Intelligent line selection optimization method based on medium-voltage power distribution network

Abstract

The invention relates to an intelligent line selection optimization method based on a medium-voltage power distribution network, and the method comprises the following steps: generating a plurality of line planning schemes, and obtaining the environment temperature and pollution degree space-time distribution characteristics of each scheme; establishing a collaborative prediction model, and predicting time series data of future temperature and pollution degree; based on material characteristics and aging test data, establishing an insulating material performance degradation model; constructing an evaluation algorithm and generating a preliminary line candidate scheme by using the prediction data and the degradation model; evaluating the long-term operation risk of the candidate scheme in the prediction environment to obtain a comprehensive reliability evaluation result; solving a Pareto optimal line scheme set meeting a multi-target weight combination by adopting a multi-target optimization model; and carrying out adaptability analysis on the optimal scheme set, and selecting the scheme with the lowest environmental sensitivity as a final optimization scheme. According to the method, the scientificity and the accuracy of line selection are improved, and the safety and the reliability of the power distribution network are enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of power system line selection, and in particular to an intelligent line selection optimization method based on a medium voltage distribution network. Background Art

[0002] With the intensification of global climate change and environmental pollution, the humidity and pollution of the geographical environment faced by medium-voltage distribution networks in the future are expected to increase significantly. In the process of line selection of distribution networks, most of the traditional medium-voltage distribution network line selection only considers the power load situation, and does not fully consider the impact of environmental pollution, environmental humidity and temperature on wire insulators. According to the impact of this high-temperature and polluted environment on the insulation degree of medium-voltage distribution networks, the line selection is optimized. The polluted environment will cause the accumulation of dirt on the surface of the insulator, reducing the insulation performance. High temperature will accelerate the aging and degradation of the insulation material, increasing the risk of leakage and flashover. At present, most practices are to predict the fault and cleanliness of the wire according to the environmental pollution, temperature and humidity of the environment after the power grid is built or in use, but these predictions are not taken into account in the line selection process. If the impact of environmental pollution can be fully considered in the line selection stage, the insulator model and material can be reasonably selected, and the layout of the tower and insulator can be optimized, it will help improve the reliability and safety of the distribution network. Summary of the invention

[0003] In order to solve the problems existing in the above-mentioned prior art, the purpose of the present invention is to provide an intelligent line selection optimization method based on a medium-voltage distribution network. The intelligent line selection optimization method based on a medium-voltage distribution network not only improves the scientificity and accuracy of line selection, but also enhances the safety and reliability of the distribution network, reduces the adverse effects of environmental factors on line operation, and provides a strong guarantee for the long-term stable operation of the power grid.

[0004] The present invention provides an intelligent line selection optimization method based on a medium voltage distribution network, comprising the following steps: S1. Generate multiple line planning schemes through the intelligent line selection system of the medium voltage distribution network, and obtain the temporal and spatial distribution characteristics of the ambient temperature and pollution degree of each line planning scheme; S2. Establish a collaborative prediction model of ambient temperature and pollution level to obtain time series prediction data of temperature and pollution level within a certain time range in the future; S3. According to the types and characteristic parameters of insulator and conductor materials, the accelerated aging test data of the materials under different temperatures and different pollution levels are analyzed to establish a performance degradation model of insulation materials under the multi-stress coupling effect of comprehensive temperature and pollution level. S4. Constructing a core evaluation algorithm of an intelligent line selection system based on the time series prediction data of temperature and contamination degree and the insulation material performance degradation model, and generating preliminary line candidate solutions; S5. Based on the preliminary line candidate schemes, evaluate their long-term operation risks under the predicted temperature and pollution conditions, calculate the insulator flashover probability and line failure rate of different schemes through simulation, and obtain the comprehensive reliability evaluation results of each scheme under the influence of different temperatures and pollution levels; S6. Based on the comprehensive reliability evaluation results, a multi-objective optimization model is used, and in combination with environmental topographic constraints, a Pareto optimal line solution set that satisfies a combination of multiple objective weights is obtained; S7. Perform adaptability analysis on the solutions in the Pareto optimal line solution set, evaluate the sensitivity of the solutions to environmental adaptability, and select the solution with the lowest sensitivity as the optimal intelligent line selection optimization solution for the medium-voltage distribution network.

[0005] Preferably, the step S1 specifically includes: Multiple line planning schemes are generated through the intelligent line selection system of the medium-voltage distribution network, and the topographic data, meteorological observation data and load distribution data of the area to be optimized are obtained from the geographic information system, meteorological department and distribution management system. Through data fusion and correlation analysis, the spatiotemporal distribution characteristics of the ambient temperature and pollution degree of each line planning scheme are obtained.

[0006] Preferably, the step S2 specifically includes: According to the spatiotemporal distribution characteristics of ambient temperature and pollution degree of each line planning scheme, historical ambient temperature data and pollution monitoring data are obtained and analyzed, the temperature change trend and pollution source distribution are predicted, and a collaborative prediction model of ambient temperature and pollution degree is established to obtain time series prediction data of temperature and pollution degree within a certain time range in the future.

[0007] Preferably, the step S3 specifically includes: The types and characteristic parameters of insulator and conductor materials are obtained from the material database. Based on the types and characteristic parameters of insulator and conductor materials, the K-means clustering algorithm is used to analyze the material characteristics and obtain the initial performance distribution of the material under temperature and contamination changes. Designing an accelerated aging test program according to the initial performance distribution, conducting aging tests on material samples under different temperature and contamination conditions, and recording changes in performance indicators of the materials at different time points to form a multidimensional data set including time, temperature, contamination degree and performance indicators; The principal component analysis method was used to reduce the dimensionality of the multidimensional data set, and support vector regression was used to establish the relationship model between temperature, contamination degree and material performance degradation rate; Combining the initial performance distribution of the material and the environmental prediction data, a Monte Carlo simulation method is used to generate a performance degradation path, set a performance degradation path threshold, and determine whether the performance degradation path reaches the performance degradation path threshold; If the performance degradation path reaches the performance degradation path threshold, the material is judged to have reached its service life, and the performance degradation probability distribution function of the material under the multi-stress coupling is obtained by statistically analyzing the performance degradation path, and a performance degradation model of the insulating material under the multi-stress coupling effect of comprehensive temperature and contamination is established; If the performance degradation path does not reach the performance degradation path threshold, it is determined that the material has not reached its service life, and the process of generating the performance degradation path using the Monte Carlo simulation method is repeated until the performance degradation path reaches the performance degradation path threshold.

[0008] Preferably, in step S3, generating the performance degradation path using the Monte Carlo simulation method includes the following steps: The number of simulations is set to A, and the temperature and contamination degree at each time point are sampled according to normal distribution. The sampling results are input into the relationship model to obtain the degradation rate, and the performance degradation value is accumulated and calculated.

[0009] Preferably, the step S4 specifically includes: Based on the time-series prediction data of temperature and contamination level and the insulation material performance degradation model, the core evaluation algorithm of the intelligent line selection system is constructed. The mapping relationship between ambient temperature and contamination level and line insulation level and reliability is obtained through training using machine learning methods. Different combinations of temperature and contamination level are obtained, and preliminary line candidate solutions are generated.

[0010] Preferably, the step S4 further comprises: By acquiring pollution combination data from different regions, simulating line insulation data under different pollution scenarios at different temperatures according to the pollution sampling time interval and pollution sampling method, the samples of the preliminary line candidate solutions are enriched.

[0011] Preferably, the step S5 specifically includes: Acquire long-term temperature prediction data and pollution degree prediction data of the area corresponding to the preliminary line candidate solution, and generate an environmental condition time series data set according to the long-term temperature prediction data and the pollution degree prediction data; The Monte Carlo method is used to sample the environmental condition time series data set, and the insulator flashover voltage is calculated by the insulator flashover theory. The insulator flashover voltage is calculated according to the following formula: U 50% = K*(L+I)*σ*(A+B*d), Among them, U 50% is the 50% flashover voltage; K is the shape factor; L is the creepage distance; l is the overhang distance; σ is the conductivity of the pollution layer; A and B are constants; d is the thickness of the pollution layer; By counting the flashover times in the Monte Carlo simulation results, the flashover probability of the candidate solutions is calculated. Constructing a fault tree, wherein the fault tree includes defining top events, intermediate events and basic events, and obtaining a line failure rate through minimum cut set analysis and calculation; Constructing a comprehensive reliability evaluation index system, wherein the comprehensive reliability evaluation index system includes insulator flashover probability and line failure rate; The weight calculation method based on historical data is used to determine the weight of each evaluation index, and the comprehensive reliability score of the candidate scheme under the influence of different temperatures and pollution levels is calculated; Calculate the cumulative risk index of the candidate solutions and obtain the comprehensive reliability evaluation results.

[0012] Preferably, the step S6 specifically includes: According to the comprehensive reliability evaluation result, a multi-objective optimization model is adopted, and the construction of the multi-objective optimization model includes the following steps: Obtaining a line reliability index, a construction cost index, and an operation and maintenance convenience index, wherein the line reliability index is defined as the complement of the annual average failure rate, the construction cost index is defined as the total investment cost, and the operation and maintenance convenience index is defined as the reciprocal of the average arrival time of maintenance personnel; Taking the line reliability index, the construction cost index and the operation and maintenance convenience index as optimization targets, and establishing the multi-objective optimization model; Combined with the environmental topographic constraints, the Pareto optimal route solution set that meets multiple objective weight combinations is solved.

[0013] Preferably, the step S7 specifically includes: Performing adaptability analysis on the schemes in the Pareto optimal line scheme set, wherein the adaptability analysis includes changing temperature and pollution conditions, and simulating the insulation performance of the line under complex environments of high temperature and heavy pollution, low temperature and light pollution, and seasonal pollution changes; The response surface method is used to construct a sensitivity model. The relative sensitivity coefficients of temperature and pollution degree are calculated through the sensitivity model, and the sensitivity of the scheme to environmental adaptability is evaluated. The scheme with the lowest sensitivity is selected as the optimal intelligent line selection optimization scheme for medium-voltage distribution network.

[0014] The intelligent line selection optimization method based on the medium voltage distribution network described in the present invention has the advantages of: The intelligent line selection optimization method based on the medium voltage distribution network of the present invention generates multiple line planning schemes that consider the spatiotemporal distribution characteristics of ambient temperature and pollution through the intelligent line selection system of the medium voltage distribution network, uses a collaborative prediction model to predict future temperature and pollution changes, combines the performance degradation model of insulating materials under different environments, and constructs a core evaluation algorithm to generate preliminary candidate schemes; evaluates the long-term operation risk of each scheme under the predicted environment through simulation calculation to achieve comprehensive reliability evaluation; adopts a multi-objective optimization model, combined with environmental terrain constraints, solves a Pareto optimal line scheme set that meets multiple objective weight combinations, and selects the optimal line scheme through adaptability analysis and environmental sensitivity evaluation of these schemes. This method not only improves the scientificity and accuracy of line selection, but also enhances the safety and reliability of the distribution network, reduces the adverse effects of environmental factors on line operation, and provides a strong guarantee for the long-term stable operation of the power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a flow chart of an intelligent line selection optimization method based on a medium voltage distribution network described in the present invention. DETAILED DESCRIPTION

[0016] like Figure 1 As shown, the intelligent line selection optimization method based on the medium voltage distribution network described in the present invention comprises the following steps: S1. Generate multiple line planning schemes through the intelligent line selection system of the medium-voltage distribution network, and obtain the spatiotemporal distribution characteristics of the ambient temperature and pollution degree of each line planning scheme; S2. Establish a collaborative prediction model of ambient temperature and pollution level to obtain time series prediction data of temperature and pollution level within a certain time range in the future; S3. According to the types and characteristic parameters of insulator and conductor materials, the accelerated aging test data of the materials under different temperatures and different pollution levels are analyzed to establish a performance degradation model of insulation materials under the multi-stress coupling effect of comprehensive temperature and pollution level. S4. Based on the time series prediction data of temperature and contamination degree and the insulation material performance degradation model, the core evaluation algorithm of the intelligent line selection system is constructed, and preliminary line candidate solutions are generated; S5. Based on the preliminary line candidate schemes, evaluate their long-term operation risks under the predicted temperature and pollution conditions, calculate the insulator flashover probability and line failure rate of different schemes through simulation, and obtain the comprehensive reliability evaluation results of each scheme under the influence of different temperatures and pollution levels; S6. Based on the comprehensive reliability evaluation results, a multi-objective optimization model is used, and combined with environmental topographic constraints, a Pareto optimal line solution set that satisfies multiple objective weight combinations is obtained; S7. Perform adaptability analysis on the solutions in the Pareto optimal line solution set, evaluate the sensitivity of the solutions to environmental adaptability, and select the solution with the lowest sensitivity as the optimal intelligent line selection optimization solution for the medium-voltage distribution network.

[0017] Furthermore, in this embodiment, step S1 specifically includes: Generate multiple line planning schemes through the intelligent line selection system of the medium-voltage distribution network, and obtain the topographic data, meteorological observation data and load distribution data of the area to be optimized from the geographic information system, meteorological department and distribution management system. Through data fusion and correlation analysis, obtain the spatiotemporal distribution characteristics of the ambient temperature and pollution degree of each line planning scheme; Specifically, through the multiple line planning schemes generated by the intelligent line selection system of the medium-voltage distribution network, the topographic data of the area to be optimized are extracted from the geographic information system, the meteorological observation data of the area is obtained from the meteorological department, and the load distribution data is extracted from the distribution management system. These heterogeneous multi-source data are cleaned by outlier detection and processing and missing value filling, and then standardized by the Z-score method and converted into a unified format; The principal component analysis method is used to reduce the dimension of the standardized multi-source data, and the main characteristic variables are extracted. Then, a multi-layer perceptron is used to establish a relationship model between factors such as terrain, weather, and load and ambient temperature and pollution. The model adopts a three-layer structure. The number of nodes in the input layer is the number of main characteristic variables, the number of nodes in the hidden layer is 1.5 times that of the input layer, and the number of nodes in the output layer is 2, corresponding to ambient temperature and pollution, respectively. The back propagation algorithm is used for training, and the ambient temperature and pollution values ​​along the line of each planning scheme are calculated based on the trained model. The calculated ambient temperature and pollution degree values ​​are interpolated in time and space using the Kriging interpolation method to generate raster data of temperature field and pollution degree along the lines of each planning scheme. The raster calculation function in the ArcGIS spatial analysis tool is used to calculate the temporal and spatial distribution characteristic parameters of temperature and pollution degree, including the average value, extreme value, and gradient. According to the temperature field and pollution degree grid data of each planning scheme, combined with the temperature and pollution grade standards in the transmission line design specifications, the weighted summation method is used to score the adaptability of each scheme, and the temperature and pollution degree grid data are reclassified according to the grade standards in the design specifications to obtain grade grid data; Calculate the area proportion of each level and calculate the weighted score according to the preset weight coefficient. The score result serves as an important basis for the optimization planning scheme of the intelligent line selection system; Here is an example: Through the five line planning schemes generated by the intelligent line selection system for medium-voltage distribution networks, the topographic data within the 1,000 square kilometers of the area to be optimized were extracted from the geographic information system, including altitude, slope, and vegetation coverage; Obtain meteorological observation data for the area in the past five years from the meteorological department, including temperature, humidity, and wind speed; Extract load distribution data from the distribution management system, including power consumption and load density; Clean the data, such as detecting and removing outliers with temperatures above 50°C or below -30°C, and filling in missing wind speed data with average values ​​from nearby time points; The Z-score method is used for standardization, and all data are converted into a unified format with a mean of 0 and a standard deviation of 1. The principal component analysis method is used to reduce the dimension of the standardized data, and the principal components that explain 95% of the variance are retained, usually 10-15 main characteristic variables are obtained; Using these characteristic variables as input, a three-layer multi-layer perceptron model is constructed, with 20 hidden layer nodes and the output layer being the ambient temperature and pollution degree; 50,000 sample points were used for training, the learning rate was set to 0.01, and a stable model was obtained after 1,000 iterations; Using the trained model, calculate the ambient temperature and pollution value of each point every 100 meters along the five planning schemes; The obtained temperature and pollution values ​​are interpolated in time and space using the Kriging interpolation method to generate raster data with a resolution of 100 meters; The mean, maximum, minimum, standard deviation and spatial autocorrelation index of temperature and pollution degree for each scheme were calculated using ArcGIS spatial analysis tools; According to the transmission line design specifications, the temperature is divided into five levels: less than 0°C, 0-10°C, 10-20°C, 20-30°C, and greater than 30°C. The pollution level is divided into four levels: light pollution, medium pollution, heavy pollution, and extremely heavy pollution. Calculate the area proportion of each level in each scheme, and calculate the weighted score based on the weights of temperature accounting for 60% and pollution accounting for 40%; Finally, the scores of the five schemes are obtained, and the scheme with the highest score is output as the preferred scheme to the intelligent line selection system for further optimization.

[0018] Furthermore, in this embodiment, step S2 specifically includes: According to the spatiotemporal distribution characteristics of ambient temperature and pollution level of each line planning scheme, historical ambient temperature data and pollution monitoring data are obtained and analyzed, the temperature change trend and pollution source distribution are predicted, and a collaborative prediction model of ambient temperature and pollution level is established to obtain the time series prediction data of temperature and pollution level within a certain time range in the future; Specifically, obtain ambient temperature data and pollution monitoring data provided by the meteorological department and the environmental protection department, including historical records of the areas where each line planning scheme is located in the past ten years; According to the historical records, the 3-times standard deviation method was used to detect and eliminate outliers, and the missing values ​​were filled by multiple interpolation to obtain the standardized ambient temperature and pollution degree data sets. For the standardized ambient temperature data set, the seasonal trend decomposition method is used to extract the temperature change pattern, and the daily temperature forecast value in the next year is obtained by interpolation method; Obtain land use type data and meteorological data from the geographic information system, use the Kriging spatial interpolation algorithm to construct a pollution source distribution map, and establish a relationship model between pollution degree and environmental factors through random forest regression; A collaborative prediction model of ambient temperature and pollution degree is constructed using a long short-term memory neural network. The input of the collaborative prediction model includes temperature distribution, pollution degree distribution and other related environmental factors. After the collaborative prediction model training is completed, the daily temperature and pollution time series prediction data along each line planning scheme in the next year will be output; Here is an example: The ambient temperature data of the areas where each line planning scheme is located in the past ten years were obtained from the meteorological department, and the pollution monitoring data of the corresponding areas were obtained from the environmental protection department. These historical data were cleaned and standardized, and the 3-times standard deviation method was used to detect and eliminate outliers. The multiple interpolation method was used to fill in missing values ​​to obtain the standardized historical ambient temperature and pollution degree data set; The seasonal trend decomposition method is used to analyze the processed historical ambient temperature data, extract the seasonal change pattern and long-term change trend of the temperature, and obtain the daily temperature forecast value in the next year through interpolation. Based on the temperature forecast value and the terrain data in the geographic information system, the temperature distribution along each line planning scheme is calculated using a multivariate linear regression model that takes into account terrain factors such as altitude and slope. At the same time, a preliminary analysis of the correlation between temperature and pollution degree is conducted to provide input for subsequent pollution degree analysis; Based on the processed historical pollution monitoring data, combined with the land use type data and meteorological data in the geographic information system, the Kriging spatial interpolation algorithm is used to construct a pollution source distribution map, and a relationship model between pollution degree and environmental factors is established through random forest regression. This model considers multiple influencing factors such as temperature, humidity, wind speed, and terrain, and calculates the pollution degree distribution along each line planning scheme; A collaborative prediction model of ambient temperature and pollution degree is constructed using a long short-term memory neural network. The network structure includes an input layer, two hidden layers, and an output layer. The Adam optimizer is used for training. The temperature distribution, pollution degree distribution, and other related environmental factors obtained in the previous steps are used as input features. At the same time, spatiotemporal distribution features such as spatial autocorrelation index and temporal autocorrelation coefficient are added as additional inputs. The model is trained and verified, and finally the daily temperature and pollution degree time series prediction data along each line planning scheme in the next year are output; For example, we obtained 3,650 records of daily ambient temperature data for the areas where each line planning scheme is located in the past ten years from the meteorological department, and 520 records of weekly pollution monitoring data for the corresponding areas from the environmental protection department; We cleaned these historical data and used the 3-times standard deviation method to detect outliers. We found 12 outliers in the temperature data and 5 outliers in the pollution data, and removed them. For missing values, there are 23 missing points in the temperature data and 7 missing points in the pollution data, and multiple interpolation method is used to fill these missing values; The Z-score method was used for data standardization, which converted the temperature and pollution data into a standard normal distribution with a mean of 0 and a standard deviation of 1. The seasonal trend decomposition method was applied to the processed temperature data to decompose the time series into three components: trend, seasonality, and residual. The trend component shows that the temperature has risen by 1.2°C over the decade, and the seasonal component shows that the average summer temperature is 12.5°C higher than the winter temperature; Use the ARIMA model to forecast the trend and seasonal components and obtain the temperature forecast for each day in the next year; Combined with the terrain data in the geographic information system, including altitude and slope aspect, a multivariate linear regression model was established, and the predicted temperature = reference temperature + (-0.006*altitude) + (0.01*cos(slope aspect)) was used to calculate the temperature distribution along each line planning scheme; The Pearson correlation coefficient between temperature and filthiness was calculated to be 0.65, indicating that the two were moderately positively correlated; Using the Kriging spatial interpolation algorithm, a pollution source distribution map with a resolution of 100 meters was constructed based on the pollution monitoring data of 500 sampling points; The random forest regression model was used, and the input variables included 10 features such as temperature, humidity, wind speed, altitude, slope aspect, and land use type. A pollution degree prediction model was established through the integration of 500 decision trees, and the root mean square error of the model was 0.18; Construct a long short-term memory neural network with 15 nodes in the input layer and 64 nodes in each of the two hidden layers. Use the ReLU activation function, and the two nodes in the output layer correspond to the predicted values ​​of temperature and pollution degree. Using the Adam optimizer, the learning rate was set to 0.001, 100 epochs were trained on 80% of the data, and the remaining 20% ​​of the data were used for validation. The final model had a mean absolute error of 0.5°C for temperature and 0.08 for contamination on the test set.

[0019] Furthermore, in this embodiment, step S3 specifically includes: The types and characteristic parameters of insulator and conductor materials are obtained from the material database. Based on the types and characteristic parameters of insulator and conductor materials, the K-means clustering algorithm is used to analyze the material characteristics and obtain the initial performance distribution of the material under temperature and contamination changes. Design an accelerated aging test plan based on the initial performance distribution, conduct aging tests on material samples under different temperatures and contamination conditions, and record the changes in performance indicators of the materials at different time points to form a multidimensional data set containing time, temperature, contamination and performance indicators; record the changes in performance indicators of the materials at different time points through automated testing equipment; The principal component analysis method was used to reduce the dimensionality of the multidimensional data set, and support vector regression was used to establish the relationship model between temperature, contamination degree and material performance degradation rate; Combining the initial performance distribution of the material and the environmental prediction data, the Monte Carlo simulation method is used to generate the performance degradation path, set the performance degradation path threshold, and determine whether the performance degradation path reaches the performance degradation path threshold; If the performance degradation path reaches the performance degradation path threshold, the material is judged to have reached its service life. The performance degradation path is statistically analyzed, and the kernel density estimation method is used to obtain the performance degradation probability distribution function of the material under multi-stress coupling, and a performance degradation model of the insulation material under multi-stress coupling that integrates the influence of temperature and contamination is established. If the performance degradation path does not reach the performance degradation path threshold, it is determined that the material has not reached its service life, and the process of generating the performance degradation path using the Monte Carlo simulation method is repeated until the performance degradation path reaches the performance degradation path threshold; Specifically, the types and characteristic parameters of insulator and conductor materials are obtained from the material database, including the chemical composition, physical properties, and key electrical performance indicators of the materials. The material characteristics are analyzed using the K-means clustering algorithm. The number of clusters is set to 5 and the number of iterations is set to 100. The initial performance distribution of different types of materials under temperature and pollution changes is obtained, and the center point, standard deviation, and dispersion of each cluster are calculated to provide a basis for the design of subsequent test plans. According to the initial performance distribution of the material, an accelerated aging test scheme is designed. Aging tests are carried out on material samples under different combinations of temperature range -10℃ to 60℃ and contamination levels from light contamination to very heavy contamination. Automated testing equipment is used to record the changes in performance indicators of the material at different time points. Specific test parameters include dielectric strength, surface resistivity, and mechanical strength, forming a multidimensional data set containing time, temperature, contamination level, and performance indicators; The principal component analysis method is used to reduce the dimensionality of the multidimensional data set, retain the principal components that explain 90% of the variance, and extract the main factors affecting the degradation of material performance. Then, support vector regression is used to establish a relationship model between temperature, contamination degree and material performance degradation rate. The radial basis function kernel is selected, and the parameters C and gamma are optimized through grid search. The model takes into account the interaction between temperature and contamination degree, and reflects the comprehensive influence of multi-stress coupling on material performance. Based on the established relationship model between temperature, contamination degree and material performance degradation rate, combined with the initial performance distribution of the material and the temperature and contamination degree time series prediction data in the use environment, the Monte Carlo simulation method is used to generate performance degradation paths. These paths are statistically analyzed, and the kernel density estimation method is used to obtain the performance degradation probability distribution function of different materials under multi-stress coupling. Finally, a performance degradation model of insulating materials under multi-stress coupling with the comprehensive influence of temperature and contamination degree is formed.

[0020] Furthermore, in this embodiment, in step S3, using the Monte Carlo simulation method to generate the performance degradation path includes the following steps: Set the number of simulations to A, perform normal distribution sampling on the temperature and contamination degree at each time point, input the sampling results into the relational model, obtain the degradation rate, and accumulate the performance degradation value; the optimal value of the number of simulations A is 10000; An example of step S3 is as follows: The characteristic parameters of 100 insulator and conductor materials were extracted from the material database, including chemical composition (e.g. SiO2 content, Al2O3 content), physical properties (e.g. density, thermal expansion coefficient) and electrical properties (e.g. dielectric constant, breakdown voltage); The K-means clustering algorithm was used, the number of clusters was set to 5, and after 100 iterations, the initial performance distribution of the five material categories was obtained; The dielectric strength of the center point of cluster 1 is 22kV / mm, the surface resistivity is 10^12Ω, and the mechanical strength is 500MPa. Based on the clustering results, a temperature range of -10℃ to 60℃, a gradient of 10℃, and a contamination level ranging from 0.03mg / cm² light contamination to 0.60mg / cm² very heavy contamination, a total of 4 levels of accelerated aging test schemes were designed; The material performance indicators were recorded at 0, 100, 500, 1000, and 2000 hours using automated testing equipment. It was found that after 1000 hours at 20°C and light pollution conditions, the dielectric strength of cluster 1 material decreased to 20.5 kV / mm, and the surface resistivity decreased to 9.8×10 11 Ω, and the mechanical strength dropped to 480 MPa. Principal component analysis was applied to the collected multidimensional data sets, and the first three principal components were retained, with a cumulative explained variance of 91.5%; The performance degradation model was established using support vector regression, the RBF kernel function was selected, and the optimal parameters C=10 and gamma=0.01 were determined through grid search; The model shows that at 30°C and 0.20mg / cm² medium contamination conditions, the material performance degradation rate is 0.5% / 1000 hours; Based on the relationship model between temperature, pollution degree and material performance degradation rate, combined with the predicted temperature and pollution degree time series data, for example, the average temperature of a line in the next year is 25℃, and the pollution degree is 0.15mg / cm², and 10,000 Monte Carlo simulations are performed; The kernel density estimation method was used to obtain the performance degradation probability distribution function, which predicted that within the 5-year service life, there was a 95% probability that the performance of the material would decrease by no more than 8.2%.

[0021] Furthermore, in this embodiment, step S4 specifically includes: Based on the time series prediction data of temperature and contamination and the insulation material performance degradation model, the core evaluation algorithm of the intelligent line selection system is constructed. The mapping relationship between ambient temperature and contamination and line insulation level and reliability is obtained through machine learning training, and different temperature and contamination combinations are obtained, and preliminary line candidate solutions are generated; Specifically, obtaining the time series prediction data of temperature and pollution degree, and preprocessing the time series prediction data, wherein the preprocessing includes using the 3 times standard deviation method to detect outliers, using the linear interpolation method to fill missing values, and performing data standardization; According to the preprocessed time series prediction data, the sliding window method is used to convert the time series prediction data into a feature matrix and a label matrix, wherein the feature matrix contains historical temperature and pollution degree data, and the label matrix contains the temperature and pollution degree values ​​at future time points; A deep neural network is used to build a mapping relationship model between ambient temperature and contamination degree and line insulation level and reliability. The deep neural network contains 3 fully connected layers and 2 LSTM layers. The input layer receives the feature matrix, and the output layer predicts the insulation level and reliability index. Obtain the output results of the deep neural network model and the prediction results of the insulation material performance degradation model; According to the output results and prediction results, weighted fusion is performed in a ratio of 6:4 to obtain the final evaluation index; Different combinations of temperature and pollution degree are traversed. If the evaluation index meets the preset threshold, the temperature and pollution degree combination is used as a preliminary line candidate solution; Here is an example: According to the temperature and pollution time series forecast data obtained in the early stage, the data is preprocessed, including using the 3-times standard deviation method for outlier detection, using the linear interpolation method to fill missing values, and data standardization. Then, the time series data is converted into a feature matrix and a label matrix using the sliding window method. The window size is set to 30 days and the step length is 1 day. The feature matrix contains historical temperature and pollution data, and the label matrix contains the temperature and pollution values ​​at future time points. A deep neural network is used to construct a mapping relationship model between ambient temperature and contamination and line insulation level and reliability. The network structure includes three fully connected layers with 64, 32, 16 nodes and two LSTM layers with 128 units each. The input layer receives the feature matrix, and the output layer predicts the insulation level and reliability index. The model is trained using the back propagation algorithm and the Adam optimizer, and the early stopping method is used to prevent overfitting. The judgment criterion is that the validation set loss has not improved for five consecutive epochs. The trained deep neural network model is integrated with the insulation material performance degradation model to build the core evaluation algorithm of the intelligent line selection system. The algorithm weights and fuses the output of the deep neural network model with the prediction results of the insulation material performance degradation model in a ratio of 6:4 to form the final evaluation index. The input includes the predicted temperature and pollution time series data and line design parameters, and the output is the insulation level and reliability evaluation results of the line. Based on the core evaluation algorithm, different combinations of temperature and contamination degree are traversed through the grid search method. The temperature search range is -10℃ to 60℃, with a step size of 5℃, and the contamination degree search range is 0.03mg / cm² to 0.60mg / cm², with a step size of 0.03mg / cm². The line insulation level and reliability index under each combination are calculated. According to the preset evaluation index thresholds, such as the insulation level is not less than 1.5 times the nominal value, and the reliability index is not less than 0.99, the temperature and contamination degree combinations that meet the requirements are screened out, and combined with the terrain data in the geographic information system, the preliminary line candidate schemes that meet the conditions are generated; The intelligent line selection system first processed the temperature and pollution time series data of a certain area over the past five years, a total of 1,825 days of records; Thirty-seven temperature outliers and 24 pollution outliers were detected using the 3-times standard deviation method, and 53 missing data points were filled by linear interpolation; After data standardization, a sliding window method with a window size of 30 days and a step size of 1 day was used to generate 1795 sets of feature matrices and label matrices; The constructed deep neural network contains 3 fully connected layers with 64, 32, and 16 nodes and 2 LSTM layers with 128 units each. It is trained using the Adam optimizer with a learning rate of 0.001. On the validation set, the early stopping mechanism is triggered when the loss function improves by less than 0.0001 for 5 consecutive epochs, and convergence is finally achieved at the 78th epoch. The average relative error of the insulation level predicted by the trained deep neural network model is 2.3%, and the average absolute error of the reliability index is 0.0035; This model is combined with the insulation material performance degradation model established earlier in a 6:4 ratio to form the core evaluation algorithm; Using this algorithm, a grid search was performed on the temperature range of -10°C to 60°C, with a step size of 5°C, and the contamination range of 0.03mg / cm² to 0.60mg / cm², with a step size of 0.03mg / cm², generating a total of 225 combinations of temperature and contamination; The thresholds of insulation level not less than 1.5 times the nominal value and reliability index not less than 0.99 were set, and 43 combinations that met the conditions were screened out; Combined with terrain data in the geographic information system, such as altitude, slope, and vegetation cover, five preliminary line candidate plans were finally generated, with line lengths ranging from 50km to 80km, an average insulation level of 1.68 times the nominal value, and an average reliability index of 0.9937.

[0022] Furthermore, in this embodiment, step S4 also includes: By obtaining pollution combination data from different regions, simulating line insulation data under different pollution scenarios at different temperatures based on pollution sampling time intervals and pollution sampling methods, we enrich the samples of preliminary line candidate solutions; Specifically, the pollution combination data of different regions in the environmental monitoring database are obtained, and the pollution combination data include pollution degree, chemical composition and particle size distribution; Seasonal and trend decomposition algorithms are used to process pollution combination data to obtain seasonal variation patterns and long-term variation trends of pollution levels; According to the seasonal variation pattern and long-term variation trend, a temperature-pollution relationship model is constructed. The temperature-pollution relationship model is used to simulate the effect of temperature on the adhesion, accumulation and cleaning process of pollutants. The pollution scenario data generated by the temperature-pollution relationship model is combined with the line insulation characteristic data to calculate the line insulation level under different pollution scenarios; If the line insulation level meets the preset conditions, the line insulation level is applied to the preliminary line candidate scheme to generate insulation level data of different sections of each candidate line, wherein the preset conditions include a threshold range of the line insulation level; Here is an example: Obtain pollution combination data of different regions from the environmental monitoring database, including pollution degree, chemical composition and particle size distribution, and record the sampling time interval and sampling method of each data point, such as direct measurement method or equivalent salt density method, and classify and organize these data by geographical location and time series to form a pollution characteristic data set; For the data measured by the equivalent salt density method, the data are converted into the pollution value of the direct measurement method through a pre-established conversion function to ensure the consistency and comparability of the data; The pollution characteristic data set is processed using seasonal and trend decomposition algorithms to extract seasonal variation patterns and long-term variation trends of pollution. Based on these patterns and trends, the cubic spline interpolation method is used to generate pollution data with higher temporal resolution, thereby filling the data gaps caused by different sampling time intervals. The generated high-time-resolution pollution degree data are matched with the temperature data at the corresponding time points to form a temperature-pollution degree paired data set; Based on the temperature-pollution degree paired data set, a temperature-pollution relationship model is constructed, which considers the effect of temperature on the adhesion, accumulation and cleaning process of contamination. The random forest regression method is used to establish the quantitative relationship between temperature and pollution degree change rate. The input variables include meteorological factors such as temperature, humidity, and wind speed, and the output variable is the pollution degree change rate. Through this model, pollution scenarios under different temperature conditions are simulated to generate pollution degree time series data covering the whole year. The simulated pollution scenario data is combined with the line insulation characteristic data, and the line insulation level under different pollution scenarios is calculated using the insulator flashover theory; The specific calculation process takes into account key parameters such as the conductivity of the contamination layer, the surface area of ​​the insulator, and the leakage distance, and obtains the insulation level by solving the key flashover voltage equation; Considering the impact of temperature on the performance of insulation materials, the insulation level calculation results are adjusted to ultimately generate a comprehensive data set that includes geographic location, temperature, pollution level, and insulation level; The comprehensive data set is applied to the preliminary line candidate schemes, and the corresponding insulation level data are generated for different sections of each candidate line through interpolation and extrapolation methods, thereby enriching and improving the sample library of the preliminary line candidate schemes; In a distribution network planning project, a total of 3,650 groups of pollution combination data were obtained from the environmental monitoring database in 5 different regions, including pollution degree 0.03-0.60 mg / cm², main chemical components Na+, Cl-, SO4²- content and particle size distribution, i.e. PM2.5 and PM10 proportion; Sampling intervals ranged from daily to monthly, and sampling methods included direct measurement and equivalent salt density methods; Through the pre-established conversion function y=1.2x+0.05, y is the direct measurement value, x is the equivalent salt density value, all data are unified into the pollution value of the direct measurement method; Use the STL algorithm to decompose the pollution characteristic data and extract seasonal components and long-term trends; Based on the decomposition results, cubic spline interpolation method is used to generate daily pollution degree data, and the interpolation node interval is 7 days; The generated high-resolution pollution degree data were matched with the temperature data to form 36,500 sets of temperature-pollution degree paired data; A temperature-pollution relationship model was constructed using random forest regression, with input variables including temperature, relative humidity, and wind speed, and the output being the daily pollution change rate; The model uses 500 decision trees with a maximum depth of 10 and a minimum number of leaf node samples of 5. Through this model, daily pollution scenario data for five regions throughout the year were simulated and generated. Combined with line insulation characteristic data, such as insulator creepage distance of 300-400mm and shape factor of 0.7-0.9, the insulator flashover theory was used to calculate the line insulation level under different pollution scenarios; The conductivity of the contamination layer and the surface area parameters of the insulator are considered in the calculation, and the insulation level is obtained by solving the key flashover voltage equation; The effect of temperature on insulation material performance was adjusted using correction factors, resulting in a comprehensive dataset of 182,500 data points covering insulation levels in five regions, 365 days, and under different temperature and pollution conditions.

[0023] Furthermore, in this embodiment, step S5 specifically includes: Obtaining long-term temperature prediction data and pollution degree prediction data for the area corresponding to the preliminary line candidate scheme, and generating an environmental condition time series data set based on the long-term temperature prediction data and pollution degree prediction data; The Monte Carlo method is used to sample the environmental condition time series data set, and the insulator flashover voltage is calculated by the insulator flashover theory. The insulator flashover voltage is calculated according to the following formula: U 50% = K*(L+I)*σ*(A+B*d), Among them, U 50%is the 50% flashover voltage; K is the shape factor; L is the creepage distance; l is the overhang distance; σ is the conductivity of the pollution layer; A and B are constants; d is the thickness of the pollution layer; By counting the flashover times in the Monte Carlo simulation results, the flashover probability of the candidate solutions is calculated. Construct a fault tree, which includes defining top events, intermediate events and basic events, and calculate the line failure rate through minimum cut set analysis; Construct a comprehensive reliability evaluation index system, which includes insulator flashover probability and line failure rate; The weight calculation method based on historical data is used to determine the weight of each evaluation index, and the comprehensive reliability score of the candidate scheme under the influence of different temperatures and pollution levels is calculated; Calculate the cumulative risk index of candidate solutions and obtain comprehensive reliability evaluation results; Specifically, based on the geographic information of the preliminary candidate route plan, the long-term temperature forecast data of the corresponding area is extracted from the meteorological database, and the corresponding pollution degree forecast data is obtained from the multivariate regression pollution degree forecast model based on environmental factors. The model considers the influence of temperature, humidity, wind speed and other factors on pollution degree; The two sets of data are matched in time series to form the environmental condition time series data set for each candidate solution. The Monte Carlo method is used to perform 10,000 simulation tests on each candidate solution. In each simulation, samples from the environmental condition time series data set are randomly selected. The insulator flashover voltage under the environmental conditions is calculated by combining the insulator characteristic parameters and line operation parameters through the insulator flashover theory. The insulator flashover voltage is calculated according to the following formula: U 50% = K*(L+I)*σ*(A+B*d), Among them, U 50% is the 50% flashover voltage; K is the shape factor; L is the creepage distance; l is the overhang distance; σ is the conductivity of the pollution layer; A and B are constants; d is the thickness of the pollution layer; If the calculated actual voltage exceeds the flashover voltage, it is determined that a flashover has occurred; Based on the Monte Carlo simulation results, the number of insulator flashovers of each candidate solution under different combinations of temperature and pollution degree is counted, and the flashover probability is calculated; According to the line topology and protection device configuration, the line fault rate is derived using the fault tree analysis method; The fault tree construction process includes defining top events (e.g., line faults), intermediate events (e.g., equipment failures), external factors, and basic events (e.g., insulator flashovers, lightning strikes); The line failure rate is calculated by minimum cut set analysis, and a comprehensive reliability evaluation index system is constructed by using the analytic hierarchy process, taking the insulator flashover probability and line failure rate as evaluation factors. Use a weight calculation method based on historical data to determine the weight of each factor, such as the information entropy weight method; The comprehensive reliability score of each candidate solution under the influence of different temperatures and contamination levels is calculated, and the time dimension is introduced to calculate the cumulative risk index of 5 and 10 years, quantify the long-term operation risk, and finally output the short-term and long-term comprehensive reliability evaluation results of each solution; Here is an example: In a distribution network planning project, a long-term operation risk assessment was conducted for five preliminary line candidate solutions, and temperature forecast data for the next 20 years was extracted from the meteorological database, with a temperature range of -10°C to 40°C; A multiple regression model based on environmental factors was used to predict the degree of contamination, taking into account temperature, relative humidity and wind speed, and the range of contamination was 0.03-0.60 mg / cm²; The two sets of data were matched at hourly intervals to form a time series data set of environmental conditions with 175,200 data points. 10,000 Monte Carlo simulations were performed on each line, and environmental condition samples were randomly selected; The flashover voltage is calculated using the insulator flashover theory, where the insulator creepage distance is 380 mm, the overhang distance is 150 mm, and the shape factor is 0.8; By comparing the actual voltage with the flashover voltage, it is determined whether flashover occurs. The statistical results show that the average flashover probability of the five lines is 0.015%, 0.022%, 0.018%, 0.020% and 0.017% respectively; Construct a fault tree analysis model, where the top event is defined as "line fault", the intermediate events include "insulation failure" and "external impact", and the basic events include "insulator flashover", "lightning strike" and "mechanical damage"; Through the minimum cut set analysis, the average annual failure rates of the five lines were calculated to be 0.32, 0.45, 0.38, 0.41 and 0.35 times / 100 km·year respectively; The analytic hierarchy process was used to construct an evaluation index system, including flashover probability, failure rate and maintenance cost. The information entropy weight method was used to calculate the weight of each index, and the comprehensive reliability scores of the five lines were 85.2, 79.8, 82.6, 81.3 and 83.9 respectively. Calculation of the 5- and 10-year cumulative risk indicators showed that after 10 years the scores would drop to 81.7, 75.2, 78.9, 77.1 and 80.3, respectively.

[0024] Furthermore, in this embodiment, step S6 specifically includes: According to the comprehensive reliability evaluation results, a multi-objective optimization model is adopted. The construction of the multi-objective optimization model includes the following steps: Obtain line reliability index, construction cost index and operation and maintenance convenience index. Define the line reliability index as the complement of the annual average failure rate, the construction cost index as the total investment cost, and the operation and maintenance convenience index as the inverse of the average arrival time of maintenance personnel. Take line reliability index, construction cost index and operation and maintenance convenience index as optimization targets and establish a multi-objective optimization model; Combined with the environmental topography constraints, the Pareto optimal route solution set that meets multiple objective weight combinations is solved; Environmental topography and geomorphic constraints are obtained by using geographic information system to extract environmental topography and geomorphic data; Generate a continuous terrain surface function through Kriging interpolation method, and transform the terrain surface function into the constraint condition of the optimization model; According to the preset population size, evolutionary generations, crossover probability and mutation probability parameters, a non-dominated sorting genetic algorithm is used to solve the multi-objective optimization model; For discrete decision variables, grid processing is used; Through iterative optimization, non-dominated solutions are obtained to form the Pareto optimal frontier; If the Pareto optimal front is obtained, the solution on the Pareto optimal front is post-processed; The fuzzy analytic hierarchy process is used to determine the weight range; uniform sampling is performed within the weight range to generate weight combinations; Calculate the weighted objective function value, and select the optimal solution under different weight combinations according to the weighted objective function value; Here is an example: According to the results of the previous evaluation, a multi-objective optimization model is constructed, in which the line reliability index f1(x) is defined as 1 minus the average failure rate of the previous year, the construction cost index f2(x) is defined as the total investment cost, and the operation and maintenance convenience index f3(x) is defined as the inverse of the average arrival time of maintenance personnel, where x represents the decision variable vector of the line plan, including line path, tower location, and conductor type parameters; These objective functions and environmental constraints together form the basis of the optimization model, which uses geographic information systems to extract environmental topographic data, including altitude, slope, and vegetation cover information; The Kriging interpolation method is used to generate continuous terrain surface functions, and these data are converted into constraints for the optimization model. For example, g1(x)≤0 means that the slope of the line does not exceed 30 degrees, and g2(x)≤0 means that the minimum distance between the line and the environmentally sensitive area is not less than 500 meters. These constraints and the objective function together define the feasible solution space, and a non-dominated sorting genetic algorithm is used to solve the multi-objective optimization problem; Based on the problem scale and convergence analysis, the population size is set to 200, the evolutionary generations to 1000, the crossover probability to 0.9, and the mutation probability to 0.1; For discrete decision variables, such as tower locations, grid processing is used to divide the continuous space into 50m × 50m grids, and towers can only be placed at grid intersections; A series of non-dominated solutions are obtained through repeated iterative optimization to form the Pareto optimal frontier. The solutions on the Pareto optimal frontier are post-processed and the weight range is determined by fuzzy analytic hierarchy process, for example, w1∈[0.3,0.5], w2∈[0.2,0.4], w3∈[0.2,0.4]. Uniform sampling is performed within this range to generate 100 sets of weight combinations. Calculate the weighted objective function value F(x)=w1f1(x)+w2f2(x)+w3f3(x), select the optimal solution under different weight combinations, and finally obtain a set of Pareto optimal route solutions that meet multiple objective weight combinations; In a distribution network planning project, a multi-objective optimization model was constructed for the optimal design of a 110kV transmission line; The line reliability index f1(x) is defined as 1 minus the average annual failure rate, with an initial value of 0.9975. The construction cost index f2(x) is defined as the total investment cost, with an initial estimate of 100 million yuan. The operation and maintenance convenience index f3(x) is defined as the inverse of the average arrival time of maintenance personnel, with an initial value of (0.5 hours). -1 ; GIS was used to extract topographic data within an area of ​​100 km × 2 km along the route, including altitude, slope, and vegetation coverage; Ordinary Kriging interpolation method is used to generate a continuous terrain surface function with a resolution of 10m. The constraints are set as follows: the line slope does not exceed 30°, and the minimum distance from environmentally sensitive areas is not less than 500m; The NSGA-II algorithm was used for optimization, the population size was set to 200, and the evolution was 1000 generations; The line path is discretized into 2000 50m×50m grids, and the tower locations are restricted to the grid intersections; After optimization, the Pareto frontier composed of 300 non-dominated solutions was obtained. The fuzzy analytic hierarchy process was used to determine the weight range as w1∈[0.35,0.45], w2∈[0.25,0.35], w3∈[0.25,0.35]. Within this range, 100 groups of weight combinations were uniformly sampled and the weighted objective function value was calculated. Finally, 10 typical Pareto optimal solutions were selected, among which the index values ​​of the best solution were f1(x)=0.9988, f2(x)=95 million yuan, f3(x)=(0.67 hours) -1 Compared with the initial solution, reliability increased by 0.13%, cost decreased by 5%, and operation and maintenance efficiency increased by 34%.

[0025] Furthermore, in this embodiment, step S7 specifically includes: Conduct adaptability analysis on the schemes in the Pareto optimal line scheme set, including changing temperature and pollution conditions, and simulating the insulation performance of the line under complex environments of high temperature and heavy pollution, low temperature and light pollution, and seasonal pollution changes; The response surface method is used to construct a sensitivity model. The relative sensitivity coefficients of temperature and pollution are calculated through the sensitivity model. The sensitivity of the scheme to environmental adaptability is evaluated, and the scheme with the lowest sensitivity is selected as the optimal intelligent line selection optimization scheme for medium-voltage distribution networks. Specifically, the key parameters of each solution in the optimal line solution set are obtained, and a digital twin model is constructed according to the key parameters. The digital twin model includes a line geometry model and an electrical characteristic model. Receive historical temperature and pollution degree data, use ARIMA model to fit the time series data of temperature and pollution degree, and obtain the time series change model of temperature and pollution degree; Generate temperature and pollution degree combination samples according to the time series variation model, input the combination samples into the digital twin model, and obtain the environmental parameters of the electric field simulation; If the maximum electric field strength on the insulator surface exceeds the preset flashover field strength threshold, flashover is determined to have occurred. The flashover probability and insulation margin of each scheme under different environmental conditions are statistically analyzed to obtain the insulation performance index. The response surface method is used to construct a sensitivity model. The relative sensitivity coefficients of temperature and pollution are calculated through the sensitivity model to determine the comprehensive sensitivity of each scheme. The scheme with the lowest comprehensive sensitivity is selected as the intelligent line selection optimization scheme for the medium-voltage distribution network. Here is an example: Extract key parameters of each solution from the optimal line solution set, including line path, tower location, conductor type and insulator configuration, and build a digital twin model of each solution, including line geometry model and electrical characteristic model; At the same time, the historical temperature and pollution data of the area are obtained from the meteorological database, and the time series data of temperature and pollution are fitted using the ARIMA model to establish a time series change model of temperature and pollution to predict the changes in environmental conditions in the next year; The Monte Carlo method is used to generate 10,000 sets of temperature and pollution combination samples, covering high temperature heavy pollution, such as 40℃, 0.5mg / cm², low temperature light pollution, such as -10℃, 0.1mg / cm², and seasonal pollution changes, such as 0.3mg / cm² in spring and 0.1mg / cm² in winter, etc., to ensure that the samples are evenly distributed in the temperature range of -10℃ to 40℃ and the pollution degree range of 0.1mg / cm² to 0.5mg / cm²; These samples are input into the digital twin model of each scheme as the environmental parameters for electric field simulation. The electric field strength and leakage current of the insulator surface of each scheme under different temperature and pollution degree combinations are calculated using electric field simulation software; According to the insulator flashover theory, flashover is determined to occur when the calculated maximum electric field strength on the insulator surface exceeds 21.2kV / cm and the AC flashover field strength is 50%; The flashover probability and insulation margin of each scheme under different environmental conditions were statistically analyzed to obtain the insulation performance index. The sensitivity coefficient of the insulation performance index relative to the change of environmental conditions was calculated for each scheme. The response surface method was used to construct the sensitivity model. The specific steps included designing an orthogonal test scheme and selecting 5 levels of temperature and contamination degree. Conduct electric field simulation tests; establish a quadratic polynomial regression model, verify the model significance through variance analysis, calculate the relative sensitivity coefficients of temperature and contamination, and the comprehensive sensitivity is defined as the square root of the sum of the squares of the two coefficients; The comprehensive sensitivity of each scheme is compared, and the scheme with the lowest comprehensive sensitivity is selected as the final medium-voltage distribution network intelligent line selection optimization scheme. In a medium-voltage distribution network intelligent line selection project, the adaptability of 5 optimal line schemes is analyzed; Build a digital twin model of each line, including a 110km long line geometry model and an electrical characteristic model that takes into account the 110kV voltage level; The ARIMA (2,1,2) model was used to fit the temperature and pollution data of the past 10 years to predict the changes in environmental conditions in the next year. The temperature range is -15℃ to 45℃, and the pollution range is 0.08mg / cm² to 0.63mg / cm². The Monte Carlo method was used to generate 10,000 sets of combined samples of temperature and pollution, including a high-temperature heavy pollution scenario of 40°C and 0.55 mg / cm², a low-temperature light pollution scenario of -12°C and 0.09 mg / cm², and a seasonal pollution change scenario of 0.35 mg / cm² in spring and 0.12 mg / cm² in winter. COMSOL multi-physics simulation software was used to calculate the insulator surface electric field strength and leakage current of each scheme under different environmental conditions. Flashover is determined to occur when the maximum electric field strength exceeds 21.2kV / cm; The statistical results show that the annual average flashover probabilities of the five line schemes are 0.015%, 0.022%, 0.018%, 0.020% and 0.017% respectively; The response surface methodology was used to construct the sensitivity model, 25 groups of orthogonal experiments were designed, and a quadratic polynomial regression model was established. The variance analysis showed that the F value of the model was 78.6, and the p value was less than 0.001, which was significant. The relative sensitivity coefficients of temperature and pollution degree were calculated, and the comprehensive sensitivities of the five lines were 0.082, 0.115, 0.097, 0.106, and 0.089, respectively; Finally, the first line scheme with the lowest comprehensive sensitivity was selected as the optimal intelligent line selection scheme for the medium-voltage distribution network. This scheme has the smallest insulation performance fluctuation under different environmental conditions and the strongest adaptability.

[0026] In the description of the present invention, it is necessary to understand that the directions or positional relationships indicated by directional words such as "front, back, up, down, left, right", "lateral, vertical, horizontal" and "top, bottom" are usually based on the directions or positional relationships shown in the drawings. They are only for the convenience of describing the present invention and simplifying the description. Unless otherwise specified, these directional words do not indicate or imply that the device or element referred to must have a specific direction or be constructed and operated in a specific direction. Therefore, they cannot be understood as limiting the scope of protection of the present invention.

[0027] For those skilled in the art, various other corresponding changes and deformations can be made according to the technical solutions and concepts described above, and all of these changes and deformations should fall within the protection scope of the claims of the present invention.

Claims

1. An intelligent line selection optimization method based on a medium voltage distribution network, characterized in that: The following steps are involved: S1. Generate multiple line planning schemes through the intelligent line selection system of the medium voltage distribution network, and obtain the temporal and spatial distribution characteristics of the ambient temperature and pollution degree of each line planning scheme; S2. Establish a collaborative prediction model of ambient temperature and pollution level to obtain time series prediction data of temperature and pollution level within a certain time range in the future; S3. According to the types and characteristic parameters of insulator and conductor materials, the accelerated aging test data of the materials under different temperature and different pollution degree combinations are analyzed to establish the insulation material performance degradation model under the multi-stress coupling effect of comprehensive temperature and pollution degree; S4. Constructing a core evaluation algorithm of an intelligent line selection system based on the time series prediction data of temperature and contamination degree and the insulation material performance degradation model, and generating preliminary line candidate solutions; S5. Based on the preliminary line candidate schemes, evaluate their long-term operation risks under the predicted temperature and pollution conditions, calculate the insulator flashover probability and line failure rate of different schemes through simulation, and obtain the comprehensive reliability evaluation results of each scheme under the influence of different temperatures and pollution levels; S6. Based on the comprehensive reliability evaluation results, a multi-objective optimization model is used, and in combination with environmental topographic constraints, a Pareto optimal line solution set that satisfies a combination of multiple objective weights is obtained; S7. Perform adaptability analysis on the solutions in the Pareto optimal line solution set, evaluate the sensitivity of the solutions to environmental adaptability, and select the solution with the lowest sensitivity as the optimal intelligent line selection optimization solution for the medium-voltage distribution network.

2. The intelligent line selection optimization method based on the medium voltage distribution network according to claim 1 is characterized in that: The step S1 specifically includes: Multiple line planning schemes are generated through the intelligent line selection system of the medium-voltage distribution network, and the topographic data, meteorological observation data and load distribution data of the area to be optimized are obtained from the geographic information system, meteorological department and distribution management system. Through data fusion and correlation analysis, the spatiotemporal distribution characteristics of the ambient temperature and pollution degree of each line planning scheme are obtained.

3. The intelligent line selection optimization method based on the medium voltage distribution network according to claim 1 is characterized in that: The step S2 specifically includes: According to the spatiotemporal distribution characteristics of ambient temperature and pollution degree of each line planning scheme, historical ambient temperature data and pollution monitoring data are obtained and analyzed, the temperature change trend and pollution source distribution are predicted, and a collaborative prediction model of ambient temperature and pollution degree is established to obtain time series prediction data of temperature and pollution degree within a certain time range in the future.

4. The intelligent line selection optimization method based on the medium voltage distribution network according to claim 1 is characterized in that: The step S3 specifically includes: The types and characteristic parameters of insulator and conductor materials are obtained from the material database. Based on the types and characteristic parameters of insulator and conductor materials, the K-means clustering algorithm is used to analyze the material characteristics and obtain the initial performance distribution of the material under temperature and contamination changes. Designing an accelerated aging test program according to the initial performance distribution, conducting aging tests on material samples under different temperature and contamination conditions, and recording changes in performance indicators of the materials at different time points to form a multidimensional data set including time, temperature, contamination degree and performance indicators; The principal component analysis method was used to reduce the dimensionality of the multidimensional data set, and support vector regression was used to establish the relationship model between temperature, contamination degree and material performance degradation rate; Combining the initial performance distribution of the material and the environmental prediction data, a Monte Carlo simulation method is used to generate a performance degradation path, set a performance degradation path threshold, and determine whether the performance degradation path reaches the performance degradation path threshold; If the performance degradation path reaches the performance degradation path threshold, the material is judged to have reached its service life, and the performance degradation probability distribution function of the material under the multi-stress coupling is obtained by statistically analyzing the performance degradation path, and a performance degradation model of the insulating material under the multi-stress coupling effect of comprehensive temperature and contamination is established; If the performance degradation path does not reach the performance degradation path threshold, it is determined that the material has not reached its service life, and the process of generating the performance degradation path using the Monte Carlo simulation method is repeated until the performance degradation path reaches the performance degradation path threshold.

5. The intelligent line selection optimization method based on the medium voltage distribution network according to claim 4 is characterized in that: In step S3, generating a performance degradation path using a Monte Carlo simulation method includes the following steps: The number of simulations is set to A, and the temperature and contamination degree at each time point are sampled according to normal distribution. The sampling results are input into the relationship model to obtain the degradation rate, and the performance degradation value is accumulated and calculated.

6. The intelligent line selection optimization method based on the medium voltage distribution network according to claim 1 is characterized in that: The step S4 specifically includes: Based on the time-series prediction data of temperature and contamination level and the insulation material performance degradation model, the core evaluation algorithm of the intelligent line selection system is constructed. The mapping relationship between ambient temperature and contamination level and line insulation level and reliability is obtained through training using machine learning methods. Different combinations of temperature and contamination level are obtained, and preliminary line candidate solutions are generated.

7. The intelligent line selection optimization method based on the medium voltage distribution network according to claim 6 is characterized in that: The step S4 further comprises: By acquiring pollution combination data from different regions, simulating line insulation data under different pollution scenarios at different temperatures according to the pollution sampling time interval and pollution sampling method, the samples of the preliminary line candidate solutions are enriched.

8. The intelligent line selection optimization method based on the medium voltage distribution network according to claim 1 is characterized in that: The step S5 specifically includes: Acquire long-term temperature prediction data and pollution degree prediction data of the area corresponding to the preliminary line candidate solution, and generate an environmental condition time series data set according to the long-term temperature prediction data and the pollution degree prediction data; The Monte Carlo method is used to sample the environmental condition time series data set, and the insulator flashover voltage is calculated by the insulator flashover theory. The insulator flashover voltage is calculated according to the following formula: U 50% = K*(L+I)*σ*(A+B*d), Among them, U 50% is the 50% flashover voltage; K is the shape factor; L is the creepage distance; l is the overhang distance; σ is the conductivity of the pollution layer; A and B are constants; d is the thickness of the pollution layer; By counting the flashover times in the Monte Carlo simulation results, the flashover probability of the candidate solutions is calculated. Constructing a fault tree, wherein the fault tree includes defining top events, intermediate events and basic events, and obtaining a line failure rate through minimum cut set analysis and calculation; Constructing a comprehensive reliability evaluation index system, wherein the comprehensive reliability evaluation index system includes insulator flashover probability and line failure rate; The weight calculation method based on historical data is used to determine the weight of each evaluation index, and the comprehensive reliability score of the candidate scheme under the influence of different temperatures and pollution levels is calculated; Calculate the cumulative risk index of the candidate solutions and obtain the comprehensive reliability evaluation results.

9. The intelligent line selection optimization method based on the medium voltage distribution network according to claim 1 is characterized in that: The step S6 specifically includes: According to the comprehensive reliability evaluation result, a multi-objective optimization model is adopted, and the construction of the multi-objective optimization model includes the following steps: Obtaining a line reliability index, a construction cost index, and an operation and maintenance convenience index, wherein the line reliability index is defined as the complement of the annual average failure rate, the construction cost index is defined as the total investment cost, and the operation and maintenance convenience index is defined as the reciprocal of the average arrival time of maintenance personnel; Taking the line reliability index, the construction cost index and the operation and maintenance convenience index as optimization targets, and establishing the multi-objective optimization model; Combined with the environmental topographic constraints, the Pareto optimal route solution set that meets multiple objective weight combinations is solved.

10. The intelligent line selection optimization method based on the medium voltage distribution network according to claim 1 is characterized in that: The step S7 specifically includes: Performing adaptability analysis on the schemes in the Pareto optimal line scheme set, wherein the adaptability analysis includes changing temperature and pollution conditions, and simulating the insulation performance of the line under complex environments of high temperature and heavy pollution, low temperature and light pollution, and seasonal pollution changes; The response surface method is used to construct a sensitivity model. The relative sensitivity coefficients of temperature and pollution degree are calculated through the sensitivity model, and the sensitivity of the scheme to environmental adaptability is evaluated. The scheme with the lowest sensitivity is selected as the optimal intelligent line selection optimization scheme for medium-voltage distribution network.

Images