An automatic modeling method and system for intelligent substation configuration files

Abstract

The present application discloses an automatic modeling method and system for intelligent substation configuration files. The method includes: receiving configuration files of multiple manufacturers and multiple formats, extracting the structural features of the configuration files, and performing data cleaning on the data records in the configuration files; constructing a configuration file parsing tree based on the structural features, and extracting the association relationships between key nodes and key nodes; performing data standardization on the data records; extracting the key features of the data records; constructing a feature association graph according to the key features, key nodes and association relationships; identifying target features in the feature association graph, and obtaining corresponding difference data sets; training a machine learning model based on the difference data sets and the feature association graph until the number of training times reaches a preset number threshold; performing hyperparameter tuning and model integration on the trained machine learning model to obtain an optimized machine learning model; deploying the optimized machine learning model to a production environment to generate a target substation configuration file.

Description

Technical Field

[0001] The embodiments of the present application relate to the technical field of model construction, and in particular, to a method and system for automatically modeling the configuration files of intelligent substations. Background Art

[0002] With the rapid development of the power system, as a core component of the power system, the safety and reliability of intelligent substations are crucial for the stable operation of the entire power grid. The configuration files of intelligent substations are the basis for substation operation and contain key information such as equipment parameters, connection relationships, and logical controls. These configuration files directly determine the operation status and performance of the substation, so their accuracy and consistency are the prerequisite for ensuring the safe operation of the substation.

[0003] However, in actual applications, due to the fact that substation equipment comes from different manufacturers, there are significant differences in the formats and contents of configuration files, such as ccd files, ini files, txt files, or xml files, etc. Due to the non-uniform format and complex content of configuration files, version management has become a major problem. There may be conflicts or inconsistencies between different versions of configuration files, increasing the risk of operation accidents in substations. Summary of the Invention

[0004] The embodiments of the present application provide a method and system for automatically modeling the configuration files of intelligent substations, which are used to solve the problem of conflicts or inconsistencies existing in different versions of configuration files and effectively reduce the risk of operation accidents in substations.

[0005] To achieve the above object, the embodiments of the present application adopt the following technical solutions:

[0006] In a first aspect, a method for automatically modeling the configuration files of intelligent substations is provided, and the method includes:

[0007] In response to receiving configuration files of multiple manufacturers and multiple formats, extract the structural features of the configuration files, and perform data cleaning on the data records in the configuration files to obtain the data records after data cleaning;

[0008] Based on the structural features, construct a configuration file parsing tree, and extract the key nodes and the association relationships between the key nodes through the configuration file parsing tree;

[0009] Perform data standardization on the data records after data cleaning to obtain the data records after standardization;

[0010] Based on the AUTO ML framework, extract the key features of the data records after standardization, where the key features include equipment failure rate, load change, number of channels, polarity, and precision compensation coefficient;

[0011] Construct a feature association graph according to the key features, the key nodes, and the association relationships, where the feature association graph is used to describe the dependency relationships and weights between the key features;

[0012] Use the AUTO ML framework to identify target features strongly related to substation configuration in the feature association graph, and obtain a corresponding difference data set from a preset difference database based on the target features;

[0013] Train a machine learning model based on the difference data set and the feature association graph until the number of training times reaches a preset number threshold, where the machine learning model is used to generate a substation configuration file;

[0014] Adjust the hyperparameters and perform model integration on the trained machine learning model, and perform structural verification on the substation configuration file through the configuration file parsing tree to obtain an optimized machine learning model, where performing structural verification on the substation configuration file is used to ensure the logical consistency of the substation configuration file;

[0015] Deploy the optimized machine learning model to a production environment to generate a target substation configuration file corresponding to the production environment.

[0016] In a possible implementation manner of the first aspect, before deploying the optimized machine learning model to a production environment to generate a target substation configuration file corresponding to the production environment, it further includes:

[0017] Collect the operation results of the optimized machine learning model in real time, and compare the operation results with preset expected results to obtain a comparison result, where the operation results are used to represent the target substation configuration file;

[0018] Calculate the error rate of the optimized machine learning model according to the comparison result;

[0019] The step of deploying the optimized machine learning model to a production environment to generate a target substation configuration file corresponding to the production environment includes:

[0020] Trigger a model optimization process to optimize the machine learning model to obtain a target machine learning model in the case that the error rate is greater than a preset threshold, and deploy the target machine learning model to a production environment to generate a target substation configuration file corresponding to the production environment;

[0021] In the case that the error rate is not greater than a preset threshold, deploy the optimized machine learning model to a production environment to generate a target substation configuration file corresponding to the production environment.

[0022] In another possible implementation of the first aspect, the model optimization process includes:

[0023] Adopt a dynamic update step to optimize the configuration file parse tree to optimize the feature association graph, and obtain an optimized feature association graph;

[0024] Use the AUTO ML framework to identify, in the optimized feature association graph, a second target feature strongly related to the substation configuration, and obtain a corresponding second difference data set in a preset difference database based on the second target feature;

[0025] Use the second difference data set and the optimized feature association graph as the input of the optimized machine learning model to obtain an optimized target substation configuration file.

[0026] In another possible implementation of the first aspect, after generating the target substation configuration file corresponding to the production environment, it includes:

[0027] Parse the target substation configuration file to obtain virtual loop information and physical loop information, where the virtual loop information includes logical connection relationships, and the physical loop information includes physical connection relationships;

[0028] Map the virtual loop information to a virtual loop diagram, and map the physical loop information to a physical loop diagram;

[0029] Align the nodes of the virtual loop diagram and the physical loop diagram so that the nodes of the virtual loop diagram correspond one by one to the nodes of the physical loop diagram;

[0030] After node alignment, generate a comparison relationship table between the virtual loop and the physical loop;

[0031] Verify the virtual loop and the physical loop according to the comparison relationship table, and obtain and output a verification result.

[0032] In another possible implementation of the first aspect, the verifying the virtual loop and the physical loop according to the comparison relationship table, obtaining and outputting a verification result includes:

[0033] Based on the comparison relationship table, compare item by item the nodes of the virtual loop and the physical loop and the connection relationships between the nodes;

[0034] In the case where the nodes or the connection relationships of the virtual loop and the physical loop are inconsistent, generate a list of inconsistent items and output the list of inconsistent items as the verification result;

[0035] When the nodes and the connection relationships of the virtual loop and the physical loop are all the same, generate a verification passed signal, and output the verification passed signal as the verification result.

[0036] In another possible implementation manner of the first aspect, the step of dynamically updating is used to optimize the configuration file parsing tree to optimize the feature association graph, and an optimized feature association graph is obtained, including:

[0037] Real-time collect the update information of the configuration file in the production environment, where the update information includes newly added configuration files, modified configuration files, and deleted configuration files;

[0038] According to the update information, dynamically adjust the nodes and edges of the configuration file parsing tree so that the configuration file parsing tree is consistent with the configuration file in the production environment;

[0039] Based on the dynamically adjusted configuration file parsing tree, re-extract the second key nodes and the second association relationships, and update the feature association graph based on the second key nodes and the second association relationships to obtain an optimized feature association graph.

[0040] In another possible implementation manner of the first aspect, the structural feature includes a hierarchical structure, and the hierarchical structure includes the attributes of the configuration file and the association relationships between the attributes. Building a configuration file parsing tree based on the structural feature and extracting the key nodes and the association relationships between the key nodes through the configuration file parsing tree includes:

[0041] Use the attributes of the configuration file as nodes and the association relationships between the attributes as edges to build a configuration file parsing tree;

[0042] Traverse the configuration file parsing tree to extract the key nodes and the association relationships between the key nodes. The key nodes include equipment failure rate, load change, number of channels, polarity, and precision compensation coefficient.

[0043] In another possible implementation manner of the first aspect, performing hyperparameter tuning and model integration on the trained machine learning model, and performing structural verification on the substation configuration file through the configuration file parsing tree to obtain an optimized machine learning model, including:

[0044] When the number of machine learning models is at least two, for each trained machine learning model, calculate the model evaluation result, and adjust the hyperparameters of the machine learning model according to the model evaluation result. The hyperparameters include learning rate, regularization coefficient, and tree depth;

[0045] Obtain the prediction results of each of the machine learning models after hyperparameter tuning, and perform weighted averaging on the prediction results of all the machine learning models to obtain an ensemble model. Take the weighted average prediction result as the model output result of the ensemble model, and the model output result is the substation configuration file;

[0046] Import the model output result into the configuration file parsing tree, so that the configuration file parsing tree performs structural verification on the substation configuration file;

[0047] If the verification passes, take the ensemble model as the optimized machine learning model;

[0048] If the verification fails, readjust the hyperparameters until the verification passes to obtain the optimized machine learning model.

[0049] In a second aspect, the present application provides an electronic device, including:

[0050] A memory configured to store instructions; and

[0051] A processor configured to call the instructions from the memory and be capable of implementing the above-mentioned intelligent substation configuration file automatic modeling method when executing the instructions.

[0052] In a third aspect, the present application provides an intelligent substation configuration file automatic modeling system, including:

[0053] The above-mentioned electronic device.

[0054] Through the above technical solutions, by responding to the receipt of configuration files in multiple formats from multiple manufacturers, extracting the structural features of the configuration files, and performing data cleaning on the data records in the configuration files to obtain the data records after data cleaning, the problem of data processing complexity brought by multi-format configuration files is effectively solved, ensuring the integrity and accuracy of the configuration file data; based on the structural features, a configuration file parsing tree is constructed, and the key nodes and the association relationships between the key nodes are extracted through the configuration file parsing tree, solving the problem of complex configuration file content and providing a basis for subsequent feature extraction and modeling; the data records after data cleaning are standardized to obtain the standardized data records, further unifying the data format and providing convenience for subsequent processing; based on the AUTO ML framework, the key features of the standardized data records are extracted, where the key features include equipment failure rate, load change, number of channels, polarity, and precision compensation coefficient, solving the problem of difficult manual verification and modeling and improving the accuracy and efficiency of modeling; according to the key features, key nodes, and association relationships, a feature association graph is constructed, and the feature association graph is used to describe the dependence relationships and weights between the key features, providing a more comprehensive input for model training; the AUTO ML framework is used to identify the target features strongly related to the substation configuration in the feature association graph, and the corresponding difference data set is obtained based on the target features in the preset difference database, ensuring the relevance and effectiveness of the model training data; the machine learning model is trained based on the difference data set and the feature association graph until the number of training times reaches the preset number threshold, and the machine learning model is used to generate the substation configuration file, solving the problem of low verification efficiency and realizing the automatic verification and optimization of the configuration file; the hyperparameters of the trained machine learning model are adjusted and the model is integrated, and the structure of the substation configuration file is verified through the configuration file parsing tree to obtain the optimized machine learning model, where verifying the structure of the substation configuration file is used to ensure the logical consistency of the substation configuration file, solving the problem of difficult version management and ensuring the logical consistency and security of the configuration file; the optimized machine learning model is deployed to the production environment to generate the target substation configuration file corresponding to the production environment, finally realizing the automatic generation and deployment of the configuration file, significantly improving the management efficiency and accuracy of the substation configuration file, reducing the risk of operation accidents, and providing a strong guarantee for the safe operation of the intelligent substation. It not only solves the problem of data processing complexity brought by multi-format configuration files, but also improves the accuracy and consistency of the configuration file through automatic modeling and verification, effectively reducing the risk of substation operation accidents and providing solid technical support for the stable operation of the power system.

[0055] Other features and advantages of the embodiments of the present application will be described in detail in the subsequent specific implementation part. Brief Description of the Drawings

[0056] Figure 1 Schematic flowchart of an automatic modeling method for intelligent substation configuration files provided by an embodiment of the present application;

[0057] Figure 2 Schematic structural diagram of a configuration file parsing tree provided by an embodiment of the present application;

[0058] Figure 3 Schematic structural diagram of a feature association graph provided by an embodiment of the present application;

[0059] Figure 4 Schematic flowchart of a model optimization process provided by an embodiment of the present application. Detailed implementation manners

[0060] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present application. It should be understood that the specific implementation manners described herein are only for explaining and illustrating the embodiments of the present application, and are not used to limit the embodiments of the present application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.

[0061] It should be noted that if there are directional indications (such as up, down, left, right, front, back,...) involved in the embodiments of the present application, such directional indications are only used to explain the relative positional relationship and movement conditions between components in a specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indications will also change accordingly.

[0062] In addition, if there are descriptions involving "first", "second", etc. in the embodiments of the present application, such descriptions of "first", "second", etc. are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one such feature. In addition, the technical solutions between various embodiments may be combined with each other, but it must be based on the fact that those of ordinary skill in the art can implement them. When the combination of technical solutions results in contradictions or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the protection scope required by the present application.

[0063] Figure 1 Schematically shows a schematic flowchart of an automatic modeling method for intelligent substation configuration files according to an embodiment of the present application. As Figure 1 shown, an embodiment of the present application provides an automatic modeling method for intelligent substation configuration files, and the method may include the following steps.

[0064] S110. In response to receiving configuration files of multiple manufacturers and multiple formats, extract the structural features of the configuration files, and perform data cleaning on the data records in the configuration files to obtain the data records after data cleaning;

[0065] S120. Based on the structural features, construct a configuration file parsing tree, and extract the key nodes and the association relationships between the key nodes through the configuration file parsing tree;

[0066] S130. Perform data standardization on the data records after data cleaning to obtain the data records after standardization;

[0067] S140. Based on the AUTO ML framework, extract the key features of the data records after standardization, where the key features include equipment failure rate, load change, number of channels, polarity, and precision compensation coefficient;

[0068] S150. According to the key features, key nodes, and association relationships, construct a feature association graph, which is used to describe the dependency relationships and weights between the key features;

[0069] S160. Use the AUTO ML framework to identify the target features strongly related to the substation configuration in the feature association graph, and obtain the corresponding difference data set in the preset difference database based on the target features;

[0070] S170. Based on the difference data set and the feature association graph, train the machine learning model until the number of training times reaches the preset number of times threshold. The machine learning model is used to generate substation configuration files;

[0071] S180. Perform hyperparameter tuning and model integration on the trained machine learning model, and perform structural verification on the substation configuration file through the configuration file parsing tree to obtain the optimized machine learning model. Among them, performing structural verification on the substation configuration file is used to ensure the logical consistency of the substation configuration file;

[0072] S190. Deploy the optimized machine learning model to the production environment to generate the target substation configuration file corresponding to the production environment.

[0073] During the operation and maintenance process of intelligent substations, the sources of configuration files are diverse, and the configuration file formats provided by different manufacturers vary, such as ccd files, ini files, txt files, xml files, etc. The differences in file formats are not only reflected in the file extensions but also in the internal structure and content organization of the files. To process these multi-format configuration files, it is first necessary to extract the structural features of the configuration files. The structural features include the hierarchical structure of the file, node types, the association relationships between nodes, and the data storage method, etc. For example, xml files usually represent the hierarchical structure in a way of nested tags, while ini files store data in the form of key-value pairs. By parsing the structural features of these files, they can be converted into a unified internal representation form for subsequent processing. After extracting the structural features, it is necessary to perform data cleaning on the data records in the configuration files. The purpose of data cleaning is to remove invalid, incorrect, and incomplete data records to ensure the integrity and accuracy of the data. The specific steps of data cleaning include removing duplicate records, filling in missing values, correcting incorrect data, and filtering out irrelevant data, etc. For example, some configuration files may contain records of equipment failure rates, but due to errors in the data collection or transmission process, some records may be missing or abnormal. Through data cleaning, these problems can be identified and fixed to ensure the reliability of subsequent processing. The data records after data cleaning not only have a unified format but also complete content, providing a high-quality data foundation for subsequent parsing and modeling.

[0074] After extracting the structural features of the configuration file, a configuration file parsing tree is constructed based on these structural features. Figure 2 A schematic structural diagram of a configuration file parsing tree according to an embodiment of the present application is shown, as Figure 2As shown, the configuration file parsing tree is a tree-shaped data structure used to represent the hierarchical structure and node relationships of a configuration file. Each node in the tree represents an attribute or parameter in the configuration file, and the edges between nodes represent the association relationships between attributes. For example, in an xml file, each tag can be used as a node, and the nested relationships between tags can be represented as parent-child node relationships. By constructing the configuration file parsing tree, the structure of the configuration file can be intuitively displayed, facilitating subsequent node extraction and relationship analysis. After constructing the parsing tree, by traversing the tree structure, the key nodes and the association relationships between key nodes are extracted. Key nodes refer to attributes or parameters that have an important impact on the substation configuration, such as equipment failure rate, load change, number of channels, polarity, and precision compensation coefficient, etc. These nodes are usually located at higher levels of the parsing tree and have strong associations with other nodes. By extracting key nodes and association relationships, the structure of the configuration file can be simplified, highlighting the core information therein, and providing a basis for subsequent feature extraction and modeling. For example, when extracting the equipment failure rate node, the association relationship between it and the load change node can be extracted simultaneously to analyze the mutual influence between the two. Through the configuration file parsing tree, not only can key information be efficiently extracted, but also the structure and logical relationships of the configuration file can be clearly displayed, providing strong support for subsequent processing.

[0075] After the data cleaning is completed, data standardization processing needs to be performed on the data records. The purpose of data standardization is to convert data in different formats and units into a unified representation form to ensure the consistency and comparability of the data. The specific steps of data standardization include data format conversion, unit unification, and data normalization, etc. For example, the unit of the equipment failure rate in some configuration files may be a percentage, while in other files it may be in decimal form. Through data standardization, all equipment failure rates can be converted into a unified percentage form for subsequent comparison and analysis. In addition, data standardization also includes normalizing the data, scaling the data to the same range, for example, normalizing the equipment failure rate to between 0 and 1. The normalization process can eliminate the dimensional differences between the data and improve the accuracy and stability of subsequent modeling. The data records after data standardization not only have a unified format but also a consistent numerical range, providing a high-quality data basis for subsequent feature extraction and modeling. For example, in the data records after standardization, key features such as equipment failure rate, load change, number of channels, polarity, and precision compensation coefficient are all represented in a unified format and unit, facilitating subsequent analysis and processing. Through data standardization, not only the consistency and comparability of the data are improved, but also the complexity of subsequent processing is simplified, providing strong support for automated modeling.

[0076] After data standardization is completed, key features of the standardized data records are extracted based on the AUTO ML framework. The AUTO ML framework is an automated machine learning tool that can automatically identify and extract features strongly correlated with the target variable. Key features refer to attributes or parameters that have an important impact on substation configuration, such as equipment failure rate, load variation, number of channels, polarity, and precision compensation coefficient, etc. Through the AUTO ML framework, the feature importance in the data records can be automatically analyzed, and the features that have the greatest impact on substation configuration can be screened out. For example, the equipment failure rate is an important indicator to measure the operating state of equipment, the load variation reflects the load fluctuation of the substation, and the number of channels, polarity, and precision compensation coefficient directly affect the performance and precision of the equipment. By extracting these key features, the dimension of the data records can be simplified, the core information can be highlighted, and high-quality input can be provided for subsequent modeling. The AUTO ML framework can not only automatically extract key features, but also preprocess the features, such as feature scaling, feature encoding, and feature combination, etc., to further improve the quality and usability of the features. Through the AUTO ML framework, not only the efficiency and accuracy of feature extraction are improved, but also the need for manual intervention is reduced, providing strong support for automated modeling.

[0077] After extracting the key features, a feature association graph is constructed based on the key features, key nodes, and association relationships. Figure 3 The structure diagram of a feature association graph provided by an embodiment of the present application is shown, as Figure 3 shown, the feature association graph is a graph structure used to describe the dependence relationship and weight between key features. Each key node in the graph represents a key feature, the edge represents the dependence relationship between features, and the weight represents the strength of the dependence relationship. For example, there may be a strong dependence relationship with a high weight between the equipment failure rate node and the load variation node; while the dependence relationship between the number of channels node and the polarity node may be weak with a low weight. By constructing the feature association graph, the relationship between key features can be intuitively displayed, facilitating subsequent analysis and modeling. The construction process of the feature association graph includes steps such as node definition, edge connection, and weight calculation. Node definition means taking the key features as the nodes of the graph, edge connection means connecting the nodes according to the association relationship between key nodes, and weight calculation means calculating the weight of the edge according to the dependence strength between features. For example, the dependence strength between features can be calculated by methods such as correlation coefficient or mutual information and used as the weight of the edge. Through the feature association graph, not only can the relationship between key features be clearly displayed, but also important input information can be provided for subsequent model training, improving the accuracy and interpretability of modeling.

[0078] After constructing the feature association graph, use the AUTO ML framework to identify target features in the graph that are strongly correlated with the substation configuration. Target features refer to features that have a significant impact on the substation configuration, such as equipment failure rate, load change, number of channels, polarity, and precision compensation coefficient. Through the AUTO ML framework, the nodes and edges in the feature association graph can be automatically analyzed to screen out target features that are strongly correlated with the substation configuration. For example, the edge weight between the equipment failure rate node and the load change node is relatively high, indicating that both have a significant impact on the substation configuration, so they can be used as target features. After identifying the target features, obtain the corresponding difference dataset in the preset difference database based on these features. The difference dataset refers to the records of substation configuration files over a period of time in the past, which contains the historical values of the target features and the corresponding configuration results. Through the difference dataset, rich sample data can be provided for subsequent model training, improving the accuracy and generalization ability of modeling. For example, according to the difference data of equipment failure rate and load change, a model can be trained to predict future substation configuration requirements. Through the AUTO ML framework, not only the efficiency and accuracy of target feature identification are improved, but also a high-quality data foundation is provided for model training, providing strong support for automated modeling.

[0079] After obtaining the difference dataset, train a machine learning model based on the difference dataset and the feature association graph. A machine learning model refers to an algorithm that can learn patterns from data and make predictions, such as decision trees, random forests, or neural networks. Through the difference dataset, rich training samples can be provided for the model, enabling it to learn the patterns and rules of substation configuration. The feature association graph provides important input information for the model, helping the model understand the relationships between key features. The training process includes steps such as data preprocessing, model selection, parameter setting, and model evaluation. Data preprocessing refers to cleaning and standardizing the difference dataset to ensure the consistency and comparability of the data. Model selection refers to selecting a suitable machine learning algorithm according to the task requirements. For example, decision trees are suitable for classification tasks, and neural networks are suitable for complex non-linear tasks. Parameter setting refers to setting hyperparameters for the model, such as learning rate, regularization coefficient, and tree depth, to optimize the model performance. Model evaluation refers to evaluating the accuracy and generalization ability of the model through cross-validation or a test set. The training process continues until the number of training times reaches the preset number threshold to ensure that the model performance reaches the optimal. The trained machine learning model can automatically generate substation configuration files based on the input key features, providing strong support for the operation and maintenance of substations.

[0080] After training is completed, hyperparameter tuning and model ensemble are performed on the machine learning model to further optimize the model performance. Hyperparameter tuning refers to finding the optimal combination of hyperparameters, such as learning rate, regularization coefficient, and tree depth, through methods like grid search or random search. Model ensemble means performing weighted averaging or voting on the prediction results of multiple models to improve the accuracy and stability of the model. For example, the prediction results of decision trees, random forests, and neural networks can be integrated to generate the final substation configuration file. After model optimization is completed, the generated substation configuration file is subjected to structural verification through a configuration file parsing tree. The purpose of structural verification is to ensure the logical consistency of the configuration file, such as checking whether the relationship between equipment failure rate and load change meets expectations, and whether the configuration between the number of channels and polarity is reasonable. Through structural verification, logical errors in the configuration file can be identified and repaired to ensure its reliability and security in practical applications. The optimized machine learning model not only has better performance but also generates a configuration file with consistent logic, providing high-quality support for the operation and maintenance of substations.

[0081] After model optimization and verification are completed, the optimized machine learning model is deployed to the production environment to generate the target substation configuration file corresponding to the production environment. The production environment refers to the actual operating substation system, and its configuration requirements may differ from the differential data. By deploying the optimized model, a substation configuration file that meets the current requirements can be automatically generated based on real-time production environment data. For example, when the equipment failure rate or load changes in the production environment, the model can adjust the configuration file in real time to ensure the safe and stable operation of the substation. The generated configuration file is verified for its correctness and security through simulation testing and field testing. Simulation testing means running the configuration file in a simulation environment to check whether it meets expectations; field testing means deploying the configuration file in an actual substation and monitoring its operating status. Through testing, the reliability and effectiveness of the generated configuration file in practical applications can be ensured. The optimized machine learning model can not only efficiently generate configuration files but also adjust in real time according to changes in the production environment, providing strong support for the operation and maintenance of substations.

[0082] In this embodiment, by responding to the receipt of configuration files in multiple formats from multiple manufacturers, extracting the structural features of the configuration files, and performing data cleaning on the data records in the configuration files to obtain the data records after data cleaning, the problem of data processing complexity brought by multi-format configuration files is effectively solved, ensuring the integrity and accuracy of the configuration file data; based on the structural features, a configuration file parsing tree is constructed, and the key nodes and the association relationships between the key nodes are extracted through the configuration file parsing tree, solving the problem of complex configuration file content and providing a basis for subsequent feature extraction and modeling; the data records after data cleaning are standardized to obtain the standardized data records, further unifying the data format and facilitating subsequent processing; based on the AUTO ML framework, the key features of the standardized data records are extracted, where the key features include equipment failure rate, load change, number of channels, polarity, and precision compensation coefficient, solving the problem of difficult manual verification and modeling, and improving the accuracy and efficiency of modeling; according to the key features, key nodes, and association relationships, a feature association graph is constructed, and the feature association graph is used to describe the dependence relationship and weight between the key features, providing a more comprehensive input for model training; the AUTO ML framework is used to identify the target features strongly related to the substation configuration in the feature association graph, and the corresponding difference data set is obtained from the preset difference database based on the target features, ensuring the relevance and effectiveness of the model training data; the machine learning model is trained based on the difference data set and the feature association graph until the number of training times reaches the preset number threshold, and the machine learning model is used to generate the substation configuration file, solving the problem of low verification efficiency and realizing the automatic verification and optimization of the configuration file; the hyperparameters of the trained machine learning model are adjusted and the model is integrated, and the structure of the substation configuration file is verified through the configuration file parsing tree to obtain the optimized machine learning model, where verifying the structure of the substation configuration file is used to ensure the logical consistency of the substation configuration file, solving the problem of difficult version management and ensuring the logical consistency and security of the configuration file; the optimized machine learning model is deployed to the production environment to generate the target substation configuration file corresponding to the production environment, finally realizing the automatic generation and deployment of the configuration file, significantly improving the management efficiency and accuracy of the substation configuration file, reducing the risk of operation accidents, and providing a strong guarantee for the safe operation of the intelligent substation. It not only solves the problem of data processing complexity brought by multi-format configuration files, but also improves the accuracy and consistency of the configuration file through automatic modeling and verification, effectively reducing the risk of substation operation accidents and providing solid technical support for the stable operation of the power system.

[0083] In one implementation manner of this embodiment, before deploying the optimized machine learning model to the production environment to generate the target substation configuration file corresponding to the production environment, the following steps are further included:

[0084] S210. Collect the operation results of the optimized machine learning model in real time, compare the operation results with the preset expected results to obtain a comparison result, where the operation results are used to represent the target substation configuration file;

[0085] S220. Calculate the error rate of the optimized machine learning model according to the comparison result;

[0086] Deploy the optimized machine learning model to the production environment to generate a target substation configuration file corresponding to the production environment, including:

[0087] S230. Trigger a model optimization process when the error rate is greater than the preset threshold to optimize the machine learning model to obtain a target machine learning model, and deploy the target machine learning model to the production environment to generate a target substation configuration file corresponding to the production environment;

[0088] S240. When the error rate is not greater than the preset threshold, deploy the optimized machine learning model to the production environment to generate a target substation configuration file corresponding to the production environment.

[0089] Before the optimized machine learning model is deployed to the production environment, it is necessary to collect the running results of the model in real time and compare them with the preset expected results to evaluate the performance and accuracy of the model. The running results refer to the substation configuration files generated by the model, which contain key parameters such as equipment failure rate, load change, number of channels, polarity, and precision compensation coefficient. These parameters directly determine the operating status and performance of the substation, so their accuracy is crucial. The preset expected results refer to the standard values or ranges set according to the differential data and expert experience, which are used to measure the rationality of the configuration files generated by the model. For example, the expected result of the equipment failure rate may be set to not exceed 5%, and the expected result of the load change may be set to fluctuate within the range of ±10%. By collecting the running results in real time, the performance of the model in actual applications can be obtained, providing data support for subsequent error rate calculation and model optimization. The comparison process includes steps such as data extraction, data alignment, and difference calculation. Data extraction refers to extracting key parameters from the running results and expected results, such as equipment failure rate, load change, etc. Data alignment refers to matching the parameters in the running results and expected results to ensure that the comparison objects are consistent. Difference calculation refers to calculating the difference between the running results and the expected results. For example, if the actual value of the equipment failure rate is 6% and the expected value is 5%, the difference is 1%. Through comparison, detailed comparison results can be generated for subsequent error rate calculation and model optimization. For example, the comparison results may show that the difference in the equipment failure rate is 1%, the difference in the load change is 2%, and the difference in the number of channels is 0, etc. Through real-time collection and comparison, not only can problems in the configuration files generated by the model be discovered in a timely manner, but also data support can be provided for subsequent error rate calculation and model optimization, ensuring the accuracy and reliability of the model in actual applications.

[0090] After generating the comparison result, calculate the error rate of the optimized machine learning model based on the comparison result. The error rate refers to the degree of difference between the configuration file generated by the model and the expected result, and is used to measure the accuracy and performance of the model. The calculation of the error rate includes steps such as difference summarization, weight assignment, and error rate calculation. Difference summarization means summarizing the differences in the comparison result. For example, the difference in equipment failure rate is 1%, the difference in load change is 2%, and the difference in the number of channels is 0. Weight assignment means assigning different weights to each parameter according to its importance to the substation configuration. For example, the equipment failure rate has a greater impact on the substation configuration and may be assigned a higher weight, such as 0.5; the impact of load change is secondary and may be assigned a medium weight, such as 0.3; the impact of the number of channels is smaller and may be assigned a lower weight, such as 0.2. Error rate calculation means calculating the overall error rate of the model based on the differences and weights. For example, the difference in equipment failure rate is 1% and the weight is 0.5, then the error contributed by it is 0.5%; the difference in load change is 2% and the weight is 0.3, then the error contributed by it is 0.6%; the difference in the number of channels is 0 and the weight is 0.2, then the error contributed by it is 0%. The overall error rate is the sum of the errors contributed by each parameter, that is, 0.5% + 0.6% + 0% = 1.1%. By calculating the error rate, the difference between the configuration file generated by the model and the expected result can be quantified, providing a basis for subsequent model optimization. For example, an error rate of 1.1% indicates that the accuracy of the configuration file generated by the model is relatively high, but there is still room for improvement. By calculating the error rate, not only can the performance of the model be objectively evaluated, but also data support can be provided for subsequent model optimization to ensure the accuracy and reliability of the model in practical applications.

[0091] When the error rate is greater than the preset threshold, the model optimization process is triggered to further optimize the machine learning model and improve the accuracy and reliability of the generated configuration file. The preset threshold refers to the upper limit of the error rate set according to actual application requirements, such as 1%. When the error rate exceeds the preset threshold, it indicates that the accuracy of the model in generating the configuration file does not meet the requirements and needs to be optimized. The model optimization process includes steps such as re-collecting data, re-selecting features, re-training the model, and re-adjusting hyperparameters. Re-collecting data means re-collecting data from the production environment to ensure the timeliness and representativeness of the training data. Re-selecting features means re-selecting features that are strongly correlated with the substation configuration based on the latest data, such as equipment failure rates and load changes. Re-training the model means using the re-collected data and re-selected features to re-train the model to improve its learning ability and generalization ability. Re-adjusting hyperparameters means re-adjusting the hyperparameters of the model, such as the learning rate, regularization coefficient, and tree depth, through methods such as grid search or random search to optimize the model performance. Through the model optimization process, the target machine learning model can be obtained, and the accuracy and reliability of the generated configuration file are significantly improved. For example, the error rate of the optimized model is reduced from 1.1% to 0.8%, indicating a significant improvement in the accuracy of the generated configuration file. The target machine learning model is deployed to the production environment to generate the target substation configuration file corresponding to the production environment, ensuring the safe and stable operation of the substation. Through the model optimization process, not only can the performance of the model be improved, but also the accuracy and reliability of the generated configuration file can be ensured, providing strong support for the operation and maintenance of the substation.

[0092] When the error rate is not greater than the preset threshold, it indicates that the accuracy of the optimized machine learning model in generating the configuration file has met the requirements, and it can be directly deployed to the production environment to generate the target substation configuration file corresponding to the production environment. The preset threshold refers to the upper limit of the error rate set according to the actual application requirements, such as 1%. When the error rate does not exceed the preset threshold, it indicates that the accuracy of the model in generating the configuration file is relatively high and no further optimization is required. The deployment process includes steps such as model export, configuration file generation, and configuration file verification. Model export means exporting the optimized machine learning model as an executable file or API interface for easy invocation in the production environment. Configuration file generation means calling the model to generate the substation configuration file according to the real-time data of the production environment, such as equipment failure rate, load change, number of channels, polarity, and precision compensation coefficient. Configuration file verification means conducting simulation tests and on-site tests on the generated configuration file to ensure its correctness and security. For example, simulation testing means running the configuration file in a simulation environment to check whether it meets the expectations; on-site testing means deploying the configuration file in an actual substation and monitoring its operation status. By deploying the optimized machine learning model, the substation configuration file that meets the requirements of the production environment can be efficiently generated, ensuring the safe and stable operation of the substation. For example, the equipment failure rate in the generated configuration file is 4.9%, the load change is ±9%, the number of channels is 10, and the polarity and precision compensation coefficient both meet the requirements, indicating its high accuracy and reliability. By deploying the optimized model, not only can the efficiency of generating the configuration file be improved, but also its accuracy and reliability in actual applications can be ensured, providing strong support for the operation and maintenance of the substation.

[0093] In this embodiment, by collecting the operation results of the optimized machine learning model in real time and comparing them with the preset expected results, problems in the configuration file generated by the model can be discovered in a timely manner, providing data support for subsequent error rate calculation and model optimization. Calculating the error rate based on the comparison results can objectively evaluate the performance of the model and provide a basis for subsequent model optimization. When the error rate is greater than the preset threshold, the model optimization process is triggered to further optimize the machine learning model and improve the accuracy and reliability of the configuration file it generates. When the error rate is not greater than the preset threshold, the optimized model is directly deployed to the production environment to generate the target substation configuration file corresponding to the production environment, ensuring the safe and stable operation of the substation. Through the above steps, not only can the accuracy and reliability of the model in generating the configuration file be improved, but also its efficiency and security in actual applications can be ensured, providing strong support for the operation and maintenance of the substation.

[0094] In one implementation of this embodiment, as Figure 4 shown, the model optimization process includes the following steps:

[0095] S310. Optimize the configuration file parsing tree using dynamic update steps to optimize the feature association graph and obtain an optimized feature association graph;

[0096] S320. Use the AUTO ML framework to identify second target features strongly related to the substation configuration in the optimized feature association graph, and obtain the corresponding second difference dataset in the preset difference database based on the second target features;

[0097] S330. Use the second difference dataset and the optimized feature association graph as the input of the optimized machine learning model to obtain an optimized target substation configuration file.

[0098] During the automatic modeling process of the intelligent substation configuration file, the configuration file parsing tree is one of the core data structures, which is used to describe the hierarchical structure of the configuration file and the association relationships between nodes. However, with the dynamic changes of the configuration file in the production environment, such as adding, modifying, or deleting the configuration file, the original configuration file parsing tree may not accurately reflect the current actual state, resulting in deviations in the feature association graph. Therefore, adopting dynamic update steps to optimize the configuration file parsing tree is the key to ensuring the accuracy of the feature association graph.

[0099] The implementation principle of dynamic update lies in real-time monitoring of the changes in the configuration file in the production environment and adjusting the node and edge structures of the parsing tree according to these changes. The specific implementation methods include: First, collect the configuration file update information in the production environment in real time through a file monitoring mechanism, such as newly added configuration files, modified configuration file contents, or deleted configuration file records. Second, dynamically adjust the node and edge structures of the parsing tree according to the update information. For example, when a new configuration file is added, a new node will be added to the parsing tree, and an association edge will be established with the existing nodes according to its attributes; when a configuration file is modified, the corresponding node and its association edges in the parsing tree will be updated; when a configuration file is deleted, the corresponding node and its association edges in the parsing tree will be removed. Finally, based on the dynamically adjusted parsing tree, extract the key nodes and association relationships again and update the feature association graph. This process ensures that the feature association graph always remains consistent with the configuration file in the production environment, thereby improving the accuracy of modeling.

[0100] The dynamic update steps significantly improve the real-time performance and adaptability of the configuration file parsing tree, avoid modeling deviations caused by configuration file changes, and provide a reliable data basis for subsequent feature extraction and model training.

[0101] In the optimized feature association graph, identifying the second target features that are strongly correlated with the substation configuration is the core step of the model optimization process. The AUTO ML framework can efficiently extract the features most influential on the substation configuration from the complex feature association graph through automated feature selection and model training.

[0102] The specific implementation methods include: First, input the optimized feature association graph into the AUTO ML framework. The framework will automatically screen out the second target features that are strongly correlated with the substation configuration according to the dependency relationships and weights between features. For example, features such as equipment failure rate, load change, and number of channels may be identified as the second target features. Second, based on the identified second target features, retrieve the relevant second difference dataset in the preset difference database. The difference database stores the past configuration file data and their corresponding operation results. By matching the second target features, the difference data similar to the current configuration file scenario can be extracted. For example, if the second target features include equipment failure rate and load change, all the configuration file data containing these features and their operation results will be retrieved from the difference database. Finally, use the extracted second difference dataset as the input data for model optimization. This ensures that the data basis for model optimization has high relevance and representativeness. Through the automated feature identification and difference data retrieval of the AUTO ML framework, the efficiency and accuracy of model optimization are significantly improved, providing reliable data support for generating high-quality substation configuration files.

[0103] After obtaining the second difference dataset and the optimized feature association graph, use them as the input of the machine learning model to perform model optimization and generate the target substation configuration file. Specifically, first, input the second difference dataset and the optimized feature association graph into the machine learning model. The model will perform training and optimization based on these data. For example, use the configuration file data and their operation results in the second difference dataset to train the model to learn the mapping relationship between the configuration file and the operation result. Second, during the training process, the model will adjust the feature weights according to the optimized feature association graph to ensure that the features strongly correlated with the substation configuration obtain higher weights. For example, the weights of the second target features such as equipment failure rate and load change in the model will be significantly increased. Finally, through multiple iterative trainings, the model will gradually optimize its prediction ability and generate the target substation configuration file corresponding to the production environment. For example, the model will generate a configuration file containing equipment parameters, connection relationships, and logical control information according to the requirements of the current production environment. In terms of implementation effects, by using the second difference dataset and the optimized feature association graph as the input, the prediction accuracy of the machine learning model and the quality of the generated configuration file have been significantly improved, ensuring a high degree of matching between the generated configuration file and the production environment.

[0104] In this embodiment, by dynamically updating the configuration file parsing tree, using the AUTO ML framework to identify the second target features, and taking the second difference data set and the optimized feature association graph as model inputs, the model optimization process significantly improves the accuracy and efficiency of automatic modeling of intelligent substation configuration files. The dynamic update step ensures the real-time performance and adaptability of the configuration file parsing tree and the feature association graph, avoiding modeling deviations caused by changes in the configuration file. The automatic feature recognition and difference data retrieval of the AUTO ML framework improve the data relevance and representativeness of model optimization. Finally, through the optimized training of the machine learning model, the generated substation configuration file highly matches the production environment, ensuring the safety and reliability of substation operation. This process not only solves the complexity problems brought by multi-vendor and multi-format configuration files, but also significantly reduces the time cost of configuration file generation and verification, providing strong technical support for the efficient operation and maintenance of intelligent substations.

[0105] In one implementation of this embodiment, after generating the target substation configuration file corresponding to the production environment, the following steps are included:

[0106] S410. Parse the target substation configuration file to obtain virtual loop information and physical loop information, where the virtual loop information includes logical connection relationships, and the physical loop information includes physical connection relationships;

[0107] S420. Map the virtual loop information to a virtual loop diagram, and map the physical loop information to a physical loop diagram;

[0108] S430. Align the nodes of the virtual loop diagram and the physical loop diagram so that the nodes of the virtual loop diagram correspond one by one to the nodes of the physical loop diagram;

[0109] S440. After node alignment, generate a comparison relationship table between the virtual loop and the physical loop;

[0110] S450. Verify the virtual loop and the physical loop according to the comparison relationship table, obtain the verification result and output it.

[0111] After generating the target substation configuration file corresponding to the production environment, it is first necessary to parse this configuration file to extract the virtual circuit information and physical circuit information from it. The virtual circuit information mainly describes the logical connection relationships between devices in the substation, such as signal transmission paths, control logics, etc. These information usually exist in the configuration file in the form of logical nodes and logical connections. The physical circuit information describes the physical connection relationships between devices in the substation, such as cable connections, device interfaces, etc. These information usually exist in the configuration file in the form of physical nodes and physical connections. The parsing process identifies and extracts the relevant data of the virtual circuit and physical circuit by reading specific fields and structures of the configuration file. For example, for the virtual circuit information, it can be obtained by parsing the logical node tags and logical connection tags in the configuration file; for the physical circuit information, it can be obtained by parsing the physical node tags and physical connection tags in the configuration file. After parsing, the virtual circuit information and physical circuit information are respectively stored as structured data for subsequent processing.

[0112] After extracting the virtual circuit information and physical circuit information, it is necessary to map this information to a virtual circuit diagram and a physical circuit diagram respectively. The virtual circuit diagram is a graphical representation used to show the logical connection relationships between devices in the substation. Each logical node is represented as a vertex in the diagram, and each logical connection is represented as an edge between the vertices. The physical circuit diagram is used to show the physical connection relationships between devices in the substation. Each physical node is represented as a vertex in the diagram, and each physical connection is represented as an edge between the vertices. The mapping process is achieved by converting the parsed virtual circuit information and physical circuit information into a graph data structure. For example, for the virtual circuit information, each logical node can be mapped to a vertex in the graph, and each logical connection can be mapped to an edge in the graph; for the physical circuit information, each physical node can be mapped to a vertex in the graph, and each physical connection can be mapped to an edge in the graph. After mapping, the virtual circuit diagram and the physical circuit diagram respectively show the logical and physical connection relationships of the substation in a graphical form. Finally, the complex virtual circuit and physical circuit information are converted into an intuitive graphical representation, facilitating subsequent node alignment and verification operations.

[0113] After generating the virtual circuit diagram and the physical circuit diagram, it is necessary to align the nodes of these diagrams to ensure that each logical node in the virtual circuit diagram corresponds one-to-one with each physical node in the physical circuit diagram. The process of node alignment is achieved by comparing the node attributes in the virtual circuit diagram and the physical circuit diagram. For example, matching can be performed based on attributes such as the name, type, and location of the nodes. For each logical node in the virtual circuit diagram, search for the physical node with the same or similar attributes in the physical circuit diagram and align them. If a logical node cannot find a corresponding physical node in the physical circuit diagram, it is marked as an unaligned node. After the node alignment is completed, the nodes in the virtual circuit diagram and the physical circuit diagram form a one-to-one correspondence, providing a basis for subsequent comparison and verification, thus effectively ensuring that the node relationship between the virtual circuit and the physical circuit is clear and definite, and reducing errors and omissions in the subsequent verification process.

[0114] After completing the node alignment, it is necessary to generate a comparison relationship table between the virtual circuit and the physical circuit to record the corresponding relationship and differences between the virtual circuit diagram and the physical circuit diagram. The comparison relationship table includes the following content: the corresponding relationship between the logical nodes in the virtual circuit diagram and the physical nodes in the physical circuit diagram, the corresponding relationship between the logical connections in the virtual circuit diagram and the physical connections in the physical circuit diagram, the unaligned nodes and connections, and the description of the differences between the nodes and connections. The process of generating the comparison relationship table is achieved by traversing the virtual circuit diagram and the physical circuit diagram and comparing the attributes of the nodes and connections item by item. For example, for each aligned node, record its attributes such as name, type, and location; for each aligned connection, record its start point, end point, type, etc.; for the unaligned nodes and connections, record their specific information and possible reasons. The comparison relationship table is stored in a structured form for subsequent verification and output.

[0115] After generating the comparison relationship table, it is necessary to verify the virtual circuit and the physical circuit according to this table to determine their consistency and correctness. The verification process is achieved by checking the nodes and connections in the comparison relationship table item by item. For example, for each aligned node, check whether its attributes are consistent; for each aligned connection, check whether its start point, end point, and type are consistent; for the unaligned nodes and connections, analyze their possible reasons and impacts. After the verification is completed, a verification result is generated, including the consistent nodes and connections, the inconsistent nodes and connections, and the specific information and reason analysis of the unaligned nodes and connections. The verification result is output in a structured form, such as generating a verification report or a visualization chart, which is convenient for operation and maintenance personnel to view and process, so that the consistency and correctness between the virtual circuit and the physical circuit can be comprehensively checked, potential problems can be discovered and solved in time, and the accuracy and security of the substation configuration file can be ensured.

[0116] Through the above steps, it is possible to achieve a comprehensive parsing, mapping, alignment, comparison, and verification of the configuration file of the target substation. First, the key information of the virtual circuit and the physical circuit is accurately extracted from the configuration file and converted into an intuitive graphical representation. Then, the node alignment is performed to ensure that the node relationship between the virtual circuit and the physical circuit is clear and definite. Next, a comparison relationship table is generated to comprehensively record the corresponding relationship and differences between the virtual circuit and the physical circuit. Finally, verification is carried out according to the comparison relationship table to promptly discover and solve potential problems. The implementation effect of this series of operations is that it can ensure the logical consistency and physical correctness of the substation configuration file, reduce the risk of operation accidents, and improve the safety and reliability of the substation. At the same time, through the automated parsing and verification process, the operation and maintenance efficiency is significantly improved, and the complexity and error rate of manual operations are reduced.

[0117] In one implementation manner of this embodiment, the virtual circuit and the physical circuit are verified according to the comparison relationship table, and the verification result is obtained and output, including the following steps:

[0118] S510. Based on the comparison relationship table, compare item by item the nodes and the connection relationships between the nodes of the virtual circuit and the physical circuit;

[0119] S520. In the case where the nodes or the connection relationships of the virtual circuit and the physical circuit are inconsistent, generate a list of inconsistent items and output the list of inconsistent items as the verification result;

[0120] S530. In the case where the nodes and the connection relationships of the virtual circuit and the physical circuit are both consistent, generate a verification passed signal and output the verification passed signal as the verification result.

[0121] Based on the comparison relation table, compare the nodes and the connection relationships between the virtual loop and the physical loop item by item. The nodes of the virtual loop and the physical loop respectively represent the key points in logical connections and physical connections, such as devices, interfaces, or signal transmission points. Each node has a clear identifier and attribute description in the comparison relation table, such as node name, type, location, etc. The connection relationship between nodes describes the logical or physical link between nodes, such as signal transmission paths, cable connections, etc. The process of item-by-item comparison starts from the nodes first, checking whether the attributes of each node in the virtual loop and the physical loop are consistent. For example, a logical node in the virtual loop may correspond to a physical device in the physical loop, and it is necessary to ensure that their names, types, and function descriptions are exactly the same. If it is found that the node attributes are inconsistent, such as different names or mismatched types, it is recorded as an inconsistent item. Next, compare the connection relationships between nodes. The logical connection relationships in the virtual loop need to correspond one by one to the physical connection relationships in the physical loop. For example, the signal transmission path between two logical nodes in the virtual loop must be consistent with the corresponding physical cable connection in the physical loop. If it is found that the connection relationships are inconsistent, such as there is a logical path in the virtual loop but the corresponding physical connection is missing in the physical loop, it is recorded as an inconsistent item. By comparing item by item, the consistency between the virtual loop and the physical loop can be comprehensively checked to ensure that the logical design is completely matched with the physical implementation. This detailed comparison process can effectively discover potential design errors or implementation deviations, providing a clear basis for subsequent corrections.

[0122] In the case where the nodes or connection relationships of the virtual loop and the physical loop are inconsistent, generate a list of inconsistent items and output the list of inconsistent items as the verification result. Among them, the list of inconsistent items is a detailed record containing all the discovered inconsistent nodes and connection relationships. Each inconsistent item includes the following information: type of inconsistency (node or connection relationship), description in the virtual loop, description in the physical loop, specific content of the inconsistency, and possible impacts. For example, if a logical node in the virtual loop cannot find the corresponding physical device in the physical loop, the list of inconsistent items will record the name, type, and function description of the node, and indicate that the corresponding physical device is missing in the physical loop. If a logical path in the virtual loop is missing the corresponding physical connection in the physical loop, the list of inconsistent items will record the start point, end point, type of transmitted signal of the path, and indicate the missing physical cable connection in the physical loop. The generation process of the list of inconsistent items is automated. Based on the comparison results in the comparison relation table, the inconsistent nodes and connection relationships can be automatically extracted and a detailed record can be generated. The output form of the list of inconsistent items can be a text file, a table, or a graphical interface, which is convenient for operation and maintenance personnel to view and analyze. By generating the list of inconsistent items, the differences between the virtual loop and the physical loop can be quickly located, providing clear guidance for subsequent corrections. This detailed recording and output method can significantly improve the efficiency of problem troubleshooting and reduce the workload of manual inspection.

[0123] When the nodes and connection relationships of the virtual circuit and the physical circuit are both consistent, a verification passed signal is generated and output as the verification result. The verification passed signal is a concise indication that the virtual circuit and the physical circuit are completely consistent, and the logical design matches the physical implementation without error. The generation process of the verification passed signal is based on the comparison results in the comparison relationship table. When all nodes and connection relationships are consistent, this signal is automatically generated. The output form of the verification passed signal can be a text prompt, a graphical flag, or a sound prompt, which is convenient for operation and maintenance personnel to quickly confirm the verification result. For example, in a graphical interface, the verification passed signal can be displayed as a green tick flag with a text prompt of "Verification Passed". In a command-line interface, the verification passed signal can be displayed as a text prompt of "SUCCESS". By generating the verification passed signal, the consistency between the virtual circuit and the physical circuit can be quickly confirmed, reducing the workload of manual inspection. This concise output method can significantly improve the operation and maintenance efficiency and ensure the accuracy and security of the substation configuration file.

[0124] In this embodiment, by comparing the nodes and connection relationships of the virtual circuit and the physical circuit item by item, an inconsistent item list or a verification passed signal is generated, which can comprehensively check the consistency between the logical design and the physical implementation. The inconsistent item list details all the discovered inconsistent nodes and connection relationships, providing a clear basis for subsequent correction. The verification passed signal quickly confirms the consistency between the virtual circuit and the physical circuit, reducing the workload of manual inspection. This detailed comparison and recording method can significantly improve the efficiency of problem troubleshooting and ensure the accuracy and security of the substation configuration file. By automatically generating the inconsistent item list and the verification passed signal, the workload of manual inspection can be reduced and the operation and maintenance efficiency can be improved. This technical solution provides an efficient and accurate solution for the verification of the substation configuration file, significantly reducing the risk of operation accidents.

[0125] In one implementation of this embodiment, a dynamic update step is adopted to optimize the configuration file parsing tree to optimize the feature association graph, and the optimized feature association graph is obtained, including the following steps:

[0126] S610. Real-time collect the update information of the configuration file in the production environment. The update information includes newly added configuration files, modified configuration files, and deleted configuration files;

[0127] S620. According to the update information, dynamically adjust the nodes and edges of the configuration file parsing tree to make the configuration file parsing tree consistent with the configuration file in the production environment;

[0128] Based on the dynamically adjusted configuration file parsing tree, re-extract the second key nodes and the second association relationships, and update the feature association graph based on the second key nodes and the second association relationships to obtain an optimized feature association graph.

[0129] Real-time collect the update information of the configuration files in the production environment. The update information includes newly added configuration files, modified configuration files, and deleted configuration files. The configuration files in the production environment will change with the update, maintenance, or upgrade of substation equipment. These changes may include adding device configuration files, modifying the parameters of existing configuration files, or deleting unused configuration files. To ensure the accuracy and real-time nature of the configuration file parsing tree, it is necessary to monitor the changes of the configuration files in the production environment in real time and collect this update information. The collection of the update information can be achieved through file system monitoring tools or database triggers. For example, use the inotify tool in the Linux system to monitor the changes of the configuration file directory, or capture the add, modify, and delete operations of the configuration files through database triggers. The collected update information includes the paths and contents of the newly added configuration files, the paths and modified contents of the modified configuration files, and the paths of the deleted configuration files. This update information provides a data basis for subsequent dynamic adjustments. By collecting the update information in real time, it can be ensured that the configuration file parsing tree can timely reflect the latest state in the production environment, thereby avoiding modeling errors or operation accidents caused by lagging configuration files.

[0130] According to the update information, dynamically adjust the nodes and edges of the configuration file parsing tree to make the configuration file parsing tree consistent with the configuration files in the production environment. The configuration file parsing tree is a tree-shaped data structure used to represent the hierarchical structure and node relationships of the configuration files. Each node represents an attribute or parameter in the configuration file, and the edge represents the association relationship between the attributes.

[0131] When the configuration files in the production environment change, it is necessary to dynamically adjust the nodes and edges of the parsing tree to maintain its consistency with the production environment. For newly added configuration files, first parse their structural features, extract the key nodes and association relationships, and then add these nodes and edges to the parsing tree. For modified configuration files, find the corresponding nodes in the parsing tree and update their attribute values or association relationships. For deleted configuration files, remove the corresponding nodes and edges from the parsing tree. During the dynamic adjustment process, it is necessary to ensure the integrity and consistency of the parsing tree, such as avoiding the appearance of isolated nodes or circular dependencies. Through dynamic adjustment, the configuration file parsing tree can timely reflect the latest configuration in the production environment, providing an accurate basis for subsequent optimization of the feature association graph.

[0132] Based on the dynamically adjusted configuration file parsing tree, re-extract the second key nodes and the second association relationships, and update the feature association graph based on the second key nodes and the second association relationships to obtain an optimized feature association graph. The feature association graph is a graph structure used to describe the dependency relationships and weights between key features in the configuration file. When the configuration file parsing tree is dynamically adjusted, it is necessary to re-extract the key nodes and association relationships to update the feature association graph. First, traverse the dynamically adjusted parsing tree to extract the second key nodes, which include key features such as equipment failure rate, load change, number of channels, polarity, and precision compensation coefficient. Then, analyze the association relationships between these nodes, such as the dependency relationship between the equipment failure rate and the load change, or the weight relationship between the number of channels and the polarity. Based on the extracted second key nodes and the second association relationships, update the nodes and edges in the feature association graph. For the newly added nodes and edges, directly add them to the feature association graph; for the modified nodes and edges, update their attribute values or weights; for the deleted nodes and edges, remove them from the feature association graph. By updating the feature association graph, it can be ensured that it is consistent with the configuration file in the production environment, thereby improving the accuracy and reliability of the machine learning model.

[0133] In this embodiment, by collecting the configuration file update information in the production environment in real time, dynamically adjusting the configuration file parsing tree, and updating the feature association graph, the real-time performance and accuracy of the modeling process can be ensured. This method can effectively cope with the dynamic changes of the configuration file in the production environment and avoid modeling errors or operation accidents caused by the lag of the configuration file. The dynamic adjustment and update process has a high degree of automation, reducing the need for manual intervention and improving the modeling efficiency. At the same time, the optimized feature association graph provides more accurate feature dependency relationships for the machine learning model, thereby improving the prediction performance and reliability of the model. Finally, this method provides an efficient and reliable solution for the automatic modeling of the configuration file of the intelligent substation, significantly reducing the risk of operation accidents and improving the safety and stability of the substation.

[0134] In one implementation of this embodiment, the structural features include a hierarchical structure, and the hierarchical structure includes the attributes of the configuration file and the association relationships between the attributes. Based on the structural features, construct a configuration file parsing tree, and extract the key nodes and the association relationships between the key nodes through the configuration file parsing tree, including the following steps:

[0135] S710: Use the attributes of the configuration file as nodes and the association relationships between the attributes as edges to construct a configuration file parsing tree;

[0136] S720: Traverse the configuration file parsing tree to extract the key nodes and the association relationships between the key nodes. The key nodes include equipment failure rate, load change, number of channels, polarity, and precision compensation coefficient.

[0137] In the process of automatic modeling of intelligent substation configuration files, the structural features of the configuration files usually exhibit a hierarchical structure, which is composed of the attributes of the configuration files and the association relationships between them. Attributes are the basic units in the configuration files, such as equipment failure rate, load change, number of channels, polarity, and precision compensation coefficient, etc. These attributes describe the specific characteristics and operating states of substation equipment. The association relationships describe the logical connections between these attributes, such as the causal relationship between the equipment failure rate and the load change, or the configuration dependency relationship between the number of channels and the polarity. In order to transform these structural features into a computable form, a configuration file parsing tree needs to be constructed. The configuration file parsing tree is a tree-like data structure, where each node represents an attribute, and the edges represent the association relationships between the attributes. The process of constructing the parsing tree first requires extracting all the attributes from the configuration file and adding these attributes as nodes to the tree. Then, according to the association relationships between the attributes, edges are added between the nodes to form a complete tree structure. For example, assume that the configuration file contains three attributes: equipment failure rate, load change, and number of channels, and there is an association relationship between the equipment failure rate and the load change, and there is also an association relationship between the load change and the number of channels. Then the parsing tree will contain three nodes, representing these three attributes respectively, and an edge will be added between the equipment failure rate and the load change, and between the load change and the number of channels respectively. In this way, the structural features of the configuration file are transformed into a computable form, providing a basis for subsequent feature extraction and modeling.

[0138] After the configuration file parsing tree is constructed, it is necessary to traverse the tree to extract the key nodes and the association relationships between them. Key nodes refer to those attributes that have an important impact on the substation configuration, such as equipment failure rate, load change, number of channels, polarity, and precision compensation coefficient, etc. These attributes are usually directly related to the operating state and performance of the substation, so they need to be particularly concerned about in the modeling process. The process of traversing the parsing tree can adopt algorithms such as depth-first search (DFS) or breadth-first search (BFS). Starting from the root node, each node is visited in turn, and its attributes and the association relationships with other nodes are recorded. For example, assume that the root node of the parsing tree is the equipment failure rate, its child node is the load change, and the child node of the load change is the number of channels. Then during the traversal process, first visit the equipment failure rate node, record its attribute value and the association relationship with the load change node; then visit the load change node, record its attribute value and the association relationship with the number of channels node; finally visit the number of channels node, record its attribute value. In this way, all the key nodes and the association relationships between them can be extracted. These key nodes and association relationships will be used for subsequent construction of the feature association graph and training of the machine learning model, so as to ensure that the generated substation configuration file can accurately reflect the actual operating state and configuration requirements of the substation.

[0139] In this embodiment, by taking the attributes of the configuration file as nodes and the association relationships between the attributes as edges to construct a configuration file parsing tree, and traversing the tree to extract the key nodes and their association relationships, the complex configuration file structure features can be transformed into a computable form. This method can not only clearly describe the hierarchical structure of the configuration file and the logical relationships between the attributes, but also provide a reliable basis for subsequent feature extraction and modeling. By extracting the key nodes and their association relationships, it can be ensured that the generated substation configuration file can accurately reflect the actual operating status and configuration requirements of the substation, thereby improving the accuracy and reliability of the configuration file modeling. In addition, this method also has strong scalability and can adapt to configuration files of different formats and complexities, providing strong technical support for the automated modeling and configuration management of intelligent substations.

[0140] In one implementation of this embodiment, hyperparameter tuning and model integration are performed on the trained machine learning model, and the structure of the substation configuration file is verified through the configuration file parsing tree to obtain an optimized machine learning model, including the following steps:

[0141] S810. When the number of machine learning models is at least two, for each trained machine learning model, calculate the model evaluation result, and adjust the hyperparameters of the machine learning model according to the model evaluation result. The hyperparameters include the learning rate, regularization coefficient, and tree depth.

[0142] S820. Obtain the prediction results of each machine learning model after hyperparameter tuning, and perform weighted averaging on the prediction results of all machine learning models to obtain an integrated model. Use the weighted average prediction result as the model output result of the integrated model. The model output result is the substation configuration file.

[0143] S830. Import the model output result into the configuration file parsing tree so that the configuration file parsing tree verifies the structure of the substation configuration file.

[0144] S840. When the verification passes, use the integrated model as the optimized machine learning model.

[0145] S850. When the verification fails, readjust the hyperparameters until the verification passes to obtain the optimized machine learning model.

[0146] In the process of automatic modeling of intelligent substation configuration files, the performance of machine learning models directly affects the accuracy and reliability of the generated configuration files. To improve the performance of the models, it is usually necessary to train multiple machine learning models and perform hyperparameter tuning on these models. Hyperparameters are parameters that need to be manually set during the model training process, such as learning rate, regularization coefficient, and tree depth. These parameters have an important impact on the training effect and prediction ability of the model. First, for each trained machine learning model, its model evaluation results need to be calculated. Model evaluation results are usually measured by performance metrics on cross-validation or test sets, such as accuracy, recall, F1-score, or mean squared error, etc.

[0147] For example, suppose there are three trained models, which are trained with different learning rates respectively. By calculating the accuracy of each model through cross-validation, the accuracy of model A is 90%, the accuracy of model B is 92%, and the accuracy of model C is 88%. According to these evaluation results, the hyperparameters of the model can be adjusted to optimize its performance. For example, if the accuracy of model B is the highest, you can try to further adjust its learning rate to see if the performance can be further improved. The adjustment of the regularization coefficient and tree depth is similar. By comparing the model performance under different parameter settings, the optimal combination of hyperparameters is selected. In this way, the prediction ability of machine learning models can be significantly improved, providing a reliable basis for generating high-quality substation configuration files.

[0148] After completing the hyperparameter tuning, it is necessary to obtain the prediction results of each machine learning model and integrate these results to improve the accuracy and stability of the prediction. An ensemble model is a method of combining the prediction results of multiple models. Common ensemble methods include weighted average, voting method, and stacking method, etc. First, obtain the prediction results of each machine learning model after hyperparameter tuning. For example, suppose there are three models, which respectively predict the equipment failure rate of a certain substation configuration file. The predicted value of model A is 5%, the predicted value of model B is 4.8%, and the predicted value of model C is 5.2%. Next, perform a weighted average on these prediction results. The weights of the weighted average can be determined according to the performance metrics of the models. For example, the higher the accuracy of the model, the greater its weight. Suppose the accuracy of model A is 90%, the accuracy of model B is 92%, and the accuracy of model C is 88%. Then weights 0.3, 0.4, and 0.3 can be assigned to them respectively. Through weighted average calculation, the prediction result of the ensemble model is 5%×0.3 + 4.8%×0.4 + 5.2%×0.3 = 4.98%. This weighted average prediction result will be used as the final output of the ensemble model for generating substation configuration files. By integrating the prediction results of multiple models, the accuracy and stability of the prediction can be significantly improved, reducing the bias and error of a single model.

[0149] After obtaining the prediction results of the integrated model, these results need to be imported into the configuration file parsing tree for structural verification. The configuration file parsing tree is a tree-like data structure used to represent the hierarchical structure and node relationships of the configuration file. By importing the model output results into the parsing tree, it can be checked whether the generated substation configuration file conforms to the expected structure and logical relationships. For example, assume that the model output results include key parameters such as equipment failure rate, load change, number of channels, polarity, and accuracy compensation coefficient. These parameters will be added to the corresponding nodes of the parsing tree. Then, traverse the parsing tree to check whether the attribute values of each node and its association with other nodes meet the expectations. For example, check whether the association between the equipment failure rate and the load change is reasonable, or whether the configuration between the number of channels and the polarity is consistent. In this way, the structure and logical relationships of the generated substation configuration file can be comprehensively checked to ensure its accuracy and reliability. If it is found that the attribute value or association relationship of a certain node does not meet the expectations, it is recorded as a non-passed item in the verification and the corresponding error message is generated. Through structural verification, errors in the configuration file can be timely discovered and corrected to ensure its reliability and safety in practical applications.

[0150] In the case where the structural verification passes, it indicates that the generated substation configuration file conforms to the expected structure and logical relationships, and the integrated model can be used as the optimized machine learning model. The optimized model not only has high prediction accuracy but also can generate configuration files that meet the actual requirements. For example, assume that through structural verification, it is found that the key parameters such as equipment failure rate, load change, number of channels, polarity, and accuracy compensation coefficient in the generated configuration file all meet the expectations, and the association relationships between the nodes are reasonable. Then, the integrated model can be used as the final optimized model. The optimized model can be deployed to the production environment to be used for real-time generation of substation configuration files. In this way, the accuracy and reliability of the generated configuration file in practical applications can be ensured, and the operation efficiency and safety of the substation can be improved. The optimized model can also be used for subsequent configuration file updates and maintenance, providing strong support for the long-term stable operation of the substation.

[0151] In the case where the structure verification fails, it indicates that there are errors or inconsistencies in the generated substation configuration file, and it is necessary to readjust the hyperparameters of the machine learning model until the verification passes. The process of readjusting the hyperparameters is similar to S810. First, calculate the evaluation results of the model, and then adjust the hyperparameters such as the learning rate, regularization coefficient, and tree depth according to the evaluation results. For example, assuming that through the structure verification, it is found that the correlation between the equipment failure rate and the load change in the generated configuration file is unreasonable, the learning rate of the model can be tried to be adjusted to see if it can improve the prediction results. Then, retrain the model and obtain new prediction results. Then, import the new prediction results into the configuration file parsing tree for structure verification. If the verification still fails, continue to adjust the hyperparameters until the generated configuration file passes the structure verification. In this way, the performance of the machine learning model can be gradually optimized to ensure that the generated configuration file conforms to the expected structure and logical relationship. Finally, an optimized machine learning model is obtained, which can generate accurate and reliable substation configuration files, providing strong support for the safe and stable operation of the substation.

[0152] In actual implementation, the accuracy of the optimized machine learning model trained by the embodiments of the present application can reach 100%.

[0153] This embodiment can significantly improve the prediction accuracy of the model and the reliability of the configuration file by performing hyperparameter tuning and model integration on the trained machine learning model and performing structure verification on the generated substation configuration file through the configuration file parsing tree. Hyperparameter tuning optimizes the performance of the model, model integration improves the stability and accuracy of the prediction, and structure verification ensures that the generated configuration file conforms to the expected structure and logical relationship. This method can not only generate high-quality substation configuration files, but also timely discover and correct errors in the configuration files, ensuring their reliability and safety in actual applications. Finally, the optimized machine learning model provides strong technical support for the automated modeling and configuration management of intelligent substations, significantly reducing the risk of operation accidents and improving the operation efficiency and safety of substations.

[0154] The embodiments of the present application also provide an electronic device, including:

[0155] A memory configured to store instructions; and

[0156] A processor configured to call instructions from the memory and capable of implementing the above-mentioned method for automatically modeling intelligent substation configuration files when executing the instructions.

[0157] The embodiments of the present application also provide an intelligent substation configuration file automatic modeling system, including:

[0158] The above-mentioned electronic device.

[0159] The electronic device can be a tablet computer, desktop, laptop, handheld computer, wearable device, notebook computer, ultra-mobile personal computer (UMPC), netbook, or other devices with a processor. Of course, the electronic device can also be a server. The embodiments of the present application do not impose special restrictions on the specific form of the electronic device.

[0160] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, system, or computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0161] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing device to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing device produce a means for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0162] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufactured article including an instruction means that implements the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0163] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0164] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0165] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0166] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0167] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0168] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. An automatic modeling method for the configuration file of an intelligent substation, characterized in that, Including: In response to receiving configuration files of multiple manufacturers and multiple formats, extracting the structural features of the configuration files, and performing data cleaning on the data records in the configuration files to obtain the data records after data cleaning; Based on the structural features, constructing a configuration file parsing tree, and extracting key nodes and the association relationships between the key nodes through the configuration file parsing tree; Performing data standardization on the data records after data cleaning to obtain the data records after standardization; Based on the AUTO ML framework, extracting the key features of the data records after standardization, where the key features include equipment failure rate, load change, number of channels, polarity, and precision compensation coefficient; According to the key features, the key nodes, and the association relationships, constructing a feature association graph, which is used to describe the dependency relationships and weights between the key features; Using the AUTO ML framework to identify target features strongly related to the substation configuration in the feature association graph, and obtaining the corresponding difference data set based on the target features in a preset difference database; Training a machine learning model based on the difference data set and the feature association graph until the number of training times reaches a preset number threshold, where the machine learning model is used to generate substation configuration files; Performing hyperparameter tuning and model integration on the trained machine learning model, and performing structural verification on the substation configuration file through the configuration file parsing tree to obtain an optimized machine learning model, where performing structural verification on the substation configuration file is used to ensure the logical consistency of the substation configuration file; Deploying the optimized machine learning model to a production environment to generate a target substation configuration file corresponding to the production environment.

2. The method according to claim 1, characterized in that, Before the step of deploying the optimized machine learning model to a production environment to generate a target substation configuration file corresponding to the production environment, it further includes: Real-time collecting the operation results of the optimized machine learning model, and comparing the operation results with preset expected results to obtain a comparison result, where the operation results are used to represent the target substation configuration file; Calculating the error rate of the optimized machine learning model according to the comparison result; The step of deploying the optimized machine learning model to a production environment to generate a target substation configuration file corresponding to the production environment includes: Triggering a model optimization process when the error rate is greater than a preset threshold to optimize the machine learning model to obtain a target machine learning model, and deploying the target machine learning model to a production environment to generate a target substation configuration file corresponding to the production environment; When the error rate is not greater than the preset threshold, deploying the optimized machine learning model to a production environment to generate a target substation configuration file corresponding to the production environment.

3. The method according to claim 2, characterized in that The model optimization process includes: Adopting dynamic update steps to optimize the configuration file parsing tree to optimize the feature association graph to obtain an optimized feature association graph; Identify the second target features strongly related to the substation configuration in the optimized feature association graph using the AUTO ML framework, and obtain the corresponding second difference dataset in the preset difference database based on the second target features; Use the second difference dataset and the optimized feature association graph as the input of the optimized machine learning model to obtain the optimized target substation configuration file.

4. The method according to claim 2, wherein After generating the target substation configuration file corresponding to the production environment, it includes: Parse the target substation configuration file to obtain virtual loop information and physical loop information, where the virtual loop information includes logical connection relationships and the physical loop information includes physical connection relationships; Map the virtual loop information to a virtual loop diagram and map the physical loop information to a physical loop diagram; Align the nodes of the virtual loop diagram and the physical loop diagram so that the nodes of the virtual loop diagram correspond one by one to the nodes of the physical loop diagram; After node alignment, generate a comparison relationship table between the virtual loop and the physical loop; Verify the virtual loop and the physical loop according to the comparison relationship table, obtain the verification result and output it.

5. The method according to claim 4, characterized in that The verifying the virtual loop and the physical loop according to the comparison relationship table, obtaining the verification result and outputting it includes: Based on the comparison relationship table, compare item by item the nodes of the virtual loop and the physical loop and the connection relationships between the nodes; In the case where the nodes or the connection relationships of the virtual loop and the physical loop are inconsistent, generate a list of inconsistent items and output the list of inconsistent items as the verification result; In the case where the nodes and the connection relationships of the virtual loop and the physical loop are both consistent, generate a verification passed signal and output the verification passed signal as the verification result.

6. The method according to claim 3, wherein The optimizing the feature association graph by optimizing the configuration file parsing tree using the dynamic update step to obtain the optimized feature association graph includes: Real-time collect the update information of the configuration file in the production environment, where the update information includes newly added configuration files, modified configuration files, and deleted configuration files; According to the update information, dynamically adjust the nodes and edges of the configuration file parsing tree so that the configuration file parsing tree is consistent with the configuration file in the production environment; Based on the dynamically adjusted configuration file parsing tree, re-extract the second key nodes and the second association relationships, and update the feature association graph based on the second key nodes and the second association relationships to obtain the optimized feature association graph.

7. The method according to claim 1, wherein The structural features include a hierarchical structure, and the hierarchical structure includes the attributes of the configuration file and the association relationships between the attributes. The constructing a configuration file parsing tree based on the structural features and extracting the key nodes and the association relationships between the key nodes through the configuration file parsing tree includes: Use the attributes of the configuration file as nodes and the association relationships between the attributes as edges to construct a configuration file parsing tree; Traverse the configuration file parsing tree to extract the key nodes and the association relationships between the key nodes. The key nodes include equipment failure rate, load change, number of channels, polarity, and precision compensation coefficient.

8. The method according to claim 1, wherein Performing hyperparameter tuning and model integration on the trained machine learning model, and performing structural verification on the substation configuration file through the configuration file parsing tree to obtain an optimized machine learning model, including: When the number of machine learning models is at least two, for each trained machine learning model, calculating a model evaluation result, and adjusting the hyperparameters of the machine learning model according to the model evaluation result, where the hyperparameters include a learning rate, a regularization coefficient, and a tree depth; Obtaining the prediction results of each machine learning model after hyperparameter tuning, and performing weighted averaging on the prediction results of all the machine learning models to obtain an integrated model, and using the weighted average prediction result as the model output result of the integrated model, where the model output result is the substation configuration file; Importing the model output result into the configuration file parsing tree, so that the configuration file parsing tree performs structural verification on the substation configuration file; When the verification passes, using the integrated model as the optimized machine learning model; When the verification fails, readjusting the hyperparameters until the verification passes to obtain an optimized machine learning model.

9. An electronic device, characterized in that, Including: A memory configured to store instructions; And A processor configured to call the instructions from the memory and capable of implementing the intelligent substation configuration file automatic modeling method according to any one of claims 1 to 8 when executing the instructions.

10. An intelligent substation configuration file automatic modeling system, characterized in that, Including: The electronic device according to claim 9.

Images