Power transmission line inspection abnormity early warning method based on multi-mode perception and digital twinning

Abstract

The invention discloses a power transmission line inspection abnormity early warning method based on multi-mode perception and digital twinning, and belongs to the technical field of power transmission line monitoring. The problems that in the prior art, inspection efficiency is low, multi-source data utilization is insufficient, early warning lags behind and closed-loop control is lacked are solved, and efficient on-site processing and feature fusion of data are achieved by deploying a multi-modal sensing terminal cluster and a layered edge computing node; a digital twinborn model fusing multi-physical field simulation and data-driven AI is constructed at a cloud end, so that virtual-real interaction and deep analysis of a physical mechanism and real data are realized; the health state of the power transmission line is evaluated based on the deep analysis result, multi-stage early warning can be generated accordingly, a reverse control instruction is triggered, closed-loop control from state perception and intelligent diagnosis to active intervention is finally formed, and the intelligent level and operation and maintenance safety efficiency of power transmission line inspection are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of power transmission line monitoring technology, specifically to a method for early warning of abnormalities in power transmission line inspections based on multimodal sensing and digital twins. Background Technology

[0002] With the continuous expansion of the power system and the ongoing improvement of voltage levels, transmission lines, as key channels for power transmission, are crucial for ensuring the reliability of the power grid and for socio-economic development. However, transmission lines typically span vast geographical areas and are exposed to complex and ever-changing natural environments for extended periods. They are highly susceptible to multiple factors, including meteorological conditions, mechanical loads, electrical stresses, and external environmental conditions, leading to a significant increase in risks such as line aging, component damage, and even outages.

[0003] Traditional transmission line inspections mainly rely on manual patrols or simple automated equipment, which suffers from low efficiency, high risk, and poor real-time performance. It is difficult to achieve accurate all-weather, all-round perception of the line status. Some inspection methods have begun to use single sensors for data collection, but single-modal data cannot fully reflect the comprehensive status of the line in complex environments. Therefore, they do not meet the current needs. To address this, we propose a transmission line inspection anomaly early warning method based on multimodal perception and digital twins. Summary of the Invention

[0004] The purpose of this invention is to provide a method for early warning of anomalies in power transmission line inspection based on multimodal perception and digital twins. By constructing a digital twin model in the cloud that integrates multi-physics field coupled simulation and data-driven AI analysis, deploying a cluster of multimodal intelligent sensing terminals on the power transmission line side for comprehensive data collection, deploying hierarchical computing nodes on the edge side to achieve real-time data preprocessing and feature fusion, and using deep spatiotemporal fusion analysis to identify the deviation between physical entities and virtual models, the invention finally generates multi-level early warnings based on the evaluation results and triggers reverse control commands, thus solving the problems mentioned in the background technology.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for early warning of anomalies in transmission line inspection based on multimodal sensing and digital twins, the method comprising:

[0006] A digital twin model is constructed in the cloud, which includes a multiphysics coupled simulation model and a data-driven AI analysis model;

[0007] Deploy multimodal intelligent sensing terminal clusters in the towers, conductors, and corridors of transmission lines to collect multimodal data of the transmission lines;

[0008] Edge computing nodes with hierarchical computing capabilities are deployed at the edge to preprocess, extract features, and perform preliminary fusion on the collected raw multimodal data to generate a set of structured feature vectors.

[0009] The generated set of structured feature vectors is transmitted to the digital twin model in the cloud via a communication network. The digital twin model then performs in-depth analysis using a multiphysics coupled simulation model and a data-driven AI analysis model.

[0010] The multiphysics coupling simulation model simulates the theoretical operating state of the line under the current environment and load. The data-driven AI analysis model performs deep spatiotemporal fusion analysis based on the structured feature vector set and the output of the multiphysics coupling simulation model to identify deviations from the actual state of the physical entity.

[0011] Based on the results of in-depth analysis, the health status of the transmission line is assessed and multi-level early warnings are generated, namely Level 1, Level 2 and Level 3 early warnings, and the reverse control command of the digital twin model is triggered according to the early warning level.

[0012] Based on the warning level, the warning is visually labeled on the visualization interface, and a warning report containing the abnormal location, type, risk level and handling suggestions is automatically generated and pushed to the operation and maintenance personnel.

[0013] Furthermore, the multimodal intelligent sensing terminal cluster includes an optical image sensor, an infrared thermal imager, a lidar, an acoustic fingerprint sensor, a micro-meteorological sensor, and a conductor status sensor. Based on the micro-meteorological sensor data and task scheduling instructions, the sampling frequency of each node in the multimodal intelligent sensing terminal cluster is dynamically adjusted.

[0014] Furthermore, based on the micro-meteorological sensor data and task scheduling instructions, the sampling frequency of each node in the multimodal intelligent sensing terminal cluster is dynamically adjusted, including:

[0015] Edge computing nodes continuously receive real-time data uploaded by micro-weather sensors;

[0016] Edge computing nodes compare the received micro-meteorological data with preset meteorological condition thresholds to determine the degree of impact of current meteorological conditions on the operation of transmission lines.

[0017] Based on the comparison results between micro-meteorological data and thresholds, the edge computing nodes generate corresponding task scheduling instructions, which include the requirements for adjusting the sampling frequency of each node in the multimodal intelligent sensing terminal cluster.

[0018] Edge computing nodes send task scheduling instructions to each node in the multimodal intelligent sensing terminal cluster through the communication network. After receiving the instructions, each node adjusts its own sampling frequency according to the instructions.

[0019] After adjusting the sampling frequency, each node continues to collect the corresponding multimodal data at the new frequency and transmits it back to the edge computing node for further processing.

[0020] Furthermore, the theoretical operating state of the line under the current environment and load is simulated using the multiphysics coupling simulation model, including:

[0021] Obtain a preprocessed and feature-extracted set of structured feature vectors from edge computing nodes;

[0022] The acquired structured feature vector set is input into a multiphysics coupling simulation model, which comprehensively considers the interaction of electric field, magnetic field, thermal field, mechanical field and fluid field;

[0023] The finite element method was used to solve the multiphysics coupling simulation model. The transmission line was divided into multiple finite element elements, and the response of each element in the electric field, magnetic field, thermal field, mechanical field and fluid field was calculated.

[0024] The coupling relationship between the fields is solved by iterative algorithm, and finally the theoretical operating state of the entire transmission line under the current environment and load is obtained.

[0025] The output of the multiphysics coupling simulation model is interfaced with the input of the data-driven AI analysis model. The output includes parameters such as the temperature, stress, deformation, electric field strength, and magnetic field strength of the conductor.

[0026] Furthermore, the data-driven AI analysis model performs deep spatiotemporal fusion analysis, including:

[0027] Construct a spatiotemporal graph neural network model, modeling each sensor or observation point as a graph node, and modeling physical connections and spatial correlations as edges;

[0028] The output of the structured feature vector set coupled with the multiphysics simulation model is used as the dynamic attribute of the node;

[0029] By using the message passing mechanism of graph neural networks, we can capture the spatial propagation and temporal evolution of abnormal states, locate the root cause of the anomaly, and predict its development trend.

[0030] Furthermore, based on the results of in-depth analysis, the health status of transmission lines is assessed and multi-level early warnings are generated, including:

[0031] After the digital twin model completes the in-depth analysis of the structured feature vector set using a multiphysics coupled simulation model and a data-driven AI analysis model, the in-depth analysis results are compared with the preset health status assessment standards.

[0032] Based on the degree of deviation between the various indicators in the in-depth analysis results and the evaluation standards, the overall health status of the transmission line is quantitatively evaluated. The evaluation results are presented in the form of Health Status Index (HSI). The higher the HSI value, the better the health status of the line, and vice versa.

[0033] Based on the HSI value and the abnormality of various indicators, the warning level is determined. The warning levels are divided into Level 1, Level 2, and Level 3, with each level corresponding to a different degree of risk and handling priority.

[0034] Furthermore, based on the warning level, the digital twin model triggers reverse control commands, including:

[0035] When the warning level reaches Level II or above, the corresponding reverse control command is selected from the preset control strategy library according to the specific type and severity of the warning.

[0036] The selected reverse control command is sent to the field execution equipment of the connected power transmission line through the cloud control interface;

[0037] The on-site equipment executes the corresponding operations according to the reverse control commands.

[0038] Furthermore, after completing the operation, the on-site execution equipment feeds the operation results back to the cloud, specifically as follows:

[0039] After receiving operation feedback information, the cloud will update the relevant parameters and status in the digital twin model to reflect the actual operation changes of the transmission line.

[0040] Meanwhile, the cloud continues to monitor and analyze the updated digital twin model in real time to evaluate the effectiveness of the reverse control commands;

[0041] If the abnormal situation is alleviated, the warning level will be reduced or lifted, and normal monitoring of the transmission lines will continue.

[0042] If the abnormal situation is not improved or a new abnormality occurs, the warning level will be reassessed based on the new in-depth analysis results, and further reverse control commands will be triggered until the health status of the transmission line returns to normal.

[0043] Furthermore, edge computing nodes with hierarchical computing capabilities are deployed at the edge, including:

[0044] The first-layer computing nodes are responsible for preprocessing the raw multimodal data, including data cleaning, noise reduction and format standardization, and time synchronization of the multimodal data;

[0045] The second-layer computing nodes are responsible for feature extraction from the preprocessed data and generating preliminary feature vectors.

[0046] The third-layer computing node is responsible for the initial fusion of feature vectors. Based on the preset spatiotemporal alignment rules, it splices and associates different modal features from the same time period, the same device, or the same region to form a structured feature vector set.

[0047] Furthermore, each vector in the structured feature vector set uniquely corresponds to the comprehensive state of the monitored object at the current monitoring time, and is tagged with a device ID, timestamp, and location tag. Finally, the structured feature vector set is uploaded to the digital twin model in the cloud through a communication network.

[0048] Compared with the prior art, the beneficial effects of the present invention are:

[0049] This invention constructs a digital twin in the cloud that integrates a multi-physics field coupled simulation model and a data-driven AI analysis model. By utilizing deep spatiotemporal fusion analysis technology, it achieves accurate mapping and simulation of the state of transmission lines, thereby overcoming the limitations of single-model analysis and significantly improving the accuracy and foresight of anomaly identification and state assessment. Furthermore, it assesses the health status of transmission lines based on the deep analysis results, thereby determining the early warning level and triggering reverse control commands. This solves the problems of traditional early warning methods being singular and passive in handling, and significantly improves the automation level and safety assurance capabilities of transmission line operation and maintenance. Attached Figure Description

[0050] Figure 1 The flowchart shows the transmission line inspection anomaly early warning method based on multimodal sensing and digital twin of the present invention.

[0051] Figure 2 This is an execution diagram of the transmission line inspection anomaly early warning method based on multimodal perception and digital twin of the present invention. Detailed Implementation

[0052] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0053] To address the issues of low inspection efficiency, insufficient utilization of multi-source data, delayed early warning, and lack of closed-loop control in existing technologies, please refer to [link / reference needed]. Figures 1-2 This embodiment provides the following technical solution:

[0054] A method for early warning of anomalies in power transmission line inspections based on multimodal sensing and digital twins, the method comprising:

[0055] A digital twin model is constructed in the cloud, which includes a multiphysics coupled simulation model and a data-driven AI analysis model;

[0056] Deploy multimodal intelligent sensing terminal clusters in the towers, conductors, and corridors of transmission lines to collect multimodal data of the transmission lines;

[0057] Edge computing nodes with hierarchical computing capabilities are deployed at the edge to preprocess, extract features, and perform preliminary fusion on the collected raw multimodal data to generate a set of structured feature vectors.

[0058] The generated set of structured feature vectors is transmitted to the digital twin model in the cloud via a communication network. The digital twin model then performs in-depth analysis using a multiphysics coupled simulation model and a data-driven AI analysis model.

[0059] The multiphysics coupling simulation model simulates the theoretical operating state of the line under the current environment and load. The data-driven AI analysis model performs deep spatiotemporal fusion analysis based on the structured feature vector set and the output of the multiphysics coupling simulation model to identify deviations from the actual state of the physical entity.

[0060] Based on the results of in-depth analysis, the health status of the transmission line is assessed and multi-level early warnings are generated, namely Level 1, Level 2 and Level 3 early warnings, and the reverse control command of the digital twin model is triggered according to the early warning level.

[0061] Based on the warning level, the warning is visually labeled on the visualization interface, and a warning report containing the abnormal location, type, risk level and handling suggestions is automatically generated and pushed to the operation and maintenance personnel.

[0062] Specifically, a Level 1 warning is indicated by a flashing red icon at the corresponding location on the transmission line, a Level 2 warning by an orange icon, and a Level 3 warning by a yellow icon.

[0063] The technical effects of the above solution are as follows: By constructing an architecture that coordinates a cloud-based digital twin model with layered computing nodes on the edge, efficient processing and in-depth analysis of multimodal sensing data of transmission lines are achieved. The digital twin model, through the fusion of multi-physics field coupled simulation and data-driven AI model, can accurately identify the deviation between the line's operating status and the theoretical model, thereby enabling early warning and precise location of abnormal situations. Based on the results of in-depth analysis, multi-level warnings can be generated and reverse control commands can be triggered, forming a closed-loop management system from state perception and intelligent diagnosis to active control. This significantly improves the intelligence level and operation and maintenance efficiency of transmission line inspection, effectively reduces the risk of faults caused by line anomalies, and ensures the safe and stable operation of the power grid.

[0064] The multimodal intelligent sensing terminal cluster includes optical image sensors, infrared thermal imagers, lidar, acoustic fingerprint sensors, micro-meteorological sensors, and wire status sensors.

[0065] Based on the micro-meteorological sensor data and task scheduling instructions, the sampling frequency of each node in the multimodal intelligent sensing terminal cluster is dynamically adjusted, specifically as follows:

[0066] Edge computing nodes continuously receive real-time data (such as wind speed, precipitation, humidity, and temperature) uploaded by micro-meteorological sensors.

[0067] Edge computing nodes compare the received micro-meteorological data with preset meteorological condition thresholds to determine the impact of current meteorological conditions on the operation of transmission lines. These meteorological condition thresholds are set in advance based on the operating characteristics of the transmission lines, historical meteorological data, and expert experience. For example, when the wind speed exceeds a certain threshold, it may increase the risk of conductor galloping; when the temperature is below freezing and the humidity is high, it may cause the conductors to become icy.

[0068] Based on the comparison results between micro-meteorological data and thresholds, the edge computing nodes generate corresponding task scheduling instructions. These instructions include requirements for adjusting the sampling frequency of each node in the multimodal intelligent sensing terminal cluster. For example, if a significant increase in wind speed is detected, the task scheduling instructions may require the optical image sensor to increase its sampling frequency to capture conductor swaying more promptly. Simultaneously, the sampling frequency of the lidar may also be increased accordingly to more accurately measure the displacement and deformation of the conductor. Conversely, if the current weather conditions are stable and the transmission line is operating well, the task scheduling instructions may appropriately reduce the sampling frequency of some sensors to save energy and reduce data transmission volume.

[0069] Edge computing nodes send task scheduling instructions to each node in the multimodal intelligent sensing terminal cluster via the communication network. After receiving the instructions, each node adjusts its own sampling frequency according to the instructions. For example, an optical image sensor may adjust from collecting images once per minute to collecting images once every 10 seconds; a voiceprint sensor may adjust from collecting data once every 5 seconds to collecting data once per second.

[0070] After adjusting the sampling frequency, each node continues to collect the corresponding multimodal data at the new frequency and transmits it back to the edge computing node for further processing.

[0071] The technical effects of the above solution are as follows: based on the intelligent comparison of real-time micro-meteorological data and preset thresholds, the sampling frequency of various sensors is adaptively adjusted, thereby improving the monitoring accuracy and timeliness of risks such as power line galloping and icing under severe weather conditions. At the same time, when the weather is stable, the sampling rate of non-critical sensors is automatically reduced, effectively saving energy consumption and data transmission load, and realizing the synergistic improvement of dynamic optimization of sensing resources and operation and maintenance efficiency.

[0072] The multiphysics coupling simulation model simulates the theoretical operating state of the line under the current environment and load, including:

[0073] Obtain a preprocessed and feature-extracted set of structured feature vectors from edge computing nodes;

[0074] The acquired structured feature vector set is input into a multiphysics coupling simulation model, which comprehensively considers the interactions of electric, magnetic, thermal, mechanical, and fluid fields. Specifically:

[0075] Regarding the electric field, the electric field distribution around the conductors is calculated based on the voltage level and conductor layout of the transmission line, as well as the resulting corona loss and electromagnetic interference.

[0076] Regarding the magnetic field, the magnetic field strength around the conductor and its electromagnetic influence on the surrounding environment are calculated based on the magnitude and distribution of the current in the line.

[0077] Regarding the thermal field, the temperature distribution of the conductor is calculated by considering factors such as the current heating effect of the conductor, solar radiation, and ambient temperature.

[0078] In terms of the mechanical field, the stress and deformation of the conductor are calculated based on factors such as conductor tension, wind load, and icing load.

[0079] In terms of fluid fields, meteorological conditions such as wind speed and wind direction are considered to calculate the airflow field around the conductor and the aerodynamic effects on the conductor.

[0080] The finite element method was used to solve the multiphysics coupling simulation model. The transmission line was divided into multiple finite element elements, and the response of each element in the electric field, magnetic field, thermal field, mechanical field and fluid field was calculated.

[0081] The coupling relationship between the fields is solved by iterative algorithm, and finally the theoretical operating state of the entire transmission line under the current environment and load is obtained.

[0082] The output of the multiphysics coupling simulation model is interfaced with the input of the data-driven AI analysis model. The output includes parameters such as the temperature, stress, deformation, electric field strength, and magnetic field strength of the conductor.

[0083] The technical effects of the above-mentioned technical solution are as follows: by integrating multi-physics field coupling simulation of electric field, magnetic field, thermal field, mechanical field and fluid field, it is possible to simulate the theoretical operating state of transmission lines under complex working conditions with high precision, thereby comprehensively revealing key parameters such as potential corona loss, electromagnetic interference, thermal stress distribution and structural deformation of transmission lines, and providing comprehensive and reliable physical mechanism input for data-driven AI analysis models, effectively improving the accuracy and scientific nature of status assessment and risk warning.

[0084] Data-driven AI analysis models perform deep spatiotemporal fusion analysis, including:

[0085] Construct a spatiotemporal graph neural network model, modeling each sensor or observation point as a graph node, and modeling physical connections and spatial correlations as edges;

[0086] The output of the structured feature vector set coupled with the multiphysics simulation model is used as the dynamic attribute of the node;

[0087] By using the message passing mechanism of graph neural networks, we can capture the spatial propagation and temporal evolution of abnormal states, locate the root cause of the anomaly, and predict its development trend.

[0088] The technical effects of the above-mentioned technical solution are as follows: The data-driven AI analysis model, by constructing a spatiotemporal graph neural network, deeply integrates physical space and dynamic attributes, thereby accurately capturing the spatiotemporal propagation patterns and evolution trends of abnormal states in sensor networks. This enables precise localization of the root causes of power transmission line anomalies and effective prediction of their development trends, significantly improving the insight of state perception and the foresight of operation and maintenance decisions.

[0089] Based on the results of in-depth analysis, the health status of transmission lines is assessed and multi-level early warnings are generated, including:

[0090] After the digital twin model completes the in-depth analysis of the structured feature vector set using a multiphysics coupled simulation model and a data-driven AI analysis model, the in-depth analysis results are compared with the preset health status assessment standards. These standards are formulated based on the operating parameters of the transmission line, historical data, and industry specifications, covering multiple dimensions such as the line's electrical performance, mechanical performance, and environmental adaptability. For example, parameters such as the insulation resistance of the line, the temperature of the conductor, and the tilt angle of the tower all have corresponding normal ranges and warning thresholds.

[0091] Based on the degree of deviation between the various indicators in the in-depth analysis results and the evaluation standards, the overall health status of the transmission line is quantitatively evaluated. The evaluation results are presented in the form of Health Status Index (HSI). The higher the HSI value, the better the health status of the line, and vice versa.

[0092] Based on the HSI value and the abnormality of various indicators, the warning level is determined. The warning levels are divided into Level 1 warning (severe abnormality), Level 2 warning (relatively severe abnormality), and Level 3 warning (general abnormality). Each level corresponds to a different degree of risk and handling priority.

[0093] The technical effects of the above solution are as follows: by comparing the results of in-depth analysis with the preset health status assessment standards, the overall health status index (HSI) of the transmission line can be quantitatively presented, thereby providing operation and maintenance personnel with an intuitive reference for the health status. In addition, the multi-level early warning mechanism determined based on the HSI value and the abnormal conditions of various indicators can clearly distinguish abnormal conditions with different risk levels and handling priorities, enabling operation and maintenance personnel to quickly and accurately identify problems and take corresponding measures, effectively improving the operation and maintenance efficiency and reliability of the transmission line, reducing the risk of fault occurrence, and ensuring the stable operation of the power system.

[0094] Based on the warning level, the digital twin model triggers reverse control commands, including:

[0095] When the warning level reaches Level II or above, the corresponding reverse control command is selected from the preset control strategy library according to the specific type and severity of the warning. The control strategy library is pre-formulated based on the operating characteristics, safety specifications and historical experience of the transmission line. It includes various possible abnormal situations and their corresponding control measures. For example, if the warning is caused by the conductor temperature being too high, the reverse control command may be to reduce the load current of that section of the line.

[0096] The selected reverse control command is sent to the field execution equipment of the connected transmission line, such as circuit breaker, voltage regulator, tension controller, etc., through the cloud control interface.

[0097] The on-site execution equipment performs corresponding operations according to the reverse control command, such as adjusting the opening and closing state of the circuit breaker to change the load current, adjusting the output of the voltage regulator to adjust the line voltage, or controlling the tension controller to change the line tension, etc.

[0098] After completing the operation, the on-site execution equipment will send the operation results back to the cloud, specifically:

[0099] After receiving operation feedback information, the cloud will update the relevant parameters and status in the digital twin model to reflect the actual operation changes of the transmission line.

[0100] Meanwhile, the cloud continues to monitor and analyze the updated digital twin model in real time to evaluate the effectiveness of the reverse control commands;

[0101] If the abnormal situation is alleviated, the warning level will be reduced or lifted, and normal monitoring of the transmission lines will continue.

[0102] If the abnormal situation is not improved or a new abnormality occurs, the warning level will be reassessed based on the new in-depth analysis results, and further reverse control commands will be triggered until the health status of the transmission line returns to normal.

[0103] The technical effects of the above solution are as follows: by triggering the reverse control command of the digital twin model through the early warning level, real-time response and automatic control of abnormal conditions of transmission lines are realized, thereby achieving rapid closed-loop intervention and dynamic optimization of abnormal conditions of transmission lines. At the same time, by leveraging operational feedback and model updates, a continuous effect evaluation and strategy adjustment mechanism is formed, thereby significantly improving the system's autonomy and fault recovery efficiency. This closed-loop feedback control mechanism effectively improves the operational safety and reliability of transmission lines, reduces operation and maintenance costs, and ensures the stable operation of the power system.

[0104] Deploying edge computing nodes with hierarchical computing capabilities at the edge includes:

[0105] The first-layer computing nodes are responsible for preprocessing the raw multimodal data, including data cleaning, noise reduction and format standardization, and for synchronizing the multimodal data in time to ensure that all data has a unified timestamp and data source identifier.

[0106] The second-layer computing nodes are responsible for feature extraction from the preprocessed data and generating preliminary feature vectors.

[0107] The third-layer computing node is responsible for the initial fusion of feature vectors. Based on the preset spatiotemporal alignment rules, it splices and associates different modal features (such as image features, point cloud features, voiceprint features, and meteorological parameters) from the same time period, the same device, or the same region to form a unified, high-dimensional set of structured feature vectors.

[0108] The spatiotemporal alignment rules are preset based on the structured digital model of the monitored physical entity (such as towers, insulators, and conductor segments) and the analysis objectives of the inspection task. The spatiotemporal alignment rules include alignment methods such as global timestamps and data interpolation. The alignment methods used in the spatiotemporal alignment rules are existing technologies in this field and are not inventive solutions of this application, so they will not be elaborated here.

[0109] Each vector in the structured feature vector set uniquely corresponds to the comprehensive status of a monitored object (such as a tower, insulator string, conductor segment, etc.) at the current monitoring time, and is tagged with a device ID, timestamp, and location tag. Finally, the structured feature vector set is uploaded to the digital twin model in the cloud through the communication network.

[0110] The technical effects of the above solution are as follows: the hierarchical processing and fusion of multimodal data are realized through the edge-side hierarchical computing architecture, thereby efficiently transforming the original multimodal data into structured feature vectors with a unified spatiotemporal benchmark, significantly improving data quality and cloud analysis efficiency, thus providing accurate and associative object state representations for digital twin models, and effectively reducing cloud computing and communication load.

[0111] Working principle: Through a cluster of multimodal intelligent sensing terminals deployed on-site, multimodal data of the transmission line is continuously collected. At the edge, edge computing nodes sequentially perform real-time preprocessing, feature extraction, and multimodal feature fusion on the collected multimodal data to form a structured feature vector set with a unified spatiotemporal label, thereby providing high-quality input for cloud analysis. The digital twin model in the cloud, on the one hand, accurately reproduces and simulates the theoretical operating state of the line under the coupling of real electric, magnetic, thermal, force, and fluid fields in virtual space through multiphysics field coupling simulation. On the other hand, by constructing a data-driven AI model of spatiotemporal graph neural network, the simulation results are deeply integrated with the measured features to deeply explore the spatiotemporal propagation law and evolution trend of abnormal states. Based on this in-depth analysis results, the health status of the line is quantitatively assessed and multi-level early warnings are generated. When the early warning level reaches level two or above, reverse control commands are automatically triggered, thereby realizing rapid closed-loop intervention and dynamic optimization of abnormal states of the transmission line.

[0112] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0113] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention.

Claims

1. A power transmission line inspection anomaly early warning method based on multi-modal perception and digital twinning, characterized in that, The method comprises the following steps: Building a digital twin model in the cloud, which includes a multi-physics field coupling simulation model and a data-driven AI analysis model; Deploying a multi-modal intelligent sensing terminal cluster in the tower, ground wire and channel environment of the transmission line to collect multi-modal data of the transmission line; Deploying an edge computing node with hierarchical computing capability on the edge side to preprocess, extract features and preliminarily fuse the collected raw multi-modal data, and generate a set of structured feature vectors; Transmitting the generated set of structured feature vectors to the digital twin model in the cloud through a communication network, and using the multi-physics field coupling simulation model and the data-driven AI analysis model for deep analysis; Simulating the theoretical running state of the line under the current environment and load through the multi-physics field coupling simulation model, and performing deep spatio-temporal fusion analysis based on the structured feature vector set and the output of the multi-physics field coupling simulation model to identify the deviation from the real state of the physical entity; Based on the deep analysis result, the health status of the transmission line is evaluated and multi-level early warning is generated, including first-level early warning, second-level early warning and third-level early warning, and the reverse control instruction of the digital twin model is triggered according to the early warning level; According to the early warning level, the hierarchical early warning visualization is marked on the visualization interface, and an early warning report containing the abnormal position, type, risk level and disposal suggestion is automatically generated and pushed to the operation and maintenance personnel.

2. The power transmission line inspection anomaly early warning method based on multi-modal perception and digital twinning of claim 1, characterized in that, The multi-modal intelligent sensing terminal cluster includes an optical image sensor, an infrared thermal imager, a laser radar, a voiceprint sensor, a microclimate sensor and a ground wire state sensor. Based on the microclimate sensor data and task scheduling instructions, the sampling frequency of each node in the multi-modal intelligent sensing terminal cluster is dynamically adjusted.

3. The power transmission line inspection anomaly early warning method based on multi-modal perception and digital twinning of claim 2, characterized in that, Based on the microclimate sensor data and task scheduling instructions, the sampling frequency of each node in the multi-modal intelligent sensing terminal cluster is dynamically adjusted, which comprises: The edge computing node continuously receives real-time data uploaded by the microclimate sensor; The edge computing node compares the received microclimate data with the preset threshold of meteorological conditions to determine the influence of the current meteorological conditions on the running state of the transmission line; According to the comparison result of the microclimate data and the threshold, the edge computing node generates corresponding task scheduling instructions, which contain the adjustment requirements for the sampling frequency of each node in the multi-modal intelligent sensing terminal cluster; The edge computing node sends the task scheduling instructions to each node in the multi-modal intelligent sensing terminal cluster through a communication network, and each node adjusts its sampling frequency according to the instruction requirements after receiving the instruction; After adjusting the sampling frequency, each node continues to collect corresponding multi-modal data at the new frequency and transmits it back to the edge computing node for subsequent processing.

4. The power transmission line inspection anomaly early warning method based on multi-modal perception and digital twinning of claim 1, characterized in that, The method for simulating the theoretical running state of the line under the current environment and load through the multi-physics field coupling simulation model comprises: Obtaining the set of structured feature vectors after preprocessing and feature extraction from the edge computing node; The obtained structured feature vector set is input into a multi-physical field coupling simulation model which comprehensively considers the interaction of electric field, magnetic field, thermal field, mechanical field and fluid field; The multi-physical field coupling simulation model is solved by using a finite element analysis method, the power transmission line is divided into a plurality of finite element units, and the response in the electric field, magnetic field, thermal field, mechanical field and fluid field is calculated for each unit; And the coupling relationship between each field is solved by an iterative algorithm, and finally the theoretical operating state of the entire power transmission line under the current environment and load is obtained; The output results of the multi-physical field coupling simulation model are connected with the input of the data-driven AI analysis model, and the output results include the temperature, stress, deformation, electric field strength and magnetic field strength parameters of the conductor.

5. The power transmission line inspection anomaly early warning method based on multi-modal perception and digital twinning of claim 1, characterized in that, The data-driven AI analysis model performs deep spatio-temporal fusion analysis, including: A spatio-temporal graph neural network model is constructed, each sensor or observation point is modeled as a graph node, and physical connection and spatial correlation are modeled as edges; The structured feature vector set and the output of the multi-physical field coupling simulation model are used as dynamic attributes of the nodes; Through the message passing mechanism of the graph neural network, the propagation of abnormal state in space and the evolution law in time are captured, the abnormal source is located and the development trend is predicted.

6. The power transmission line inspection anomaly early warning method based on multi-modal perception and digital twinning of claim 1, wherein, Based on the deep analysis results, the health status of the power transmission line is evaluated and multi-level warnings are generated, including: After the digital twin model completes the deep analysis of the structured feature vector set by using the multi-physical field coupling simulation model and the data-driven AI analysis model, the deep analysis results are compared with the preset health status evaluation standard; According to the deviation degree of each index in the deep analysis results and the evaluation standard, the overall health status of the power transmission line is quantitatively evaluated, and the evaluation result is presented in the form of health status index HSI, the higher the HSI value, the better the line health status, and vice versa; According to the HSI value and the abnormal situation of each index, the warning level is determined, and the warning level is divided into first-level warning, second-level warning and third-level warning, each level corresponds to different risk degree and processing priority.

7. The power transmission line inspection anomaly early warning method based on multi-modal perception and digital twinning of claim 1, wherein, Trigger the reverse control instruction of the digital twin model according to the warning level, including: When the warning level reaches the second-level warning or above, according to the specific type and severity of the warning, select the corresponding reverse control instruction from the preset control strategy library; The selected reverse control instruction is sent to the on-site execution equipment connected to the power transmission line through the control interface of the cloud; The on-site execution equipment performs the corresponding operation according to the reverse control instruction.

8. The power transmission line inspection anomaly early warning method based on multi-modal perception and digital twinning of claim 7, characterized in that, After the on-site execution equipment completes the operation, the operation result is fed back to the cloud, specifically: After receiving the operation feedback information, the cloud updates the related parameters and states in the digital twin model to reflect the actual operating changes of the power transmission line; At the same time, the cloud continues to monitor and analyze the updated digital twin model to evaluate the effect of the reverse control instruction; If the abnormal situation is alleviated, the warning level is reduced or removed, and the normal monitoring of the power transmission line is continued. If the abnormal situation does not improve or a new abnormality occurs, the early warning level is re-evaluated according to the new deep analysis result, and further reverse control instructions are triggered until the health status of the power transmission line returns to normal.

9. The power transmission line inspection anomaly early warning method based on multi-modal perception and digital twinning of claim 1, characterized in that, An edge computing node with layered computing capability is deployed on the edge, including: A first layer computing node is responsible for preprocessing the original multi-modal data, including data cleaning, denoising, and format standardization, and time synchronization of the multi-modal data; A second layer computing node is responsible for feature extraction on the preprocessed data to generate a preliminary feature vector; A third layer computing node is responsible for preliminary fusion of the feature vector, and according to a preset time-space alignment rule, different modal features of the same time period, same device or region are spliced and associated to form a structured feature vector set.

10. The power transmission line inspection anomaly early warning method based on multi-modal perception and digital twinning of claim 9, wherein, Each vector in the structured feature vector set uniquely corresponds to the comprehensive state of a monitored object at the current monitoring time, and is labeled with a device ID, a timestamp and a location tag. Finally, the structured feature vector set is uploaded to a digital twin model in the cloud through a communication network.

Images