A method and system for managing inspection data of transmission lines

Abstract

The present invention provides a method and system for managing inspection data assets of transmission lines. Through various sensors installed on transmission lines and the surrounding environment, as well as sensors carried by helicopters and unmanned aerial vehicles, multi-modal data of transmission lines is obtained. By processing the multi-modal data based on a spatio-temporal association model, a fused data set is generated. An adaptive feature extraction algorithm is used to analyze the fused data set to generate structured transmission line data assets including environmental features, defect features, and potential hazard features, and based on the association between environmental features and defect features, the possibility of environmental conditions triggering defects is evaluated. According to potential hazard features and environmental conditions, a graph network model with transmission lines as nodes and sensor data as edge weights is constructed to analyze the defect propagation path and predict the temporal and spatial distribution of potential hazards. Based on the prediction results, a defect risk level warning is generated, and combined with environmental meteorological data, a potential hazard distribution map of transmission lines is generated to provide precise assistance for inspection and maintenance plans.

Description

Technical Field

[0001] The present invention relates to the technical field of transmission lines, and particularly to a method and system for managing inspection data of transmission lines. Background Art

[0002] Transmission lines are key infrastructure in the power system, and their safe and stable operation is of great significance to power supply reliability and economy. At present, traditional inspection methods for transmission lines mainly rely on manual inspections and limited on-line monitoring technologies. These methods usually involve inspectors observing the condition of transmission lines on-site, recording the problems found, and using some sensors to monitor equipment parameters such as current, voltage, and temperature changes. In addition, helicopter / large and medium-sized unmanned aerial vehicles and satellite remote sensing technologies have been introduced in recent years for the inspection of transmission lines.

[0003] Although the existing technologies can identify potential hazards of transmission lines to a certain extent, they generally have the following problems: First, the data sources are single or incomplete, and it is impossible to comprehensively consider the multi-modal information of transmission lines, such as the combined effects of environmental factors, equipment status, and appearance defects; Second, the data processing and analysis methods are relatively traditional, usually only focusing on a single time point or a single equipment status, and it is difficult to provide a comprehensive analysis of multi-temporal and spatial correlations; Third, the prediction of potential hazard risks lacks accuracy, especially in the prediction of the propagation path of hazards and the spatio-temporal distribution of potential risks, and it is impossible to provide an accurate basis for maintenance decisions.

[0004] In view of the above problems, the present application proposes a method and system for managing inspection data of transmission lines. Summary of the Invention

[0005] The present application provides a method and system for managing inspection data of transmission lines to improve the inspection efficiency and decision-making accuracy of transmission lines.

[0006] The present application provides a method for managing inspection data of transmission lines, including:

[0007] Obtaining multi-modal data related to the inspection of transmission lines through various sensors installed on the transmission lines and the surrounding environment, mobile terminals carried by inspectors, and data acquisition devices of environmental monitoring satellites, where the multi-modal data includes physical state data of the transmission lines, electrical parameter data, environmental meteorological data, appearance photo and video data of the transmission lines, temperature distribution data captured by infrared cameras, defect data of the transmission lines, and potential hazard data of the transmission channels;

[0008] Synchronously calibrating and dynamically fusing the multi-modal data based on a spatio-temporal correlation model to obtain a fused data set with time consistency and spatial accuracy;

[0009] Parse the fused dataset using an adaptive feature extraction algorithm to generate transmission line data assets with structured features. The transmission line data assets include environmental feature tags, defect feature tags, and potential hazard feature tags. Among them, the environmental feature tags describe the potential impacts of weather, temperature, humidity, and wind speed on the operation of transmission lines.

[0010] Evaluate the triggering probability of environmental conditions on transmission line defects based on the association between the environmental feature tags and the defect feature tags.

[0011] Construct a graph network model with transmission lines as nodes and sensor data as edge weights based on the evaluation results of the potential hazard feature tags and environmental conditions. The graph network model is used to analyze the defect propagation paths of transmission lines and their corridors. Predict the occurrence time of potential hazards using the graph network model based on historical inspection data and environmental dynamic parameters.

[0012] Generate a defect risk level warning based on the prediction results of the graph network model, and generate a transmission line potential hazard distribution map in combination with environmental meteorological data. The potential hazard distribution map is used to assist in formulating inspection and maintenance plans.

[0013] The beneficial effects of the technical solution provided by this application include:

[0014] (1) By combining multi-modal data collected by sensors, mobile terminals, and satellites, comprehensively covering multi-dimensional information such as the physical state, electrical parameters, environmental meteorology, and appearance defects of transmission lines, it solves the problems of single data source or limited coverage in traditional methods, and improves the comprehensiveness and reliability of potential hazard data collection. (2) Use a spatio-temporal correlation model to synchronously calibrate and dynamically fuse multi-modal data to ensure that the generated dataset has time consistency and spatial accuracy. This processing method can effectively combine the data characteristics of different acquisition devices and provide a high-quality data basis for subsequent feature extraction and analysis. (3) By constructing a graph network model with transmission lines as nodes and sensor data as edge weights, it can dynamically simulate the propagation paths of transmission line potential hazards, and predict the time and spatial distribution of potential hazards in combination with historical data and environmental dynamic parameters, providing an accurate dynamic assessment of potential hazard propagation. (4) Generate a defect risk level warning based on the prediction results, and generate a transmission line potential hazard distribution map in combination with environmental meteorological data. The distribution map can intuitively display high-risk areas and their intensity distributions, assist maintenance personnel in reasonably planning inspection and maintenance tasks, and improve inspection efficiency and decision-making accuracy. Description of the Drawings

[0015] Figure 1 is a flowchart of a method for managing transmission line inspection data assets provided in the first embodiment of this application.

[0016] Figure 2It is a schematic diagram of a transmission line inspection data asset management system provided by the second embodiment of the present application. Detailed implementation manners

[0017] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application. However, the present application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar generalizations without departing from the connotation of the present application. Therefore, the present application is not limited by the specific implementations disclosed below.

[0018] The first embodiment of the present application provides a method for managing transmission line inspection data assets. Please refer to Figure 1 , which is a schematic diagram of the first embodiment of the present application. The following will Figure 1 describe in detail a method for managing transmission line inspection data assets provided by the first embodiment of the present application.

[0019] Step S101: Obtain multi-modal data related to transmission line inspections through various sensors installed on the transmission line and its surrounding environment, sensors carried by helicopters and drones, mobile terminals carried by inspection personnel, and data acquisition devices of environmental monitoring satellites. Among them, the multi-modal data includes physical state data of the transmission line, electrical parameter data, three-dimensional laser point cloud data, environmental meteorological data, appearance photos and video data of the transmission line, temperature distribution data captured by infrared cameras and infrared photos, defect data of the transmission line, and potential hazard data of the transmission line corridor.

[0020] Step S101 involves collecting multi-modal data from the transmission line and its surrounding environment to comprehensively cover the operating status, environmental conditions, and potential hazard information of the transmission line. The following details the implementation manners of this step.

[0021] First, install various sensors on the transmission line and its surrounding environment to collect real-time data related to the operation of the transmission line. These sensors include but are not limited to: vibration sensors for monitoring the mechanical state of the transmission line; temperature sensors for measuring the temperature change of the wire; current and voltage sensors for collecting electrical parameters of the transmission line, such as current, voltage, and their fluctuations. In addition, use meteorological sensors to collect environmental meteorological data in real time, including parameters such as wind speed, humidity, temperature, and precipitation.

[0022] The mobile terminals carried by inspection personnel are equipped with high-definition cameras, infrared cameras, and positioning modules, which are used to take photos and videos of the appearance of the transmission line during the inspection process, and capture the temperature distribution data of the transmission line and its key components through infrared imaging. These mobile terminals can also upload the collected data to the data processing center in real time through the wireless communication module.

[0023] Meanwhile, deploy data acquisition devices for environmental monitoring satellites to monitor the wide-area environment of the transmission line corridor through high-resolution imaging technology and optical remote sensing technology, and obtain data such as topographic changes, vegetation coverage, and potential external hazards (such as tree falls or fire risks) around the transmission line. Satellite data can be collected at fixed intervals or an emergency monitoring mode can be initiated according to specific events or requirements.

[0024] During the data acquisition process, each device calibrates the timestamp through a time synchronization protocol (such as the Network Time Protocol NTP) to ensure the consistency of data from different sources in the time dimension. In addition, a label is assigned to each sensor, inspection terminal, and satellite data acquisition device through a unique identifier, so that the data source can be accurately identified in the subsequent data processing and integration stages.

[0025] The multimodal data obtained in the above manner includes but is not limited to: environmental parameters such as the mechanical vibration frequency of the transmission line, wire temperature, current fluctuation, wind speed, and humidity; visual features of potential defects (such as cracks and corrosion areas) in the appearance photos and videos of the transmission line; temperature anomaly hotspots reflected in the infrared imaging; and potential external hazard information (such as vegetation coverage changes and the location of fallen trees) in the transmission line corridor.

[0026] Finally, all the collected data is stored in a unified data management platform and classified and labeled according to time, space, and data type, providing high-quality raw input data for subsequent steps (such as data calibration, fusion, and feature extraction).

[0027] Step S102: Synchronously calibrate and dynamically fuse the multimodal data based on the spatio-temporal association model to obtain a fused data set with time consistency and spatial accuracy.

[0028] Step S102 is to synchronously calibrate and dynamically fuse the multimodal data obtained through various methods to ensure the consistency of the data in time and space, providing an accurate basis for subsequent data processing and analysis.

[0029] First, it is necessary to classify and organize the multimodal data for subsequent processing. The obtained data includes time series data (such as physical state data, electrical parameter data, and environmental meteorological data) and spatial data (such as appearance photos, videos, infrared temperature distributions, and transmission line corridor hazard data). These data are initially classified and indexed according to their timestamps, acquisition device identifiers, and geographical location information to ensure that all data can accurately match its source and acquisition time.

[0030] In the time dimension, the dynamic time warping method is used to synchronously calibrate time series data. First, the timestamps of all sensors and data acquisition devices are unified, and the errors caused by device clock drift are eliminated. Then, the dynamic time warping algorithm is used to align the data with different acquisition frequencies in time. For example, the electrical parameter data sampled at a high frequency is matched with the environmental meteorological data sampled at a low frequency to generate a data sequence with consistent time. For image and video data, they are associated with the time series data through timestamps to ensure that relevant data from multiple sensors and devices can be integrated at the same time point.

[0031] In the space dimension, it is necessary to perform spatial mapping on the data according to the geographical locations of the acquisition devices and the actual distribution of the transmission lines. By combining the geographical topological structure of the transmission lines, all data is mapped into a unified geographical coordinate system. For example, for the image and video data captured by the inspection personnel, the geographical location corresponding to the image is calibrated through the longitude and latitude information provided by the positioning device; for the temperature distribution data collected by the infrared camera, according to the installation location and shooting direction of the sensor, combined with the distance correction algorithm, the data is projected onto the spatial model of the transmission line. In this way, the spatial coordinates of all data can accurately reflect the actual position of the transmission line.

[0032] Subsequently, after the time consistency and spatial mapping are completed, dynamic fusion of the multi-modal data is performed. The key steps of the fusion include weighted superposition and feature extraction of the data. First, weights are assigned to each data type. For example, the weights of physical state data and electrical parameter data in potential hazard assessment may be higher than those of environmental meteorological data. The fusion process performs weighted calculations on the data according to these weights to obtain a fusion data set that can comprehensively reflect the current state of the transmission line. During this process, abnormal data is eliminated. For example, abnormal values that deviate significantly from other data are identified through statistical analysis and smoothed to ensure data consistency.

[0033] Finally, the fused data is stored as a time-space fusion data set in a unified format, which contains the comprehensive state information at each time point and spatial location. This data set not only has time consistency but also can be accurately located in the space dimension, laying a reliable data foundation for feature extraction and potential hazard assessment in subsequent steps.

[0034] Furthermore, the synchronous calibration and dynamic fusion of the multi-modal data based on the spatio-temporal correlation model to obtain a fusion data set with time consistency and spatial accuracy includes:

[0035] Obtain multi-modal data of the transmission line;

[0036] According to the timestamp information of the multimodal data, sort by the acquisition time to generate a preliminary time series; use the dynamic time warping method to calibrate the data in the preliminary time series, align the multimodal data collected at different times, and generate time series data with consistent time.

[0037] Based on the time-consistent time series data and the geographical location information of the acquisition devices in the multimodal data, combined with the geographical topology of the transmission line, map the data to a unified geographical grid; use the geographically weighted regression method to interpolate and match the spatial distribution characteristics of the mapped data, eliminate the spatial deviation caused by the position offset of the acquisition device, and generate spatially consistent multimodal data.

[0038] Based on the source characteristics, historical accuracy, and environmental condition impacts of each data type, assign priorities to physical state data, electrical parameter data, environmental meteorological data, infrared temperature distribution data, and visual feature data.

[0039] According to the real-time monitoring results of the operating state of the transmission line, dynamically adjust the priorities of the physical state data, electrical parameter data, environmental meteorological data, infrared temperature distribution data, and visual feature data.

[0040] According to the adjusted data priorities, perform weighted superposition and fusion on the time-consistent and spatially consistent multimodal data to generate a fusion dataset with time consistency and spatial accuracy.

[0041] First, obtain the multimodal data of the transmission line from multiple data acquisition devices, including sensors installed on the transmission line, mobile terminals carried by inspection personnel, and environmental monitoring satellites. The multimodal data includes physical state data (such as vibration, stress), electrical parameter data (such as current, voltage), environmental meteorological data (such as temperature, humidity, wind speed), temperature distribution data captured by infrared cameras, and appearance photo and video data collected by inspection personnel. Each type of data is accompanied by a timestamp and the geographical location information of the acquisition device.

[0042] Next, according to the timestamp information of the data, sort all the collected data in chronological order to generate a preliminary time series. Since there may be differences in the sampling frequencies and time synchronization accuracies of different devices, use the dynamic time warping method to calibrate the time series to ensure the alignment of various types of data in the time dimension. For example, for the high-frequency sampled electrical parameter data and the low-frequency sampled meteorological data, adjust the time step through the dynamic time warping algorithm to align them to a common time axis. After calibration, the generated time series data is consistent in the time dimension.

[0043] After completing time calibration, using the geographical location information of the acquisition device and the geographical topology of the transmission line, the multi-modal data is mapped into a unified geographical grid. The geographical grid is based on the actual distribution of the transmission line and is divided into small regional units. The boundaries and central positions of each unit are defined according to the specific spatial layout of the transmission line. For the data within each unit, the geographically weighted regression method is used to interpolate and match the spatial distribution characteristics of the data, thereby compensating for the errors caused by the position offset or uneven spatial distribution of the acquisition device. For example, when the number of temperature data sampling points in some areas is small, the temperature distribution is estimated by the interpolation method to generate spatially consistent multi-modal data.

[0044] Based on the spatially consistent data, priorities are assigned to each data type. The determination of priorities is based on the source characteristics of the data, historical accuracy, and the influence of environmental conditions. For example, under weather conditions of high wind speed and high humidity, the importance of environmental meteorological data for defect assessment increases, and its priority is correspondingly increased, while the priority of other data such as electrical parameters may be relatively reduced. Through statistical analysis in historical monitoring results, the data priorities are further adjusted to make the same type of data have dynamic adaptability in different operating scenarios.

[0045] Based on the above priority adjustment results, weighted superposition and fusion are performed on the time-consistent and spatially consistent multi-modal data. The fusion process comprehensively considers the importance of each data type and its impact on specific problems. For example, physical state data and electrical parameter data may have higher weights in mechanical defect analysis, while visual feature data dominates in appearance defect recognition. Through the weighted superposition algorithm, the feature values of various types of data are combined according to their priorities, and finally a fusion data set with time consistency and spatial accuracy is generated.

[0046] The generated fusion data set fully reflects the state changes of the transmission line in the time and space dimensions and simultaneously has the comprehensive characteristics of multi-modal data. This data set provides a high-quality basis for subsequent feature extraction, hidden danger assessment, and risk prediction, ensuring the accuracy and applicability of data processing.

[0047] Furthermore, dynamically adjusting the priorities of the physical state data, electrical parameter data, environmental meteorological data, infrared temperature distribution data, and visual feature data according to the real-time monitoring results of the operating state of the transmission line includes:

[0048] Based on the monitoring results of the transmission line operating status, the transmission line is divided into a normal state and an abnormal state. Among them, the normal state means that the operating parameters of the transmission line are within the set safety threshold range, and the abnormal state means that the operating parameters of the transmission line are outside the set safety threshold range. Among them, for the situation where monitoring results cannot be obtained, the defect distribution situation is considered. When the operating state is the normal state, the initial priority of the data type is maintained, and the data type with higher historical accuracy and credibility is processed preferentially. When the operating state is the abnormal state, the data priority is dynamically adjusted according to the abnormal category.

[0049] First, through the real-time monitoring results of the transmission line operating status, the current operating status is classified. The monitoring data of the transmission line operating status includes physical state data (such as vibration frequency, tension value), electrical parameter data (such as current, voltage), environmental meteorological data (such as wind speed, humidity, temperature), etc. The real-time monitoring system continuously collects these data and compares them with the pre-set safety threshold. If all monitoring parameters are within the set safety threshold range, the operating state is classified as the normal state; if any parameter exceeds the safety threshold range, the operating state is classified as the abnormal state.

[0050] In the normal state, the system defaults to maintaining the initial priority of each data type. The initial priority is determined according to the historical accuracy and credibility of the data type. For example, physical state data and electrical parameter data may be given higher priorities because they usually directly come from sensor devices with higher precision, while environmental meteorological data and visual feature data may have relatively lower priorities due to greater external interference. During the data processing process, the system preferentially processes data with higher priorities to ensure that key data is analyzed and used in a timely manner.

[0051] When the operating state is the abnormal state, the system will dynamically adjust the priority of the data type according to the specific category of the abnormality. For example, if the abnormal category is an electrical parameter abnormality (such as voltage or current exceeding the normal range), the priority of the electrical parameter data will be significantly increased, and at the same time, the corresponding physical state data will also be given a higher priority due to possible associated problems (such as the risk of wire breakage). For abnormalities that may be caused by environmental conditions (such as excessive wind speed or abnormal temperature), the priority of the environmental meteorological data will be increased, and at the same time, the infrared temperature distribution data may be given a higher priority due to the importance of identifying the temperature rise area. In this dynamic adjustment, the priority of the visual feature data may remain unchanged unless the abnormality involves appearance defects (such as insulator damage or line slack).

[0052] The dynamically adjusted priorities will directly affect the order of data processing and resource allocation. Based on the adjusted priorities, the system first processes high-priority data to ensure that key information related to anomalies can be extracted and analyzed in a timely manner. Although low-priority data will still be processed, it may be delayed until after the critical data analysis is completed or, when necessary, the computational load is reduced through compression processing.

[0053] In this way, the system can efficiently utilize computing resources under normal conditions, and at the same time, can quickly respond and concentrate resources to process critical data under abnormal conditions, ensuring the timely identification and handling of abnormal situations.

[0054] Step S103: Use an adaptive feature extraction algorithm to analyze the fused dataset to generate power transmission line data assets with structured features. The power transmission line data assets include environmental feature tags, defect feature tags, and potential hazard feature tags. Among them, the environmental feature tags describe the potential impacts of weather, temperature, humidity, and wind speed on the operation of the power transmission line.

[0055] The core of step S103 is to use an adaptive feature extraction algorithm to deeply analyze the fused dataset to extract and label the key features of the operation status of the power transmission line, thereby generating structured power transmission line data assets. The following details the implementation process of this step.

[0056] First, extract different types of data from the fused dataset, including environmental data, physical state data, electrical parameter data, and image and video data, etc. These data have been calibrated in terms of time and space in step S102 and already have consistency and accurate geographical mapping. For each type of data, feature extraction is performed according to its characteristics and application scenarios. For example, for environmental data, features such as the average wind speed, the change trend of humidity, and the occurrence frequency of extreme weather events are extracted; for electrical parameter data, the current fluctuation range, voltage abnormal peak value, and short-term power change rate, etc. are extracted.

[0057] Next, for image and video data, a convolutional neural network (CNN) is used to extract the key visual features of the images. These features include the crack edges on the surface of the power transmission line, the distribution of corrosion areas, and the visualization information of insulator loss. To improve the extraction accuracy, the original image data can be preprocessed, such as optimizing the image quality through edge enhancement and denoising algorithms, so as to ensure that the extracted features are accurate and have physical significance.

[0058] For the infrared temperature distribution data, the distribution of hot spots is identified through a thermal imaging feature extraction algorithm. Specifically, the infrared imaging data is associated with the spatial coordinates of the transmission line, and the area, temperature gradient, and temperature rise amplitude relative to the normal area of the hot spot area are calculated. Through this process, potential high-temperature hazard areas on the transmission line can be judged, such as conductor overload or joint overheating.

[0059] All the extracted feature data is further organized into structured data assets. In this process, an adaptive feature extraction algorithm is used to dynamically weight various features to ensure that the most relevant features are preferentially extracted in different scenarios. For example, under strong wind weather conditions, the weight of the wind speed change in the environmental features will be dynamically increased to prioritize marking the potential impact of wind speed on the operation of the transmission line. The finally generated data assets are stored in a multi-dimensional structure, including environmental feature labels, defect feature labels, and hazard feature labels. Environmental feature labels mainly describe the potential impact of weather, temperature, humidity, and wind speed on the operation of the transmission line; defect feature labels are used to mark equipment problems such as cracks, corrosion, and insulator loss; hazard feature labels include high-risk area identifications based on electrical parameters and thermal imaging data.

[0060] The structured data assets generated by the above method cover the multi-dimensional operation status information of the transmission line and can provide high-quality inputs for hazard assessment and risk prediction in subsequent steps. This process not only ensures the accuracy and comprehensiveness of the data but also improves the applicability and dynamic response ability of the feature extraction results.

[0061] Furthermore, the adaptive feature extraction algorithm is used to analyze the fusion data set to generate transmission line data assets with structured features, including:

[0062] Extract multi-modal data from the fusion data set, including environmental data, physical state data, electrical parameter data, infrared temperature distribution data, and visual feature data, and group and preprocess the multi-modal data. Among them, the environmental data and electrical parameter data are normalized, and the visual feature data is subjected to image enhancement processing;

[0063] Based on the preprocessed multi-modal data, environmental features, defect features, and hazard features are respectively extracted using feature extraction algorithms. Specifically, environmental features describing the impact of weather, temperature, humidity, and wind speed on the operation of the transmission line are extracted based on environmental data; defect features of the crack or corrosion area on the surface of the transmission line are extracted based on visual feature data and infrared temperature distribution data; hazard features that may lead to the operation risk of the transmission line are extracted based on physical state data, the situation of buildings and trees in the transmission line corridor, and electrical parameter data;

[0064] Generate environmental feature tags, defect feature tags, and potential hazard feature tags for the extracted environmental features, defect features, and potential hazard features respectively, and associate them with their time and space information to generate labeled transmission line data tags.

[0065] According to the generated labeled transmission line data tags, organize the environmental feature tags, defect feature tags, and potential hazard feature tags into a structured transmission line data asset. The transmission line data asset includes a time dimension, a space dimension, and a feature dimension, and is used for the assessment of the operating state of the transmission line and the prediction of potential hazards.

[0066] Extract multimodal data from the fusion dataset, including environmental data, physical state data, electrical parameter data, infrared temperature distribution data, and visual feature data. These data have been calibrated in time and space and dynamically fused in the previous steps, and have time consistency and space accuracy. To improve the accuracy and efficiency of data processing, it is necessary to group and preprocess the extracted data. Specifically, due to the differences in numerical ranges and units between environmental data and electrical parameter data, normalization processing is performed to eliminate the influence of dimensions. For example, data such as wind speed, humidity, and voltage are standardized to the range of 0 to 1. Visual feature data undergoes image enhancement processing, including adjusting the image contrast and brightness, removing low-quality images, and improving the recognizability of crack or corrosion areas through edge enhancement algorithms.

[0067] Based on the preprocessed multimodal data, use an adaptive feature extraction algorithm to extract environmental features, defect features, and potential hazard features respectively. The extraction of environmental features mainly targets environmental data, and generates feature descriptions by analyzing the impacts of weather, temperature, humidity, wind speed, etc. on the operation of transmission lines. For example, high wind speed may cause an increase in the swing amplitude of the conductor, and rising humidity may accelerate the corrosion of insulators. Defect feature extraction targets visual feature data and infrared temperature distribution data, and uses specific algorithms to detect crack, corrosion areas, and infrared temperature rise hot spots in images. For example, the length and depth of cracks are extracted through image segmentation technology, and the scope of the corrosion area is marked; abnormal hot spot areas are identified through temperature gradient analysis. The extraction of potential hazard features is based on physical state data and electrical parameter data. For example, potential mechanical structure problems are identified through abnormal changes in vibration frequency, and potential short circuit or overload hazards are detected through current or voltage fluctuations.

[0068] After extraction, generate corresponding tags for environmental features, defect features, and potential hazard features, and associate them with time and space information. Each feature tag not only contains the descriptive information of the feature but also records the time point and spatial location where the feature appears. For example, an environmental feature tag may indicate that the wind speed is 20 meters per second at a certain time point, and the affected area is a specific section of a certain transmission line; a defect feature tag may mark that a crack appears on the insulator of a certain pole tower, and the crack length is 2 centimeters; a potential hazard feature tag may indicate that the current fluctuation of a certain section of the transmission line exceeds the safety threshold at a specific time point.

[0069] Finally, all feature tags are organized into a structured transmission line data asset, which includes a time dimension, a space dimension, and a feature dimension. For example, the time dimension records the occurrence time of the feature, the space dimension records the geographical location or the corresponding transmission line section of the feature, and the feature dimension classifies and records the detailed descriptions of environmental features, defect features, and potential hazard features. This structured data asset provides a basis for a comprehensive assessment of the operating state of the transmission line and also supports the accurate prediction of potential hazards.

[0070] Step S104: Evaluate the triggering possibility of environmental conditions on transmission line defects based on the association between the environmental feature tags and the defect feature tags.

[0071] The goal of step S104 is to evaluate the triggering possibility of environmental conditions on transmission line defects based on the association between the environmental feature tags and the defect feature tags. The following is a detailed description of the specific implementation process.

[0072] First, extract environmental feature tags and defect feature tags from the structured transmission line data asset. Environmental feature tags include key parameters such as wind speed, temperature, and humidity. These parameters are collected through sensors or satellites and have been calibrated and dynamically fused in time and space, having time and space consistency. Defect feature tags include information such as the crack length, corrosion area, insulator missing situation on the surface of the transmission line, and the temperature rise area identified in the infrared thermal imaging. These tags are generated by the feature extraction algorithm in the previous steps.

[0073] Next, establish the association relationship between environmental features and defect features. Specifically, by analyzing historical data, determine the influence degree of each environmental parameter on the triggering of specific defects. For example, the association between wind speed and crack propagation can be quantified by the growth rate of crack length with the change of wind speed; the arc effect caused by temperature and insulator missing is represented by the statistical relationship between temperature rise and failure rate. These association relationships can be modeled using regression models, decision trees, or other statistical analysis methods to form a complete set of environment-defect association rules.

[0074] Then, the above rules are used to evaluate the real-time collected environmental feature data and defect feature data. Specifically, by calculating one by one the possibility of each defect being triggered under the current environmental conditions. For example, to calculate the possibility of wind speed causing crack propagation under the current conditions, the current wind speed can be substituted into the correlation model, and combined with the initial value of the crack length and its propagation rate for prediction. Similarly, for the possibility of temperature rise causing insulator loss, by calculating the abnormality degree of the temperature rise area and combining with the historical defect records of the insulators for analysis. For complex correlation situations, a multi-variable model or a conditional probability-based analysis method can be introduced to more accurately evaluate the comprehensive impact of the environment on defect triggering.

[0075] After the above calculations are completed, the triggering possibilities of each defect are standardized to ensure that the possibility values of different categories can be compared on the same scale. By comprehensively scoring the possibilities of various defects, an evaluation result of the overall defect triggering risk is generated. This score can be divided into multiple levels according to needs, such as low risk, medium risk, and high risk, for use in subsequent steps for hidden danger propagation path analysis and risk warning.

[0076] Finally, all evaluation results are associated with the time and space information corresponding to the environmental and defect features, forming a set of detailed defect triggering possibility evaluation data. These data can be directly used for subsequent hidden danger propagation path modeling, or can provide targeted suggestions for operation and maintenance personnel, thereby optimizing the inspection and maintenance strategies of transmission lines. The whole process is simple and efficient, ensuring the accuracy and practicality of the evaluation results.

[0077] Furthermore, the evaluating the triggering possibility of environmental conditions on transmission line defects according to the association between the environmental feature labels and the defect feature labels includes:

[0078] Obtain environmental feature labels, including key environmental parameters such as wind speed, temperature and humidity, precipitation, and their corresponding spatio-temporal distributions;

[0079] Obtain defect feature labels, including crack length on the surface of the transmission line, corrosion area, and infrared temperature anomaly hot spot distribution.

[0080] First, extract key parameters from the environmental feature labels, including wind speed , humidity , precipitation and their corresponding time and space distributions. These data are sourced from meteorological sensors installed around the transmission line or environmental monitoring satellites. Secondly, extract the physical damage information of the transmission line from the defect feature labels, such as crack length corrosion area , and the hot spot temperature identified by infrared imaging These characteristic data have been generated through the previous steps to ensure temporal and spatial consistency.

[0081] According to the following formula (1), calculate the possibility of crack propagation under different environmental conditions:

[0082] ;

[0083] Where is the crack trigger factor, reflecting the possibility of crack propagation under different environmental conditions; is the crack length; is the wind speed; , , are sensitivity parameters obtained through historical data and model training;

[0084] The crack trigger factor reflects the possibility of crack propagation under the current environmental conditions. The crack length is identified through a visual feature extraction algorithm. For example, the initial crack length on a certain transmission line is 10 millimeters. The wind speed is sourced from real-time meteorological data. For example, the current wind speed is 20 meters per second. The sensitivity parameters are obtained through regression analysis or model training of historical crack propagation data. For example, under similar wind speed conditions, for every 1 millimeter increase in crack length, the propagation speed shows a certain growth trend. The term in the exponential part of the formula quantifies the coupling effect between wind speed and crack length, that is, the higher the wind speed, the more significant the impact on the propagation of long cracks.

[0085] According to the following formula (2), calculate the possibility of corrosion development:

[0086] ;

[0087] Where is the corrosion trigger factor, reflecting the possibility of corrosion development; is the corrosion area; is the humidity; is the precipitation; , , are sensitivity parameters obtained through historical data and model training;

[0088] The corrosion trigger factor quantifies the accelerating effect of humidity and precipitation on corrosion development. The corrosion area is obtained by a visual feature extraction algorithm. For example, the corrosion area of a certain region is 15 square millimeters. The humidity and precipitation It is provided by a meteorological sensor. For example, the current humidity is , and the precipitation is 5 mm. The parameter is obtained by analyzing the expansion data of the corrosion area under different humidity and precipitation conditions. For example, when the humidity is high, the logarithmic growth factor of the corrosion rate will increase significantly, and the influence of precipitation is further amplified by the exponential term.

[0089] According to the following formula (3), calculate the possibility of high temperature accelerating crack propagation:

[0090] ;

[0091] where is the temperature rise trigger factor, reflecting the possibility of high temperature accelerating crack propagation; is the temperature value based on the infrared temperature hot spot; is the crack length; , are influence coefficients, obtained through historical data and model training;

[0092] The temperature rise trigger factor reflects the combined effect of the infrared temperature hot spot and the crack length. The hot spot temperature is captured by infrared imaging. For example, the temperature at a certain point is 80 degrees Celsius. The parameter represents the direct and indirect effects of high temperature on crack propagation, usually obtained by fitting experimental data. The square root term of the crack length describes the non-linear acceleration effect of high temperature on longer cracks, that is, the longer the crack, the greater the potential risk of high temperature on its propagation.

[0093] According to the following formula (4), calculate the comprehensive defect trigger possibility value :

[0094] ;

[0095] where , , are weight coefficients, obtained through historical data and model training;

[0096] If is greater than or equal to the specified threshold, generate a defect trigger warning and record the risk type; if is less than the specified threshold, record the low influence degree of the current environmental conditions on defect triggering.

[0097] Step S105: According to the evaluation results of the hidden danger characteristic tags and environmental conditions, construct a graph network model with the transmission line as the node and the sensor data as the edge weight. The graph network model is used to analyze the defect propagation path of the transmission line and its corridor; based on historical inspection data and environmental dynamic parameters, use the graph network model to predict the time when hidden dangers occur.

[0098] The core of step S105 is to construct a graph network model with the transmission line as the node and the sensor data as the edge weight based on the evaluation results of the hidden danger characteristic tags and environmental conditions, and use this model to analyze the defect propagation path and predict the time when hidden dangers occur. The following is the detailed implementation process of this step.

[0099] First of all, it is necessary to define the basic structure of the graph network model. The transmission line and its affiliated facilities (such as poles, insulators, etc.) are used as the nodes of the graph network. The attributes of each node include the evaluation results of its hidden danger characteristic tags and environmental conditions, such as crack length, corrosion area, hot spot temperature, and wind speed and humidity in adjacent areas. The edges between nodes represent physical or functional correlations, such as electrical connections between transmission lines, geographical proximity, or correlations due to common exposure to similar environmental conditions. The weight of the edge is determined by the following factors: one is the geographical distance, and nodes closer in distance have a higher propagation probability; the other is the correlation degree of hidden danger characteristics, for example, both nodes have similar cracks or temperature anomalies; the third is the similarity of environmental conditions, for example, adjacent nodes are affected by the same wind speed or humidity.

[0100] Next, construct the basic topological structure of the graph network by setting the initial attributes of nodes and edges. In the node initialization stage, the hidden danger characteristics and environmental conditions of each node are used as input vectors. For example, the initial hidden danger characteristic vector of a node may include crack length, temperature rise amplitude, and the defect triggering probability calculated based on historical data. The weight of the edge is initialized based on the above geographical, hidden danger, and environmental correlations, and specific values are calculated through formulas. After the initialization of all nodes and edges is completed, a complete weighted graph network is formed.

[0101] Subsequently, analyze the defect propagation path on the graph network. Through the graph network propagation algorithm (such as graph convolutional network, GCN), the hidden danger propagation state of each node is iteratively updated. Specifically, the hidden danger propagation state of each node is calculated by weighting its own characteristics and the characteristics of neighboring nodes through the edge weight. For example, the hidden danger propagation possibility of a node is affected by the hidden danger characteristics and propagation states of adjacent nodes, and at the same time, the role of the edge weight is considered, so that nodes with a longer distance or lower hidden danger correlation contribute less to the propagation. After multiple rounds of propagation iteration, the hidden danger propagation state tends to be stable, and at this time, the main path of hidden danger propagation and high-risk areas can be determined.

[0102] On this basis, by combining historical inspection data and environmental dynamic parameters, the time of potential hazard occurrence is further predicted. The time prediction is mainly based on the hazard propagation state of the nodes and the time distribution of hazard development under similar conditions in historical data. For example, the crack growth rate of a node combined with the time model of historical data can predict when the crack may develop to the critical value. Similarly, the growth trend of the hot spot temperature combined with the overheating fault time in the historical records can estimate the time point of failure. Through these calculations, the time of potential hazard occurrence for each node is assigned and recorded in the graph network model.

[0103] Finally, the results of the entire graph network model include the analysis of the hazard propagation path and the prediction of the time of potential hazard occurrence for each node. The propagation path provides the possible trend of hazard expansion, while the time prediction calibrates the time distribution of hazard risk for each node. These results provide a direct basis for generating risk warnings and formulating maintenance plans in the subsequent steps, and ensure the accuracy and practicality of the prediction.

[0104] Furthermore, according to the evaluation results of the hazard feature tags and environmental conditions, a graph network model with transmission lines as nodes and sensor data as edge weights is constructed, including:

[0105] Taking the transmission line as a node of the graph network, the attributes of the node include hazard feature tags, environmental feature tags, and historical inspection data features.

[0106] In the construction of the graph network, first, the transmission line and its affiliated facilities (such as poles, insulators, etc.) are defined as nodes of the graph network. The attributes of each node include hazard feature tags (such as crack length, corrosion area, temperature anomaly value, etc.), environmental feature tags (such as wind speed, humidity, precipitation, etc.), and historical inspection data features (such as the number of defect repairs, historical hazard records, etc.). These attributes are obtained through the multi-modal data extracted and processed in the previous steps and have temporal and spatial consistency.

[0107] According to the following formula 5, calculate the physical adjacency weight between nodes:

[0108] ;

[0109] where represents the physical adjacency weight between node and node ; represents the physical distance between node and node ; represents the electrical transmission distance between node and node ; is the distance sensitivity coefficient, obtained through historical analysis, and is used to adjust the influence of distance on the weight;

[0110] According to the following formula (6), calculate the environmental similarity weight between nodes:

[0111] ;

[0112] where, represents the environmental similarity weight between node and node ; is the Euclidean distance between node and node , and the Euclidean distance is calculated based on the similarity of environmental characteristics between nodes.

[0113] For example, if the environmental characteristics of node are wind speed of 10 m / s, humidity , and precipitation of 5 mm, and the environmental characteristics of node are wind speed of 12 m / s, humidity , and precipitation of 4 mm, then their Euclidean distance is:

[0114] ;

[0115] When the environmental conditions are more similar, the Euclidean distance is smaller, and the environmental similarity weight is higher; conversely, it is lower.

[0116] According to the following formula (7), calculate the hidden danger correlation weight between nodes:

[0117] ;

[0118] where, represents the hidden danger correlation weight between node and node ; is the included angle between the hidden danger feature vectors of node and node .

[0119] represents the hidden danger correlation weight between node and node . The parameter is the included angle between the hidden danger feature vectors of the two nodes. For example, the hidden danger characteristics of node and are crack length of 10 mm and corrosion area of 20 square millimeters, and crack length of 8 mm and corrosion area of 25 square millimeters respectively, and their hidden danger feature vectors are and The cosine value of the included angle can be calculated by the inner product formula:

[0120] ;

[0121] The closer the cosine value is to 1, the higher the similarity of the hidden danger features of the two nodes.

[0122] According to the following formula (8), calculate the edge weight between nodes:

[0123] ;

[0124] where, is the edge weight between node and node ; is the weighting coefficient, determined according to historical data, and satisfies .

[0125] Construct a weighted graph based on the nodes and edge weights. Construct a weighted graph , where represents the set of transmission line nodes, represents the set of edges; Initialize the attribute vector of each node, and combine the node features with the weight distribution of its adjacent nodes to generate the initial hidden danger propagation state.

[0126] Furthermore, based on the historical inspection data and environmental dynamic parameters, using the graph network model to predict the time of hidden danger occurrence includes:

[0127] First, based on the nodes and edge weights constructed by the graph network model, simulate the propagation process of hidden dangers. In the graph network, each node represents a part of the transmission line or related facilities, and its attributes include hidden danger feature labels and environmental feature labels. The edge weights reflect the physical distance, environmental similarity, and hidden danger correlation between nodes. The simulation of hidden danger propagation is based on the graph convolutional network (GCN), and the hidden danger propagation state of the nodes is updated through layer-by-layer iterative calculation.

[0128] According to the following formula (9), for each node, combine the attribute vectors and edge weights of its adjacent nodes to update the hidden danger propagation state of the node:

[0129] ;

[0130] where, represents the state of node after the -th layer of propagation; represents the state of node after the -th layer of propagation; is the node The set of neighbor nodes of is the weight matrix obtained through training; is the activation function;

[0131] Repeat the propagation iteration until the node state converges.

[0132] In this formula, represents the node after propagation in the th layer. The initial state is the initial hidden feature of the node , such as the feature vector obtained by comprehensively calculating the crack length, corrosion area, and temperature rise value. The set of neighbor nodes includes all nodes connected to the node . For example, if the node is connected to three other nodes, its neighbor set is . represents the state of the neighbor node after propagation in the th layer.

[0133] The weight is the edge weight between the node and , which is obtained from the weight calculation formula (such as the combination of physical adjacency weight, environmental similarity weight, and hidden danger correlation weight). The matrix is the weight matrix obtained through training with historical data, used to perform weighted combination of different features of hidden danger propagation. For example, if historical data shows that hidden dangers related to cracks propagate faster, the matrix has a higher weight value corresponding to the crack feature. The activation function is a non - linear transformation function. Commonly used choices include ReLU or sigmoid, and its role is to limit the range of state values to make them suitable for subsequent calculations.

[0134] The propagation process is achieved through hierarchical iteration. Each time of propagation updates the state of the node until the node state converges. For example, after the first - layer propagation, the state of the node will be updated by integrating the states of its neighbor nodes and edge weight information; during the second - layer propagation, the updated state continues to be fused with information from farther nodes. The number of iterations of this process is controlled by the parameter , usually depending on the depth of the network and the range of information transmission.

[0135] After the update of the hidden danger propagation state is completed, according to formula 10 below, calculate the occurrence time of the potential hidden danger of the node :

[0136] ;

[0137] Among them, is the initial hidden danger occurrence time, is the propagation iteration times; represents the layer propagation after the node status.

[0138] In the formula, is the potential hidden danger occurrence time of node , is the initial hidden danger occurrence time, for example, set as the current time or the first discovery time of the known hidden danger based on historical data. The hidden danger propagation status after each propagation is mapped to the interval by the sigmoid function, reflecting the cumulative influence degree of the hidden danger. The possibility of the cumulative hidden danger is calculated by summing the status values of all propagation layers.

[0139] For example, assume that the initial hidden danger status of node is . After three propagations, the status values are . When calculating the hidden danger occurrence time, the initial time , then:

[0140] ;

[0141] Specific calculations show that each item is respectively and 0.750. Therefore, the hidden danger occurrence time , indicating that the hidden danger of node may occur after about 2 unit times.

[0142] Step S106: Based on the prediction results of the graph network model, generate a defect risk level warning, and generate a hidden danger distribution map of the transmission line in combination with environmental meteorological data. The hidden danger distribution map is used to assist in formulating inspection and maintenance plans.

[0143] The goal of step S106 is to generate a defect risk level warning based on the prediction results of the graph network model, and generate a hidden danger distribution map of the transmission line in combination with environmental meteorological data to assist in formulating efficient inspection and maintenance plans. The implementation method of this step is described in detail below.

[0144] First, based on the prediction results of the graph network model, comprehensively analyze the hidden danger propagation status, risk likelihood, and predicted hidden danger occurrence time for each node. For each node, calculate its defect risk score by combining its hidden danger triggering likelihood, environmental dynamic parameters, and importance on the propagation path. This score can be determined by integrating multiple factors through a weighted formula. For example, the risk score is inversely proportional to the predicted hidden danger occurrence time, that is, nodes with earlier hidden danger occurrence have higher risk scores, while also considering the impact of hidden danger propagation of other nodes on this node on the propagation path. These scores are divided into multiple levels, such as low risk, medium risk, and high risk, and each node is labeled with the corresponding risk level.

[0145] When generating the risk level warning, mark the high-risk areas as the key. By analyzing the propagation path of the graph network model, identify the central nodes of hidden danger propagation and the scope they may affect. In the high-risk areas, record the risk level and the main risk types, such as the potential wire breakage risk caused by crack propagation, the overheating hidden danger caused by the temperature rise hot spot, etc. These warning messages will be classified and sorted to output the risk level warning data for subsequent analysis.

[0146] Next, combine the environmental meteorological data to map the risk information to the geographical space and generate a hidden danger distribution map of the transmission line. The hidden danger distribution map intuitively presents the overall operation risk status of the transmission line by overlaying the environmental dynamic data and the risk level information. For example, high-risk areas on the map can be marked with colors, and the darker the color, the higher the risk; in addition, environmental data such as wind speed direction and humidity distribution can be overlaid on the map to provide context information for further analysis. The hidden danger distribution map also includes detailed information about specific nodes, such as the specific location, risk level, and risk type of the node. This information can help the operation personnel quickly locate the high-risk areas.

[0147] Finally, use the hidden danger distribution map as the basis for the inspection and maintenance plan. According to the risk level and hidden danger type in the distribution map, prioritize arranging inspection personnel to check the high-risk areas, such as areas with significant crack propagation or abnormal temperature rise. At the same time, a maintenance plan can be formulated based on the distribution map, such as repairing in advance the hidden danger nodes that may cause serious consequences or optimizing the inspection path to cover the high-risk areas. This process provides precise support for operation and maintenance by analyzing the spatial characteristics and risk levels of the hidden danger distribution map, ensuring the reasonable allocation of inspection and maintenance resources.

[0148] Furthermore, generating a defect risk level warning based on the prediction results of the graph network model and generating a hidden danger distribution map of the transmission line in combination with environmental meteorological data includes:

[0149] According to the hidden danger propagation state values of each node in the graph network model, combined with environmental meteorological data and historical inspection records, calculate the defect risk score of each node, and classify the nodes into low-risk, medium-risk, and high-risk levels according to the preset level standard;

[0150] Based on the geographical location and risk level of the nodes, map the risk information of the transmission line and its affiliated facilities to the geographical space to generate a hidden danger distribution map of the transmission line, which is used to display the distribution range and risk intensity of high-risk areas;

[0151] According to the hidden danger distribution map, identify the key nodes and influence range of high-risk areas, generate defect risk warning information, and provide corresponding inspection and maintenance suggestions.

[0152] First, obtain the hidden danger propagation state values of each node according to the prediction results of the graph network model. These values represent the degree of risk accumulation of the nodes in the hidden danger propagation path. Combine environmental meteorological data (such as wind speed, humidity, precipitation, etc.) and historical inspection records (such as the number of defect repairs and inspection frequencies) to calculate the defect risk score of each node. The calculation of the risk score comprehensively considers the influence of the hidden danger propagation state value, the promotion effect of the current environmental conditions on the triggering of hidden dangers, and the probability of defects occurring at this node in historical data. For example, if a node has a high hidden danger propagation state value, and the wind speed and precipitation in the area where the node is located are at abnormal levels, and the historical records show that there are multiple defect repair records for this node, then the risk score of this node will increase significantly.

[0153] The calculated defect risk scores are standardized to a unified range, and the nodes are classified into low-risk, medium-risk, and high-risk levels according to the preset risk level standard. The level classification can be achieved by setting the threshold of the risk score. For example, low risk is a score less than 0.4, medium risk is 0.4 to 0.7, and high risk is above 0.7. This level classification is convenient for clarifying the risk degree of each node and providing a basis for subsequent visualization and decision-making.

[0154] Next, associate the risk level of the nodes with their geographical locations to display the risk distribution in the geographical space. By mapping the location information and risk level of each node to the geographical space, a hidden danger distribution map of the transmission line is generated. In the hidden danger distribution map, high-risk areas are marked with prominent colors or icons. For example, red is used to represent high-risk areas, yellow is used to represent medium-risk areas, and green is used to represent low-risk areas. The hidden danger distribution map not only shows the node risk levels but also displays the intensity distribution of high-risk areas in a gradient manner. For example, nodes with higher risk scores and their adjacent areas will be shown as dark red, while nodes with lower risks will be shown as light yellow or green. This visualization method intuitively reflects the risk status of the transmission line and is convenient for operators to quickly identify key hidden dangers.

[0155] Based on the hidden danger distribution map, the key nodes and influence scope of high-risk areas can be further analyzed. Key nodes refer to nodes with a relatively high risk level and significant influence on the spread of hidden dangers, such as nodes located on the main propagation path or nodes with strong associations with multiple neighboring nodes. The influence scope refers to the potential hidden danger area spread from high-risk nodes to surrounding nodes. By analyzing the key nodes and influence scope, the main concentrated areas of risks and the possibility of hidden danger propagation can be determined, thus generating targeted defect risk warning information.

[0156] Finally, combined with the generated warning information, corresponding inspection and maintenance suggestions are provided. For example, for high-risk areas, it is recommended to arrange priority inspections, check the actual status of key nodes, and conduct detailed inspections on relevant equipment; for medium-risk areas, regular maintenance can be planned to reduce the possibility of hidden danger accumulation; for low-risk areas, it is recommended to maintain the normal inspection frequency. Such maintenance suggestions can optimize the allocation of inspection resources, improve inspection efficiency, and ensure that high-risk areas receive priority attention.

[0157] Through the above steps, based on the prediction results of the graph network model, not only the accurate assessment of transmission line hidden dangers is achieved, but also clear guidance is provided for actual operation and maintenance through visualization and the generation of warning information.

[0158] Furthermore, the transmission line inspection data asset management method described above further includes:

[0159] A visualization model of the transmission line is constructed through laser point cloud data and a three-dimensional geographic information system. The visualization model is used to dynamically display the defect hot spot map and hidden danger distribution map of the transmission line, and support users to select display dimensions and analysis results as needed.

[0160] First, the three-dimensional geographic information system is used to model the spatial structure of the transmission line. By collecting the geographic coordinate data of the transmission line and its surrounding environment, including tower positions, line directions, heights, etc., and combining with three-dimensional terrain data, a basic three-dimensional geometric model of the transmission line is generated. This model needs to accurately reflect the distribution of the transmission line in the real space, including the starting point, ending point of the line, and the relative height differences and distances between each tower. In addition, the bottom terrain of the transmission line model is rendered using high-precision terrain data to provide a real geographic background.

[0161] Based on the 3D model, dynamic data is further superimposed to achieve the visualization of temperature flow distribution, defect hotspot map, and potential hazard distribution map. The data of temperature flow distribution is sourced from the surface temperature information of the transmission line collected by infrared cameras or temperature sensors. Through interpolation calculation and rendering of the temperature data, the temperature distribution is superimposed on the surface of the transmission line in the form of color gradient. The defect hotspot map is generated based on the data of defect feature tags. For example, areas with a relatively large crack length can be marked in dark red, and areas with severe corrosion are marked in orange. The potential hazard distribution map is generated based on the analysis results of the graph network model, and areas with a higher risk score are highlighted, while the corresponding potential hazard types are marked on the map, such as high-temperature hotspots, mechanical damage, or external threats, etc.

[0162] To enhance the user's interactive experience, the visualization model supports users to select display dimensions and analysis results as needed. For example, users can select through the interface to only view the temperature flow distribution in a specific area, or view only the areas involving high-risk potential hazards through filtering conditions. Users can also adjust the perspective to zoom in and view the detailed information of a certain tower, or zoom out to view the panoramic view of the entire transmission line. In addition, the system supports the hierarchical display function, and users can choose to superimpose the temperature flow distribution and the potential hazard distribution map simultaneously, or view the defect hotspot map alone.

[0163] The visualization model also supports the real-time update function. Through the linkage with the transmission line sensors and the background analysis system, the model can dynamically update the status information of the transmission line. For example, when the sensor detects a new temperature anomaly area or the analysis result of the potential hazard propagation path changes, the model can re-render and display the latest distribution situation within seconds, thus ensuring that users obtain the latest status of the transmission line.

[0164] Through the above implementation method, the 3D geographic information system can not only provide the spatial visualization of the transmission line, but also intuitively present the complex operation status and potential hazard information to users, while providing flexible interaction functions, enabling users to customize analysis and display solutions according to specific needs. This method has important application value in the operation monitoring, potential hazard investigation, and maintenance plan formulation of transmission lines.

[0165] In the above embodiment, a method for managing transmission line inspection data assets is provided. Correspondingly, the present application also provides a system for managing transmission line inspection data assets. Please refer to Figure 2 which is a schematic diagram of an embodiment of the system for managing transmission line inspection data assets of the present application. Since this embodiment, that is, the second embodiment, is basically similar to the method embodiment, the description is relatively simple, and for related parts, reference can be made to the partial description of the method embodiment. The system embodiment described below is only illustrative.

[0166] A power transmission line inspection data asset management system provided by the second embodiment of the present application includes:

[0167] An acquisition unit 201, configured to obtain multimodal data related to power transmission line inspections through various sensors installed on the power transmission line and its surrounding environment, sensors carried by helicopters and drones, mobile terminals carried by inspection personnel, and data acquisition devices of environmental monitoring satellites. Among them, the multimodal data includes physical state data of the power transmission line, electrical parameter data, environmental meteorological data, appearance photo and video data of the power transmission line, temperature distribution data captured by infrared cameras, defect data of the power transmission line, and potential hazard data of the power transmission channel;

[0168] An obtaining unit 202, configured to perform synchronous calibration and dynamic fusion on the multimodal data based on a spatio-temporal association model to obtain a fusion data set with time consistency and spatial accuracy;

[0169] A generating unit 203, configured to parse the fusion data set by using an adaptive feature extraction algorithm to generate power transmission line data assets with structured features. The power transmission line data assets include environmental feature tags, defect feature tags, and potential hazard feature tags. Among them, the environmental feature tags describe the potential impact of weather, temperature, humidity, and wind speed on the operation of the power transmission line;

[0170] An evaluation unit 204, configured to evaluate the triggering possibility of environmental conditions on power transmission line defects according to the association between the environmental feature tags and the defect feature tags;

[0171] A constructing unit 205, configured to construct a graph network model with the power transmission line as nodes and sensor data as edge weights according to the evaluation results of the potential hazard feature tags and environmental conditions. The graph network model is used to analyze the defect propagation path of the power transmission line and its channel, and predict the time and spatial distribution of potential hazards based on historical inspection data and environmental dynamic parameters;

[0172] A combining unit 206, configured to generate a defect risk level warning based on the prediction result of the graph network model, and generate a power transmission line potential hazard distribution map in combination with environmental meteorological data. The potential hazard distribution map is used to assist in formulating inspection and maintenance plans.

[0173] A third embodiment of the present application provides an electronic device, and the electronic device includes:

[0174] A processor;

[0175] A memory, configured to store a program, and when the program is read and executed by the processor, it executes a power transmission line inspection data asset management method provided in the first embodiment of the present application.

[0176] The fourth embodiment of the present application provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, it executes a method for managing inspection data of power transmission lines provided in the first embodiment of the present application.

[0177] Although the present application is disclosed above in preferred embodiments, it is not intended to limit the present application. Any person skilled in the art can make possible changes and modifications without departing from the spirit and scope of the present application. Therefore, the protection scope of the present application should be subject to the scope defined by the claims of the present application.

Claims

1. A method for managing transmission line patrol data assets, characterized in that: include: Through the use of various sensors installed on the transmission lines and the surrounding environment, sensors carried by helicopters and drones, mobile terminals carried by patrol personnel, and data acquisition devices of environmental monitoring satellites, multimodal data related to the inspection of transmission lines are obtained, wherein the multimodal data includes physical state data of the transmission lines, electrical parameter data, three-dimensional laser point cloud data, environmental meteorological data, appearance photos and video data of the transmission lines, temperature distribution data and infrared photos captured by infrared cameras, defect data of the transmission lines, and hidden danger data of the transmission channels; Based on the spatiotemporal correlation model, the multimodal data are synchronously calibrated and dynamically fused to obtain a fused data set with temporal consistency and spatial accuracy; The fused data set is parsed using an adaptive feature extraction algorithm to generate a transmission line data asset with structured features, wherein the transmission line data asset includes an environmental feature label, a defect feature label, and a hidden danger feature label, wherein the environmental feature label describes the potential impact of weather, temperature and humidity, and wind speed on the operation of the transmission line; According to the association between the environmental characteristic tag and the defect characteristic tag, evaluating the possibility of environmental conditions triggering the transmission line defect; According to the evaluation results of the hidden danger feature labels and environmental conditions, a graph network model is constructed with the transmission line as the node and the sensor data as the edge weight, and the graph network model is used to analyze the defect propagation path of the transmission line and its channel; based on the historical inspection data and environmental dynamic parameters, the graph network model is used to predict the time when the hidden danger occurs; Based on the prediction results of the graph network model, a defect risk level warning is generated, and a transmission line hidden danger distribution map is generated in combination with environmental meteorological data. The hidden danger distribution map is used to assist in formulating inspection and maintenance plans.

2. The power transmission line patrol data asset management method according to claim 1, characterized in that: Also includes: A visualization model of the transmission line is constructed through laser point cloud data and a three-dimensional geographic information system. The visualization model is used to dynamically display the defect hotspot map and hidden danger distribution map of the transmission line, and supports users to select display dimensions and analysis results as needed.

3. The power transmission line patrol data asset management method according to claim 1, characterized in that: The synchronous calibration and dynamic fusion of the multimodal data based on the spatiotemporal correlation model to obtain a fused data set with temporal consistency and spatial accuracy includes: Acquire multimodal data of transmission lines; According to the timestamp information of the multimodal data, the data is sorted by the acquisition time to generate a preliminary time series; the data in the preliminary time series is calibrated by a dynamic time warping method, and the multimodal data collected at different times are time-aligned to generate time series data with consistent time; Based on the temporally consistent time series data and the geographic location information of the acquisition equipment in the multimodal data, combined with the geographic topology of the transmission line, the data is mapped to a unified geographic grid; the spatial distribution characteristics of the mapped data are interpolated and matched using a geographically weighted regression method to eliminate the spatial deviation caused by the position offset of the acquisition equipment, and generate spatially consistent multimodal data; Assign priorities to physical state data, electrical parameter data, ambient meteorological data, infrared temperature distribution data, and visual feature data based on the source characteristics, historical accuracy, and environmental conditions of each data type; According to the monitoring results of the transmission line operation status, dynamically adjust the priorities of the physical status data, electrical parameter data, environmental meteorological data, infrared temperature distribution data and visual feature data; According to the adjusted data priority, the temporally consistent and spatially consistent multimodal data are weightedly superimposed and fused to generate a fused dataset with temporal consistency and spatial accuracy.

4. The transmission line patrol data asset management method according to claim 3, characterized in that: The method of dynamically adjusting the priorities of the physical state data, electrical parameter data, environmental meteorological data, infrared temperature distribution data and visual feature data according to the real-time monitoring results of the transmission line operation status includes: According to the monitoring results of the transmission line operation status, the transmission line is divided into a normal state and an abnormal state; wherein the normal state means that the transmission line operation parameters are within the set safety threshold range, and the abnormal state means that the transmission line operation parameters are outside the set safety threshold range; wherein, in the case where the monitoring results cannot be obtained, the defect distribution is considered; When the operating status is normal, the initial priority of the data type is maintained, and data types with higher historical accuracy and credibility are given priority; when the operating status is abnormal, the data priority is dynamically adjusted according to the abnormal category.

5. The power transmission line patrol data asset management method according to claim 1, characterized in that: The method of parsing the fused data set using an adaptive feature extraction algorithm to generate a transmission line data asset with structured features includes: Extracting multimodal data from the fused data set, including environmental data, physical state data, electrical parameter data, infrared temperature distribution data, and visual feature data, and grouping and preprocessing the multimodal data, wherein the environmental data and the electrical parameter data are normalized, and the visual feature data are image enhanced; Based on the pre-processed multimodal data, environmental features, defect features and hidden danger features are extracted respectively using a feature extraction algorithm, specifically including extracting environmental features that describe the impact of weather, temperature, humidity and wind speed on the operation of the transmission line based on environmental data; extracting defect features of cracks or corrosion areas on the surface of the transmission line based on visual feature data and infrared temperature distribution data; extracting hidden danger features that may cause transmission line operation risks based on physical state data, the conditions of buildings and trees in the transmission line channel and electrical parameter data; The extracted environmental features, defect features and hidden danger features are respectively generated into environmental feature labels, defect feature labels and hidden danger feature labels, and are associated with their time and space information to generate annotated transmission line data labels; According to the generated annotated transmission line data labels, environmental feature labels, defect feature labels and hidden danger feature labels are organized into structured transmission line data assets. The transmission line data assets include time dimension, space dimension and feature dimension, which are used for transmission line operation status assessment and hidden danger prediction.

6. The power transmission line patrol data asset management method according to claim 1, characterized in that: The step of evaluating the possibility of environmental conditions triggering transmission line defects according to the association between the environmental feature tag and the defect feature tag includes: Obtain environmental feature labels, including key environmental parameters such as wind speed, temperature and humidity, and precipitation, as well as their corresponding spatiotemporal distribution; Obtain defect feature labels, including crack length, corrosion area, and infrared temperature abnormality hotspot distribution on the surface of the transmission line; According to the following formula (1), the possibility of crack extension under different environmental conditions is calculated: ； in, is the crack triggering factor, reflecting the possibility of crack extension under different environmental conditions; is the crack length; is the wind speed; , , is a sensitivity parameter obtained through historical data and model training; The possibility of corrosion development is calculated according to the following formula (2): ； in, It is a corrosion trigger factor, reflecting the possibility of corrosion development; is the corrosion area; for humidity; is the precipitation; , , is a sensitivity parameter obtained through historical data and model training; The possibility of high temperature aggravating crack growth is calculated according to the following formula (3): ； in, It is the temperature rise trigger factor, reflecting the possibility of high temperature aggravating crack growth; is the temperature value based on the infrared temperature hotspot; is the crack length; , is the influence coefficient, obtained through historical data and model training; According to the following formula (4), calculate the comprehensive defect trigger probability value : ； in, , , is the weight coefficient, obtained through historical data and model training; if If it is greater than or equal to the specified threshold, a defect trigger warning is generated and the risk type is recorded; if If it is less than the specified threshold, the low impact of the current environmental conditions on the defect trigger is recorded.

7. The power transmission line patrol data asset management method according to claim 1, characterized in that: The method of constructing a graph network model with transmission lines as nodes and sensor data as edge weights according to the evaluation results of the hidden danger feature labels and environmental conditions includes: The transmission line is regarded as a node of the graph network, and the attributes of the node include a hidden danger feature label, an environmental feature label, and a historical inspection data feature; According to the following formula 5, the physical proximity weight between nodes is calculated: ； in, Representation Node and nodes The physical proximity weight between Representation Node and nodes The physical distance between Representation Node and nodes The electrical transmission distance between is the distance sensitivity coefficient; According to the following formula (6), the environmental similarity weight between nodes is calculated: ； in, Representation Node and nodes The weight of the environmental similarity between them; To represent the node and nodes The Euclidean distance between the nodes is calculated based on the similarity of the environmental characteristics between the nodes; According to the following formula (7), the hidden danger correlation weight between nodes is calculated: ； in, Representation Node and nodes The hidden danger correlation weight between them; For Node and nodes The angle between the hidden danger feature vectors of two nodes; According to the following formula (8), the edge weights between nodes are calculated: ； in, For Node and nodes The edge weights between ; is the weighting coefficient, determined according to historical data and satisfies ; Construct a weighted graph based on the nodes and edge weights. , in represents the set of transmission line nodes, Represents the edge set; initializes the attribute vector of each node, combines the node characteristics with the weight distribution of its adjacent nodes, and generates the initial hidden danger propagation state.

8. The power transmission line patrol data asset management method according to claim 7, characterized in that: The method of predicting the time of occurrence of hidden dangers by using the graph network model based on historical patrol data and environmental dynamic parameters includes: The hidden danger propagation simulation is carried out on the graph network, and the hidden danger propagation probability of the node is calculated based on the graph convolutional network, including: According to the following formula (9), for each node, the hidden danger propagation state of the node is updated by combining the attribute vector and edge weight of its adjacent nodes: ； in, Indicates Layer propagation node Status; Indicates Layer propagation node Status; For Node The set of neighbor nodes of is the weight matrix obtained through training; is the activation function; Repeat the propagation iteration until the node status converges; According to the following formula 10, calculate the node Time of occurrence of potential hazards : ； in, is the time when the initial hidden danger occurred, is the number of propagation iterations; Indicates Layer propagation node status.

9. The power transmission line patrol data asset management method according to claim 1, characterized in that: The prediction results based on the graph network model generate a defect risk level warning, and generate a transmission line hidden danger distribution map in combination with environmental meteorological data, including: According to the hidden danger propagation status value of each node in the graph network model, combined with environmental meteorological data and historical inspection records, the defect risk score of each node is calculated, and the nodes are divided into low risk, medium risk and high risk levels according to the preset level standards; Based on the geographical location and risk level of the node, the risk information of the transmission line and its ancillary facilities is mapped to the geographic space to generate a transmission line hidden danger distribution map, wherein the hidden danger distribution map is used to show the distribution range and risk intensity of high-risk areas; According to the hidden danger distribution map, the key nodes and impact range of the high-risk area are identified, defect risk warning information is generated, and corresponding inspection and maintenance suggestions are provided.

10. A transmission line patrol data asset management system, characterized in that: include: An acquisition unit is used to acquire multimodal data related to the inspection of the transmission line through various sensors installed on the transmission line and the surrounding environment, sensors carried by helicopters and drones, mobile terminals carried by inspection personnel, and data acquisition devices of environmental monitoring satellites, wherein the multimodal data includes physical state data of the transmission line, electrical parameter data, environmental meteorological data, appearance photos and video data of the transmission line, temperature distribution data captured by infrared cameras, defect data of the transmission line, and hidden danger data of the transmission channel; An acquisition unit, used for synchronously calibrating and dynamically fusing the multimodal data based on a spatiotemporal correlation model to obtain a fused data set with temporal consistency and spatial accuracy; A generating unit, configured to parse the fused data set using an adaptive feature extraction algorithm to generate a transmission line data asset with structured features, wherein the transmission line data asset includes an environmental feature label, a defect feature label, and a hidden danger feature label, wherein the environmental feature label describes the potential impact of weather, temperature and humidity, and wind speed on the operation of the transmission line; An evaluation unit, configured to evaluate the possibility of environmental conditions triggering transmission line defects based on the association between the environmental feature tags and the defect feature tags; A construction unit, configured to construct a graph network model with transmission lines as nodes and sensor data as edge weights according to the hidden danger feature labels and the evaluation results of the environmental conditions, wherein the graph network model is used to analyze the defect propagation path of the transmission line and its channel, and to predict the time and spatial distribution of hidden danger occurrence based on historical inspection data and environmental dynamic parameters; The combining unit is used to generate a defect risk level warning based on the prediction results of the graph network model, and to generate a transmission line hidden danger distribution map in combination with environmental meteorological data, wherein the hidden danger distribution map is used to assist in formulating inspection and maintenance plans.

Images