Method and device for visualizing hidden danger identification of overhead transmission line

Abstract

This invention discloses a method and device for visual monitoring and hazard identification of overhead transmission lines, belonging to the field of power technology. First, a hazard identification model is used to detect hazards in periodically captured images. When the hazard type is a foreign object on the conductor, it is determined as a critical alarm. When the hazard type is construction machinery or smoke / wildfire, based on the movement trend of the machinery or smoke / wildfire, and the positional relationship between the alarm location and the protection zone of the overhead transmission line, it is determined as a general alarm, a key alarm, or a critical alarm. For key alarms, periodic intensive image capture is performed; for critical alarms, video streams of the alarm location are collected and analyzed. This invention reduces the workload of maintenance personnel in handling alarms, avoids serious accidents such as power outages, personal injury, and vehicle damage caused by untimely alarm handling, and can effectively monitor the development and changes of hazards while saving equipment power.

Description

Technical Field

[0001] This invention relates to the field of power technology, and in particular to a method and device for visually monitoring and identifying potential hazards in overhead transmission lines. Background Technology

[0002] Currently, the visual monitoring algorithm for identifying potential hazards in overhead transmission lines uses a target detection scheme to identify hazards. After the front-end device or back-end service identifies a hazard, it sends an alarm to the platform. After receiving the alarm, the monitoring personnel decide on the next steps based on their experience. For urgent alarms, they resolve them through audible and visual alarms or on-site inspections. For non-urgent alarms, they closely monitor them but do not take any action. For alarms that pose no threat at all, they do not take any action.

[0003] As the number of visual monitoring devices increases and the data collection vacuum period shortens, the number of alarms on the platform is increasing day by day. Personnel monitoring is limited by the number of people and staff fatigue, which may lead to missed detections. This results in urgent alarms being silenced in the alarm messages, which may lead to serious accidents such as power outages, personal injury, or vehicle damage. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method and device for visual monitoring and identifying potential hazards in overhead transmission lines. This reduces the workload of maintenance personnel in handling alarms, avoids serious accidents such as power outages, personal injury, and vehicle damage caused by untimely alarm handling, and enables effective monitoring of the development and changes of potential hazards while saving equipment power.

[0005] The technical solution provided by this invention is as follows:

[0006] A method for identifying potential hazards in overhead transmission lines through visual monitoring, the method comprising:

[0007] S1: Acquire images captured periodically by a monitoring device, and use a trained hazard identification model to detect hazards in the images, determine whether there are hazards and the type of hazard; wherein, the hazard types include construction machinery, wires and foreign objects, and smoke and wildfires;

[0008] S2: When the hazard type is determined to be construction machinery or smoke / wildfire, execute S3; when the hazard type is determined to be wire or foreign object, the alarm level is determined to be a critical alarm, and execute S4.

[0009] S3: Based on the movement trend of the construction machinery or smoke and wildfire and the relationship between the alarm location of the construction machinery or smoke and wildfire and the protected area of ​​the set overhead transmission line, the alarm level is determined as a general alarm, a key alarm or a critical alarm.

[0010] S4: For general alarms, no action is taken, and the process waits for the next inspection cycle to start from S1; for key alarms, the monitoring device performs timed and intensive snapshots, outputting key alarm information and timed and intensively snapshot images; for critical alarms, the position and focus of the monitoring device are adjusted, video streams of the alarm location are collected, the video streams are analyzed, and critical alarm information and video stream analysis results are output.

[0011] Furthermore, the protection zone of the overhead transmission line is set through the following process:

[0012] Acquire target images including conductors and towers of overhead transmission lines, and detect the towers and conductors in the target images using tower detection models and conductor detection models respectively to obtain the tower positions and conductor positions;

[0013] The main tower is determined based on the relative position information of the conductor position and the tower position, and the outermost conductors on both sides of the main tower are determined using the conductor position. The area between the main tower and the outermost conductor is determined as the initial protection zone.

[0014] The initial protection zone is expanded outward based on the voltage level of the overhead transmission line to obtain the protection zone of the overhead transmission line.

[0015] Furthermore, the initial protection zone is expanded outward through the following process:

[0016] Set the expansion coefficient K and calculate the expansion distance R;

[0017] Where, (R=X) R -X L )*K / 2, X R and X L These are the X coordinates of the right edge point and the left edge point of the initial protection zone, respectively;

[0018] For each contour point P of the initial protection zone i When the contour point P i When the point is convex, concave, or flat, its expanded contour point Q is calculated using the following formulas (1), (2), and (3), respectively. i ;

[0019]

[0020]

[0021]

[0022] in,

[0023] P i-1 and P i+1These are the contour points P. i The previous contour point and the next contour point, θ i Let P be the line segment i-1 P i With P i+ 1P i The included angle, e i For P i-1 Point to P i unit vector, e i+1 For P i Point to P i+1 The unit vector.

[0024] Furthermore, S3 includes:

[0025] S31: Calculate the intersection-union ratio (IOU1) of the detection boxes of the smoke and wildfire alarm locations in the two images before and after, and compare the IOU1 with the set threshold T1. If IOU1 is less than T1, the alarm level is determined to be a key alarm; otherwise, execute S32.

[0026] Calculate the intersection-union ratio (IOU2) of the detection boxes of the alarm positions of the construction machinery in the two images before and after. Compare the IOU2 with the set threshold T2. If IOU2 is greater than T2, the alarm level is determined to be a general alarm. Otherwise, execute S32.

[0027] S32: Determine whether the detection frame of the alarm location of construction machinery or smoke and wildfire overlaps with the protection zone of the overhead transmission line. If not, the alarm level is determined to be a general alarm. If yes, execute S33.

[0028] S33: If the alarm location of construction machinery or smoke and wildfire is located outside the protection zone of overhead transmission line or the nearest distance between the alarm location of construction machinery or smoke and wildfire and the conductor location is greater than the first distance, the alarm level will be determined as a general alarm.

[0029] If the alarm location of construction machinery or smoke / wildfire is outside the protection zone of overhead transmission line, or if the closest distance between the alarm location of construction machinery or smoke / wildfire and the conductor location is between the second distance and the first distance, then the alarm level will be determined as a key alarm.

[0030] If the alarm location of construction machinery or smoke / wildfire is located within the protection zone of overhead transmission line, or if the nearest distance between the alarm location of construction machinery or smoke / wildfire and the conductor location is less than the second distance, the alarm level will be determined as a critical alarm.

[0031] Wherein, the second distance is less than the first distance.

[0032] Furthermore, the video stream is analyzed through the following process:

[0033] The video detection model detects potential hazards in each frame of the video stream, the target tracking model performs continuous analysis of the detected hazards, and the trend determination method calculates the motion state of construction machinery or the spread trend of smoke and wildfire.

[0034] Furthermore, the position of the monitoring device is adjusted through the following process:

[0035] Calculate the X-direction adjustment scale Angle_x and Y-direction adjustment scale Angle_y of the monitoring device, and adjust the position of the monitoring device according to Angle_x and Angle_y;

[0036] Angle_x=(bbox_center_x-image_center_x)*FOV_width / width

[0037] Angle_y=(bbox_center_y-image_center_y)*FOV_height / height

[0038] Where bbox_center_x and bbox_center_y are the X and Y coordinates of the center of the alarm location, image_center_x and image_center_y are the X and Y coordinates of the center of the image, FOV_width and FOV_height are the width and height viewing angles of the monitoring device, and width and height are the width and height of the image, respectively.

[0039] Adjust the focal length of the monitoring device using the following procedure:

[0040] Calculate the new focal length New_focal of the monitoring device, and adjust the focal length of the monitoring device according to the new focal length New_focal;

[0041] New_focal=current_focal+bbox_width / width*scale

[0042] Where current_focal is the current focal length of the monitoring device, bbox_width is the width of the monitoring device, and scale is the set adjustment ratio coefficient.

[0043] A visual monitoring and hazard identification device for overhead transmission lines, the device comprising:

[0044] The hazard detection module is used to acquire images captured periodically by a monitoring device, and to detect hazards in the images using a trained hazard recognition model to determine whether there are hazards and the type of hazard; wherein, the hazard types include construction machinery, wires and foreign objects, and smoke and wildfires;

[0045] The first-level judgment module is used to execute the second-level judgment module when the hazard type is judged to be construction machinery or smoke and wildfire; when the hazard type is judged to be wire or foreign object, the alarm level is judged to be a critical alarm and the alarm processing module is executed.

[0046] The second-level judgment module is used to determine the alarm level as a general alarm, a key alarm, or a critical alarm based on the movement trend of the construction machinery or smoke and wildfire and the positional relationship between the alarm location of the construction machinery or smoke and wildfire and the protection zone of the set overhead transmission line.

[0047] The alarm processing module is used to handle general alarms without processing them, waiting for the next inspection cycle to be executed by the hidden danger detection module; for key alarms, it uses a monitoring device to perform timed and intensive snapshots, outputting key alarm information and timed and intensively snapshot images; for critical alarms, it adjusts the position and focus of the monitoring device to acquire video streams of the alarm location, analyzes the video streams, and outputs critical alarm information and video stream analysis results.

[0048] Furthermore, the protection zone of the overhead transmission line is set through the following process:

[0049] Acquire target images including conductors and towers of overhead transmission lines, and detect the towers and conductors in the target images using tower detection models and conductor detection models respectively to obtain the tower positions and conductor positions;

[0050] The main tower is determined based on the relative position information of the conductor position and the tower position, and the outermost conductors on both sides of the main tower are determined using the conductor position. The area between the main tower and the outermost conductor is determined as the initial protection zone.

[0051] The initial protection zone is expanded outward based on the voltage level of the overhead transmission line to obtain the protection zone of the overhead transmission line.

[0052] Furthermore, the initial protection zone is expanded outward through the following process:

[0053] Set the expansion coefficient K and calculate the expansion distance R;

[0054] Where, (R=X) R -X L )*K / 2, X R and X LThese are the X coordinates of the right edge point and the left edge point of the initial protection zone, respectively;

[0055] For each contour point P of the initial protection zone i When the contour point P i When the point is convex, concave, or flat, its expanded contour point Q is calculated using the following formulas (1), (2), and (3), respectively. i ;

[0056]

[0057]

[0058]

[0059] in,

[0060] P i-1 and P i+1 These are the contour points P. i The previous contour point and the next contour point, θ i Let P be the line segment i-1 P i With P i+ 1P i The included angle, e i For P i-1 Point to P i unit vector, e i+1 For P i Point to P i+1 The unit vector.

[0061] Furthermore, the second-level judgment module includes:

[0062] The motion trend judgment unit is used to calculate the intersection-union ratio (IOU1) of the detection boxes of the alarm positions of smoke and wildfire in the two images before and after. The IOU1 is compared with the set threshold T1. If the IOU1 is less than T1, the alarm level is judged as a key alarm. Otherwise, the qualitative analysis unit is executed.

[0063] Calculate the intersection-union ratio (IOU2) of the detection boxes of the alarm positions of the construction machinery in the two images before and after. Compare the IOU2 with the set threshold T2. If IOU2 is greater than T2, the alarm level is determined to be a general alarm. Otherwise, the qualitative analysis unit is executed.

[0064] The qualitative analysis unit is used to determine whether the detection frame of the alarm location of construction machinery or smoke and wildfire overlaps with the protection zone of the overhead transmission line. If not, the alarm level is determined to be a general alarm; if so, the quantitative analysis unit is executed.

[0065] The quantitative analysis unit is used to determine the alarm level as a general alarm if the alarm location of construction machinery or smoke and wildfire is outside the protection zone of the overhead transmission line or the nearest distance between the alarm location of construction machinery or smoke and wildfire and the conductor location is greater than the first distance.

[0066] If the alarm location of construction machinery or smoke / wildfire is outside the protection zone of overhead transmission line, or if the closest distance between the alarm location of construction machinery or smoke / wildfire and the conductor location is between the second distance and the first distance, then the alarm level will be determined as a key alarm.

[0067] If the alarm location of construction machinery or smoke / wildfire is located within the protection zone of overhead transmission line, or if the nearest distance between the alarm location of construction machinery or smoke / wildfire and the conductor location is less than the second distance, the alarm level will be determined as a critical alarm.

[0068] Wherein, the second distance is less than the first distance.

[0069] Furthermore, the video stream is analyzed through the following process:

[0070] The video detection model detects potential hazards in each frame of the video stream, the target tracking model performs continuous analysis of the detected hazards, and the trend determination method calculates the motion state of construction machinery or the spread trend of smoke and wildfire.

[0071] Furthermore, the position of the monitoring device is adjusted through the following process:

[0072] Calculate the X-direction adjustment scale Angle_x and Y-direction adjustment scale Angle_y of the monitoring device, and adjust the position of the monitoring device according to Angle_x and Angle_y;

[0073] Angle_x=(bbox_center_x-image_center_x)*FOV_width / width

[0074] Angle_y=(bbox_center_y-image_center_y)*FOV_height / height

[0075] Where bbox_center_x and bbox_center_y are the X and Y coordinates of the center of the alarm location, image_center_x and image_center_y are the X and Y coordinates of the center of the image, FOV_width and FOV_height are the width and height viewing angles of the monitoring device, and width and height are the width and height of the image, respectively.

[0076] Adjust the focal length of the monitoring device using the following procedure:

[0077] Calculate the new focal length New_focal of the monitoring device, and adjust the focal length of the monitoring device according to the new focal length New_focal;

[0078] New_focal=current_focal+bbox_width / width*scale

[0079] Where current_focal is the current focal length of the monitoring device, bbox_width is the width of the monitoring device, and scale is the set adjustment ratio coefficient.

[0080] A computer-readable storage medium for visual monitoring and hazard identification of overhead transmission lines includes a memory for storing processor-executable instructions, which, when executed by the processor, implement the steps of the method for visual monitoring and hazard identification of overhead transmission lines.

[0081] A device for visual monitoring and hazard identification of overhead transmission lines includes at least one processor and a memory storing computer-executable instructions. When the processor executes the instructions, it implements the steps of the visual monitoring and hazard identification method for overhead transmission lines.

[0082] The present invention has the following beneficial effects:

[0083] 1. This invention classifies the threat level of potential hazards into three levels: critical alarms, important alarms, and general alarms. Critical alarms are those requiring intervention from maintenance personnel; important alarms are those requiring close monitoring to prepare for potential crisis alarms; and general alarms are those that do not require attention and pose no threat to conductors or towers. This invention can conduct targeted analysis based on the type of hazard, fully exploring the hazard characteristics of transmission line service scenarios to define effective alarms. Only critical and important alarms are effective alarms requiring attention, while general alarms are ineffective alarms that do not require attention. Maintenance personnel only need to focus on key alarms and prioritize handling critical alarms; general alarms do not require attention. This reduces the workload of maintenance personnel in handling alarms, avoids overlooking alarms, and prevents serious accidents such as power outages, personal injury, or vehicle damage caused by alarms.

[0084] 2. Due to power consumption requirements, the monitoring device cannot operate for extended periods for video analysis. Furthermore, video analysis models generally have significantly lower recognition accuracy than image analysis models. Therefore, this invention divides the process into two parts: image analysis and video analysis. Normally, the monitoring device periodically captures and analyzes images. When a key alarm is detected, intensive image capture is performed. Only in the event of a critical alarm is video collected and analyzed. Through effective handling of key and critical alarms, effective monitoring of the development and changes of potential hazards can be achieved while conserving device power. Attached Figure Description

[0085] Figure 1 This is a flowchart of the method for identifying potential hazards in overhead transmission lines using visual monitoring, as described in this invention.

[0086] Figure 2 A schematic diagram of the initial protection zone for an overhead transmission line;

[0087] Figure 3 This is a schematic diagram showing the expansion of the initial protected area.

[0088] Figure 4 This is a schematic diagram of the expanded protective zone after the initial protective zone was established.

[0089] Figure 5 A schematic diagram of IOU;

[0090] Figure 6 This is a schematic diagram of the overhead transmission line visual monitoring and hidden danger identification device of the present invention. Detailed Implementation

[0091] To make the technical problems, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with the accompanying drawings and specific embodiments. Obviously, the described embodiments are only a part of the embodiments of this invention, and not all of them. The components of the embodiments of this invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without inventive effort are within the scope of protection of this invention.

[0092] This invention provides a method for identifying potential hazards in overhead transmission lines through visual monitoring, such as... Figure 1 As shown, the method includes:

[0093] S1: Acquire images captured at regular intervals by the monitoring device, and use the trained hazard identification model to detect hazards in the images, determine whether there are hazards and the type of hazard; among them, hazard types include construction machinery, wires and foreign objects, and smoke and wildfires.

[0094] The need for hazard analysis in power transmission line operations involves collecting scene images using monitoring devices such as cameras installed on transmission line towers. A typical image is taken every 10-30 minutes, with the interval adjustable. Taking 13,000 overhead transmission lines installed in a certain city as an example, there are 13,000 scenes. Statistics show that approximately 20% of these scenes pose a hazard. With one image taken every 10 minutes, 6 images are needed per hour. The alarm probability for these 13,000 scenes is 13,000 * 6 * 0.2 = 15,600 images. Every 10 minutes, 13,000 * 0.2 = 2,600 hazard images need to be reviewed. A typical maintenance team, typically fewer than 10 people, is responsible for 13,000 devices. Each person is responsible for 260 scenes, not only confirming alarms but also assessing the threat level, and requires long-term monitoring, resulting in significant fatigue.

[0095] In some scenarios, potential hazards may remain stationary or move far from the conductors, posing no threat to the towers, and therefore there is no need to report them to maintenance personnel for follow-up. However, for other hazards, such as foreign objects hanging from the conductors, which are dangerous and must be dealt with immediately, it is necessary to address them promptly. For fires, if the fire is small and controllable, only monitoring is required, and no further action is needed. Similarly, for cranes, if they are far from the conductors, there is no need to report them to maintenance personnel. Therefore, it is necessary to classify the threat level of potential hazards and determine whether to report them to maintenance personnel based on their severity level, thereby reducing the workload of maintenance personnel.

[0096] Because the level of a hazard is related to its type, and different hazard types require different methods for determining their levels, this invention first performs hazard detection on time-captured images to determine the presence and type of a hazard, and then locates the hazard to obtain its position. The hazard position can typically be represented by a detection bounding box.

[0097] There are various conventional hazard identification models, such as those based on YOLOv5, Faster R-CNN, and Cascade R-CNN. These hazard identification models need to be trained before use.

[0098] Taking the YOLOv5 target detection model as an example, the labeled hazard data (including three main categories: construction machinery, wires and foreign objects, and smoke and wildfires, with construction machinery including cranes, cement pump trucks, pile drivers, dump trucks, etc.) is input into the YOLOv5 network. After performing Mosaic enhancement, adaptive anchor box calculation, and image scaling on the training data, it is input into the backbone network to extract image features. Finally, it is input into the FPN and PAN structures to fuse multi-scale features, and finally input into the loss function to calculate the loss. The error is backpropagated to realize the parameter training process of the model.

[0099] After training, the system can detect potential hazards in the input images. If no hazards are identified, it continues to wait for the next image acquisition. If a hazard is detected, a detection box is provided at the hazard location, and the target type is determined, such as construction machinery, wires, foreign objects, smoke, or wildfire.

[0100] S2: When the hazard type is determined to be construction machinery or smoke and wildfire, execute S3; when the hazard type is determined to be wire or foreign object, the alarm level is determined to be a critical alarm and S4 is executed.

[0101] Different types of alarm objects are classified in different ways due to their varying degrees of threat. If it is a foreign object in the conductor, which poses an immediate threat to the safe operation of the conductor, it must be dealt with urgently and is therefore directly classified as a critical alarm, reminding maintenance personnel to take immediate action until the alarm is extinguished. If it is smoke, fire, or construction machinery, further analysis is required.

[0102] S3: Based on the movement trend of construction machinery or smoke and wildfire, and the relationship between the alarm location of construction machinery or smoke and wildfire and the protected area of ​​the overhead transmission line, the alarm level is determined as a general alarm, a key alarm, or a critical alarm.

[0103] The movement trends and relative locations of construction machinery, smoke, and wildfires pose varying threats to overhead power lines; therefore, it is necessary to determine the alarm level based on both movement trends and relative locations. While the types of smoke and construction machinery differ, the judgment logic is similar.

[0104] S4: For general alarms, no action is taken, and there is no need to report them to the maintenance personnel for follow-up. Wait for the next inspection cycle to start from S1.

[0105] For key alarms, the monitoring device performs timed and intensive snapshots, outputting key alarm information and timed and intensively captured images.

[0106] Once a critical alarm is triggered, the monitoring device is notified to take intensive snapshots to closely observe the further development and changes of the potential hazard. The image analysis monitoring rule is generally to take pictures every 10-30 minutes, but the interval for intensive monitoring can be shortened to 1-5 minutes. This interval is only an example and can be set according to actual circumstances. After intensive snapshots, the level of the critical alarm and the captured images are reported to the maintenance personnel for follow-up.

[0107] For emergency alarms, adjust the position and focus of the monitoring device, collect the video stream at the alarm location, analyze the video stream, and output the emergency alarm information and the analysis results of the video stream.

[0108] After the emergency alarm is activated, adjust the camera gimbal and camera focus, and initiate either real-time video analysis or silent video analysis. The purpose of adjusting the camera gimbal and focus is to keep the target object in the center of the frame and sufficiently large. The real-time and silent video analysis schemes are used to further detect potential hazards in the acquired video stream data, providing a more rigorous analysis of the hazards. The hazard identification method for video analysis is similar to that for image analysis. After analysis, the emergency alarm information and the video stream analysis results are reported to maintenance personnel for intervention. The video stream analysis results can be in different forms, such as real-time video, short video, sampled and synthesized short video, or dense image sequences.

[0109] The present invention has the following beneficial effects:

[0110] 1. This invention classifies the threat level of potential hazards into three levels: critical alarms, important alarms, and general alarms. Critical alarms are those requiring intervention from maintenance personnel; important alarms are those requiring close monitoring to prepare for potential crisis alarms; and general alarms are those that do not require attention and pose no threat to conductors or towers. This invention can conduct targeted analysis based on the type of hazard, fully exploring the hazard characteristics of transmission line service scenarios to define effective alarms. Only critical and important alarms are effective alarms requiring attention, while general alarms are ineffective alarms that do not require attention. Maintenance personnel only need to focus on key alarms and prioritize handling critical alarms; general alarms do not require attention. This reduces the workload of maintenance personnel in handling alarms, avoids overlooking alarms, and prevents serious accidents such as power outages, personal injury, or vehicle damage caused by alarms.

[0111] 2. Due to power consumption requirements, the monitoring device cannot operate for extended periods for video analysis. Furthermore, video analysis models generally have significantly lower recognition accuracy than image analysis models. Therefore, this invention divides the process into two parts: image analysis and video analysis. Normally, the monitoring device periodically captures and analyzes images. When a key alarm is detected, intensive image capture is performed. Only in the event of a critical alarm is video collected and analyzed. Through effective handling of key and critical alarms, effective monitoring of the development and changes of potential hazards can be achieved while conserving device power.

[0112] This invention defines a protection zone for overhead transmission lines, which can be determined empirically or set using image recognition methods. For example, the protection zone for an overhead transmission line can be set through the following process:

[0113] 1. Obtain target images including overhead transmission line conductors and towers. Detect the towers and conductors in the target images using tower detection models and conductor detection models respectively to obtain the tower positions and conductor positions.

[0114] Among them, pole detection can adopt a deep learning-based method. The pole detection model uses MobileNet as the backbone network and the loss function is GFLoss (General FocalLoss), specifically:

[0115] GFLoss=QFL+γDFL

[0116] The hyperparameter γ = 0.25, and the QFL and DFL are as follows:

[0117] QFL(p) = -|y - σ| β ((1-y)log(1-σ)+ylog(σ))

[0118] DFL(S i ,S i+1 )=-((y i+1 -y)log(S i )+(yy i )log(S i+1 ))

[0119]

[0120]

[0121] In QFL, y takes the value 0 or 1, representing the IoU value between the predicted bounding box and the ground truth bounding box. σ is the class prediction value after the classification branch passes through the sigmoid activation function, ranging from 0 to 1. β is a hyperparameter set to 2. In DFL, y is the prediction boundary value. iThe value of the actual border is the integer value after being divided equally. Here, the hyperparameter i takes the value of an integer from 1 to 16.

[0122] The conductor detection part also adopts a deep learning-based method, and the model settings are the same as those for the tower detection model. The difference is that the loss function is different during the training process of conductor detection. The loss function for conductor detection is CELoss.

[0123] 2. Determine the main tower based on the relative position information of the conductor and the tower, and use the conductor position to determine the outermost conductors on both sides of the main tower. Define the area between the main tower and the outermost conductor as the initial protection zone.

[0124] There are multiple towers in the image; we need to identify the towers corresponding to the conductors, i.e., the main towers. The main towers are shown below. Figure 2 As shown in the middle tower, the position can be determined based on the relationship between the conductor and the tower. Generally, the conductor is directly connected to the main tower and the position is close to it, so the position of the main tower can be used to determine its location.

[0125] The outermost guideline can be determined based on its slope. Due to the imaging characteristics of the camera, the guideline is projected centrally in the image. The inner and outer guidelines have different slopes, with the outer guideline having a smaller slope than the inner guideline. Figure 2 As shown. Therefore, the outermost conductors on both sides of the main tower can be determined based on the slope of the conductor.

[0126] Finally, the initial protection zone can be determined based on the main tower and the outermost conductor. The initial protection zone can be constructed by mapping the outermost two conductors to the bottom of the main tower, such as... Figure 2 As shown.

[0127] 3. Expand the initial protection zone according to the voltage level of the overhead transmission line to obtain the protection zone of the overhead transmission line.

[0128] Since the voltage levels of overhead transmission lines are different, their safety zones are also different. Therefore, the initial protection zone can be expanded outward according to the voltage level of the overhead transmission line to obtain the final protection zone.

[0129] The method of expansion can be manually adjusted based on experience, or it can be automatically expanded based on the initial protection zone.

[0130] For example, the initial protected area can be expanded outward through the following process:

[0131] 1. Set the expansion coefficient K and calculate the expansion distance R.

[0132] Where, (R=X) R -X L )*K / 2, X R and X LThese are the X coordinates of the right edge point and the left edge point of the initial protection zone, respectively.

[0133] 2. For each contour point P in the initial protection zone i When the contour point P i When the point is convex, concave, or flat, its expanded contour point Q is calculated using the following formulas (1), (2), and (3), respectively. i .

[0134]

[0135]

[0136]

[0137] in,

[0138] P i-1 and P i+1 These are the contour points P. i The previous contour point and the next contour point, θ i Let P be the line segment i-1 P i With P i+ 1P i The included angle, e i For P i-1 Point to P i unit vector, e i+1 For P i Point to P i+1 The unit vector.

[0139] Convex points, concave points, or flat points can be obtained by differentiating from the points near the contour points. The outward expansion of convex points, concave points, or flat points is as follows: Figure 3 As shown in (a), (b), and (c), an example of an expanded protective zone is shown below. Figure 4 As shown, the expansion coefficient K = 1.5.

[0140] As one embodiment of the present invention, the aforementioned S3 includes:

[0141] S31: Calculate the intersection-union ratio (IOU1) of the detection boxes at the alarm locations of smoke and wildfire in the two images before and after. Compare the IOU1 with the set threshold T1. If the IOU1 is less than T1, the alarm level is determined to be a key alarm. Otherwise, execute S32.

[0142] Intersection over Union (IOU) is used to represent the degree of overlap between two bounding boxes, as illustrated in the diagram below. Figure 5 As shown, its formula is as follows:

[0143]

[0144] IOU1 is used to judge the fire trend of fireworks. The detection box of the warning position of the smoke wildfire includes information on the flame and smoke areas, and the spread of the smoke can be used to judge whether the fire has spread. It is inferred by the intersection over union IOU1 of the two detection results before and after. If the intersection over union IOU1 < T1, it means that the fire has spread, and key warnings will be issued to remind the operation and maintenance personnel to continuously pay attention. If the intersection over union < T1, it is considered that the fire has not spread, and qualitative and quantitative analysis of S32 will be carried out.

[0145] Calculate the intersection over union IOU2 of the detection boxes of the warning positions of construction machinery in two consecutive images, and compare the intersection over union IOU2 with the set threshold T2. If IOU2 is greater than T2, the warning level is determined to be a general warning, otherwise execute S32.

[0146] The warning of construction machinery is similar to that of smoke wildfire. The movement trend of construction machinery is judged by the intersection over union IOU2. If IOU2 is less than T2 and the construction machinery has a large movement, qualitative or quantitative analysis of S32 will be carried out. If IOU2 is greater than T2 and the construction machinery hardly moves, a general warning will be issued.

[0147] The values of the thresholds T1 and T2 are adjustable. Exemplarily, 0.8 can be taken.

[0148] S32: Judge whether there is an intersection between the detection box of the warning position of construction machinery or smoke wildfire and the protection area of the overhead transmission line. If not, the warning level is determined to be a general warning. If so, execute S33.

[0149] This step is used to conduct qualitative analysis first to judge whether the warning position of construction machinery or smoke wildfire is within the protection area. The judgment method can use the positional relationship between the smoke detection box and the protection area position box. The specific method is to judge whether there is an intersection between the detection box and the protection area. If there is an intersection, it is considered that the construction machinery or smoke wildfire is at the edge or inside the protection area, and quantitative analysis of S33 will be carried out. If there is no intersection, it is considered that the hidden danger is far from the protection area, and a general warning is feedback.

[0150] S33: If the warning position of construction machinery or smoke wildfire is outside the protection area of the overhead transmission line or the minimum distance between the warning position of construction machinery or smoke wildfire and the conductor position is greater than the first distance, the warning level is determined to be a general warning. Among them, the second distance is less than the first distance.

[0151] For example, if the smoke or machinery occurs at the edge of the protection area and the minimum distance from the conductor exceeds 50 meters, a general warning will be issued.

[0152] If the alarm location of construction machinery or smoke / wildfire is outside the protection zone of the overhead transmission line, or if the closest distance between the alarm location of construction machinery or smoke / wildfire and the conductor location is between the second distance and the first distance, then the alarm level will be determined as a key alarm.

[0153] For example, a key warning will be issued if smoke or construction machinery occurs at the edge of the protected area or within a clearance distance of 30-50 meters.

[0154] If the alarm location of construction machinery or smoke / wildfire is located within the protection zone of overhead transmission lines, or if the nearest distance between the alarm location of construction machinery or smoke / wildfire and the conductor location is less than the second distance, the alarm level will be determined as a critical alarm.

[0155] For example, an emergency alarm will be triggered if smoke or construction machinery is inside the protected area or within 30 meters of the power line. The above range can be set according to actual needs and is not limited to the numbers 30 or 50.

[0156] This invention designs an effective alarm analysis method that utilizes multi-dimensional characteristic indicators of potential hazards, including historical alarm indicators, current alarm indicators, target movement state analysis, qualitative and quantitative estimation, etc., to classify alarm levels, prioritize sending critical alarms to monitoring personnel, and track critical alarms in real time. This achieves timely early warning of potential hazards while recording the process of critical alarms, preventing tripping accidents, and enabling traceability of the hazard process, thus protecting the safe and stable operation of transmission lines.

[0157] As another improvement to this embodiment of the invention, the video stream can be analyzed through the following process:

[0158] The system acquires each frame of the video stream, detects potential hazards in each frame using a video detection model, performs continuous analysis on the detected hazards using a target tracking model, and then calculates the motion state of construction machinery or the spread trend of smoke and wildfires using a trend determination method.

[0159] The fundamental methods relied upon in this invention, such as target detection methods, target tracking methods, smoke diffusion methods, and construction machinery motion analysis methods, are not specifically limited. For example, target tracking algorithms can be common single-target or multi-target algorithms such as KCF, Sort, and DeepSort. Smoke diffusion and the motion state analysis of construction machinery can be performed using methods based on optical flow or bounding box movement. Real-time video analysis methods analyze at least 15 frames per second, while silent video analysis methods analyze up to 10 images per second, such as one image per second.

[0160] Furthermore, the position of the monitoring device can be adjusted through the following process:

[0161] Calculate the X-direction adjustment scale Angle_x and Y-direction adjustment scale Angle_y of the monitoring device, and adjust the position of the monitoring device according to Angle_x and Angle_y.

[0162] Angle_x=(bbox_center_x-image_center_x)*FOV_width / width

[0163] Angle_y=(bbox_center_y-image_center_y)*FOV_height / height

[0164] Where bbox_center_x and bbox_center_y are the X and Y coordinates of the center of the alarm location, respectively; image_center_x and image_center_y are the X and Y coordinates of the center of the image; FOV_width and FOV_height are the width and height viewing angles of the monitoring device, respectively. These can be represented by a lookup table, where width and height are the width and height of the image, respectively. For cameras with non-uniform pixel distribution and viewing angles, using a lookup table to calculate the offset angles corresponding to these pixels is more accurate.

[0165] Accordingly, the focal length of the monitoring device can be adjusted through the following process:

[0166] Calculate the new focal length New_focal of the monitoring device, and adjust the focal length of the monitoring device according to the new focal length New_focal.

[0167] New_focal=current_focal+bbox_width / width*scale

[0168] Here, `current_focal` is the current focal length of the monitoring device, `bbox_width` is the width of the monitoring device, and `scale` is the set adjustment ratio coefficient. `scale` is related to the position of the bottom of the target; the closer the bottom of the target is to the top of the image, the farther the target is from the camera, and the larger the ratio. Different cameras have different parameters.

[0169] This invention also provides a visual monitoring and hazard identification device for overhead transmission lines, such as... Figure 6 As shown, the device includes:

[0170] The hazard detection module 1 is used to acquire images captured by the monitoring device at regular intervals, and to detect hazards in the images using a trained hazard recognition model to determine whether there are hazards and the type of hazard. Hazard types include construction machinery, wires and foreign objects, and smoke and wildfires.

[0171] The first-level judgment module 2 is used to execute the second-level judgment module 3 when the hazard type is judged to be construction machinery or smoke and wildfire; when the hazard type is judged to be wire or foreign object, the alarm level is judged to be a critical alarm and the alarm processing module 4 is executed.

[0172] The second-level judgment module 3 is used to determine the alarm level as a general alarm, a key alarm, or a critical alarm based on the movement trend of construction machinery or smoke and wildfire and the positional relationship between the alarm location of construction machinery or smoke and wildfire and the protection zone of the set overhead transmission line.

[0173] Alarm processing module 4 is used to handle general alarms without processing them, waiting for the next inspection cycle to be executed by hazard detection module 1; for key alarms, it uses a monitoring device to perform timed and intensive snapshots, outputting key alarm information and timed and intensively snapshot images; for critical alarms, it adjusts the position and focus of the monitoring device, collects video streams of the alarm location, analyzes the video streams, and outputs critical alarm information and video stream analysis results.

[0174] This invention defines a protection zone for overhead transmission lines, which can be determined empirically or set using image recognition methods. For example, the protection zone for an overhead transmission line can be set through the following process:

[0175] The target image, including the conductors and towers of the overhead transmission line, is acquired. The towers and conductors in the target image are detected by the tower detection model and the conductor detection model, respectively, to obtain the tower position and conductor position.

[0176] The main tower is determined based on the relative position information of the conductor and the tower. The outermost conductors on both sides of the main tower are determined using the conductor position. The area between the main tower and the outermost conductor is defined as the initial protection zone.

[0177] The initial protection zone is expanded outward based on the voltage level of the overhead transmission line to obtain the protection zone of the overhead transmission line.

[0178] Furthermore, the initial protected area can be expanded outward through the following process:

[0179] Set the expansion coefficient K and calculate the expansion distance R.

[0180] Where, (R=X) R -X L )*K / 2, X R and X L These are the X coordinates of the right edge point and the left edge point of the initial protection zone, respectively.

[0181] For each contour point P in the initial protection zone iWhen the contour point P i When the point is convex, concave, or flat, its expanded contour point Q is calculated using the following formulas (1), (2), and (3), respectively. i .

[0182]

[0183]

[0184]

[0185] in,

[0186] P i-1 and P i+1 These are the contour points P. i The previous contour point and the next contour point, θ i Let P be the line segment i-1 P i With P i+ 1P i The included angle, e i For P i-1 Point to P i unit vector, e i+1 For P i Point to P i+1 The unit vector.

[0187] As an improvement to this embodiment of the invention, the aforementioned second-level judgment module includes:

[0188] The motion trend judgment unit is used to calculate the intersection-union ratio (IOU1) of the detection boxes of the smoke and wildfire alarm locations in the two images before and after. The IOU1 is compared with the set threshold T1. If the IOU1 is less than T1, the alarm level is judged as a key alarm. Otherwise, the qualitative analysis unit is executed.

[0189] Calculate the intersection-union ratio (IOU2) of the detection boxes at the alarm locations of the construction machinery in the two images before and after. Compare the IOU2 with the set threshold T2. If the IOU2 is greater than T2, the alarm level is determined to be a general alarm. Otherwise, the qualitative analysis unit is executed.

[0190] The qualitative analysis unit is used to determine whether the detection frame of the alarm location of construction machinery or smoke and wildfire overlaps with the protection zone of overhead transmission lines. If not, the alarm level is determined to be a general alarm; if so, the quantitative analysis unit is executed.

[0191] The quantitative analysis unit is used to determine the alarm level as a general alarm if the alarm location of construction machinery or smoke / wildfire is located outside the protection zone of the overhead transmission line or the nearest distance between the alarm location of construction machinery or smoke / wildfire and the conductor location is greater than a first distance.

[0192] If the alarm location of construction machinery or smoke / wildfire is outside the protection zone of the overhead transmission line, or if the closest distance between the alarm location of construction machinery or smoke / wildfire and the conductor location is between the second distance and the first distance, then the alarm level will be determined as a key alarm.

[0193] If the alarm location of construction machinery or smoke / wildfire is located within the protection zone of overhead transmission lines, or if the nearest distance between the alarm location of construction machinery or smoke / wildfire and the conductor location is less than the second distance, the alarm level will be determined as a critical alarm.

[0194] The second distance is smaller than the first distance.

[0195] As another improvement to this embodiment of the invention, the video stream can be analyzed through the following process:

[0196] The video detection model detects potential hazards in each frame of the video stream, the target tracking model performs continuous analysis of the detected hazards, and the trend determination method calculates the motion status of construction machinery or the spread trend of smoke and wildfire.

[0197] Furthermore, the position of the monitoring device can be adjusted through the following process:

[0198] Calculate the X-direction adjustment scale Angle_x and Y-direction adjustment scale Angle_y of the monitoring device, and adjust the position of the monitoring device according to Angle_x and Angle_y.

[0199] Angle_x=(bbox_center_x-image_center_x)*FOV_width / width

[0200] Angle_y=(bbox_center_y-image_center_y)*FOV_height / height

[0201] Where bbox_center_x and bbox_center_y are the X and Y coordinates of the center of the alarm location, image_center_x and image_center_y are the X and Y coordinates of the center of the image, FOV_width and FOV_height are the width and height viewing angles of the monitoring device, and width and height are the width and height of the image, respectively.

[0202] Accordingly, the focal length of the monitoring device can be adjusted through the following process:

[0203] Calculate the new focal length New_focal of the monitoring device, and adjust the focal length of the monitoring device according to the new focal length New_focal.

[0204] New_focal=current_focal+bbox_width / width*scale

[0205] Where current_focal is the current focal length of the monitoring device, bbox_width is the width of the monitoring device, and scale is the set adjustment ratio coefficient.

[0206] The apparatus provided in the above embodiments corresponds one-to-one with the embodiments of the aforementioned methods in terms of its implementation principle and the resulting technical effects. For the sake of brevity, any parts of the apparatus not mentioned in the embodiments can be referred to the corresponding content in the embodiments of the aforementioned methods. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the modules and units described in this apparatus can all be referred to the corresponding processes in the embodiments of the aforementioned methods, and will not be repeated here.

[0207] The methods described in the above embodiments of the present invention can implement business logic through a computer program and record it on a storage medium. This storage medium can be read and executed by a computer, achieving the effects of the scheme described in the method embodiments of this specification. Therefore, the embodiments of the present invention also provide a computer-readable storage medium for visual monitoring and hazard identification of overhead transmission lines, including a memory for storing processor-executable instructions. When these instructions are executed by the processor, they implement the steps of the visual monitoring and hazard identification method for overhead transmission lines as described in the foregoing embodiments.

[0208] The storage medium may include a physical device for storing information, typically digitizing the information and then storing it using electrical, magnetic, or optical methods. The storage medium may include: devices that store information using electrical energy, such as various types of memory, like RAM and ROM; devices that store information using magnetic energy, such as hard disks, floppy disks, magnetic tapes, magnetic core memory, bubble memory, and USB flash drives; and devices that store information using optical methods, such as CDs or DVDs. Of course, there are other readable storage media, such as quantum memories and graphene memories.

[0209] The storage medium described above may also include other implementation methods according to the description of the method embodiments. The implementation principle and technical effects of this embodiment are the same as those of the foregoing method embodiments. For details, please refer to the description of the relevant method embodiments, which will not be repeated here.

[0210] This invention also provides a device for visual monitoring and hazard identification of overhead transmission lines. The device can be a standalone computer, or it can include an actual operating device that uses one or more of the methods or embodiments described in this specification. The device for visual monitoring and hazard identification of overhead transmission lines may include at least one processor and a memory storing computer-executable instructions. When the processor executes the instructions, it implements the steps of any one or more of the methods for visual monitoring and hazard identification of overhead transmission lines.

[0211] The device described above may also include other implementation methods according to the description of the method embodiments. The implementation principle and technical effects of this embodiment are the same as those of the foregoing method embodiments. For details, please refer to the description of the relevant method embodiments, which will not be repeated here.

[0212] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments within the scope of the technology disclosed in the present invention, or make equivalent substitutions for some of the technical features; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. All should be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for identifying potential hazards in overhead transmission lines through visual monitoring, characterized in that, The method includes: S1: Acquire images captured periodically by a monitoring device, and use a trained hazard identification model to detect hazards in the images, determine whether there are hazards and the type of hazard; wherein, the hazard types include construction machinery, wires and foreign objects, and smoke and wildfires; S2: When the hazard type is determined to be construction machinery or smoke / wildfire, execute S3; when the hazard type is determined to be wire or foreign object, the alarm level is determined to be a critical alarm, and execute S4. S3: Based on the movement trend of the construction machinery or smoke and wildfire and the relationship between the alarm location of the construction machinery or smoke and wildfire and the protected area of ​​the set overhead transmission line, the alarm level is determined as a general alarm, a key alarm or a critical alarm. S4: For general alarms, no action is taken, and the process waits for the next inspection cycle to start from S1; for key alarms, the monitoring device performs timed and intensive snapshots, and outputs key alarm information and timed and intensively snapshot images; for critical alarms, the position and focus of the monitoring device are adjusted, the video stream of the alarm location is collected, the video stream is analyzed, and critical alarm information and video stream analysis results are output. The protection zone of the overhead transmission line is set through the following process: Acquire target images including conductors and towers of overhead transmission lines, and detect the towers and conductors in the target images using tower detection models and conductor detection models respectively to obtain the tower positions and conductor positions; The main tower is determined based on the relative position information of the conductor and the tower. The outermost conductors on both sides of the main tower are determined using the conductor positions. The area between the main tower and the outermost conductors is defined as the initial protection zone. The initial protection zone is formed by mapping the outermost conductors on both sides to the bottom of the main tower. The initial protection zone is expanded outward according to the voltage level of the overhead transmission line to obtain the protection zone of the overhead transmission line. The initial protected area is expanded outward through the following process: Set the expansion coefficient K and calculate the expansion distance R; Where, (R= X R - X L ) K / 2, X R and X L These are the X coordinates of the right edge point and the left edge point of the initial protection zone, respectively; For each contour point P of the initial protection zone i When the contour point P i When the point is a convex point, a concave point, or a flat point, its expanded contour point Q is calculated using the following formulas (1), (2), and (3), respectively. i ; Equation (1) Equation (2) Equation (3) in, , ; P i-1 and P i+1 These are the contour points P. i The previous contour point and the next contour point, θ i Let P be the line segment i-1 P i With P i+1 P i The included angle, e i For P i-1 Point to P i unit vector, e i+1 For P i Point to P i+1 , unit vector; S3 includes: S31: Calculate the intersection-union ratio (IOU1) of the detection boxes of the smoke and wildfire alarm locations in the two images before and after, and compare the IOU1 with the set threshold T1. If IOU1 is less than T1, the alarm level is determined to be a key alarm; otherwise, execute S32. Calculate the intersection-union ratio (IOU2) of the detection boxes of the alarm positions of the construction machinery in the two images before and after. Compare the IOU2 with the set threshold T2. If IOU2 is greater than T2, the alarm level is determined to be a general alarm. Otherwise, execute S32. S32: Determine whether the detection frame of the alarm location of construction machinery or smoke and wildfire overlaps with the protection zone of the overhead transmission line. If not, the alarm level is determined to be a general alarm. If yes, execute S33. S33: If the nearest distance between the alarm position of construction machinery or smoke / wildfire and the position of the conductor is greater than the first distance, the alarm level will be determined as a general alarm. If the nearest distance between the alarm position of construction machinery or smoke / wildfire and the position of the power line is between the second distance and the first distance, the alarm level will be determined as a critical alarm. If the alarm location of construction machinery or smoke / wildfire is located within the protection zone of overhead transmission line, or if the nearest distance between the alarm location of construction machinery or smoke / wildfire and the conductor location is less than the second distance, the alarm level will be determined as a critical alarm. Wherein, the second distance is less than the first distance.

2. The method for identifying hidden dangers in overhead transmission lines through visual monitoring according to claim 1, characterized in that, The video stream is analyzed using the following process: The video detection model detects potential hazards in each frame of the video stream, the target tracking model performs continuous analysis of the detected hazards, and the trend determination method calculates the motion state of construction machinery or the spread trend of smoke and wildfire.

3. The method for identifying hidden dangers in overhead transmission lines through visual monitoring according to claim 2, characterized in that, Adjust the position of the monitoring device using the following procedure: Calculate the X-direction adjustment scale Angle_x and Y-direction adjustment scale Angle_y of the monitoring device, and adjust the position of the monitoring device according to Angle_x and Angle_y; Angle_x=(bbox_center_x-image_center_x) FOV_width / width Angle_y=(bbox_center_y-image_center_y) FOV_height / height Where bbox_center_x and bbox_center_y are the X and Y coordinates of the center of the alarm location, image_center_x and image_center_y are the X and Y coordinates of the center of the image, FOV_width and FOV_height are the width and height viewing angles of the monitoring device, and width and height are the width and height of the image, respectively. Adjust the focal length of the monitoring device using the following procedure: Calculate the new focal length New_focal of the monitoring device, and adjust the focal length of the monitoring device according to the new focal length New_focal; New_focal=current_focal+bbox_width / width scale Where current_focal is the current focal length of the monitoring device, bbox_width is the width of the monitoring device, and scale is the set adjustment ratio coefficient.

4. A visual monitoring and hazard identification device for overhead transmission lines, characterized in that, The device includes: The hazard detection module is used to acquire images captured periodically by a monitoring device, and to detect hazards in the images using a trained hazard recognition model to determine whether there are hazards and the type of hazard; wherein, the hazard types include construction machinery, wires and foreign objects, and smoke and wildfires; The first-level judgment module is used to execute the second-level judgment module when the hazard type is judged to be construction machinery or smoke and wildfire; when the hazard type is judged to be wire or foreign object, the alarm level is judged to be a critical alarm and the alarm processing module is executed. The second-level judgment module is used to determine the alarm level as a general alarm, a key alarm, or a critical alarm based on the movement trend of the construction machinery or smoke and wildfire and the positional relationship between the alarm location of the construction machinery or smoke and wildfire and the protection zone of the set overhead transmission line. The alarm processing module is used to leave general alarms unprocessed and wait for the next inspection cycle to be executed by the hidden danger detection module; for key alarms, it uses a monitoring device to take timed and intensive snapshots and outputs key alarm information and timed and intensive snapshot images; for critical alarms, it adjusts the position and focus of the monitoring device to collect video streams of the alarm location, analyzes the video streams, and outputs critical alarm information and video stream analysis results. The protection zone of the overhead transmission line is set through the following process: Acquire target images including conductors and towers of overhead transmission lines, and detect the towers and conductors in the target images using tower detection models and conductor detection models respectively to obtain the tower positions and conductor positions; The main tower is determined based on the relative position information of the conductor position and the tower position, and the outermost conductors on both sides of the main tower are determined using the conductor position. The area between the main tower and the outermost conductor is determined as the initial protection zone. The initial protection zone is expanded outward according to the voltage level of the overhead transmission line to obtain the protection zone of the overhead transmission line. The initial protected area is expanded outward through the following process: Set the expansion coefficient K and calculate the expansion distance R; Where, (R= X R - X L ) K / 2, X R and X L These are the X coordinates of the right edge point and the left edge point of the initial protection zone, respectively; For each contour point P of the initial protection zone i When the contour point P i When the point is a convex point, a concave point, or a flat point, its expanded contour point Q is calculated using the following formulas (1), (2), and (3), respectively. i ; Equation (1) Equation (2) Equation (3) in, , ; P i-1 and P i+1 These are the contour points P. i The previous contour point and the next contour point, θ i Let P be the line segment i-1 P i With P i+1 P i The included angle, e i For P i-1 Point to P i unit vector, e i+1 For P i Point to P i+1 , unit vector; The second-level judgment module includes: The motion trend judgment unit is used to calculate the intersection-union ratio (IOU1) of the detection boxes of the alarm positions of smoke and wildfire in the two images before and after. The IOU1 is compared with the set threshold T1. If the IOU1 is less than T1, the alarm level is judged as a key alarm. Otherwise, the qualitative analysis unit is executed. Calculate the intersection-union ratio (IOU2) of the detection boxes of the alarm positions of the construction machinery in the two images before and after. Compare the IOU2 with the set threshold T2. If IOU2 is greater than T2, the alarm level is determined to be a general alarm. Otherwise, the qualitative analysis unit is executed. The qualitative analysis unit is used to determine whether the detection frame of the alarm location of construction machinery or smoke and wildfire overlaps with the protection zone of the overhead transmission line. If not, the alarm level is determined to be a general alarm; if so, the quantitative analysis unit is executed. The quantitative analysis unit is used to determine the alarm level as a general alarm if the nearest distance between the alarm position of construction machinery or smoke and wildfire and the position of the guide wire is greater than the first distance. If the nearest distance between the alarm position of construction machinery or smoke / wildfire and the position of the power line is between the second distance and the first distance, the alarm level will be determined as a critical alarm. If the alarm location of construction machinery or smoke / wildfire is located within the protection zone of overhead transmission line, or if the nearest distance between the alarm location of construction machinery or smoke / wildfire and the conductor location is less than the second distance, the alarm level will be determined as a critical alarm. Wherein, the second distance is less than the first distance.

5. A computer-readable storage medium for visual monitoring and hazard identification of overhead transmission lines, characterized in that, It includes a memory for storing processor-executable instructions, which, when executed by the processor, implement the steps of the overhead transmission line visual monitoring and hidden danger identification method according to any one of claims 1-3.

6. A device for visual monitoring and hazard identification of overhead transmission lines, characterized in that, It includes at least one processor and a memory storing computer-executable instructions, wherein the processor executes the instructions to implement the steps of the overhead transmission line visual monitoring and hidden danger identification method according to any one of claims 1-3.

Images