Variable grid group state monitoring method, system, equipment and medium

Abstract

The invention discloses a variable grid group state monitoring method, system and device and a medium. The method comprises the steps that multi-dimensional climate data are collected, and joint feature representation is generated through association fusion; obtaining the influence degree of each climatic factor parameter and the operation parameter of the power distribution network through multi-dimensional stratified sampling; constructing a meteorological power grid coupling index matrix, and obtaining node vulnerability; constructing a variable network group vulnerability thermodynamic diagram; generating a variable grid group load curve by comparing the correction deviation; and generating a variable grid group state monitoring report. According to the method, multi-dimensional climate data is associated and fused to generate joint feature representation; obtaining the influence degree of each climatic factor parameter change on the operation parameters of the power distribution network through multi-dimensional stratified sampling, and constructing a meteorological power grid coupling index matrix to obtain node vulnerability; finally, a variable grid group state monitoring report is generated, the problems of fixed grid deviation and centralized processing lag are solved, the monitoring accuracy and timeliness are improved, and scientific decision support is provided for power grid operation and maintenance.

Description

Technical Field

[0001] This invention relates to the field of power system monitoring technology, and in particular to a method, system, device and medium for monitoring the status of a variable grid group. Background Technology

[0002] As power systems evolve towards intelligence and distributed computing, the distribution network, as a crucial link in power transmission and user-side connections, is significantly affected by factors such as climate and load fluctuations. Currently, the field of distribution network condition monitoring generally adopts a "grid-based management + cloud-edge collaboration" architecture. This involves deploying sensing nodes at the edge to collect multi-dimensional climate data, such as temperature, humidity, and wind speed, along with grid operating parameters, such as voltage, power, and equipment status. Data fusion, feature analysis, and load forecasting are then performed on a centralized cloud platform. Simultaneously, fixed grid partitioning is used to assess distribution network vulnerability, providing fundamental support for grid dispatching, fault early warning, and safe operation. The application of intelligent technologies in this field has become an important means of ensuring the reliable operation of the distribution network.

[0003] However, existing methods for monitoring the status of distribution networks still have significant limitations: On the one hand, fixed grid division models are difficult to adapt to dynamic environmental changes, such as the drift of distribution network topology caused by extreme weather and seasonal load adjustments. The fixed grid boundaries cause discrepancies between the network status representation and the actual operating conditions, making it impossible to accurately reflect the real-time vulnerability differences of power grids in different regions. On the other hand, centralized data fusion and computing models are highly dependent on cloud computing power. Data interaction between edge nodes and the cloud has a time delay, and the efficiency of multi-node collaborative processing is low, making it difficult to meet the need for real-time correction of load forecast deviations. This results in delayed early warning responses, affecting the timeliness and accuracy of distribution network operation adjustments. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a variable grid group status monitoring method, system, device, and medium to address the limitations of existing distribution network status monitoring methods: First, fixed grid division is difficult to adapt to dynamic changes and is prone to deviations in grid group status representation due to topological correlation drift, making it impossible to accurately reflect the real-time vulnerability of the regional power grid; Second, centralized data processing relies on cloud computing power, resulting in time delays in interaction, low collaborative efficiency, and difficulty in real-time correction of load forecast deviations, leading to delayed early warnings.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for monitoring the state of a variable grid group, comprising: Collect multidimensional climate data, extract power grid environmental features based on the multidimensional climate data, and use a first algorithm to correlate and fuse the power grid environmental features to generate a joint feature representation. Based on the joint feature characterization, the mapping relationship between each climate factor parameter and the distribution network operation parameter is obtained through a multi-dimensional stratified sampling method. Based on the mapping relationship, the degree of influence of the changes in each climate factor parameter on the distribution network operation parameter is analyzed. Calculate the weights corresponding to each climate factor parameter based on the degree of influence; construct a meteorological power grid coupling index matrix by combining the weights, degree of influence, and power grid operation parameters; The identification results are obtained based on the meteorological power grid coupling index matrix. The vulnerability of each node in the distribution network is obtained based on the identification results. A variable network group vulnerability heat map is constructed based on the vulnerability of each node. The variable grid group load is predicted by the variable grid group vulnerability heat map, and the variable grid group load curve is generated by comparing and correcting the deviation between the variable grid group load and the distribution network operating parameters. The variable grid group status is identified by the load curve of the variable grid group, the variable grid group status is integrated, and a variable grid group status monitoring report is generated.

[0007] As a preferred embodiment of the variable grid group state monitoring method described in this invention, the step of generating joint feature representation includes: The collected multidimensional climate data is subjected to time-uniform calibration, interference signal filtering, and data scale standardization to generate standardized climate data. The standardized climate data are subjected to a feature extraction algorithm to extract key features and generate a basic environmental feature set; redundant features in the basic environmental feature set are removed by a feature filtering algorithm to obtain a refined environmental feature set. The refined environment feature set is encrypted and fused using a first algorithm to generate encrypted joint features; the encrypted joint features are then verified using a data verification mechanism to generate verification joint features; and the verification joint features are then decrypted and restored to generate a joint feature representation.

[0008] The beneficial effects of this preferred technical solution are as follows: it improves the quality of climate data through standardized processing, simplifies effective information through feature extraction and screening, and achieves encrypted fusion and verification by combining the first algorithm, which not only ensures the integrity and correlation of the data, but also enhances the security of features, provides a reliable foundation for subsequent impact degree analysis and coupling matrix construction, and improves monitoring accuracy.

[0009] As a preferred embodiment of the variable grid group state monitoring method of the present invention, the step of analyzing the impact of changes in various climate factor parameters on the operating parameters of the distribution network based on the mapping relationship includes: Based on the range of each climate factor parameter and the type of distribution network operation parameter in the joint feature representation, the sampling parameter space is obtained through the sampling dimension and boundary operation of the multi-dimensional stratified sampling method; the sampling parameter space is divided into multi-dimensional intervals to generate a stratified sampling framework. The sample positions of the stratified sampling framework are independently and randomly ordered to generate an initial sample matrix; the initial sample matrix is ​​then filtered according to the electrical operation safety threshold of the distribution network to obtain a valid sample set. Parallel simulations are performed on the parameters of each climate factor in the effective sample set to generate an operational parameter response dataset. A rate of change analysis is then performed on the operational parameter response dataset to obtain the degree of influence of changes in each climate factor parameter on the operational parameters of the distribution network.

[0010] The beneficial effects of this preferred technical solution are as follows: It constructs a parameter space and sampling framework through multi-dimensional stratified sampling, ensures the validity of the initial sample matrix by combining it with safety threshold screening, and then analyzes the impact degree through parallel simulation and rate of change. This improves both the comprehensiveness and reliability of the initial sample matrix and accelerates data processing efficiency, providing a basis for subsequent calculation of climate factor weights and construction of meteorological-power grid coupling index matrices, thus ensuring the scientific nature of the monitoring process.

[0011] As a preferred embodiment of the variable grid group state monitoring method described in this invention, the step of constructing the meteorological power grid coupling index matrix is ​​as follows: Based on the degree of impact, the grey relational analysis method is used to analyze the correlation between climate factors and power grid parameters, and a meteorological power grid evaluation hierarchy framework is constructed by combining distribution network operation parameters and equipment status data. The importance of the meteorological power grid evaluation hierarchy framework is compared pairwise to generate an initial judgment matrix; the relative weights in the initial judgment matrix are then corrected to generate an optimized judgment matrix. A weight set is generated by weighting the optimized judgment matrix; an index fusion algorithm is used to correlate the weight set with the power distribution network operating parameters in multiple dimensions to construct a meteorological power grid coupled index matrix.

[0012] The beneficial effects of this preferred technical solution are as follows: by analyzing the relationship between climate factors and power grid parameters through grey relational analysis, and constructing a meteorological and power grid coupled index matrix through optimized judgment matrix weighting and combined index fusion algorithm, the weight of climate factors is determined, and multidimensional correlation between meteorological and power grid data is realized, thereby improving the timeliness and correlation of the matrix and providing a reliable analytical basis for subsequent state identification.

[0013] As a preferred embodiment of the variable grid group state monitoring method described in this invention, the step of constructing a variable grid group vulnerability heatmap includes: Based on the weight distribution in the meteorological power grid coupling index matrix and the distribution network operating parameters, identify the set of nodes with weak voltage stability. By combining the line transmission power limit with the set of weak voltage stability nodes, a dual-threshold collaborative identification method is used to identify the set of branches with abnormal load balance. Information is aggregated from the set of nodes with weak voltage stability and the set of branches with abnormal load balance to obtain node spatial dependency feature vectors; matrix transformation is performed on the node spatial dependency feature vectors to obtain the node vulnerability score matrix corresponding to the vulnerability of each node. The node vulnerability score matrix is ​​mapped to a color 3D space to construct a vulnerability heatmap for variable network clusters.

[0014] The beneficial effects of this preferred technical solution are as follows: It identifies voltage and load anomaly nodes through a meteorological power grid coupling index matrix, obtains node vulnerability through information aggregation and matrix transformation, and then maps it to a color space to generate a heat map. This not only assesses multi-dimensional vulnerability but also visualizes the risk distribution of the power grid cluster, providing spatial characteristic basis for subsequent load forecasting and status identification, thus improving the intuitiveness and accuracy of monitoring.

[0015] As a preferred embodiment of the variable grid group state monitoring method of the present invention, the step of generating the variable grid group load curve includes: Based on the three-dimensional spatial distribution of colors in the variable network vulnerability heatmap, a node risk color value matrix is ​​obtained using the risk analysis method; highly vulnerable areas in the node risk color value matrix are matched to the corresponding load fluctuation sensitivity coefficients to generate a node load sensitivity vector. The node load sensitivity vector is propagated using the topological neighborhood association propagation method to generate a network-level load fluctuation feature tensor; the network-level load fluctuation feature tensor is then extrapolated using a trend prediction algorithm to output the initial load prediction value of the variable network group. The initial load forecast value and the distribution network operation parameters are corrected by using a constraint correction method to generate network group load time series data; the peak-valley fluctuation optimization is performed on the network group load time series data by using a curve optimization algorithm to generate a variable grid group load curve.

[0016] The beneficial effects of this preferred technical solution are as follows: Load sensitivity coefficients are obtained using risk analysis based on the vulnerability heatmap of the variable network group. Initial load forecast values ​​are generated through topological neighborhood association propagation and trend prediction. Then, load curves for the variable network group are generated through deviation correction and curve optimization. This improves prediction accuracy, reduces fluctuation interference, provides high-quality data support for subsequent network group status identification, and ensures the reliability of monitoring results.

[0017] As a preferred embodiment of the variable grid group state monitoring method of the present invention, the step of generating a variable grid group state monitoring report is as follows: Spatiotemporal features are extracted from the variable grid group load curve, and the spatiotemporal features are fused using a weighted correlation method to generate an enhanced load feature matrix; The enhanced load feature matrix is ​​analyzed using a spatiotemporal clustering algorithm to capture the spatial correlation and temporal continuity of the variable network state and generate a spatiotemporal dependent weight distribution. The spatiotemporal dependent weight distribution is then evaluated for clustering quality using a dual-index evaluation method to remove noisy clusters and obtain effective clusters. The effective clusters are mapped to typical grid group state patterns; the statistical characteristics of the effective clusters are fused with real-time power grid alarm information to generate multi-dimensional state indicators; and a variable grid group state monitoring report is output based on the multi-dimensional state indicators.

[0018] The beneficial effects of this preferred technical solution are as follows: it extracts spatiotemporal features from the load curve of the variable grid group, obtains effective clusters through cluster analysis and quality assessment, maps the state pattern and integrates alarm information to generate a variable grid group state monitoring report, making the variable grid group state monitoring report both scientific and practical, and providing a clear decision-making basis for power grid dispatching and operation and maintenance.

[0019] In a second aspect, the present invention provides a variable grid group state monitoring system, comprising: The data processing module is used to collect multidimensional climate data, filter and standardize the multidimensional climate data, and output standardized climate data. The feature fusion module is used to extract distribution network environmental features based on standardized climate data, and to perform correlation and fusion of distribution network environmental features using the first algorithm to generate and output joint feature representations. The impact degree analysis module is used to obtain the mapping relationship between each climate factor parameter and the distribution network operation parameter based on the joint feature characterization and the multi-dimensional stratified sampling method, and to obtain the impact degree of each climate factor parameter change on the distribution network operation parameter based on the mapping relationship. The matrix construction module is used to calculate the weights of each climate factor according to its degree of influence, and to construct and output a meteorological power grid coupling index matrix by combining power distribution network operating parameters. The heat map construction module is used to obtain the identification results based on the meteorological power grid coupling index matrix, obtain the vulnerability of each node in the distribution network based on the identification results, and construct a variable network group vulnerability heat map. The load curve generation module is used to predict the load of the variable network group based on the vulnerability heat map of the variable network group, compare and correct the deviation between the predicted variable network group load and the distribution network operation parameters, and generate the load curve of the variable network group. The report generation module is used to identify the status of variable grid groups based on the variable grid group load curve, integrate the variable grid group status, and generate and output a variable grid group status monitoring report.

[0020] Thirdly, the present invention provides an electronic device, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions, which, when executed by the processor, implement the steps of the variable grid group state monitoring method.

[0021] Fourthly, the present invention provides a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the variable grid group state monitoring method.

[0022] Compared with existing technologies, the beneficial effects of this invention are as follows: First, multidimensional climate data is standardized and fused to generate joint feature representations; then, the impact of changes in various climate factor parameters on distribution network operation parameters is obtained through multidimensional hierarchical sampling, and a meteorological-power grid coupling index matrix is ​​constructed based on the impact degree to determine node vulnerability; finally, a variable grid group status monitoring report is generated through load forecasting correction and spatiotemporal clustering analysis. This solves the problems of fixed grid bias and centralized processing lag, while improving monitoring accuracy and timeliness, providing scientific decision support for power grid operation and maintenance. Attached Figure Description

[0023] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a schematic diagram of the overall process of a variable grid group state monitoring method according to an embodiment of the present invention. Detailed Implementation

[0025] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0026] Example 1, referring to Figure 1 As an embodiment of the present invention, a method for monitoring the state of a variable grid group is provided, comprising the following steps S100~S600: S100. Collect multi-dimensional climate data, extract distribution network environmental features based on the multi-dimensional climate data, and use the first algorithm to correlate and fuse the distribution network environmental features to generate a joint feature representation.

[0027] S200. Based on the joint feature characterization, the mapping relationship between each climate factor parameter and the distribution network operation parameter is obtained through a multi-dimensional stratified sampling method. Based on the mapping relationship, the degree of influence of the changes in each climate factor parameter on the distribution network operation parameter is analyzed.

[0028] S300. Calculate the weights corresponding to each climate factor parameter based on the degree of influence; construct a meteorological power grid coupling index matrix by combining the weights, degree of influence, and power grid operation parameters.

[0029] S400. Based on the meteorological power grid coupling index matrix, the identification result is obtained. Based on the identification result, the vulnerability of each node of the distribution network is obtained. Based on the vulnerability of each node, a variable network group vulnerability heat map is constructed.

[0030] S500. The variable grid group load is predicted by the variable grid group vulnerability heat map. The variable grid group load curve is generated by comparing the deviation between the variable grid group load and the distribution network operation parameters and making corrections.

[0031] S600. Identify the variable grid group status through the variable grid group load curve, integrate the variable grid group status, and generate a variable grid group status monitoring report.

[0032] It should be noted that the operating status of the distribution network is affected by multi-dimensional climate factors and load. Traditional monitoring uses fixed grid division, which is difficult to adapt to topological correlation drift, resulting in deviations in the network status representation. Furthermore, centralized data processing relies on cloud computing power, resulting in high latency in edge-cloud interaction, low efficiency in multi-node collaboration, and inability to correct load forecast deviations in real time, causing delayed early warnings and exacerbating the pressure on power grid operation and maintenance.

[0033] Therefore, to address the issues of poor adaptability of fixed grids and lagging centralized processing, the following steps from S100 to S600 are employed: first, reliable joint features are generated through data standardization and security fusion; then, the degree of impact is determined by multi-dimensional hierarchical sampling, and a meteorological-grid coupling index matrix is ​​constructed in conjunction with distribution network operating parameters to identify node vulnerability; finally, load is predicted and deviations are corrected based on the variable grid group vulnerability heat map, and the variable grid group status monitoring report is identified by combining the variable grid group load curve, providing scientific support for distribution network dynamic monitoring, risk warning, and operation and maintenance scheduling.

[0034] Example 2, refer to Figure 1 As an embodiment of the present invention, a variable grid group state monitoring method is provided based on the above embodiment.

[0035] In this embodiment, the application scenario is an industrial park power distribution network in a temperate continental climate zone, covering 10 power distribution zones, including a mixed power consumption scenario of high-energy-consuming industries, general commercial areas, and employee dormitories. The specific implementation steps for multi-dimensional climate data processing, power distribution network environmental feature extraction, and joint feature representation generation in S100 (A1~A3) are as follows: A1. Perform time-uniform calibration, interference signal filtering, and data scale standardization on the collected multidimensional climate data to generate standardized climate data.

[0036] Specifically, the collected multidimensional climate data includes basic meteorological parameters such as temperature -12℃ to 38℃, wind speed 0 to 17m / s, relative humidity 18% to 95%, and daily precipitation 0 to 42mm. The sampling frequency is 15 minutes / time, and a total of 8,832 multidimensional climate data points were collected over 92 days. Using the park's standard clock as a reference, 8832 multidimensional climate data points were collected. A multi-stage collaborative synchronization method was used to map 127 data points with time misalignments to their corresponding 15-minute time windows. After mapping, a fixed 15-minute time window strategy was employed for alignment verification, generating 8832 synchronized climate data points. Noise was filtered out from the synchronized climate data using a sliding window filtering method. Five extreme outliers were removed using the 3σ criterion, such as a temperature jump of 12℃ or a sudden increase in wind speed to 22m / s. Nineteen missing data points were completed by linear interpolation, such as interpolating the range from -5℃ to -3℃ as -5℃→-4.5℃→-4℃→-3.5℃→-3℃, resulting in 8832 denoised climate data points. The Z-score standardization method was used to scale the denoised climate data. The mean (temperature 15.2℃, wind speed 3.8m / s) and standard deviation (temperature 8.7℃, wind speed 2.1m / s) of the denoised climate data were calculated and transformed. For example: temperature 25℃ → (25-15.2) / 8.7≈1.13; wind speed 1.2m / s → (1.2-3.8) / 2.1≈-1.24. After transformation, the data conforms to a standard normal distribution, and standardized climate data is generated.

[0037] A2. Use a feature extraction algorithm to extract key features from the standardized climate data to generate a basic environmental feature set; use a feature filtering algorithm to remove redundant features from the basic environmental feature set to obtain a refined environmental feature set.

[0038] Specifically, key features include, but are not limited to, temperature and humidity co-feedback features, strong wind interference features, and precipitation impact features. The PCA algorithm is used to extract key features from standardized climate data. The loading coefficients for temperature and humidity co-feedback features are 0.42 for air temperature and 0.58 for humidity; 0.91 for wind speed; and 0.87 for precipitation impact features, generating a basic environmental feature set containing these three key features. Redundancy is filtered through correlation analysis, and correlation coefficients between features are calculated. For example, the correlation coefficient between temperature and humidity co-feedback features and strong wind interference features is 0.12, and the correlation coefficient with precipitation impact features is 0.08, both less than the set threshold of 0.7, indicating no redundant features and thus a refined environmental feature set.

[0039] A3. The refined environment feature set is encrypted and fused using the first algorithm to generate encrypted joint features; the encrypted joint features are verified using a data verification mechanism to generate verification joint features; the verification joint features are decrypted and restored to generate joint feature representations.

[0040] Specifically, the first algorithm is a secure aggregation algorithm; the 10 power distribution zones serve as edge nodes, and a homomorphic encryption algorithm is used to encrypt the refined environmental feature set to generate an encrypted feature matrix; The encrypted feature matrix is ​​fragmented and transmitted to the cloud aggregation node through a sparse k-regular graph topology. The cloud uses a secure multi-party computation protocol to perform weighted fusion of the encrypted data from each node. The weights are allocated according to the load ratio of the partition: 0.4 for high-energy-consuming industrial partitions, 0.3 for commercial partitions, and 0.3 for dormitory partitions, to generate encrypted joint features. A hash algorithm is used to verify the integrity of the encrypted joint feature. The cloud calculates the hash value and compares it with the hash values ​​uploaded by each edge node. Once the consistency is confirmed, a verification joint feature is generated. By using a threshold decryption mechanism, which requires ≥7 edge nodes to provide key shards to verify the joint feature decryption, a three-dimensional joint feature characterization is obtained, including the correlation between temperature and humidity and load, strong wind and line loss coefficient, precipitation and insulation performance index.

[0041] In an optional implementation, step S100 can further adjust the feature selection threshold based on the electricity load type of different power distribution zones in the park. The steps are as follows: In the redundant feature removal stage of step A2, the correlation coefficient threshold can be dynamically adjusted based on the park's electricity load type: for high-energy-consuming industrial zones, the threshold for determining strong correlation features is relaxed from 0.7 to 0.75, allowing more potential related features to be retained; for commercial and dormitory zones, the threshold is tightened to 0.65, strictly removing redundant features. Verification using 30 days of summer data shows that the adjusted feature selection strategy allows for more complete preservation of climate and load correlation features in high-energy-consuming zones, and a more streamlined feature dimension for commercial and dormitory zones. This enables the generated joint feature representation to better adapt to the subsequent analysis needs of different load type zones, improving the method's adaptability to diverse electricity consumption scenarios within the park.

[0042] In another optional implementation, step S100 can be further expanded with sample expansion and cross-validation. The steps are as follows: After generating the joint feature representation in step A3, supplement the sample with three months of climate-grid correlation data from adjacent industrial parks in the same climate zone, including eight distribution zones, primarily focusing on precision manufacturing and light industry, increasing the total sample size from 92 days to 184 days. A 5-fold cross-validation method is used to evaluate the feature extraction effect, whereby the merged samples are randomly divided into five groups: four groups are used to optimize the number of principal components in PCA, and one group is used to verify the feature discrimination and effectiveness. Cross-validation avoids the feature overfitting problem caused by samples from a single industrial park, enabling the joint feature representation to adapt to more diverse industrial distribution network scenarios and further expanding the applicability of the method.

[0043] In this embodiment of the application, step S200, which involves analyzing the impact of changes in various climate factor parameters on the operating parameters of the distribution network based on the mapping relationship, includes steps B1 to B3: B1. Based on the range of each climate factor parameter and the type of distribution network operation parameter in the joint feature representation, the sampling parameter space is obtained through the sampling dimension and boundary operation of the multi-dimensional stratified sampling method; the sampling parameter space is divided into multi-dimensional intervals to generate a stratified sampling framework.

[0044] Specifically, based on the joint feature representation, the actual physical range of climate factor parameters is extracted, such as air temperature -12℃~38℃, wind speed 0~17m / s, relative humidity 18%~95%, and daily precipitation 0~42mm; the types and safe ranges of distribution network operation parameters are determined, such as 10kV node voltage, line active power 0~120%, rated capacity of 500kVA, transformer load rate 0~100%, and outdoor equipment temperature -15℃~85℃; the sampling dimensions are defined as four types of climate factor parameters and four types of distribution network operation parameters, and parameter combinations that exceed the equipment tolerance limits are eliminated, such as equipment temperature >85℃ and voltage <9.5kV, to obtain the compliant sampling parameter space; the space is divided into multi-dimensional intervals according to the principle of uniform coverage, such as air temperature every 5℃, wind speed every 3m / s, and voltage every 0.2kV, finally forming a stratified sampling framework containing 135 stratified units.

[0045] B2. Randomize the sample positions of the stratified sampling framework to generate an initial sample matrix; filter the initial sample matrix according to the electrical operation safety threshold of the distribution network to obtain a valid sample set.

[0046] Specifically, the sample locations within each hierarchical unit are randomly sorted to generate an initial sample matrix containing 2800 parameter combinations. Based on the electrical safety thresholds of the distribution network (voltage deviation ≤ ±5%, transformer load rate ≤ 100%, line power ≤ 120%, and rated capacity), 420 samples exceeding the thresholds, such as combinations with voltage of 9.3kV and load rate of 105%, are filtered out, resulting in a set of 2380 valid samples containing parameter combinations from multiple typical climate scenarios.

[0047] B3. Parallel simulation and deduction are performed on the parameters of each climate factor in the effective sample set to generate an operating parameter response dataset; the rate of change analysis is performed on the operating parameter response dataset to obtain the degree of influence of the changes of each climate factor parameter on the operating parameters of the distribution network.

[0048] Specifically, the distribution network digital simulation method was used to simulate the climate factor parameters in the combined effective samples, such as temperature 35℃, humidity 90%, wind speed 2m / s and temperature -8℃, humidity 30%, wind speed 15m / s, generating a dataset of 2380 operating parameter responses. The rate of change of the operating parameter response dataset was calculated to obtain the degree of influence. For example, when the temperature rises from 25℃ to 35℃, the load rate increases from 60% to 62.8%; when the wind speed increases from 5m / s to 15m / s, the power loss increases from 3% to 7.5%; when the humidity increases from 50% to 80%, the insulation resistance decreases from 1000MΩ to 856MΩ, etc.

[0049] In an optional implementation, step S200 may further include sampling accuracy feedback adjustment. The steps are as follows: retrieve the 92-day standardized data from S100, and calculate the goodness of fit with the effective sample set of B2 using the R² index. If R² < 0.91, then reduce the granularity of the climate parameter intervals, such as reducing the temperature interval from 5℃ to 3℃ and the wind speed interval from 3m / s to 2m / s, and regenerate the hierarchical framework and effective samples. If R² ≥ 0.91, no adjustment is required, ensuring the consistency between the response dataset and the actual operating data, and improving the accuracy of the impact degree analysis.

[0050] In another optional implementation, extreme climate sample enhancement can also be performed in step S200. The steps are as follows: screen extreme climate data of climate factor parameters in S100, such as temperature >35℃ or <-8℃, wind speed >12m / s, and daily precipitation >30mm. Based on the extreme climate data, 25 extreme climate stratification units are added separately in the B1 stratification framework. After simulation and deduction in B3, the quantification error of the impact of extreme climate on the operation parameters of the distribution network is reduced, which is suitable for the monitoring needs under extreme weather.

[0051] In this embodiment of the application, the step of constructing the meteorological power grid coupling index matrix in step S300 includes C1~C3: C1. Based on the degree of influence, the grey relational analysis method is used to analyze the correlation between climate factors and power grid parameters, and a meteorological power grid evaluation hierarchy framework is constructed by combining the distribution network operation parameters and equipment status data.

[0052] Specifically, based on climatic factor parameters and distribution network operation parameters, the distribution network operation parameter sequence is used as a reference sequence, and the climatic factor sequence is used as a comparison sequence. The reference and comparison sequences are averaged and standardized to eliminate dimensional differences. The correlation coefficients between each climatic factor and the distribution network operation parameters are calculated, and the arithmetic mean of the correlation coefficient sequence is obtained to generate a grey relational degree. For example, temperature and transformer load rate are 0.82, wind speed and line active power are 0.78, relative humidity and outdoor equipment temperature are 0.65, and daily precipitation and 10kV node voltage are 0.52. Based on the descending order of correlation degree, temperature > wind speed > relative humidity > daily precipitation, the dominant climatic factors are identified as temperature and wind speed. Combining the distribution network operation parameters, a hierarchical framework for meteorological power grid evaluation is constructed: the target layer is the evaluation of the meteorological power grid coupling state; the criteria layer includes voltage stability, power transmission efficiency, equipment load safety, and equipment environmental tolerance; and the indicator layer represents the correlation relationships between various climatic power grid parameters.

[0053] C2. Perform pairwise importance comparisons on the meteorological power grid evaluation hierarchy framework to generate an initial judgment matrix; then correct the relative weight assignments in the initial judgment matrix to generate an optimized judgment matrix.

[0054] Specifically, the AHP-entropy weight combination method is used to perform pairwise importance comparisons on the elements of the criterion layer. The relative importance is quantified according to the 1-9 scale method. For example, equipment load safety is slightly more important than voltage stability, so it is assigned a value of 3; power transmission efficiency is significantly more important than equipment environmental tolerance, so it is assigned a value of 5, etc. All comparison results are filled into the corresponding positions in the matrix to form the initial judgment matrix. The initial judgment matrix is ​​then corrected using the analytic hierarchy process. The maximum eigenvalue of the initial judgment matrix is ​​calculated to be 4.12, and the consistency ratio CR≈0.044<0.1, thus obtaining the optimized judgment matrix.

[0055] C3. A weight set is generated by weighting the optimized judgment matrix; an index fusion algorithm is used to correlate the weight set with the power distribution network operating parameters in multiple dimensions to construct a meteorological power grid coupled index matrix.

[0056] Specifically, the sum-product method is used to solve the optimization judgment matrix, and the weight vector corresponding to the element weights of the criterion layer is obtained as [0.22, 0.15, 0.52, 0.11]. Adjustment coefficients are calculated based on the distribution network operation parameters such as the average voltage deviation rate of 1.2% and the average line power loss rate of 4.5%, and the weight vector is corrected to generate a weight set. An index fusion algorithm is used to fuse the weight set with power grid equipment data. After fusion, the coupling relationship between meteorological and power grid parameters is established through linear regression analysis, and a meteorological and power grid coupling index matrix is ​​generated.

[0057] In an optional implementation, step S300 can also incorporate an expert review mechanism. The steps are as follows: invite 5 experts in power distribution network operation and meteorology to independently review the scale value of the initial judgment matrix in C2. Based on the experts' opinions, such as if the experts believe that the importance of equipment environmental tolerance should be equivalent to the power transmission efficiency, adjust the original scale value 2 to 3, correct the initial judgment matrix, and then perform weight calculation and consistency verification using the analytic hierarchy process to make the generated weight set more consistent with actual engineering experience and improve the practicality of the meteorological power grid coupling index matrix.

[0058] In another optional implementation, step S300 can further enhance the application of historical fault data. The steps are as follows: supplement the historical fault data of the park's power distribution network in the past 3 years, and record a total of 87 faults caused by climate factors, including 42 caused by high temperature, 28 caused by strong wind, 12 caused by high humidity, and 5 caused by precipitation. When calculating the grey relational degree in C1, the influence weight of the fault data is increased, making the fault risk correlation of the constructed meteorological power grid evaluation hierarchy framework more obvious, and the generated meteorological power grid coupling index matrix more accurate.

[0059] In this embodiment of the application, the step of constructing the variable network group vulnerability heatmap in step S400 includes D1~D4: D1. Identify the set of nodes with weak voltage stability based on the weight distribution in the meteorological power grid coupling index matrix and the distribution network operating parameters.

[0060] Specifically, based on the meteorological power grid coupling index matrix, a voltage stability weight of 0.22 was extracted, along with the coupling relationship between temperature, wind speed, and 10kV node voltage. Combined with the 10kV node voltage data from the distribution network operation, specifically the average voltage of nodes in the high-energy-consuming industrial zone (9.85kV, deviation rate 1.5%), the average voltage of nodes in the commercial zone (10.1kV, deviation rate 1.0%), and the average voltage of nodes in the dormitory zone (10.05kV, deviation rate 0.5%) over 92 days, a weak voltage stability threshold was set: node voltage deviation rate > 1.2% and voltage value < 9.9kV or > 10.1kV. After screening, three 10kV nodes in the high-energy-consuming industrial zone, numbered N1, N2, and N3, met the threshold requirements, constituting a set of nodes with weak voltage stability.

[0061] D2. Combining the line transmission power limit with the set of weak voltage stability nodes, a dual-threshold collaborative identification method is used to identify the set of branches with abnormal load balance.

[0062] Specifically, the power transmission limit of the distribution network line is the rated capacity, which is 500kVA; the dual threshold is set as follows: branch load rate > 85% and load rate fluctuation in adjacent time periods > 15%; for nodes N1, N2, and N3 in the set of nodes with weak voltage stability, the power data of their associated branches, a total of 8 branches, numbered L1-L8, are retrieved. Among them, L1 has a load rate of 92% and rises from 78% to 92% within 1 hour, L3 has a load rate of 88% and rises from 72% to 88% within 30 minutes, and L5 has a load rate of 90% and rises from 75% to 90% within 45 minutes; after dual threshold collaborative identification, L1, L3, and L5 meet the abnormal judgment and constitute a set of branches with abnormal load balance.

[0063] D3. Aggregate information from the set of nodes with weak voltage stability and the set of branches with abnormal load balance to obtain node spatial dependency feature vectors; perform matrix transformation on the node spatial dependency feature vectors to obtain the node vulnerability score matrix corresponding to the vulnerability of each node.

[0064] Specifically, taking the abnormal branches L1, L3, and L5 as the objects, the neighborhood range is determined according to the inter-branch impedance distance: L1-L3 = 0.8Ω, L1-L5 = 1.2Ω, and L3-L5 = 0.9Ω. The density values ​​are calculated as 2.5 for L1 and L3, and 2.22 for L5, generating an initial topology map. Based on the initial topology map, combined with the voltage deviation rates of N1 (1.5%), N2 (1.3%), and N3 (1.2%), and the neighboring branch data (L1 load rate 92%), the initial topology map is used to determine the neighborhood range range. With L3 load at 88% and L5 load at 90%, spatial dependency feature vectors [0.82, 0.71, 0.65] are generated through dynamic graph convolution. Fluctuation correction values ​​(N1 + 0.03, N2 + 0.01, N3 + 0.03) are processed and calculated using a 30-minute sliding window, and then summed to obtain [0.85, 0.72, 0.68]. Weighted by 0.5 for both voltage and load channels, and normalized to the 0.6-0.9 range, a node vulnerability score matrix is ​​generated. D4. Map the node vulnerability score matrix to a color 3D space to construct a variable network group vulnerability heatmap.

[0065] Specifically, a three-color gradient mapping score is used; N1, N2, and N3 are marked in the distribution network topology map, and the load rate and fluctuation information of the associated abnormal branches L1, L3, and L5 are marked to generate a variable network group vulnerability heat map, which shows the voltage and load-related vulnerable areas of the variable network group in the high-energy-consuming industrial zone.

[0066] In an optional implementation, step S400 may also introduce an update mechanism, which involves: collecting distribution network operating parameters every 15 minutes, repeating steps D1-D3 to update the node vulnerability score matrix, and then refreshing the variable network group vulnerability heat map, so that the display of vulnerable areas is synchronized with the power grid operating status and the monitoring timeliness is improved.

[0067] In another optional implementation, step S400 can also include historical vulnerability comparison, which involves retrieving historical data of the node vulnerability score matrix for the past 3 months, totaling approximately 1440 sets, comparing them with the current matrix over time, and displaying the vulnerability change trend by overlaying a variable network cluster vulnerability heatmap. For example, the score of node N1 has increased from 0.75 to 0.84 in the past week, indicating an increase in vulnerability, which helps maintenance personnel predict potential risks.

[0068] In this embodiment of the application, the step of generating the variable grid group load curve in step S500 includes E1~E3: E1. Based on the three-dimensional spatial distribution of colors in the variable network vulnerability heatmap, a node risk color value matrix is ​​obtained using the risk analysis method; the highly vulnerable areas in the node risk color value matrix are matched to the corresponding load fluctuation sensitivity coefficients to generate a node load sensitivity vector.

[0069] Specifically, based on the HSV color space of the variable network vulnerability heatmap, the vulnerability level of nodes is analyzed. Using the HSV risk analysis method, and based on the hue-vulnerability mapping rule, 0°~60° is the high vulnerability range, and a node risk color value matrix is ​​generated. N1, N2, and N3, whose phase values ​​are in the range of 0° to 60°, are selected as highly vulnerable nodes. N1 and N2 correspond to high-energy-consuming industrial loads with a basic sensitivity coefficient of 0.8. N3 corresponds to commercial loads with a basic sensitivity coefficient of 0.6. The basic sensitivity coefficients are weighted and adjusted, such as N1: 0.8×0.85=0.68, N2: 0.8×0.72=0.576, and N3: 0.6×0.68=0.408, generating a node load sensitivity vector [0.68, 0.576, 0.408].

[0070] E2. The node load sensitivity vector is propagated using the topological neighborhood association propagation method to generate a network-level load fluctuation feature tensor; the network-level load fluctuation feature tensor is extrapolated using a trend prediction algorithm to output the initial load prediction value of the variable network group.

[0071] Specifically, using the graph convolution diffusion method, an adjacency matrix is ​​defined, such as adjacency weights of 0.9, 0.8, and 0.7 for N1 and L1, N2 and L3, and N3 and L5, respectively. The node load sensitivity vector is used as the initial node feature matrix, and the network-level load fluctuation feature tensor is generated by propagating along the power grid topology. The trend prediction method is used to perform trend fitting calculation on the time dimension of the network-level load fluctuation feature tensor. Combined with 92 days of distribution network operation parameters, the average load of the high-energy-consuming industrial zone is 420 kVA and that of the commercial zone is 280 kVA, and the initial load prediction value is generated. That is, the load of the high-energy-consuming industrial zone is 450-500 kVA from 8:00 to 18:00 the next day, and that of the commercial zone is 300-350 kVA.

[0072] E3. The initial load forecast value and the distribution network operation parameters are corrected by using the constraint correction method to generate network group load time series data; the peak-valley fluctuation optimization is performed on the network group load time series data by using the curve optimization algorithm to generate a variable grid group load curve.

[0073] Specifically, distribution network operating parameters are collected synchronously. The actual load of the high-energy-consuming industrial zone is 430kVA, and that of the commercial zone is 290kVA. The absolute deviation vector between the initial predicted value and the actual value is calculated [20,10,10]. A physical constraint hybrid correction method is adopted, based on the formula... ,in For actual load value, To predict load values, For a rated capacity of 500kVA, Taking 1.5, the initial grid group load time series data is generated after correcting the deviation vector; using the curve smoothing algorithm, the initial grid group load time series data is optimized through multiple rounds of iterative smoothing and peak-valley constraints to obtain the grid group load time series data, namely, the high-energy-consuming industrial zone fluctuates at 430-490kVA and the commercial zone fluctuates at 290-340kVA; the grid group load time series data is transformed into a variable grid group load curve.

[0074] In an optional implementation, step S500 can also dynamically adjust the sensitivity coefficient according to the load type of different power distribution zones in the park. The steps are as follows: In the load fluctuation sensitivity coefficient matching stage of step E1, for high-energy-consuming industrial zones, if real-time climate factors, such as temperature > 30℃, trigger a surge in production load, the basic sensitivity coefficient is increased from 0.8 to 0.9; for commercial zones, if it is during non-business hours, the basic sensitivity coefficient is reduced from 0.6 to 0.5, so that the generated node load sensitivity vector is more in line with the actual load dynamics and the accuracy of the load curve is improved.

[0075] In another optional implementation, step S500 can also introduce multi-source data fusion verification. The steps are as follows: before the deviation correction in step E3, the power distribution network operation parameters of adjacent industrial parks, namely precision manufacturing parks in the same climate zone and containing 8 power distribution zones, are fused as auxiliary verification sources for 3 months. The cosine similarity algorithm is used to match the similarity with the local park. If the similarity is ≥ threshold 0.8, the initial load forecast value is corrected by combining the auxiliary data to enhance the robustness of the forecast result and make the generated variable grid group load curve more suitable for the overall load characteristics of the region.

[0076] In this embodiment of the application, the step of generating a variable grid group state monitoring report in step S600 includes F1 to F3: F1. Extract the spatiotemporal dimension features from the variable grid group load curve, and fuse the spatiotemporal dimension features using a weighted correlation method to generate an enhanced load feature matrix.

[0077] Specifically, based on the variable grid group load curve, time-dimensional features are extracted, such as the fluctuation sequence of the load of high-energy-consuming industrial zone (430-490kVA) and commercial zone (290-340kVA) from 8:00 to 18:00 daily, and spatial-dimensional features, such as the load distribution ratios of high-energy-consuming industrial, commercial, and dormitory zones (40%, 30%, and 30%, respectively). Using a weighted correlation method, the time-dimensional features reflecting the dynamic changes of load over time are assigned a weight of 0.6, and the spatial-dimensional features reflecting the spatial distribution of load in different zones are assigned a weight of 0.4. Feature fusion is then performed to generate an enhanced load feature matrix.

[0078] F2. The enhanced load feature matrix is ​​analyzed using a spatiotemporal clustering algorithm to capture the spatial correlation and temporal continuity of the variable network state and generate a spatiotemporal dependent weight distribution. The spatiotemporal dependent weight distribution is then evaluated for clustering quality using a dual-index evaluation method to remove noisy clusters and obtain effective clusters.

[0079] Specifically, a spatiotemporal clustering algorithm is used to process the features of each time slice in the enhanced load feature matrix, extracting low-dimensional spatial feature vectors. For example, the load feature vector [0.8, 0.7, 0.6] of the high-energy-consuming industrial zone represents the spatial state of the variable network group. An adaptive propagation clustering algorithm is used to determine the number of clusters as 3. The similarity matrix between samples is iteratively updated to generate cluster centers. Nodes represent different spatial state categories, and edge weights are calculated using a deep spatiotemporal clustering algorithm to generate spatiotemporal dependent weights. The node self-loop weights are set as follows: cluster 1 is 0.8, cluster 2 is 0.7, and cluster 3 is 0.6; edge weights are set as follows: cluster 1-cluster 2 is 0.5, and cluster 2-cluster 3 is 0.4. The silhouette coefficient of each cluster is calculated: cluster 1 is 0.82, cluster 2 is 0.78, and cluster 3 is 0.75. The Davidson-Bolding index is set as follows: cluster 1 is 250, cluster 2 is 230, and cluster 3 is 210. The top 3 clusters with the joint evaluation value of the two indices are set as effective clusters. Meaningless noise clusters are removed, and 3 effective clusters are obtained.

[0080] F3. Map the effective clusters to typical grid group state patterns; integrate the statistical characteristics of the effective clusters with real-time power grid alarm information to generate multi-dimensional state indicators; output a variable grid group state monitoring report based on the multi-dimensional state indicators.

[0081] Specifically, based on the statistical characteristics of effective clusters, such as Cluster 1: high-energy-consuming industrial zone with an average load of 460kVA, variance of 50kVA², and peak-to-valley difference of 80kVA; Cluster 2: commercial zone with an average load of 310kVA, variance of 30kVA², and peak-to-valley difference of 50kVA; and Cluster 3: dormitory zone with an average load of 180kVA, variance of 20kVA², and peak-to-valley difference of 30kVA, typical grid group state patterns are defined: Cluster 1 corresponds to the high-load, high-stress state of industry; Cluster 2 corresponds to the stable, low-load state of commercial sector; and Cluster 3 corresponds to the low-load, relaxed state of dormitory sector. A state label is assigned to each effective cluster and associated with real-time grid alarms. Information such as two overvoltage alarms and one overcurrent alarm in cluster 1; no alarms in cluster 2; no alarms in cluster 3; generating alarm event vectors; determining weight coefficients based on alarm event vectors and status labels, i.e., overvoltage alarm weight 0.6, overcurrent alarm weight 0.4; calculating multi-dimensional status indicators, obtaining a comprehensive status indicator of 0.85 for cluster 1, 0.7 for cluster 2, and 0.6 for cluster 3; integrating cluster statistical characteristics and alarm event vectors, calculating load balancing degree using a weighted average method; and outputting a variable network group status monitoring report, such as recommending enhanced line inspection in high-energy-consuming industrial zones under cluster 1 mode.

[0082] In an optional implementation, step S600 may also include a weight adaptive adjustment mechanism, the steps of which are as follows: In the weighted adjacency method of F1, the shape similarity weight is adjusted in real time according to the distribution network operation parameters such as load fluctuation amplitude and alarm frequency. If the load fluctuation amplitude of the high-energy-consuming industrial zone increases by 20% compared with the historical average, the shape similarity weight is increased from 0.6 to 0.7, so that the generated enhanced load feature matrix is ​​more in line with the current power grid operation status and improves the accuracy of subsequent clustering and reporting.

[0083] In another optional implementation, step S600 can also be extended to integrate multi-source alarm information. The steps are as follows: when integrating real-time power grid alarm information in F3, in addition to overvoltage and overcurrent alarms, temperature limit alarms are added, such as outdoor equipment temperature monitoring data and temperature alarms triggered by equipment temperatures exceeding 80°C in high-energy-consuming industrial zones. Additional weighting coefficients are determined based on temperature alarms and status labels, and multi-dimensional status index calculations are incorporated to make the monitoring report more comprehensively reflect the operating status under multiple types of power grid alarms.

[0084] In summary, this invention takes the power distribution network of an industrial park in a temperate continental climate zone as a scenario. It first processes multi-dimensional climate data to generate joint feature representations, then analyzes the impact of climate factors on the operating parameters of the power distribution network, and constructs a meteorological-power grid coupling index matrix. Subsequently, it generates a variable grid group vulnerability heat map, outputs the initial load forecast value of the variable grid group based on the variable grid group vulnerability heat map, generates the variable grid group load curve, and finally outputs a variable grid group status monitoring report. It also provides optional methods such as dynamic adjustment and multi-source verification, providing support for power distribution network status monitoring and operation and maintenance scheduling, and improving the stability and adaptability of power grid operation.

[0085] Example 3 illustrates a schematic scheme for a variable grid group state monitoring method. It should be noted that the technical solution of this variable grid group state monitoring system belongs to the same concept as the technical solution of the variable grid group state monitoring method described above. Details not described in detail in this embodiment can be found in the description of the technical solution of the variable grid group state monitoring method described above.

[0086] This embodiment also provides a variable grid group state monitoring system, including: The data processing module is used to collect multidimensional climate data, filter and standardize the multidimensional climate data, and output standardized climate data. The feature fusion module is used to extract distribution network environmental features based on standardized climate data, and to perform correlation and fusion of distribution network environmental features using the first algorithm to generate and output joint feature representations. The impact degree analysis module is used to obtain the mapping relationship between each climate factor parameter and the distribution network operation parameter based on the joint feature characterization and the multi-dimensional stratified sampling method, and to obtain the impact degree of each climate factor parameter change on the distribution network operation parameter based on the mapping relationship. The matrix construction module is used to calculate the weights of each climate factor according to its degree of influence, and to construct and output a meteorological power grid coupling index matrix by combining power distribution network operating parameters. The heat map construction module is used to obtain the identification results based on the meteorological power grid coupling index matrix, obtain the vulnerability of each node in the distribution network based on the identification results, and construct a variable network group vulnerability heat map. The load curve generation module is used to predict the load of the variable network group based on the vulnerability heat map of the variable network group, compare and correct the deviation between the predicted variable network group load and the distribution network operation parameters, and generate the load curve of the variable network group. The report generation module is used to identify the status of variable grid groups based on the variable grid group load curve, integrate the variable grid group status, and generate and output a variable grid group status monitoring report.

[0087] This embodiment also provides an electronic device suitable for variable grid group state monitoring, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the variable grid group state monitoring method proposed in the above embodiment.

[0088] This embodiment also provides a storage medium on which a computer program is stored. When the program is executed by a processor, it implements the variable grid group state monitoring method proposed in the above embodiments.

[0089] The storage medium proposed in this embodiment and the method for monitoring the state of a variable grid group proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0090] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0091] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for monitoring the state of a variable grid group, characterized in that, include: Collect multidimensional climate data, extract power grid environmental features based on the multidimensional climate data, and use a first algorithm to correlate and fuse the power grid environmental features to generate a joint feature representation. Based on the joint feature characterization, the mapping relationship between each climate factor parameter and the distribution network operation parameter is obtained through a multi-dimensional stratified sampling method. Based on the mapping relationship, the degree of influence of the changes in each climate factor parameter on the distribution network operation parameter is analyzed. Calculate the weights corresponding to each climate factor parameter based on the degree of influence described above; A meteorological power grid coupling index matrix is ​​constructed by combining the aforementioned weights, degree of influence, and power grid operating parameters; The identification results are obtained based on the meteorological power grid coupling index matrix. The vulnerability of each node in the distribution network is obtained based on the identification results. A variable network group vulnerability heat map is constructed based on the vulnerability of each node. The variable grid group load is predicted by the variable grid group vulnerability heat map, and the variable grid group load curve is generated by comparing and correcting the deviation between the variable grid group load and the distribution network operating parameters. The variable grid group status is identified by the load curve of the variable grid group, the variable grid group status is integrated, and a variable grid group status monitoring report is generated.

2. The variable grid group state monitoring method as described in claim 1, characterized in that, The steps for generating joint feature representations are as follows: The collected multidimensional climate data is subjected to time-uniform calibration, interference signal filtering, and data scale standardization to generate standardized climate data. The standardized climate data are subjected to a feature extraction algorithm to extract key features and generate a basic environmental feature set. The redundant features in the basic environmental feature set are removed by a feature filtering algorithm to obtain a refined environmental feature set. The refined environment feature set is encrypted and fused using a first algorithm to generate encrypted joint features; the encrypted joint features are then verified using a data verification mechanism to generate verified joint features. The joint verification features are decrypted and restored to generate a joint feature representation.

3. The variable grid group state monitoring method as described in claim 2, characterized in that, The steps for analyzing the impact of changes in various climate factor parameters on the operating parameters of the power distribution network based on the aforementioned mapping relationship are as follows: Based on the range of each climate factor parameter and the type of distribution network operation parameter in the joint feature representation, the sampling parameter space is obtained through the sampling dimension and boundary operation of the multi-dimensional stratified sampling method; the sampling parameter space is divided into multiple intervals to generate a stratified sampling framework. The sample positions of the stratified sampling framework are independently and randomly ordered to generate an initial sample matrix; the initial sample matrix is ​​then filtered according to the electrical operation safety threshold of the distribution network to obtain a valid sample set. Parallel simulations are performed on the parameters of each climate factor in the effective sample set to generate an operational parameter response dataset. A rate of change analysis is then performed on the operational parameter response dataset to obtain the degree of influence of changes in each climate factor parameter on the operational parameters of the distribution network.

4. The variable grid group state monitoring method as described in claim 3, characterized in that, The steps for constructing the meteorological power grid coupling index matrix are as follows: Based on the degree of impact, the grey relational analysis method is used to analyze the correlation between climate factors and power grid parameters, and a meteorological power grid evaluation hierarchy framework is constructed by combining distribution network operation parameters and equipment status data. A pairwise importance comparison is performed on the meteorological power grid evaluation hierarchy framework to generate an initial judgment matrix; The relative weights in the initial judgment matrix are corrected to generate an optimized judgment matrix; A weight set is generated by weighting the optimized judgment matrix; An index fusion algorithm is used to correlate the weight set with the operating parameters of the distribution network in multiple dimensions, and a meteorological power grid coupled index matrix is ​​constructed.

5. The variable grid group state monitoring method as described in claim 4, characterized in that, The steps to construct a vulnerability heatmap for a variable network group are as follows: Based on the weight distribution in the meteorological power grid coupling index matrix and the distribution network operating parameters, identify the set of nodes with weak voltage stability. By combining the line transmission power limit with the set of weak voltage stability nodes, a dual-threshold collaborative identification method is used to identify the set of branches with abnormal load balance. Information is aggregated from the set of nodes with weak voltage stability and the set of branches with abnormal load balance to obtain the node spatial dependency feature vector; The node spatial dependency feature vector is transformed into a matrix to obtain the node vulnerability score matrix corresponding to the vulnerability of each node. The node vulnerability score matrix is ​​mapped to a color 3D space to construct a vulnerability heatmap for variable network clusters.

6. The variable grid group state monitoring method as described in claim 5, characterized in that, The steps for generating variable grid group load curves are as follows: Based on the three-dimensional spatial distribution of colors in the variable network vulnerability heatmap, the node risk color value matrix is ​​obtained using the risk analysis method. The highly vulnerable regions in the node risk color value matrix are matched to the corresponding load fluctuation sensitivity coefficients to generate a node load sensitivity vector; The node load sensitivity vector is propagated using the topological neighborhood association propagation method to generate a network-level load fluctuation feature tensor; The trend prediction algorithm is used to extrapolate the trend of the network-level load fluctuation characteristic tensor to output the initial load prediction value of the variable network group. The initial load forecast value and the distribution network operation parameters are corrected by using a constraint correction method to generate network group load time series data; the peak-valley fluctuation optimization is performed on the network group load time series data by using a curve optimization algorithm to generate a variable grid group load curve.

7. The variable grid group state monitoring method as described in claim 6, characterized in that, The steps to generate a variable grid group status monitoring report are as follows: Spatiotemporal features are extracted from the variable grid group load curve, and the spatiotemporal features are fused using a weighted correlation method to generate an enhanced load feature matrix; The enhanced load feature matrix is ​​analyzed using a spatiotemporal clustering algorithm to capture the spatial correlation and temporal continuity of the variable network state and generate a spatiotemporal dependent weight distribution. The spatiotemporal dependent weight distribution is evaluated for cluster quality using a dual-index evaluation method to remove noisy clusters and obtain effective clusters. The effective clusters are mapped to typical grid group state patterns; the statistical characteristics of the effective clusters are fused with real-time power grid alarm information to generate multi-dimensional state indicators; and a variable grid group state monitoring report is output based on the multi-dimensional state indicators.

8. A variable grid group state monitoring system, employing the method described in any one of claims 1-7, characterized in that, include: The data processing module is used to collect multidimensional climate data, filter and standardize the multidimensional climate data, and output standardized climate data. The feature fusion module is used to extract distribution network environmental features based on standardized climate data, and to perform correlation and fusion of distribution network environmental features using the first algorithm to generate and output joint feature representations. The impact degree analysis module is used to obtain the mapping relationship between each climate factor parameter and the distribution network operation parameter based on the joint feature characterization and the multi-dimensional stratified sampling method, and to obtain the impact degree of each climate factor parameter change on the distribution network operation parameter based on the mapping relationship. The matrix construction module is used to calculate the weights of each climate factor according to its degree of influence, and to construct and output a meteorological power grid coupling index matrix by combining power distribution network operating parameters. The heat map construction module is used to obtain the identification results based on the meteorological power grid coupling index matrix, obtain the vulnerability of each node in the distribution network based on the identification results, and construct a variable network group vulnerability heat map. The load curve generation module is used to predict the load of variable grid groups based on the vulnerability heat map of variable grid groups, compare and correct the deviation between the predicted variable grid group load and the distribution network operating parameters, and generate the load curve of variable grid groups. The report generation module is used to identify the status of variable grid groups based on the variable grid group load curve, integrate the variable grid group status, and generate and output a variable grid group status monitoring report.

9. An electronic device, comprising: Memory and processor; The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions, which, when executed by the processor, implement the steps of the variable grid group state monitoring method according to any one of claims 1 to 7.

10. A computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the variable grid group state monitoring method according to any one of claims 1 to 7.

Images