Multi-dimensional online monitoring method based on transmission line visualization

Abstract

The application discloses a multi-dimension online monitoring method based on power transmission line visualization and belongs to the technical field of power transmission line online monitoring. The multi-dimension online monitoring method based on power transmission line visualization has the following specific implementation steps: step S1, data acquisition and multi-source fusion; step S2, edge end preprocessing and feature extraction; step S3, multi-dimension state evaluation model construction; step S4, dynamic risk early warning and hierarchical response; step S5, visualization interaction and decision support; and step S6, model iteration optimization processing. The method can reduce the false alarm rate, ensure high precision and high stability of the algorithm, replace the traditional static threshold alarm, improve the active defense capability and emergency disposal efficiency of the power transmission line, and realize fine risk control.

Description

Technical Field

[0001] This invention relates to the field of online monitoring technology for power transmission lines, and more specifically, to a multi-dimensional online monitoring method based on the visualization of power transmission lines. Background Technology

[0002] Online monitoring of transmission lines is a system that utilizes sensors, communication, and artificial intelligence technologies to achieve all-weather perception of the line's operational status. It deploys micro-meteorological, icing, and galloping sensors on conductors, installs video surveillance and tilt monitoring devices on towers, and integrates distributed fiber optic temperature measurement in the ground wire composite optical cable, enabling real-time collection of temperature, tension, sag, and channel environmental data. Information is transmitted back to the main station via a private wireless network or 5G. The platform automatically identifies anomalies such as external damage hazards, insulator flashover, and hardware overheating, triggering tiered alarms. This system upgrades traditional manual inspection to "unmanned operation + intelligent early warning," significantly improving emergency response speed under extreme weather conditions and reducing operation and maintenance costs. It is a key technological means to ensure the safe and stable operation of the large power grid.

[0003] Currently, in the online monitoring process of transmission lines: ① When shooting from a distance, the existing algorithm is prone to blurring the features of small targets such as insulators exploding, resulting in a high false negative rate. Furthermore, the model is difficult to adapt to "concept drift" caused by environmental changes, and has poor long-term stability. ② Traditional monitoring relies on static threshold alarms, which cannot dynamically assess the risk of the conductor based on real-time meteorological data. Furthermore, on-site inspections lack virtual and real integration methods, resulting in delayed emergency response under extreme weather conditions and a passive defense mode.

[0004] Therefore, in view of this, we will study and improve the existing structure and provide a multi-dimensional online monitoring method based on the visualization of transmission lines, in order to achieve a more practical value. Summary of the Invention

[0005] 1. Technical problems to be solved To address the problems existing in the prior art, the purpose of this invention is to provide a multi-dimensional online monitoring method based on the visualization of transmission lines. This method can reduce the false alarm rate, ensure high accuracy and stability of the algorithm, and replace traditional static threshold alarms, thereby improving the active defense capability and emergency response efficiency of transmission lines and achieving refined risk management.

[0006] 2. Technical Solution To solve the above problems, the present invention adopts the following technical solution.

[0007] A multi-dimensional online monitoring method based on power transmission line visualization is described below. Step S1: Data Acquisition and Multi-Source Fusion Deploy high-definition visible light cameras, infrared thermal imagers, lidar, and environmental sensor groups on transmission towers and lines to collect video stream data, equipment temperature field data, 3D point cloud data, and micro-meteorological data in the transmission channel in real time; By accessing power grid GIS geographic information data and historical inspection log data, a multi-source heterogeneous dataset containing spatiotemporal benchmarks is constructed. Step S2, Edge Preprocessing and Feature Extraction: The collected data is denoised, enhanced, and compressed at the edge computing nodes; Deep learning object detection algorithms are used to identify wires, insulators, hardware, and foreign objects in the video stream; Extract features of high-temperature anomaly regions from infrared images; Filter the laser point cloud data to restore the sag shape of the conductor; Step S3: Construction of a multi-dimensional state assessment model: The preprocessed data is then input into the cloud server. Construct a visualization foundation based on digital twins to map the physical power grid into a virtual three-dimensional model; Establish a comprehensive evaluation matrix that includes four dimensions: electrical performance, mechanical properties, thermal stability, and external environmental interference; The comprehensive health index of transmission lines is calculated based on a weighted fusion algorithm. Step S4: Dynamic Risk Warning and Tiered Response: The line's operating status is determined based on a comprehensive health index and preset thresholds. When wildfires, external damage, or extreme weather are detected, a millisecond-level alarm is triggered. By combining power flow distribution data, the fault propagation path is simulated, and a work order containing handling suggestions is generated; Step S5: Visual Interaction and Decision Support The assessment results, alarm information and 3D models are rendered using WebGL technology and displayed on the monitoring screen as a panoramic view of the corridor and equipment details. Through multi-view switching, historical data review, and emergency response drills, a panoramic, visualized, and closed-loop management system for monitoring data is implemented. Step S6: Perform iterative optimization of the model. As the runtime increases, the algorithm model is gradually optimized and its accuracy is improved.

[0008] Furthermore, in step S1: The environmental sensor group includes: wind speed and direction sensors, temperature and humidity sensors, rain gauges, and ice thickness monitors; Multi-source heterogeneous datasets are time-stamp aligned using dual-mode timing from BeiDou and GPS to ensure that data collected by different sensors are synchronized in the time dimension, providing an accurate time reference for subsequent multi-dimensional correlation analysis.

[0009] Furthermore, in step S2: During edge preprocessing, a strategy is controlled based on the adaptive bitrate of the H.265 encoding standard. The steps are as follows: The video stream's encoding bitrate and frame rate are dynamically adjusted based on the current network bandwidth and the degree of drastic changes in the image. While ensuring the clarity of images for critical alarm events, the wireless transmission bandwidth utilization is reduced to adapt to the unstable communication environment in remote mountainous areas.

[0010] Furthermore, in step S2: After feature extraction, minute defect detection is performed based on the improved YOLO algorithm, and the steps are as follows: To address the issue of blurred features of small targets such as insulator self-explosion and missing pins caused by long-distance shooting, an attention mechanism module is introduced to enhance the network's ability to extract features of small targets. By fusing shallow texture information and deep semantic information in the feature pyramid network, the recall rate of defect identification in complex backgrounds can be improved.

[0011] Furthermore, in step S3: During the construction of the digital twin visualization base, oblique photogrammetry technology is used to obtain the topography of the passage, and laser point cloud data is combined to fit the three-dimensional spatial coordinates of the conductor. BIM technology is used to refine the modeling of key towers and substation equipment, so as to complete the geometric appearance consistency and attribute information full element mapping between the physical entity and the virtual model.

[0012] Furthermore, in step S3: The comprehensive evaluation matrix also incorporates a dynamic calculation model for conductor sag; Based on real-time collected ambient temperature, wind speed, and icing data, combined with conductor physical parameters, the model calculates the real-time tension and sag of the conductor under the current working conditions using the finite element analysis method. The calculation results are then used as core indicators of the mechanical dimension and input into the health index calculation.

[0013] Furthermore, in step S3: The weighted fusion algorithm adopts a combined weighting method that combines objective weighting based on entropy weighting with analytic hierarchy process based on expert experience.

[0014] Furthermore, in step S4: When conducting dynamic risk warnings, the steps for predicting the spread trend of wildfires are as follows: Infrared thermal imaging data was used to determine the center point of the fire source, and combined with wind direction and speed sensor data and vegetation type distribution map, a forest fire spread rate model was run. Predict the area that wildfires may affect within a specified timeframe; Automatically match available fire-fighting resources and line shutdown strategies in the vicinity.

[0015] Furthermore, step S5: In visual interaction, the steps for AR-assisted inspection are as follows: Inspection personnel use handheld terminals or smart helmets to view on-site equipment: The system overlays the device parameters, historical defect records, and real-time monitoring data from the background digital twin model onto the real device screen in an augmented reality manner, assisting maintenance personnel in on-site diagnosis and troubleshooting.

[0016] Furthermore, in step S6: During the model's iterative optimization, the system automatically stores the false alarm samples and missed alarm samples that have been manually verified into the training sample library and periodically starts the incremental learning process. The parameters of the deep learning object detection model in step S2 and the state evaluation model in step S3 are optimized so that the accuracy of the algorithm model continuously improves as the running time increases.

[0017] 3. Beneficial Effects Compared with the prior art, the advantages of this invention are: ① This solution, by introducing attention and improving the YOLO algorithm, effectively solves the problem of blurred features of small targets such as insulator self-explosion and missing pins under long-distance shooting, and significantly reduces the false negative rate.

[0018] By adopting an H.265 adaptive bitrate strategy, transmission parameters are dynamically adjusted based on network bandwidth, ensuring data transmission continuity in remote mountainous areas with weak signals and fluctuating bandwidth.

[0019] The model's self-iterative mechanism based on incremental learning can automatically absorb false positives and false negatives for optimization, overcoming the "concept drift" problem caused by seasonal changes or environmental changes, and ensuring that the algorithm maintains high accuracy and high stability in long-term operation. In summary, this method can significantly improve the recognition accuracy and system robustness in complex environments.

[0020] ②This solution, by integrating oblique photography, laser point cloud and BIM technology to build a high-fidelity digital twin base, combined with a finite element-based dynamic calculation model of conductor sag, can accurately quantify the conductor stress state based on real-time micro-meteorological data, replacing the traditional static threshold alarm. AR-assisted inspections have enabled the fusion of virtual and real data on-site, significantly shortening on-site diagnosis time and improving the overall proactive defense capability and emergency response efficiency of transmission lines under extreme weather and external damage. In summary, this method can achieve refined risk management from passive alarm to proactive defense, while ensuring management efficiency. Attached Figure Description

[0021] Figure 1 This is a flowchart illustrating the multi-dimensional online monitoring method based on power transmission line visualization in this invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0023] Example: Please see Figure 1 A multi-dimensional online monitoring method based on transmission line visualization is proposed. The specific implementation steps of the multi-dimensional online monitoring method are as follows: Step S1: Data Acquisition and Multi-Source Fusion Deploy high-definition visible light cameras, infrared thermal imagers, lidar, and environmental sensor groups on transmission towers and lines to collect video stream data, equipment temperature field data, 3D point cloud data, and micro-meteorological data in the transmission channel in real time; By accessing power grid GIS geographic information data and historical inspection log data, a multi-source heterogeneous dataset containing spatiotemporal benchmarks is constructed. Specifically, in step S1: The environmental sensor group includes: wind speed and direction sensors, temperature and humidity sensors, rain gauges, and ice thickness monitors; Multi-source heterogeneous datasets are time-stamp aligned using dual-mode timing from BeiDou and GPS to ensure that data collected by different sensors are synchronized in the time dimension, providing an accurate time reference for subsequent multi-dimensional correlation analysis.

[0024] By introducing multiple micro-meteorological sensors and performing high-precision time synchronization, the consistency of multi-source data over time was ensured. This provides a precise data foundation for subsequent analysis of the impact of wind speed on conductor galloping and the impact of temperature changes on sag, avoiding analytical errors caused by asynchronous data.

[0025] Step S2, Edge Preprocessing and Feature Extraction: The collected data is denoised, enhanced, and compressed at the edge computing nodes; Deep learning object detection algorithms are used to identify wires, insulators, hardware, and foreign objects in the video stream; Extract features of high-temperature anomaly regions from infrared images; Filter the laser point cloud data to restore the sag shape of the conductor; Specifically, in step S2: During edge preprocessing, a strategy is controlled based on the adaptive bitrate of the H.265 encoding standard. The steps are as follows: The video stream's encoding bitrate and frame rate are dynamically adjusted based on the current network bandwidth and the degree of drastic changes in the image. While ensuring the clarity of images for critical alarm events, the wireless transmission bandwidth utilization is reduced to adapt to the unstable communication environment in remote mountainous areas.

[0026] By adopting an adaptive bit rate control strategy, the transmission problems of weak 4G / 5G signals and large bandwidth fluctuations in remote areas are effectively solved.

[0027] While ensuring the quality of critical images, it saves on data traffic costs and improves the system's survivability and data transmission efficiency in harsh communication environments.

[0028] Specifically, in step S2: After feature extraction, minute defect detection is performed based on the improved YOLO algorithm, and the steps are as follows: To address the issue of blurred features of small targets such as insulator self-explosion and missing pins caused by long-distance shooting, an attention mechanism module is introduced to enhance the network's ability to extract features of small targets. By fusing shallow texture information and deep semantic information in the feature pyramid network, the recall rate of defect identification in complex backgrounds can be improved.

[0029] To address the common pain point of "difficulty in identifying small targets" in power transmission line monitoring, attention mechanisms and feature fusion technology are introduced.

[0030] It significantly improves the detection rate of minor defects such as missing pins and insulator cracks, reduces the risk of missed detection, and enhances the robustness of the system in complex environments.

[0031] Step S3: Construction of a multi-dimensional state assessment model: The preprocessed data is then input into the cloud server. Construct a visualization foundation based on digital twins to map the physical power grid into a virtual three-dimensional model; Establish a comprehensive evaluation matrix that includes four dimensions: electrical performance, mechanical properties, thermal stability, and external environmental interference; The comprehensive health index of transmission lines is calculated based on a weighted fusion algorithm. Specifically, in step S3: During the construction of the digital twin visualization base, oblique photogrammetry technology is used to obtain the topography of the passage, and laser point cloud data is combined to fit the three-dimensional spatial coordinates of the conductor. BIM technology is used to refine the modeling of key towers and substation equipment, so as to complete the geometric appearance consistency and attribute information full element mapping between the physical entity and the virtual model.

[0032] By combining oblique photogrammetry, laser point cloud, and BIM technology, a high-fidelity digital twin foundation is constructed. This makes the virtual model not only visually appealing but also accurate, realistically reflecting the terrain and spatial relationships of equipment, providing precise geographical references for UAV autonomous inspection path planning and tree obstacle analysis.

[0033] Specifically, in step S3: The comprehensive evaluation matrix also incorporates a dynamic calculation model for conductor sag; Based on real-time collected ambient temperature, wind speed, and icing data, combined with conductor physical parameters, the model calculates the real-time tension and sag of the conductor under the current working conditions using the finite element analysis method. The calculation results are then used as core indicators of the mechanical dimension and input into the health index calculation.

[0034] A dynamic calculation model for sag based on real-time environment and physical parameters is introduced.

[0035] Compared to traditional static threshold alarms, this model can more accurately reflect the stress on the conductor under extreme weather conditions, effectively preventing conductor breakage or ground discharge accidents caused by icing or strong winds.

[0036] Specifically, in step S3: The weighted fusion algorithm adopts a combined weighting method that combines objective weighting based on entropy weighting with analytic hierarchy process based on expert experience.

[0037] This method eliminates the subjective arbitrariness of human-assigned weights while retaining the experience-based judgment of power industry experts on specific fault modes, thus improving the scientific rigor and accuracy of the comprehensive health index assessment results.

[0038] A weighting method combining subjective and objective approaches is adopted. It utilizes both the objective laws of big data mining (entropy weighting method) and the industry experience of power experts (AHP), making the calculation results of the comprehensive health index more consistent with the actual operating logic of the power system and improving the credibility of the assessment conclusions.

[0039] Step S4: Dynamic Risk Warning and Tiered Response: The line's operating status is determined based on a comprehensive health index and preset thresholds. When wildfires, external damage, or extreme weather are detected, a millisecond-level alarm is triggered. By combining power flow distribution data, the fault propagation path is simulated, and a work order containing handling suggestions is generated; Specifically, in step S4: When conducting dynamic risk warnings, the steps for predicting the spread trend of wildfires are as follows: Infrared thermal imaging data was used to determine the center point of the fire source, and combined with wind direction and speed sensor data and vegetation type distribution map, a forest fire spread rate model was run. Predict the area that wildfires may affect within a specified timeframe; Automatically match available fire-fighting resources and line shutdown strategies in the vicinity.

[0040] It can not only identify existing wildfires but also predict their spread. This proactive risk assessment enables maintenance personnel to take proactive measures to cut off or isolate power lines before fires threaten them, minimizing disaster losses and reflecting a shift from "post-disaster repair" to "pre-disaster prevention."

[0041] Step S5: Visual Interaction and Decision Support The assessment results, alarm information and 3D models are rendered using WebGL technology and displayed on the monitoring screen as a panoramic view of the corridor and equipment details. Through multi-view switching, historical data review, and emergency response drills, a panoramic, visualized, and closed-loop management system for monitoring data is implemented. Specifically, the S5 steps are as follows: In visual interaction, the steps for AR-assisted inspection are as follows: Inspection personnel use handheld terminals or smart helmets to view on-site equipment: The system overlays the device parameters, historical defect records, and real-time monitoring data from the background digital twin model onto the real device screen in an augmented reality manner, assisting maintenance personnel in on-site diagnosis and troubleshooting.

[0042] By combining big data from the backend with AR technology on-site, a "virtual-real fusion" has been achieved. Inspection personnel can intuitively obtain information about the entire lifecycle of equipment on-site without repeatedly consulting paper drawings or backend systems, significantly shortening the time for fault location and equipment verification.

[0043] Step S6: Perform iterative optimization of the model. As the runtime increases, the algorithm model is gradually optimized and its accuracy is improved.

[0044] Specifically, in step S6: During the model's iterative optimization, the system automatically stores the false alarm samples and missed alarm samples that have been manually verified into the training sample library and periodically starts the incremental learning process. The parameters of the deep learning object detection model in step S2 and the state evaluation model in step S3 are optimized so that the accuracy of the algorithm model continuously improves as the running time increases.

[0045] A self-evolution mechanism for the algorithm model was established. By continuously collecting on-site feedback data to optimize model parameters, the "concept drift" problem, in which the accuracy of AI models decreases due to environmental changes after actual deployment, was solved, ensuring the long-term stability and advanced nature of the system.

[0046] Working principle: The multi-dimensional online monitoring method constructs a complete technical closed loop of "perception-edge computing-digital twin-decision-visualization".

[0047] Through comprehensive multi-dimensional (electrical, mechanical, thermal, and environmental) assessments, the problems of limited monitoring methods and data silos in traditional systems have been solved. This has enabled a shift from passive alarms to proactive health management, significantly improving the intelligence level and decision-making efficiency of power transmission line operation and maintenance.

[0048] The above description is merely a preferred embodiment of the present invention; however, the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and its improved concepts, should be covered within the scope of protection of the present invention.

Claims

1. A multi-dimensional online monitoring method based on transmission line visualization, characterized in that: The specific implementation steps of the multi-dimensional online monitoring method are as follows: Step S1: Data Acquisition and Multi-Source Fusion Deploy high-definition visible light cameras, infrared thermal imagers, lidar, and environmental sensor groups on transmission towers and lines to collect video stream data, equipment temperature field data, 3D point cloud data, and micro-meteorological data in the transmission channel in real time; By accessing power grid GIS geographic information data and historical inspection log data, a multi-source heterogeneous dataset containing spatiotemporal benchmarks is constructed. Step S2, Edge Preprocessing and Feature Extraction: exist Edge computing nodes perform noise reduction, enhancement, and compression processing on the collected data; Deep learning object detection algorithms are used to identify wires, insulators, hardware, and foreign objects in the video stream; Extract features of high-temperature anomaly regions from infrared images; Filter the laser point cloud data to restore the sag shape of the conductor; Step S3: Construction of a multi-dimensional state assessment model: The preprocessed data is then input into the cloud server. Construct a visualization foundation based on digital twins to map the physical power grid into a virtual three-dimensional model; Establish a comprehensive evaluation matrix that includes four dimensions: electrical performance, mechanical properties, thermal stability, and external environmental interference; The comprehensive health index of transmission lines is calculated based on a weighted fusion algorithm. Step S4: Dynamic Risk Warning and Tiered Response: The line's operating status is determined based on a comprehensive health index and preset thresholds. When wildfires, external damage, or extreme weather are detected, a millisecond-level alarm is triggered. By combining power flow distribution data, the fault propagation path is simulated, and a work order containing handling suggestions is generated; Step S5: Visual Interaction and Decision Support The assessment results, alarm information and 3D models are rendered using WebGL technology and displayed on the monitoring screen as a panoramic view of the corridor and equipment details. Through multi-view switching, historical data review, and emergency response drills, a panoramic, visualized, and closed-loop management system for monitoring data is implemented. Step S6: Perform iterative optimization of the model. As the runtime increases, the algorithm model is gradually optimized and its accuracy is improved.

2. The multi-dimensional online monitoring method based on transmission line visualization according to claim 1, characterized in that: In step S1: The environmental sensor group includes: wind speed and direction sensors, temperature and humidity sensors, rain gauges, and ice thickness monitors; Multi-source heterogeneous datasets are time-stamp aligned using dual-mode timing from BeiDou and GPS to ensure that data collected by different sensors are synchronized in the time dimension, providing an accurate time reference for subsequent multi-dimensional correlation analysis.

3. The multi-dimensional online monitoring method based on transmission line visualization according to claim 1, characterized in that: In step S2: During edge preprocessing, a strategy is controlled based on the adaptive bitrate of the H.265 encoding standard. The steps are as follows: The video stream's encoding bitrate and frame rate are dynamically adjusted based on the current network bandwidth and the degree of drastic changes in the image. While ensuring the clarity of images for critical alarm events, the wireless transmission bandwidth utilization is reduced to adapt to the unstable communication environment in remote mountainous areas.

4. The multi-dimensional online monitoring method based on transmission line visualization according to claim 1, characterized in that: In step S2: After feature extraction, minute defect detection is performed based on the improved YOLO algorithm, and the steps are as follows: To address the issue of blurred features of small targets such as insulator self-explosion and missing pins caused by long-distance shooting, an attention mechanism module is introduced to enhance the network's ability to extract features of small targets. By fusing shallow texture information and deep semantic information in the feature pyramid network, the recall rate of defect identification in complex backgrounds can be improved.

5. The multi-dimensional online monitoring method based on transmission line visualization according to claim 1, characterized in that: In step S3: During the construction of the digital twin visualization base, oblique photogrammetry technology is used to obtain the topography of the passage, and laser point cloud data is combined to fit the three-dimensional spatial coordinates of the conductor. BIM technology is used to refine the modeling of key towers and substation equipment, so as to complete the geometric appearance consistency and attribute information full element mapping between the physical entity and the virtual model.

6. The multi-dimensional online monitoring method based on transmission line visualization according to claim 1, characterized in that: In step S3: The comprehensive evaluation matrix also incorporates a dynamic calculation model for conductor sag; Based on real-time collected ambient temperature, wind speed, and icing data, combined with conductor physical parameters, the model calculates the real-time tension and sag of the conductor under the current working conditions using the finite element analysis method. The calculation results are then used as core indicators of the mechanical dimension and input into the health index calculation.

7. The multi-dimensional online monitoring method based on transmission line visualization according to claim 1, characterized in that: In step S3: The weighted fusion algorithm adopts a combined weighting method that combines objective weighting based on entropy weighting with analytic hierarchy process based on expert experience.

8. The multi-dimensional online monitoring method based on transmission line visualization according to claim 1, characterized in that: In step S4: When conducting dynamic risk warnings, the steps for predicting the spread trend of wildfires are as follows: Infrared thermal imaging data was used to determine the center point of the fire source, and combined with wind direction and speed sensor data and vegetation type distribution map, a forest fire spread rate model was run. Predict the area that wildfires may affect within a specified timeframe; Automatically match available fire-fighting resources and line shutdown strategies in the vicinity.

9. The multi-dimensional online monitoring method based on transmission line visualization according to claim 1, characterized in that: Step S5: In visual interaction, the steps for AR-assisted inspection are as follows: Inspection personnel use handheld terminals or smart helmets to view on-site equipment: The system overlays the device parameters, historical defect records, and real-time monitoring data from the background digital twin model onto the real device screen in an augmented reality manner, assisting maintenance personnel in on-site diagnosis and troubleshooting.

10. The multi-dimensional online monitoring method based on transmission line visualization according to claim 1, characterized in that: In step S6: During the model's iterative optimization, the system automatically stores the false alarm samples and missed alarm samples that have been manually verified into the training sample library and periodically starts the incremental learning process. The parameters of the deep learning object detection model in step S2 and the state evaluation model in step S3 are optimized so that the accuracy of the algorithm model continuously improves as the running time increases.

Images