Power transmission line galloping early warning method and device, computer equipment and storage medium

Abstract

The invention relates to the technical field of power transmission line state monitoring, and discloses a power transmission line galloping early warning method and device, computer equipment and a storage medium, and the method comprises the steps: obtaining meteorological monitoring data covering a power transmission line region; galloping physical criterion features are extracted based on meteorological monitoring data; training a preset deep learning model based on the meteorological monitoring data and the galloping physical criterion features to obtain a power transmission line galloping early warning model; and inputting the actual meteorological monitoring data into the power transmission line galloping early warning model to obtain an original galloping prediction result, and carrying out early warning on the power transmission line galloping risk based on the original galloping prediction result. And the method is not suitable for a large-range and multi-scene power transmission line galloping risk identification task.

Description

Technical Field

[0001] This invention relates to the field of power transmission line condition monitoring technology, specifically to a method, device, computer equipment, and storage medium for early warning of power transmission line galloping. Background Technology

[0002] As a critical infrastructure of the power system, the safe and stable operation of transmission lines directly affects the reliability of power transmission and the security of the power supply system. During winter or transitional seasons, transmission lines often experience conductor galloping under complex weather conditions. Galloping refers to the low-frequency, large-amplitude, quasi-periodic self-excited oscillations generated by icy conductors under crosswinds. Once this phenomenon occurs, it can cause conductor contact, tripping, insulator damage, and hardware fatigue; in severe cases, it can even lead to tower collapse and widespread power outages, making it one of the significant disaster risks in power system operation.

[0003] The physical causes of galloping are relatively clear. It is mainly caused by icing on the conductor, leading to cross-sectional asymmetry, which, under the influence of lateral wind, creates unstable aerodynamic forces, thus inducing self-excited vibrations in the structure. The classic Den Hartog criterion states that when the derivative of the conductor lift with respect to the angle of attack is negative and the drag coefficient is large, the system will exhibit negative damping, resulting in galloping. This theory has a wide influence in wind engineering and aeroelasticity research and is considered the core criterion for galloping. However, the Den Hartog criterion is difficult to apply directly in practical engineering. Its key parameters, such as the lift coefficient and the derivative of the angle of attack, are usually obtained through wind tunnel tests or CFD simulations, and are difficult to acquire in real time through conventional monitoring methods. Therefore, galloping judgment is mostly based on empirical rules or subjective analysis, lacking a unified, quantifiable, and generalizable technical approach.

[0004] In traditional practice, gooseneck early warning methods mainly rely on a combination of meteorological condition threshold judgments and manual inspections, such as setting empirical thresholds for variables like wind speed, temperature, and humidity for prediction. This method is highly dependent on local experience, making it difficult to adapt to widespread deployment across multiple regions and meteorological conditions, and it cannot effectively characterize the coupling relationships between different meteorological factors. In recent years, with the improvement of meteorological monitoring and intelligent operation and maintenance levels, data-driven methods such as deep learning have begun to be introduced into the field of gooseneck prediction. These methods improve the recognition accuracy to a certain extent by constructing prediction models (such as LSTM, Transformer, etc.) based on multi-variable meteorological sequences such as wind speed, wind direction, temperature, humidity, and precipitation, and can achieve automatic model updates and batch deployment.

[0005] However, purely data-driven methods have significant limitations. On the one hand, conductor galloping samples are extremely scarce in monitoring data, often several orders of magnitude lower than non-galloping samples, which can easily lead to model bias towards the main class and difficulty in identifying real galloping events. On the other hand, these methods often ignore the physical formation mechanism of galloping, resulting in poor model interpretability and difficulty in providing clear physical explanations for the results, thus limiting their credibility and practicality in engineering scenarios such as scheduling control and emergency response. Summary of the Invention

[0006] In view of this, the present invention provides a method, device, computer equipment and storage medium for early warning of transmission line galloping, in order to solve the problem that the early warning model of transmission line galloping in the prior art has poor interpretability and cannot be applied to the task of identifying transmission line galloping risks in a wide range and multiple scenarios.

[0007] In a first aspect, the present invention provides a method for early warning of transmission line galloping, the method comprising: Acquire meteorological monitoring data covering the area of ​​power transmission lines; Extracting physiological characteristics of dancing animals based on meteorological monitoring data; A pre-set deep learning model was trained based on meteorological monitoring data and physical criteria for power line galloping to obtain a power line galloping early warning model. Actual meteorological monitoring data is input into the transmission line galloping early warning model to obtain the original galloping prediction results, and the risk of transmission line galloping is warned based on the original galloping prediction results.

[0008] This invention provides a method for early warning of power transmission line galloping. By acquiring meteorological monitoring data covering the transmission line area, it ensures that the data is directly related to the actual environment of the transmission line, avoiding the problem of generalized meteorological data being disconnected from the actual operating conditions of the line. Simultaneously, it extracts physical evidence features of galloping based on meteorological monitoring data, focusing on the core physical influencing factors of galloping, screening key features from a principle level, reducing irrelevant data interference, and providing high-quality input for subsequent model training, thereby improving the accuracy of the early warning results. The method employs a joint training of a deep learning model using meteorological monitoring data and galloping physical evidence features, balancing the data-driven pattern learning capability with the interpretability of physical mechanisms, avoiding the data noise fitting or overfitting problems that may occur with purely data-driven models. This results in an early warning model with stronger generalization ability and higher reliability under different lines and meteorological conditions. By inputting actual meteorological data into the early warning model, the original galloping prediction results can be obtained in advance, and early warnings of galloping risks can be issued based on this. This process transforms traditional post-event handling into pre-event early warning, giving maintenance personnel ample time to take preventative measures. It effectively reduces the risk of accidents such as line tripping, line breakage, and tower damage caused by galloping, ensuring the safe and stable operation of transmission lines. It is applicable to large-scale, multi-scenario transmission line galloping risk identification tasks and solves the problem that existing transmission line galloping early warning models have poor interpretability and cannot be applied to large-scale, multi-scenario transmission line galloping risk identification tasks.

[0009] In an alternative implementation, before extracting the physical criteria features of the dance based on meteorological monitoring data, the method further includes: Standardized meteorological characteristic data are obtained by performing time alignment and data fusion of meteorological monitoring data, imputation of missing values, removal of outliers and normalization of data preprocessing.

[0010] This invention provides a method for early warning of power transmission line galloping. Through data preprocessing such as time alignment, missing value imputation, outlier removal, and normalization, it can ensure the consistency, integrity, and accuracy of meteorological data, unify feature dimensions, provide a high-quality data foundation for subsequent physical criterion feature extraction and model training, reduce the interference of data noise and bias on the early warning results, and improve the reliability of early warning.

[0011] In one alternative implementation, the meteorological monitoring data includes wind speed, wind direction, route alignment, and meteorological factors; Physiological criteria features of dancing animals were extracted based on meteorological monitoring data, including: Wind deflection angle is extracted based on wind direction and route orientation; The wind speed is decomposed to obtain the lateral wind speed vector; An icing coefficient was constructed based on meteorological factors. A comprehensive galloping coefficient is constructed based on wind deflection angle, lateral wind speed vector, and icing coefficient. Wind deflection angle, lateral wind speed component, icing coefficient, and comprehensive dancing coefficient are used as physical criteria for dancing behavior.

[0012] This invention provides a method for early warning of power transmission line galloping. By selectively extracting wind yaw angle, lateral wind speed component, icing coefficient, and comprehensive galloping coefficient as physical criteria features, it transforms raw meteorological data into core features that closely align with the galloping mechanism: the wind yaw angle and lateral wind speed component accurately capture the direction and intensity of wind forces inducing galloping; the icing coefficient focuses on the key trigger of low-temperature icing; and the comprehensive galloping coefficient achieves the synergistic quantification of multiple physical factors. These features not only strengthen the correlation between data and galloping risk, reflecting the physical triggers of galloping, but also provide interpretable physical evidence for the model, reducing redundant information interference and effectively improving the feature learning efficiency and early warning accuracy of subsequent early warning models.

[0013] In one optional implementation, the preset deep learning model includes a first sub-channel, a second sub-channel, a two-layer fully connected network corresponding to the first sub-channel, an independent fully connected network corresponding to the second sub-channel, a fusion layer, a shared feature learning channel, a multilayer perceptron, and a Dropout regularization layer. A pre-defined deep learning model is trained based on meteorological monitoring data and physical criteria for power line galloping to obtain a power line galloping early warning model, including: Standardized meteorological feature data is input into the first sub-channel and then extracted into high-dimensional meteorological features through a corresponding two-layer fully connected network. The physical criteria features of the dance are input into the second sub-channel, and the dance inducement expression features are extracted through the corresponding independent fully connected network; High-dimensional meteorological features and dance-inducing factor expression features are fused through a fusion layer; The fused features are input into a shared feature learning channel for learning. After processing by a multilayer perceptron and a Dropout regularization layer, the original galloping prediction result is output, and a transmission line galloping early warning model is obtained.

[0014] In one optional implementation, a pre-set deep learning model is trained based on meteorological monitoring data and physical criteria features of power line galloping to obtain a power line galloping early warning model, which further includes: The pre-defined deep learning model is trained using a weighted binary cross-entropy loss function as the optimization objective function. The formula for the weighted binary cross-entropy loss function is as follows: ; ; Where N is the total number of samples of meteorological characteristic data and dance animal physical judgment characteristics. The values ​​0 and 1 represent "no dancing" and "dancing," respectively. These are the weighting coefficients; For category The number of samples.

[0015] This invention provides a method for early warning of transmission line galloping. In terms of model structure design, it proposes a dual-channel deep neural network structure that integrates physical criteria and meteorological time-series features. This network contains two independent input channels: one channel receives physical indicators constructed based on galloping theory, and the other channel processes the original meteorological variable time series. The two channels extract high-level semantic features through independent neural network structures, which are then fused in the middle to form a unified joint feature representation. This representation is then used to predict the galloping probability through a subsequent shared fully connected layer. This structure effectively integrates prior physical information while maintaining the modeling capabilities of deep learning, improving the model's generalization and interpretability, and making it suitable for large-scale, multi-scenario transmission line galloping risk identification tasks.

[0016] In one alternative implementation, after obtaining the original dance prediction result, the method further includes: A correction function is constructed based on the comprehensive goofing coefficients, and the original goofing prediction results are corrected based on the correction function to obtain the final goofing prediction results; the correction function is expressed by the following formula: ; in, This is the original dance prediction result. For the final dance prediction results, To achieve a comprehensive dancing coefficient, Let be the Sigmoid activation function, such that The range is (0,1). This is the preset correction factor.

[0017] This invention provides a method for early warning of power transmission line galloping. Considering the potential for misjudgment and insufficient reliability of deep learning models under highly imbalanced sample conditions, a prediction result correction mechanism based on galloping criterion coefficients is proposed. This mechanism utilizes pre-constructed galloping criterion coefficients (such as those aggregated from indicators like wind deflection angle, crosswind speed, and precipitation pattern) as an external physical reference to post-process the probability values ​​output by the model. Specifically, the model output values ​​and galloping coefficients are fused using a nonlinear mapping function to form the final corrected galloping probability. This significantly enhances the model's recognition sensitivity near boundary values, providing compensatory guidance, especially for situations with low confidence but significant physical risks. This mechanism, as a lightweight structure decoupled from the original model, not only improves the reliability of the early warning model but also enhances the physical interpretability of the results and its auxiliary value for manual review.

[0018] In one optional implementation, early warning of transmission line galloping risk is provided based on galloping prediction results, including: The final galloping prediction result is compared with a preset threshold. If the final galloping prediction result is greater than or equal to the preset threshold, it is determined that there is a risk of transmission line galloping; if the final galloping prediction result is less than the preset threshold, it is determined that there is no risk of transmission line galloping.

[0019] This invention provides a method for early warning of transmission line galloping. By comparing a preset threshold with the final prediction result, risk assessment is performed. This transforms the abstract prediction value output by the model into a clear and intuitive risk conclusion ("risk exists" or "no risk"), providing maintenance personnel with a clear basis for decision-making. This standardized assessment method ensures the consistency and operability of the early warning results, avoids ambiguous interpretations, facilitates rapid identification of high-risk lines and timely initiation of prevention and control measures, and improves the practicality and response efficiency of galloping risk early warning.

[0020] In a second aspect, the present invention provides a transmission line galloping early warning device, the device comprising: The meteorological monitoring data acquisition module is used to acquire meteorological monitoring data covering the area of ​​the power transmission line. The dance animal physiology criterion feature extraction module is used to extract dance animal physiology criterion features based on meteorological monitoring data; The model training module is used to train a preset deep learning model based on meteorological monitoring data and the physical criteria features of power line galloping, so as to obtain a power line galloping early warning model. The galloping risk early warning module is used to input actual meteorological monitoring data into the transmission line galloping early warning model to obtain the original galloping prediction results, and to issue early warnings on the risk of transmission line galloping based on the original galloping prediction results.

[0021] Thirdly, the present invention provides a computer device, comprising: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the transmission line galloping early warning method of the first aspect or any corresponding embodiment described above.

[0022] Fourthly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute the transmission line galloping early warning method of the first aspect or any corresponding embodiment described above.

[0023] Fifthly, the present invention provides a computer program product, including computer instructions, which are used to cause a computer to execute the transmission line galloping early warning method of the first aspect or any corresponding embodiment described above. Attached Figure Description

[0024] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0025] Figure 1 This is a flowchart illustrating a transmission line galloping early warning method according to an embodiment of the present invention; Figure 2 This is a flowchart illustrating another transmission line galloping early warning method according to an embodiment of the present invention; Figure 3 This is a flowchart illustrating another transmission line galloping early warning method according to an embodiment of the present invention; Figure 4 This is a structural block diagram of a power transmission line galloping early warning device according to an embodiment of the present invention; Figure 5 This is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. Detailed Implementation

[0026] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] Currently, some studies have begun to introduce physical mechanisms and prior knowledge to assist deep learning modeling, transforming wind deflection angle, wind speed components, and icing coefficients in the galloping process into physical criterion features to enhance the model's ability to identify galloping boundary conditions. These methods extract physically meaningful prior information from meteorological data and construct derived features such as the "galloping coefficient" as model input or post-processing basis, showing promising development prospects. However, a systematic and universal mechanism for fusing galloping physical criteria with deep learning models is still lacking, particularly in feature engineering, structural design, sample imbalance handling, and posterior correction, where significant room for improvement remains. Therefore, there is an urgent need to propose a galloping early warning technology scheme that integrates physical mechanisms and deep neural networks to enhance the model's discriminative ability and interpretability, adapting to practical application needs under complex meteorological conditions. This invention provides a transmission line galloping early warning method that integrates physical mechanisms and deep neural networks to form a galloping early warning technology scheme, enhancing the model's discriminative ability and interpretability to adapt to practical application needs under complex meteorological conditions.

[0028] According to an embodiment of the present invention, a method for early warning of transmission line galloping is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0029] This embodiment provides a method for early warning of transmission line galloping, which can be used in the aforementioned transmission line galloping early warning device. Figure 1 This is a flowchart of a transmission line galloping early warning method according to an embodiment of the present invention, such as... Figure 1 As shown, the process includes the following steps: Step S101: Obtain meteorological monitoring data covering the area of ​​the power transmission line.

[0030] Specifically, acquiring meteorological monitoring data covering the transmission line area aims to provide high-quality input feature data for the subsequent construction of physical criteria features and deep learning prediction models. For example, meteorological monitoring data involves core meteorological data such as wind speed, wind direction, line alignment, temperature, humidity, and precipitation.

[0031] Ground-based meteorological stations and automatic meteorological monitoring terminals are installed along key sections of transmission lines (such as areas prone to vibration and near towers) to collect basic meteorological parameters such as wind speed, wind direction, temperature, humidity, and precipitation in real time. This data is then combined with publicly available data from regional meteorological departments (such as meteorological station network data, satellite remote sensing data, and radar monitoring data) to supplement meteorological information over a wide area around the transmission lines, thus expanding the coverage.

[0032] It can also transmit real-time data from each monitoring point to the data center via wireless communication (such as 4G / 5G, LoRa) or wired networks, while integrating historical meteorological databases to form a complete meteorological dataset covering the line area, providing raw data support for subsequent analysis.

[0033] This embodiment selects meteorological monitoring data covering the transmission line area over the past five years. The core meteorological variables selected are shown in Table 1 below: Table 1 Core meteorological variables

[0034] Step S102: Extract the physical criteria features of the dancing animals based on meteorological monitoring data.

[0035] Specifically, the physical characteristics of power line galloping refer to key physical indicators extracted from meteorological monitoring data based on the mechanism of power line galloping, which quantify the likelihood of galloping. These characteristics are directly related to the core triggers of galloping (such as wind force and icing conditions) and serve as the physical basis for judging whether a line is at risk of galloping.

[0036] Transmission line galloping is a low-frequency, large-amplitude vibration phenomenon caused by aerodynamic instability under specific meteorological conditions (such as wind and icing). Its occurrence is directly related to physical factors such as the direction and intensity of wind force and the state of icing on the line. The physical criteria for galloping transform these abstract physical influences into quantifiable indicators, accurately capturing the key conditions for galloping to occur.

[0037] Step S103: Based on meteorological monitoring data and the physical criteria features of power line galloping, a preset deep learning model is trained to obtain a power line galloping early warning model.

[0038] Specifically, based on meteorological monitoring data and physical evidence features of galloping, a deep learning model for predicting the risk of transmission line galloping was designed and trained. This model employs a dual-channel structure, processing the raw meteorological monitoring data input and the physical evidence features of galloping separately, and achieves joint feature learning through a fusion layer to improve the galloping detection capability. This structure effectively integrates prior physical information while maintaining the modeling capabilities of deep learning, enhancing the model's generalization and interpretability, making it suitable for large-scale, multi-scenario transmission line galloping risk identification tasks.

[0039] Step S104: Input the actual meteorological monitoring data into the transmission line galloping early warning model to obtain the original galloping prediction results, and issue an early warning for the risk of transmission line galloping based on the original galloping prediction results.

[0040] Specifically, based on the comparison between the prediction results and the set threshold, the transmission line galloping risk assessment result is output.

[0041] The transmission line galloping early warning method provided in this embodiment acquires meteorological monitoring data covering the transmission line area, ensuring that the data is directly related to the actual environment of the transmission line and avoiding the problem of generalized meteorological data being disconnected from the actual operating conditions of the line. Simultaneously, it extracts galloping physical criteria features based on meteorological monitoring data, focusing on the core physical influencing factors of galloping, screening key features from a principle level, reducing irrelevant data interference, and providing high-quality input for subsequent model training, thereby improving the accuracy of the early warning results. The method employs a joint training of a deep learning model using meteorological monitoring data and galloping physical criteria features, balancing the data-driven pattern learning capability with the interpretability of physical mechanisms, avoiding the data noise fitting or overfitting problems that may occur with purely data-driven models. This results in a more robust and reliable early warning model with stronger generalization ability under different lines and meteorological conditions. By inputting actual meteorological data into the early warning model, the original galloping prediction results can be obtained in advance, and early warnings of galloping risks can be issued based on these predictions. This process transforms traditional post-event handling into pre-event early warning, giving maintenance personnel ample time to take preventative measures. It effectively reduces the risk of accidents such as line tripping, line breakage, and tower damage caused by galloping, ensuring the safe and stable operation of transmission lines. It is applicable to large-scale, multi-scenario transmission line galloping risk identification tasks and solves the problem that existing transmission line galloping early warning models have poor interpretability and cannot be applied to large-scale, multi-scenario transmission line galloping risk identification tasks.

[0042] This embodiment provides a method for early warning of power transmission line galloping, which can be used in power transmission line galloping early warning devices. Figure 2 This is a flowchart of a transmission line galloping early warning method according to an embodiment of the present invention, such as... Figure 2 As shown, the process includes the following steps: Step S201: Obtain meteorological monitoring data covering the area of ​​the transmission line. For details, please refer to [link to relevant documentation]. Figure 1 Step S101 of the illustrated embodiment will not be described again here.

[0043] Step S202 involves preprocessing the meteorological monitoring data by time alignment, missing value imputation, outlier removal, and normalization to obtain standardized meteorological characteristic data.

[0044] Specifically, step S202 includes: 1) Time alignment and data integration: Since different data sources may have different sampling frequencies and timestamps, time alignment is required. Time alignment and data integration include: resampling all meteorological monitoring data to a daily scale; retaining daily timed data (e.g., 9:00 AM); and prioritizing the use of data from the nearest station when merging data from multiple stations. A time-series format of meteorological monitoring data samples is constructed for each transmission line, as shown in Table 2 below: Table 2. Sample of meteorological monitoring data in time series format.

[0045] 2) Missing value imputation: Missing data often occurs in actual meteorological monitoring data collection. The missing value imputation strategy is as follows: If data for a single day is missing, use data from the previous day to fill in the missing data (forward imputation); if a whole segment is missing for more than 3 consecutive days, directly delete the record for that segment; for nonlinear variables such as precipitation and wind speed, interpolation is used to fill in the missing data.

[0046] 3) Outlier removal: A combination of rule-based thresholds and statistical methods is used to determine outliers: data with wind speeds exceeding 60 m / s, temperatures below -50°C or above 50°C, and humidity of 0 or above 100% are considered outliers and are removed.

[0047] (1); Simultaneously, the 3σ principle is used to eliminate extreme values. If the condition is met, the value is considered an extreme outlier and is eliminated. The formula is as follows: (2); in: σ is the average of all meteorological monitoring data samples; σ is the standard deviation of the sample, representing the dispersion of the data. This is a sample of meteorological monitoring data.

[0048] 4) Data Normalization: To improve the training efficiency and convergence speed of deep learning models, continuous variables need to be normalized. Normalization methods are as follows: (3); in, For the normalized data, , These are the minimum and maximum values ​​in the meteorological monitoring data sample, respectively.

[0049] All normalization parameters are computed on the training set and standardized on the test set using the same parameters to ensure consistency.

[0050] Step S203: Extract the physical criteria features of the dancing animals based on meteorological monitoring data.

[0051] Specifically, meteorological monitoring data includes wind speed, wind direction, line alignment, and meteorological factors, including temperature, humidity, and precipitation. Based on the physical mechanism of transmission line galloping, and combined with wind speed, wind direction, and meteorological factors, multiple quantifiable galloping physical criteria features are constructed to enhance the deep model's understanding of the actual galloping mechanism. The above step S203 includes: Step S2031: Extract the wind deflection angle based on wind direction and route orientation.

[0052] Specifically, the wind deflection angle is the angle between the wind direction and the transmission line's orientation, and it is one of the fundamental physical quantities for determining whether galloping has occurred. Galloping often occurs under the influence of crosswinds, therefore, it is necessary to extract this angular characteristic, expressed by the following formula: (4); in, For wind deflection angle, For wind direction, For the route direction, This is a modulo operation, also known as the remainder operation.

[0053] Step S2032: Decompose the wind speed to obtain the lateral wind speed vector.

[0054] Specifically, since only the wind force component perpendicular to the conductor direction will cause galloping, the wind speed needs to be decomposed, retaining only the lateral wind speed component, as expressed by the following formula: (5); in, This represents the lateral wind speed component. This refers to wind speed.

[0055] Step S2033: Construct the icing coefficient based on meteorological factors. The icing coefficient is constructed based on temperature, humidity and precipitation.

[0056] Specifically, since icing is a crucial condition for inducing galloping, this embodiment of the invention constructs an icing coefficient that considers temperature, humidity, and precipitation. This coefficient is non-zero only at low temperatures (e.g., <0°C), and the specific formula is as follows: (6); in, For precipitation, Relative humidity, For temperature.

[0057] Step S2034: Construct a comprehensive galloping coefficient based on wind deflection angle, lateral wind speed vector, and icing coefficient.

[0058] Specifically, this invention integrates multiple physical factors to construct a comprehensive dancing coefficient. It is used to determine the probability of dance occurring, serving as an auxiliary channel input for subsequent deep learning models or a post-processing correction index.

[0059] Comprehensive dancing coefficient The formula is expressed as follows: (7); in, and The dance weighting coefficient is set based on experience.

[0060] Step S2035 uses wind deflection angle, lateral wind speed component, icing coefficient, and comprehensive galloping coefficient as physical criteria for galloping.

[0061] It should be noted that after extracting the physical criteria features of the dance, these features need to be added as new features to each meteorological monitoring data sample record to form a fused new feature, which can be used directly by the subsequent deep learning model or analyzed independently, as shown in Table 3 below: Table 3 New Features of Fusion

[0062] Step S204: Based on meteorological monitoring data and the physical characteristics of power line galloping, a pre-set deep learning model is trained to obtain a power transmission line galloping early warning model. For details, please refer to [link to relevant documentation]. Figure 1 Step S103 of the illustrated embodiment will not be described again here.

[0063] Step S205 involves inputting actual meteorological monitoring data into the transmission line galloping early warning model to obtain the raw galloping prediction results, and then issuing an early warning for transmission line galloping risk based on these results. For details, please refer to [link to relevant documentation]. Figure 1 Step S104 of the illustrated embodiment will not be described again here.

[0064] The transmission line galloping early warning method provided in this embodiment extracts wind yaw angle, lateral wind speed component, icing coefficient, and comprehensive galloping coefficient as physical criteria features. This transforms raw meteorological data into core features that align with the galloping mechanism: the wind yaw angle and lateral wind speed component accurately capture the direction and intensity of wind forces inducing galloping; the icing coefficient focuses on the key trigger of low-temperature icing; and the comprehensive galloping coefficient achieves the synergistic quantification of multiple physical factors. These features not only strengthen the correlation between data and galloping risk, reflecting the physical causes of galloping, but also provide interpretable physical evidence for the model, reducing redundant information interference and effectively improving the feature learning efficiency and early warning accuracy of subsequent early warning models.

[0065] This embodiment provides a method for early warning of power transmission line galloping, which can be used in power transmission line galloping early warning devices. Figure 3 This is a flowchart of a transmission line galloping early warning method according to an embodiment of the present invention, such as... Figure 3 As shown, the process includes the following steps: Step S301: Obtain meteorological monitoring data covering the area of ​​the transmission line. For details, please refer to [link to relevant documentation]. Figure 2 Step S201 of the illustrated embodiment will not be described again here.

[0066] Step S302: Extract physiological criteria features of the dancing animals based on meteorological monitoring data. For details, please refer to [link to relevant documentation]. Figure 2 Step S203 of the illustrated embodiment will not be described again here.

[0067] Step S303: Based on meteorological monitoring data and the physical criteria features of power line galloping, a preset deep learning model is trained to obtain a power line galloping early warning model.

[0068] Specifically, based on preprocessed standard meteorological feature data and physical criteria features for galloping, a deep learning model for predicting the risk of transmission line galloping is designed and trained. This model employs a dual-channel structure, processing the original meteorological input and physical criteria input separately, and achieves joint feature learning through a fusion layer to improve galloping discrimination capability. The pre-defined deep learning model includes a first sub-channel, a second sub-channel, a two-layer fully connected network corresponding to the first sub-channel, an independent fully connected network corresponding to the second sub-channel, a fusion layer, a shared feature learning channel, a multilayer perceptron, and a Dropout regularization layer.

[0069] 1) Model input design: The model's input is divided into two sub-channels: First sub-channel: Raw meteorological data: wind speed, wind direction, route, temperature, humidity, precipitation; Second sub-channel: wind deflection angle, lateral wind speed, icing coefficient, and galloping coefficient.

[0070] 2) Model structure design: In order to fully integrate the original meteorological features and physical criteria features, a dual-channel neural network structure was constructed for the recognition of dancing events.

[0071] One sub-channel receives physical indicators constructed based on galloping theory, while the other processes the original meteorological variable time series. The two sub-channels extract high-level semantic features through independent neural network structures, which are then fused in the middle to form a unified joint feature representation. This representation is then used to predict the galloping probability through a subsequent shared fully connected layer. This structure effectively integrates prior physical information while maintaining the modeling capabilities of deep learning, improving the model's generalization and interpretability. It is suitable for large-scale, multi-scenario transmission line galloping risk identification tasks.

[0072] Step S303 above includes: In step S3031, standardized meteorological feature data is input into the first sub-channel and high-dimensional meteorological features are extracted through the corresponding two-layer fully connected network.

[0073] Specifically, the first sub-channel takes conventional meteorological variables such as wind speed, wind direction, route orientation, temperature, humidity, and precipitation as input, and extracts high-dimensional features through two layers of fully connected networks.

[0074] Step S3032: Input the dance physiology criteria features into the second sub-channel, and extract the dance motive expression features through the corresponding independent fully connected network.

[0075] Specifically, the second sub-channel takes physical characteristics such as wind deflection angle, lateral wind speed, icing coefficient and galloping coefficient as input, and extracts the physical-level galloping causation expression features through an independent fully connected network.

[0076] Step S3033: The high-dimensional meteorological features and the dance-induced features are fused through the fusion layer.

[0077] Specifically, the output features of the two sub-channels are fused by vector concatenation. The specific fusion method can be found in relevant technologies and will not be elaborated here.

[0078] Step S3034: The fused features are input into the shared feature learning channel for learning. After being processed by a multilayer perceptron and a Dropout regularization layer, the original galloping prediction result is output, and the transmission line galloping early warning model is obtained.

[0079] Specifically, the fused features are input into a shared feature learning channel, and after further processing by a multilayer perceptron and a Dropout regularization layer, the probability value of galloping is output, which is the original galloping prediction result. The corresponding deep learning model is then converted into a transmission line galloping early warning model.

[0080] Step S3035: The pre-defined deep learning model is trained using the weighted binary cross-entropy loss function as the optimization objective function.

[0081] Specifically, the model uses the Sigmoid activation function for binary classification, employs class-weighted binary cross-entropy as the loss function, and introduces a sample imbalance handling mechanism to improve sensitivity and recognition ability for the minority class (dancing samples). This includes: Considering that dancing events are a typical example of an extreme minority class problem, with highly imbalanced data (dancing samples are far fewer than non-dancing samples), this method uses a weighted binary cross-entropy loss function as the optimization objective function. By assigning higher loss weights to the dancing category (positive class), the network's bias towards the majority class (non-dancing) is effectively suppressed, improving the ability to identify the minority class. The weight ratio can be automatically calculated based on the total number of samples in each category, or it can be set and adjusted according to actual business needs. Furthermore, to prevent numerical instability and gradient explosion during training, the network uses a sigmoid activation function to output the dancing probability, and dynamically adjusts the learning rate and regularization coefficients during training to improve the model's generalization ability and convergence stability. The weight coefficient formula is expressed as: (8); The formula for the weighted binary cross-entropy loss function constructed based on the weighting coefficients is as follows: (9); Where N is the total number of samples of meteorological characteristic data and dance animal physical judgment characteristics. The values ​​0 and 1 represent "no dancing" and "dancing," respectively. These are the weighting coefficients; For category The number of samples.

[0082] Step S304: Construct a correction function based on the comprehensive dancing coefficient, and correct the original dancing prediction result based on the correction function to obtain the final dancing prediction result.

[0083] Specifically, although deep learning models can achieve goofing prediction by learning complex nonlinear relationships, they may experience missed reports or low-confidence predictions due to limitations such as training sample size, class imbalance, and meteorological uncertainties. To alleviate this problem, the goofing criterion coefficients constructed earlier are used as auxiliary physical quantities. A correction function is introduced to perform a secondary adjustment on the model's output probability, resulting in a corrected output. A coefficient smoothing mechanism is used to construct the correction function. Combining the original model output and the goofing coefficient values, the correction function is expressed by the following formula: (10); in, The original dance prediction result (between 0 and 1). For the final dance prediction results, To achieve a comprehensive dancing coefficient, Let be the Sigmoid activation function, such that The range is (0,1). The preset correction coefficient controls the extent to which the physical criteria enhance the model results (β can be set to 0.3).

[0084] Step S305: Input the actual meteorological monitoring data into the transmission line galloping early warning model to obtain the original galloping prediction results, and issue an early warning for the risk of transmission line galloping based on the original galloping prediction results.

[0085] Specifically, the revised prediction results For the final determination of the risk of dancing, step S305 above includes: Step a: Compare the final galloping prediction result with a preset threshold. If the final galloping prediction result is greater than or equal to the preset threshold, it is determined that there is a risk of transmission line galloping; if the final galloping prediction result is less than the preset threshold, it is determined that there is no risk of transmission line galloping. The formula for determining galloping risk is expressed as follows: (11); in, This is a preset threshold, which can be set according to the actual situation; no specific restrictions are set here.

[0086] The transmission line galloping early warning method provided in this embodiment addresses the potential misjudgments and insufficient reliability of deep learning models under highly imbalanced samples. It proposes a prediction result correction mechanism based on galloping criterion coefficients. This mechanism utilizes pre-constructed galloping criterion coefficients (such as those aggregated from indicators like wind deflection angle, crosswind speed, and precipitation patterns) as an external physical reference to post-process the probability values ​​output by the model. Specifically, the model output values ​​and galloping coefficients are fused using a nonlinear mapping function to form the final corrected galloping probability. This significantly enhances the model's sensitivity near boundary values, providing compensatory guidance, especially for situations with low confidence levels but significant physical risks. This mechanism, as a lightweight structure decoupled from the original model, not only improves the reliability of the early warning model but also enhances the physical interpretability of the results and its auxiliary value for manual review. By comparing the final prediction results with preset thresholds for risk assessment, the abstract prediction values ​​output by the model can be transformed into intuitive and clear risk conclusions ("risk exists" or "no risk"), providing maintenance personnel with a clear basis for decision-making. This standardized judgment method ensures the consistency and operability of the early warning results, avoids ambiguous interpretations, facilitates the rapid identification of high-risk routes and timely activation of prevention and control measures, and improves the practicality and response efficiency of the risk warning system.

[0087] As one or more specific application embodiments of the present invention, the transmission line galloping early warning method provided by the present invention will be further described in detail as follows: Step 1, meteorological data acquisition and preprocessing: This step aims to provide high-quality input feature data for the subsequent construction of the dancing coefficient criterion and deep learning prediction model. Specifically, it includes determining the data source, selecting variables, time alignment, missing value imputation, outlier removal, and normalization.

[0088] 1) Select meteorological monitoring data covering the transmission line area in the past five years. The core meteorological variables selected are shown in Table 1 above.

[0089] 2) Time Alignment and Data Integration: Since the sampling frequencies and timestamps of different data sources may differ, time alignment is required. Methods include: resampling all data to a daily scale; retaining daily timed data (e.g., 9:00 AM); and prioritizing the use of data from the nearest station when merging data from multiple stations. Sample data in time-series format is constructed for each transmission line, as shown in Table 2 above.

[0090] 3) Missing value imputation: Missing data often occurs in actual meteorological data collection. The missing value imputation strategy is as follows: If data for a single day is missing, use data from the previous day to fill in the missing data (forward imputation); if a whole segment is missing for more than 3 consecutive days, directly delete the record for that segment; for nonlinear variables such as precipitation and wind speed, interpolation is used to fill in the missing data.

[0091] 4) Use both rule thresholds and statistical methods to judge anomalies: data with wind speeds exceeding 60m / s, temperatures below -50°C or above 50°C, and humidity of 0 or above 100% are considered outliers and are removed.

[0092] (1); Simultaneously, the 3σ principle is used to eliminate extreme values. If the condition is met, the value is considered an extreme outlier and is eliminated. The formula is as follows: (2); in: σ is the average of all meteorological monitoring data samples; σ is the standard deviation of the sample, representing the dispersion of the data. This is a sample of meteorological monitoring data.

[0093] 5) Data Normalization: To improve the training efficiency and convergence speed of deep learning models, continuous variables need to be normalized. Normalization methods are as follows: (3); in, For the normalized data, , These are the minimum and maximum values ​​in the meteorological monitoring data sample, respectively.

[0094] Step 2: Extraction and construction of dance physiology criteria features: This step is based on the physical mechanism of transmission line galloping, combined with wind speed, wind direction and meteorological factors, to construct multiple quantifiable dance physiology criteria features to enhance the deep model's ability to understand the actual dance mechanism.

[0095] 1) Wind deflection angle: The wind deflection angle is the angle between the wind direction and the transmission line's orientation, and is one of the fundamental physical quantities for determining whether galloping has occurred. Galloping often occurs under the influence of crosswinds, therefore, it is necessary to extract this angular characteristic. The formula is as follows: (4); in, For wind deflection angle, For wind direction, For the route direction, This is a modulo operation, also known as the remainder operation.

[0096] 2) Lateral wind speed component: Since only the wind force component perpendicular to the conductor direction will cause galloping, the wind speed needs to be decomposed, retaining only the lateral wind speed component. The formula is as follows: (5); in, This represents the lateral wind speed component. This refers to wind speed.

[0097] 3) Icing Coefficient: Since icing is a crucial factor inducing galloping, this embodiment of the invention constructs an icing coefficient that considers temperature, humidity, and precipitation. This coefficient is non-zero only at low temperatures (e.g., <0°C), and the specific formula is as follows: (6); in, For precipitation, Relative humidity, For temperature.

[0098] 4) Dancing Coefficient: This invention integrates multiple physical factors to construct a comprehensive dancing coefficient. It is used to determine the probability of dance occurring, serving as an auxiliary channel input for subsequent deep learning models or a post-processing correction index.

[0099] Comprehensive dancing coefficient The formula is expressed as follows: (7); in, and The dance weighting coefficient is set based on experience.

[0100] 5) Sample construction with new features: After the criterion is constructed, it needs to be added as a new feature to each sample record of the processed meteorological monitoring data for direct use by the model or independent analysis. The new features after fusion are shown in Table 3 above.

[0101] Step 3: Building and training a deep learning model that integrates physical criteria: Based on preprocessed meteorological and physical criterion features, a deep learning model for predicting transmission line galloping risk was designed and trained. The model employs a dual-channel structure, processing the original meteorological input and the physical criterion input separately, and achieves joint feature learning through a fusion layer to improve galloping detection capability.

[0102] 1) Model input design: The model's input is divided into two sub-channels: Channel 1: Raw meteorological data: wind speed, wind direction, wind path, temperature, humidity, precipitation; Channel 2: Wind deflection angle, crosswind speed, icing coefficient, galloping coefficient; 2) Model Structure Design: To fully integrate original meteorological features and physical criteria features, a dual-channel neural network structure was constructed for gooseing event recognition. One channel takes conventional meteorological variables such as wind speed, wind direction, route alignment, temperature, humidity, and precipitation as input, and extracts high-dimensional features through two fully connected layers. The other channel takes physical features such as wind deflection angle, lateral wind speed, icing coefficient, and gooseing coefficient as input, and extracts the physical causes of gooseing through an independent fully connected network. The output features of the two channels are fused by vector concatenation and enter a shared feature learning channel. After further processing by a multilayer perceptron and Dropout regularization layer, the probability value of gooseing is output. The model uses the Sigmoid activation function for binary classification, the loss function is class-weighted binary cross-entropy, and a sample imbalance handling mechanism is introduced to improve the sensitivity and recognition ability of the minority class (gooseing samples).

[0103] 3) Loss Function: Considering that dancing events are a typical extreme minority class problem with highly imbalanced data (dancing samples are far fewer than non-dancing samples), this method uses a weighted binary cross-entropy loss function as the optimization objective function. By assigning higher loss weights to the dancing category (positive class), the network's bias towards the majority class (non-dancing) is effectively suppressed, improving the ability to identify the minority class. The weight ratio can be automatically calculated based on the total number of samples in each category, or it can be set and adjusted according to actual business needs. Furthermore, to prevent numerical instability and gradient explosion during training, the network uses the sigmoid activation function to output the dancing probability, and dynamically adjusts the learning rate and regularization coefficient during training to improve the model's generalization ability and convergence stability. The weight coefficient formula is expressed as: (8); The formula for the weighted binary cross-entropy loss function constructed based on the weighting coefficients is as follows: (9); Where N is the total number of samples of meteorological characteristic data and dance animal physical judgment characteristics. The values ​​0 and 1 represent "no dancing" and "dancing," respectively. These are the weighting coefficients; For category The number of samples.

[0104] Step 4: Model Correction Mechanism Guided by Dancing Coefficients: Although deep learning models can achieve dancing prediction by learning complex nonlinear relationships, they may experience missed reports or low-confidence predictions due to limitations such as training sample size, class imbalance, and meteorological uncertainties. To alleviate this problem, the dancing criterion coefficients constructed above are used as auxiliary physical quantities, and a correction function is introduced to perform a secondary adjustment on the model's output probability, resulting in a corrected output.

[0105] 1) Construction of the correction function: The correction function is constructed using a coefficient smoothing mechanism. Combining the original output of the model with the dancing coefficient values, the correction function is expressed by the following formula: (10); in, The original dance prediction result (between 0 and 1). For the final dance prediction results, To achieve a comprehensive dancing coefficient, Let be the Sigmoid activation function, such that The range is (0,1). The preset correction coefficient controls the extent to which the physical criteria enhance the model results (β can be set to 0.3).

[0106] 2) Dancing judgment strategy: Corrected prediction results The formula used for the final determination of goofing risk is expressed as follows: (11); in, This is a preset threshold, which can be set according to the actual situation; no specific restrictions are set here.

[0107] The transmission line galloping early warning method provided in this embodiment has the following innovative features: 1) Meteorological feature construction method based on dance physiology criteria: To address the issue that traditional deep learning models rely on raw meteorological variables and lack physical mechanism constraints, a feature construction strategy integrating the physical mechanisms of icing is proposed. By introducing typical icing theories such as the Den Hartog criterion (one of the classic theories explaining the icing mechanism of transmission lines), raw meteorological data such as wind speed, wind direction, temperature, humidity, and precipitation are processed to construct criterion-based feature variables with clear physical meanings, such as wind deflection parameters, lateral wind speed, icing coefficient, and icing coefficient. These features not only preserve the dynamic changes of the input meteorological variables but also reflect the physical causes of icing, providing a more discriminative and interpretable input data foundation for subsequent model training.

[0108] (ii) A dual-channel neural network structure integrating physical indicators and deep representations of meteorological time series: In terms of model structure design, a dual-channel deep neural network structure integrating physical criteria and meteorological time-series features is proposed. This network contains two independent input channels: one channel receives physical indicators constructed based on galloping theory, and the other processes the original meteorological variable time series. The two channels extract high-level semantic features through independent neural network structures, which are then fused in the middle to form a unified joint feature representation. This representation is then used to predict the galloping probability through a subsequent shared fully connected layer. This structure effectively integrates prior physical information while maintaining the modeling capabilities of deep learning, improving the model's generalization and interpretability, and making it suitable for large-scale, multi-scenario transmission line galloping risk identification tasks.

[0109] (iii) Model correction mechanism guided by dancing coefficients: To address the potential for misjudgments and insufficient reliability in deep learning models under highly imbalanced sample conditions, a prediction result correction mechanism based on galloping criterion coefficients is proposed. This mechanism utilizes pre-constructed galloping criterion coefficients (aggregated from indicators such as wind deflection angle, crosswind speed, and precipitation pattern) as an external physical reference to post-process the probability values ​​output by the model. Specifically, the model output values ​​and galloping coefficients are fused using a nonlinear mapping function to form the final corrected galloping probability. This significantly enhances the model's sensitivity near boundary values, providing compensatory guidance, especially for cases with low confidence but significant physical risks. As a lightweight structure decoupled from the original model, this mechanism not only improves the reliability of the early warning model but also enhances the physical interpretability of the results and its value for manual review.

[0110] This embodiment also provides a transmission line galloping early warning device, which is used to implement the above embodiments and preferred embodiments, and will not be repeated as already described. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0111] This embodiment provides a power transmission line galloping early warning device, such as... Figure 4 As shown, it includes: Meteorological monitoring data acquisition module 401 is used to acquire meteorological monitoring data covering the area of ​​the power transmission line.

[0112] The dance animal physiology criteria feature extraction module 402 is used to extract dance animal physiology criteria features based on meteorological monitoring data.

[0113] The model training module 403 is used to train a preset deep learning model based on meteorological monitoring data and the physical criteria features of power line galloping to obtain a power line galloping early warning model.

[0114] The galloping risk early warning module 404 is used to input actual meteorological monitoring data into the transmission line galloping early warning model to obtain the original galloping prediction results, and to provide early warning of transmission line galloping risk based on the original galloping prediction results.

[0115] In some alternative implementations, the transmission line galloping early warning device further includes: The data preprocessing module is used to perform time alignment and data fusion, missing value imputation, outlier removal and normalization of the meteorological monitoring data before extracting the physical criteria features based on the meteorological monitoring data, so as to obtain standardized meteorological feature data.

[0116] In some optional implementations, meteorological monitoring data includes wind speed, wind direction, line direction, and meteorological factors; the dance physiology feature extraction module 402 includes: The wind deflection angle extraction unit is used to extract the wind deflection angle based on wind direction and route alignment.

[0117] The wind speed decomposition unit is used to decompose wind speed to obtain the lateral wind speed vector.

[0118] The icing coefficient construction unit is used to construct the icing coefficient based on meteorological factors.

[0119] The integrated galloping coefficient construction unit is used to construct the integrated galloping coefficient based on wind deflection angle, lateral wind speed vector and icing coefficient.

[0120] The dancing behavior criterion feature determination unit is used to use wind deflection angle, lateral wind speed component, icing coefficient and comprehensive dancing coefficient as dancing behavior criterion features.

[0121] In some optional implementations, the preset deep learning model includes a first sub-channel, a second sub-channel, a two-layer fully connected network corresponding to the first sub-channel, an independent fully connected network corresponding to the second sub-channel, a fusion layer, a shared feature learning channel, a multilayer perceptron, and a Dropout regularization layer; the model training module 403 includes: The first sub-channel processing unit is used to input standardized meteorological feature data into the first sub-channel and extract high-dimensional meteorological features through the corresponding two-layer fully connected network.

[0122] The second sub-channel processing unit is used to input the dance physiology criteria features into the second sub-channel and extract the dance motive expression features through the corresponding independent fully connected network.

[0123] The feature fusion unit is used to fuse high-dimensional meteorological features and dance-induced features through the fusion layer.

[0124] The prediction unit is used to input the fused features into the shared feature learning channel for learning. After learning, the features are processed by a multilayer perceptron and a Dropout regularization layer, and the original galloping prediction result is output, thus obtaining the transmission line galloping early warning model.

[0125] In some alternative implementations, the model training module 403 further includes: The loss function determination unit is used to train a predefined deep learning model using a weighted binary cross-entropy loss function as the optimization objective function. The formula for the weighted binary cross-entropy loss function is as follows: ; ; Where N is the total number of samples of meteorological characteristic data and dance animal physical judgment characteristics. The values ​​0 and 1 represent "no dancing" and "dancing," respectively. These are the weighting coefficients; For category The number of samples.

[0126] In some alternative implementations, the transmission line galloping early warning device further includes: The correction module is used to construct a correction function based on the comprehensive goof coefficients after obtaining the original goof prediction result, and then correct the original goof prediction result based on the correction function to obtain the final goof prediction result; the correction function is expressed by the following formula: ; in, This is the original dance prediction result. For the final dance prediction results, To achieve a comprehensive dancing coefficient, Let be the Sigmoid activation function, such that The range is (0,1). This is the preset correction factor.

[0127] In some optional implementations, the dancing risk warning module 404 includes: The galloping risk assessment unit is used to compare the final galloping prediction result with a preset threshold. If the final galloping prediction result is greater than or equal to the preset threshold, it is determined that there is a risk of transmission line galloping; if the final galloping prediction result is less than the preset threshold, it is determined that there is no risk of transmission line galloping.

[0128] Further functional descriptions of the above modules and units are the same as those in the corresponding embodiments described above, and will not be repeated here.

[0129] In this embodiment, the transmission line galloping early warning device is presented in the form of a functional unit. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0130] This invention also provides a computer device having the above-described features. Figure 4 The image shows a power line galloping early warning device.

[0131] Please see Figure 5 , Figure 5 This is a schematic diagram of the structure of a computer device provided in an optional embodiment of the present invention, such as... Figure 5 As shown, the computer device includes one or more processors 10, memory 20, and interfaces for connecting the components, including high-speed interfaces and low-speed interfaces. The components communicate with each other via different buses and can be mounted on a common motherboard or otherwise installed as needed. The processors can process instructions executed within the computer device, including instructions stored in or on memory to display graphical information of a GUI on external input / output devices (such as display devices coupled to the interfaces). In some alternative implementations, multiple processors and / or multiple buses can be used with multiple memories and multiple memory modules, if desired. Similarly, multiple computer devices can be connected, each providing some of the necessary operations (e.g., as a server array, a group of blade servers, or a multiprocessor system). Figure 5 Take a processor 10 as an example.

[0132] Processor 10 may be a central processing unit, a network processor, or a combination thereof. Processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit (ASIC), a programmable logic device (PLD), or a combination thereof. The programmable logic device may be a complex programmable logic device (CAMP), a field-programmable gate array (FPGA), a general-purpose array logic (GDA), or any combination thereof.

[0133] The memory 20 stores instructions executable by at least one processor 10 to cause the at least one processor 10 to perform the method shown in the above embodiments.

[0134] The memory 20 may include a program storage area and a data storage area. The program storage area may store the operating system and applications required for at least one function; the data storage area may store data created based on the use of the computer device. Furthermore, the memory 20 may include high-speed random access memory and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, the memory 20 may optionally include memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0135] The memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk or solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0136] The computer device also includes an input device 30 and an output device 40. The processor 10, memory 20, input device 30, and output device 40 can be connected via a bus or other means. Figure 5 Taking the example of a connection between China and Israel via a bus.

[0137] Input device 30 can receive input numerical or character information, and generate key signal inputs related to user settings and function control of the computer device, such as a touchscreen, keypad, mouse, trackpad, touchpad, joystick, one or more mouse buttons, trackball, joystick, etc. Output device 40 may include display devices, auxiliary lighting devices (e.g., LEDs), and haptic feedback devices (e.g., vibration motors). The aforementioned display devices include, but are not limited to, liquid crystal displays, light-emitting diodes, displays, and plasma displays. In some alternative embodiments, the display device may be a touchscreen.

[0138] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code, which, when accessed and executed by the computer, processor, or hardware, implements the methods shown in the above embodiments.

[0139] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0140] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for early warning of transmission line galloping, characterized in that, include: Acquire meteorological monitoring data covering the area of ​​power transmission lines; Based on the meteorological monitoring data, physical criteria features of the dancing animals were extracted; Based on the meteorological monitoring data and the physical criteria features of power line galloping, a preset deep learning model is trained to obtain a power line galloping early warning model. Actual meteorological monitoring data is input into the transmission line galloping early warning model to obtain the original galloping prediction results, and the risk of transmission line galloping is warned based on the original galloping prediction results.

2. The method according to claim 1, characterized in that, Before extracting the physical criteria features of the dancing animals based on the meteorological monitoring data, the process also includes: Standardized meteorological characteristic data are obtained by performing time alignment and data fusion, missing value imputation, outlier removal and normalization on the meteorological monitoring data.

3. The method according to claim 1, characterized in that, The meteorological monitoring data includes wind speed, wind direction, route alignment, and meteorological factors; The extraction of physiological criteria features based on the meteorological monitoring data includes: The wind deflection angle is extracted based on the wind direction and the route orientation. The wind speed is decomposed to obtain the lateral wind speed vector; An icing coefficient was constructed based on meteorological factors. A comprehensive galloping coefficient is constructed based on wind deflection angle, lateral wind speed vector, and icing coefficient. Wind deflection angle, lateral wind speed component, icing coefficient, and comprehensive dancing coefficient are used as physical criteria for dancing behavior.

4. The method according to claim 2, characterized in that, The preset deep learning model includes a first sub-channel, a second sub-channel, a two-layer fully connected network corresponding to the first sub-channel, an independent fully connected network corresponding to the second sub-channel, a fusion layer, a shared feature learning channel, a multilayer perceptron, and a Dropout regularization layer. The method of training a preset deep learning model based on the meteorological monitoring data and the physical criteria features of power line galloping to obtain a power line galloping early warning model includes: The standardized meteorological feature data is input into the first sub-channel and then processed through a corresponding two-layer fully connected network to extract high-dimensional meteorological features. The dance physiology criteria features are input into the second sub-channel and the dance motive expression features are extracted through the corresponding independent fully connected network. The high-dimensional meteorological features and the dance-inducing factor expression features are fused through a fusion layer; The fused features are input into a shared feature learning channel for learning. After being processed by the multilayer perceptron and Dropout regularization layer, the original galloping prediction result is output, and a transmission line galloping early warning model is obtained.

5. The method according to claim 4, characterized in that, The step of training a preset deep learning model based on the meteorological monitoring data and the physical criteria features of power line galloping to obtain a power line galloping early warning model also includes: A weighted binary cross-entropy loss function is used as the optimization objective function to train a pre-defined deep learning model. The formula for the weighted binary cross-entropy loss function is as follows: ； ； Where N is the total number of samples of meteorological characteristic data and dance animal physical judgment characteristics. The values ​​0 and 1 represent "no dancing" and "dancing," respectively. These are the weighting coefficients; For category The number of samples.

6. The method according to claim 3, characterized in that, After obtaining the original dance prediction results, the method further includes: A correction function is constructed based on the comprehensive dancing coefficients, and the original dancing prediction result is corrected based on the correction function to obtain the final dancing prediction result; the correction function is expressed by the following formula: ； in, This is the original dance prediction result. For the final dance prediction results, To achieve a comprehensive dancing coefficient, Let be the Sigmoid activation function, such that The range is (0,1). This is the preset correction factor.

7. The method according to claim 6, characterized in that, Based on the galloping prediction results, early warning of transmission line galloping risk is provided, including: The final galloping prediction result is compared with a preset threshold. If the final galloping prediction result is greater than or equal to the preset threshold, it is determined that there is a risk of transmission line galloping; if the final galloping prediction result is less than the preset threshold, it is determined that there is no risk of transmission line galloping.

8. A transmission line galloping early warning device, characterized in that, The device comprises: The meteorological monitoring data acquisition module is used to acquire meteorological monitoring data covering the area of ​​the power transmission line. The dance animal physiology criteria feature extraction module is used to extract dance animal physiology criteria features based on the meteorological monitoring data; The model training module is used to train a preset deep learning model based on the meteorological monitoring data and the physical criteria features of power line galloping to obtain a power line galloping early warning model. The galloping risk early warning module is used to input actual meteorological monitoring data into the transmission line galloping early warning model to obtain the original galloping prediction results, and to issue an early warning of transmission line galloping risk based on the original galloping prediction results.

9. The apparatus according to claim 8, characterized in that, The device further includes: The data preprocessing module is used to perform time alignment and data fusion, missing value imputation, outlier removal and normalization on the meteorological monitoring data before extracting the physical criteria features based on the meteorological monitoring data, so as to obtain standardized meteorological feature data.

10. The apparatus according to claim 8, characterized in that, Meteorological monitoring data includes wind speed, wind direction, route alignment, and meteorological factors; the dance physiology feature extraction module includes: A wind deflection angle extraction unit is used to extract the wind deflection angle based on the wind direction and the route orientation. A wind speed decomposition unit is used to decompose the wind speed to obtain a lateral wind speed vector; Ice accretion coefficient construction unit, used to construct the ice accretion coefficient based on meteorological factors; A comprehensive galloping coefficient construction unit is used to construct a comprehensive galloping coefficient based on wind deflection angle, lateral wind speed vector, and icing coefficient; The dancing behavior criterion feature determination unit is used to use wind deflection angle, lateral wind speed component, icing coefficient and comprehensive dancing coefficient as dancing behavior criterion features.

11. The apparatus according to claim 9, characterized in that, The preset deep learning model includes a first sub-channel, a second sub-channel, a two-layer fully connected network corresponding to the first sub-channel, an independent fully connected network corresponding to the second sub-channel, a fusion layer, a shared feature learning channel, a multilayer perceptron, and a Dropout regularization layer. The model training module includes: The first sub-channel processing unit is used to input the standardized meteorological feature data into the first sub-channel and extract high-dimensional meteorological features through the corresponding two-layer fully connected network. The second sub-channel processing unit is used to input the dance physiology criteria features into the second sub-channel and extract the dance motive expression features through the corresponding independent fully connected network. The feature fusion unit is used to fuse the high-dimensional meteorological features and the dance-inducing factor expression features through the fusion layer; The prediction unit is used to input the fused features into the shared feature learning channel for learning. After learning, the features are processed by the multilayer perceptron and the Dropout regularization layer, and the original galloping prediction result is output, thus obtaining the transmission line galloping early warning model.

12. The apparatus according to claim 8, characterized in that, The model training module also includes: The loss function determination unit is used to train a preset deep learning model using a weighted binary cross-entropy loss function as the optimization objective function. The formula for the weighted binary cross-entropy loss function is as follows: ； ； Where N is the total number of samples of meteorological characteristic data and dance animal physical judgment characteristics. The values ​​0 and 1 represent "no dancing" and "dancing," respectively. These are the weighting coefficients; For category The number of samples.

13. The apparatus according to claim 10, characterized in that, The device further includes: The correction module is used to construct a correction function based on the comprehensive dance coefficients after obtaining the original dance prediction result, and to correct the original dance prediction result based on the correction function to obtain the final dance prediction result; the correction function is expressed by the following formula: ； in, This is the original dance prediction result. For the final dance prediction results, To achieve a comprehensive dancing coefficient, Let be the Sigmoid activation function, such that The range is (0,1). This is the preset correction factor.

14. The apparatus according to claim 13, characterized in that, The dance risk warning module includes: The galloping risk assessment unit is used to compare the final galloping prediction result with a preset threshold. If the final galloping prediction result is greater than or equal to the preset threshold, it is determined that there is a risk of transmission line galloping; if the final galloping prediction result is less than the preset threshold, it is determined that there is no risk of transmission line galloping.

15. A computer device, characterized in that, include: The system includes a memory and a processor, which are communicatively connected to each other. The memory stores computer instructions, and the processor executes the computer instructions to perform the transmission line galloping early warning method according to any one of claims 1 to 7.

16. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the transmission line galloping early warning method according to any one of claims 1 to 7.

Images