Safety Assessment Method for High-Voltage Transmission Lines Based on Intelligent Algorithms

Abstract

This invention belongs to the field of power system safety technology. It discloses a safety assessment method for high-voltage transmission lines based on intelligent algorithms. The method constructs a database through multi-source data acquisition and fusion processing, utilizes deep learning models to analyze the correlation between meteorological conditions and icing formation mechanisms and explore their evolutionary patterns, and combines the physical characteristics of transmission lines and topographic information to conduct multi-scenario simulations and sensitivity analyses, establishing a risk zoning assessment system for transmission line icing. This method calculates the critical values ​​of line mechanical strength and icing thickness based on a mechanical model, formulates graded early warning threshold standards, constructs a time-series prediction algorithm and a risk propagation model to achieve intelligent early warning, and forms a prevention and control strategy through optimized allocation of anti-icing resources and adaptive generation of de-icing schemes. This invention effectively improves the ability of high-voltage transmission lines to resist icing disasters and enhances the safe and stable operation level of the power grid.

Description

Technical Field

[0001] This invention relates to the field of power system safety technology, and more specifically, to a method for safety assessment of high-voltage transmission lines based on intelligent algorithms. Background Technology

[0002] High-voltage transmission lines, as a key component of the power system, undertake the task of transmitting electricity over long distances and in large quantities. Their safe and stable operation is directly related to national economic development and social order. In southern, southwestern, and central my country, due to special geographical and climatic conditions, frequent low temperatures, rain, snow, and freezing weather in winter easily lead to severe icing on transmission lines. Icing increases the weight of the conductors, alters their aerodynamic characteristics, and under the influence of wind, causes severe vibration, galloping, and even breakage of the conductors, resulting in large-scale power outages.

[0003] There are still unresolved pain points in the current high-voltage transmission line icing safety assessment technology. In actual operation and maintenance, monitoring equipment is scattered and of varying quality, resulting in isolated data collection and an inability to form a comprehensive and coherent icing risk data chain. Existing icing predictions rely heavily on empirical judgment, lacking in-depth research into micrometeorological conditions and the dynamics of icing formation, leading to high model prediction bias rates and difficulty in meeting the need for accurate predictions under complex weather conditions. Furthermore, traditional risk assessment methods are too crude, failing to fully consider the complex interaction between line characteristics and topography. For example, on a mountainous transmission line, icing thickness on different tower sections can vary by more than three times due to terrain differences, yet a uniform assessment standard is applied. Early warning mechanisms lack a scientific computational basis; threshold settings often rely on empirical estimations, resulting in insufficient accuracy under extreme weather conditions, causing power grid departments to frequently initiate unnecessary emergency responses or miss critical defense opportunities. Furthermore, the allocation of emergency response resources failed to differentiate according to the risk levels of different regions, resulting in insufficient resources in key areas and idle resources in low-risk areas. This led to low efficiency in emergency resource dispatch, causing delays in de-icing of critical lines, resulting in large-scale power outages and huge economic losses.

[0004] In view of this, the present invention proposes a safety assessment method for high-voltage transmission lines based on intelligent algorithms to solve the above problems. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: a high-voltage transmission line safety assessment method based on intelligent algorithms, comprising:

[0006] Step 1: Collect and fuse multi-source data on meteorological parameters, line load status, and historical icing events along the high-voltage transmission line to obtain a database of spatiotemporal icing risk correlation factors.

[0007] Step 2: Based on the spatiotemporal icing risk correlation factor database, construct a deep learning model, conduct correlation analysis and evolution law mining between meteorological conditions and icing formation mechanism, and obtain a dynamic prediction model for icing growth.

[0008] Step 3: Based on the dynamic prediction model for icing growth and combined with the physical characteristics and topographic information of the transmission line, conduct multi-scenario simulation and sensitivity analysis to obtain a risk zoning assessment system for line icing.

[0009] Step 4: Based on the aforementioned line icing risk zoning assessment system, calculate the line mechanical strength and determine the critical value of icing thickness to obtain the graded early warning threshold standard;

[0010] Step 5: Based on the graded early warning threshold standard and combined with the collected real-time meteorological monitoring data and power grid operation status, construct a time-series prediction algorithm and a risk propagation model to obtain an intelligent early warning mechanism for icing risk;

[0011] Step 6: Based on the aforementioned intelligent early warning mechanism for icing risk, optimize the allocation of anti-icing resources and adaptively generate de-icing plans to obtain differentiated prevention and control strategies and emergency response plans.

[0012] The technical effects and advantages of the intelligent algorithm-based high-voltage transmission line safety assessment method of this invention are as follows:

[0013] This invention improves the overall level of safety management for high-voltage transmission lines. By establishing a complete icing risk assessment and early warning system, it significantly enhances the power grid's ability to withstand severe weather conditions. This invention greatly improves the completeness and reliability of data utilization, significantly improves the accuracy and timeliness of icing prediction, effectively reduces false alarm and missed alarm rates, and provides a more scientific and reliable basis for operation and maintenance decisions. Through refined risk zoning management, it achieves precise allocation and optimized configuration of prevention and control resources, improves resource utilization efficiency, and reduces power grid operation and maintenance costs. The intelligent early warning mechanism of this invention realizes the transformation from passive response to proactive prevention and control, shortens emergency response time, improves repair efficiency, effectively reduces line faults and power outages caused by icing disasters, and significantly enhances the reliability and stability of power grid supply. This invention also achieves precise regional management through differentiated prevention and control strategies, greatly improving the power grid safety level under severe weather conditions. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of the high-voltage transmission line safety assessment method based on intelligent algorithms of the present invention;

[0015] Figure 2 This is a detailed flowchart illustrating step 6 of the present invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] This application provides a method for safety assessment of high-voltage transmission lines based on intelligent algorithms. The execution entities of this method include, but are not limited to: mechanical equipment, data processing platforms, cloud server nodes, and network upload devices, which can be considered general computing nodes in this application. The data processing platform includes, but is not limited to: at least one of a transmission line monitoring system, a meteorological data management system, an icing early warning platform, and a power grid safety assessment system.

[0018] Please see Figure 1 This invention provides a method for safety assessment of high-voltage transmission lines based on intelligent algorithms, comprising the following steps:

[0019] Step 1: Collect and fuse multi-source data on meteorological parameters, line load status, and historical icing events along the high-voltage transmission line to obtain a database of spatiotemporal icing risk correlation factors.

[0020] Step 2: Construct a deep learning model based on the spatiotemporal icing risk correlation factor database, conduct correlation analysis and evolution law mining between meteorological conditions and icing formation mechanism, and obtain a dynamic prediction model for icing growth.

[0021] Step 3: Based on the dynamic prediction model of icing growth and combined with the physical characteristics and topographic information of the transmission line, conduct multi-scenario simulation and sensitivity analysis to obtain the line icing risk zoning assessment system.

[0022] Step 4: Based on the line icing risk zoning assessment system, calculate the mechanical strength of the line and determine the critical value of icing thickness to obtain the graded early warning threshold standard;

[0023] Step 5: Based on the graded early warning threshold standard and combined with the collected real-time meteorological monitoring data and power grid operation status, construct a time series prediction algorithm and a risk propagation model to obtain an intelligent early warning mechanism for icing risk;

[0024] Step 6: Based on the intelligent early warning mechanism for icing risk, optimize the allocation of anti-icing resources and adaptively generate de-icing plans to obtain differentiated prevention and control strategies and emergency response plans.

[0025] This invention constructs a spatiotemporal icing risk correlation factor database through multi-source data acquisition and fusion processing, providing a reliable data foundation for subsequent analysis. Based on a deep learning model, it performs correlation analysis between meteorological conditions and icing formation mechanisms to achieve an accurate understanding of the icing formation process. Utilizing a dynamic prediction model for icing growth combined with line physical characteristics for multi-scenario simulation helps identify high-risk areas. Line mechanical strength calculation and determination of critical icing thickness provide a scientific basis for early warning thresholds. The construction of time-series prediction algorithms and risk propagation models enables automated and precise intelligent early warning. The formulation of differentiated prevention and control strategies and emergency response plans significantly improves the power grid's disaster prevention and mitigation capabilities, ensuring the safe and stable operation of high-voltage transmission lines under complex meteorological conditions.

[0026] In this embodiment of the invention, the detailed implementation steps of step 1 include:

[0027] Data on temperature, humidity, wind speed, wind direction, and precipitation from meteorological stations along high-voltage transmission lines were collected and their quality controlled to obtain a standardized meteorological parameter matrix.

[0028] Real-time acquisition and outlier processing of current, voltage, power factor, and line vibration data of high-voltage transmission lines are performed to obtain the line load state sequence.

[0029] The occurrence time, icing thickness, duration and impact range of historical icing events on high-voltage transmission lines were extracted in a structured manner to obtain a historical icing event record table;

[0030] Spatiotemporal alignment and data fusion were performed on the standardized meteorological parameter matrix, line load state sequence, and historical icing event record table to obtain a multidimensional feature tensor.

[0031] Missing values ​​were filled and outliers were detected in the multidimensional feature tensor to obtain a complete database of spatiotemporal icing risk correlation factors.

[0032] In this embodiment, meteorological data is first collected from meteorological stations along the high-voltage transmission line, including parameters such as temperature (°C), relative humidity (%), wind speed (m / s), wind direction, and precipitation (mm), ensuring that the temporal resolution (e.g., hourly or higher) and spatial resolution (station coverage density) of the data meet the analysis requirements. Quality control is performed on the collected meteorological data, including missing value detection (identifying missing points in the data sequence), outlier detection (identifying values ​​that significantly deviate from the normal range), and consistency checks (checking whether the physical relationships between related parameters are reasonable). Standardization methods (such as Z-score standardization, Min-Max standardization, etc.) are used to process the meteorological data to eliminate the influence of different units, making the parameters comparable. The processed meteorological parameters are arranged according to time series to form a standardized meteorological parameter matrix, where rows represent time points and columns represent different meteorological parameters and station locations. A real-time monitoring system is used to collect electrical parameter data of the high-voltage transmission line, including current (A), voltage (kV), power factor, and line vibration data, ensuring that the frequency and accuracy of data acquisition meet the analysis requirements.Outlier processing is performed on the collected line parameter data to identify and handle abnormal data caused by equipment failures, communication interruptions, etc. Statistical methods or machine learning methods (such as Isolation Forest, One-Class SVM, etc.) are used for anomaly detection. The processed line parameter data is arranged according to time series to form a line load state sequence. Historical icing event data of high-voltage transmission lines is extracted from the historical record system, including information such as event occurrence time, icing thickness (mm), duration (h), and affected area (km). Unstructured data (such as accident reports, maintenance records, etc.) is structured by extracting key information from the text using text mining or natural language processing techniques. The extracted historical icing event information is organized into a structured tabular form to form a historical icing event record table. The standardized meteorological parameter matrix, line load state sequence, and historical icing event record table are aligned in time and space to ensure that data from different sources are consistent in time and space. Correspondingly, spatiotemporal interpolation methods (such as IDW, Kriging, etc.) can be used to process spatially unevenly distributed data. Data fusion techniques (such as Kalman filtering, Bayesian fusion, etc.) can be used to fuse data from different sources into a unified multidimensional feature tensor. This tensor comprehensively represents multidimensional information such as time, space, meteorological conditions, line status, and historical events. Missing values ​​in the multidimensional feature tensor can be filled in using various methods: temporal interpolation methods (such as linear interpolation, spline interpolation, etc.), machine learning methods (such as K-nearest neighbors, random forest, etc.), deep learning methods (such as LSTM autoencoders, etc.). Outlier detection and processing are then performed on the filled data. Statistical methods (such as the 3σ rule, box plot method, etc.) or machine learning methods (such as density clustering, support vector machines, etc.) are used to identify and process outliers. After missing value filling and outlier detection processing, a complete spatiotemporal icing risk correlation factor database is formed. This database provides high-quality training data for the subsequent construction of deep learning models.

[0033] In this embodiment of the invention, the detailed implementation steps of step 2 include:

[0034] Feature engineering was performed on the database of spatiotemporal icing risk correlation factors to obtain a set of key influencing factors of icing.

[0035] Based on the set of key influencing factors of icing, a deep learning architecture combining recurrent neural networks and attention mechanisms was constructed to obtain an initial deep learning model;

[0036] The parameters and structure of the initial deep learning model are tuned and optimized to obtain the optimized deep learning model.

[0037] The multidimensional correlation between meteorological conditions and ice formation was analyzed using an optimized deep learning model, resulting in an ice formation mechanism correlation network.

[0038] Based on the correlation network of ice formation mechanism, the temporal evolution law of ice growth is explored, and a dynamic prediction model of ice growth is constructed.

[0039] In this embodiment, feature engineering is performed on the spatiotemporal icing risk correlation factor database, including feature extraction, feature selection, and feature transformation. Meaningful features are extracted from the original data, such as temperature and humidity combination features, wind speed and direction composite features, and precipitation features. Feature selection methods (such as analysis of variance, mutual information method, L1 regularization, etc.) are used to select features highly correlated with icing formation. Feature transformation techniques (such as polynomial transformation, logarithmic transformation, etc.) are used to enhance the expressive power of the data, forming a set of key influencing factors of icing. This set contains key factors that have a significant impact on icing formation and development. Based on the set of key influencing factors of icing, a deep learning model architecture is constructed. Combining the temporal processing capability of recurrent neural networks (RNNs) and the important feature capture capability of attention mechanisms, long short-term memory networks (LSTM) or gated recurrent units (GRUs) can be selected as basic network units to process temporal features and introduce attention mechanisms, such as self-attention (Se). Multi-head attention (FHAttention) enhances the model's ability to perceive key features. A hierarchical network structure is designed, such as an input layer, multiple RNN layers, attention layers, fully connected layers, and an output layer, forming an initial deep learning model. Parameter tuning and structural optimization are then performed on this initial model, including: learning rate adjustment (using learning rate scheduling strategies such as cosine annealing and learning rate decay), regularization (such as L1 / L2 regularization and Dropout to prevent overfitting), batch normalization (to accelerate training and improve stability), and structure search (such as adjusting the number of network layers and neurons). Cross-validation and other methods are used to evaluate model performance, avoiding overfitting and underfitting. Hyperparameter optimization methods (such as grid search and Bayesian optimization) can be used to automatically adjust model parameters, resulting in an optimized deep learning model.

[0040] The multidimensional correlation analysis between meteorological conditions and icing formation can be expressed by the following formula:

[0041] ;in, This represents the correlation coefficient between the i-th meteorological factor and the j-th icing characteristic. This represents the value of the i-th meteorological factor at time t. This represents the j-th icing feature value at time t. and These represent the average values ​​of the corresponding variables, T represents the length of the time series, and t represents the index of the time point.

[0042] An optimized deep learning model was used to analyze the multidimensional correlation between meteorological conditions and ice formation. Feature representations and attention weights of the model's intermediate layers were extracted to analyze the contribution and influence of different meteorological parameters on ice formation. Feature importance ranking and feature interaction analysis were constructed to identify key influencing factors and their interactions, forming an ice formation mechanism correlation network. This network represents the complex relationship between different meteorological conditions and ice formation. Based on this network, the temporal evolution of ice growth was further explored, analyzing the growth rate, saturation value, and decay characteristics of ice under different conditions. A dynamic model of ice thickness variation over time was established, which can be described using ordinary differential equations or difference equations, such as: ;in, This represents the icing thickness at time t. The icing growth coefficient is indicated by the influence of temperature T, humidity H, and wind speed W (this coefficient is larger when the temperature is slightly below 0°C, the humidity is high, and the wind speed is moderate, and icing forms quickly). This represents the icing melting coefficient affected by temperature T and solar radiation intensity S (this coefficient is larger when the temperature is above 0°C or the solar radiation is strong, and the icing melts faster). This represents the maximum possible icing thickness, the theoretical maximum thickness that icing can reach under specific conditions. It is determined by both meteorological conditions and the physical properties of the conductor and is used as a saturation value in icing growth models.

[0043] Taking into account meteorological forecasts, ice growth patterns, and historical data, a dynamic prediction model for ice growth is constructed. This model can predict the thickness and distribution of ice at different future time points based on meteorological conditions, providing a scientific basis for subsequent risk assessment and early warning.

[0044] In this embodiment of the invention, step 3 involves feature engineering the spatiotemporal icing risk correlation factor database to obtain a set of key icing influencing factors, including:

[0045] Correlation analysis was performed on the database of spatiotemporal icing risk-related factors to obtain a feature correlation matrix;

[0046] Principal component analysis and feature importance ranking are performed based on the feature correlation matrix to obtain a preliminary feature subset;

[0047] The feature interaction effect is evaluated on the initial feature subset to obtain the feature interaction matrix;

[0048] A simplified feature set is obtained by combining and generating features and eliminating redundant features based on the feature interaction matrix.

[0049] Time delay effect analysis and nonlinear transformation were performed on the simplified feature set to obtain the set of key influencing factors of icing.

[0050] In this embodiment, the correlation coefficients between features in the spatiotemporal icing risk correlation factor database are first calculated. Methods such as Pearson correlation coefficient, Spearman rank correlation coefficient, or Kendall's staunch correlation coefficient can be used. An appropriate correlation measurement method is selected based on the data type and distribution characteristics of different features to construct a feature correlation matrix. This matrix displays the strength and direction of the correlation between all feature pairs. Based on the feature correlation matrix, principal component analysis (PCA) is performed to reduce feature dimensionality while retaining key information. The mathematical representation of principal component analysis is as follows:

[0051] Where X is the original data matrix, W is the eigenvector matrix, and Y is the data matrix after dimensionality reduction.

[0052] Simultaneously, features are ranked using feature importance assessment methods (such as the reduction of average impurity in random forests and the feature importance assessment in gradient boosting trees). Combining PCA results with feature importance ranking, features with significant explanatory power are selected to form a preliminary feature subset. This subset contains features highly correlated with icing formation and development. Interaction effects are assessed on the features in this preliminary subset to analyze the impact of feature interactions on icing formation. Partial dependency plots (PDPs), ICE plots, or H-statistics can be used to evaluate feature interaction effects. A feature interaction matrix is ​​constructed, quantifying the interaction strength between different feature pairs. Based on the feature interaction matrix, meaningful feature combinations are generated, such as the product of temperature and humidity, or the combination of wind speed and precipitation. These combined features may have stronger predictive power than single features. In addition to improving the ability to predict icing, the model identifies and eliminates redundant features, such as highly correlated features or features with insignificant interaction effects. Methods like variance inflation factor (VIF) and L1 regularization can be used to assist in the judgment, forming a simplified feature set. This set retains key information while avoiding the curse of dimensionality. Time-lag effect analysis is then performed on the simplified feature set to study the lagged impact of meteorological conditions on icing formation. Methods such as cross-correlation function (CCF) or Granger causality test are used to analyze the correlation under different time lags. Considering the nonlinear relationships between features, nonlinear transformations are applied to the features, such as polynomial transformations, exponential transformations, and logarithmic transformations, to enhance the model's ability to capture nonlinear relationships. After time-lag effect analysis and nonlinear transformations, the final set of key influencing factors for icing is obtained, which includes all factors that significantly contribute to icing prediction.

[0053] In this embodiment of the invention, the detailed implementation steps of step 3 include:

[0054] The conductor type, erection height, tower spacing and line direction of the transmission line are parametrically expressed to obtain the line physical characteristic vector;

[0055] Spatial analysis was performed on the elevation, slope, vegetation cover and water system distribution along the transmission line to obtain a topographic feature layer.

[0056] By combining the dynamic prediction model of icing growth, the physical characteristic vector of the line and the topographic feature layer, a combination of meteorological conditions for multiple scenarios is designed to obtain the scenario simulation matrix.

[0057] Numerical simulations of the ice formation and development process were performed on the scene simulation matrix to obtain an ice risk distribution map;

[0058] Sensitivity analysis and clustering based on the icing risk distribution map were conducted to obtain a zoning assessment system for line icing risk.

[0059] In this embodiment, the physical characteristics of high-voltage transmission lines are parameterized. Basic line parameters are collected, including: conductor type (such as aluminum-coated steel conductors (ACSR), aluminum alloy conductors, etc.) and their specifications (such as cross-sectional area, diameter, etc.), erection height (m), tower spacing (m), and line direction (azimuth). Considering the physical characteristics of different conductor types, such as wire diameter, mass per unit length, linear expansion coefficient, and elastic modulus, a conductor characteristic parameter table is established. Parameters of transmission towers are collected, such as tower type, height, and strength grade. Considering the spatial distribution of towers and the geometric arrangement of the line, the collected line physical parameters are converted into standardized numerical values. The data is used to form a physical characteristic vector of the transmission line, which contains key parameters describing the physical characteristics of the line. Spatial analysis of the topography along the transmission line is performed, using digital elevation model (DEM) data to obtain the elevation distribution along the line, generating an elevation profile map of the line. Topographic slope is analyzed, calculating slope values ​​and aspects at different locations along the line, identifying special terrain features (such as valleys, ridges, steep slopes, etc.). Vegetation cover along the line is analyzed using remote sensing data or land cover data, including vegetation type, coverage, and height, considering the impact of vegetation on local microclimates. The distribution of water systems along the line is analyzed, including the location and area of ​​water bodies such as rivers, lakes, and reservoirs. The impact of water bodies on local humidity and wind field is investigated. Integrating the aforementioned topographic information, a Geographic Information System (GIS) layer is generated, forming a topographic feature layer. This layer contains detailed topographic information along the transmission line. Combining the dynamic prediction model for icing growth, the line's physical characteristic vector, and the topographic feature layer, various possible combinations of meteorological conditions are designed, covering different seasons, weather systems, and extreme conditions. Different combinations of parameters such as temperature, humidity, wind speed, wind direction, and precipitation are considered, with particular attention paid to meteorological conditions in historical icing events. A systematic design of meteorological parameter combinations, such as factor design or orthogonal experimental design, is employed to ensure comprehensive scenario coverage. To ensure comprehensiveness and representativeness, a scenario simulation matrix was formed, which includes parameter settings for various meteorological condition combinations. Numerical simulations of the icing formation and development process were performed for each meteorological condition combination in the scenario simulation matrix. An icing growth dynamic prediction model was used to predict the icing growth process under different conditions. The influence of the physical characteristics of the line on the icing process was considered, such as conductor diameter and erection height. Local microclimate effects, such as valley wind effect and lake effect, were considered in combination with topographic features. The icing thickness at different locations and time points was calculated to form an icing risk distribution map. This map shows the icing risk level of each section of the line under different meteorological conditions.

[0060] The following formula can be used to calculate icing load:

[0061] ;in, Indicates the weight of the ice layer (N). Let r represent the ice density (kg / m³), r represent the conductor radius (m), b represent the ice thickness (m), L represent the length of the calculated section (m), and g represent the gravitational acceleration (m / s²).

[0062] Sensitivity analysis is conducted based on the icing risk distribution map to assess the impact of different factors on icing risk. Sensitivity analysis methods (such as ANOVA, Morris method, Sobol index, etc.) are used to quantify the importance and scope of influence of each factor, identify key influencing factors and high-sensitivity areas, and classify the transmission lines into risk clusters based on the sensitivity analysis results and icing risk levels. Clustering algorithms (such as K-means, DBSCAN, hierarchical clustering, etc.) or expert knowledge can be used to divide the transmission lines into different risk level areas. Considering geographical continuity and management convenience, risk zone boundaries are reasonably set to form a transmission line icing risk zoning assessment system. This system divides the transmission lines into different risk zones, each with relatively consistent risk characteristics, which facilitates differentiated management and early warning.

[0063] In this embodiment of the invention, the detailed implementation steps of step 4 include:

[0064] Mechanical models were established for transmission lines in each risk zone of the line icing risk zoning assessment system to obtain the line stress calculation framework;

[0065] Static load analysis was performed on the line stress calculation framework under different icing thickness conditions to obtain the icing load stress distribution;

[0066] Based on the stress distribution of icing load, the mechanical strength limit state analysis of the line is carried out to obtain the critical icing thickness curve.

[0067] The critical icing thickness curve is corrected for safety factors and adjusted for regional adaptation to obtain regionalized critical icing thickness values.

[0068] Based on the regional icing thickness critical value, a rule for classifying early warning levels was designed, resulting in a graded early warning threshold standard.

[0069] In this embodiment, a mechanical model is established for the transmission lines in each risk zone of the line icing risk zoning assessment system. Corresponding material mechanics models are established for different types of conductors (such as steel-cored aluminum stranded wire, carbon fiber composite conductors, etc.). Considering parameters such as the conductor's elastic modulus and plastic deformation characteristics, a catenary model is established to describe the geometric shape of the conductor under its own weight and external loads. The catenary equation can be expressed as:

[0070] Where y is the vertical coordinate of the conductor at position x, a is the catenary parameter, and s is the horizontal distance between the two support points.

[0071] Considering the mechanical properties and support conditions of the towers, a comprehensive stress calculation framework for power lines is established. This framework can simulate the stress state of conductors and towers under icing conditions. Based on this framework, static load analysis is performed under different icing thicknesses to calculate the additional weight caused by icing. The icing load can be calculated using the following formula:

[0072] Where G_ice is the weight of the ice (N), ρ_ice is the density of the ice (kg / m³), g is the acceleration due to gravity (m / s²), r is the radius of the conductor (m), h_ice is the thickness of the ice (m), and L is the length of the conductor (m).

[0073] Formula for calculating wind pressure load:

[0074] Where F_wind is wind pressure (N), ρ_air is air density (kg / m³), v is wind speed (m / s), C_d is drag coefficient, and A is windward area (m²).

[0075] Calculation formula for conductor tension under combined load:

[0076] Where T is the conductor tension (N), m is the mass per unit length of the conductor (kg / m), f is the conductor sag (m), and θ is the angle between the resultant load and the vertical direction.

[0077] The stress distribution of different tower sections was analyzed to identify the points of maximum stress and weak points, resulting in an ice load stress distribution map. This map shows the stress state of various parts of the line under different ice thicknesses. Based on the ice load stress distribution, a mechanical strength limit state analysis of the line was conducted to determine the mechanical strength limit of the conductor. Considering the tensile strength and safety factor of the conductor, the ultimate bearing capacity of the conductor was determined. The bearing capacity of the towers was analyzed, and the ultimate bearing capacity of the towers was determined according to the tower design standards and material properties. Considering the strength limitations of connectors (such as insulators and hardware), and comprehensively considering the limit states of all components of the line, the critical state of the overall system was determined. Limit state analysis was conducted based on different locations and conditions. A critical icing thickness curve is plotted, which represents the icing thickness at which the line reaches its limit under different conditions (such as different temperatures and wind speeds). The critical icing thickness curve is then corrected for safety factors and adjusted for regional adaptability. Considering the engineering safety factor, which is generally between 1.5 and 2.5, the critical thickness is reduced to ensure a safety margin. Based on the line icing risk zoning assessment system, the critical thickness of different risk areas is adjusted differently. A more conservative safety factor is used in high-risk areas. Considering seasonal factors and extreme weather conditions, a dynamic adjustment mechanism is set up to form a regionalized critical icing thickness value. This critical value is the upper limit of safe icing thickness under different regions and conditions.

[0078] In this embodiment of the invention, the critical icing thickness curve is corrected for safety factors and adjusted for regional adaptability to obtain regionalized critical icing thickness values, including:

[0079] By introducing multi-level safety factors into the critical icing thickness curve, a family of safety margin curves is obtained;

[0080] Based on historical icing accident data, the family of safety margin curves was verified and calibrated to obtain the corrected critical thickness curve.

[0081] Based on the regional characteristics in the line icing risk zoning assessment system, the corrected critical thickness curve is mapped to different zones to obtain the zone critical value matrix.

[0082] The critical value matrix of the zones is corrected for seasonal and extreme weather conditions to obtain a table of dynamic adjustment coefficients;

[0083] By combining the regional critical value matrix and the dynamic adjustment coefficient table, regionalized icing thickness critical values ​​are generated.

[0084] In this embodiment, a multi-level safety factor is introduced for the critical icing thickness curve. Different levels of safety factors are set according to the importance of the line, reliability requirements, and risk tolerance. For example, a level 1 safety factor (e.g., 2.5) is suitable for UHV main lines and important load power supply lines; a level 2 safety factor (e.g., 2.0) is suitable for conventional high-voltage transmission lines; and a level 3 safety factor (e.g., 1.5) is suitable for general lines in low-risk areas. The mathematical expression of the safety factor is:

[0085] Where h_safe is the safe icing thickness, h_critical is the critical icing thickness, and SF is the safety factor, which refers to the coefficient that provides a safety margin in engineering design, usually greater than 1. For example, an important line may use a safety factor of 2.5, while an ordinary line may use a safety factor of 1.5.

[0086] A family of safety margin curves is generated using multi-level safety factors. This family includes safe icing thickness curves under different safety levels. Historical icing accident data is collected and analyzed, including icing thickness, meteorological conditions, and fault types leading to line faults. This data is then compared with the safety margin curve family to verify the rationality of the safety margin curves. Based on actual accident data, the safety factors and critical thickness curves are adjusted to better reflect real-world conditions. Statistical regression or machine learning methods can be used for parameter optimization to form a corrected critical thickness curve. This curve considers both theoretical calculations and historical experience. Based on the regional characteristics in the line icing risk zoning assessment system, the corrected critical thickness curve is mapped to different zones. To address the characteristics of different risk zones (such as climate conditions, topographic features, and historical icing records), the critical thickness values ​​are adjusted to form a zone-specific critical value matrix. This matrix sets a specific critical icing thickness value for each risk zone. Considering the impact of seasonal factors on icing risk, different seasons may require different critical value standards; for example, the critical value may need to be more stringent in winter. Taking into account the amplifying effect of extreme weather conditions (such as sharp temperature drops and blizzards) on icing risk, a dynamic adjustment mechanism is designed, forming a dynamic adjustment coefficient table. This table defines the adjustment coefficients for critical values ​​under different conditions. Combining the zone-specific critical value matrix and the dynamic adjustment coefficient table, the final regionalized critical icing thickness values ​​are generated. The critical value calculation formula can be expressed as:

[0087] Where h_thre(r, s, w) represents the critical value of icing thickness for region r under season s and weather conditions w; h_base(r) represents the basic critical value for region r, which refers to the safe threshold of icing thickness for a specific region r under standard conditions. It is a benchmark value determined based on the physical characteristics of the line and the basic risk level in that region, without considering seasonal and special weather changes; C_s represents the seasonal adjustment coefficient (for example, it may be set to 0.8 in winter and close to 1 in summer); C_w represents the weather condition adjustment coefficient (in extreme weather such as blizzards and sharp temperature drops, this coefficient may be set to below 0.7 to increase the safety margin).

[0088] Based on the regionalized critical values ​​for icing thickness, a rule for classifying warning levels is designed, generally resulting in four warning levels: blue (mild risk), yellow (moderate risk), orange (relatively severe risk), and red (severe risk). The triggering conditions for each warning level are defined, such as:

[0089] Blue alert: Ice thickness is predicted to reach 60%-70% of the critical value;

[0090] Yellow alert: Ice thickness is predicted to reach 70%-80% of the critical value;

[0091] Orange alert: The predicted ice thickness has reached 80%-90% of the critical value;

[0092] Red alert: The predicted ice thickness has reached more than 90% of the critical value;

[0093] Considering the speed and trend of icing development, the early warning triggering conditions are adjusted. For example, if the development speed is fast, an early warning can be triggered in advance, forming a graded early warning threshold standard. This standard provides a decision-making basis for the subsequent intelligent early warning mechanism.

[0094] In this embodiment of the invention, the detailed implementation steps of step 5 include:

[0095] Quality control and missing value repair are performed on real-time meteorological monitoring data to obtain an effective meteorological monitoring data stream;

[0096] Real-time acquisition and feature extraction of power grid operation status data yields a power grid operation status feature sequence;

[0097] Based on effective meteorological monitoring data streams and power grid operation status characteristic sequences, a time series prediction algorithm is constructed to obtain an ice accretion risk prediction model.

[0098] The icing risk prediction model is combined with the graded early warning threshold standard to determine the threshold trigger and obtain the initial risk warning signal;

[0099] A risk propagation model is constructed based on the initial risk warning signal, resulting in an intelligent early warning mechanism for icing risk.

[0100] In this embodiment, quality control and missing value repair are performed on real-time meteorological monitoring data. Real-time meteorological data, including parameters such as temperature, humidity, wind speed, wind direction, and precipitation, is received from meteorological stations, online monitoring systems, and meteorological radars. Quality control is performed on the real-time data to detect anomalies (such as sudden changes, frozen values, and unreasonable values). Multiple methods are used to determine data validity, such as physical constraint methods and statistical detection methods. Missing or outlier values ​​are repaired using temporal interpolation methods (such as linear interpolation and spline interpolation), spatial interpolation methods (such as IDW and Kriging), or machine learning methods (such as K-nearest neighbors and random forests), forming an effective meteorological monitoring data stream. This data stream is then provided... This system provides continuous and reliable real-time meteorological information, enabling real-time acquisition and feature extraction of power grid operating status data. It collects operating status data of transmission lines, such as parameters like current, voltage, power, phase, and temperature, using SCADA systems, PMUs (phasor measurement units), and online monitoring systems to acquire real-time data. Useful features are extracted from the raw power grid data, including load characteristics (such as load level and load change rate) and operating characteristics (such as line temperature and vibration status). Derived features, such as power factor and line impedance changes, are calculated to form a power grid operating status feature sequence. This sequence reflects the real-time operating status of the power grid. Based on the effective meteorological monitoring data stream and the power grid operating status feature sequence, a time series forecasting mechanism is constructed. The algorithm can select appropriate time series forecasting methods, such as Autoregressive Integrated Moving Average (ARIMA), Long Short-Term Memory (LSTM), and Gated Recurrent Unit (GRU), and combine meteorological principles and power grid physical models to design feature engineering and forecasting strategies. The forecasting model is trained using historical data, and its performance is evaluated using cross-validation. The model parameters are continuously optimized to form an icing risk forecasting model. This model can predict icing risk in future time periods (e.g., 24 hours, 48 ​​hours, 72 hours). The output of the icing risk forecasting model is compared with the tiered warning threshold standards to determine whether a warning threshold has been triggered. Specific threshold requirements for different regions and time periods are considered in the design. Threshold triggering logic, such as continuous threshold exceeding time and threshold exceeding magnitude, forms an initial risk warning signal. This signal contains information such as warning level, location, and time range. Based on the initial risk warning signal, a risk propagation model is constructed, considering the propagation path and impact range of icing risk in the power grid. For example, an icing-induced line break may trigger a chain reaction. The risk cascading effect is simulated, and the probability and severity of the chain impact of risk events are calculated. The impact of external factors (such as road traffic conditions and the accessibility of repair resources) on risk response is considered. By integrating risk prediction, threshold judgment, and risk propagation model, an intelligent icing risk early warning mechanism is formed. This mechanism can provide early warning of possible icing risks and assess the severity and impact range of the risks.

[0101] In this embodiment of the invention, a time series prediction algorithm is constructed based on effective meteorological monitoring data streams and power grid operation status characteristic sequences to obtain an icing risk prediction model, including:

[0102] Time-series features are extracted from effective meteorological monitoring data streams and power grid operation status feature sequences to obtain a multi-source time-series feature matrix;

[0103] A long short-term memory network model structure was constructed, and a sliding window processing method was applied to the multi-source temporal feature matrix to obtain the training sample set;

[0104] Based on the training sample set, the parameters of the long short-term memory network are learned and the model is trained to obtain the initial time series prediction model.

[0105] An enhanced time series prediction model is obtained by performing ensemble learning and adaptive adjustment on the initial time series prediction model.

[0106] The performance of the enhanced time-series prediction model for icing probability prediction and uncertainty quantification are performed to obtain the icing risk prediction model.

[0107] In this embodiment, time-series features are extracted from effective meteorological monitoring data streams and power grid operation status feature sequences. Time-series features are extracted from the original time-series data, including: statistical features (such as mean, standard deviation, maximum, minimum, etc.), trend features (such as slope, periodicity), and frequency domain features (obtained through Fourier transform or wavelet transform). Feature fusion is performed on multi-source heterogeneous data to unify the time scale and spatial range of different data sources, generating a multi-source time-series feature matrix. This matrix contains comprehensive time-series features from meteorological monitoring and power grid status. A Long Short-Term Memory (LSTM) network model structure is constructed, and the network architecture is designed, including an input layer and an LST layer. The model consists of an M-layer (which may contain multiple LSTM layers), a fully connected layer, and an output layer. Model parameters are determined, such as the number of LSTM layers, the number of neurons per layer, and the type of activation function. A sliding window process is applied to the multi-source temporal feature matrix, dividing the continuous time series data into overlapping time windows. Each window contains a time series and a corresponding target value (e.g., future ice thickness or risk probability). Appropriate window sizes and sliding step sizes are determined to balance prediction accuracy and computational efficiency. A training sample set is generated, containing numerous input windows and their corresponding target outputs. Based on this training sample set, the LSTM network undergoes parameter learning and model training, using backpropagation. The algorithm updates model parameters and minimizes the loss function (such as mean squared error or cross-entropy loss). Optimization algorithms such as Adam and RMSprop can be used. Batch training is implemented, setting appropriate batch size and training epochs. An early stopping strategy is adopted to avoid overfitting, forming an initial time-series prediction model. This model can predict future icing risk based on historical data. The initial time-series prediction model undergoes ensemble learning and adaptive adjustment, constructing a model ensemble framework, such as model averaging, Bagging, and Boosting, combining the prediction results of multiple base models to improve prediction stability and accuracy. This achieves adaptive model adjustment based on the latest data and prediction errors. The model parameters are dynamically updated using online learning or incremental learning methods to form an enhanced time-series prediction model. This model has stronger generalization and adaptability. The performance of the enhanced time-series prediction model in predicting icing probability is evaluated using various evaluation metrics, such as mean squared error (MSE), mean absolute error (MAE), ROC curve, and precision-recall curve. Cross-validation and independent test set validation are performed to assess the model's predictive performance under different conditions. The uncertainty of the prediction results is quantified, and the reliability and range of variation of the prediction results can be expressed using Monte Carlo methods, Bayesian methods, or prediction intervals. The uncertainty quantification formula can be expressed as:

[0108] Where P(y|x) is the posterior probability distribution of the output y given the input x, P(y|x, θ) is the conditional probability of the output y given the model parameters θ and the input x, and P(θ|D) is the posterior distribution of the model parameters θ given the training data D.

[0109] The final icing risk prediction model was developed. This model can not only predict the level of icing risk, but also quantify the uncertainty of the prediction, providing a more comprehensive reference for decision-making.

[0110] Please see Figure 2 The diagram below illustrates the detailed implementation steps of step 6. In this embodiment of the invention, the detailed implementation steps of step 6 include:

[0111] Based on the intelligent early warning mechanism for icing risk, a demand assessment of personnel, equipment and materials for icing prevention is conducted to obtain a resource demand matrix.

[0112] Based on the resource demand matrix and the existing resource distribution, resource allocation is optimized to obtain an optimized resource allocation scheme for anti-icing;

[0113] Differentiated de-icing technology selection strategies were designed for different icing risk levels and icing types, resulting in a de-icing technology solution library;

[0114] Based on the de-icing technology solution library and combined with line characteristics and environmental conditions, adaptive solution generation is performed to obtain line de-icing adaptive solution.

[0115] Spatiotemporal collaborative optimization of the anti-icing resource allocation scheme and the line de-icing adaptive scheme was carried out to obtain differentiated prevention and control strategies and emergency response plans.

[0116] In this embodiment, based on the intelligent early warning mechanism for icing risk, a demand assessment is conducted on various resources required for icing prevention work. This assessment includes: evaluating the demand for icing prevention personnel, including the number and professional structure of technical personnel (such as electricians and line maintenance workers), managers, and emergency repair personnel; evaluating the demand for icing prevention equipment, including the type and quantity of de-icing equipment (such as mechanical de-icers and thermal de-icing devices), monitoring equipment (such as icing monitors and drones), and emergency repair equipment; and evaluating the demand for icing prevention materials, including the types and quantities of de-icing agents, spare wires, spare parts, and fuel. Based on the risk warning level and impact range, the demand for different resources is calculated, forming a resource demand matrix. This matrix describes the quantity of various resources required under different regions and risk levels. Based on the source demand matrix and existing resource distribution, resource allocation optimization is performed. The distribution of existing anti-icing resources is investigated, including personnel allocation, equipment reserves, and material inventory in each region. A resource allocation optimization model is established, employing operations research methods such as linear programming and integer programming. The objective function can be to minimize resource scheduling costs or maximize risk coverage. Constraints include total resource constraints, response time constraints, and personnel skill constraints. Solving the resource allocation optimization problem yields the optimal resource scheduling scheme, forming an optimized anti-icing resource allocation scheme. This scheme guides how to efficiently allocate limited resources to cope with icing risks. Differentiated de-icing technology selection strategies are designed for different icing risk levels and icing types, and the applicability conditions of different de-icing technologies are analyzed. The advantages, disadvantages, and cost-effectiveness of various de-icing technologies, such as mechanical de-icing (suitable for light icing), thermal melting (suitable for moderate icing), and chemical de-icing (special cases), are analyzed. Decision rules are established to select the most suitable de-icing technology based on the type of icing (e.g., hard ice, soft ice, mixed ice) and thickness. Environmental impact and safety factors are considered to avoid environmental damage or safety accidents, resulting in a de-icing technology solution library. This library contains technical solutions for various icing conditions. Based on this library, and combined with line characteristics and environmental conditions, de-icing solutions are adaptively generated, considering line characteristics such as conductor type, erection height, and tower structure, and analyzing the impact of environmental conditions such as temperature, wind force, terrain conditions, and accessibility. Automatic adjustments are then made based on specific circumstances. A customized de-icing solution is generated, including the selection of de-icing technology, operation procedures, and safety measures, forming an adaptive de-icing solution for the line. This solution is tailored to the specific line segment and conditions, providing the optimal de-icing solution. The anti-icing resource optimization allocation scheme and the adaptive de-icing solution are spatiotemporally optimized, analyzing the time arrangement and spatial distribution of each scheme to identify potential temporal and spatial conflicts. For example, if resource demand in multiple areas exceeds supply capacity simultaneously, collaborative optimization algorithms (such as genetic algorithms and particle swarm optimization) are used to resolve spatiotemporal conflicts and optimize resource utilization efficiency. An emergency response process is designed, clarifying the responsibilities of each department and role, forming a complete emergency plan, including specific measures for the early warning, preparation, response, and recovery stages.A differentiated prevention and control strategy and emergency response plan were developed. This plan provides customized prevention and control strategies and emergency response measures for different risk areas and risk levels, ensuring the scientific nature and effectiveness of the anti-icing work.

[0117] In this embodiment of the invention, the anti-icing resource optimization allocation scheme and the line de-icing adaptive scheme are spatiotemporally coordinated for optimization to obtain differentiated prevention and control strategies and emergency response schemes, including:

[0118] By constructing the objective function and constraints for anti-icing resource scheduling optimization, the resource scheduling optimization problem is obtained.

[0119] An improved particle swarm optimization algorithm is used to solve the resource scheduling optimization problem and obtain the optimal resource scheduling scheme.

[0120] The execution timing plan and priority setting of the line de-icing adaptive scheme are performed to obtain the de-icing execution timing diagram;

[0121] By combining the optimal resource scheduling scheme and the de-icing execution sequence diagram, spatiotemporal conflict detection and resolution are performed to obtain a collaborative optimization scheme;

[0122] By designing emergency response procedures and clarifying the division of responsibilities for collaborative optimization schemes, executable differentiated prevention and control strategies and emergency response plans are obtained.

[0123] Simulation verification and effectiveness evaluation of differentiated prevention and control strategies and emergency response plans were conducted to obtain a feedback mechanism for plan implementation.

[0124] In this embodiment, an objective function and constraints for optimizing anti-icing resource scheduling are constructed. The objective function is defined, such as minimizing the total response time, minimizing the resource mobilization cost, or maximizing the icing risk coverage rate. The objective function can be expressed as:

[0125] Where c_ij represents the cost of allocating resource i to location j, t_ij represents the corresponding response time, x_ij and y_ij are decision variables, representing the amount of resource allocation and whether to allocate, respectively, and Z is the objective function value.

[0126] Define constraints, including: total resource constraints (limited number of available resources), response time constraints (must respond within a specified time), coverage constraints (must cover all high-risk areas), and resource complementarity constraints (some resources must be used in combination). Combine the objective function and constraints to form a resource scheduling optimization problem. Use an improved particle swarm optimization (PSO) algorithm to solve the resource scheduling optimization problem. Design a particle encoding scheme so that each particle represents a feasible resource scheduling scheme. Introduce improved strategies, such as inertia weight adjustment, crossover and mutation operations, and local search, to improve the algorithm's convergence and global search capability. Run the algorithm for iterative optimization until the termination condition is met (such as reaching the maximum number of iterations or meeting the accuracy requirements). The optimal resource scheduling scheme is obtained, which provides the optimal allocation strategy for anti-icing resources. The execution sequence planning of the line de-icing adaptive scheme is performed, analyzing the priority, duration, and dependencies of each de-icing task. A task execution network diagram is constructed, and the critical path method (CPM) or Project Evaluation and Review Technique (PERT) is used to determine critical tasks and their timing arrangements. Priorities are set based on factors such as line importance, icing risk level, and power supply reliability requirements. High-priority tasks receive priority resource allocation and execution time, generating a de-icing execution sequence diagram. This diagram clearly defines the execution order, time window, and resource requirements of each de-icing task. The optimal resource scheduling scheme and the de-icing execution sequence diagram are then combined. This involves detecting and resolving spatiotemporal conflicts, such as multiple demands for the same resource within a given timeframe or spatial overlap of multiple tasks within the same region. Conflict resolution strategies are developed, including task rearrangement, resource reallocation, and task decomposition. Heuristic algorithms or constraint fulfillment techniques can be used to resolve conflicts, forming a collaborative optimization scheme. This scheme coordinates resource allocation and task execution in both time and space dimensions, avoiding conflicts and improving efficiency. Based on this collaborative optimization scheme, an emergency response process is designed, including early warning information reception, response decision-making, resource mobilization, on-site operations, and situation feedback. The operational procedures and standards for each stage are clearly defined, responsibilities are clearly defined, and the specific duties and powers of relevant departments and personnel in the emergency response are determined. The system establishes a clear command structure and communication mechanism, forming an executable differentiated prevention and control strategy and emergency response plan. This plan provides customized prevention and control strategies and detailed emergency response procedures for different risk areas. The differentiated prevention and control strategies and emergency response plans are simulated and verified using computer simulation tools to simulate the emergency response process under different icing conditions, verifying the feasibility and effectiveness of the plan, evaluating the implementation effect of the plan, setting evaluation indicators (such as response time, resource utilization rate, icing treatment effect, etc.), conducting effect evaluation under multiple scenarios, and establishing a plan implementation feedback mechanism, including real-time monitoring, effect evaluation, problem feedback, and plan optimization, forming a closed-loop management to ensure that the plan can be continuously improved and optimized.

[0127] Through the above implementation methods, the high-voltage transmission line safety assessment method based on intelligent algorithms provided by the present invention can comprehensively collect and process multi-source data, construct an accurate icing prediction model, realize scientific risk assessment and zoning, establish reasonable early warning thresholds, form an effective early warning mechanism, and provide differentiated prevention and control strategies and emergency response plans, which greatly improves the ability of high-voltage transmission lines to resist icing disasters and ensures the safe and stable operation of the power grid.

[0128] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0129] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0130] In the description of this invention, it should be understood that the terms "first," "second," etc., are used only for distinguishing descriptions and should not be construed as indicating or implying relative importance.

[0131] In the description of this invention, unless otherwise stated, "a plurality of" means two or more.

[0132] In the description of this invention, "several" means one or more, and "a large number" means two or more.

[0133] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0134] All formulas in this manual are dimensionless and calculated numerically. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0135] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for safety assessment of high-voltage transmission lines based on intelligent algorithms, characterized in that, include: Step 1: Collect and fuse multi-source data on meteorological parameters, line load status, and historical icing events along the high-voltage transmission line to obtain a database of spatiotemporal icing risk correlation factors. Step 2: Based on the aforementioned spatiotemporal icing risk correlation factor database, construct a deep learning model to conduct correlation analysis and evolution law mining between meteorological conditions and icing formation mechanisms, and obtain an icing growth dynamic prediction model, including: The spatiotemporal icing risk correlation factor database is subjected to feature engineering processing to obtain a set of key influencing factors of icing. Based on the set of key influencing factors of icing, a deep learning architecture combining recurrent neural networks and attention mechanisms is constructed to obtain an initial deep learning model; The initial deep learning model is subjected to parameter tuning and structure optimization to obtain an optimized deep learning model; The optimized deep learning model was used to analyze the multidimensional correlation between meteorological conditions and icing formation, and a network of associations for icing formation mechanisms was obtained. Based on the aforementioned ice formation mechanism association network, the temporal evolution law of ice growth is mined, and a dynamic prediction model for ice growth is constructed. Step 3: Based on the dynamic prediction model for icing growth and combined with the physical characteristics and topographic information of the transmission line, conduct multi-scenario simulation and sensitivity analysis to obtain a risk zoning assessment system for line icing. Step 4: Based on the aforementioned line icing risk zoning assessment system, calculate the line mechanical strength and determine the critical value of icing thickness to obtain the graded early warning threshold standard, including: A mechanical model is established for each risk zone of the transmission line in the aforementioned line icing risk zoning assessment system to obtain the line stress calculation framework; Static load analysis was performed on the stress calculation framework of the line under different icing thickness conditions to obtain the stress distribution of icing load; Based on the stress distribution of the icing load, the mechanical strength limit state analysis of the line is performed to obtain the critical icing thickness curve. The critical icing thickness curve is corrected for safety factor and adjusted for regional adaptation to obtain the regionalized critical icing thickness value. Based on the regionalized icing thickness critical value, a warning level classification rule is designed to obtain a graded warning threshold standard; The step of performing safety factor correction and zonal adaptive adjustment on the critical icing thickness curve to obtain regionalized critical icing thickness values ​​includes: A family of safety margin curves is obtained by introducing multiple levels of safety factors into the critical icing thickness curve. The safety margin curve family is verified and calibrated based on historical icing accident data to obtain the corrected critical thickness curve. Based on the regional characteristics in the line icing risk zoning assessment system, the corrected critical thickness curve is mapped to different zones to obtain a zone critical value matrix. The partition critical value matrix is ​​corrected for seasonal and extreme weather conditions to obtain a dynamic adjustment coefficient table; By combining the aforementioned partition critical value matrix and dynamic adjustment coefficient table, regionalized icing thickness critical values ​​are generated. Step 5: Based on the graded early warning threshold standard and combined with the collected real-time meteorological monitoring data and power grid operation status, construct a time-series prediction algorithm and a risk propagation model to obtain an intelligent early warning mechanism for icing risk; Step 6: Based on the aforementioned intelligent early warning mechanism for icing risk, optimize the allocation of anti-icing resources and adaptively generate de-icing plans to obtain differentiated prevention and control strategies and emergency response plans.

2. The method for safety assessment of high-voltage transmission lines based on intelligent algorithms according to claim 1, characterized in that, The process involves multi-source data collection and fusion processing of meteorological parameters, line load status, and historical icing events along high-voltage transmission lines to obtain a database of spatiotemporal icing risk correlation factors, including: Temperature, humidity, wind speed, wind direction, and precipitation data from meteorological stations along the high-voltage transmission line were collected and their quality controlled to obtain a standardized meteorological parameter matrix. Real-time acquisition and outlier processing of current, voltage, power factor, and line vibration data of the high-voltage transmission line are performed to obtain the line load state sequence. The occurrence time, icing thickness, duration and impact range of historical icing events on the high-voltage transmission line are extracted in a structured manner to obtain a historical icing event record table; Spatiotemporal alignment and data fusion are performed on the standardized meteorological parameter matrix, line load state sequence and historical icing event record table to obtain a multidimensional feature tensor; Missing values ​​are filled and outliers are detected in the multidimensional feature tensor to obtain a complete database of spatiotemporal icing risk correlation factors.

3. The method for safety assessment of high-voltage transmission lines based on intelligent algorithms according to claim 1, characterized in that, The feature engineering process performed on the spatiotemporal icing risk correlation factor database yields a set of key icing influencing factors, including: Correlation analysis was performed on the aforementioned database of spatiotemporal icing risk-related factors to obtain a feature correlation matrix; Principal component analysis and feature importance ranking are performed based on the feature correlation matrix to obtain a preliminary feature subset; The feature interaction effect is evaluated on the preliminary feature subset to obtain the feature interaction matrix; Based on the feature interaction matrix, feature combination generation and redundant feature elimination are performed to obtain a simplified feature set; Time delay effect analysis and nonlinear transformation are performed on the simplified feature set to obtain the set of key influencing factors of icing.

4. The method for safety assessment of high-voltage transmission lines based on intelligent algorithms according to claim 1, characterized in that, The aforementioned dynamic prediction model for icing growth, combined with the physical characteristics and topographic information of the transmission line, is used to conduct multi-scenario simulations and sensitivity analyses to obtain a zoning assessment system for transmission line icing risk, including: The conductor type, erection height, tower spacing, and line direction of the transmission line are parametrically expressed to obtain the line physical characteristic vector; Spatial analysis was performed on the elevation, slope, vegetation cover and water system distribution along the transmission line to obtain a topographic feature layer. Combining the aforementioned dynamic prediction model for icing growth, the physical characteristic vector of the line, and the topographic feature layer, a combination of meteorological conditions for multiple scenarios is designed to obtain a scenario simulation matrix; Numerical simulations of the ice formation and development process were performed on the aforementioned scenario simulation matrix to obtain an ice risk distribution map; Sensitivity analysis and clustering were performed based on the icing risk distribution map to obtain a line icing risk zoning assessment system.

5. The method for safety assessment of high-voltage transmission lines based on intelligent algorithms according to claim 1, characterized in that, Based on the tiered early warning threshold standard, combined with real-time meteorological monitoring data and power grid operation status, a time-series prediction algorithm and a risk propagation model are constructed to obtain an intelligent early warning mechanism for icing risk, including: The real-time meteorological monitoring data is subjected to quality control and missing value repair to obtain an effective meteorological monitoring data stream; The power grid operation status data is collected in real time and features are extracted to obtain a power grid operation status feature sequence; Based on the effective meteorological monitoring data stream and the power grid operation status characteristic sequence, a time series prediction algorithm is constructed to obtain an ice accretion risk prediction model. The icing risk prediction model is combined with the graded early warning threshold standard to determine the threshold trigger and obtain the initial risk warning signal; Based on the initial risk warning signal, a risk propagation model is constructed to obtain an intelligent early warning mechanism for icing risk.

6. The method for safety assessment of high-voltage transmission lines based on intelligent algorithms according to claim 1, characterized in that, The intelligent early warning mechanism for icing risk optimizes the allocation of anti-icing resources and adaptively generates de-icing plans, resulting in differentiated prevention and control strategies and emergency response plans, including: Based on the aforementioned intelligent early warning mechanism for icing risk, a demand assessment of personnel, equipment, and materials for icing prevention is conducted to obtain a resource demand matrix. Based on the resource demand matrix and the existing resource distribution, resource allocation is optimized to obtain an optimized resource allocation scheme for anti-icing; Differentiated de-icing technology selection strategies were designed for different icing risk levels and icing types, resulting in a de-icing technology solution library; Based on the aforementioned de-icing technology solution library, and combined with line characteristics and environmental conditions, an adaptive de-icing solution for the line is generated. The aforementioned anti-icing resource optimization scheme and line de-icing adaptive scheme are spatiotemporally optimized to obtain differentiated prevention and control strategies and emergency response plans.

7. The method for safety assessment of high-voltage transmission lines based on intelligent algorithms according to claim 5, characterized in that, The step of constructing a time series prediction algorithm based on the effective meteorological monitoring data stream and the power grid operation status characteristic sequence to obtain an icing risk prediction model includes: Time-series feature extraction is performed on the effective meteorological monitoring data stream and power grid operation status feature sequence to obtain a multi-source time-series feature matrix; A long short-term memory network model structure is constructed, and the multi-source temporal feature matrix is ​​processed by a sliding window to obtain a training sample set; Based on the training sample set, parameter learning and model training are performed on the long short-term memory network to obtain an initial time series prediction model; The initial time series prediction model is integrated and adaptively adjusted to obtain an enhanced time series prediction model; The enhanced time-series prediction model is evaluated for its icing probability prediction performance and its uncertainty is quantified to obtain an icing risk prediction model.

8. The method for safety assessment of high-voltage transmission lines based on intelligent algorithms according to claim 6, characterized in that, The spatiotemporal collaborative optimization of the anti-icing resource allocation scheme and the line de-icing adaptive scheme yields differentiated prevention and control strategies and emergency response plans, including: By constructing the objective function and constraints for anti-icing resource scheduling optimization, the resource scheduling optimization problem is obtained. An improved particle swarm optimization algorithm is used to solve the resource scheduling optimization problem to obtain the optimal resource scheduling scheme. The execution timing plan and priority setting of the line de-icing adaptive scheme are performed to obtain the de-icing execution timing diagram; By combining the optimal resource scheduling scheme and the de-icing execution timing diagram, spatiotemporal conflict detection and resolution are performed to obtain a collaborative optimization scheme; The collaborative optimization scheme is designed with emergency response procedures and clear division of responsibilities to obtain executable differentiated prevention and control strategies and emergency response plans; The differentiated prevention and control strategies and emergency response plans were simulated and evaluated to obtain a feedback mechanism for plan implementation.

Images