Power transmission tower power transmission line galloping monitoring system

Abstract

The invention belongs to the technical field of power transmission line monitoring, and discloses a power transmission tower power transmission line galloping monitoring system which comprises a data acquisition module, a data processing module, an intelligent prediction module, an early warning response module and a man-machine interaction module. The galloping prediction precision is improved through multi-source data fusion and physical constraint modeling, a mixed architecture of time sequence analysis and dynamic characteristic fusion is adopted, complex correlation characteristics of meteorological parameters and conductor dynamic behaviors are effectively captured, an intelligent prediction model is optimized in combination with physical equation constraints, and the galloping prediction accuracy is improved. The physical rationality and extreme scene adaptability of a prediction result are obviously enhanced; the line state change is adapted in real time based on a dynamic threshold adjustment mechanism, and the early warning sensitivity and reliability are optimized; through spatio-temporal feature alignment and a multi-mode galloping mode identification technology, a composite vibration form is accurately analyzed, medium and long term trend pre-judgment and short-time risk early warning are synchronously realized, and a multi-dimensional decision support is provided for line safety regulation and control.

Description

Technical Field

[0001] The invention belongs to the technical field of transmission line monitoring, and particularly relates to a dancing monitoring system for transmission towers and transmission lines. Background Art

[0002] The dancing of transmission towers and transmission lines is a low-frequency and large-amplitude self-excited vibration phenomenon, mainly caused by uneven icing of the conductors under wind excitation. This phenomenon may lead to tripping of transmission lines, loosening of some bolts on the cross arms of iron towers, and damage to the cross arms, jumpers, insulators, and fittings of iron towers, seriously threatening the safe and stable operation of the power grid. Dancing usually occurs under specific meteorological conditions, such as when the temperature is between 0 and -10 °C or lower, the wind speed is between 2 and 25 m / s or higher, and the included angle between the wind direction and the line direction is within the range of 45° to 90°. The energy generated by dancing is huge, and long-term persistence may lead to serious accidents such as broken strands and broken wires of conductors, phase-to-phase flashovers, and mixed lines, and even cause the inclination or collapse of the tower poles.

[0003] The dancing of transmission lines is a major hidden danger threatening the safe operation of the power grid. Traditional monitoring systems mostly rely on single-sensor data and static experience thresholds for early warning, with significant technical defects. Existing prediction models usually based on statistical regression or shallow machine learning algorithms are difficult to capture the non-linear coupling relationship between meteorological parameters such as wind speed and ice thickness and conductor dynamic parameters such as tension and natural frequency, resulting in insufficient recognition accuracy for multi-modal composite dancing, such as vertical-torsional coupling vibration; secondly, most methods lack physical law constraints, and the prediction results are easily interfered by data noise, especially prone to misjudgments that violate the principles of aerodynamics under extreme meteorological conditions, such as predicting a swing amplitude exceeding the material limit of the conductor but not triggering an early warning; in addition, traditional systems use fixed early warning thresholds and cannot dynamically adapt to time-varying factors such as line aging and material fatigue, resulting in high false alarm rates and missed alarm rates. Therefore, it is necessary to improve them. Summary of the Invention

[0004] The purpose of the present invention is to provide a dancing monitoring system for transmission towers and transmission lines to solve the problems raised in the above background art.

[0005] To achieve the above purpose, the present invention provides the following technical solution: A dancing monitoring system for transmission towers and transmission lines, comprising: a data acquisition module, a data processing module, an intelligent prediction module, an early warning response module, and a human-computer interaction module;

[0006] The data acquisition module is used to collect three-dimensional meteorological data within the line corridor range in real time, including parameters such as wind speed, wind direction, temperature, humidity, and precipitation. At the same time, it integrates a meteorological radar data interface to obtain regional meteorological forecast information for the next 3 hours and constructs a micro-meteorological monitoring network;

[0007] The data processing module is responsible for processing the collected data, feature extraction, and pattern recognition;

[0008] The intelligent prediction module is used to predict the galloping trend of the transmission line in the future;

[0009] The warning response module automatically generates warning prompts according to the intelligent prediction module;

[0010] The human-machine interaction module provides a real-time data monitoring interface and a high-definition image stream. Maintenance personnel can adjust the warning threshold through the human-machine interaction module in combination with the line aging degree and material fatigue parameters, and support manual intervention in the output results of the prediction model.

[0011] Preferably, the data acquisition module includes a historical data storage unit, a meteorological parameter monitoring unit, an image acquisition unit, and a line state perception unit. The historical data storage unit is used to build a distributed database cluster to classify and store historical galloping event data, equipment maintenance records, and fault handling files. A time series database is used to store high-frequency sampling data, and a data life cycle management strategy is designed to realize the intelligent retrieval and rapid call of historical data. The meteorological parameter monitoring unit is responsible for collecting environmental data such as wind speed, wind direction, temperature, humidity, air pressure, and precipitation intensity in real time. The image acquisition unit uses a multi-spectral imaging device to achieve high-definition video acquisition at 30 frames per second, and cooperates with infrared thermal imaging technology for all-weather real-time monitoring.

[0012] Preferably, the line state perception unit obtains physical parameters such as wire amplitude, swing frequency, tension change, and tilt angle through a distributed sensor network, and the monitoring accuracy is controlled within the range of ±0.5 degrees. At the same time, a line dynamic characteristic database is established to record the natural vibration modes of wires in different sections.

[0013] Preferably, the data processing module includes a data cleaning unit, a feature extraction and fusion unit, and a pattern recognition unit. The data cleaning unit applies an outlier detection algorithm to eliminate sensor false alarm data, eliminates random noise interference through the sliding window mean method, and implements spatio-temporal correlation compensation for missing data to ensure the integrity and reliability of the input data. The feature extraction and fusion unit extracts time-frequency domain features through Fourier analysis, aligns meteorological parameters, mechanical parameters, and geographical information in space and time, constructs a feature engineering processing pipeline, extracts a combination of key influencing factors, and generates a standardized data set of multi-dimensional feature vectors. The pattern recognition unit uses wavelet transform to analyze the vibration spectrum characteristics, identifies typical galloping patterns, including vertical oscillation, horizontal swing, and torsional deformation; establishes a line dynamic behavior classification model to realize the automatic recognition and pattern coding of different galloping forms.

[0014] Preferably, the intelligent prediction module includes a model training unit, a real-time prediction unit, and a result optimization unit. The model training unit is used to train an intelligent prediction model; the real-time prediction unit generates a future 3-hour galloping risk probability distribution through the intelligent prediction model and outputs key indicators such as the maximum swing prediction value, the dangerous duration, and the energy accumulation index; the result optimization unit applies a fuzzy logic algorithm to perform confidence weighting on the multi-model prediction results and establishes a prediction error compensation mechanism.

[0015] Preferably, the model training unit first extracts historical complete monitoring record data from the historical data storage unit, constructs a training data set containing multiple feature dimensions, uses the sliding window technique to generate time series samples, sets the window length to 3 hours, and sets the sliding step to 30 minutes. At the same time, in the data annotation stage, line dynamics constraints are introduced: the maximum swing of the conductor does not exceed 1 / 20 of the span, the galloping frequency is negatively correlated with the conductor tension, and the ice thickness and the critical wind speed have an exponential decay relationship, so as to establish a constraint rule base and automatically filter abnormal data samples that violate physical laws. During the prediction model training process, the loss function is designed as the MSE loss + the physical consistency loss term. The physical term includes the residual of the conductor motion equation: λ(∂²y / ∂t² - (T / m)∂²y / ∂x² + c∂y / ∂t)^2, where λ is the constraint strength coefficient, set to 0.3; when the predicted swing exceeds 1.2 times the historical maximum value, the regularization strength is automatically enhanced, and at the same time, anti-aircraft dynamic adversarial samples are generated during the training process to improve the robustness of the model.

[0016] Preferably, the intelligent prediction model trained by the model training unit adopts a three-level prediction framework. The basic layer is an LSTM network for processing time series features; the middle layer is a random forest regressor for fusing static parameters; the output layer is a physical equation constraint layer. At the same time, offline training is carried out using historical data, similar scenarios are retrieved from the historical data storage unit, and the prediction model is used for prediction under the condition of ensuring the consistency of the time dimension, space dimension, and environmental dimension. The prediction results of the prediction model are compared with the historical data, and the hyperparameter configuration is optimized through cross-validation to form an intelligent prediction model suitable for different climate zones.

[0017] Preferably, the early warning response module has four levels of response criteria: blue, yellow, orange, and red. It gives early warning prompts according to the prediction results of the intelligent prediction model and generates response measure suggestions corresponding to the levels. Among them, the blue early warning means that the maximum swing amplitude ≤ 1.50% of the span, the energy index < 300 kJ / m, and the wind speed ≤ 5 m / s, and it is in the initial vibration stage. The response measure is to automatically generate a monitoring report and self-check the status of the sensor network at the same time; the yellow early warning means that 1.50% < swing amplitude ≤ 3.00% of the span, 300 ≤ energy index < 800, and 5 m / s < wind speed ≤ 10 m / s, and it is in the medium amplitude movement stage. The response measure is to increase the sampling frequency of the sensor and prepare the de-icing device to be on standby; the orange early warning means that 3.00% < swing amplitude ≤ 5.00% of the span, 800 ≤ energy index < 1500, and 10 m / s < wind speed ≤ 15 m / s, and it is in the multi-modal composite vibration stage. The response measures are to start the de-icing device, transfer the line load to the standby line, have the engineering repair team on standby at the scene, and restrict the approach of surrounding vehicles and personnel; the red early warning means that the swing amplitude > 5.00% of the span, the energy index ≥ 1500 kJ / m, and the wind speed > 15 m / s, and it is in the three-dimensional violent galloping state. The response measures are to forcibly cut off the power supply of the line, evacuate the surrounding personnel, and start the vibration suppression device.

[0018] Preferably, the human-computer interaction module includes a real-time monitoring unit and an interaction unit. Among them, the real-time monitoring unit provides the real-time data of the transmission line collected by the data acquisition module and real-time images; the interaction unit is used for the operation and maintenance personnel to automatically optimize the alarm trigger conditions according to parameters such as the line operation years and material fatigue degree and correct the prediction results of the prediction model.

[0019] The beneficial effects of the present invention are as follows:

[0020] By multi-source data fusion and physical constraint modeling, the galloping prediction accuracy is improved. A hybrid architecture that combines time series analysis and dynamic characteristics fusion is adopted to effectively capture the complex correlation characteristics between meteorological parameters and the dynamic behavior of the conductor. Combining physical equation constraints to optimize the intelligent prediction model significantly enhances the physical rationality and extreme scenario adaptability of the prediction results; based on the dynamic threshold adjustment mechanism, it adapts to the line state changes in real time and optimizes the early warning sensitivity and reliability; through spatio-temporal feature alignment and multi-modal galloping mode recognition technology, it accurately analyzes the composite vibration form, and synchronously realizes medium- and long-term trend prediction and short-term risk early warning, providing multi-dimensional decision-making support for line safety control. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] Figure 1 It is a system block diagram of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0022] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0023] As Figure 1 shown, the embodiment of the present invention provides a dancing monitoring system for a transmission tower transmission line, including: a data acquisition module, a data processing module, an intelligent prediction module, an early warning response module, and a human-computer interaction module;

[0024] The data acquisition module is used to collect three-dimensional meteorological data within the line corridor in real time, including parameters such as wind speed, wind direction, temperature, humidity, and precipitation. At the same time, it integrates a meteorological radar data interface to obtain regional meteorological forecast information for the next 3 hours and constructs a micro-meteorological monitoring network;

[0025] The data processing module is responsible for processing the collected data, feature extraction, and pattern recognition;

[0026] The intelligent prediction module is used to predict the dancing trend of the transmission line in the future;

[0027] The early warning response module automatically generates an early warning prompt according to the intelligent prediction module;

[0028] The human-computer interaction module provides a real-time data monitoring interface and a high-definition image stream. The operation and maintenance personnel can adjust the early warning threshold through the human-computer interaction module in combination with the line aging degree and material fatigue parameters, and support manual intervention in the output results of the prediction model.

[0029] By collecting three-dimensional meteorological data and future meteorological forecast information of the line corridor in real time through the data acquisition module, the data processing module performs data processing and feature extraction, the intelligent prediction module predicts the dancing trend of the line, the early warning response module automatically generates an early warning prompt, and the human-computer interaction module provides a real-time monitoring interface and allows the operation and maintenance personnel to adjust the early warning threshold and manually intervene in the prediction results in combination with the line aging degree and material fatigue parameters, forming a closed-loop control system that synergistically optimizes machine intelligence and expert experience.

[0030] Among them, the data acquisition module includes a historical data storage unit, a meteorological parameter monitoring unit, an image acquisition unit and a line status perception unit. The historical data storage unit is used to build a distributed database cluster, classify and store historical dancing event data, equipment maintenance records, and fault handling archives, use a time series database to store high-frequency sampling data, and design a data lifecycle management strategy to realize intelligent retrieval and rapid call of historical data; the meteorological parameter monitoring unit is responsible for real-time collection of environmental data such as wind speed, wind direction, temperature and humidity, air pressure, and precipitation intensity; the image acquisition unit uses multi-spectral imaging equipment to achieve high-definition video acquisition at 30 frames per second, and cooperates with infrared thermal imaging technology for all-weather real-time monitoring.

[0031] By building a distributed database cluster and a multi-source data collection system, efficient management of historical data and accurate acquisition of real-time monitoring data are achieved. The time series database is used to store high-frequency sampling data, and the data lifecycle management strategy is combined to ensure the intelligent retrieval and rapid call of historical dance event data, equipment maintenance records, and fault handling archives.

[0032] Among them, the line status perception unit obtains physical parameters such as conductor amplitude, swing frequency, tension change, inclination angle, etc. through a distributed sensor network. The monitoring accuracy is controlled within the range of ±0.5 degrees. At the same time, a line dynamic characteristics database is established to record the inherent vibration modes of conductors in different sections.

[0033] The line status perception unit enables the system to accurately distinguish the characteristic differences between normal wind vibration and dangerous dancing, especially in complex terrain areas. By establishing a correlation map between the dynamic parameters of the conductor and micro-meteorological conditions, it provides physical field data with spatiotemporal continuity for subsequent pattern recognition, effectively improving the system's analysis depth of the conductor dancing mechanism and the timeliness of early warning.

[0034] Among them, the data processing module includes a data cleaning unit, a feature extraction and fusion unit and a pattern recognition unit. The data cleaning unit uses an outlier detection algorithm to eliminate sensor false alarm data, eliminates random noise interference through the sliding window mean method, and implements time-space correlation compensation for missing data to ensure the integrity and reliability of the input data; the feature extraction and fusion unit extracts time-frequency domain features through Fourier analysis, aligns meteorological parameters, mechanical parameters, and geographic information in time and space, constructs a feature engineering processing pipeline, extracts key influencing factor combinations, and generates a standardized data set of multi-dimensional feature vectors; the pattern recognition unit uses wavelet transform to analyze vibration spectrum characteristics and identify typical dancing patterns, including vertical oscillation, horizontal swing, and torsional deformation; a line dynamic behavior classification model is established to realize automatic recognition and pattern encoding of different dancing forms.

[0035] The feature extraction and fusion unit combines Fourier analysis, wavelet transform, and spatio-temporal alignment techniques, enabling it to simultaneously capture the non-linear coupling relationships among abrupt changes in meteorological parameters, abnormal fluctuations in mechanical parameters, and geographical environment features. By constructing a multi-dimensional feature vector set, it provides input features with both physical significance and statistical value for the machine learning model, significantly improving the accuracy and interpretability of galloping pattern classification.

[0036] Among them, the intelligent prediction module includes a model training unit, a real-time prediction unit, and a result optimization unit. The model training unit is used to train the intelligent prediction model; the real-time prediction unit generates the galloping risk probability distribution for the next 3 hours through the intelligent prediction model, and outputs key indicators such as the predicted maximum swing value, dangerous duration, and energy accumulation index; the result optimization unit applies the fuzzy logic algorithm to perform confidence weighting on the prediction results of multiple models and establishes a prediction error compensation mechanism.

[0037] The result optimization unit uses the fuzzy logic algorithm to perform confidence weighting on the outputs of multiple models and establishes a prediction error compensation mechanism. Specifically for special working conditions such as gust impacts and ice shedding, it automatically corrects the prediction deviation through physical constraint rules. This technical route that deeply integrates data-driven and mechanism models enables the system to not only capture complex non-linear relationships but also ensure that the prediction results conform to the principles of aerodynamics, significantly enhancing the engineering credibility of early warning decisions and the value of operation guidance.

[0038] Among them, the model training unit first extracts the historical complete monitoring record data from the historical data storage unit, constructs a training data set containing multiple feature dimensions, and uses the sliding window technique to generate time series samples. The window length is set to 3 hours, and the sliding step is set to 30 minutes. At the same time, in the data annotation stage, line dynamics constraints are introduced: the maximum swing of the conductor does not exceed 1 / 20 of the span, the galloping frequency is negatively correlated with the conductor tension, and there is an exponential decay relationship between the ice thickness and the critical wind speed, thereby establishing a constraint rule library to automatically filter out abnormal data samples that violate physical laws. During the training process of the prediction model, the loss function is designed as the MSE loss + physical consistency loss term. The physical term includes the residual of the conductor motion equation: λ(∂²y / ∂t² - (T / m)∂²y / ∂x² + c∂y / ∂t)^2, where λ is the constraint intensity coefficient, set to 0.3; when the predicted swing exceeds 1.2 times the historical maximum value, the regularization intensity is automatically enhanced, and at the same time, anti-aircraft dynamic adversarial samples are generated during the training process to improve the robustness of the model.

[0039] By designing a composite loss function that includes the MSE loss and the residual term of the motion equation, the problem of violating mechanical laws that may occur in pure data-driven models is fundamentally solved. The dynamic regularization mechanism automatically adjusts the constraint strength according to the predicted swing amplitude, and the adversarial sample generation technology is introduced in the model training stage to simulate the conductor vibration mode under extreme meteorological conditions, enhancing the model's adaptability to rare working conditions.

[0040] Among them, the intelligent prediction model trained by the model training unit adopts a three-level prediction framework. The basic layer is an LSTM network to process time series features; the middle layer is a random forest regressor to fuse static parameters; the output layer is a physical equation constraint layer. At the same time, offline training is carried out using historical data, similar scenarios are retrieved from the historical data storage unit, and the prediction model is used for prediction under the condition of ensuring the consistency of the time dimension, space dimension and environment dimension. The prediction results of the prediction model are reviewed with the historical data, and the hyperparameter configuration is optimized through cross-validation to form an intelligent prediction model suitable for different climate zones.

[0041] Through the adoption of a three-level prediction framework, the deep fusion and engineering application of multi-source heterogeneous data are realized, ensuring that the prediction results conform to the catenary dynamics principle. At the same time, a similar scenario retrieval mechanism is constructed relying on the historical database, and the model parameters are dynamically optimized through comparison and verification, so that the intelligent prediction model can still maintain high precision and strong robustness in the complex and changeable field environment.

[0042] Among them, the early warning of the early warning response module is divided into four-level response standards of blue, yellow, orange and red. Early warning prompts are given according to the prediction results of the intelligent prediction model, and corresponding-level response measure suggestions are generated. Among them, the blue early warning is that the maximum swing amplitude ≤ 1.50% of the span, the energy index < 300 kJ / m, and the wind speed ≤ 5 m / s, in the initial vibration stage, and the response measure is to automatically generate a monitoring report and self-check the status of the sensor network at the same time; the yellow early warning is that 1.50% < swing amplitude ≤ 3.00% of the span, 300 ≤ energy index < 800, 5 m / s < wind speed ≤ 10 m / s, in the medium-amplitude motion, and the response measure is to increase the sensor sampling frequency and prepare the de-icing device to stand by; the orange early warning is that 3.00% < swing amplitude ≤ 5.00% of the span, 800 ≤ energy index < 1500, 10 m / s < wind speed ≤ 15 m / s, in the multi-modal composite vibration, and the response measure is to start the de-icing device, transfer the line load to the standby line, the engineering repair team stands by on site, and restrict the approach of surrounding vehicles and personnel; the red early warning is that the swing amplitude > 5.00% of the span, the energy index ≥ 1500 kJ / m, and the wind speed > 15 m / s, in the three-dimensional violent dancing state, and the response measure is to forcibly cut off the line power supply, evacuate the surrounding personnel, and start the vibration suppression device.

[0043] Through the established four - level early warning system, operation and maintenance personnel can start the emergency plan 2 - 3 hours before the conductor enters three - dimensional violent galloping, greatly reducing the probability of tower collapse and wire breakage accidents.

[0044] Among them, the human - machine interaction module includes a real - time monitoring unit and an interaction unit. The real - time monitoring unit provides the real - time data of the transmission line and real - time images collected by the data acquisition module. The interaction unit is used for operation and maintenance personnel to automatically optimize the alarm trigger conditions according to parameters such as the line operation years and material fatigue degree and correct the prediction results of the prediction model.

[0045] The design of the interaction unit enables operation and maintenance personnel to dynamically adjust functions based on the thresholds of equipment health indicators such as material fatigue parameters and operation years.

[0046] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non - exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article or device.

[0047] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principle and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A monitoring system for galloping of a transmission tower transmission line, characterized in that, It includes: data acquisition module, data processing module, intelligent prediction module, early warning response module, and human-computer interaction module; The data acquisition module is used to collect three-dimensional meteorological data within the line corridor in real time, including parameters such as wind speed, wind direction, temperature, humidity, and precipitation. At the same time, it integrates the meteorological radar data interface to obtain regional meteorological forecast information for the next 3 hours and build a micro-meteorological monitoring network; The data processing module is responsible for processing, feature extraction and pattern recognition of the collected data; The intelligent prediction module is used to predict the dancing trend of the power transmission line in the future; The early warning response module automatically generates early warning prompts according to the intelligent prediction module; The human-computer interaction module provides a real-time data monitoring interface and high-definition image stream. Operation and maintenance personnel can adjust the warning threshold value based on the line aging degree and material fatigue parameters through the human-computer interaction module, and support manual intervention in the output results of the prediction model.

2. The dancing monitoring system for a transmission tower transmission line according to claim 1, characterized in that: The data acquisition module includes a historical data storage unit, a meteorological parameter monitoring unit, an image acquisition unit and a line status perception unit, wherein the historical data storage unit is used to build a distributed database cluster, classify and store historical dancing event data, equipment maintenance records, and fault handling archives, use a time series database to store high-frequency sampling data, design a data life cycle management strategy, and realize intelligent retrieval and rapid call of historical data; the meteorological parameter monitoring unit is responsible for real-time collection of environmental data such as wind speed, wind direction, temperature and humidity, air pressure, and precipitation intensity; the image acquisition unit uses a multi-spectral imaging device to achieve high-definition video acquisition at 30 frames per second, and cooperates with infrared thermal imaging technology for all-weather real-time monitoring.

3. The dancing monitoring system for a transmission tower transmission line according to claim 2, wherein: The line state sensing unit obtains physical parameters such as conductor amplitude, swing frequency, tension change, tilt angle, etc. through a distributed sensor network, and the monitoring accuracy is controlled within the range of ±0.5 degrees. At the same time, a line dynamic characteristics database is established to record the natural vibration modes of conductors in different sections.

4. A power transmission tower power transmission line galloping monitoring system according to claim 1, characterized in that: The data processing module includes a data cleaning unit, a feature extraction and fusion unit, and a pattern recognition unit, wherein the data cleaning unit uses an outlier detection algorithm to remove sensor false alarm data, eliminates random noise interference through a sliding window mean method, and implements spatiotemporal correlation compensation for missing data to ensure the integrity and reliability of input data; The extraction and fusion unit extracts time-frequency domain features through Fourier analysis, aligns meteorological parameters, mechanical parameters, and geographic information in time and space, builds a feature engineering processing pipeline, extracts key influencing factor combinations, and generates a standardized data set of multi-dimensional feature vectors; the pattern recognition unit uses wavelet transform to analyze vibration spectrum characteristics and identify typical dancing patterns, including vertical oscillation, horizontal swing, and torsional deformation; a line dynamic behavior classification model is established to realize automatic recognition and pattern coding of different dancing forms.

5. The dancing monitoring system for a transmission tower transmission line according to claim 1, characterized in that: The intelligent prediction module includes a model training unit, a real-time prediction unit, and a result optimization unit. The model training unit is used to train the intelligent prediction model. The real-time prediction unit generates the future 3-hour galloping risk probability distribution through the intelligent prediction model and outputs key indicators such as the maximum swing prediction value, the dangerous duration, and the energy accumulation index. The result optimization unit applies the fuzzy logic algorithm to perform confidence weighting on the multi-model prediction results and establishes a prediction error compensation mechanism.

6. The dancing monitoring system for a transmission tower transmission line according to claim 1, wherein: The model training unit first extracts the historical complete monitoring record data from the historical data storage unit, constructs a training data set containing multiple feature dimensions, and uses the sliding window technique to generate time series samples. The window length is set to 3 hours, and the sliding step is set to 30 minutes. At the same time, in the data annotation stage, line dynamics constraints are introduced: the maximum swing of the wire does not exceed 1 / 20 of the span, the galloping frequency is negatively correlated with the wire tension, and there is an exponential decay relationship between the ice thickness and the critical wind speed, so as to establish a constraint rule library and automatically filter abnormal data samples that violate the physical laws. During the prediction model training process, the loss function is designed as the MSE loss + the physical consistency loss term. The physical term includes the wire motion equation residual: λ(∂²y / ∂t² - (T / m)∂²y / ∂x² + c∂y / ∂t)^2, where λ is the constraint intensity coefficient, set to 0.

3. When the predicted swing exceeds 1.2 times the historical maximum value, the regularization intensity is automatically enhanced, and at the same time, anti-aircraft dynamic adversarial samples are generated during the training process to improve the robustness of the model.

7. A dancing monitoring system for a transmission tower and transmission line according to claim 1, characterized in that: The intelligent prediction model trained by the model training unit adopts a three-level prediction framework. The basic layer is an LSTM network to process time series features. The middle layer is a random forest regressor to fuse static parameters. The output layer is a physical equation constraint layer. At the same time, offline training is carried out using historical data. Similar scenarios are retrieved from the historical data storage unit, and the prediction model is used for prediction under the condition of ensuring the consistency of the time dimension, space dimension, and environment dimension. The prediction results of the prediction model are reviewed with the historical data, and the hyperparameter configuration is optimized through cross-validation to form an intelligent prediction model suitable for different climate zones.

8. A power transmission tower power transmission line galloping monitoring system according to claim 1, characterized in that: The early warning of the early warning response module is divided into four levels: blue, yellow, orange, and red. According to the prediction results of the intelligent prediction model, early warning prompts are given, and response measures suggestions corresponding to the levels are generated. Among them, the blue early warning is that the maximum swing amplitude ≤ 1.50% of the span, the energy index < 300 kJ / m, and the wind speed ≤ 5 m / s. It is in the initial vibration stage, and the response measure is to automatically generate a monitoring report and self-check the status of the sensor network at the same time; the yellow early warning is that 1.50% < swing amplitude ≤ 3.00% of the span, 300 ≤ energy index < 800, and 5 m / s < wind speed ≤ 10 m / s. It is in the medium-amplitude motion, and the response measure is to increase the sensor sampling frequency and prepare the de-icing device to standby; the orange early warning is that 3.00% < swing amplitude ≤ 5.00% of the span, 800 ≤ energy index < 1500, and 10 m / s < wind speed ≤ 15 m / s. It is in the multi-modal composite vibration, and the response measures are to start the de-icing device, transfer the line load to the standby line, have the engineering repair team on standby at the scene, and restrict the approach of surrounding vehicles and personnel; the red early warning is that the swing amplitude > 5.00% of the span, the energy index ≥ 1500 kJ / m, and the wind speed > 15 m / s. It is in the three-dimensional violent galloping state, and the response measures are to forcibly cut off the line power supply, evacuate the surrounding personnel, and start the vibration suppression device.

9. The dancing monitoring system for a transmission tower transmission line according to claim 1, wherein: The human-computer interaction module includes a real-time monitoring unit and an interaction unit. The real-time monitoring unit provides the real-time data of the transmission line collected by the data acquisition module and real-time images. The interaction unit is used for the operation and maintenance personnel to automatically optimize the alarm trigger conditions according to parameters such as the line operation years and material fatigue degree and correct the prediction results of the prediction model.

Images