A method, system, device, and medium for multi-level grid diagram module splicing

Abstract

The application discloses a kind of multi-level power grid chart module splicing method, system, equipment and medium, belong to power grid dispatching automation and digital twin field, including: acquisition original multi-level power grid model file, construct global ontology knowledge base, carry out semantic coding and coordinate reduction to multi-level power grid model, generate chart module layered index, retrieve spatial overlap area, calculate the spatial Euclidean distance between the nodes of multi-level power grid model, adopt the matching strategy of preestablished and establish node mapping and branch coupling relationship, extract topological structure and node state vector, construct graph structure, carry out topological consistency check and parameter coordination optimization, obtain boundary correction parameter, update multi-level power grid model splicing section, carry out logic suture to multi-level power grid model, output multi-level power grid chart module.The application forms global power grid chart module that is electrically continuous, topologically consistent and supports panoramic analysis, overcomes the difficulty that multi-level power grid model is difficult to seamlessly integrate and collaborative application.

Description

Technical Field

[0001] This invention relates to the fields of power grid dispatch automation and digital twin technology, specifically to a method, system, equipment, and medium for multi-level power grid diagram splicing. Background Technology

[0002] With the rapid development of new power systems and the advancement of integrated power generation, grid, load, and storage construction, the power grid is gradually evolving from a traditional unidirectional transmission network to a complex interconnected system encompassing ultra-high-voltage main grids, distribution networks, distributed microgrids, and a massive number of user terminals. To achieve real-time situational awareness, accurate fault prediction, and comprehensive carbon flow tracking across the entire network, constructing a digital twin power grid model with physical topology connectivity, consistent electrical parameters, and coverage of all aspects has become an urgent industry requirement.

[0003] However, existing technologies still have shortcomings. In terms of semantic fusion of multi-source heterogeneous data, they mainly focus on time-series alignment and denoising normalization of real-time running data, lacking a unified semantic processing mechanism for deep-level heterogeneous data from microgrids and user sides. When faced with model files of different formats, it is difficult to automatically resolve issues of homonyms or conflicting attribute definitions, resulting in the inability to build a four-level global standardized index covering primary, secondary, micro, and user domains. Regarding the accuracy of cross-level topology matching, it mainly relies on known physical connection points or topology structures generated based on optimization algorithms to locate boundary nodes. When facing scenarios with complex spatial relationships or missing topology files, it lacks a basic... Advanced verification methods for graph-based fingerprints are prone to mismatches or missed matches due to reliance on spatial location or simple electrical properties. In terms of eliminating splicing gaps and handling consistency, traditional power flow calculation methods are mainly used to forcibly adjust the boundary power balance. However, when there are significant differences in measurement error accuracy between the main network and microgrid / user-side models, such as the mixing of PMU and SCADA data, or inconsistent model simplification, traditional iterative algorithms are prone to non-convergence and cannot intelligently smooth numerical cliffs through nonlinear fitting like physical information neural networks, making it difficult to generate a truly globally consistent graph model that satisfies physical constraints and has smooth data. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by this invention is that the existing technology is unable to overcome the huge differences in semantic definition and spatial coordinates of the multi-source heterogeneous models of the main grid, distribution network, microgrid and user side, and cannot achieve automated and accurate matching of cross-level topologies in complex power grid environments. Moreover, it is difficult to effectively eliminate the gaps and physical violations of electrical parameters in the splicing section caused by differences in measurement accuracy and model heterogeneity, so it is impossible to construct a globally consistent panoramic power grid model that supports accurate power flow calculation and step-by-step drilling down.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a method for splicing multi-level power grid diagrams, comprising, Collect original multi-level power grid model files, construct a global ontology knowledge base, perform semantic encoding and coordinate reduction on the multi-level power grid model, and generate a hierarchical index of the graph model; The system calls the hierarchical index of the graph model to retrieve spatially overlapping areas, calculates the spatial Euclidean distance between nodes of the multi-level power grid model, and uses a preset matching strategy to identify and establish the node mapping and branch coupling relationship. Based on the node mapping and branch coupling relationship, the topology and node state vectors are extracted, a graph structure is constructed, the composite loss function value is calculated, topology consistency verification and parameter coordination optimization are performed, boundary correction parameters are obtained, the splicing section of the multi-level power grid model is updated, the multi-level power grid model is logically stitched together, and a multi-level power grid graph model is output.

[0007] As a preferred embodiment of the multi-level power grid model stitching method described in this invention, the steps of collecting original multi-level power grid model files, constructing a global ontology knowledge base, performing semantic encoding and coordinate reduction on the multi-level power grid model, and generating a hierarchical index for the model include... Collect original multi-level power grid model files and construct a global ontology knowledge base; Based on a global ontology knowledge base, the spatial coordinates of a multi-level power grid model are transformed through spatial reference system transformation to generate a globally coordinated element set; Based on the global coordinate element set, normalized coordinate data is extracted, a top-level sparse grid is constructed, and a hierarchical index of the graph model is generated.

[0008] As a preferred embodiment of the multi-level power grid model splicing method described in this invention, the steps of calling the model hierarchical index to retrieve spatially overlapping areas, calculating the spatial Euclidean distance between nodes of the multi-level power grid model, and identifying and establishing node mapping and branch coupling relationships using a preset matching strategy include: Call the hierarchical index of the graph model to identify the spatial overlap area of ​​the multi-level power grid model in the Earth fixed coordinate reference system; Based on the spatially overlapping region, the spatial Euclidean distance between nodes in the multi-level power grid model is calculated, and the set of candidate spatially overlapping nodes is screened. Based on the spatially overlapping candidate node set, fingerprint vectors are generated and the cosine similarity of the fingerprint vectors is calculated. The topologically isomorphic node set is then selected, the Jacobian sensitivity matrix is ​​constructed, and the node mapping and branch coupling relationship is established.

[0009] In a preferred embodiment of the multi-level power grid model splicing method described in this invention, the following steps are performed: Based on node mapping and branch coupling relationships, topology and node state vectors are extracted to construct a graph structure; composite loss function values ​​are calculated; topology consistency verification and parameter coordination optimization are conducted to derive boundary correction parameters; the splicing section of the multi-level power grid model is updated; the multi-level power grid model is logically stitched together; and the output multi-level power grid model includes… Based on the node mapping and branch coupling relationship, the topology and node state vectors are extracted to construct a graph structure. Based on the graph structure, the composite loss function value is calculated, topology consistency verification and parameter coordination optimization are performed, and boundary correction parameters are obtained. Based on the boundary correction parameters, the splicing section of the multi-level power grid model is updated, the multi-level power grid model is logically stitched together, and the multi-level power grid diagram is output.

[0010] In a preferred embodiment of the method for stitching together multi-level power grid diagrams according to the present invention, the step of transforming the spatial coordinates of the multi-level power grid model through spatial reference system transformation includes... The spatial transformation is performed using the Bursa seven-parameter coordinate transformation model, expressed as follows: in, The coordinates are the transformed coordinates. The translation factor is... For rotation factor, As a scale factor, These are the coordinates before the transformation.

[0011] As a preferred embodiment of the multi-level power grid diagram splicing method described in this invention, the calculation of the cosine similarity of the fingerprint vector includes, The cosine similarity of fingerprint vectors is calculated using the following expression: in, Cosine similarity between spatially overlapping candidate node pairs. The topological fingerprint vector is generated by iteratively processing the first node in the paired nodes using the Weisfeller-Lyman algorithm. The topological fingerprint vector is generated by iteratively applying the Weisfeller-Lyman algorithm to the second node in the paired nodes. This is the operation on the magnitude of a vector.

[0012] This invention uses the Wiesfeld-Lyman algorithm to generate fingerprint vectors and calculate cosine similarity, overcoming the mismatches that may result from relying solely on spatial distance. It can reliably distinguish heterogeneous equipment nodes that are geographically close but belong to different electrical islands or voltage levels, thus improving the accuracy of automatic splicing of cross-level power grid models.

[0013] As a preferred embodiment of the multi-level power grid diagram splicing method described in this invention, the calculation of the composite loss function value includes, Construct a composite loss function, the mathematical expression of which is: in, The value of the composite loss function. These are the pre-defined weighting coefficients for the data fitting terms, used to adjust the penalty for deviations between predicted and measured values. These are the weighting coefficients for preset physical constraints, used to adjust the penalty for violating the physical operating rules of the power grid. Predict state vectors for boundary nodes. These are the original measurement values. This is the residual function of nodal power imbalance constructed based on Kirchhoff's laws. This is the preset L2 norm operator.

[0014] This invention integrates the data fitting deviation between the predicted state of the boundary node and the original measurement value, as well as the physical constraint of node power imbalance based on Kirchhoff's laws, into a unified mathematical framework for joint optimization. During the parameter coordination process, it synchronously approximates the actual measurement and strictly satisfies the basic physical laws of the power grid, eliminating the problems of electrical parameter discontinuity and power imbalance caused by model heterogeneity and measurement errors at the splicing section.

[0015] This invention provides a system for splicing multi-level power grid diagrams.

[0016] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a system for splicing multi-level power grid diagrams, comprising: a construction module, a matching module, and a splicing and generation module. The construction module is to collect the original multi-level power grid model files, build a global ontology knowledge base, perform semantic encoding and coordinate reduction on the multi-level power grid model, and generate a hierarchical index of the graph model; The matching module calls the hierarchical index of the graph model to retrieve spatially overlapping areas, calculates the spatial Euclidean distance between nodes of the multi-level power grid model, and uses a preset matching strategy to identify and establish the node mapping and branch coupling relationship. The splicing and generation module extracts the topology and node state vectors based on the node mapping and branch coupling relationship, constructs the graph structure, calculates the composite loss function value, performs topology consistency verification and parameter coordination optimization, obtains boundary correction parameters, updates the splicing section of the multi-level power grid model, logically stitches the multi-level power grid model, and outputs the multi-level power grid graph model.

[0017] The present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, characterized in that the processor executes the computer program to implement the steps of the method for splicing multi-level power grid diagrams.

[0018] The present invention provides a computer-readable storage medium having a computer program stored thereon, characterized in that the computer program, when executed by a processor, implements the steps of the method for splicing multi-level power grid diagrams.

[0019] The beneficial effects of this invention are as follows: This invention achieves unified calculation of semantics and coordinates of multi-source heterogeneous models by constructing a global ontology knowledge base, accurately locates cross-level topology connection points by using graph model hierarchical indexing and spatial overlap retrieval, and establishes node mapping and branch coupling relationships by integrating spatial distance, topology fingerprint similarity and sensitivity analysis; then, based on graph structure and composite loss function, topology verification and parameter coordination are performed to eliminate electrical parameter gaps and physical violations in model splicing sections, and finally form a globally consistent panoramic power grid model that supports power flow calculation and step-by-step drill-down, solving the technical problem of automated cross-level topology matching and seamless model splicing in complex power grid environments. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a general flowchart of a method for splicing multi-level power grid diagrams according to an embodiment of the present invention. Detailed Implementation

[0022] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0023] Example 1, referring to Figure 1 This is one embodiment of the present invention, which provides a method for splicing multi-level power grid diagrams, including: To address the challenges of existing technologies in overcoming the significant differences in semantic definitions and spatial coordinates among heterogeneous models of the main grid, distribution network, microgrid, and user side, which prevent automated and accurate matching of cross-level topologies in complex power grid environments, and which struggle to effectively eliminate gaps in electrical parameters and physical violations at splicing sections caused by differences in measurement accuracy and model heterogeneity, thus hindering the construction of a globally consistent panoramic power grid model that supports accurate power flow calculation and step-by-step drill-down, this invention provides a method for splicing multi-level power grid models.

[0024] S1: Collect original multi-level power grid model files, construct a global ontology knowledge base, perform semantic encoding and coordinate reduction on the multi-level power grid model, and generate a hierarchical index of the graph model; S2: Call the graph model hierarchical index to retrieve spatially overlapping areas, calculate the spatial Euclidean distance between nodes of the multi-level power grid model, and use a preset matching strategy to identify and establish node mapping and branch coupling relationships; S3: Based on the node mapping and branch coupling relationship, extract the topology and node state vectors, construct the graph structure, calculate the composite loss function value, perform topology consistency verification and parameter coordination optimization, obtain boundary correction parameters, update the splicing section of the multi-level power grid model, logically stitch the multi-level power grid model, and output the multi-level power grid graph model.

[0025] Therefore, to solve the semantic and spatial alignment problem of multi-source heterogeneous models, a hierarchical index and multi-level matching strategy are used to identify and associate cross-level power grid topologies in spatially overlapping areas. Through joint optimization based on physical constraints and data fitting, parameter inconsistencies in splicing sections are coordinated and corrected, thereby forming a global power grid model that is electrically continuous, topologically consistent, and supports panoramic analysis and calculation, overcoming the difficulty of seamless integration and collaborative application of multi-level power grid models.

[0026] Example 2, an embodiment of the present invention, provides a method for splicing multi-level power grid diagrams based on the previous embodiment, including: In this embodiment of the application, step S1 involves collecting the original multi-level power grid model file, constructing a global ontology knowledge base, performing semantic encoding and coordinate reduction on the multi-level power grid model, and generating a hierarchical index of the graph model, including the following steps A1-A3: A1: Collect original multi-level power grid model files and construct a global ontology knowledge base.

[0027] A global ontology knowledge base is established by acquiring CIM / E files from the main network, SVG / G graphic files from the distribution network, BIM model files from the microgrid, and EMS / AMI model files from the user side. Through an entity alignment algorithm, private object tags and private attribute fields in the CIM / E files, SVG / G graphic files, BIM model files, and EMS / AMI model files are mapped to the global ontology knowledge base, generating a global semantic alignment model object set. The global ontology knowledge base is defined by global power equipment, which includes primary transmission and transformation equipment, distribution automation equipment, distributed power sources, and user-side energy terminals.

[0028] A2: Based on the global ontology knowledge base, the spatial coordinates of the multi-level power grid model are transformed through spatial reference system transformation to generate a global coordinate element set.

[0029] Identify the WGS84, Xi'an 80, or local independent coordinate system used in the multi-level power grid model.

[0030] By using preset reference voltage and reference power, the voltage parameters, power parameters and impedance parameters of the multi-level power grid model are normalized to generate a normalized electrical feature set. The multi-level power grid model includes the main grid model, distribution network model, microgrid model and user-side model.

[0031] To eliminate the order-of-magnitude differences caused by voltage levels, such as 500kV and 380V, a system reference power is selected. and reference voltages at all levels The actual measured values ​​are converted into dimensionless per-unit values, and numerical per-unit normalization is performed to generate a normalized electrical feature set.

[0032] A3: Based on the global coordinate element set, extract normalized coordinate data, construct a top-level sparse grid, and generate a hierarchical index for the graph model.

[0033] Based on the binary approximation method and latitude-longitude binary conversion, the normalized geographic coordinates of the main network model elements are extracted and denoted as longitude. and latitude The initial interval for Earth's longitude is set as follows: The initial interval of latitude is .

[0034] Longitude is encoded, and the longitude interval is divided into two sub-intervals, left and right. If the longitude... If the interval falls within the right subinterval, the binary bit is set to 1, and the interval for the next iteration is updated to the right subinterval. If the longitude... If the value falls within the left sub-interval, the binary bit is marked as 0, and the interval for the next iteration is updated to be the left sub-interval. This process is repeated n times, where n is the preset precision in bits, to generate the longitude binary string. .

[0035] The latitude is encoded using the same binary search method to determine the latitude. At the current latitude range, generate a length of latitudinal binary string .

[0036] Construct a length of The mixed binary sequence is merged according to the rule of placing longitude in even positions and latitude in odd positions, or alternating longitude and latitude. The The bit is placed in the first position of the mixed sequence. The 0th, 2nd, 4th... digits will The The bit is placed in the first position of the mixed sequence. Bits, i.e., bits 1, 3, 5... are used to generate the merged Z-order curve binary stream.

[0037] The Base32 character mapping divides the merged binary stream into groups of 5 bits, calculates the corresponding decimal value for each group, ranging from 0 to 31. It then calls a pre-defined Base32 encoding mapping table, which typically contains the digits 0-9 and the letters bz, removing the easily confused letters a, i, l, and o, mapping each decimal value to its corresponding Base32 character. All characters are then concatenated sequentially to generate the GeoHash string encoding of the mainnet model element.

[0038] For the hierarchical elements of the distribution network, microgrid, and user side within the grid, the R-Tree algorithm is constructed to attach the physical IDs of the hierarchical elements of the distribution network, microgrid, and user side to the corresponding index leaf nodes, thereby generating a graph model hierarchical index.

[0039] In this embodiment of the application, the spatial coordinate transformation of the multi-level power grid model in step A2 through spatial reference system transformation specifically adopts the Bursa seven-parameter coordinate transformation model: The spatial transformation is performed using the Bursa seven-parameter coordinate transformation model, expressed as follows: in, The coordinates are the transformed coordinates. The translation factor is... For rotation factor, As a scale factor, These are the coordinates before the transformation.

[0040] In an optional implementation, the spatial coordinate transformation of the multi-level power grid model in step A2 by spatial reference system transformation can also be carried out by differential grid interpolation. The grid file is loaded, the source coordinates to be transformed are located in a specific cell of the grid file, the coordinate difference values ​​stored at the four corner points of the cell are calculated by bilinear interpolation to obtain the coordinate correction of the point, and finally the correction is added to the source coordinates to obtain the target coordinates.

[0041] In another alternative implementation, the spatial coordinate transformation of the multi-level power grid model in step A2 by spatial reference system transformation can also be carried out using a three-parameter model. Three translation parameters (offset values ​​in the east-west, north-south, and elevation directions) are determined. During the transformation, the target coordinates can be obtained by directly adding the corresponding fixed offset value to the coordinates of each point in the source coordinate system.

[0042] In the embodiments of this application, step A2 transforms the spatial coordinates of the multi-level power grid model through spatial reference system transformation, which solves the inconsistency problem of the spatial reference of the multi-level power grid model from different mapping systems or independent coordinate systems. By uniformly estimating translation, rotation and scale parameters, the scattered local coordinates can be transformed to a unified Earth fixed coordinate reference system.

[0043] In this embodiment of the application, step S2 involves calling the graph model hierarchical index to retrieve spatially overlapping regions, calculating the spatial Euclidean distance between nodes in the multi-level power grid model, and using a preset matching strategy to identify and establish node mapping and branch coupling relationships, including the following steps B1-B3: B1: Call the layered index of the model, the spatial overlap area of ​​the multi-level power grid model in the fixed coordinate reference system of the earth.

[0044] Call the hierarchical index of the model to retrieve the spatial overlap region of the multi-level power grid model in the Earth fixed coordinate reference system.

[0045] B2: Based on the spatially overlapping region, calculate the spatial Euclidean distance between nodes in the multi-level power grid model and filter the set of candidate nodes with spatial overlap.

[0046] Matching strategies include boundary matching strategy, topology matching strategy, and parameter matching strategy.

[0047] Boundary matching strategies specifically include: The spatial Euclidean distance between nodes of the multi-level power grid model is calculated for the spatially overlapping region. The spatial Euclidean distance between nodes of the multi-level power grid model includes the spatial Euclidean distance between the main grid model node and the distribution network model node, the spatial Euclidean distance between the distribution network model node and the microgrid model node, and the spatial Euclidean distance between the microgrid model node and the user-side model node.

[0048] B3: Based on the spatially overlapping candidate node set, generate fingerprint vectors and calculate the cosine similarity of fingerprint vectors, filter the topologically isomorphic node set, construct the Jacobian sensitivity matrix, and establish the node mapping and branch coupling relationship.

[0049] Taking each spatially overlapping candidate node in the set of spatially overlapping candidate nodes as the center node, the k-hop neighborhood subgraph of each spatially overlapping candidate node in the corresponding multi-level power grid model is extracted.

[0050] The k-hop neighborhood subgraph is iteratively hashed using the Weisfeller-Lehman algorithm to generate a topological fingerprint vector, which is a local connection structure of nodes.

[0051] Calculate the cosine similarity of the topological fingerprint vectors between spatially overlapping candidate node pairs, and output the paired nodes with a cosine similarity greater than a preset topological isomorphism threshold as a set of topologically isomorphic nodes.

[0052] A topology matching strategy is used for fine screening to address the misjudgment of overlapping but electrically disconnected lines due to grade-separated intersections or lines erected on the same pole. The Weisfeller-Lyman graph kernel algorithm is introduced. The k-hop neighborhood subgraph of candidate nodes is extracted, and label aggregation is performed. By iteratively aggregating neighbor labels, a topological fingerprint vector containing local topological structure is generated. The cosine similarity of the fingerprint vector is calculated using the following expression: in, Cosine similarity between spatially overlapping candidate node pairs. The topological fingerprint vector is generated by iteratively applying the Weisfeller-Lyman algorithm to the first node in the paired nodes. The topological fingerprint vector is generated by iteratively applying the Weisfeller-Lyman algorithm to the second node in the paired nodes. This is the operation on the magnitude of a vector.

[0053] Only when the topological structure is highly isomorphic, that is Only node pairs that exceed the preset topological isomorphism threshold will be included in the topological isomorphic node set.

[0054] The parameter matching strategy specifically includes: Based on a set of topologically isomorphic nodes, a Jacobian sensitivity matrix is ​​constructed, which includes the partial derivatives of voltage amplitude with respect to active power and the partial derivatives of voltage amplitude with respect to reactive power.

[0055] The expression for the constructed Jacobian sensitivity matrix is ​​as follows: in, The Jacobian sensitivity matrix is... The sensitivity of active power to voltage phase angle, The sensitivity of active power to voltage amplitude, The sensitivity of reactive power to voltage phase angle. This refers to the sensitivity of reactive power to voltage amplitude.

[0056] The consistency of voltage and power responses of topologically isomorphic node pairs under a unit electrical disturbance is verified by using the Jacobi sensitivity matrix.

[0057] The topologically isomorphic node pairs that pass the sensitivity consistency verification will be output as entity mapping relationships, and virtual tie-line branch coupling based on entity mapping relationships will be established between multi-level power grid models.

[0058] Sensitivity verification is performed using a parameter matching strategy. Utilizing electrical physics principles, nodes at the same level will respond consistently to electrical disturbances. The sensitivity coefficients of topologically isomorphic node pairs are compared; if the deviation is within acceptable limits, a physical mapping relationship is confirmed, and virtual interconnections are established between multi-level power grid models.

[0059] In this embodiment of the application, the calculation of the spatial Euclidean distance between nodes in the multi-level power grid model in step S2 is specifically manifested as follows: The nodes whose spatial Euclidean distance is less than the preset spatial tolerance threshold are output as a set of spatially overlapping candidate nodes.

[0060] A coarse screening is performed using a boundary matching strategy, and the spatial Euclidean distance between cross-level node pairs is calculated using the following expression: in, For cross-level nodes to middle nodes With nodes The spatial Euclidean distance between them For nodes Three-dimensional spatial position coordinate vector in a unified Earth fixed coordinate reference system For nodes A three-dimensional spatial position coordinate vector in a unified Earth fixed coordinate reference system.

[0061] like For example, 50 meters, then the node pair Enter the set of candidate nodes with spatial overlap to solve the problem of proximity.

[0062] In an optional implementation, the spatial Euclidean distance between nodes in the multi-level power grid model in step S2 can also be calculated using a weighted Euclidean distance calculation method. Weight coefficients are introduced to different dimensions of the three-dimensional spatial coordinates of the nodes to construct a weighted distance metric. Smaller weights are assigned to the elevation direction (Z-axis) or coordinate dimensions with larger errors, while larger weights are assigned to the planar positions (X, Y axes). The distance calculation results are still used to compare with a preset spatial tolerance threshold to filter the set of candidate nodes with spatial overlap.

[0063] In another optional implementation, the spatial Euclidean distance between nodes in the multi-level power grid model in step S2 can also be calculated using the point-to-line minimum distance calculation method. The node-attached branches in the power grid model are abstracted into three-dimensional spatial line segments. The minimum spatial distance from the cross-level node to the corresponding line geometric center line or branch segment is calculated. When the minimum distance is less than the preset spatial tolerance threshold, the node pair is included in the spatial overlapping candidate node set.

[0064] In the embodiments of this application, step S2 calculates the spatial Euclidean distance between nodes of the multi-level power grid model to achieve the initial screening of cross-level power grid topology matching, and quickly converges the potential matching range from all network nodes to spatially adjacent local areas, thereby improving processing efficiency while ensuring recall rate.

[0065] In this embodiment of the application, the calculation of the cosine similarity of the fingerprint vectors in step B3 is specifically manifested as follows: The cosine similarity of fingerprint vectors is calculated using the following expression: in, Cosine similarity between spatially overlapping candidate node pairs. The topological fingerprint vector is generated by iteratively applying the Weisfeller-Lyman algorithm to the first node in the paired nodes. The topological fingerprint vector is generated by iteratively applying the Weisfeller-Lyman algorithm to the second node in the paired nodes. This is the operation on the magnitude of a vector.

[0066] In an optional implementation, the cosine similarity of the fingerprint vector in step B3 can also be calculated using a weighted cosine similarity calculation method. In the cosine similarity calculation, weight coefficients are introduced for different dimensions in the fingerprint vector, and higher weights are given to dimensions that reflect key branch connection relationships or higher-order neighborhood structures. The calculated similarity is used to compare with a preset topological isomorphism threshold.

[0067] In another optional implementation, the cosine similarity of the fingerprint vectors in step B3 can also be calculated using a similarity calculation method based on normalized dot product. The topological fingerprint vectors are normalized, and the dot product of the two normalized fingerprint vectors is directly calculated as a similarity index. When the similarity is greater than a preset threshold, they are determined to be a pair of topologically isomorphic nodes.

[0068] In this embodiment of the application, step B3 calculates the cosine similarity of the fingerprint vectors to identify power grid nodes that are close in spatial distance but have essential differences in local connection patterns. This eliminates false matches caused by common engineering scenarios such as line crossings and pole erection, and makes up for the shortcomings of simply relying on spatial distance for matching, thereby improving the accuracy and reliability of node association when automatically splicing cross-level power grid models.

[0069] It should be noted that spatial Euclidean distance is used to quickly filter candidate node pairs with similar locations, overcoming the blindness of the initial search. The Weisfeller-Lyman algorithm is introduced to extract and compare the local topological fingerprints of nodes, eliminating mismatched objects that overlap in location but are electrically isolated due to line crossings or being erected on the same pole. By verifying the consistency of the electrical disturbance response exhibited by the nodes in the Jacobian sensitivity matrix, the correctness of the node mapping relationship is confirmed from the physical law level. This improves the accuracy of automatic node mapping in cross-level, multi-source heterogeneous power grid models.

[0070] In this embodiment of the application, step S3 involves extracting the topology and node state vectors based on node mapping and branch coupling relationships, constructing a graph structure, calculating the composite loss function value, performing topology consistency verification and parameter coordination optimization, obtaining boundary correction parameters, updating the splicing section of the multi-level power grid model, logically stitching the multi-level power grid model, and outputting the multi-level power grid graph model, including the following steps C1-C3: C1: Based on the relationship between node mapping and branch coupling, extract the topology and node state vectors to construct the graph structure.

[0071] Nodes in the entity mapping relationship and adjacent nodes that have a physical connection with the nodes in the entity mapping relationship are considered as graph nodes.

[0072] The physical connection lines and virtual tie lines within the multi-level power grid model are used as graph edges.

[0073] The per-unit values ​​of voltage amplitude, phase angle, active power, and reactive power of the corresponding nodes in the multi-level power grid model are used as the node characteristic attributes.

[0074] C2: Based on the graph structure, calculate the composite loss function value, perform topology consistency verification and parameter coordination optimization, and obtain boundary correction parameters.

[0075] The local graph structure dataset of the spliced ​​region is input into the pre-trained elastic graph convolutional network. Through the multi-head attention mechanism in the elastic graph convolutional network, the feature aggregation weights of the virtual connection line are obtained. Based on the feature aggregation weights, the node features at both ends of the virtual connection line are weighted, aggregated, and nonlinearly transformed to output the boundary node prediction state vector. The boundary node prediction state vector integrates multi-source information from the main network, distribution network, microgrid, and user side.

[0076] C3: Based on the boundary correction parameters, update the splicing section of the multi-level power grid model, logically stitch the multi-level power grid model, and output the multi-level power grid diagram.

[0077] Based on the reference voltage and reference power, the boundary correction parameters are de-standardized and recalculated to obtain the parameter correction values ​​under physical dimensions. The parameter correction values ​​are then superimposed onto the original electrical measurement attribute fields of the multi-level power grid model at the splicing section to update the electrical parameters of the splicing section in a consistent manner.

[0078] By invoking entity mapping relationships, cross-level physical connection edges are constructed. These cross-level physical connection edges include physical connection edges connecting main grid model nodes and distribution network model nodes, physical connection edges connecting distribution network model nodes and microgrid model nodes, and physical connection edges connecting microgrid model nodes and user-side model nodes. The dispersed multi-level power grid model topology is logically merged through these cross-level physical connection edges to generate a single-connection physical topology architecture for the entire network.

[0079] The updated electrical parameters are loaded into the entire network single-channel physical topology architecture and associated with the hierarchical index of the graph model to output a globally consistent multi-level power grid graph model that supports drill-down across voltage levels and power flow calculation across the entire network.

[0080] In this embodiment of the application, the calculation of the composite loss function value in step S3 is specifically manifested as follows: Construct a composite loss function, the mathematical expression of which is: in, The value of the composite loss function. These are the preset weighting coefficients for the data fitting terms, used to adjust the penalty for deviations between predicted and measured values. They can be manually set. These are the weighting coefficients for preset physical constraints, used to adjust the penalty for violating the physical operating rules of the power grid. They can be manually set. Predict state vectors for boundary nodes. These are the original measurement values. This is the residual function of nodal power imbalance constructed based on Kirchhoff's laws. This is the preset L2 norm operator.

[0081] In an optional implementation, step S3 can further employ a loss function based on the sum of absolute values. This involves calculating the sum of the absolute values ​​of the differences between the predicted state and the original measurement across each dimension, as well as the sum of the absolute values ​​of the power imbalance components based on Kirchhoff's laws. These two sums are then multiplied by preset weighting coefficients and summed to obtain the total loss value.

[0082] In another alternative implementation, the calculation of the composite loss function value in step S3 can also employ an adaptive weighted loss function based on measurement accuracy. This assigns an independent weight coefficient to each specific measurement value, which is typically inversely proportional to the uncertainty of that measurement value. When calculating the data fitting term, the squared difference between each prediction and measurement is multiplied by its corresponding weight before summing. Physical constraint terms can remain unchanged or be handled similarly.

[0083] In the embodiments of this application, the composite loss function value is calculated in step S3. By weighted fusion of data fitting terms and physical constraint terms, the optimization result can closely approximate the field measurement value while strictly satisfying the power balance of Kirchhoff's law, thus achieving a balance between measurement uncertainty and the rigidity of physical laws.

[0084] In summary, this invention constructs a global data foundation with unified semantics and coordinates, and achieves efficient spatial organization of massive model elements through a hybrid index of GeoHash and R-Tree. First, candidate nodes are coarsely screened by spatial distance, and then structural similarity screening is performed using the topological fingerprint generated by the Weisfeller-Lehman algorithm. An electrical consistency verification is performed by introducing a Jacobi sensitivity matrix based on the power flow equation. Cross-level node mapping is established in complex scenarios, and a global graph structure containing virtual connecting lines is constructed. Finally, a composite loss function that integrates measurement data fitting terms and Kirchhoff's law physical constraint terms is used to automatically coordinate and optimize the splicing boundary parameters, solving the accuracy and consistency problems of automatic splicing of multi-source heterogeneous models.

[0085] Example 3 is an embodiment of the present invention, which provides a system for splicing multi-level power grid diagrams, including a construction module, a matching module, and a splicing and generation module. The building module is a collection of original multi-level power grid model files, a global ontology knowledge base, semantic encoding and coordinate reduction of the multi-level power grid model, and generation of a hierarchical index of the graph model; The matching module calls the hierarchical index of the graph model to retrieve spatially overlapping areas, calculates the spatial Euclidean distance between nodes of the multi-level power grid model, and uses a preset matching strategy to identify and establish node mapping and branch coupling relationships. The splicing and generation module extracts the topology and node state vectors based on the node mapping and branch coupling relationship, constructs the graph structure, calculates the composite loss function value, performs topology consistency verification and parameter coordination optimization, obtains boundary correction parameters, updates the splicing section of the multi-level power grid model, logically stitches the multi-level power grid model, and outputs the multi-level power grid graph model.

[0086] This embodiment also provides an electronic device applicable to a method for splicing multi-level power grid diagrams, comprising: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to implement the method for splicing multi-level power grid diagrams as proposed in the above embodiment.

[0087] This embodiment also provides a storage medium storing a computer program that, when executed by a processor, implements a method for splicing multi-level power grid diagrams as proposed in the above embodiments.

[0088] The storage medium proposed in this embodiment and the method for splicing a multi-level power grid diagram proposed in the above embodiments belong to the same inventive concept. Technical details not described in detail in this embodiment can be found in the above embodiments, and this embodiment has the same beneficial effects as the above embodiments.

[0089] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of the various embodiments of the present invention.

[0090] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for splicing multi-level power grid diagrams, characterized in that: include, Collect original multi-level power grid model files, construct a global ontology knowledge base, perform semantic encoding and coordinate reduction on the multi-level power grid model, and generate a hierarchical index of the graph model; The graph model hierarchical index is called to retrieve spatially overlapping areas, the spatial Euclidean distance between nodes of the multi-level power grid model is calculated, and a preset matching strategy is used to identify and establish the node mapping and branch coupling relationship. Based on the node mapping and branch coupling relationship, the topology and node state vectors are extracted, a graph structure is constructed, the composite loss function value is calculated, topology consistency verification and parameter coordination optimization are performed, boundary correction parameters are obtained, the splicing section of the multi-level power grid model is updated, the multi-level power grid model is logically stitched together, and a multi-level power grid graph model is output.

2. The method for splicing multi-level power grid diagrams as described in claim 1, characterized in that: The process involves collecting original multi-level power grid model files, constructing a global ontology knowledge base, performing semantic encoding and coordinate reduction on the multi-level power grid model, and generating a hierarchical index of the graph model. Collect original multi-level power grid model files and construct a global ontology knowledge base; Based on a global ontology knowledge base, the spatial coordinates of a multi-level power grid model are transformed through spatial reference system transformation to generate a globally coordinated element set; Based on the global coordinate element set, normalized coordinate data is extracted, a top-level sparse grid is constructed, and a hierarchical index of the graph model is generated.

3. The method for splicing multi-level power grid diagrams as described in claim 2, characterized in that: The process of calling the hierarchical index of the graph model to retrieve spatially overlapping regions, calculating the spatial Euclidean distance between nodes in the multi-level power grid model, and identifying and establishing node mapping and branch coupling relationships using a preset matching strategy includes... Call the hierarchical index of the graph model to identify the spatial overlap area of ​​the multi-level power grid model in the Earth fixed coordinate reference system; Based on the spatially overlapping region, the spatial Euclidean distance between nodes in the multi-level power grid model is calculated, and the set of candidate spatially overlapping nodes is screened. Based on the spatially overlapping candidate node set, fingerprint vectors are generated and the cosine similarity of the fingerprint vectors is calculated. The topologically isomorphic node set is then selected, the Jacobian sensitivity matrix is ​​constructed, and the node mapping and branch coupling relationship is established.

4. The method for splicing multi-level power grid diagrams as described in claim 3, characterized in that: The process involves extracting topology and node state vectors based on node mapping and branch coupling, constructing a graph structure, calculating composite loss function values, performing topology consistency verification and parameter coordination optimization, deriving boundary correction parameters, updating the splicing sections of the multi-level power grid model, logically stitching the multi-level power grid model, and outputting a multi-level power grid graph model. Based on the node mapping and branch coupling relationship, the topology and node state vectors are extracted to construct a graph structure; Based on the graph structure, the composite loss function value is calculated, topology consistency verification and parameter coordination optimization are performed, and boundary correction parameters are obtained. Based on the boundary correction parameters, the splicing section of the multi-level power grid model is updated, the multi-level power grid model is logically stitched together, and the multi-level power grid diagram is output.

5. The method for splicing multi-level power grid diagrams as described in claim 4, characterized in that: The transformation of the spatial coordinates of the multi-level power grid model through spatial reference frame transformation includes... The spatial transformation is performed using the Bursa seven-parameter coordinate transformation model, expressed as follows: in, The coordinates are the transformed coordinates. The translation factor is... For rotation factor, As a scale factor, These are the coordinates before the transformation.

6. The method for splicing multi-level power grid diagrams as described in claim 5, characterized in that: The calculation of the cosine similarity of the fingerprint vectors includes, The cosine similarity of fingerprint vectors is calculated using the following expression: in, Cosine similarity between spatially overlapping candidate node pairs. The topological fingerprint vector is generated by iteratively applying the Weisfeller-Lyman algorithm to the first node in the paired nodes. The topological fingerprint vector is generated by iteratively applying the Weisfeller-Lyman algorithm to the second node in the paired nodes. This is the operation on the magnitude of a vector.

7. The method for splicing multi-level power grid diagrams as described in claim 6, characterized in that: The calculation of the composite loss function value includes, Construct a composite loss function, the mathematical expression of which is: in, The value of the composite loss function. These are the pre-defined weighting coefficients for the data fitting terms, used to adjust the penalty for deviations between predicted and measured values. These are the weighting coefficients for preset physical constraints, used to adjust the penalty for violating the physical operating rules of the power grid. Predict state vectors for boundary nodes. These are the original measurement values. This is the residual function of nodal power imbalance constructed based on Kirchhoff's laws. This is the preset L2 norm operator.

8. A system for splicing multi-level power grid diagrams, using the method for splicing multi-level power grid diagrams as described in any one of claims 1 to 7, characterized in that, This includes a building module, a matching module, and a splicing and generation module. The construction module is to collect the original multi-level power grid model files, build a global ontology knowledge base, perform semantic encoding and coordinate reduction on the multi-level power grid model, and generate a hierarchical index of the graph model; The matching module calls the hierarchical index of the graph model to retrieve spatially overlapping areas, calculates the spatial Euclidean distance between nodes of the multi-level power grid model, and uses a preset matching strategy to identify and establish the node mapping and branch coupling relationship. The splicing and generation module extracts the topology and node state vectors based on the node mapping and branch coupling relationship, constructs the graph structure, calculates the composite loss function value, performs topology consistency verification and parameter coordination optimization, obtains boundary correction parameters, updates the splicing section of the multi-level power grid model, logically stitches the multi-level power grid model, and outputs the multi-level power grid graph model.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method for splicing multi-level power grid diagrams according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for splicing multi-level power grid diagrams according to any one of claims 1 to 7.

Images