Substation SCD model consistency matching method and system based on scanning algorithm

Abstract

The invention relates to the technical field of power system automation, in particular to a transformer substation SCD model consistency matching method and system based on a scanning algorithm, and the method comprises the steps: analyzing an SCD file, and generating an LN feature vector based on a logic node LN, a data object DO and a data attribute DA; performing clustering analysis on the LN feature vector based on an SCAN algorithm to generate an LN cluster; generating an SCD file matching model based on the LN clusters; performing consistency matching on the first substation SCD file matching model and the second substation SCD file matching model, including traversing all LN clusters of the two substation SCD file matching models, and judging whether each corresponding LN cluster is matched or not; and summarizing matching results of all the LN clusters of the two substation SCD file matching models, and generating a substation SCD model consistency verification report. According to the SCD model consistency verification method, efficient and accurate SCD model consistency verification is realized by introducing the SCAN algorithm.

Description

Technical Field

[0001] The present invention relates to the technical field of power system automation, and in particular to a method and system for matching the consistency of a substation SCD model based on a scanning algorithm. Background Art

[0002] In smart substation systems, the Substation Configuration Description (SCD) file, as the core configuration file, carries key information such as the substation's primary system structure, secondary equipment configuration, and communication parameters. It is an important foundation for ensuring the stable operation and efficient operation and maintenance of smart substations. SCD files, constructed according to the IEC 61850 standard, enable information exchange and functional collaboration between various intelligent electronic devices (IEDs) within the substation. The accuracy of their configuration is directly related to the reliability of core functions such as protection control and status monitoring. However, in the construction and operation and maintenance of smart substations, the consistency verification of SCD files faces significant challenges, including:

[0003] Model complexity and heterogeneity: SCD files must comprehensively describe multiple configuration information, including substation topology, equipment logical node associations, and communication service mapping. Their complex hierarchical models and massive data volumes, coupled with differences among vendors in their understanding of specifications such as the Abstract Communication Service Interface (ACSI) and data object modeling in the IEC61850 standard, easily lead to model heterogeneity when integrating SCD files across vendors, increasing the difficulty of ensuring configuration consistency.

[0004] Dynamic Updates and Version Management: As power grid operations adjust and technology upgrades occur, SCD files require frequent updates to accommodate new equipment or expanded functionality. Traditional manual maintenance methods struggle to accurately track configuration changes throughout their lifecycle, leading to version management confusion and the risk of configuration drift, which can impact the stable operation of substations.

[0005] Limitations of traditional verification methods: Existing SCD file consistency verification methods have obvious shortcomings. The manual comparison method relies on expert experience to check each item one by one, which is inefficient and easily affected by subjective factors, and it is difficult to meet the needs of refined verification of large-scale SCD files. The rule matching method requires the pre-construction of a rule base covering all constraint rules, which has high maintenance costs. With the update and expansion of the IEC 61850 standard, the rule base needs to evolve synchronously, which has the risk of lag. The model comparison method adopts a full parsing and traversal strategy. The computational complexity increases exponentially with the growth of file size, making it difficult to meet real-time requirements. It mainly focuses on structural matching at the grammatical level, lacks in-depth exploration of the semantic relevance of the SCD model, and is difficult to discover potential logical contradictions or configuration conflicts.

[0006] These issues lead to SCD file configuration errors, which have become a major cause of frequent failures in smart substations and severely hampered the progress of the intelligent transformation of power grids. Therefore, developing efficient and intelligent SCD file consistency verification technology is of great significance for improving the operation and maintenance efficiency of smart substations and ensuring the reliable operation of power systems. Summary of the Invention

[0007] The purpose of the present invention is to provide a substation SCD model consistency matching method and system based on a scanning algorithm in order to solve at least one of the above technical problems. By introducing the SCAN algorithm to perform cluster analysis on the SCD model and combining it with semantic similarity calculation, efficient and accurate substation SCD model consistency verification can be achieved.

[0008] The present invention achieves the above-mentioned purpose through the following technical solutions:

[0009] A substation SCD model consistency matching method based on a scanning algorithm, the method comprising the following steps:

[0010] Parse the SCD file and generate the LN feature vector based on the logical node LN, data object DO and data attribute DA;

[0011] Performing cluster analysis on the LN feature vector based on the SCAN algorithm to generate an LN cluster; generating an SCD file matching model based on the LN cluster;

[0012] Performing consistency matching on the first substation SCD file matching model and the second substation SCD file matching model, including: traversing all LN clusters of the two substation SCD file matching models, and determining whether each corresponding LN cluster matches;

[0013] The matching results of all LN clusters of the two substation SCD file matching models are summarized to generate a substation SCD model consistency verification report.

[0014] Further, generating an LN feature vector based on the logical node LN, the data object DO and the data attribute DA includes: generating the LN feature vector based on the data object DO and the data attribute DA of each logical node LN in combination with a semantic weight allocation rule defined according to the IEC 61850 standard;

[0015] The semantic weight distribution rules include: the DO weight of the data object corresponding to the key function is 1.0, and the DO weight of the data object corresponding to the auxiliary function is 0.3; the DA weight of the data attribute corresponding to the mandatory attribute is 0.5, and the DA weight of the data attribute corresponding to the optional attribute is 0.2.

[0016] Furthermore, cluster analysis is performed on the LN feature vector based on the SCAN algorithm to generate LN clusters, including:

[0017] Based on the preset neighborhood threshold and minimum number of points, LN feature vectors with similar functions are divided into the same LN cluster;

[0018] The identification condition of the core point is: the number of similar LN feature vectors in its neighborhood is not less than the minimum number of points;

[0019] The LN feature vectors that cannot be classified into any LN cluster are marked as noise nodes.

[0020] Furthermore, the neighborhood threshold is 0.7-0.8, and the minimum number of points is 3.

[0021] Furthermore, determining whether each corresponding LN cluster matches includes:

[0022] Determining whether the LN clusters match based on the structure of the LN clusters and the composition of the LN feature vectors within the LN clusters;

[0023] The similarity of the corresponding LN feature vectors in the two SCD file matching models is calculated based on the cosine similarity, and whether the LN feature vectors match is determined based on the similarity threshold.

[0024] Furthermore, determining whether the LN clusters match includes determining whether the LN clusters match based on a difference in the number of LN feature vectors within the LN clusters and an average similarity of the LN feature vectors within the LN clusters.

[0025] Furthermore, the LN cluster matching criteria include: the number difference of LN feature vectors within the LN cluster does not exceed 10%; and the average similarity of LN feature vectors within the LN cluster is not less than 0.75.

[0026] Furthermore, the calculation formula of the cosine similarity is:

[0027] sim(LN1,LN2)=(V(LN1)·V(LN2)) / (||V(LN1)||*||V(LN2)||)

[0028] Where sim(LN1,LN2) represents the cosine similarity between the LN1 feature vector and the LN2 feature vector; V(LN1) represents the LN1 feature vector; V(LN2) represents the LN2 feature vector; “·” represents the dot product of the vector; “||||” represents the modulus of the vector;

[0029] The similarity threshold is 0.75-0.85.

[0030] Furthermore, the method further comprises:

[0031] Display the substation SCD model consistency verification report through a visual interface;

[0032] Dynamically update the rule base, store matching rules and exception information, and use a version control mechanism to maintain the rule base;

[0033] Provides a standard API interface that can be integrated with other power system automation systems;

[0034] For large-scale SCD files, distributed computing technology is used to assign the consistency verification task of the SCD file matching model to multiple computing nodes for parallel execution;

[0035] Supports parsing of the SCL file format that complies with the IEC 61850 standard and can be extended to other SCD file formats.

[0036] The substation SCD model consistency matching system based on the scanning algorithm includes:

[0037] Parsing module, used to parse the SCD file and generate LN feature vector based on logical node LN, data object DO and data attribute DA;

[0038] A clustering module, configured to perform cluster analysis on the LN feature vectors based on a SCAN algorithm to generate LN clusters; and generate an SCD file matching model based on the LN clusters;

[0039] A matching module is used to perform consistency matching on the SCD file matching model of the first substation and the SCD file matching model of the second substation, including: traversing all LN clusters of the SCD file matching models of the two substations, and determining whether each corresponding LN cluster matches;

[0040] The output module is used to summarize the matching results of all LN clusters of the two substation SCD file matching models and generate a substation SCD model consistency verification report.

[0041] The beneficial effects of the present invention are:

[0042] 1. Efficiency improvement: The present invention uses the SCAN algorithm to perform cluster analysis on LN feature vectors, compressing the high-dimensional feature space into a set of semantically related LN clusters, thereby narrowing the matching range and improving verification efficiency.

[0043] 2. Accuracy breakthrough: This invention constructs semantic weight distribution rules based on the IEC 61850 standard and combines it with the cosine similarity algorithm to improve the LN feature vector matching accuracy and key function node matching accuracy.

[0044] 3. Dynamic maintenance: This invention uses JSON format to store matching rules and exception pattern libraries, combined with the Git version control mechanism, to support hot updates and parallel development of the rule library, thereby improving the efficiency of rule iteration.

[0045] 4. Ecological expansion: This invention provides a RESTful API standard interface to achieve seamless integration with third-party systems such as control systems and operation and maintenance platforms, and build a complete SCD model verification ecosystem.

[0046] 5. Large-scale processing: For large-scale SCD files, Apache Spark is used to implement fine-grained task segmentation and parallel computing, shortening verification time.

[0047] 6. Elastic expansion: This invention supports dynamic allocation of computing resources and combines it with in-memory columnar storage technology to improve the speed of feature vector calculation.

[0048] 7. High reliability: The present invention integrates Spark automatic retry mechanism and Saga transaction compensation mode, automatically migrates tasks when a node fails, and ensures the integrity of the verification process.

[0049] 8. Interactive Analysis: This paper generates interactive verification reports based on D3.js, supports mismatched cluster topology highlighting and anomaly tracing analysis, and improves the efficiency of manual review.

[0050] The present invention solves the pain points of low verification efficiency, poor accuracy and weak scalability of traditional SCD models, and provides an intelligent solution for substation automation system configuration management. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 This is a flow chart of a method for matching the consistency of a substation SCD model based on a scanning algorithm according to an embodiment of the present invention;

[0052] Figure 2 This is a flow chart of a method for matching the consistency of a substation SCD model based on a scanning algorithm according to another embodiment of the present invention;

[0053] Figure 3 This is a schematic diagram of SCD file parsing according to one embodiment of the present invention;

[0054] Figure 4 This is a schematic diagram of code for calculating LN feature vector similarity based on cosine similarity according to an embodiment of the present invention;

[0055] Figure 5 A schematic diagram of code for calling a verification service through an external system according to an embodiment of the present invention;

[0056] Figure 6 Schematic diagram of a substation SCD model consistency matching system based on a scanning algorithm according to an embodiment of the present invention;

[0057] Figure 7 This is the substation SCD model file information of manufacturer A;

[0058] Figure 8 This is the substation SCD model file information of manufacturer B. DETAILED DESCRIPTION

[0059] The present invention will now be discussed with reference to exemplary embodiments. It should be understood that the embodiments discussed are only intended to enable those skilled in the art to better understand and implement the present invention, rather than to imply any limitation on the scope of the present invention.

[0060] As used herein, the term "including" and variations thereof are to be interpreted as open-ended terms meaning "including, but not limited to." The term "based on" is to be interpreted as "based, at least in part, on." The terms "one embodiment" and "an embodiment" are to be interpreted as "at least one embodiment."

[0061] Example 1

[0062] Figure 1 This is a flow chart of a method for matching the consistency of a substation SCD model based on a scanning algorithm according to an embodiment of the present invention. Figure 1 As shown, according to one embodiment of the present invention, a substation SCD model consistency matching method based on a scanning algorithm is used for substation SCD model consistency verification, comprising the following steps:

[0063] Step S102: Parse the SCD file and generate an LN feature vector based on the logical node LN, the data object DO, and the data attribute DA;

[0064] Preferably, step S102 comprises: generating an LN feature vector according to the data object DO and data attribute DA of each logical node LN in combination with a semantic weight allocation rule defined according to the IEC 61850 standard;

[0065] The semantic weight allocation rules include: the DO weight of the data object corresponding to the key function is 1.0, and the DO weight of the data object corresponding to the auxiliary function is 0.3; the DA weight of the data attribute corresponding to the mandatory attribute is 0.5, and the DA weight of the data attribute corresponding to the optional attribute is 0.2.

[0066] Step S104: performing cluster analysis on the LN feature vectors based on the SCAN algorithm to generate LN clusters; generating an SCD file matching model based on the LN clusters;

[0067] Step S106 , performing consistency matching on the first substation SCD file matching model and the second substation SCD file matching model, including: traversing all LN clusters of the two substation SCD file matching models, and determining whether each corresponding LN cluster matches;

[0068] Step S108 : Summarize the matching results of all LN clusters of the two substation SCD file matching models and generate a substation SCD model consistency verification report.

[0069] In this implementation, a method for matching substation SCD models based on the SCAN algorithm is proposed. Its core principle is to verify the consistency of different substation SCD models through feature extraction, cluster analysis, and similarity matching. First, the SCD file to be verified for consistency is obtained and an SCD file matching model is constructed. Specifically, the SCD file is parsed, logical nodes (LNs) and their associated data objects (DOs) and data attributes (DAs) are extracted, and a multidimensional LN feature vector is generated based on the semantic weight assignment rules defined in the IEC 61850 standard. For example, the weight of the DO data object corresponding to a critical function (such as protection action instructions) is set to 1.0, and the weight of the DO data object corresponding to an auxiliary function (such as condition monitoring) is set to 0.3. The weight of the DA data attribute corresponding to a mandatory attribute (such as svID) is set to 0.5, and the weight of the DA data attribute corresponding to an optional attribute (such as desc) is set to 0.2. The unstructured SCD file is converted into a structured feature space, laying the foundation for subsequent cluster analysis. The SCAN algorithm (density-based clustering algorithm) was used to perform cluster analysis on LN feature vectors, compressing the high-dimensional feature space into a set of semantically related LN clusters to form an SCD file matching model. Next, a dual-model comparison was performed between the two substation SCD file matching models, including both LN cluster-level matching and LN feature vector-level matching. Finally, the matching results were summarized to generate a substation SCD model consistency verification report containing the matching results for all LN clusters in the two substation SCD file matching models, enabling engineers to quickly identify configuration discrepancies.

[0070] The present invention realizes the automation and intelligence of SCD model verification through feature extraction, cluster analysis and similarity matching. It automatically clusters and analyzes LN feature vectors through the SCAN algorithm, and generates a structured feature space in combination with semantic weight rules to achieve accurate positioning of configuration differences and significantly improve detection efficiency. By unifying the semantic association rules of LN clusters, it can effectively integrate SCD files from different equipment manufacturers and solve the compatibility issues of multi-source heterogeneous models. By automatically generating comparison reports, it shortens the project acceptance cycle and significantly improves the intelligent level of SCD model management.

[0071] According to one embodiment of the present invention, step S104 includes:

[0072] Based on the preset neighborhood threshold and minimum number of points, LN feature vectors with similar functions are divided into the same LN cluster;

[0073] The identification condition of the core point is: the number of similar LN feature vectors in its neighborhood is not less than the minimum number of points;

[0074] LN feature vectors that cannot be classified into any LN cluster are marked as noise nodes.

[0075] Preferably, the neighborhood threshold is 0.7-0.8 and the minimum number of points is 3.

[0076] In this embodiment, the contents of step S104 are further defined. The SCAN algorithm is used to perform cluster analysis on LN feature vectors to construct functionally similar LN clusters. The SCAN algorithm defines the LN cluster structure using a dual threshold. The neighborhood threshold ε (preferably 0.7-0.8) is used to measure the similarity between LN feature vectors (e.g., cosine similarity). A neighborhood relationship is considered only when the similarity between two vectors is ≥ ε. A minimum number of points, MinPts (preferably 3), is set: the neighborhood of a core point must contain at least MinPts similar LN feature vectors to ensure the density stability of the LN cluster. If an LN feature vector's ε neighborhood contains ≥ MinPts vectors, it is marked as a core point. Core points expand LN clusters using density reachability: if vector A is a core point and vector B is within A's ε neighborhood, then B belongs to the LN cluster containing A. This process is iterated until no new members can be added. LN feature vectors that do not meet the core point criteria and are not included in any LN cluster are marked as noise nodes (e.g., customized LNs or abnormal configurations). The proportion of noise nodes is usually low and they need to be manually verified before being added to the whitelist.

[0077] The neighborhood threshold ε balances the compactness and inclusiveness of LN clusters. When the SCD file is large (total number of LNs > 2000), ε can be reduced (e.g., to 0.65) to avoid over-merging; for small files (total number of LNs < 500), ε can be appropriately increased (e.g., to 0.85). The minimum number of points, MinPts, ensures the significance of LN clusters. Experiments show that a too low MinPts easily generates small cluster noise, while a too high MinPts misses sparse clusters. In this embodiment, preferably, MinPts = 3.

[0078] The present invention realizes efficient clustering analysis of LN feature vectors through the SCAN algorithm. By adopting the dual constraints of the neighborhood threshold ε and the minimum number of points MinPts, LN feature vectors with similar functions are divided into highly cohesive clusters, effectively preserving the semantic association of SCD files. By automatically marking LN feature vectors that cannot be classified (such as customized configurations) and combining them with a manual verification mechanism, the efficiency of abnormal configuration detection is significantly improved.

[0079] According to one embodiment of the present invention, step S106 includes:

[0080] Based on the structure of the LN cluster and the composition of the LN feature vectors within the LN cluster, determine whether the LN cluster matches;

[0081] The similarity of the corresponding LN feature vectors in the two SCD file matching models is calculated based on the cosine similarity, and whether the LN feature vectors match is determined based on the similarity threshold.

[0082] Preferably, determining whether the LN clusters match includes determining whether the LN clusters match based on the difference in the number of LN feature vectors within the LN clusters and the average similarity of the LN feature vectors within the LN clusters.

[0083] Preferably, the LN cluster matching criteria include: the number difference of LN feature vectors within the LN cluster does not exceed 10%; and the average similarity of LN feature vectors within the LN cluster is not less than 0.75.

[0084] Preferably, the calculation formula of cosine similarity is:

[0085] sim(LN1,LN2)=(V(LN1)·V(LN2)) / (||V(LN1)||*||V(LN2)||)

[0086] Where sim(LN1,LN2) represents the cosine similarity between the LN1 feature vector and the LN2 feature vector; V(LN1) represents the LN1 feature vector; V(LN2) represents the LN2 feature vector; “·” represents the dot product of the vector; “||||” represents the modulus of the vector;

[0087] Preferably, the similarity threshold is 0.75-0.85.

[0088] In this embodiment, the content of step S106 is further limited. In step S106, a two-level matching strategy based on the similarity of LN cluster structure and LN feature vector is used to quantify the topological consistency of LN clusters and the semantic similarity of LN feature vectors to achieve high-precision comparison of SCD models. It includes LN cluster-level matching and LN feature vector-level matching. LN cluster-level matching uses dual constraints of quantity difference threshold (≤10%) and average similarity threshold (≥0.75). Quantity difference measures the scale difference of LN feature vectors within two LN clusters to prevent mismatching due to the large difference in the number of nodes. For example, if cluster A contains 100 LN vectors and cluster B contains 95, the difference is 5%, which meets the conditions. Average similarity evaluates the overall semantic consistency of the LN cluster by calculating the mean cosine similarity of all LN feature vectors in the LN cluster. Cosine similarity is used to quantify the semantic similarity between vectors, with a similarity threshold set between 0.75 and 0.85. The threshold is dynamically adjusted based on the actual application scenario to balance matching strictness and fault tolerance. The threshold for critical protection LNs (such as PTOC) is raised to 0.85, while the threshold for auxiliary function LNs (such as MMXU) is set to 0.75.

[0089] The present invention achieves efficient and accurate comparison of the SCD model by quantifying the topological consistency of LN clusters and the semantic similarity of LN feature vectors: LN cluster-level matching adopts dual constraints of quantity difference threshold and average similarity threshold to ensure cluster structure stability and functional consistency; LN feature vector-level matching quantifies semantic similarity through cosine similarity, significantly improving the strictness of core function matching; the similarity threshold is dynamically adjusted according to the scenario, balancing matching accuracy and fault tolerance, and improving the generalization ability of the algorithm in different scenarios.

[0090] According to one embodiment of the present invention, the method further comprises:

[0091] Display the substation SCD model consistency verification report through a visual interface;

[0092] Dynamically update the rule base, store matching rules and exception information, and use version control mechanism to maintain the rule base;

[0093] Provides a standard API interface that can be integrated with other power system automation systems;

[0094] For large-scale SCD files, distributed computing technology is used to assign the consistency verification task of the SCD file matching model to multiple computing nodes for parallel execution;

[0095] Supports parsing of the SCL file format that complies with the IEC 61850 standard and can be extended to other SCD file formats.

[0096] Preferably, dynamically updating the rule base includes: storing matching rules and exception information in JSON format; using Git for version control and recording update logs.

[0097] Preferably, a standard API interface is provided that can be integrated with other power system automation systems, including: providing a RESTful API interface for integration with an external power system automation platform.

[0098] Preferably, for large-scale SCD files, distributed computing technology is used to distribute the consistency verification task of the SCD file matching model to multiple computing nodes for parallel execution, including:

[0099] Use Apache Spark to split the SCD file into subtasks based on logical nodes or functional clusters;

[0100] Assign to multiple computing nodes for parallel execution through the task scheduler;

[0101] Automatically detect and redistribute tasks from failed nodes during distributed computing;

[0102] Ensure the integrity of the validation process and the consistency of results.

[0103] This implementation further defines the consistency matching method for the substation SCD model based on the scanning algorithm. Specifically, it develops a visual interface to display verification results. Using D3.js, an interactive chart is generated to intuitively display LN cluster matching results. Color coding (e.g., red marking of mismatched clusters) is used to enhance readability. Mismatched LNs and anomaly information are highlighted, and detailed inconsistency analysis is provided. Clicking on an mismatched LN cluster triggers a detailed difference analysis, including an LN feature vector comparison table, semantic weight change tracking, and IEC 61850 compliance checks, helping engineers quickly locate configuration deviations. Version control logs (e.g., Git commit records) are linked to the substation SCD model consistency verification report to achieve full traceability of configuration changes. A rule base and knowledge management framework utilizes a lightweight JSON format to store matching rules (e.g., threshold parameters) and anomaly pattern libraries (e.g., common noise node characteristics), facilitating dynamic rule updates and cross-platform migration. Branch management enables parallel development and version rollback of the rule base, and compliance with the Semantic Versioning (SemVer) specification ensures the stability of rule iterations. System Integration Capabilities: Utilizing a RESTful API design, this system provides standardized interfaces compliant with the OpenAPI specification. It supports SCD file uploads, verification task submissions, and report downloads, facilitating integration with third-party systems such as control systems and operation and maintenance platforms. The Celery framework is used for asynchronous backend execution of verification tasks, with the frontend receiving real-time progress notifications via WebSocket, enhancing the user experience. Distributed Computing Architecture: For large-scale SCD files (e.g., containing over 10,000 LNs), Spark is used to partition files into subtasks based on functional clusters or voltage levels, enabling fine-grained parallel computing. Integrating Spark's automatic retry mechanism, the task scheduler automatically reassigns subtasks to healthy nodes when a compute node fails, and a checkpointing mechanism ensures that intermediate results are not lost. In-memory columnar storage (such as Apache Arrow) is used to accelerate feature vector computation and improve computation speed. Format Compatibility and Extensibility: ANTLR is used to implement deep parsing of IEC 61850 SCL files, supporting XML Schema validation and fast DOM tree traversal. A plug-in architecture supports parsing modules for new SCD file formats (such as the domestically extended SCD-E format), achieving system scalability. Fault tolerance and consistency assurance: Distributed transaction management uses the Saga model to implement transaction compensation for cross-node tasks, ensuring the eventual consistency of distributed computing results. Hash checksums are introduced during the task merging phase to prevent tampering or loss of subtask results.

[0104] This paper uses the dynamic reporting and exception location functions of D3.js to improve the efficiency of configuration difference analysis; adopts JSON storage + Git version control to support rapid rule iteration and reduce rule maintenance costs; adopts RESTful API and asynchronous task processing to achieve seamless connection with the control system and improve the automation rate of the verification process; adopts Spark architecture to speed up the processing of 10,000-level LN files, and combines fault-tolerant mechanisms to ensure the stability of large-scale verification; adopts ANTLR parser and plug-in design to support multi-standard SCD files, compatible with domestic extension formats, and improve scalability.

[0105] Example 2

[0106] Figure 2 FIG1 is a flow chart of a method for matching the consistency of a substation SCD model based on a scanning algorithm according to another embodiment of the present invention. Figure 2 As shown, according to one embodiment of the present invention, a method for matching the consistency of a substation SCD model based on a scanning algorithm includes the following steps:

[0107] Step S202, SCD file parsing;

[0108] Parse the SCD file to be verified and extract information such as logical nodes (LN), data objects (DO) and data attributes (DA). Figure 3 This is a schematic diagram of SCD file parsing according to an embodiment of the present invention. Figure 3 As shown, use Python's xml.etree.ElementTree library to parse the SCD file in XML format.

[0109] Step S204, constructing LN feature vector;

[0110] Based on the SCD file parsing results, a semantically-based LN feature vector is constructed for each LN. For each LN, the data object (DO) and data attribute (DA) information contained therein are extracted. Combined with the semantic information defined in the IEC 61850 standard, a semantically-based LN feature vector is constructed to measure the similarity between LNs.

[0111] The construction of the LN feature vector specifically includes:

[0112] For each LN, its eigenvector can be expressed as: V(LN)=[w1,w2,...,w i ,...,w n ];

[0113] Among them, "w i " represents the weight of the i-th feature item, reflecting the importance of the feature item to the LN feature vector, i = 1, 2, ..., n; n is the dimension of the feature vector; "wi " can be set according to the importance of data objects (DO) and data attributes (DA).

[0114] Weight allocation: The importance of DO / DA is defined according to the IEC 61850 standard. For example, the DO weight of a critical protection function (such as PTOC) is 1.0; the DA weight of an auxiliary parameter (such as Mod) is 0.3.

[0115] Example: If the LN name is PTOC1 and contains DOStr (weight 1.0) and DAt (weight 0.5), the feature vector is [1.0, 0.5].

[0116] Step S206, SCAN cluster analysis;

[0117] LN feature vectors are clustered using the SCAN algorithm. Logical nodes (LNs) in the SCD model are treated as data points. The SCAN algorithm is used to cluster LNs based on their connectivity. LN feature vectors with similar functions are grouped into the same LN cluster, providing a basis for subsequent consistency matching. After clustering, only similar clusters need to be compared, reducing the amount of computation.

[0118] Parameter settings include: ε neighborhood threshold: set the distance threshold (such as 0.7); minimum number of points: each cluster contains at least 3 LN feature vectors.

[0119] Algorithm Flow: Traverse all LN feature vectors and calculate their cosine similarity with other LN feature vectors. If the number of similar LN feature vectors in the neighborhood of a certain LN feature vector is greater than or equal to the minimum number of points, an LN cluster is formed. The LN cluster is expanded and adjacent LN feature vectors with high similarity are merged.

[0120] Step S208, LN feature vector matching;

[0121] An LN matching algorithm based on cosine similarity is designed. The similarity between the corresponding LN feature vectors in different SCD models is calculated using cosine similarity, and whether the LN feature vectors match is determined based on a preset similarity threshold. The semantic weight is combined with cosine similarity to improve the matching accuracy.

[0122] Figure 4 This is a schematic diagram of a code for calculating the similarity of LN feature vectors based on cosine similarity according to an embodiment of the present invention. Figure 4 As shown, the cosine similarity calculation method includes:

[0123] For two LN feature vectors V(LN1) and V(LN2), their cosine similarity can be expressed as: sim(LN1,LN2)=(V(LN1)·V(LN2)) / (||V(LN1)||*||V(LN2)||);

[0124] Where sim(LN1, LN2) represents the calculation of the cosine similarity between two logical nodes (LN1 and LN2), that is, the cosine similarity between the feature vector of LN1 and the feature vector of LN2; “·” represents the dot product of the vectors; and “||||” represents the modulus of the vector.

[0125] Step S210, LN cluster matching;

[0126] Based on the cluster matching model consistency verification strategy, LN cluster matching is performed. Based on the LN feature vector matching, the structure and LN feature vector composition of the corresponding LN clusters in different SCD models are further compared to determine whether the LN clusters match, thereby achieving consistency verification of the entire SCD model. The consistency of the LN feature vector and LN cluster is determined by cosine similarity.

[0127] For the corresponding LN clusters in different SCD models, the cosine similarity is used to calculate the similarity between the LN feature vectors in the LN cluster, and the preset similarity threshold is used to determine whether the LN feature vectors match.

[0128] The similarity threshold can be adjusted according to actual conditions. If the similarity is greater than the preset similarity threshold, the two LN feature vectors are considered to match. In this embodiment, the similarity threshold is set to 0.8. If two LN clusters meet the following matching criteria, they are considered to match:

[0129] The difference in the number of LN eigenvectors within an LN cluster is ≤10%.

[0130] The average similarity of LN feature vectors in an LN cluster is ≥ 0.75.

[0131] Compare the structure and LN feature vector composition of corresponding clusters in different SCD models to determine whether the LN clusters match. For corresponding LN clusters in two SCD models, compare the number and type of LN feature vectors in the LN clusters. If the LN feature vectors in the LN clusters have a high degree of matching, the two LN clusters are considered to match.

[0132] Step S212, consistency verification;

[0133] Based on the results of LN feature vector matching and LN cluster matching, the consistency of the two substation SCD models is determined. If all corresponding LN clusters and LN feature vectors match, the two SCD models are considered consistent. If there are any mismatched LN clusters or LN feature vectors, the inconsistencies are recorded and a report is generated.

[0134] Implement dynamic updates to the rule base based on version control; store matching rules and exception information found during the verification process in the rule base and update the rule base version. By introducing a version control mechanism, dynamic updates and maintenance of the rule base are achieved. During the rule base update process, new rules are stored in JSON format, for example:

[0135] {

[0136] "rule_id":"R001",

[0137] "description":"The DO of PTOC type LN must contain Str",

[0138] "threshold":0.8

[0139] }

[0140] Use Git for version control and record each update log.

[0141] Step S214, result display.

[0142] Develop a visual interface to display the verification results; provide an intuitive visual interface to display the results of SCD model consistency verification, including matching LNs, mismatching LNs, and abnormal information, etc., to facilitate user analysis and processing.

[0143] Verification results are visually displayed to users, including matching LNs, mismatching LNs, and exception information. A detailed report lists all inconsistencies and their possible causes. Interactive charts are generated using D3.js to highlight mismatching LNs.

[0144] The present invention combines semantic feature modeling with SCAN clustering to achieve efficient consistency verification of the SCD model: 1) The semantic weight allocation mechanism strengthens the matching priority of key functional nodes and improves the accuracy of core function matching; 2) Dynamic threshold adjustment takes into account both strict and loose scenario requirements, and improves the sensitivity of anomaly detection; 3) Closed-loop rule updating and visual reporting functions shorten the system adaptation cycle and significantly improve verification efficiency compared with traditional methods.

[0145] According to one embodiment of the present invention, the substation SCD model consistency matching method based on the scanning algorithm also includes the following features:

[0146] 1. Support parsing of multiple SCD file formats:

[0147] Supports parsing of SCL file formats that comply with the IEC 61850 standard and can be extended to other types of SCD file formats.

[0148] 2. Provide API interface for integration with other systems:

[0149] Provides standard API interfaces (such as RESTful API interfaces) to facilitate integration with other power system automation systems, supports external system call verification services, and realizes the automation of SCD model consistency verification. The code diagram of external system call verification service is as follows Figure 5 shown.

[0150] 3. Use distributed computing to improve verification efficiency:

[0151] For large-scale SCD files, distributed computing technology is used to assign verification tasks to multiple computing nodes for parallel execution, thereby improving verification efficiency.

[0152] Use Apache Spark to split the SCD file into multiple subtasks, parallelize the verification process, and improve verification efficiency. The specific operation method is as follows:

[0153] Task decomposition: Decompose the SCD file verification task into multiple subtasks. For example, divide the SCD file into multiple subsets based on logical nodes (LNs) or functional clusters (Clusters). Each subset contains a certain number of LNs or clusters, which serve as independent computing units.

[0154] Task allocation: Use a task scheduler (such as Apache Spark Driver) to assign subtasks to multiple computing nodes (Workers) in the cluster.

[0155] Through multi-format compatible parsing and standardized interface design, the present invention realizes the cross-platform verification capability of heterogeneous SCD models of power systems and improves the efficiency of API integration; the parallel verification mechanism based on the distributed computing framework significantly shortens the processing time of large-scale SCD files and meets real-time requirements.

[0156] The present invention can effectively solve the problems existing in the prior art, such as the difficulty in maintaining the rule base, low efficiency in model comparison, and lack of consideration for semantic information, and has the following beneficial effects: a) Improving verification efficiency: By introducing the SCAN algorithm for model clustering, the matching range is narrowed and the verification efficiency is improved. b) Improving matching accuracy: By constructing a semantic-based LN feature vector and designing an LN feature vector matching algorithm based on cosine similarity, the matching accuracy is improved. c) Enhancing scalability: By implementing dynamic updates of the rule base based on version control and providing an API interface, the scalability of the method is enhanced. d) Reducing maintenance costs: By automated verification and dynamic updates of the rule base, manual maintenance costs are reduced. e) Increasing fault tolerance: Automatically handling node failures to ensure task completion. f) Significantly shortening verification time: Simultaneously supporting the verification of large-scale SCD files, and significantly shortening the verification time through parallel computing.

[0157] Example 3

[0158] Figure 6 Schematic diagram of a substation SCD model consistency matching system based on a scanning algorithm according to an embodiment of the present invention. Figure 6 As shown, according to one embodiment of the present invention, a substation SCD model consistency matching system based on a scanning algorithm includes:

[0159] Parsing module 10, used to parse the SCD file and generate an LN feature vector based on the logical node LN, data object DO and data attribute DA;

[0160] The clustering module 20 is used to perform cluster analysis on the LN feature vectors based on the SCAN algorithm to generate LN clusters; and generate an SCD file matching model based on the LN clusters;

[0161] The matching module 30 is configured to perform consistency matching on the first substation SCD file matching model and the second substation SCD file matching model, including: traversing all LN clusters of the two substation SCD file matching models, and determining whether each corresponding LN cluster matches;

[0162] The output module 40 is used to summarize the matching results of all LN clusters of the two substation SCD file matching models and generate a substation SCD model consistency verification report.

[0163] According to one embodiment of the present invention, the system further includes:

[0164] A display module 50 is used to display the substation SCD model consistency verification report through a visual interface;

[0165] Database module 60, used to store matching rules and exception information, dynamically update the rule base, and use version control mechanism to maintain the rule base;

[0166] The interface module 70 is used to provide a standard API interface that can be integrated with other power system automation systems.

[0167] According to one embodiment of the present invention, the parsing module 10 also supports parsing the SCL file format that complies with the IEC 61850 standard, and is extended to other SCD file formats;

[0168] The matching module 30 is also used to allocate the consistency verification task of the SCD file matching model to multiple computing nodes for parallel execution for large-scale SCD files by using distributed computing technology.

[0169] The present invention realizes efficient consistency verification of the substation SCD model by integrating a multi-module collaborative mechanism: 1) The LN feature vector clustering model constructed based on the SCAN algorithm significantly improves the accuracy and anti-interference ability of model matching; 2) The distributed computing architecture supports large-scale parallel processing of SCD files, breaking through the performance bottleneck of traditional single-point computing; 3) The dynamic rule base and version control mechanism ensure that the verification rules are updated in real time to adapt to the evolution requirements of the power system model; 4) The standardized API interface and visual report output not only ensure the convenience of system integration, but also enhance the traceability of the verification results; 5) The cross-format parsing capability is compatible with the IEC 61850 standard and extended format, significantly enhancing the scope of application of the system. Through the deep integration of algorithm optimization, architectural innovation and engineering design, the present invention has constructed an SCD model consistency verification system with high precision, strong scalability and easy maintenance.

[0170] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working process of the system described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0171] Example 4

[0172] Figure 7 This is the SCD model file information of the substation of manufacturer A. Figure 8 The substation SCD model file information of manufacturer B. The consistency of the substation SCD models of manufacturer A and manufacturer B is verified by using a substation SCD model consistency matching method based on a scanning algorithm of the present invention, which specifically includes the following steps:

[0173] Step S302, SCD file parsing;

[0174] Parse the SCD files of manufacturer A and manufacturer B respectively, extract information such as LN, DO, and DA in the SCD files, and obtain target extraction information.

[0175] Step S304, constructing LN feature vector;

[0176] Based on the target extraction information, a semantic-based LN feature vector is constructed. For the LN feature vector, the semantic weight distribution rules of the LN feature vector include:

[0177] Data Object (DO): Assigns weights based on functional importance as defined in the IEC 61850 standard.

[0178] Key features (such as PTOC's Str) have a weight of 1.0.

[0179] Auxiliary functions (such as Mod) have a weight of 0.3.

[0180] Data Attributes (DA): Assign weights based on parameter necessity.

[0181] Required attributes (such as t) have a weight of 0.5.

[0182] Optional attributes (such as q) have a weight of 0.2.

[0183] For example, the LN name is PTOC1, which contains DOStr (weight 1.0) and DAt (weight 0.5). Then the LN feature vector is: V(PTOC1) = [1.0, 0.5].

[0184] Step S306, clustering based on the SCAN algorithm;

[0185] Cluster the LNs in the SCD file by function (such as protection, measurement) to form LN clusters with similar functions. For example, LN cluster 1 is protection-type LNs (PTOC, PDIS, etc.). LN cluster 2 is measurement-type LNs (MMXU, MMTR, etc.). The algorithm steps are as follows:

[0186] Step 1: Parameter definition

[0187] ε neighborhood: Set a similarity threshold (such as 0.8). If the cosine similarity of two LN feature vectors is ≥ε, they are considered to be nodes in the neighborhood.

[0188] Minimum number of points (MinPts): The minimum number of LNs that each LN cluster contains (e.g., 3).

[0189] Step 2: Core point identification:

[0190] Traverse all LN feature vectors. If the number of LNs in the ε neighborhood of a certain LN feature vector is ≥ MinPts, it is marked as a core point.

[0191] Step 3: LN cluster expansion:

[0192] Starting from the core point, all reachable LN feature vectors in its neighborhood are recursively merged to form an LN cluster.

[0193] Step 4: Noise processing:

[0194] LN feature vectors that cannot be classified into any LN cluster are marked as noise, which may be an abnormal configuration.

[0195] Step S308, LN cluster matching;

[0196] For the corresponding LN clusters in the two SCD models, the cosine similarity is used to calculate the similarity between the LN feature vectors in the LN clusters, and the LN feature vectors are judged to be matched based on the preset similarity threshold. The cosine similarity calculation formula is as follows:

[0197]

[0198] The matching rules are:

[0199] If the cosine similarity is ≥ 0.8 (similarity threshold), the two LN feature vectors are considered to match; otherwise, they are considered to not match.

[0200] The similarity threshold can be dynamically adjusted according to actual needs. For example, in a strict scenario, the similarity threshold is set to 0.85; in a loose scenario, the similarity threshold is set to 0.75.

[0201] For example, in this embodiment, the cosine similarity of `PTOC1` is calculated to be 0.85 (exceeding the similarity threshold of 0.8), and is determined to be a match.

[0202] Step S310, LN cluster matching;

[0203] Compare the structures and LN feature vector compositions of the corresponding LN clusters in the two SCD models to determine whether the LN clusters match.

[0204] The matching criteria are:

[0205] Structural consistency: the difference in the number of LN feature vectors between two LN clusters is ≤10% (quantity difference threshold).

[0206] Functional consistency: the average similarity of LN feature vectors within an LN cluster is ≥ 0.75 (average similarity threshold).

[0207] For example, in this embodiment, LN cluster 1 of manufacturer A contains 10 LN feature vectors, and LN cluster 1 of manufacturer B contains 9 LN feature vectors (quantity difference = 10%), and the average similarity is 0.78 (greater than 0.75), so LN cluster 1 is determined to be a match.

[0208] Step S312, consistency verification;

[0209] Based on the results of LN feature vector matching and LN cluster matching, it is determined whether the two SCD models are consistent.

[0210] Step S314, result display;

[0211] The verification results are displayed to the user visually, including matching LNs, mismatching LNs, and exception information. The two SCD files are identical in key functions, but vendor B is missing a secondary LN, generating a report prompt.

[0212] Step S316, rule base update;

[0213] The matching rules and exception information found during the verification process are stored in the rule base and the rule base version is updated.

[0214] According to the above method, the comparison results of manufacturer A and manufacturer B are as follows:

[0215] Step 1: Parse the file

[0216] Use Python's xml.etree.ElementTree to parse the SCD files of manufacturer A and manufacturer B.

[0217] Extract logical nodes (LN), data objects (DO) and data attributes (DA).

[0218] Step 2: Feature vector construction

[0219] PTOC1 of manufacturer A: DOStr (weight 1.0), DAt (weight 0.5) → LN feature vector [1.0, 0.5].

[0220] PTOC1 of manufacturer B: DOStr (weight 1.0), DAt (weight 0.5) → LN feature vector [1.0, 0.5].

[0221] Step 3: SCAN clustering

[0222] Parameter settings: ε=0.7, MinPts=3.

[0223] Manufacturer A is clustered into 5 LN clusters, and manufacturer B is clustered into 4 LN clusters.

[0224] Step 4: LN feature vector matching

[0225] The cosine similarity calculated for PTOC1 is 1.0 (perfect match).

[0226] Step 5: LN cluster matching

[0227] The difference in the number of LN feature vectors between LN cluster 1 (manufacturer A) and LN cluster 1 (manufacturer B) is 0%, and the average similarity is 0.92, so they are determined to be a match.

[0228] Result: The core functions of the two files are the same, but manufacturer B is missing an auxiliary LN. A report is generated and a prompt is given.

[0229] The present invention combines semantic feature modeling with the SCAN clustering algorithm to achieve high-precision consistency verification of the manufacturer's SCD model: 1) The semantic weight allocation mechanism strengthens the matching priority of key functional nodes and improves the accuracy of core function matching; 2) Dynamic threshold adjustment takes into account both strict and loose scenario requirements, and improves the sensitivity of anomaly detection; 3) The noise processing mechanism effectively identifies redundant configurations and reduces the false alarm rate; 4) The versioned rule base automatic update function shortens the system adaptation cycle; 5) The visual report can accurately locate the missing auxiliary LN of manufacturer B, and the verification efficiency is significantly improved compared with traditional methods.

[0230] Example 5

[0231] In this embodiment, a certain SCD file in actual operation is compared with a standard SCD model, and a substation SCD model consistency matching method based on a scanning algorithm of the present invention is adopted, which specifically includes:

[0232] Step 1: Parsing and feature extraction

[0233] MMXU1 in the actual file: DOPhsA (weight 0.8), DAmag (weight 0.4) → LN feature vector [0.8, 0.4].

[0234] MMXU1 in the standard model: DOPhsA (weight 1.0), DAmag (weight 0.5) → LN eigenvector [1.0, 0.5].

[0235] Step 2: Similarity calculation

[0236] The cosine similarity between the LN feature vector in the actual file and the LN feature vector in the standard model is calculated. The cosine similarity is 0.72 (lower than the threshold value of 0.8), which is determined to be a mismatch.

[0237] Step 3: Update the rule base

[0238] Based on the matching situation, the rule base content is updated and new rules are added:

[0239] {

[0240] "rule_id":"R002",

[0241] "description":"MMXU class LN must contain PhsA",

[0242] "threshold":0.8

[0243] }

[0244] The present invention achieves precise positioning of SCD model deviations through the linkage of semantic feature vector matching and dynamic rule base: 1) Cosine similarity calculation based on weight difference effectively identifies MMXU1 configuration offsets; 2) Automatic generation of rule R002 strengthens PhsA parameter verification, thereby improving the detection rate of similar problems; 3) A threshold-driven mismatch judgment mechanism ensures 100% core function verification coverage; 4) A closed-loop rule update process shortens the cycle of system adaptation standard evolution.

[0245] The present invention can effectively solve the problems existing in the prior art, such as the difficulty in maintaining the rule base, the low efficiency of model comparison, and the lack of consideration of semantic information, and has high practical value and promotion prospects.

[0246] The above description is merely a preferred embodiment of the present application and an illustration of the technical principles employed. Those skilled in the art should understand that the scope of the invention herein is not limited to the technical solutions formed by the specific combination of the above-mentioned technical features, but also encompasses other technical solutions formed by any combination of the above-mentioned technical features or their equivalents without departing from the inventive concept. For example, a technical solution formed by replacing the above-mentioned features with (but not limited to) technical features having similar functions disclosed in this application.

[0247] It should be understood that the size of the serial numbers of each step in the content of the invention and the embodiments of the present invention does not absolutely mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

Claims

1. The substation SCD model consistency matching method based on scanning algorithm is characterized by: The method comprises the following steps: Parse the SCD file and generate the LN feature vector based on the logical node LN, data object DO and data attribute DA; Performing cluster analysis on the LN feature vector based on the SCAN algorithm to generate an LN cluster; generating an SCD file matching model based on the LN cluster; Performing consistency matching on the first substation SCD file matching model and the second substation SCD file matching model, including: traversing all LN clusters of the two substation SCD file matching models, and determining whether each corresponding LN cluster matches; The matching results of all LN clusters of the two substation SCD file matching models are summarized to generate a substation SCD model consistency verification report.

2. The substation SCD model consistency matching method based on the scanning algorithm according to claim 1 is characterized in that: Generating an LN feature vector based on the logical node LN, the data object DO, and the data attribute DA, including: generating the LN feature vector based on the data object DO and the data attribute DA of each logical node LN in combination with a semantic weight allocation rule defined in the IEC 61850 standard; The semantic weight distribution rules include: the DO weight of the data object corresponding to the key function is 1.0, and the DO weight of the data object corresponding to the auxiliary function is 0.3; the DA weight of the data attribute corresponding to the mandatory attribute is 0.5, and the DA weight of the data attribute corresponding to the optional attribute is 0.

2.

3. The substation SCD model consistency matching method based on scanning algorithm according to claim 1 is characterized in that: Performing cluster analysis on the LN feature vector based on the SCAN algorithm to generate an LN cluster, including: Based on the preset neighborhood threshold and minimum number of points, LN feature vectors with similar functions are divided into the same LN cluster; The identification condition of the core point is: the number of similar LN feature vectors in its neighborhood is not less than the minimum number of points; The LN feature vectors that cannot be classified into any LN cluster are marked as noise nodes.

4. The substation SCD model consistency matching method based on the scanning algorithm according to claim 3 is characterized by: The neighborhood threshold is 0.7-0.8, and the minimum number of points is 3.

5. The substation SCD model consistency matching method based on scanning algorithm according to claim 1 is characterized in that: Determine whether each corresponding LN cluster matches, including: Determining whether the LN clusters match based on the structure of the LN clusters and the composition of the LN feature vectors within the LN clusters; The similarity of the corresponding LN feature vectors in the two SCD file matching models is calculated based on the cosine similarity, and whether the LN feature vectors match is determined based on the similarity threshold.

6. The substation SCD model consistency matching method based on scanning algorithm according to claim 5 is characterized in that: Determining whether the LN clusters match includes determining whether the LN clusters match according to a difference in the number of LN feature vectors within the LN clusters and an average similarity of the LN feature vectors within the LN clusters.

7. The method for matching the substation SCD model consistency based on the scanning algorithm according to claim 6 is characterized in that: The LN cluster matching criteria include: the number difference of LN feature vectors within the LN cluster does not exceed 10%; the average similarity of LN feature vectors within the LN cluster is not less than 0.

75.

8. The substation SCD model consistency matching method based on scanning algorithm according to claim 5 is characterized by: The calculation formula of the cosine similarity is: sim(LN1,LN2)=(V(LN1)·V(LN2)) / (||V(LN1)||*||V(LN2)||) Where sim(LN1,LN2) represents the cosine similarity between the LN1 feature vector and the LN2 feature vector; V(LN1) represents the LN1 feature vector; V(LN2) represents the LN2 feature vector; "·" represents the dot product of the vector; "||||" represents the modulus of the vector; The similarity threshold is 0.75-0.

85.

9. The substation SCD model consistency matching method based on scanning algorithm according to claim 1 is characterized in that: The method further comprises: Display the substation SCD model consistency verification report through a visual interface; Dynamically update the rule base, store matching rules and exception information, and use a version control mechanism to maintain the rule base; Provides a standard API interface that can be integrated with other power system automation systems; For large-scale SCD files, distributed computing technology is used to assign the consistency verification task of the SCD file matching model to multiple computing nodes for parallel execution; Supports parsing of the SCL file format that complies with the IEC 61850 standard and can be extended to other SCD file formats.

10. The substation SCD model consistency matching system based on scanning algorithm is characterized by: The system comprises: Parsing module, used to parse the SCD file and generate LN feature vector based on logical node LN, data object DO and data attribute DA; A clustering module, configured to perform cluster analysis on the LN feature vectors based on a SCAN algorithm to generate LN clusters; and generate an SCD file matching model based on the LN clusters; A matching module is used to perform consistency matching on the SCD file matching model of the first substation and the SCD file matching model of the second substation, including: traversing all LN clusters of the SCD file matching models of the two substations, and determining whether each corresponding LN cluster matches; The output module is used to summarize the matching results of all LN clusters of the two substation SCD file matching models and generate a substation SCD model consistency verification report.

Images