliability and practicability of the whole system are improved.

Description

Technical Field

[0001] The present application relates to the field of intelligent control technology, and more specifically, to an unmanned live-working control system based on an autonomous decision-making algorithm. Background Art

[0002] In the power transmission system, the maintenance and inspection of high-voltage transmission lines is a key link to ensure a stable power supply. Traditional live-line work usually requires manual participation, which is not only inefficient but also has high safety risks. Especially when dealing with high-voltage lines, workers may face dangers such as electric shock. This method is difficult to meet the needs of rapid response of modern power grids. With the development of technology, unmanned equipment such as drones and robots have been introduced into the power industry to achieve more efficient and safer live-line work.

[0003] Unmanned aerial vehicles (UAVs) and crawling robots can carry various sensors, such as high-definition cameras, infrared thermal imagers, and laser radars (LiDAR), and can replace manual work on high-voltage lines without direct human intervention, such as inspections, inspections, fault repairs, or maintenance. These unmanned devices can work in complex terrain and severe weather conditions, greatly improving work efficiency and safety.

[0004] However, traditional unmanned live-line operation control solutions usually rely on pre-programmed paths or remote operator instructions, lacking real-time adaptability and autonomous decision-making capabilities. This means that they cannot effectively respond or adjust strategies when encountering unforeseen situations. In addition, traditional image processing technology and sensor data analysis methods may not provide sufficient resolution and accuracy to identify subtle line damage or other problems, which may lead to missed faults or false alarms.

[0005] Therefore, an optimized unmanned live working control system is desired. Summary of the invention

[0006] This summary is provided to introduce concepts in a brief form that will be described in detail in the detailed description below. This summary is not intended to identify key features or essential features of the claimed technical solution, nor is it intended to limit the scope of the claimed technical solution.

[0007] In a first aspect, an unmanned live operation control system based on an autonomous decision-making algorithm is provided, which includes: A UAV flight control module is used to control the UAV to fly along the power transmission line to be inspected; A transmission line segment image acquisition module, used for acquiring a surface state image of a predetermined line segment of the transmission line to be detected through a camera of the drone to obtain a time queue of the surface state image of the transmission line segment; An image level difference coefficient calculation module, used for calculating the image level difference coefficient between every two adjacent transmission line segment surface state images in the time queue of the transmission line segment surface state images to obtain a time queue of image set difference coefficients; A suspected outlier determination module, used to determine suspected outliers based on a comparison between the time queue of the image set difference coefficient and a preset threshold; An autonomous decision-making module, used to input the surface state image of the power transmission line section corresponding to the suspected abnormal point into an autonomous decision-making model based on deep learning to obtain a decision result, wherein the decision result is whether the power transmission line section is damaged; The drone hovering and collecting module is used to control the drone to hover and collect high-definition images of the transmission line section and transmit them back to the ground server in response to the decision result that the transmission line section is damaged.

[0008] Optionally, the autonomous decision-making module includes: an image shallow feature extraction unit, used to extract shallow features of the transmission line segment image from the surface state image of the transmission line segment to obtain a shallow feature map of the surface state of the transmission line segment; an image deep semantic feature extraction unit, used to extract deep semantic features of the transmission line segment image from the surface state image of the transmission line segment to obtain a deep semantic feature map of the surface state of the transmission line segment; a feature joint perception processing unit, used to perform feature joint perception processing based on semantic information field guidance on the shallow feature map of the surface state of the transmission line segment and the deep semantic feature map of the surface state of the transmission line segment to obtain a deep and shallow joint perception coding feature of the surface state of the transmission line segment; a decision result determination unit, used to determine the decision result based on the deep and shallow joint perception coding feature of the surface state of the transmission line segment.

[0009] Optionally, the image shallow feature extraction unit is used to: extract shallow features of the transmission line segment image from the transmission line segment surface state image through a transmission line segment surface state multi-scale feature extractor based on a hollow pyramid network to obtain a shallow feature map of the transmission line segment surface state.

[0010] Optionally, the image deep semantic feature extraction unit is used to: extract deep semantic features of the transmission line segment image from the transmission line segment surface state multi-scale feature extractor based on the hollow pyramid network to obtain the transmission line segment surface state deep semantic feature map.

[0011] Optionally, the feature joint perception processing unit includes: an upsampling subunit, which is used to upsample the deep semantic feature map of the surface state of the transmission line segment to obtain an upsampled deep semantic feature map of the surface state of the transmission line segment, wherein the upsampled deep semantic feature map of the surface state of the transmission line segment and the shallow feature map of the surface state of the transmission line segment have the same size; a semantic information field calculation subunit, which is used to calculate the semantic information field between the deep semantic feature map of the surface state of the upsampled transmission line segment and the shallow feature map of the surface state of the transmission line segment to obtain the transmission line segment. The surface state semantic information field is obtained by mapping the upsampled transmission line segment surface state deep semantic feature map to the transmission line segment surface state semantic information field to obtain the field-domain modulated transmission line segment surface state deep semantic feature map; a feature joint perception subunit is used to input the transmission line segment surface state shallow feature map and the field-domain modulated transmission line segment surface state deep semantic feature map into a feature joint perception module based on a multiple attention structure to obtain the transmission line segment surface state deep-shallow joint perception coding feature map as the transmission line segment surface state deep-shallow joint perception coding feature.

[0012] Optionally, the semantic information field calculation subunit includes: a semantic information field prediction secondary subunit, which is used to perform feature connection on the deep semantic feature map of the surface state of the upsampled transmission line segment and the shallow feature map of the surface state of the transmission line segment, and then input the feature maps into a gated convolution-based semantic information field predictor to obtain the semantic information field of the surface state of the transmission line segment; and a feature map mapping secondary subunit, which is used to map the deep semantic feature map of the surface state of the upsampled transmission line segment to the semantic information field of the surface state of the transmission line segment to obtain the deep semantic feature map of the surface state of the field-domain modulated transmission line segment.

[0013] Optionally, the semantic information field prediction secondary subunit is used to: perform feature connection on the upsampled transmission line segment surface state deep semantic feature map and the transmission line segment surface state shallow feature map, and then perform convolution kernel to obtain The convolution layer is processed to obtain a multi-scale semantic fusion feature map of the surface state of the transmission line segment; the multi-scale semantic fusion feature map of the surface state of the transmission line segment is point convolution processed to obtain a multi-dimensional semantic modulation feature map of the surface state of the transmission line segment as the semantic information field of the surface state of the transmission line segment.

[0014] Optionally, the feature joint perception subunit is used to: perform point-by-point convolution processing on the deep semantic feature map of the surface state of the field modulated transmission line segment and then input it into the batch normalization layer, and then use the Sigmoid function to perform activation processing to obtain the semantic attention optimization expression feature map of the surface state of the transmission line segment; perform multiple attention optimization expression on the shallow feature map of the surface state of the transmission line segment to obtain the semantic attention optimization expression feature map of the surface state of the transmission line segment; and perform position point-by-position optimization on the semantic attention optimization expression feature map of the surface state of the transmission line segment and the semantic attention optimization expression feature map of the surface state of the transmission line segment to obtain the deep and shallow joint perception coding feature map of the surface state of the transmission line segment.

[0015] Optionally, the decision result determination unit is used to: input the depth joint perceptual coding feature map of the surface state of the transmission line section into a classifier-based decision module to obtain the decision result.

[0016] The above technical solution uses drones to perform flight inspections on power transmission lines along a predetermined path, and uses image processing technology to analyze the status of power transmission lines in real time; then, a multi-stage feature extraction method is used, that is, shallow features are first extracted from the original image, and then deep semantic features are further mined, and finally the information is integrated through the feature joint perception module to generate the final decision result. In addition, a feedback mechanism is designed. When a suspected abnormality is detected, the drone can automatically adjust its position and hover over the target location to collect more detailed high-definition images, and transmit these data back to the ground server for further analysis. In this way, not only the speed of fault diagnosis is improved, but also the possibility of false alarms is reduced, thereby improving the reliability and practicality of the entire system.

[0017] Other features and advantages of the present application will be described in detail in the subsequent detailed description. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be 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 any creative work.

[0019] Figure 1 It is a block diagram of an unmanned live working control system based on an autonomous decision-making algorithm according to an embodiment of the present application.

[0020] Figure 2 It is a block diagram of the autonomous decision-making module in the unmanned live operation control system based on the autonomous decision-making algorithm according to an embodiment of the present application.

[0021] Figure 3 It is a block diagram of the feature joint perception processing unit in the unmanned live operation control system based on the autonomous decision-making algorithm according to an embodiment of the present application.

[0022] Figure 4 It is a block diagram of the semantic information field calculation subunit in the unmanned live operation control system based on the autonomous decision-making algorithm according to an embodiment of the present application.

[0023] Figure 5 It is a flow chart of an unmanned live working control method based on an autonomous decision-making algorithm according to an embodiment of the present application. DETAILED DESCRIPTION

[0024] The embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although certain embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be construed as being limited to the embodiments described herein. Instead, these embodiments are provided to provide a more thorough and complete understanding of the present application. It should be understood that the drawings and embodiments of the present application are only for exemplary purposes and are not intended to limit the scope of protection of the present application.

[0025] It should be understood that the various steps described in the method implementation of the present application can be performed in different orders and / or performed in parallel. In addition, the method implementation may include additional steps and / or omit the steps shown. The scope of the present application is not limited in this respect.

[0026] The term "including" and its variations used herein are open inclusions, i.e., "including but not limited to". The term "based on" means "based at least in part on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description.

[0027] It should be noted that the concepts such as "first" and "second" mentioned in this application are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0028] It should be noted that the modifications of "one" and "plurality" mentioned in the present application are illustrative rather than restrictive, and those skilled in the art should understand that unless otherwise clearly indicated in the context, it should be understood as "one or more".

[0029] The names of the messages or information exchanged between multiple devices in the embodiments of the present application are only used for illustrative purposes and are not used to limit the scope of these messages or information.

[0030] The specific implementation of the present application is described in detail below with reference to the accompanying drawings.

[0031] In view of the above technical problems, in the technical solution of the present application, an unmanned live operation control system based on an autonomous decision-making algorithm is proposed, which combines advanced drone technology and artificial intelligence algorithms, especially deep learning, to improve the intelligent level of automatic inspection of transmission lines. The system uses drones to perform flight inspections on transmission lines according to a predetermined path, and uses image processing technology to analyze the status of transmission lines in real time. In order to improve the accuracy of fault detection, the system adopts a multi-stage feature extraction method, that is, first extracting shallow features from the original image, then further mining deep semantic features, and finally integrating this information through a feature joint perception module to generate the final decision result. In addition, the system also designs a feedback mechanism. When a suspected abnormality is detected, the drone can automatically adjust its position and hover over the target location to collect more detailed high-definition images, and transmit these data back to the ground server for further analysis. This not only improves the speed of fault diagnosis, but also reduces the possibility of false alarms, thereby improving the reliability and practicality of the entire system.

[0032] Figure 1 is a block diagram of an unmanned live operation control system based on an autonomous decision-making algorithm according to an embodiment of the present application, such as Figure 1 As shown, according to an embodiment of the present application, an unmanned live working control system 100 based on an autonomous decision-making algorithm includes: The UAV flight control module 110 is used to control the UAV to fly along the power transmission line to be inspected; A transmission line segment image acquisition module 120, configured to acquire a surface state image of a predetermined line segment of the transmission line to be inspected through a camera of the drone to obtain a time queue of the surface state image of the transmission line segment; An image level difference coefficient calculation module 130, used to calculate the image level difference coefficient between every two adjacent transmission line segment surface state images in the time queue of the transmission line segment surface state images to obtain a time queue of image set difference coefficients; A suspected outlier determination module 140, configured to determine a suspected outlier based on a comparison between the time queue of the image set difference coefficient and a preset threshold; An autonomous decision-making module 150, configured to input the surface state image of the power transmission line section corresponding to the suspected abnormal point into an autonomous decision-making model based on deep learning to obtain a decision result, wherein the decision result is whether the power transmission line section is damaged; The drone hovering acquisition module 160 is used to control the drone to hover and acquire high-definition images of the transmission line section and transmit them back to the ground server in response to the decision result that the transmission line section is damaged.

[0033] In the above-mentioned unmanned live-line operation control system 100 based on the autonomous decision-making algorithm, the UAV flight control module 110 is used to control the UAV to fly along the power transmission line to be inspected. Specifically, the UAV flight control module 110 first needs to use geographic information system (GIS) data, combined with high-resolution maps and terrain information to set the flight route of the UAV, which includes determining the starting point, end point and key checkpoints along the way. On this basis, taking into account factors such as weather conditions, air traffic control requirements or emergencies such as bird activities, the flight control system can adjust the flight path in real time during the flight to ensure safety and efficiency. This dynamic path adjustment capability is crucial to ensure that the UAV successfully completes its mission.

[0034] Navigation and positioning are important components of the UAV flight control module 110. The global positioning system (GPS) provides accurate location information, while the inertial measurement unit (IMU) is used to compensate for the offset caused by wind or other factors to maintain a stable flight attitude. In addition, visual navigation assistance technology is also widely used. The camera installed on the UAV captures images and uses computer vision algorithms to identify feature points such as ground landmarks or power towers as additional navigation references to ensure that the UAV flies along the predetermined path. Through these technical means, the UAV can accurately find the target and maintain the correct heading even in complex environments.

[0035] To achieve a smooth flight, the drone uses an automatic flight mode. Once the path planning is completed, the built-in flight controller will control the speed, altitude and direction according to the preset parameters. At the same time, the obstacle avoidance function also plays an important role. The application of laser radar (LiDAR), ultrasonic sensors or stereo vision technology enables the drone to automatically slow down or change course when approaching obstacles to avoid collisions. This feature greatly improves the safety and reliability of drone operations, especially in complex power transmission line environments.

[0036] In a typical power transmission line inspection mission, the UAV flight control module 110 loads the map information of the section of the power transmission line from the database and formulates a flight plan based on it. After takeoff, the UAV relies on GPS and IMU to maintain the established route, while using a high-definition camera and infrared thermal imager to scan the line. When encountering complex terrain or severe weather conditions, the flight control system will automatically adjust the flight altitude and speed to ensure the safety of the equipment. If an abnormal situation is found, such as signs of damage to an insulator somewhere, the UAV will pause and stay above the fault point to take clearer photos for subsequent analysis. The whole process reflects the high intelligence and flexibility of the UAV flight control module, providing strong technical support for the maintenance of the power system.

[0037] Suppose a report is received that an unexplained short circuit has occurred on a high-voltage line in a certain area, and a drone needs to be dispatched quickly to investigate. At this point, the flight control module quickly plans a new path that is the shortest and avoids known danger zones. Upon arriving at the scene, the drone immediately starts an all-round scanning mode, recording not only visible light images but also temperature distribution maps to determine the specific location of the fault and its severity. All acquired data is sent synchronously to the command center to provide a basis for decision-making by the repair team. This emergency response scenario demonstrates the drone flight control module's ability to respond quickly and efficiently in emergencies.

[0038] In the above-mentioned unmanned live-line operation control system 100 based on the autonomous decision-making algorithm, the transmission line segment image acquisition module 120 is used to acquire the surface state image of the predetermined line segment of the transmission line to be inspected through the camera of the drone to obtain the time queue of the surface state image of the transmission line segment. Among them, the transmission line segment image acquisition module 120 acquires the surface state image of the predetermined transmission line segment through the high-definition camera or other special sensors (such as infrared thermal imagers) installed on the drone. This process not only requires precise control of the position and posture of the drone to ensure that the image covers the complete inspection area, but also requires adjustment of camera parameters such as focal length, aperture size, etc. according to different inspection requirements to obtain high-quality image data. In addition, considering that transmission lines are usually located in complex environments and may be affected by factors such as weather conditions and light changes, the image acquisition module must also have certain adaptability and flexibility to ensure the consistency and stability of image quality.

[0039] When the drone flies along a preset path, the transmission line segment image acquisition module 120 activates the high-definition camera or other dedicated sensors on the drone to continuously capture the surface state of the transmission line segment at specific time intervals or based on distance triggering. These images are recorded to form a time series, that is, a time queue of images of the surface state of the transmission line segment. The image at each time point represents the observation result of the drone on the transmission line at a certain position, and the entire time queue constitutes a comprehensive state record of the transmission line segment. In order to ensure that the collected images can meet the needs of subsequent analysis, the transmission line segment image acquisition module 120 must consider multiple aspects, including but not limited to image resolution, viewing angle, lighting compensation, and environmental adaptability.

[0040] Specifically, in a conventional power transmission line inspection mission, after the drone takes off from the ground station, it flies along a pre-planned path. At this time, the power transmission line segment image acquisition module 120 starts working, triggering the camera to take pictures at a fixed interval (for example, once per second), and recording the geographical location information corresponding to each photo. As the drone gradually approaches the target line, the power transmission line segment image acquisition module 120 will adjust the camera settings according to the actual situation, such as switching to wide-angle mode to obtain a larger field of view, or switching to a telephoto lens to focus on certain key parts. As the drone flies over the entire inspection area, the power transmission line segment image acquisition module 120 continues to collect a large number of images and stores them as a time series. Once the inspection is completed, all images will be transmitted back to the ground station for further processing and analysis. High-resolution images can provide clearer details and help identify subtle faults or damage features; appropriate viewing angles and angles are crucial to capturing the complete transmission line structure, and drones can optimize the viewing angle by adjusting the flight altitude and direction; changes in natural light may affect image quality, so the transmission line segment image acquisition module 120 should include automatic white balance and exposure compensation functions to maintain image consistency; for image acquisition under different weather conditions, such as fog, rain or strong light, the transmission line segment image acquisition module 120 takes corresponding measures, such as using a polarizing filter to reduce interference from reflected light, or enabling infrared imaging technology to penetrate smoke.

[0041] In the above-mentioned unmanned live-line operation control system 100 based on the autonomous decision-making algorithm, the image-level difference coefficient calculation module 130 is used to calculate the image-level difference coefficient between each two adjacent transmission line segment surface state images in the time queue of the transmission line segment surface state images to obtain the time queue of the image set difference coefficient. Before formally calculating the image-level difference coefficient, it is necessary to preprocess each two adjacent transmission line segment surface state images in the time queue of the transmission line segment surface state images. This step includes converting the color image into a grayscale image to reduce the computational complexity; ensuring that all images have the same resolution and scale to facilitate subsequent feature extraction and matching; and using filters (such as Gaussian blur or median filtering) to remove random noise in the image and improve image quality. After the preprocessing is completed, the image-level difference coefficient calculation module 130 enters the core step - feature extraction and matching. Commonly used feature extraction methods include using algorithms such as Harris corner detection to find significant feature points in the image; marking the outline of the object through the Canny operator or other edge detection techniques; using local binary patterns (LBP) to describe texture features, which helps to distinguish different types of materials or surface conditions; applying scale-invariant feature transform (SIFT) / speeded up robust features (SURF) to identify stable feature points that are not affected by scale and rotation, which is suitable for matching in complex scenes.

[0042] The next step is to find the corresponding relationship, that is, to pair the feature points of each two adjacent transmission line segment surface state images. This usually involves calculating the distance between the feature points (such as the Euclidean distance) and then selecting the closest pairing. In order to improve the matching accuracy, geometric constraints can also be introduced, such as verifying whether the pairing is reasonable based on the affine transformation model. With the matched feature points, the image set difference coefficient between each two adjacent transmission line segment surface state images can be calculated. The specific calculation method of the image set difference coefficient depends on the selected feature type and the requirements of the application scenario. The following are several common calculation methods: directly compare the difference in pixel values ​​at corresponding positions of each two adjacent transmission line segment surface state images, and the mean square error (MSE) or structural similarity index (SSIM) is often used to measure the overall similarity; calculate the displacement vector length between the paired feature points to reflect the position change between each two adjacent transmission line segment surface state images; construct a color or gradient direction histogram, and use statistical indicators such as chi-square distance or cross entropy to evaluate its similarity; train a convolutional neural network (CNN) to automatically learn high-level semantic differences between images, which can capture more complex visual information. Through any one or a combination of the above methods, the image-level difference coefficient calculation module 130 generates image set difference coefficients for each pair of surface status images of two adjacent transmission line sections, which constitute a time queue of image set difference coefficients, providing basic data support for subsequent outlier detection.

[0043] In a conventional power transmission line inspection mission, the drone flies along a pre-planned path, triggering the camera to take pictures at a certain distance, forming a time queue of the surface state image of the transmission line segment. After receiving the time queue of the surface state image of the transmission line segment, the image-level difference coefficient calculation module 130 first pre-processes it, including graying, size standardization and noise removal. Then, the corner detection algorithm is applied to find a number of stable feature points in each adjacent two transmission line segment surface state images in the time queue of the transmission line segment surface state image. Subsequently, an attempt is made to establish a correspondence between the feature points in the time queue of the surface state image of the transmission line segment, and the length of the displacement vector between each paired feature point is calculated as the image set difference coefficient. Finally, all image set difference coefficients are organized into a time queue of image set difference coefficients for further analysis.

[0044] In the above-mentioned unmanned live working control system 100 based on the autonomous decision-making algorithm, the suspected abnormal point determination module 140 is used to determine the suspected abnormal point based on the comparison between the time queue of the image set difference coefficient and the preset threshold. In order to effectively distinguish between normal minor changes and possible faults, it is necessary to set a reasonable threshold first. This step is usually based on historical data and domain expertise. For example, inspection data in multiple different time periods and different weather conditions can be collected, the distribution of the image difference coefficient under normal circumstances can be statistically calculated, and then the appropriate percentile can be selected as the threshold. In addition, taking into account the changes in the actual application environment, a dynamic adjustment mechanism can also be set so that the threshold can be adaptively updated according to seasonal factors, geographical location and other factors.

[0045] Once the threshold is set, the next step is to conduct an in-depth analysis of the time series of the difference coefficients of the image set. The key here is not only to pay attention to whether a single difference coefficient exceeds the threshold, but also to consider the trend of change in the entire time series. Specifically, if a difference coefficient suddenly rises or falls sharply and exceeds the set threshold range, the point is considered a preliminary suspect; if several consecutive difference coefficients exceed the threshold, it means that there may be more serious state changes during this period of time, which requires special attention; sometimes, due to the influence of the natural environment (such as slight swings caused by changes in wind speed), the image difference coefficient may show certain periodic fluctuations. In this case, filters can be used to remove high-frequency noise and retain low-frequency components to avoid misjudgments caused by normal fluctuations.

[0046] In order to improve the accuracy of the judgment of suspected abnormal points, it is also necessary to conduct a comprehensive assessment in combination with other relevant information. For example, you can refer to auxiliary data such as geographic coordinates and meteorological conditions recorded during the flight of the drone; you can also use multi-sensor fusion technology to integrate information from different types of sensors (such as infrared thermal imagers, lidar) to form a more comprehensive understanding. Ultimately, through the auxiliary decision-making of machine learning algorithms or expert systems, it is determined which points do have potential problems and marked as "suspected abnormal points" for subsequent detailed inspection.

[0047] Suppose a certain area is hit by heavy rain, causing some power transmission lines to be damaged, and it is urgent to dispatch drones for emergency assessment. Since bad weather may cause image quality to deteriorate, additional measures must be taken to ensure the validity of the judgment results. For example, the temperature distribution map obtained by the infrared thermal imager can be used for feature extraction first, because such images are less affected by lighting conditions and are more suitable for use at night or in low visibility environments. In addition, considering the interference of reflected light caused by rain, the local binary pattern (LBP) technology can also be combined to focus on analyzing the texture changes on the surface of the pole. When analyzing the time series of the difference coefficient of the image set, not only pay attention to whether a single difference coefficient exceeds the threshold, but also pay special attention to those data points that rise rapidly in a short period of time, which may be caused by sudden physical damage. At the same time, a dynamic adjustment mechanism is introduced to appropriately relax or tighten the threshold range according to real-time meteorological data to adapt to different environmental conditions. Finally, by combining multiple methods and techniques, the suspected abnormal point determination module 140 successfully identified multiple suspected abnormal points, helping the repair team to respond quickly.

[0048] The specific method examples for determining suspected outliers show different implementations. In a simple implementation, the suspected outlier determination module 140 can determine suspected outliers by directly comparing the time series of the difference coefficient of the image set with a fixed threshold. Specifically, assume that there is a difference coefficient time series {Di}, where i represents the time point of the i-th shooting. A fixed threshold T is set, then for each Di, if Di>T, the point is considered to be a suspected outlier. This method is suitable for most cases, especially when historical data shows that the change in the difference coefficient is relatively stable.

[0049] In order to improve the accuracy of determining suspected outliers, a method based on dynamic thresholds can be used. This method not only considers the value of the current difference coefficient, but also combines the average value and standard deviation in the previous period to set the threshold. Specifically, let Di be the difference coefficient of the i-th shot, and mi and si represent the average value and standard deviation of the difference coefficient from the in-th to the i-1-th shot, respectively. Then, the dynamic threshold Ti can be defined as: Here, k is an adjustable parameter used to control the sensitivity of the threshold. If Di>Ti, the point is considered a suspected outlier. This method can better adapt to environmental changes and reduce the possibility of false alarms.

[0050] In the above-mentioned unmanned live working control system 100 based on the autonomous decision-making algorithm, the autonomous decision-making module 150 is used to input the surface state image of the transmission line section corresponding to the suspected abnormal point into the autonomous decision-making model based on deep learning to obtain a decision result, and the decision result is whether the transmission line section is damaged. Figure 2is a block diagram of the autonomous decision-making module in the unmanned live working control system based on the autonomous decision-making algorithm according to an embodiment of the present application, such as Figure 2 As shown, the autonomous decision-making module 150 includes: an image shallow feature extraction unit 151, which is used to extract the shallow features of the transmission line segment image from the surface state image of the transmission line segment to obtain a shallow feature map of the surface state of the transmission line segment; an image deep semantic feature extraction unit 152, which is used to extract the deep semantic features of the transmission line segment image from the surface state image of the transmission line segment to obtain a deep semantic feature map of the surface state of the transmission line segment; a feature joint perception processing unit 153, which is used to perform feature joint perception processing based on semantic information field guidance on the shallow feature map of the surface state of the transmission line segment and the deep semantic feature map of the surface state of the transmission line segment to obtain a deep and shallow joint perception coding feature of the surface state of the transmission line segment; and a decision result determination unit 154, which is used to determine the decision result based on the deep and shallow joint perception coding feature of the surface state of the transmission line segment.

[0051] In particular, in the above-mentioned unmanned live-line operation control system based on the autonomous decision-making algorithm, it is crucial to input the surface state image of the transmission line section corresponding to the suspected abnormal point into the autonomous decision-making model based on deep learning to perform the damage detection process of the transmission line section. This is because the transmission lines in different regions and at different times may have different aging conditions or damage characteristics. Traditional fault detection methods rely on fixed thresholds or rules, which are often not flexible and accurate enough when facing complex actual environments. The method of detecting damage to the transmission line section through the autonomous decision-making model based on deep learning can identify more subtle and complex line defect patterns through a large amount of data training, thereby significantly improving the accuracy of fault detection such as transmission line damage. At the same time, by continuously learning new sample data to adjust its own parameters, the system's ability to adapt to various environmental changes is enhanced.

[0052] Specifically, in the process of autonomous decision-making, the technical concept of this application is to analyze the surface state image of the transmission line segment corresponding to the suspected abnormal point by using an image processing and analysis algorithm based on artificial intelligence and deep learning, so as to capture the multi-scale joint perception features of the surface state of the transmission line segment in the image, so as to determine the damage of the transmission line and return the decision result. In this way, when the transmission line segment is detected to be damaged, the drone can be controlled to further collect and transmit high-definition images to ensure the safe and reliable operation of the power grid, and provide strong technical support for the efficient management and maintenance of the power transmission system.

[0053] In a specific embodiment of the present application, in the image shallow feature extraction unit 151 and the image deep semantic feature extraction unit 152, first, the shallow features of the transmission line segment image are extracted from the transmission line segment surface state image to obtain the shallow feature map of the transmission line segment surface state, and the deep semantic features of the transmission line segment image are extracted from the transmission line segment surface state image to obtain the deep semantic feature map of the transmission line segment surface state. It should be understood that since the transmission line segment surface state image contains various feature information about the surface state of the transmission line segment, specifically including the edge, texture, color and semantic features of the transmission line segment, these features can not only directly reflect the surface state of the transmission line, such as whether there are obvious cracks or corrosion marks, but also help understand the state of the entire transmission line segment, rather than just the local performance, which provides a basis for subsequent more complex analysis and transmission line segment damage defect detection. Specifically, in a specific example of the present application, a multi-scale feature extractor of the surface state of a transmission line segment based on a hollow pyramid network is used to perform multi-scale feature extraction on the surface state image of the transmission line segment, so as to obtain a shallow feature map of the surface state of the transmission line segment and a deep semantic feature map of the surface state of the transmission line segment. It is worth mentioning that the combined use of the shallow features and deep semantic features of the surface state of the transmission line segment can complement each other and provide a richer and more comprehensive description of the surface state of the transmission line segment. This can not only improve the system's accuracy in detecting damage to the transmission line segment, but also enhance its adaptability and response speed to complex environmental changes.

[0054] In a specific embodiment of the present application, Figure 3 is a block diagram of the feature joint perception processing unit in the unmanned live operation control system based on the autonomous decision-making algorithm according to an embodiment of the present application, such as Figure 3As shown, the feature joint perception processing unit 153 includes: an upsampling subunit 1531, which is used to upsample the deep semantic feature map of the surface state of the transmission line segment to obtain an upsampled deep semantic feature map of the surface state of the transmission line segment, wherein the upsampled deep semantic feature map of the surface state of the transmission line segment and the shallow feature map of the surface state of the transmission line segment have the same size; a semantic information field calculation subunit 1532, which is used to calculate the semantic information field between the upsampled deep semantic feature map of the surface state of the transmission line segment and the shallow feature map of the surface state of the transmission line segment to obtain a transmission line segment surface state deep semantic feature map; The surface state semantic information field of the line segment, and mapping the upsampled transmission line segment surface state deep semantic feature map to the transmission line segment surface state semantic information field to obtain the field domain modulated transmission line segment surface state deep semantic feature map; the feature joint perception subunit 1533 is used to input the transmission line segment surface state shallow feature map and the field domain modulated transmission line segment surface state deep semantic feature map into a feature joint perception module based on a multiple attention structure to obtain the transmission line segment surface state deep and shallow joint perception coding feature map as the transmission line segment surface state deep and shallow joint perception coding feature.

[0055] It should be understood that since the shallow feature map of the surface state of the transmission line segment and the deep semantic feature map of the surface state of the transmission line segment respectively contain shallow features and deep semantic features related to the surface state of the transmission line segment, by combining the shallow and deep features, the visual performance of the surface defects of these transmission line segments can be more comprehensively captured, thereby improving the accuracy of fault identification and damage detection. Based on this, in order to be able to more fully and comprehensively perform surface state identification and damage detection of the transmission line segment, in the technical solution of the present application, the shallow feature map of the surface state of the transmission line segment and the deep semantic feature map of the surface state of the transmission line segment are further processed by feature joint perception based on the guidance of the semantic information field to obtain the deep and shallow joint perception coding features of the surface state of the transmission line segment. Through the feature joint perception processing guided by the semantic information field, the shallow and deep features of the surface state of the transmission line segment can be combined, and the semantic information field is used as a guide to achieve the interaction and modulation between the two features, thereby generating a richer and more semantically explanatory deep and shallow joint perception coding feature of the surface state of the transmission line segment.

[0056] Specifically, the upsampling subunit 1531 is used to upsample the deep semantic feature map of the surface state of the transmission line segment to obtain an upsampled deep semantic feature map of the surface state of the transmission line segment, wherein the upsampled deep semantic feature map of the surface state of the transmission line segment has the same size as the shallow feature map of the surface state of the transmission line segment.

[0057] Specifically, compared with the traditional visual processing architecture, the feature joint perception processing guided by the semantic information field introduces upsampling and feature connection, so that the upsampled deep semantic feature map of the surface state of the transmission line segment and the shallow feature map of the surface state of the transmission line segment can interact at the same scale, solving the problem of multi-scale information fusion and improving the model's ability to understand complex scenes.

[0058] The process can be expressed as follows:

[0059] in, is the deep semantic feature map of the surface state of the transmission line segment, represents upsampling processing, It is the deep semantic feature map of the surface state of the upsampled transmission line segment.

[0060] Specifically, Figure 4 is a block diagram of the semantic information field calculation subunit in the unmanned live operation control system based on the autonomous decision-making algorithm according to an embodiment of the present application, such as Figure 4 As shown, the semantic information field calculation subunit 1532 includes: a semantic information field prediction secondary subunit 15321, which is used to perform feature connection on the deep semantic feature map of the surface state of the upsampled transmission line segment and the shallow feature map of the surface state of the transmission line segment, and then input the feature into the gated convolution-based semantic information field predictor to obtain the semantic information field of the surface state of the transmission line segment; a feature map mapping secondary subunit 15322, which is used to map the deep semantic feature map of the surface state of the upsampled transmission line segment to the semantic information field of the surface state of the transmission line segment to obtain the deep semantic feature map of the surface state of the field-domain modulated transmission line segment.

[0061] More specifically, the semantic information field prediction secondary subunit 15321 is used to connect the deep semantic feature map of the surface state of the upsampled power transmission line segment with the shallow feature map of the surface state of the power transmission line segment through a convolution kernel. The convolution layer is processed to obtain a multi-scale semantic fusion feature map of the surface state of the transmission line segment; the multi-scale semantic fusion feature map of the surface state of the transmission line segment is point convolution processed to obtain a multi-dimensional semantic modulation feature map of the surface state of the transmission line segment as the semantic information field of the surface state of the transmission line segment.

[0062] Furthermore, the feature map mapping secondary subunit 15322 constructs the semantic information field of the surface state of the transmission line segment and modulates the deep semantic feature map of the surface state of the upsampled transmission line segment accordingly, thereby ensuring that the field-modulated deep semantic feature map of the surface state of the transmission line segment not only contains rich detail information, but also is consistent with the structure of the shallow feature map of the surface state of the transmission line segment. In particular, the feature combination method using semantic information field guided migration can enable the model to better understand complex scenes, and correctly identify key objects and damage defects of the transmission line segment even in the face of occlusion or deformation. At the same time, this method can also provide a more detailed and explanatory feature representation, which helps to discover early small faults or potential transmission line damage risks, and provide more reliable support for autonomous decision-making.

[0063] The process can be expressed as follows:

[0064]

[0065] in, is the deep semantic feature map of the surface state of the upsampled transmission line segment, is a shallow characteristic map of the surface state of the transmission line section, represents the feature connection operation, The convolution kernel is The convolutional layer, For point convolution processing, is the semantic information field of the surface state of the transmission line segment, For point-by-position multiplication, It is a deep semantic feature map of the surface state of the field-modulated transmission line segment.

[0066] Specifically, the feature joint perception subunit 1533 is used to perform point-by-point convolution processing on the deep semantic feature map of the surface state of the field modulated transmission line segment and then input it into the batch normalization layer, and then use the Sigmoid function to perform activation processing to obtain the semantic attention optimization expression feature map of the surface state of the transmission line segment; perform multiple attention optimization expression on the shallow feature map of the surface state of the transmission line segment to obtain the semantic attention optimization expression feature map of the surface state of the transmission line segment; and perform position point-by-position optimization on the semantic attention optimization expression feature map of the surface state of the transmission line segment and the semantic attention optimization expression feature map of the surface state of the transmission line segment to obtain the deep and shallow joint perception coding feature map of the surface state of the transmission line segment.

[0067] The process can be expressed as follows:

[0068]

[0069]

[0070] in, For point-by-point convolution processing, is the global mean pooling operation, For channel-by-channel convolution processing, is the batch normalization layer, is the Sigmoid function, Optimize the semantic attention representation feature map of the surface state of the transmission line segment. The shallow attention optimization expression feature map of the surface state of the transmission line section is used. For positional addition, is the depth-shallowness joint perception coding feature map of the surface state of the transmission line section, It is the deep semantic feature map of the surface state of the field-modulated transmission line segment. It is a shallow characteristic map of the surface state of the transmission line section.

[0071] In a specific embodiment of the present application, the decision result determination unit 154 is used to: input the depth-joint perception coding feature map of the surface state of the transmission line segment into a decision module based on a classifier to obtain the decision result. Then, the depth-joint perception coding feature map of the surface state of the transmission line segment is input into a decision module based on a classifier to obtain the decision result. In other words, the depth-joint perception feature of the surface state of the transmission line segment is used for classification processing to make a decision on whether the transmission line segment is damaged and return the decision result. In this way, when damage is detected in the transmission line segment, the drone can be controlled to perform further high-definition image acquisition and transmission, ensuring the safe and reliable operation of the power grid, and providing strong technical support for the efficient management and maintenance of the power transmission system.

[0072] That is, when the shallow feature map of the surface state of the transmission line segment and the deep semantic feature map of the surface state of the transmission line segment respectively represent the shallow features of the surface state of the transmission line segment and the deep semantic features of the surface state of the transmission line segment determined by the shallow feature extraction and the deep semantic feature extraction of the surface state image of the transmission line segment, the shallow-depth joint perception coding feature map of the surface state of the transmission line segment will also have the joint interactive distribution diversity of the surface states of the heterogeneous transmission line segments under the cross-feature domain feature distribution based on the fine-grained cores of their respective features. That is to say, due to the information differences in different feature layers (for example, the shallow layer captures local details, and the deep layer identifies global anomalies), the feature map presents an uneven distribution problem in cross-domain fusion. Therefore, when the shallow-depth joint perception coding feature map of the surface state of the transmission line segment is classified by a classifier, the accuracy of the classification result will be affected. In this way, considering that the weight matrix of the classifier acts on the joint perceptual coding feature vector of the surface state of the transmission line segment after the joint perceptual coding feature map of the surface state of the transmission line segment is expanded, the distribution diversity of the joint perceptual coding feature map of the surface state of the transmission line segment causes the randomness of the weight response of the weight matrix of the classifier, which affects the accuracy of the classification result.

[0073] Preferably, inputting the depth-shallowness joint perceptual coding feature map of the surface state of the transmission line section into a decision module based on a classifier to obtain the decision result comprises: The parameter expectation and distribution dispersion of the depth-joint perceptual coding feature map of the surface state of the transmission line section are calculated, and the parameter expectation is normalized based on the distribution dispersion to obtain the distribution probability index of the surface state of the transmission line section: The depth-shallowness joint perception coding feature map of the transmission line segment surface state is multiplied by the inverse of the transmission line segment surface state distribution probability index to obtain a transmission line segment surface state random robust map:

[0074]

[0075] in, A depth-shallowness joint perception coding feature map representing the surface state of the transmission line section, represents dot product, and They respectively represent the parameter expectation and distribution dispersion of the joint perception coding feature map of the surface state of the transmission line section, that is, the mean and standard deviation of the set of all feature values ​​of the joint perception coding feature map of the surface state of the transmission line section, represents the probability index of the surface state distribution of the transmission line section, A stochastic robust graph representing the surface state of a transmission line segment; Analyze the number of feature values ​​in the depth-shallowness joint perception coding feature map of the surface state of the transmission line section that are greater than the expected feature value of the parameter as the probability stack base; After the transmission line segment surface state probability constraint diagram and the probability stack base are added, a binary logarithmic transformation is performed to obtain the transmission line segment surface state probability constraint diagram:

[0076] in, Indicates point addition, represents the probability stack base, A diagram showing the surface state stability constraints of a transmission line section; The probability stack base is used to perform a differential operation on the transmission line segment surface state probability constraint graph, and then an inverse element calculation is performed to obtain a transmission line segment surface state measurement series suppression graph:

[0077] in, represents the difference operation, A measurement sequence suppression diagram representing the surface state of a transmission line section; The surface state stability constraint map of the transmission line segment is feature-fused with the surface state measurement sequence suppression map of the transmission line segment to obtain an optimized surface state depth-shallowness joint perception coding feature map of the transmission line segment; The optimized transmission line section surface state depth joint perception coding feature map is input into a classifier-based decision module to obtain a decision result.

[0078] Accordingly, in this preferred embodiment, a dynamic coupling relationship is established between the physical magnitude dimension and the abstract feature dimension through the statistical distribution characteristics of the joint perception coding feature map of the surface state of the transmission line segment. In addition, a dual-channel latent factor topology is generated using cross-level short-time series data based on statistics to achieve bidirectional knowledge transfer at an intermediate scale. At the same time, the time series correlation constraints are integrated to complete the latent variable inversion compensation, enhance the optimization stability in the distribution function space, and thus improve the accuracy of the decision results output by the joint perception coding feature map of the surface state of the transmission line segment via the decision module based on the classifier.

[0079] In the above-mentioned unmanned live-line operation control system 100 based on the autonomous decision-making algorithm, the drone hovering acquisition module 160 is used to control the drone to hover and collect high-definition images of the transmission line section in response to the decision result that the transmission line section is damaged and transmit it back to the ground server. Once it reaches the vicinity of the target position, the drone hovering acquisition module 160 controls the drone to gradually slow down until it stops completely, and adjusts its posture to maintain a stable hovering state. This step is crucial for obtaining high-quality images because any unnecessary movement may cause the captured image to be blurred. In order to maintain hovering stability, the drone hovering acquisition module 160 uses an advanced flight controller algorithm, combined with the information fed back by the IMU, to dynamically adjust the motor output power, offset external interference factors (such as wind, airflow changes, etc.), and ensure that the drone can be fixed in the air.

[0080] After completing the image acquisition, the drone hovering acquisition module 160 compresses and encodes the obtained high-definition image and transmits it back to the ground server through a wireless communication link. In order to ensure the reliability and speed of data transmission, the drone hovering acquisition module 160 usually adopts a high-speed, low-latency communication protocol, such as 5G network or Wi-Fi 6 technology. At the same time, considering the possible signal blocking problem, the drone hovering acquisition module 160 will also pre-plan multiple backup transmission paths to ensure that even if one path fails, other available paths can still be found to complete the data upload. In addition, in order to improve transmission efficiency, the drone hovering acquisition module 160 can also apply image compression algorithms to reduce file size and speed up transmission rate without affecting image quality.

[0081] In one embodiment of the present application, Figure 5 Flow chart of an unmanned live working control method based on an autonomous decision-making algorithm according to an embodiment of the present application, such as Figure 5As shown, according to the unmanned live working control method based on the autonomous decision-making algorithm of the embodiment of the present application, the method includes: S210, controlling the drone to fly along the transmission line to be detected; S220, collecting the surface state image of the predetermined line segment of the transmission line to be detected through the camera of the drone to obtain the time queue of the surface state image of the transmission line segment; S230, calculating the image-level difference coefficient between each two adjacent surface state images of the transmission line segment in the time queue of the surface state image of the transmission line segment to obtain the time queue of the image set difference coefficient; S240, determining the suspected abnormal point based on the comparison between the time queue of the image set difference coefficient and the preset threshold; S250, inputting the surface state image of the transmission line segment corresponding to the suspected abnormal point into the autonomous decision-making model based on deep learning to obtain a decision result, and the decision result is whether the transmission line segment is damaged; S260, in response to the decision result that the transmission line segment is damaged, controlling the drone to hover and collect the high-definition image of the transmission line segment and transmit it back to the ground server.

[0082] Those skilled in the art will appreciate that the specific operations of each step in the above-mentioned unmanned live working control method based on the autonomous decision-making algorithm have been described in detail above. Figures 1 to 4 The description of the unmanned live working control system 100 based on the autonomous decision-making algorithm has been introduced in detail, and therefore, its repeated description will be omitted.

[0083] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of disclosure involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features with similar functions disclosed in this application (but not limited to) by each other to form a technical solution.

[0084] In addition, although each operation is described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or to be performed in a sequential order. Under certain circumstances, multitasking and parallel processing may be advantageous. Similarly, although some specific implementation details are included in the above discussion, these should not be construed as limiting the scope of the application. Some features described in the context of a separate embodiment can also be implemented in a single embodiment in combination. On the contrary, the various features described in the context of a single embodiment can also be implemented in multiple embodiments individually or in any suitable sub-combination mode.

[0085] Although the subject matter has been described in language specific to structural features and / or method logic actions, it should be understood that the subject matter defined is not necessarily limited to the specific features or actions described above. On the contrary, the specific features and actions described above are merely exemplary forms of implementation. With respect to the apparatus in the above-mentioned embodiments, the specific manner in which each module performs the operation has been described in detail in the embodiments related to the method, and will not be described in detail here.

Claims

1. An unmanned live working control system based on an autonomous decision-making algorithm, characterized in that: include: A UAV flight control module is used to control the UAV to fly along the power transmission line to be inspected; A transmission line segment image acquisition module, used for acquiring a surface state image of a predetermined line segment of the transmission line to be detected through a camera of the drone to obtain a time queue of the surface state image of the transmission line segment; An image level difference coefficient calculation module, used for calculating the image level difference coefficient between every two adjacent transmission line segment surface state images in the time queue of the transmission line segment surface state images to obtain a time queue of image set difference coefficients; A suspected outlier determination module, configured to determine suspected outliers based on a comparison between a time queue of a difference coefficient of the image set and a preset threshold; An autonomous decision-making module, used to input the surface state image of the power transmission line section corresponding to the suspected abnormal point into an autonomous decision-making model based on deep learning to obtain a decision result, wherein the decision result is whether the power transmission line section is damaged; The drone hovering and collecting module is used to control the drone to hover and collect high-definition images of the transmission line section and transmit them back to the ground server in response to the decision result that the transmission line section is damaged.

2. The unmanned live working control system based on the autonomous decision-making algorithm according to claim 1 is characterized in that: The autonomous decision-making module comprises: An image shallow feature extraction unit, used for extracting the shallow features of the transmission line segment image from the transmission line segment surface state image to obtain a transmission line segment surface state shallow feature map; An image deep semantic feature extraction unit, used for extracting the deep semantic features of the transmission line segment image from the transmission line segment surface state image to obtain a transmission line segment surface state deep semantic feature map; A feature joint perception processing unit, used for performing feature joint perception processing based on the guidance of the semantic information field on the shallow feature map of the surface state of the transmission line segment and the deep semantic feature map of the surface state of the transmission line segment to obtain the deep and shallow joint perception coding features of the surface state of the transmission line segment; A decision result determination unit is used to determine the decision result based on the depth and shallowness joint perception coding features of the surface state of the transmission line section.

3. The unmanned live working control system based on the autonomous decision-making algorithm according to claim 2 is characterized in that: The image shallow feature extraction unit is used to extract shallow features of the transmission line segment image from the transmission line segment surface state multi-scale feature extractor based on a hollow pyramid network to obtain a shallow feature map of the transmission line segment surface state.

4. The unmanned live working control system based on the autonomous decision-making algorithm according to claim 3 is characterized in that: The image deep semantic feature extraction unit is used to extract the deep semantic features of the transmission line segment image through the transmission line segment surface state multi-scale feature extractor based on the hollow pyramid network to obtain the transmission line segment surface state deep semantic feature map.

5. The unmanned live working control system based on the autonomous decision-making algorithm according to claim 4 is characterized in that: The feature joint perception processing unit includes: An upsampling subunit, configured to upsample the power line segment surface state deep semantic feature map to obtain an upsampled power line segment surface state deep semantic feature map, wherein the upsampled power line segment surface state deep semantic feature map and the power line segment surface state shallow feature map have the same size; A semantic information field calculation subunit, used for calculating the semantic information field between the upsampled transmission line segment surface state deep semantic feature map and the transmission line segment surface state shallow feature map to obtain the transmission line segment surface state semantic information field, and mapping the upsampled transmission line segment surface state deep semantic feature map to the transmission line segment surface state semantic information field to obtain a field domain modulated transmission line segment surface state deep semantic feature map; The feature joint perception subunit is used to input the shallow feature map of the surface state of the transmission line segment and the deep semantic feature map of the surface state of the field modulated transmission line segment into a feature joint perception module based on a multiple attention structure to obtain a deep and shallow joint perception coding feature map of the surface state of the transmission line segment as the deep and shallow joint perception coding feature of the surface state of the transmission line segment.

6. The unmanned live working control system based on the autonomous decision-making algorithm according to claim 5 is characterized in that: The semantic information field calculation subunit comprises: A semantic information field prediction secondary subunit is used to perform feature connection on the upsampled transmission line segment surface state deep semantic feature map and the transmission line segment surface state shallow feature map, and then input the feature map into a gated convolution-based semantic information field predictor to obtain the transmission line segment surface state semantic information field; The feature map mapping secondary subunit is used to map the upsampled transmission line segment surface state deep semantic feature map to the transmission line segment surface state semantic information field to obtain the field domain modulated transmission line segment surface state deep semantic feature map.

7. The unmanned live working control system based on the autonomous decision-making algorithm according to claim 6 is characterized in that: The semantic information field prediction secondary subunit is used to: The upsampled transmission line segment surface state deep semantic feature map and the transmission line segment surface state shallow feature map are feature-connected and then convolution kernel is used to obtain: The convolutional layer is processed to obtain a multi-scale semantic fusion feature map of the surface state of the transmission line segment; The multi-scale semantic fusion feature map of the surface state of the transmission line segment is subjected to point convolution processing to obtain a multi-dimensional semantic modulation feature map of the surface state of the transmission line segment as the semantic information field of the surface state of the transmission line segment.

8. The unmanned live working control system based on the autonomous decision-making algorithm according to claim 7 is characterized in that: The feature joint perception subunit is used to: The deep semantic feature map of the surface state of the field-modulated transmission line segment is subjected to point-by-point convolution processing and then input into a batch normalization layer, and then activated by a Sigmoid function to obtain a semantic attention-optimized expression feature map of the surface state of the transmission line segment; Performing multiple attention optimization expression on the shallow feature map of the surface state of the transmission line segment to obtain a semantic attention optimization expression feature map of the surface state of the transmission line segment; The semantic attention optimized expression feature map of the surface state of the transmission line segment and the semantic attention optimized expression feature map of the surface state of the transmission line segment are processed according to the position points to obtain the depth joint perception coding feature map of the surface state of the transmission line segment.

9. The unmanned live working control system based on the autonomous decision-making algorithm according to claim 8 is characterized in that: The decision result determination unit is used to input the depth-shallowness joint perception coding feature map of the surface state of the transmission line section into a decision module based on a classifier to obtain the decision result.

Images