Power transmission line video on-line monitoring device and system

Abstract

The invention belongs to the technical field of power transmission line monitoring, and particularly relates to a power transmission line video online monitoring device and system, which comprises an acquisition module, an analysis module, a prediction module, an early warning module, an equipment health management module and an intelligent decision support module, the prediction module in the power transmission line video on-line monitoring device combines real-time data with historical rules through the short-term hidden danger prediction sub-module, can identify potential hidden dangers in advance and reserve processing time for operation and maintenance personnel, and then can formulate precautionary measures in advance through the long-term trend analysis sub-module, namely, tree pruning is strengthened before rainy seasons, so that the early warning effect is achieved. Mountain fire monitoring is enhanced before winter, hidden danger risks are reduced from a long-term perspective, meanwhile, the early warning module triggers multi-level alarm according to abnormal levels, a solution suggestion library is generated, deep integration with a GIS platform is achieved, a hidden danger influence range thermodynamic diagram is automatically generated, and the timeliness and accuracy of early warning are improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of power transmission line monitoring, and in particular relates to a video online monitoring device and system for power transmission lines. Background Art

[0002] Transmission lines are constructed by using transformers to boost the voltage of electricity generated by generators, then connecting it to the transmission line via control devices such as circuit breakers. Transmission lines are categorized by structure as overhead transmission lines and cable lines. Overhead transmission lines are constructed above ground and consist of towers, conductors, insulators, line hardware, guy wires, tower foundations, and grounding devices. Inspection and monitoring are required to ensure proper functioning of these lines.

[0003] Chinese invention patent publication number CN119520736A discloses an online video image monitoring device for power transmission lines, belonging to the field of power transmission line monitoring technology. The device comprises an online detection board, an image recognition module mounted below the board, a stepper motor mounted above the module for adjusting the position and angle of the module, a power transmission conductor mounted below the module, and a line clamping plate for mounting a physical detection structure fixed to the outside of the conductor. The key technical solution addresses the issues with existing online image monitoring devices, such as cumbersome detection methods, inability to adapt to different monitoring environments, and poor adaptability during actual use.

[0004] However, the above technologies often have the following defects: the existing technologies lack effective models and methods for predicting the development trend of hidden dangers in transmission lines, making it difficult to take measures in advance to prevent accidents. In addition, the previous early warning mechanism has a single alarm level and cannot accurately warn according to the severity of the hidden dangers.

[0005] To this end, the present invention provides a transmission line video online monitoring device and system. Summary of the Invention

[0006] In order to make up for the deficiencies of the prior art, at least one technical problem raised in the background technology is solved.

[0007] The technical solution adopted by the present invention to solve the technical problem is as follows: the transmission line video online monitoring device of the present invention includes: an acquisition module, an analysis module, a prediction module, an early warning module, an equipment health management module and an intelligent decision support module; Acquisition module: used for all-weather video acquisition, supporting dual-mode data fusion of thermal imaging and visible light; Analysis module: Based on deep learning algorithms, it processes video streams in real time, identifies anomaly types, and outputs anomaly levels; Prediction module: Combining historical data with real-time data, using a spatiotemporal fusion prediction model to analyze and predict the development trend of hidden dangers; Equipment health management module: used to automatically generate treatment suggestions and emergency plans for abnormal transmission line conditions based on monitoring data and preset rules; Early warning module: triggers multi-level alarms based on the abnormality level and generates a solution suggestion library.

[0008] Furthermore, the acquisition module integrates a multispectral sensor with a wavelength range of 8-14μm and a thermal sensitivity of ≤50mK. It distinguishes smoke and fog by the difference in spectral reflectivity, where the distinction is based on: smoke reflectivity <15%, cloud reflectivity >30%. The acquisition module also includes a high-resolution infrared detector and a 4K visible light camera, supports HDR imaging, and supports adaptive exposure adjustment, and dynamically adjusts the infrared gain and visible light shutter speed. The acquisition module also includes an anti-shake gimbal, which uses three-axis gyroscope stabilization technology.

[0009] Furthermore, the analysis module detects abnormal targets in the video in real time based on the YOLOv7 deep learning model; integrates the DeepSORT algorithm to predict the trajectory of cranes and tower cranes, Position coordinates at a given moment , speed record , according to the DeepSORT algorithm, it is predicted that Position coordinates at the moment for: Calculate the minimum spatial distance between it and the wire, support multimodal data fusion, combine thermal imaging temperature data with visible light images, enhance wildfire identification capabilities, and the thermal imaging temperature is -20℃-150℃.

[0010] Furthermore, the analysis module is deployed on the edge computing device, and accelerated inference technology is used to reduce latency. The deployed edge computing device uses TensorRT to accelerate inference, with a latency of <200ms.

[0011] Furthermore, the spatiotemporal fusion prediction model includes: Short-term hidden danger prediction submodule: Based on the LSTM neural network, it inputs real-time monitoring data and outputs a hidden danger probability score within the next two hours; The input data includes: Using laser rangefinders and image recognition technology to monitor tree growth rates in real time; Use the global positioning system to obtain real-time location information of cranes and tower cranes; Utilize wind speed, humidity, and temperature data obtained from IoT sensors; Collect historical hidden danger database, including hidden danger type, occurrence time, location, severity and treatment results; Output hidden danger development trend score, triggering an early warning when the score exceeds the threshold, where the score range is 1-100; Long-term trend analysis submodule: Integrates the Prophet algorithm to analyze the periodic patterns in the historical hidden danger database and predict the accelerated growth of trees in the rainy season and the high incidence of wildfires in winter. The basic model of the Prophet algorithm is: in is a time series in time The observed value of is the trend term, which represents the long-term trend of the time series; is the seasonal term, which represents the periodic changes of the time series; is the holiday term, which indicates the impact of holidays on the time series; is the error term, which represents the random fluctuations in the time series that cannot be explained by the model.

[0012] Furthermore, the device health management module includes: Condition monitoring unit: used to collect equipment parameters such as conductor sag, insulator contamination, and hardware corrosion rate in real time, with a sampling frequency of ≥1Hz; Remaining life prediction unit: Calculates the remaining service life based on the equipment's degradation rate. Remaining service life: in, is the initial life of the equipment, The equipment has reached its service life. is the degradation rate; The degradation rate function is calculated by a random forest regression model with the following input parameters: Material properties, environmental stresses, mechanical loads; Maintenance decision-making unit: dynamically generate maintenance plans based on calculation results and process calculation results in sequence Devices < 6 months old.

[0013] Furthermore, the intelligent decision support module includes: Cross-system linkage engine: triggers regional collaborative protection when the following conditions are met simultaneously: If the environmental hazard score is ≥85 and the equipment service life is less than 3 months, cross-region load transfer will be initiated; Emergency plan library: preset disposal plans according to voltage levels.

[0014] Further, the multi-level alarm mechanism of the early warning module comprises: Primary alarm, score >= 70: sound and light alarm + local log recording; Secondary alarm, score >= 85: send SMS to operation and maintenance personnel, with abnormal screenshot and GPS coordinates; Tertiary alarm, score >= 95: automatically link to power grid control system, execute emergency shutdown, and need manual review and permission confirmation; The early warning module is deeply integrated with the GIS platform, automatically generates a hidden danger influence range heat map, and superimposes a power transmission line topology map, the color classification of the heat map can be dynamically adjusted according to the severity of the hidden danger; the power transmission line topology map includes the position information, tower height and conductor sag information of the tower, and the update cycle of the topology map is 7-25 days.

[0015] The power transmission line video online monitoring system comprises the power transmission line video online monitoring device and a server in wireless communication with the power transmission line video online monitoring device, and the server is used for receiving image data of the power transmission line uploaded by the power transmission line video online monitoring device and marked with an abnormal target.

[0016] Further, the server is also used for sending a neural network model update instruction to the power transmission line video online monitoring device, the neural network model update instruction is used for updating an abnormal target type, and the wireless communication between the server and the power transmission line video online monitoring device adopts an AES-256 encryption algorithm for data encryption transmission, wherein if plaintext data is , the encryption key is , the encrypted ciphertext data is , and the encryption process satisfies , wherein is an encryption function of the AES-256 encryption algorithm.

[0017] The beneficial effects of the present application are as follows: 1. The prediction module combines real-time data with historical rules through a short-term hidden danger prediction submodule, which can identify potential hidden dangers in advance and reserve disposal time for operation and maintenance personnel, and through a long-term trend analysis submodule, preventive measures can be developed in advance, i.e., tree pruning is strengthened before the rainy season and mountain fire monitoring is strengthened before winter, thereby reducing hidden danger risks from a long-term perspective, and the early warning module triggers a multi-level alarm according to the abnormal level and generates a solution suggestion library, and is deeply integrated with the GIS platform to automatically generate a hidden danger influence range heat map and improve the timeliness and accuracy of early warning based on the heat map; 2. A multi-level alarm mechanism is adopted, that is, responses are graded according to severity. Low-risk events do not require notification of all employees, and high-risk events are quickly escalated to reduce ineffective responses. Different levels are matched with different measures, such as direct shutdown for level 3 alarms, to avoid electric shock accidents, reduce the risk of human misjudgment, and thereby further improve safety and the accuracy of early warnings. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The present invention will be further described below with reference to the accompanying drawings.

[0019] Figure 1 is a diagram of the device architecture of the present invention; Figure 2 This is the LSTM network structure diagram of the present invention. DETAILED DESCRIPTION

[0020] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0021] See also Figure 1-2 : This embodiment provides: a transmission line video online monitoring device, including: an acquisition module, an analysis module, a prediction module, an early warning module, an equipment health management module and an intelligent decision support module; Acquisition module: used for all-weather video acquisition, supporting dual-mode data fusion of thermal imaging and visible light; The acquisition module integrates a multispectral sensor with a wavelength range of 8-14μm and a thermal sensitivity of ≤50mK. It distinguishes smoke and fog by the difference in spectral reflectance. The distinction is based on the following criteria: smoke reflectance <15%, cloud reflectance >30%. For example, in a certain monitoring, the reflectance of a certain area was detected to be 12%. Based on the distinction criteria of smoke reflectance <15% and cloud reflectance >30%, the area is judged to be a smoke area.

[0022] The acquisition module also includes a high-resolution infrared detector and a 4K visible light camera, supports HDR imaging, and supports adaptive exposure adjustment. It can dynamically adjust the infrared gain and visible light shutter speed. The acquisition module also includes an anti-shake gimbal, which uses three-axis gyroscope stabilization technology.

[0023] For example, when the ambient light suddenly becomes stronger, adjust the visible light shutter speed from 1 / 125s to 1 / 500s, and appropriately reduce the infrared gain to ensure image quality.

[0024] Analysis module: Based on deep learning algorithms, it processes video streams in real time, identifies anomaly types, and outputs anomaly levels; The analysis module is based on the YOLOv7 deep learning model to detect abnormal targets in the video in real time; the DeepSORT algorithm is integrated to predict the trajectory of cranes and tower cranes. Position coordinates at a given moment , speed record , according to the DeepSORT algorithm, it is predicted that Position coordinates at the moment for: Calculate the minimum spatial distance between it and the wire, For example, a crane is detected in the video, and the DeepSORT algorithm is integrated to predict the crane's trajectory. Assume that the crane's position coordinates at time t=0 are (x0=100,y0=200), its speed is (vx,0=5,vy,0=3), and the time interval is =1s, according to the formula , the crane's position coordinates at time t=1s can be predicted to be (x1=100+5×1=105, y1=200+3×1=203). The minimum distance between the crane and the conductor is then calculated. If the minimum distance is less than 5m (the safety distance threshold), a safety hazard is identified.

[0025] It also supports multimodal data fusion, combining thermal imaging temperature data with visible light images to enhance wildfire identification capabilities. The thermal imaging temperature is -20°C-150°C. For example: during the monitoring process, the thermal imaging sensor detects a temperature of T=80°C in a certain area, and the visible light image shows that the area has smoke and flame characteristics. Through multimodal data fusion analysis, wildfires can be identified more accurately.

[0026] The analysis module is deployed on edge computing devices and uses accelerated inference technology to reduce latency. The deployed edge computing devices use TensorRT to accelerate inference, with a latency of <200ms.

[0027] Prediction module: Combining historical data with real-time data, using a spatiotemporal fusion prediction model to analyze and predict the development trend of hidden dangers; The spatiotemporal fusion prediction model includes: Short-term hidden danger prediction submodule: Based on the LSTM neural network, it inputs real-time monitoring data and outputs a hidden danger probability score within the next two hours; The input data includes: Using laser rangefinders and image recognition technology to monitor tree growth rates in real time; Use the global positioning system to obtain real-time location information of cranes and tower cranes; Utilize wind speed, humidity, and temperature data obtained from IoT sensors; Collect historical hidden danger database, including hidden danger type, occurrence time, location, severity and treatment results; Output a hazard development trend score. When the score exceeds the threshold, an alert is triggered. The score range is 1-100. For example, a laser rangefinder monitors a tree growth rate of Vtree = 0.02 m / month, a global positioning system is used to obtain the crane location information (x = 150, y = 250), and an IoT sensor obtains wind speed Vwind = 5 m / s, humidity H = 60%, and temperature Tair = 25°C. At the same time, a historical hazard database is collected, including the hazard type Type (such as tree fall, crane line collision, etc.), occurrence time toccur, location (xoccur, yoccur), severity S (level 1-10), and processing result Result. This data is combined into an input vector: , input LSTM neural network, output hidden danger probability score P in the next 2 hours, if the score P=80 (exceeds the threshold Pthreshold=70), then trigger the warning.

[0028] Long-term trend analysis submodule: Integrates the Prophet algorithm to analyze the periodic patterns in the historical hidden danger database and predict the accelerated growth of trees in the rainy season and the high incidence of wildfires in winter. The basic model of the Prophet algorithm is: in is a time series in time The observed value of is the trend term, which represents the long-term trend of the time series; is the seasonal term, which represents the periodic changes of the time series; is the holiday term, which indicates the impact of holidays on the time series; The error term represents the random fluctuations in the time series that cannot be explained by the model. For example, by analyzing historical data over many years, it is found that tree growth accelerates during the rainy season (June to August) and wildfires are more frequent in winter (December to February). According to the basic model of the Prophet algorithm , we can predict the trend of hidden dangers in the future rainy season and winter, and take preventive measures in advance.

[0029] Equipment health management module: used to automatically generate treatment suggestions and emergency plans for abnormal transmission line conditions based on monitoring data and preset rules; Equipment health management module, including: Condition monitoring unit: used to collect equipment parameters such as conductor sag, insulator contamination, and hardware corrosion rate in real time, with a sampling frequency of ≥1Hz; Remaining life prediction unit: Calculates the remaining service life based on the equipment's degradation rate. Remaining service life: in, is the initial life of the equipment, The equipment has reached its service life. is the degradation rate; The degradation rate function formula is expressed as: in, Remaining service life refers to the length of time that a device or component can still work normally from the current moment.

[0030] It is a time variable, representing the length of time from the start of monitoring to the current moment, or the upper limit of the time range used for integration.

[0031] The total life of a device or component is a fixed time value, which represents the expected total length of time from the start of use to the complete failure of the device.

[0032] It may represent a set of parameters that affect device degradation, such as environmental factors (temperature, humidity, etc.), workload, etc. These parameters will affect the degradation rate of the device.

[0033] is the degradation rate function, which is a function of the total life and parameter collection This function is used to describe the degradation rate of a device under different conditions. The output value of this function reflects the degradation rate of the device at a specific moment.

[0034] Is the integral symbol, indicating the function behind In the time interval from 0 to The result of the integration is used to calculate the remaining service life. By integrating the inverse of the degradation rate, the "effective life" of the equipment over a period of time can be accumulated.

[0035] The degradation rate function is calculated by the random forest regression model with the following input parameters: Material properties (conductor type, insulator material), environmental stresses (salt density, dust density, UV intensity), mechanical loads (wind vibration frequency, ice thickness); Maintenance decision-making unit: dynamically generate maintenance plans based on calculation results and process calculation results in sequence Devices < 6 months old.

[0036] It should be noted that the prediction module focuses on environmental hazards (external threats); the health management module focuses on equipment degradation (internal status); although data is shared through the GIS platform, the analysis objectives are separated.

[0037] Early warning module: triggers multi-level alarms based on the abnormality level and generates a solution suggestion library.

[0038] The early warning module is deeply integrated with the GIS platform, automatically generating a heat map of the impact range of hidden dangers and overlaying it with a transmission line topology map. The color grading of the heat map can be dynamically adjusted according to the severity of the hidden danger; the transmission line topology map includes the location information of the tower, tower height and conductor sag information, and the topology map is updated every 7-25 days.

[0039] The multi-level alarm mechanism of the early warning module includes: Level 1 alarm, score ≥70: audible and visual alarm + local log recording; Level 2 alarm, score ≥85: SMS notification to operation and maintenance personnel, with attached abnormal screenshot and GPS coordinates; Level 3 alarm, score ≥95: Automatically link with the power grid control system and execute emergency shutdown, while manual review and authority confirmation are required; In one monitoring, if the probability score of hidden danger P=75, a level 1 alarm is triggered, the device sounds and lights an alarm and records it in the local log; if the score P=90, a level 2 alarm is triggered, and the operation and maintenance personnel are notified by SMS with a screenshot of the abnormality and GPS coordinates (120.5, 30.2); if the score P=98, a level 3 alarm is triggered, and the power grid control system is automatically linked to perform an emergency shutdown, and manual review and permission confirmation are required. Intelligent decision support modules include: Cross-system linkage engine: triggers regional collaborative protection when the following conditions are met simultaneously: If the environmental hazard score is ≥85 and the equipment service life is less than 3 months, cross-region load transfer will be initiated; Emergency plan library: preset disposal plans according to voltage levels, For example, A transmission line video online monitoring system includes a transmission line video online monitoring device and a server that wirelessly communicates with the transmission line video online monitoring device. The server is used to receive image data of the transmission line marked with abnormal targets uploaded by the transmission line video online monitoring device.

[0040] The server is also used to send a neural network model update instruction to the transmission line video online monitoring device. The neural network model update instruction is used to update the abnormal target type. The wireless communication between the server and the transmission line video online monitoring device uses the AES-256 encryption algorithm for data encryption transmission. If the plaintext data is , the encryption key is , the encrypted ciphertext data is , the encryption process satisfies ,in It is the encryption function of the AES-256 encryption algorithm.

[0041] Example 1 Wildfire warning system response Scene Enhancement Description Line parameters: 500kV double-circuit line on the same tower, conductor model LGJ-630 / 45, sag 12.6m (at 40°C); Environmental background: At 2:00 PM in the summer, the vegetation dryness index (FDI) was 85 (extremely flammable). Data Input Extensions Process optimization: LSTM model input (time window 60 seconds): The input data is represented as: ,in Output probability score: Output probability The calculation formula is: ,in = Enhanced linkage measures: Power supply cut-off: Execute the "reduce first, cut later" strategy through the SCADA system: First reduce the voltage to 220kV and run for 5 seconds Then completely disconnect the DL-4512 circuit breaker Drone swarms: Heatmap generation: Use Kernel Density Estimation (KDE): (h=50m is the bandwidth, K is the Gaussian kernel function) Effect comparison supplement: in conclusion: 1: Technical performance Response speed: The entire process from fire identification to power outage is now 8.7 seconds, 35 times more efficient than traditional manual inspections (>5 minutes). Positioning accuracy: Through multi-spectral fusion + SLAM technology, the fire point positioning error is ≤3 meters (traditional methods ≥50 meters); False alarm control: The smoke recognition accuracy rate is 98.2%, and the false alarm rate is reduced to 2.3% (the false alarm rate of traditional infrared monitoring is 18.7%). 2: Safety improvement Rapid verification through drone swarms prevents operations personnel from entering the fire scene for high-risk operations; GIS heat maps support precise dispatch of firefighting resources, improving firefighting efficiency by 60%.

[0042] Example 2 Intelligent handling of mechanical intrusion Scene Enhancement Description Line parameters: Voltage and safety distance: The system voltage U is 220kV, according to the formula =0.01U+5 to calculate the safety distance It is 7.2m.

[0043] Conductor model and environmental conditions: The conductor model used is LGJ-400 / 35. When the ambient temperature is 30°C, the conductor sag is It is 9.8m.

[0044] Tower parameters: Tower height is 42m, the span L is 450m, These parameters are important basic data for the design and operation of transmission lines and are used to determine the electrical performance and mechanical structure characteristics of the lines.

[0045] Crane parameters Trajectory prediction algorithm upgrade Motion model improvements: Equation of motion: gives the motion model of the object in the x and y directions, and Respectively expressed in The position of the object in the x and y directions at the moment, including the speed v and the time interval , angle θ, acceleration a, and process noise and ; Process noise: Process noise ( , ) obeys a two-dimensional normal distribution with a mean of 0 and a standard deviation of 0.1, which is used to simulate the uncertainty in the motion process.

[0046] Angle estimation: The angle θ is estimated in real time through the extended Kalman filter to improve the accuracy of tracking the direction of object movement. Dynamic adjustment of safety distance: safety distance Depending on wind speed Perform dynamic adjustment, the adjustment formula is = ×(1+0.05 This means that in windy conditions, the safety distance needs to be increased appropriately to ensure safety. The greater the wind speed, the greater the increase in the safety distance. Secondary alarm handling process SMS alarm enhancements: Message standard: The alarm message content complies with the IEC61850-7-2 standard, a communication standard widely used in power systems, ensuring the standardization and compatibility of the message.

[0047] Message content: The message type is "MechanicalIntrusion," indicating a mechanical intrusion event has occurred; the risk level is 2; it contains the coordinates of the event; and specifies the corresponding operation steps, such as the first step of reducing the voltage (from 220kV to 110kV) and the second step of dispatching two drones.

[0048] Wire boost protection technical details: On-load tap-changing transformer (OLTC) operation: Protection setting switching logic: When the voltage reaches 110kV, the protection settings are switched. The overcurrent protection setting is set to 12kA, with an operating time of 0.3s. At the same time, the distance protection settings for Zone 1 are set to 80% and for Zone 2 to 120%. This ensures that the protection device can accurately operate according to the preset settings under specific voltage conditions, ensuring the safe operation of the power system. Verification and test data Trajectory prediction accuracy: System response time: The time consumption of each step in the alarm response process is broken down in detail (unit: ms): Image acquisition: takes 120ms and is used to obtain on-site image information.

[0049] Target recognition: It takes 80ms to identify the target in the collected image.

[0050] Trajectory prediction: takes 150ms to predict the target's trajectory.

[0051] Risk assessment: This process takes 50ms and assesses the risk level based on target information and trajectory prediction results.

[0052] SMS sending: It takes 300ms to send the alarm information in the form of SMS.

[0053] Voltage switching: takes 2800ms and performs voltage switching when needed.

[0054] Fail-safe mechanism False alarm handling: Drone review mechanism: When the alarm continues for 5 minutes without confirmation, it will automatically take off for verification Multi-source verification rules: In order to confirm the authenticity and accuracy of the alarm event, the following three conditions need to be met: Image recognition confidence: The confidence level of image recognition results must be greater than 90% to ensure the reliability of image recognition results.

[0055] Trajectory prediction consistency: The trajectory prediction consistency must be greater than 85% to ensure the accuracy of trajectory prediction.

[0056] GPS displacement: The displacement measured by GPS must be greater than 3σ ( The validity of the alarm event can only be confirmed if all three conditions are met.

[0057] Degraded mode: in conclusion 1: Technical performance Trajectory prediction accuracy: 10-second prediction error is 0.42±0.15 meters, meeting the requirements of dynamic adjustment of safety distance; System response delay: From identification to completion of voltage switching, it takes only 2.8 seconds, and the SMS alarm delay is 300ms (in line with power emergency communication standards); Decision reliability: All 37 real intrusions were accurately identified, with zero missed reports. 2: Safety improvement The voltage grading protection strategy (220kV→110kV) reduces the risk of electric shock from 28kA to 14kA, improving safety by 50%; Automated handling reduces human misjudgment, and the recurrence rate of historical accidents has dropped to 0.

[0058] Comparison summary: Summary: The two examples jointly verified the high reliability of the monitoring device in complex scenarios, and the comprehensive accident prevention efficiency reached more than 99%.

[0059] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the scope of protection of the present invention.

[0060] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the foregoing embodiments. The foregoing embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. The transmission line video online monitoring device is characterized by: include: Acquisition module, analysis module, prediction module, early warning module, equipment health management module and intelligent decision support module; Acquisition module: used for all-weather video acquisition, supporting dual-mode data fusion of thermal imaging and visible light; Analysis module: Based on deep learning algorithms, it processes video streams in real time, identifies anomaly types, and outputs anomaly levels; Prediction module: Combining historical data with real-time data, using a spatiotemporal fusion prediction model to analyze and predict the development trend of hidden dangers; Equipment health management module: used to automatically generate treatment suggestions and emergency plans for abnormal transmission line conditions based on monitoring data and preset rules; Early warning module: triggers multi-level alarms based on the abnormality level and generates a solution suggestion library.

2. The power transmission line video online monitoring device according to claim 1, characterized in that: The acquisition module integrates a multispectral sensor with a wavelength range of 8-14μm and a thermal sensitivity of ≤50mK. It distinguishes smoke and fog by the difference in spectral reflectance, where the distinction is based on: smoke reflectance <15%, cloud reflectance >30%. The acquisition module also includes a high-resolution infrared detector and a 4K visible light camera, supports HDR imaging, and supports adaptive exposure adjustment, and dynamically adjusts the infrared gain and visible light shutter speed. The acquisition module also includes an anti-shake gimbal, which uses three-axis gyroscope stabilization technology.

3. The power transmission line video online monitoring device according to claim 1, characterized in that: The analysis module is based on the YOLOv7 deep learning model to detect abnormal targets in the video in real time; the DeepSORT algorithm is integrated to predict the trajectory of cranes and tower cranes. Position coordinates at a given moment , speed record , according to the DeepSORT algorithm, it is predicted that Position coordinates at the moment for: Calculate the minimum spatial distance between it and the wire, support multimodal data fusion, combine thermal imaging temperature data with visible light images, enhance wildfire identification capabilities, and the thermal imaging temperature is -20℃-150℃.

4. The power transmission line video online monitoring device according to claim 1, characterized in that: The analysis module is deployed on edge computing devices and uses accelerated inference technology to reduce latency. The deployed edge computing devices use TensorRT to accelerate inference, with a latency of <200ms.

5. The power transmission line video online monitoring device according to claim 1, characterized in that: The spatiotemporal fusion prediction model includes: Short-term hidden danger prediction submodule: Based on the LSTM neural network, it inputs real-time monitoring data and outputs a hidden danger probability score within the next two hours; The input data includes: Using laser rangefinders and image recognition technology to monitor tree growth rates in real time; Use the global positioning system to obtain real-time location information of cranes and tower cranes; Utilize wind speed, humidity, and temperature data obtained from IoT sensors; Collect historical hidden danger database, including hidden danger type, occurrence time, location, severity and treatment results; Output hidden danger development trend score, triggering an early warning when the score exceeds the threshold, where the score range is 1-100; Long-term trend analysis submodule: Integrates the Prophet algorithm to analyze the periodic patterns in the historical hidden danger database and predict the accelerated growth of trees in the rainy season and the high incidence of wildfires in winter. The basic model of the Prophet algorithm is: in is a time series in time The observed value of is the trend term, which represents the long-term trend of the time series; is the seasonal term, which represents the periodic changes of the time series; is the holiday term, which indicates the impact of holidays on the time series; is the error term, which represents the random fluctuations in the time series that cannot be explained by the model.

6. The power transmission line video online monitoring device according to claim 1, characterized in that: The device health management module includes: Condition monitoring unit: used to collect equipment parameters such as conductor sag, insulator contamination, and hardware corrosion rate in real time, with a sampling frequency of ≥1Hz; Remaining life prediction unit: Calculates the remaining service life based on the equipment's degradation rate. Remaining service life: in, is the initial life of the equipment, The equipment has reached its service life. is the degradation rate; The degradation rate function is calculated by a random forest regression model with the following input parameters: Material properties, environmental stresses, mechanical loads; Maintenance decision-making unit: dynamically generate maintenance plans based on calculation results and process calculation results in sequence Devices < 6 months old.

7. The power transmission line video online monitoring device according to claim 1, characterized in that: The intelligent decision support module includes: Cross-system linkage engine: triggers regional collaborative protection when the following conditions are met simultaneously: If the environmental hazard score is ≥85 and the equipment service life is less than 3 months, cross-region load transfer will be initiated; Emergency plan library: preset disposal plans according to voltage levels.

8. The power transmission line video online monitoring device according to claim 1, characterized in that: The multi-level alarm mechanism of the early warning module includes: Level 1 alarm, score ≥70: audible and visual alarm + local log recording; Level 2 alarm, score ≥85: SMS notification to operation and maintenance personnel, with attached abnormal screenshot and GPS coordinates; Level 3 alarm, score ≥95: Automatically link with the power grid control system and execute emergency shutdown, while manual review and authority confirmation are required; The early warning module is deeply integrated with the GIS platform, automatically generating a heat map of the impact range of hidden dangers and overlaying it with a transmission line topology map. The color grading of the heat map can be dynamically adjusted according to the severity of the hidden danger. The transmission line topology map includes tower location information, tower height and conductor sag information, and the topology map is updated every 7-25 days.

9. Transmission line video online monitoring system, characterized by: It comprises the transmission line video online monitoring device according to any one of claims 1 to 8, and a server that wirelessly communicates with the transmission line video online monitoring device, wherein the server is used to receive image data of the transmission line marked with abnormal targets uploaded by the transmission line video online monitoring device.

10. The power transmission line video online monitoring system according to claim 9, characterized in that: The server is also used to send a neural network model update instruction to the transmission line video online monitoring device, and the neural network model update instruction is used to update the abnormal target type. The wireless communication between the server and the transmission line video online monitoring device uses the AES-256 encryption algorithm for data encryption transmission, wherein if the plaintext data is , the encryption key is , the encrypted ciphertext data is , the encryption process satisfies ,in It is the encryption function of the AES-256 encryption algorithm.

Images