Whole-flow closed-loop management and control system for icing galloping of power transmission line
By constructing a closed-loop management and control system for the entire process of transmission line icing and galloping, the problems of insufficient monitoring accuracy, poor adaptability of prediction models, and delayed response to governance have been solved. This system enables accurate prediction and efficient governance of transmission line icing and galloping, ensuring power grid safety.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-08
- Publication Date
- 2026-04-03
AI Technical Summary
Existing technologies lack sufficient monitoring accuracy for icing and galloping on transmission lines, have poor predictive model adaptability, provide crude early warning classifications, and are passive in response to governance issues. They also lack closed-loop management throughout the entire process, resulting in inaccurate risk prediction and delayed disaster governance.
A closed-loop management and control system for the entire process of icing and galloping of transmission lines is constructed, including a multi-source sensing and monitoring module, an intelligent fusion prediction module, a dynamic hierarchical early warning module, and a linkage and collaborative governance module. Combining edge computing and cloud collaborative architecture, it realizes automated and precise closed-loop management and control of monitoring, prediction, early warning and governance.
It has significantly improved monitoring accuracy, forecasting accuracy, and governance efficiency, enabling proactive defense and precise management of ice galloping disasters and ensuring the safe and stable operation of the power grid.
Smart Images
Figure CN121786780A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of power transmission line safety protection technology, and more particularly to a closed-loop control system for the entire process of power transmission line icing and galloping. Background Technology
[0002] Icing galloping of transmission lines refers to the low-frequency, large-amplitude self-excited vibration phenomenon of conductors caused by the combined effects of eccentric icing and wind excitation. This can lead to serious accidents such as hardware damage, conductor breakage, tower collapse, and phase-to-phase flashover. Therefore, risk prediction for icing galloping of transmission lines is extremely necessary. However, existing technologies have significant limitations in predicting the risks of icing galloping of transmission lines. In terms of monitoring, manual inspections are inefficient, and fixed sensors are susceptible to environmental interference, resulting in insufficient data accuracy and continuity. In terms of prediction, traditional models ignore the influence of micro-topography, making it difficult to achieve refined regional predictions. In terms of early warning, the grading standards are too simplistic to match differentiated prevention and control needs. In terms of mitigation, de-icing and anti-galloping devices are mostly statically designed, with delayed response and poor adaptability, lacking a linkage mechanism with monitoring data.
[0003] Therefore, it is urgent to build a closed-loop management and control system covering the entire process to achieve proactive defense and precise management of ice-covered dancing disasters. Summary of the Invention
[0004] To address the aforementioned issues, this invention provides a closed-loop management system for the entire process of transmission line icing and galloping. Through the coordinated operation of various modules, it achieves automated and precise closed-loop management of the entire process of "monitoring-prediction-early warning-treatment," solving problems such as insufficient monitoring accuracy, poor adaptability of prediction models, coarse early warning classification, and passive treatment response in existing technologies. Furthermore, it innovatively introduces a prediction model adapted to micro-topography and a dynamic response treatment mechanism, combined with an edge computing and cloud collaborative architecture, which can significantly improve the prevention and control capabilities of transmission line icing and galloping disasters and ensure the safe and stable operation of the power grid.
[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: This invention provides a closed-loop management and control system for the entire process of icing and galloping of transmission lines, including a multi-source sensing and monitoring module with communication connection, an intelligent fusion prediction module, a dynamic hierarchical early warning module, a linkage and collaborative governance module, and a data hub module; The multi-source sensing and monitoring module is used to collect multi-source data of the transmission line in real time. The multi-source data includes icing status, galloping characteristics, environmental parameters and equipment operating data. The intelligent fusion prediction module, based on historical and real-time data stored in the data hub module, uses a fusion algorithm of a geographic weighted regression model and a deep learning model to predict the probability, amplitude, and frequency of ice dancing. The dynamic hierarchical early warning module is used to establish a multi-dimensional risk assessment system based on prediction results and real-time monitoring data, and generate differentiated early warning information. The linkage and collaborative governance module is used to respond to early warning commands, dispatch active anti-flashover devices and dynamic de-icing equipment to perform prevention and control operations, and provide feedback on the governance effect; The data hub module is used for the storage, encryption, sharing, and iterative updating of data across the entire data chain.
[0006] Furthermore, the multi-source sensing and monitoring module includes a distributed sensing unit and an environmental sensing unit; The distributed sensing unit includes a ring-shaped force measuring component, a three-dimensional accelerometer, an RTK positioning subunit, and an image acquisition subunit, all mounted on the conductor. The ring-shaped force measuring component acquires the icing gravity and trotting force parameters through contact measurement. The three-dimensional accelerometer is used to collect the vibration acceleration of the conductor along the X, Y, and Z axes. The RTK positioning subunit is used to acquire the spatial coordinates of the conductor in real time to locate the trotting trajectory. The image acquisition subunit uses an anti-fog high-definition camera to capture the icing morphology. The environmental sensing unit includes a micro-weather station and a terrain sensing subunit. The micro-weather station is used to collect meteorological data, and the terrain sensing subunit is used to acquire micro-terrain parameters.
[0007] Preferably, the icing status data includes icing gravity, icing morphology, and icing thickness; the galloping characteristics include vibration acceleration, galloping amplitude, galloping frequency, and galloping trajectory; the meteorological data includes wind speed, wind direction, temperature, and precipitation data along the line; the micro-topographic parameters include altitude, slope, orientation, and vegetation cover; and the equipment operating condition data includes conductor body operating condition data, hardware and support structure operating condition data, and monitoring and control equipment operating condition data.
[0008] Preferably, the distributed sensing unit is powered by a photovoltaic-battery composite energy system and has charge / discharge protection and power distribution functions.
[0009] Furthermore, the prediction process of the intelligent fusion prediction module includes the following steps: Outlier removal, noise error correction, and parameter normalization are performed on historical and real-time data. Using historical geographic data as the independent variable and historical meteorological data as the dependent variable, the Kriging interpolation method was used to fit the data, establish an initial meteorological geostatistical model, and output refined meteorological data at the conductor spacer between each tower. Using historical refined meteorological data, micro-topographic parameters, and line operating condition data as input features, a GWR-LSTM fusion model is constructed to output the probability of ice-covered galloping, the amplitude of galloping, and the frequency of galloping. The GWR-LSTM fusion model is retrained regularly using new historical data to ensure that the model adapts to changes in the environment.
[0010] Furthermore, based on the initial meteorological geostatistical model, refined meteorological data is output, including the following steps: Geographic parameters of all key points along the transmission line were extracted as independent variables, and meteorological data with the same time range as the independent variable dataset were extracted as dependent variables. The independent variables and dependent variables were then spatially matched. A semivariogram was constructed to analyze the spatial variation patterns of geographical and meteorological data. A scatter plot of the semivariogram values versus spatial distance was plotted and fitted to a theoretical model that conforms to the geographical characteristics of power transmission lines. Based on the fitting results of the semivariogram, an initial meteorological geostatistical model was constructed, and the meteorological data gaps at the conductor spacers were filled by interpolation calculations. The accuracy of the initial meteorological geostatistical model is verified using validation set data. Based on the validated initial meteorological geostatistical model, refined meteorological data for the conductor spacers between each tower is output.
[0011] Preferably, the dynamic hierarchical early warning module includes a risk assessment unit and an early warning push unit; Risk assessment unit: The weighted scoring method is adopted, and the galloping amplitude, galloping frequency, icing thickness and line load are used as assessment indicators to divide the risk level into four levels, including normal, level one risk, level two risk and level three risk. Early warning push unit: Pushes early warning information through multiple channels. The information includes risk level, specific location, forecast parameters and suggested response measures.
[0012] The coordinated governance module includes an active anti-dash unit, a dynamic de-icing unit, and an emergency response unit: Active anti-galloping unit: includes an adjustable spoiler and an intelligent detuned pendulum, which dynamically adjusts the spoiler angle and detuned pendulum parameters according to the galloping frequency, thereby suppressing galloping by changing aerodynamic characteristics; Dynamic de-icing unit: includes a distributed pulse de-icing device and an ultrasonic de-icing device. Ultrasonic de-icing is activated when there is a level 1 risk, and pulse de-icing is triggered when there is a level 2 or higher risk. The de-icing parameters are adaptively adjusted according to the predicted ice thickness. Emergency Response Unit: Used to automatically generate backup line switching plans when there is a level 3 risk, coordinate with the dispatch center to realize load transfer, and push emergency repair resource allocation instructions to the operation and maintenance team.
[0013] The collaborative governance module also includes an effect feedback unit, which compares the changes in the dancing parameters before and after governance in real time by monitoring data, generates a governance effect evaluation report, and feeds it back to the data hub module for model optimization.
[0014] The data hub module adopts a two-tier edge-cloud architecture. Edge nodes are deployed in the control cabinet of the power transmission tower and integrate edge computing units to realize real-time preprocessing of monitoring data and rapid local response. The cloud platform uses an encrypted database to store historical data, model parameters and governance records, and uses blockchain technology to achieve data traceability and support data sharing among multiple departments.
[0015] The beneficial effects of this invention are as follows: Improved monitoring accuracy: Through multi-source sensor fusion and anti-interference design, the measurement error of ice thickness is ≤1mm and the measurement error of gyratory amplitude is ≤2cm, effectively solving the problem of inaccurate traditional monitoring data.
[0016] Prediction capability optimization: Based on the GWR-LSTM fusion model combined with micro-topography and time series features, the accuracy of the dance prediction is ≥92% within 24 hours, realizing refined prediction for each monitoring point.
[0017] Precise early warning response: Multi-dimensional hierarchical standards are matched with differentiated prevention and control measures to avoid excessive or insufficient early warnings and reduce ineffective operation and maintenance costs.
[0018] Improved governance efficiency: The dynamic response de-icing and anti-galling device and emergency linkage mechanism shorten the galling suppression response time to within 5 minutes, increase the ice melting efficiency by more than 40%, and significantly reduce disaster losses.
[0019] Closed-loop management is achieved by continuously optimizing system performance through effect feedback and model iteration, forming an adaptive ice-covering dance prevention and control system that can adapt to complex and ever-changing climate and geographical conditions. Attached Figure Description
[0020] Figure 1 This is a framework diagram of a closed-loop control system for the entire process of icing and galloping of power transmission lines according to the present invention. Detailed Implementation
[0021] Please see Figure 1 As shown, the present invention relates to a closed-loop control system for the entire process of icing and galloping of transmission lines, including a multi-source sensing and monitoring module, an intelligent fusion prediction module, a dynamic hierarchical early warning module, a linkage and collaborative governance module, and a data hub module connected by communication. The multi-source sensing and monitoring module is used to collect multi-source data of the transmission line in real time. The multi-source data includes icing status, galloping characteristics, environmental parameters and equipment operating data. The multi-source sensing and monitoring module includes a distributed sensing unit and an environmental sensing unit. The distributed sensing unit includes a ring-shaped force measuring component, a three-dimensional accelerometer, an RTK positioning subunit, and an image acquisition subunit, all mounted on the conductor. The ring-shaped force measuring component acquires the icing gravity and trotting force parameters through contact measurement. The three-dimensional accelerometer is used to collect the vibration acceleration of the conductor along the X, Y, and Z axes. The RTK positioning subunit is used to acquire the spatial coordinates of the conductor in real time to locate the trotting trajectory. The image acquisition subunit uses an anti-fog high-definition camera to capture the icing morphology. The environmental sensing unit includes a micro-weather station and a terrain sensing subunit. The micro-weather station is used to collect meteorological data, and the terrain sensing subunit is used to acquire micro-terrain parameters.
[0022] The distributed sensing unit is powered by a photovoltaic-battery composite energy system, which has charge / discharge protection and power distribution functions. The photovoltaic panels are installed on the crossarm of the tower and can use 2W-18V photovoltaic panels and 6.4V-1.8AH lithium iron phosphate batteries, which can achieve stable power supply in extreme environments of -40℃ to +70℃ and can still maintain normal operation of the equipment under conditions of no sunlight for 7 consecutive days.
[0023] The icing status data includes icing gravity, icing morphology, and icing thickness; the galloping characteristics include vibration acceleration, galloping amplitude, galloping frequency, and galloping trajectory; the meteorological data includes wind speed, wind direction, temperature, and precipitation data along the line; the micro-topographic parameters include altitude, slope, orientation, and vegetation cover; and the equipment operating condition data includes conductor body operating condition data, hardware and support structure operating condition data, and monitoring and control equipment operating condition data.
[0024] Icing status acquisition: Monitoring nodes are installed at the conductor spacers between each tower. The ring force measuring component adopts a two-half structure and is fixed with bolts, and is sleeved on the outside of the conductor. When the conductor is iced or galloping, the pressure sensor converts the contact force into an electrical signal with a measurement accuracy of ±0.1N. The anti-fog high-definition camera takes an icing image every 5 minutes and is used in conjunction with an infrared supplementary light to ensure the imaging quality at night, directly capturing the icing morphology. The icing thickness is derived based on the pressure and icing images.
[0025] Galloping Feature Acquisition: A three-dimensional accelerometer with a sampling frequency of 100Hz acquires the vibration acceleration of the conductor along the X, Y, and Z axes (the galloping frequency is calculated based on the time series analysis of the vibration acceleration), covering the galloping frequency range of 0.1-5Hz, reflecting the dynamic intensity of the galloping; The RTK positioning subunit achieves spatial coordinate positioning with an accuracy of ±2cm through differential calibration with the base station, records the spatial coordinate sequence of the conductor in real time (the galloping amplitude can be derived), and forms the complete motion trajectory of the conductor's galloping, allowing for intuitive judgment of the direction and range of the galloping motion.
[0026] Conductor body operating data: Conductor tension, reflecting the stress state of the conductor, to prevent breakage due to excessive tension caused by galloping. Tension sensors (integrated with the ring force measuring component) are deployed synchronously at the conductor spacer bar. The strain gauge senses the conductor's deformation under stress and converts the mechanical tension into an electrical signal. The measurement accuracy is ±50N, and the dynamic tension value of the conductor is output in real time. Conductor temperature, related to current carrying capacity and ice melting characteristics, to prevent overheating damage. A contact-type platinum resistance temperature sensor (integrated inside the ring force measuring component) is used, directly attached to the conductor surface. The measurement range is -40℃ to +120℃, with an accuracy of ±0.3℃. Temperature data is collected once every 1 minute. Degree of conductor surface damage, such as whether there are scratches, broken strands, or other physical damage to the conductor after ice peeling.
[0027] Fittings and Support Structure Operating Data: Spacer Bar Tightness: To prevent spacer bars from loosening or falling off due to galloping, torque sensors are installed at the spacer bar bolts to monitor bolt torque values in real time. When the torque is lower than a preset threshold (e.g., 80% of the design torque), it is determined that the tightness is insufficient. The data sampling frequency is once every 5 minutes. Insulator Leakage Current: To determine the insulation performance of the insulators and avoid phase-to-phase flashover, leakage current sensors are installed at the lower end of the insulator string to monitor the leakage current value through induced current signals. The measurement range is 0-10mA, with an accuracy of ±0.1mA, and real-time detection of insulation performance anomalies. Tower Connection Node Displacement: To monitor whether the tower has structural deformation due to galloping vibration.
[0028] Monitoring and managing the equipment's own operating data: The power supply voltage and current of the distributed sensing unit are integrated with the power management unit of the photovoltaic-cell composite energy system. Voltage / current sensors are integrated to collect the output voltage (range 6-18V) and current (range 0-0.5A) in real time to determine whether the equipment's power supply is stable; The mechanical component operating status of the intelligent detuned pendulum / adjustable turbulence device is monitored. The stepper motor of the adjustable turbulence device and the drive module of the intelligent detuned pendulum have built-in operating status monitoring chips, which output data such as motor speed and pendulum length adjustment position in real time. The sampling frequency is synchronized with the equipment's operating frequency.
[0029] Environmental parameter acquisition: A micro weather station is deployed every 2 kilometers along the transmission line to collect data on wind speed (measurement range 0-60m / s, accuracy ±0.1m / s), wind direction, temperature (measurement range -40℃~+70℃, accuracy ±0.5℃) and precipitation, so as to obtain the external meteorological conditions for galloping in real time.
[0030] Equipment operating condition related data: The terrain sensing subunit integrates a GIS data interface, imports digital elevation model (DEM) data along the route, and analyzes micro-topographic parameters such as altitude, slope, orientation, and vegetation coverage at each monitoring point, providing data support for subsequent analysis of the relationship between equipment operating conditions and terrain.
[0031] The intelligent fusion prediction module, based on historical and real-time data stored in the data hub module, uses a fusion algorithm of a geographic weighted regression model and a deep learning model to predict the probability, amplitude, and frequency of ice dancing. The prediction process of the intelligent fusion prediction module includes the following steps: Outlier removal, noise error correction, and parameter normalization are performed on historical and real-time data. The 3σ criterion is used to screen and remove abnormal data caused by sensor drift, communication packet loss, lightning interference, or equipment failure, such as sudden increases in wind speed or negative icing gravity values. A Kalman filter algorithm is used to filter dynamic data such as vibration acceleration collected by the 3D accelerometer and spatial coordinates obtained by the RTK positioning subunit, reducing the impact of environmental noise on data accuracy and correcting deviations in displacement and acceleration data. All preprocessed valid data is mapped to the [0,1] interval to eliminate interference from differences in parameter magnitudes in subsequent model calculations and ensure input data consistency.
[0032] Using historical geographic data as the independent variable and historical meteorological data as the dependent variable, the Kriging interpolation method was used to fit the data, establish an initial meteorological geostatistical model, and output refined meteorological data at the conductor spacer between each tower. The process of outputting refined meteorological data based on the initial meteorological geostatistical model includes the following steps: Geographic parameters of all key points along the transmission line were extracted as independent variables, and meteorological data with the same time range as the independent variable dataset were extracted as dependent variables. The independent variables and dependent variables were then spatially matched. From the GIS database of the data hub module, the geographic parameters of all key points along the transmission line are exported, including basic geographic coordinates: latitude and longitude of each tower (accurate to 0.0001°), relative coordinates of conductor spacers (based on two adjacent towers, marking the distance of the spacer on the conductor, such as 100m and 200m from the tower); micro-topographic parameters: elevation (accuracy ±1m), slope (0-90°, accuracy ±1°), orientation (0-360°, calculated clockwise with true north as 0°), and vegetation coverage (0-100%, accuracy ±5%) of each spacer's corresponding point. This results in a dataset of independent variables: "spacer ID - latitude and longitude - elevation - slope - orientation - vegetation coverage," ensuring that each conductor spacer has a unique corresponding geographic parameter record.
[0033] Meteorological data with a time range consistent with the independent variable dataset (covering at least three icing periods to ensure data representativeness) are extracted from the historical meteorological database of the data center module. This includes historical observation data from existing meteorological stations along the route and in the surrounding area. Parameters include wind speed (unit: m / s, accuracy ±0.1 m / s), wind direction (0-360°, accuracy ±5°), temperature (unit: ℃, accuracy ±0.5℃), and precipitation (unit: mm, accuracy ±0.1 mm). The time granularity is recorded hourly.
[0034] Using "time + geographic coordinates" as the association key, the historical meteorological data of each meteorological station is spatially matched with the geographic parameters of the spacers around the meteorological station to form a corresponding dataset of "spacer ID - historical time - wind speed - wind direction - temperature - precipitation - geographic parameters", ensuring that the geographic parameters (independent variables) of each spacer have corresponding meteorological data (dependent variables) for the same period.
[0035] A semivariogram was constructed to analyze the spatial variation patterns of geographical and meteorological data. A scatter plot of the semivariogram values versus spatial distance was plotted and fitted to a theoretical model that conforms to the geographical characteristics of power transmission lines. For each meteorological parameter (wind speed, wind direction, temperature, precipitation), a semivariogram is constructed. Taking wind speed as an example, the calculation formula is as follows: ; Where γ(h) represents the semivariogram at a distance of h, and h represents the spatial distance between two conductor spacers (independent variable, the actual geographical distance calculated based on latitude, longitude, and altitude); N(h) is the number of spacer data pairs at a distance of h; Z(x i Z(x) represents the historical wind speed value (dependent variable) of the i-th spacer bar; i +h) represents the historical wind speed value of another spacer that is h away from the i-th spacer.
[0036] Plot a scatter plot of the semivariogram γ(h) versus spatial distance h. Select a theoretical model that conforms to the geographical characteristics of the transmission line (such as a spherical model or an exponential model) for fitting, determine the key parameters of the model, and ensure that the coefficient of determination R between the fitted theoretical model and the actual scatter plot is consistent. 2 A value ≥0.85 is used to ensure the reliability of spatial correlation analysis. Key parameters of the model include: nugget value, reflecting measurement errors or random variations at the microscale (such as small sensor errors), obtained through fitting; sill value, the value at which the semi-variable value reaches stability, reflecting the total degree of variation of meteorological parameters (including spatial structure variation and random variation); and range, the spatial distance at which the semi-variable value reaches the sill value. Beyond this distance, the spatial correlation of meteorological parameters is negligible.
[0037] Based on the fitting results of the semivariogram, an initial meteorological geostatistical model was constructed, and the meteorological data gaps at the conductor spacers were filled by interpolation calculations. For each conductor spacer (target point) to be interpolated, all known data points whose spatial distance is within the "variable range" are selected. Based on the Kriging interpolation principle, the weight coefficient λ of each known data point to the target point is calculated. i The weights must meet two conditions: first, the sum of the weights must be 1 to ensure that the interpolation result has no systematic bias; second, the variance of the interpolation error must be minimized.
[0038] Using historical geographic data as spatial constraints, the meteorological parameters of each known data point are multiplied by their corresponding weighting coefficients and summed to obtain the interpolated meteorological data for the target interval bars. The formula is as follows: ; Among them, Z * (x0) represents the interpolated meteorological data for the target spacing bar, λ i Z(x) represents the weight of the i-th known data point. i ) represents the historical meteorological data of the i-th known data point, and n represents the number of known data points participating in the interpolation.
[0039] The above interpolation calculations are performed on each of the conductor spacers to form an initial meteorological geostatistical model covering all spacers of the entire line. The model can output the predicted meteorological parameters of any spacer for the same historical period.
[0040] The accuracy of the initial meteorological geostatistical model is verified using validation set data. Based on the validated initial meteorological geostatistical model, refined meteorological data for the conductor spacers between each tower is output.
[0041] From the matched "geographic-meteorological" dataset, 20% of the interval bar data is randomly selected as the validation set, and the remaining 80% is used as the training set. The actual historical meteorological data for each interval bar in the validation set is compared with the interpolated prediction data for that interval bar from the initial model, and the mean absolute error (MAE), root mean square error (RMSE), and coefficient of determination R are calculated. 2 Three accuracy indicators. If the accuracy indicators are not met, return to the semivariogram analysis step, readjust the theoretical model type (e.g., replace the spherical model with an exponential model) or optimize parameters such as range and nugget value until the model accuracy meets the above requirements, and finally determine a reliable initial meteorological geostatistical model.
[0042] The system retrieves the "tower-spacer" topology data (such as tower number, number and location of spacers between adjacent towers) of the transmission lines from the data hub module. It then associates the interpolation results of the initial model with the specific spacer locations, forming a correspondence table of "tower number - spacer location (distance from tower) - refined meteorological parameters." For each spacer, it outputs the historical refined meteorological data calculated by the initial model for the same period, i.e., outputting the daily meteorological parameters (wind speed, wind direction, temperature, precipitation) for each day during the historical icing period, on an hourly basis. This data is stored in JSO format on the edge nodes of the data hub module and in the cloud database.
[0043] Using historical refined meteorological data, micro-topographic parameters, and line operating condition data as input features, a GWR-LSTM fusion model is constructed to output the probability of ice-covered galloping, the amplitude of galloping, and the frequency of galloping. The GWR model (Geographically Weighted Regression Model) is used to capture geographic spatial heterogeneity, that is, the differences in the impact of different micro-topographic conditions on ice blasting. It takes the micro-topographic parameters and refined meteorological data of each monitoring point as inputs and outputs local regression coefficients to reflect the influence weight of the meteorological-topographic combination on ice thickness and blasting intensity in the region (such as the coefficient correction for faster ice thickness growth in steep slope areas).
[0044] LSTM (Long Short-Term Memory) model: Used to learn time-series dependencies, i.e., the patterns of changes in weather conditions and icing status over time in historical data. It takes historical icing dance time-series data (such as the trends in icing thickness and dance amplitude over the past 30 days) and real-time standardized data as input from the data hub module. Gating units (input gate, forget gate, output gate) memorize long-term dependency features, outputting a predicted trend of icing dance over time.
[0045] A fully connected layer is used to fuse the local regression coefficients output by the GWR model with the time series prediction results output by the LSTM model, constructing a complete GWR-LSTM fusion model. The mean squared error is used as the loss function and optimized using the Adam optimizer, iterating until the loss value is below 0.001 to ensure prediction accuracy.
[0046] The refined meteorological data (for the next 1-24 hours), micro-topographic parameters, and line operating condition data (including conductor body and equipment operating status data, such as conductor tension, conductor temperature, spacer tightness, insulator leakage current, etc.) from each monitoring point are input into the pre-trained GWR-LSTM fusion model. The model outputs hourly prediction results, including the probability of icing galloping (0-100%, quantifying the possibility of icing galloping at each monitoring point within the next 1-24 hours; a probability ≥50% is judged as "high risk tendency," triggering subsequent amplitude and frequency predictions). If galloping is predicted, the corresponding galloping amplitude (unit: m) and galloping frequency (unit: Hz) are output.
[0047] The GWR-LSTM fusion model is retrained regularly using new historical data to ensure that the model adapts to changes in the environment.
[0048] For the monthly aggregated real-time monitoring data (icing status, galloping characteristics, meteorological parameters) and actual icing and galloping event records (such as whether galloping occurred, actual amplitude and frequency), valid data are selected and added to the historical database. Using the supplemented historical database as the training set, the GWR-LSTM fusion model is retrained using the same Adam optimizer and mean squared error loss function as the initial training. The focus is on optimizing the local regression coefficients of the GWR model (matching new micro-topography-meteorological correlation patterns) and the network weights of the LSTM model (learning new time-series trends). The retrained optimized model parameters are then distributed to edge nodes (edge computing units of transmission tower control cabinets) and the cloud platform to replace the old model, ensuring that subsequent predictions are always based on the latest data patterns and maintaining a 24-hour galloping prediction accuracy of ≥92%.
[0049] The dynamic hierarchical early warning module is used to establish a multi-dimensional risk assessment system based on prediction results and real-time monitoring data, and generate differentiated early warning information. The dynamic hierarchical early warning module includes a risk assessment unit and an early warning push unit; Risk assessment unit: The weighted scoring method is adopted, with galloping amplitude (weight 0.4), galloping frequency (weight 0.3), icing thickness (weight 0.2), and line load (weight 0.1) as assessment indicators to divide the risk into four levels, including normal (score < 30), level one risk (30 ≤ score < 50), level two risk (50 ≤ score < 80), and level three risk (score ≥ 80). The galloping amplitude threshold is set to 0.5m, the galloping frequency threshold to 1Hz, and the icing thickness threshold to 10mm. The rated load of the line is used as the benchmark value. The predicted results are compared with the set index thresholds, and the total risk score is calculated by combining the assigned weights. For example, if a monitoring point has a galloping amplitude of 0.6m (score 40), a frequency of 0.8Hz (score 24), an icing thickness of 8mm (score 16), and a line load of 80% (score 8), the calculated total risk score is 88, and it is judged as a level three risk.
[0050] Early warning push unit: Pushes early warning information through multiple channels (SMS, platform pop-ups, and operation and maintenance terminal APP, etc.). The information includes risk level, specific location, prediction parameters, and suggested handling measures.
[0051] The early warning push subunit is linked with the power grid operation and maintenance platform, emergency command center, and operation and maintenance personnel's APP. Under normal conditions, it only updates data logs; the first-level risk pushes inspection reminders, and it is recommended to increase special inspections by 2 times a day; the second-level risk pushes device control instructions, with a schematic diagram of the anti-surge device installation location; the third-level risk triggers an emergency response, and simultaneously pushes the backup line switching plan and the location information of the repair personnel.
[0052] The linkage and collaborative governance module is used to respond to early warning commands, dispatch active anti-flashover devices and dynamic de-icing equipment to perform prevention and control operations, and provide feedback on the governance effect; The collaborative governance module includes an active anti-dash unit, a dynamic de-icing unit, and an emergency response unit. Active anti-galloping unit: includes an adjustable spoiler and an intelligent detuned pendulum, which dynamically adjusts the spoiler angle and detuned pendulum parameters according to the galloping frequency, thereby suppressing galloping by changing aerodynamic characteristics; The adjustable spoiler has a built-in stepper motor. After receiving a warning command, it automatically adjusts the angle of the spoiler (adjustment range 0-90°) according to the galloping frequency to change the flow field distribution around the conductor and suppress self-excited vibration. The intelligent detuned pendulum adjusts its natural frequency by changing the pendulum length (adjustment range 0.5-1.5m) to avoid resonance with the galloping frequency of the conductor.
[0053] Dynamic de-icing unit: includes a distributed pulse de-icing device and an ultrasonic de-icing device. Ultrasonic de-icing is activated when there is a level 1 risk, and pulse de-icing is triggered when there is a level 2 or higher risk. The de-icing parameters are adaptively adjusted according to the predicted ice thickness. The ultrasonic de-icing device is activated when there is a level 1 risk. The output power is adjusted according to the predicted ice thickness (5-20W) and the surface ice is peeled off by high-frequency vibration. The distributed pulse de-icing device is put into operation when there is a level 2 or higher risk. The de-icing current (100-500A) is calculated according to the conductor material and ice thickness. It adopts a short-time pulse power supply mode to reduce the impact on line operation. The de-icing efficiency is 40% higher than that of traditional DC de-icing.
[0054] Emergency Response Unit: Used to automatically generate backup line switching plans when there is a level 3 risk, coordinate with the dispatch center to realize load transfer, and push emergency repair resource allocation instructions to the operation and maintenance team.
[0055] In the event of a Level 3 risk, the emergency response unit automatically retrieves the power grid topology and generates the optimal backup line switching plan through load flow calculation. This plan is then submitted to the dispatch center for review and execution. Simultaneously, the unit sends information on the fault location, required equipment, and safety precautions to nearby repair teams, thereby shortening the emergency response time.
[0056] The collaborative governance module also includes an effect feedback unit, which compares the changes in the dancing parameters before and after governance in real time by monitoring data, generates a governance effect evaluation report, and feeds it back to the data hub module for model optimization.
[0057] The data hub module is used for the storage, encryption, sharing, and iterative updating of data across the entire data chain.
[0058] The data hub module adopts a two-tier edge-cloud architecture. Edge nodes are deployed in the control cabinet of the power transmission tower and integrate edge computing units to realize real-time preprocessing of monitoring data and rapid local response. The cloud platform uses an encrypted database to store historical data, model parameters and governance records, and uses blockchain technology to achieve data traceability and support data sharing among multiple departments.
[0059] Edge nodes store real-time monitoring data for the past 7 days, using an SQLite database for fast local queries. The cloud platform uses a MySQL cluster to store historical data (with a 10-year retention period) and is equipped with a data backup server to prevent data loss. All transmitted data uses the AES-256 encryption algorithm, and the cloud platform records data modification logs using blockchain technology. Each data block contains a timestamp and hash value, ensuring data immutability and traceability. The cloud platform automatically aggregates monitoring data and dance event records monthly, retrains the GWR-LSTM fusion model, optimizes model parameters, and distributes the updated model to edge nodes, achieving continuous improvement in prediction accuracy.
[0060] The above embodiments are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.
Claims
1. A closed-loop control system for the entire process of icing and galloping on transmission lines, characterized in that, It includes a multi-source sensing and monitoring module for communication connectivity, an intelligent fusion and prediction module, a dynamic hierarchical early warning module, a collaborative governance module, and a data hub module; The multi-source sensing and monitoring module is used to collect multi-source data of the transmission line in real time. The multi-source data includes icing status, galloping characteristics, environmental parameters and equipment operating data. The intelligent fusion prediction module, based on historical and real-time data stored in the data hub module, uses a fusion algorithm of a geographic weighted regression model and a deep learning model to predict the probability, amplitude, and frequency of ice dancing. The dynamic hierarchical early warning module is used to establish a multi-dimensional risk assessment system based on prediction results and real-time monitoring data, and generate differentiated early warning information. The linkage and collaborative governance module is used to respond to early warning commands, dispatch active anti-flashover devices and dynamic de-icing equipment to perform prevention and control operations, and provide feedback on the governance effect; The data hub module is used for the storage, encryption, sharing, and iterative updating of data across the entire data chain.
2. The closed-loop control system for the entire process of icing and galloping of transmission lines according to claim 1, characterized in that, The multi-source sensing and monitoring module includes a distributed sensing unit and an environmental sensing unit. The distributed sensing unit includes a ring-shaped force measuring component, a three-dimensional accelerometer, an RTK positioning subunit, and an image acquisition subunit, all mounted on the conductor. The ring-shaped force measuring component acquires the icing gravity and trotting force parameters through contact measurement. The three-dimensional accelerometer is used to collect the vibration acceleration of the conductor along the X, Y, and Z axes. The RTK positioning subunit is used to acquire the spatial coordinates of the conductor in real time to locate the trotting trajectory. The image acquisition subunit uses an anti-fog high-definition camera to capture the icing morphology. The environmental sensing unit includes a micro-weather station and a terrain sensing subunit. The micro-weather station is used to collect meteorological data, and the terrain sensing subunit is used to acquire micro-terrain parameters.
3. The closed-loop control system for the entire process of icing and galloping of transmission lines according to claim 2, characterized in that, The icing status data includes icing gravity, icing morphology, and icing thickness; the galloping characteristics include vibration acceleration, galloping amplitude, galloping frequency, and galloping trajectory; the meteorological data includes wind speed, wind direction, temperature, and precipitation data along the line; the micro-topographic parameters include altitude, slope, orientation, and vegetation cover; and the equipment operating condition data includes conductor body operating condition data, hardware and support structure operating condition data, and monitoring and control equipment operating condition data.
4. The closed-loop control system for the entire process of icing and galloping of transmission lines according to claim 2, characterized in that, The distributed sensing unit is powered by a photovoltaic-battery composite energy system and has charge / discharge protection and power distribution functions.
5. A closed-loop control system for the entire process of icing and galloping of transmission lines according to claim 1, characterized in that, The prediction process of the intelligent fusion prediction module includes the following steps: Outlier removal, noise error correction, and parameter normalization are performed on historical and real-time data. Using historical geographic data as the independent variable and historical meteorological data as the dependent variable, the Kriging interpolation method was used to fit the data, establish an initial meteorological geostatistical model, and output refined meteorological data at the conductor spacer between each tower. Using historical refined meteorological data, micro-topographic parameters, and line operating condition data as input features, a GWR-LSTM fusion model is constructed to output the probability of ice-covered galloping, the amplitude of galloping, and the frequency. The GWR-LSTM fusion model is retrained regularly using new historical data to ensure that the model adapts to changes in the environment.
6. A closed-loop control system for the entire process of icing and galloping of transmission lines according to claim 5, characterized in that, The process of outputting refined meteorological data based on the initial meteorological geostatistical model includes the following steps: Geographic parameters of all key points along the transmission line were extracted as independent variables, and meteorological data with the same time range as the independent variable dataset were extracted as dependent variables. The independent variables and dependent variables were then spatially matched. A semivariogram was constructed to analyze the spatial variation patterns of geographical and meteorological data. A scatter plot of the semivariogram values versus spatial distance was plotted and fitted to a theoretical model that conforms to the geographical characteristics of power transmission lines. Based on the fitting results of the semivariogram, an initial meteorological geostatistical model was constructed, and the meteorological data gaps at the conductor spacers were filled by interpolation calculations. The accuracy of the initial meteorological geostatistical model is verified using validation set data. Based on the validated initial meteorological geostatistical model, refined meteorological data for the conductor spacers between each tower is output.
7. A closed-loop control system for the entire process of icing and galloping of transmission lines according to claim 1, characterized in that, The dynamic hierarchical early warning module includes a risk assessment unit and an early warning push unit; Risk assessment unit: The weighted scoring method is adopted, and the galloping amplitude, galloping frequency, icing thickness and line load are used as assessment indicators to divide the risk level into four levels, including normal, level one risk, level two risk and level three risk. Early warning push unit: Pushes early warning information through multiple channels. The information includes risk level, specific location, forecast parameters and suggested response measures.
8. A closed-loop control system for the entire process of icing and galloping of transmission lines according to claim 1, characterized in that, The coordinated governance module includes an active anti-dash unit, a dynamic de-icing unit, and an emergency response unit: Active anti-galloping unit: includes an adjustable spoiler and an intelligent detuned pendulum, which dynamically adjusts the spoiler angle and detuned pendulum parameters according to the galloping frequency, thereby suppressing galloping by changing aerodynamic characteristics; Dynamic de-icing unit: includes a distributed pulse de-icing device and an ultrasonic de-icing device. Ultrasonic de-icing is activated when there is a level 1 risk, and pulse de-icing is triggered when there is a level 2 or higher risk. The de-icing parameters are adaptively adjusted according to the predicted ice thickness. Emergency Response Unit: Used to automatically generate backup line switching plans when there is a level 3 risk, coordinate with the dispatch center to realize load transfer, and push emergency repair resource allocation instructions to the operation and maintenance team.
9. A closed-loop control system for the entire process of icing and galloping of transmission lines according to claim 8, characterized in that, The collaborative governance module also includes an effect feedback unit, which compares the changes in the dancing parameters before and after governance in real time by monitoring data, generates a governance effect evaluation report, and feeds it back to the data hub module for model optimization.
10. A closed-loop control system for the entire process of icing and galloping of transmission lines according to claim 1, characterized in that, The data hub module adopts a two-tier edge-cloud architecture. Edge nodes are deployed in the control cabinet of the power transmission tower and integrate edge computing units to realize real-time preprocessing of monitoring data and rapid local response. The cloud platform uses an encrypted database to store historical data, model parameters and governance records, and uses blockchain technology to achieve data traceability and support data sharing among multiple departments.