A power grid digital twin modeling method and system based on multi-paradigm fusion

Abstract

This invention discloses a multi-paradigm fusion-based digital twin modeling method and system for power grids, belonging to the field of power system digital twin technology. The method includes: standardizing multi-source heterogeneous power grid data based on the SG-CIM model to construct a high-quality modeling data foundation; then accurately reconstructing the geometric shape and spatial layout of power grid equipment through GIM file parsing and 3D modeling technology; subsequently embedding the physical parameters and operating rules of the power grid equipment to construct a realistic mechanism model; and finally, combining historical and real-time measurement data, employing a deep learning algorithm with physical constraints to complete data-driven modeling. This invention, through the fusion of geometry, information mechanism, and data-driven multi-paradigm approaches, solves the problems of difficult reuse of heterogeneous data and insufficient accuracy of a single paradigm in traditional modeling, significantly improving the fidelity and adaptability of the power grid digital twin model, and providing reliable digital twin support for power grid equipment monitoring, intelligent scheduling, and operation and maintenance optimization.

Description

Technical Field

[0001] This invention relates to the field of digital twin technology for power systems, and more specifically, to a method and system for digital twin modeling of power grids based on multi-paradigm fusion. Background Technology

[0002] Digital twin technology, as a core means to achieve efficient interaction between the physical power grid and the virtual space, is of great significance for ensuring the stable operation of the power grid. By constructing a high-fidelity virtual mapping, this technology can synchronize the status of power grid equipment in real time, accurately predict fault risks, and optimize dispatch strategies. It significantly improves the system's anti-interference capability and adaptability under conditions of high proportion of new energy access, promotes the transformation of the power grid from "passive operation and maintenance" to "proactive prevention and control," and thus enhances the reliability and resilience of power supply.

[0003] The model is the foundation for accurate mapping and intelligent decision-making in digital twins, and its quality directly determines the effectiveness of the digital twin system. An ideal power grid digital twin model needs to have multi-physics coupling description capabilities, integrate multi-dimensional characteristics such as electrical, thermal, and mechanical aspects, and achieve self-evolution and lifelong learning through dynamic data-driven processes. The model must balance mechanistic accuracy and data adaptability to support the analytical needs of complex scenarios such as transient simulation and fault prediction, thereby providing core basis for power grid state perception, risk assessment, and control optimization.

[0004] However, existing digital twin modeling methods still face multiple challenges. At the data level, the standardization of multi-source heterogeneous data is difficult, with issues such as inconsistent acquisition standards, insufficient transmission stability, and difficulties in semantic alignment, leading to challenges in ensuring the quality of model input. At the model construction level, single modeling paradigms have significant limitations: pure mechanistic models rely on precise physical equations, making it difficult to describe complex nonlinear behaviors and consuming large computational resources, increasing the modeling data cycle; pure data-driven models suffer from poor interpretability due to insufficient training samples, are prone to physical paradoxes, and have weak generalization ability. Furthermore, the lack of physical constraints in pure data-driven models makes them prone to misjudgment or omission in equipment fault diagnosis, resulting in a high misjudgment rate. In addition, multi-scale model coupling technology is still immature, and joint simulation across levels and physical fields faces efficiency bottlenecks, failing to achieve real-time synchronization from equipment components to the system level, making it difficult to meet real-time requirements. Lagging model updates further weaken its applicability in dynamic scenarios.

[0005] Therefore, researching a digital twin modeling method and system for power grids based on the fusion of multiple paradigms such as geometric modeling, information mechanism modeling, and data-driven modeling, and achieving dynamic consistency and high-fidelity mapping between the model and the physical power grid by constructing a standardized data base and breaking through key technologies such as hybrid modeling and lightweight real-time computing, has become an urgent need to overcome existing technological bottlenecks and promote the evolution of new power systems towards a higher level of intelligence.

[0006] Prior art document 1 (CN117634995A) discloses a modeling method and system for a digital twin platform for power grid operations, which achieves rapid model updates by predicting the state description matrix using a Markov network. Prior art document 2 (CN120337457A) discloses a dynamic ontology modeling method and system for power grids for digital twins, which constructs a dynamic model based on ontology and graph theory algorithms and supports visual interaction.

[0007] However, existing technology document 1 focuses on model update efficiency and simplifies calculations through state matrix prediction, but it does not address the accurate reconstruction of the geometric shape of power grid equipment and lacks multi-physics mechanism modeling support. Relying solely on data-driven state prediction is prone to results that contradict physical laws, resulting in insufficient model fidelity. Existing technology document 2 emphasizes ontology framework and dynamic interaction functions. Although it involves data preprocessing and model verification, it does not establish a unified data standard, resulting in poor reusability of heterogeneous data. Furthermore, it does not integrate deep learning algorithms to optimize the model's nonlinear representation capabilities, making it difficult to meet the high-precision modeling requirements under complex power grid operating conditions. None of the existing technologies propose a multi-paradigm fusion architecture of "geometric modeling - information mechanism modeling - data-driven modeling," nor do they achieve the technical integration of full-process data standardization based on SG-CIM combined with accurate geometric reconstruction of GIM files and deep learning modeling under physical constraints. They cannot simultaneously solve the three core problems of data reuse, model fidelity, and real-time adaptation in power grid digital twin modeling.

[0008] To address the aforementioned issues, there is an urgent need for a digital twin modeling method and system for power grids based on multi-paradigm fusion. Summary of the Invention

[0009] To address the shortcomings of existing technologies, this invention provides a multi-paradigm fusion-based digital twin modeling method and system for power grids. The method includes: standardizing multi-source heterogeneous power grid data; establishing unified data specifications based on the SG-CIM (State Grid Common Information Model) model to construct a high-quality modeling data foundation; conducting geometric modeling by combining GIM (Grid Information Model) file parsing with 3D modeling technology to accurately reconstruct the geometric shape, spatial layout, and component relationships of power grid equipment; implementing information mechanism modeling by embedding physical parameters and operating rules of power grid equipment to construct a mechanism model that conforms to actual operating laws; and performing data-driven modeling by using deep learning algorithms to mine high-dimensional data mapping features based on historical and real-time measurement data to optimize the model's nonlinear representation capabilities. This invention, through a multi-paradigm fusion modeling method of "geometry-information mechanism-data-driven," solves the problems of difficult reuse of heterogeneous data and insufficient accuracy of a single paradigm in traditional modeling, significantly improving the fidelity and adaptability of the power grid digital twin model, and providing reliable digital twin support for power grid equipment monitoring, intelligent scheduling, and operation and maintenance optimization.

[0010] The present invention adopts the following technical solution: This invention protects a digital twin modeling method for power grids based on multi-paradigm fusion, comprising the following steps: S1. Standardize the multi-source heterogeneous data of the power grid, establish unified data specifications based on the SG-CIM model, and build a high-quality modeling data base; S2. Conduct geometric modeling, and accurately restore the geometric shape, spatial layout and component relationships of power grid equipment by combining GIM file parsing and 3D modeling technology; S3. Implement information mechanism modeling, embed the physical parameters and operating rules of power grid equipment, and construct a digital twin mechanism model of power grid equipment that conforms to actual operating rules; S4. Conduct data-driven modeling, based on historical and real-time measurement data, and use deep learning algorithms to mine high-dimensional data mapping features to optimize the nonlinear representation capability of the digital twin mechanism model.

[0011] Furthermore, S1, standardizing the multi-source heterogeneous data of the power grid, establishing unified data specifications based on the SG-CIM model, and constructing a high-quality modeling data foundation includes: S11. Initiate the multi-source data automatic aggregation process to receive structured, semi-structured, and unstructured power grid data from different external systems; S12. Based on the SG-CIM model, extract the power grid equipment classes, core attributes and cross-equipment relationships, establish a precise mapping between multi-source heterogeneous data and model elements, clarify the data fields, formats, encoding and transmission protocol specifications, and unify the data format to a structured format that conforms to the IEC61970 standard. S13. Perform preprocessing operations on the aggregated heterogeneous data, use intelligent cleaning algorithms to remove duplicate data and correct erroneous data, fill missing values ​​through interpolation, eliminate data noise using smoothing techniques, and use semantic web technology to achieve semantic alignment of heterogeneous data and eliminate semantic differences between different data sources. S14. Set four quality indicators for data integrity, accuracy, consistency, and timeliness. Use a dual-drive mode of expert review and intelligent verification to conduct full-cycle quantitative evaluation of the preprocessed data. After evaluation, a standardized dataset is formed to build the data base for modeling.

[0012] Furthermore, S11, initiating the multi-source data automatic aggregation process and receiving structured, semi-structured, and unstructured power grid data from different external systems, includes: Obtain equipment diagram data from the power grid resource business platform through services; Obtain electrical and non-electrical measurement data of the equipment from the real-time measurement center through services; Obtain geographic environment and meteorological monitoring data from enterprise-level meteorological centers through services; Obtain defect, fault, power outage, and maintenance record data from the PMS3.0 business system; Furthermore, S12, based on the SG-CIM model, extracts power grid equipment classes, core attributes, and cross-equipment relationships, establishes a precise mapping between multi-source heterogeneous data and model elements, clarifies data fields, formats, encoding, and transmission protocol specifications, and unifies the data format to a structured format conforming to the IEC 61970 standard, including: Based on the core architecture of the SG-CIM model, we extract the core attributes of various power grid equipment categories, including switches, transformers, capacitors, and reactors, and sort out the cross-equipment relationships. Establish a mapping relationship between heterogeneous data and SG-CIM model elements; Establish a unified data standard system; All types of heterogeneous structured data are uniformly converted into the CIM XML structured format that conforms to the IEC61970 standard.

[0013] Furthermore, in step S2, geometric modeling is performed. By combining GIM file parsing with 3D modeling technology, the geometric shape, spatial layout, and component relationships of the power grid equipment are accurately restored. S21. Parse the GIM standard file and extract the attribute parameters, hierarchical structure relationships, and geometric parameters of the power grid equipment. S22. Based on the extracted primitive geometric parameters, the model is reconstructed and assembled in the 3D model development platform to generate the basic 3D white model of the power grid equipment. S23. Based on the hierarchical structure relationship, perform model assembly and spatial registration, establish the association mapping between components, and form a three-dimensional model with accurate geometric shape and spatial layout.

[0014] Furthermore, step S21, parsing the GIM standard file and extracting the attribute parameters, hierarchical structure relationships, and geometric parameters of the power grid equipment, includes: Load and parse the GIM compressed file package, establish a complete five-level engineering hierarchy tree from system level to component level, generate a GIM structure tree and device list described in JSON format, and clarify the parent-child dependency and spatial membership relationship between devices; Parse the device property file, extract the Chinese and English property names and values ​​line by line, and map the extracted properties to the SG-CIM standard; The primitives are extracted and transformed step by step from the local coordinate system to the global engineering coordinate system, correcting coordinate offsets caused by projection transformation or acquisition accuracy; the final output is a primitive parameter set containing geometric type, vertex coordinates, line width, and transformation matrix information.

[0015] Further, step S22, based on the extracted primitive geometric parameters, involves model reconstruction and assembly on a 3D model development platform to generate a basic 3D white model of the power grid equipment, including: Input the geometric parameters of the primitives into the 3D model development platform to generate the corresponding 3D geometry. The geometry is refined and assembled into component-level 3D white models; Based on the hierarchical structure, the assembled component-level models are placed in the correct spatial positions determined by the global coordinate system of the project, establishing parent-child relationships between components, and finally generating a three-dimensional white model of the overall foundation of the power grid facility.

[0016] Further, step S23, assembling and spatially registering the model according to the hierarchical structure relationship, establishing the association mapping between components, and forming a three-dimensional model with accurate geometric shape and spatial layout, includes: In a 3D scene, the generated component-level 3D white model is spatially positioned to form the overall 3D structural framework of the substation or line corridor. Establish relationships between components and add logical connectors; establish parent-child relationships between models; and add logical connectors representing the connection relationships of conductors and busbars based on the power grid topology. The starting and ending points of the lines in the model are mapped and associated with the nodes in the physical power grid to generate a 3D model that interacts with the physical power grid in real time.

[0017] Furthermore, S3, implementing information mechanism modeling, embedding physical parameters and operating rules of power grid equipment, and constructing a digital twin mechanism model of power grid equipment that conforms to actual operating laws includes: S31. Based on the physical characteristics of power grid equipment, embed the mechanism equations of multi-physics field behavior, and integrate electrical and non-electric measurement data to form a multi-scale physical parameter database for the equipment; S32. Based on the internal operating rules of the equipment and the response behavior to external disturbances, construct a differential algebraic equation model or finite state machine model of the equipment in transient and steady states to characterize the dynamic functional behavior of the equipment. S33. Assemble and couple the multiphysics model, behavior model and rule model of a single device, and form a digital twin full-space model that supports real-time simulation and offline deduction through model verification, validation and confirmation mechanisms.

[0018] Furthermore, in step S31, based on the physical characteristics of the power grid equipment, the mechanism equations of multi-physics behavior are embedded, and electrical and non-electrical measurement data are fused to form a multi-scale physical parameter database for the equipment, including: The core mechanism equations of the multiphysics field of power grid equipment are selected and spatiotemporally discretized. Integrate static parameters and dynamic measurement data, and correct model parameters through parameter identification and data assimilation; The physical parameters are organized and stored in a structured manner according to their physical meaning and spatiotemporal scale, forming a multi-scale physical parameter database.

[0019] Furthermore, in step S32, based on the internal operating rules of the equipment and the response behavior to external disturbances, a differential-algebraic equation model or finite state machine model of the equipment in transient and steady states is constructed to characterize the dynamic functional behavior of the equipment, including: For the dynamic characteristics of power grid equipment, for continuously changing physical quantities, a differential-algebraic equation model describing their transient and steady-state behavior is constructed based on Kirchhoff's laws and the law of electromagnetic induction; for discrete logical operation behavior, a finite state machine model based on state, event, and action as core elements is constructed. Set initial conditions, boundary conditions, and equation coefficients related to the physical parameters of the device for the differential-algebraic equation model; define all possible state sets, event sets that trigger state transitions, and action sets to be executed during state transitions for the finite state machine model; and encode the internal operating rules of the device into state transition logic. An interactive interface is established between the differential algebraic equation model and the finite state machine model, so that the state switching signal output by the finite state machine model can be used as the triggering condition for changes in the structure or parameters of the differential algebraic equation model. At the same time, the physical quantities calculated by the differential algebraic equation model are used as the event inputs of the finite state machine model, thereby realizing an integrated representation of the dynamic behavior of the device.

[0020] Furthermore, in step S33, the multiphysics model, behavior model, and rule model of a single device are assembled and coupled, and a digital twin full-space model supporting real-time simulation and offline derivation is formed through model verification, validation, and confirmation mechanisms. Define a unified data exchange format and calling interface to enable data interaction and logical association between different models; The accuracy of the model is quantified and its reliability is ensured through model verification and validation. The assembled and coupled models form a digital twin full-space model that supports real-time simulation and offline deduction, reflecting the complete information and dynamic characteristics of the power grid from equipment components to the system level.

[0021] Furthermore, in step S4, data-driven modeling is performed. Based on historical and real-time measurement data, deep learning algorithms are used to mine high-dimensional data mapping features and optimize the model's nonlinear representation capabilities, including: S41. Integrate historical operating data and real-time measurement data of power grid equipment and preprocess them to construct a high-dimensional feature space. Then, use the chi-square test to screen the feature subset that is strongly correlated with the equipment status to form a training sample set. S42. Design a deep learning architecture, input preprocessed feature data, optimize model parameters through backpropagation algorithm; introduce physical constraint terms into the loss function to improve model performance; S43. Evaluate model metrics through cross-validation, deploy an online learning mechanism to update model parameters, and generate a high-fidelity data-driven model.

[0022] In another aspect, this invention protects a power grid digital twin modeling system based on multi-paradigm fusion, the system comprising: The data governance and standardization module is used to aggregate, clean and standardize multi-source heterogeneous data of the power grid, establish unified data specifications based on the SG-CIM model and build a high-quality modeling data foundation. The 3D geometric modeling engine module is used to perform geometric modeling, accurately restoring the geometric shape, spatial layout, and component relationships of power grid equipment through GIM file parsing and 3D reconstruction technology. The multiphysics mechanism modeling module is used to implement information mechanism modeling, embed the physical parameters and operating rules of power grid equipment, and construct a digital twin mechanism model of power grid equipment that conforms to the actual operating law. The multi-paradigm fusion and model management module is used for data-driven modeling. Based on historical and real-time measurement data, it uses deep learning algorithms to mine high-dimensional data mapping features and optimize the model's nonlinear representation capabilities.

[0023] This invention also protects an electronic device, including a processor and a storage medium; The storage medium is used to store instructions; The processor is configured to operate according to the instructions to execute the steps of the above-described method for digital twin modeling of power grids based on multi-paradigm fusion.

[0024] The present invention also protects a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described multi-paradigm fusion-based digital twin modeling method for power grids.

[0025] The beneficial effects of this invention are that, compared with the prior art, 1. It solves the problem of reusing multi-source heterogeneous data from the power grid. By constructing a unified data standard through the SG-CIM model, and combining intelligent cleaning, semantic alignment and dual-round quality assessment, it achieves standardized data processing, which greatly improves data integrity, accuracy and reusability, shortens the data preparation cycle for modeling, and lays a high-quality data foundation for digital twin modeling.

[0026] 2. Breaking through the accuracy limitations of a single modeling paradigm, it pioneered a multi-paradigm fusion architecture of "geometry-information mechanism-data-driven". It not only accurately restores the geometric shape and topological relationship of power grid equipment through GIM file parsing and 3D modeling, but also embeds multi-physics mechanism equations to ensure the interpretability of the model. At the same time, it combines deep learning with physical constraints to optimize nonlinear representation capabilities, thereby achieving a dual improvement in model accuracy and adaptability.

[0027] 3. Achieve high-fidelity dynamic mapping of the power grid digital twin model. Through multi-model coupling integration and verification mechanism, ensure the consistency of information from equipment components to the system level. It also supports online learning and dynamic parameter updates, adapting to complex operating conditions such as equipment aging, meteorological disturbances, and new energy grid connection, meeting the dual needs of real-time simulation and offline simulation.

[0028] 4. Enhance the intelligence level of power grid operation and maintenance and dispatch. High-fidelity digital twin models can provide reliable support for power grid equipment status monitoring, fault diagnosis, intelligent dispatch and operation and maintenance optimization, effectively reduce the fault misjudgment rate, improve dispatch decision efficiency, reduce power grid operation and maintenance costs, and promote the transformation of the power grid from "passive operation and maintenance" to "proactive prevention and control".

[0029] 5. The model has good versatility and scalability. It is based on common standards such as IEC61970 and IEC61850 to build data and communication specifications, and is compatible with the modeling of various power grid facilities such as substations and transmission line corridors. The modules are coupled through a unified interface, which facilitates subsequent functional expansion and cross-system integration, and adapts to the development needs of new power systems. Attached Figure Description

[0030] Figure 1A schematic diagram of a power grid digital twin modeling method based on multi-paradigm fusion provided by the present invention; Figure 2 This invention provides a schematic diagram of a power grid digital twin modeling system based on multi-paradigm fusion. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The embodiments described in this application are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, other embodiments obtained by those skilled in the art without creative effort are all within the protection scope of this invention.

[0032] Example 1: like Figure 1 As shown, a digital twin modeling method and system for power grids based on multi-paradigm fusion includes the following steps: S1. Standardize the multi-source heterogeneous data of the power grid, establish unified data specifications based on the SG-CIM model, and build a high-quality modeling data base; S2. Conduct geometric modeling, and accurately restore the geometric shape, spatial layout and component relationships of power grid equipment by combining GIM file parsing and 3D modeling technology; S3. Implement information mechanism modeling, embed the physical parameters and operating rules of power grid equipment, and construct a mechanism model that conforms to the actual operating law; S4. Conduct data-driven modeling, based on historical and real-time measurement data, and use deep learning algorithms to mine high-dimensional data mapping features and optimize the model's nonlinear representation capabilities.

[0033] Furthermore, S1, standardizing the multi-source heterogeneous data of the power grid, establishing unified data specifications based on the SG-CIM model, and constructing a high-quality modeling data foundation includes: S11. Initiate the multi-source data automatic aggregation process to receive structured, semi-structured, and unstructured power grid data from different external systems.

[0034] Through multi-protocol adaptation interfaces, GIM model data of equipment such as substation transformers and circuit breakers are obtained from the power grid resource business platform; electrical / non-electrical measurement data such as voltage, current, temperature, and humidity are obtained from the real-time measurement center (sampling frequency 50Hz); meteorological data such as wind speed, precipitation, and temperature in the substation area are obtained from the enterprise-level meteorological center (sampling frequency 10min / time); and equipment defect, fault, and maintenance record data for the past 5 years are obtained from the PMS3.0 system, thus aggregating structured data, semi-structured data, and unstructured data.

[0035] S12. Based on the SG-CIM model, extract the power grid equipment classes, core attributes and cross-equipment relationships, establish a precise mapping between multi-source heterogeneous data and model elements, clarify the data fields, formats, encoding and transmission protocol specifications, and unify the data format to a structured format that conforms to the IEC61970 standard.

[0036] Based on the SG-CIM model, core equipment classes (transformers, circuit breakers, busbars, disconnectors, etc.) of substations were extracted, and 28 core attributes such as rated voltage, rated current, and topological location were selected. Three types of cross-equipment relationships, such as node-branch connections, were identified. A unified data specification was established, with field types using String / Int / Float, character encoding using UTF-8, numerical precision retaining 4 decimal places, and transmission protocol following the IEC61850-9-2 standard. All heterogeneous data was converted into CIM XML format.

[0037] S13. Perform preprocessing operations on the aggregated heterogeneous data, use intelligent cleaning algorithms to remove duplicate data and correct erroneous data, fill missing values ​​through interpolation, eliminate data noise using smoothing techniques, and use semantic web technology to achieve semantic alignment of heterogeneous data and eliminate semantic differences between different data sources. An intelligent cleaning algorithm based on isolated forests was used to remove 12,000 duplicate data entries and correct 8,000 erroneous data entries. Missing values ​​in the measurement data were filled using linear interpolation, with a filling rate of 99.8%. Data noise was eliminated using the moving average method, with a smoothing window size of 5. Semantic alignment of heterogeneous data was achieved through ontology-based semantic web technology, eliminating 16 semantic differences between data sources.

[0038] S14. Four quality indicators are set for data integrity, accuracy, consistency, and timeliness. A dual-drive model of "expert review + intelligent verification" is adopted to conduct a full-cycle quantitative evaluation of the preprocessed data. After evaluation, a standardized dataset is formed to build a high-quality modeling data foundation to support subsequent modeling. The four indicators are set as follows: data integrity ≥99%, accuracy ≥99.5%, consistency ≥99%, and timeliness ≤5s. First, an intelligent verification program written in Python performs automated evaluation, followed by manual review by four power system modeling experts. After passing the evaluation, a standardized dataset is formed to build the modeling data foundation.

[0039] Specifically, S11, initiating the multi-source data automatic aggregation process and receiving structured, semi-structured, and unstructured power grid data from different external systems, includes: S111. Obtain equipment diagram data from the power grid resource business platform through services; S112. Obtain electrical and non-electrical measurement data of the equipment from the real-time measurement center through services; S113. Obtain geographic environment and meteorological monitoring data from enterprise-level meteorological centers through services; S114. Obtain defect, fault, power outage and maintenance record data from the PMS3.0 business system; Furthermore, S12, based on the SG-CIM model, extracts power grid equipment classes, core attributes, and cross-equipment relationships, establishes a precise mapping between multi-source heterogeneous data and model elements, clarifies data fields, formats, encoding, and transmission protocol specifications, and unifies the data format to a structured format conforming to the IEC 61970 standard, including: S121. Based on the core architecture of the SG-CIM model, extract power grid equipment types covering switches, transformers, capacitors, reactors, busbars, circuit breakers, etc., filter core attributes such as rated voltage, rated current, rated power, topology location, asset number, etc., and sort out cross-equipment relationships such as node-branch connection, energy flow transfer, control signal interaction, etc. S122. For the aggregated multi-source heterogeneous data, classify the data according to its source and purpose, establish a mapping relationship between each type of data and the equipment class, core attributes and related relationships in the SG-CIM model, clarify the data semantic matching rules, and ensure the accurate association between heterogeneous data and model elements. S123. Establish a unified data specification system, in which the data field definition specification clearly defines the field name, data type, value range and semantic description; the data format standard clearly defines the character encoding, numerical precision and file storage format; the encoding rules adopt a unified device encoding, attribute encoding and association relationship encoding scheme; and the transmission protocol is required to comply with the communication protocol specification of IEC61850 standard. S124. Standardize and convert various heterogeneous data through format conversion tools, and uniformly convert unstructured text data, semi-structured tabular data and heterogeneous structured data into CIM XML structured format that conforms to IEC61970 standard to ensure data interactivity and reusability.

[0040] S2 involves conducting geometric modeling, combining GIM file parsing with 3D modeling technology to accurately reconstruct the geometric shape, spatial layout, and component relationships of power grid equipment, including: S21. Parse the GIM standard file and extract the attribute parameters, hierarchical structure relationships, and geometric parameters of the power grid equipment.

[0041] Load the substation GIM compressed package (zip format, 2.5GB in size), and after decompression, identify the project type as a 220kV substation; parse the engineering / physical / combination / geometric model directory, establish a five-level engineering hierarchy tree of system-bay-equipment-component-element, and generate a GIM structure tree in JSON format; extract equipment attribute parameters, establish attribute indexes through hash tables, merge 32 redundant attribute rows, delete 18 invalid attribute items, and map the attributes to the SG-CIM standard; extract triangular patch files from the geometric model directory, extract point coordinates and wireframe information, and combine the homogeneous transformation matrix to transform the elements from the local coordinate system to the global engineering coordinate system, correcting the maximum coordinate offset of 0.03m.

[0042] S22. Based on the extracted primitive geometric parameters, the model is reconstructed and assembled in a 3D model development platform to generate a basic 3D white model of the power grid equipment. Unity3D is selected as the 3D model development platform. The analyzed primitive parameters are input into the platform's geometry engine to reconstruct basic primitives such as points, lines, and surfaces. Components such as transformer tanks and circuit breaker arc-extinguishing chambers are subjected to surface smoothing (smoothness parameter set to 0.8) and edge chamfering (chamfer radius 5mm) to assemble them into component-level 3D white models.

[0043] S23. Based on the hierarchical structure, assemble and spatially register the model, establish the association mapping between components, and form a three-dimensional model with accurate geometric shape and spatial layout. According to the five-level hierarchical relationship, place the component-level model at the corresponding position in the global coordinate system and establish parent-child association relationship; add logical connectors such as conductors and busbars, and map the model nodes to the physical entity nodes of the substation one by one, finally generating a three-dimensional geometric model of the 220kV substation, with the spatial position error controlled within 3cm.

[0044] Furthermore, step S21, parsing the GIM standard file and extracting the attribute parameters, hierarchical structure relationships, and geometric parameters of the power grid equipment, includes: S211. Load the GIM compressed file package, decompress it and read the file header information to identify the project type; according to the directory structure defined by the GIM specification, traverse and parse the files in the project model directory, physical model directory, combined model directory and geometric model directory in sequence; establish a complete five-level project hierarchy tree from system level to component level by parsing the project root file; generate a GIM structure tree and device list described in JSON format; and clarify the parent-child dependency and spatial membership relationship between devices. S212. For each device object, parse its associated attribute file, extract the Chinese and English attribute names and values ​​line by line, and use a hash table to build an attribute index for fast retrieval; perform word segmentation and semantic parsing on the Chinese attribute names, and calculate semantic similarity through string matching or word vector models, merge redundant attribute lines, and delete attribute items with empty values ​​or invalid formats to form a structured device attribute dataset; map the extracted attributes to the SG-CIM standard to ensure that the attribute naming and value range conform to the unified specification; S213. Read parametric primitive files or triangular patch files from the geometric model directory, and extract point coordinate information and wireframe information; combine the homogeneous transformation matrix (pose matrix) extracted from each level of files, and transform the primitives from the local coordinate system to the global engineering coordinate system step by step, correcting the coordinate offset caused by projection transformation or acquisition accuracy; finally output a primitive parameter set containing information such as geometric type, vertex coordinates, line width, and transformation matrix, providing an accurate geometric data foundation for 3D model reconstruction.

[0045] S22, based on the extracted primitive geometric parameters, involves model reconstruction and assembly on a 3D model development platform to generate a basic 3D white model of the power grid equipment, including: S221. Input the geometric parameters of basic primitives such as points, lines, and surfaces obtained from the analysis into the 3D model development platform. Through the platform's built-in geometry engine, the basic primitives are reconstructed one by one in 3D according to their type, coordinates, and topological relationships to generate the corresponding 3D geometry. S222. Refine the reconstructed basic geometry, including surface smoothing, edge chamfering, and assembling multiple basic geometries according to the correct spatial position and constraint relationship based on the assembly relationship attributes in the GIM file, to form a complete component-level three-dimensional white model of power grid equipment such as transformers and circuit breakers. S223. Based on the hierarchical structure, the assembled component-level model is placed in the correct spatial position determined by the global coordinate system of the project, the parent-child relationship between components is established, and finally the overall basic three-dimensional white model of power grid facilities such as substations or line corridors with accurate geometric shape and spatial layout is generated.

[0046] Step S23, which involves assembling and spatially registering the model based on the hierarchical structure, establishing the association mapping between components, and forming a three-dimensional model with accurate geometric shape and spatial layout, includes: S231. Based on the equipment hierarchy structure and global coordinate transformation matrix of each component parsed from the GIM file, the generated component-level 3D white models are placed sequentially in the correct spatial position determined by the global coordinate system of the project in the 3D scene to initially form the overall 3D structural framework of the substation or line corridor. S232. Based on the connection relationships described in the GIM file, establish parent-child relationships between the already in place component models in the 3D model development platform, and add logical connectors to represent the connection relationships of conductors, buses, etc. according to the power grid topology to ensure that the spatial position of the model and the internal electrical connection logic are consistent with the actual power grid. S233. Establish a one-to-one mapping relationship between each node in the assembled 3D model and the corresponding node in the physical entity of the power grid. At the same time, map and associate the starting point and ending point of the line in the model with the node in the physical power grid. Finally, generate a 3D model that has both accurate geometric shape and spatial layout and fully retains the power grid topology connection information, and can interact with the physical power grid in real time.

[0047] Furthermore, S3, implementing information mechanism modeling, embedding the physical parameters and operating rules of power grid equipment, and constructing a mechanism model that conforms to actual operating laws includes: S31. Based on the physical characteristics of power grid equipment, embed the mechanism equations describing its electrical, mechanical, thermodynamic and other multi-physical field behaviors, and integrate electrical and non-electrical measurement data to form a multi-scale physical parameter database of the equipment. S32. Based on the internal operating rules of the equipment and the response behavior to external disturbances, construct a differential algebraic equation model or finite state machine model of the equipment in transient and steady states to characterize its energy flow trajectory, information flow transmission relationship and dynamic functional behavior. For continuous physical quantities such as voltage and current, a differential-algebraic equation model is constructed based on Kirchhoff's laws. The initial condition is a transformer no-load voltage of 220kV, and the boundary condition is an ambient temperature of 25℃. For discrete logic operations such as circuit breaker opening / closing, a finite state machine model is constructed, defining four states: "closing-opening-energy storage-fault". The triggering events are eight types, including overcurrent and voltage exceeding limits, and the state transition actions are switching operations and fault alarms. An interaction interface between the two models is established. The state switching signals of the finite state machine serve as the parameter triggering conditions of the differential-algebraic equation model, and the calculation results of the differential-algebraic equation serve as the event inputs of the finite state machine.

[0048] S33. Assemble and couple the multiphysics model, behavioral model, and rule model of individual devices. Ensure consistency with the actual physical object through model verification, validation, and confirmation mechanisms, ultimately forming a digital twin full-space model that supports real-time simulation and offline derivation. Assemble single device models such as transformers and circuit breakers into a substation system-level model using a unified data exchange format (JSON). Verify the accuracy of model conversion through model verification. Compare the model output with historical substation operating data. The simulation error is 6.5%. Through sensitivity testing (wind speed change ±5m / s, model output change ≤3%), confirm that the model meets the requirements of practical applications, forming a digital twin mechanism model.

[0049] Furthermore, in step S31, based on the physical characteristics of the power grid equipment, mechanistic equations describing its electrical, mechanical, thermodynamic, and other multi-physical field behaviors are embedded, and electrical and non-electrical measurement data are integrated to form a multi-scale physical parameter database for the equipment, including: S311. For specific power grid equipment such as transformers and circuit breakers, select the core mechanism equations that describe their electrical performance, mechanical structure and thermodynamic behavior based on their physical characteristics; use the finite element method or finite volume method to perform spatiotemporal discretization processing on the above partial differential equation form of the mechanism equations to generate an algebraic equation system suitable for numerical calculation. S312. Integrate static parameters and dynamic measurement data from equipment nameplates and design drawings; use parameter estimation algorithms to identify unknown or time-varying parameters in the mechanism equations; and use data assimilation technology to fuse real-time measurement data with the mechanism model output to correct model parameters and reduce model uncertainty. S313. The identified and fused physical parameters of the equipment are structured and stored according to their physical meaning, time scale, and spatial scale to form a multi-scale physical parameter database containing metadata such as parameter name, value, unit, timestamp, and data source, providing accurate and consistent parameter input for subsequent full-scale multi-physics digital twin models.

[0050] Furthermore, in step S32, based on the internal operating rules of the equipment and the response behavior to external disturbances, a differential-algebraic equation model or finite state machine model of the equipment under transient and steady-state conditions is constructed to characterize its energy flow trajectory, information flow transmission relationship, and dynamic functional behavior, including: S321. For the dynamic characteristics of power grid equipment, for continuously changing physical quantities, construct differential algebraic equation models describing their transient and steady-state behavior based on Kirchhoff's laws and the law of electromagnetic induction; for discrete logical operation behaviors, construct finite state machine models based on states, events, and actions as core elements. S322. Set initial conditions, boundary conditions, and equation coefficients related to the physical parameters of the device for the differential algebraic equation model; define all possible state sets, event sets that trigger state transitions, and action sets to be executed during state transitions for the finite state machine model, and encode the internal operating rules of the device into state transition logic. S323. Establish an interactive interface between the differential algebraic equation model and the finite state machine model, so that the state switching signal output by the finite state machine model can be used as the triggering condition for changes in the structure or parameters of the differential algebraic equation model. At the same time, the physical quantities calculated by the differential algebraic equation model are used as the event inputs of the finite state machine model, thereby comprehensively representing the energy flow trajectory, information flow transmission relationship and complete dynamic functional behavior of the device under internal rules and external disturbances.

[0051] Specifically, S33 involves assembling and coupling the multiphysics model, behavior model, and rule model of a single device, and ensuring its consistency with the actual physical object through model verification, validation, and confirmation mechanisms, ultimately forming a digital twin full-space model that supports real-time simulation and offline deduction, including: S331. Assemble the multiphysics model, behavior model and rule model of a single device, and realize the data interaction and logical association between different models by defining a unified data exchange format and calling interface, so as to ensure the synergistic effect between physical field changes, behavior response and rule constraints. S332. Ensure the accuracy of model conversion and implementation through model verification; compare the model output with historical operating data and real-time monitoring data of actual physical objects through model validation, and use error analysis and sensitivity testing to quantify the model accuracy; ensure that the model is sufficiently reliable for its intended application. S333. After passing the verification mechanism, the assembled and coupled model will form a digital twin full-space model that supports real-time simulation and offline deduction. This model can reflect the complete information and dynamic characteristics of the power grid from equipment components to the system level.

[0052] Furthermore, in step S4, data-driven modeling is performed. Based on historical and real-time measurement data, deep learning algorithms are used to mine high-dimensional data mapping features and optimize the model's nonlinear representation capabilities, including: S41. Integrate historical operating data and real-time measurement data of power grid equipment, clean, align and standardize the data; construct a high-dimensional feature space by using feature encoding, time-series feature extraction and dimensionality reduction algorithms, and screen feature subsets that are strongly correlated with equipment status through chi-square test to form a high-quality training sample set. S42. Design a deep learning architecture that includes a multilayer perceptron, convolutional neural network or long short-term memory network, input preprocessed feature data, and optimize model parameters through backpropagation algorithm; introduce physical constraint terms as regularization in the loss function to ensure that the output of the data-driven model conforms to the physical laws of the power grid, and improve its nonlinear representation ability and generalization. The design incorporates a deep learning architecture based on LSTM+CNN. The CNN layer extracts spatial features from the data (3×3 convolutional kernels and 2×2 pooling kernels), while the LSTM layer extracts temporal features (128 hidden nodes and 3 layers). The output layer represents the device state prediction result. Kirchhoff's current law constraint term is introduced into the loss function (MSE loss) as regularization, with the constraint coefficient set to 0.01 to ensure that the model output conforms to physical laws.

[0053] S43. Use cross-validation to divide the training set and test set, and evaluate the model's accuracy, root mean square error and other indicators; deploy an online learning mechanism to dynamically update the model parameters through incremental learning or federated learning frameworks to adapt to time-varying characteristics such as equipment aging and environmental changes, and finally generate a high-fidelity data-driven model that supports equipment status prediction, fault diagnosis or energy efficiency optimization.

[0054] The model was trained using the PyTorch framework with 1000 iterations and a learning rate of 0.001. The model achieved an accuracy of 98.6% on the test set and a root mean square error of 4.2%. A federated learning framework was deployed to enable online model updates. Incremental training was performed every 24 hours based on newly added measurement data to update model parameters and adapt to time-varying characteristics such as equipment aging and weather changes.

[0055] S5: Multi-paradigm integration to form a full-space digital twin model of substations. By integrating 3D geometric models, information mechanism models, and data-driven models with a multi-paradigm fusion and coupling them with a model management module, a unified API call interface is defined to achieve bidirectional data interaction for the visualization of geometric models, real-time simulation of mechanism models, and state prediction of data-driven models. Ultimately, a digital twin full-space model of a 220kV substation is formed, which can realize real-time synchronization between the physical power grid and the virtual model (synchronization delay of 45ms), supporting business scenarios such as equipment status monitoring, fault simulation and inference, and intelligent scheduling.

[0056] Example 2 Figure 2 The present invention provides a power grid digital twin modeling system based on multi-paradigm fusion, the system comprising: The data governance and standardization module is used to aggregate, clean and standardize multi-source heterogeneous data of the power grid, establish unified data specifications based on the SG-CIM model and build a high-quality modeling data foundation. This model is used to aggregate, clean, and standardize multi-source heterogeneous data from the power grid. This module acquires structured, semi-structured, and unstructured data from the power grid resource business platform, real-time measurement center, enterprise-level meteorological center, and PMS system through multi-protocol adaptation interfaces. Based on the SG-CIM model, it establishes a unified data specification and generates a standardized dataset that conforms to the IEC 61970 standard through semantic alignment, format conversion, and quality assessment, thus building a high-quality modeling data foundation.

[0057] The 3D geometric modeling engine module is used to perform geometric modeling, accurately restoring the geometric shape, spatial layout, and component relationships of power grid equipment through GIM file parsing and 3D reconstruction technology. This model is used to accurately restore the geometric shape and spatial layout of power grid equipment through GIM file parsing and 3D reconstruction technology. This module parses GIM standard files to extract attribute parameters, hierarchical relationships and geometric data of primitives, reconstructs basic primitives such as points, lines and surfaces in the 3D development platform, and performs component-level white model assembly and spatial registration according to assembly relationships to form a 3D model with accurate geometric shape and component relationship.

[0058] The multiphysics mechanism modeling module is used to implement information mechanism modeling, embed the physical parameters and operating rules of power grid equipment, and construct a digital twin mechanism model of power grid equipment that conforms to the actual operating law. This module is used to embed physical parameters and operating rules of power grid equipment to construct a mechanism model that conforms to actual operating laws. It integrates multiple physical field equations such as electrical, mechanical, and thermodynamic, combines static parameters and real-time measurement data, corrects model parameters through parameter identification and data assimilation technology, and constructs differential algebraic equation model and finite state machine model to characterize the energy flow and dynamic behavior of the equipment under transient and steady states. The data-driven analysis module is used to perform deep learning-driven modeling and optimization based on historical and real-time measurement data. This module uses feature engineering and dimensionality reduction algorithms to construct a high-dimensional feature space, mines data mapping features through multilayer perceptrons, convolutional neural networks or long short-term memory networks, and introduces physical constraint terms into the loss function to improve the model's nonlinear representation ability and generalization, supporting equipment status prediction and fault diagnosis. The multi-paradigm fusion and model management module is used for data-driven modeling. Based on historical and real-time measurement data, it uses deep learning algorithms to mine high-dimensional data mapping features and optimize the model's nonlinear representation capabilities.

[0059] This module is used to couple and integrate geometric models, mechanistic models and data-driven models. It assembles models of different paradigms by defining a unified data interface and calling specifications, and uses model verification, validation and confirmation mechanisms to ensure the consistency between the model and the actual object, forming a digital twin full-space model that supports real-time simulation and offline inference, and providing visualization and interactive services for power grid monitoring, scheduling and operation and maintenance.

[0060] In addition, the parts of the technical solutions provided in the embodiments of this application that are consistent with the implementation principles of the corresponding technical solutions in the prior art have not been described in detail, so as to avoid excessive elaboration.

[0061] Embodiment 3 of the present invention provides a terminal, including a processor and a storage medium; the storage medium is used to store instructions; the processor is used to operate according to the instructions to execute the method steps provided according to Embodiment 1.

[0062] Embodiment 4 of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method steps provided according to Embodiment 1.

[0063] This disclosure can be a system, method, and / or computer program product. A computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for causing a processor to implement various aspects of this disclosure.

[0064] Computer-readable storage media can be tangible devices capable of holding and storing instructions for use by an instruction execution device. Computer-readable storage media can be, for example—but not limited to—electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or recessed protrusions storing instructions thereon, and any suitable combination of the foregoing. The computer-readable storage media used herein are not to be construed as transient signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses through fiber optic cables), or electrical signals transmitted through wires.

[0065] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0066] Computer program instructions used to perform the operations of this disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, status setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as C or similar languages. The computer-readable program instructions may execute entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing the status information of the computer-readable program instructions to implement various aspects of this disclosure.

[0067] Finally, 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 the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A digital twin modeling method for power grids based on multi-paradigm fusion, characterized in that, Includes the following steps: S1. Standardize the multi-source heterogeneous data of the power grid, establish unified data specifications based on the SG-CIM model, and build a high-quality modeling data base; S2. Conduct geometric modeling, and accurately restore the geometric shape, spatial layout and component relationships of power grid equipment by combining GIM file parsing and 3D modeling technology; S3. Implement information mechanism modeling, embed the physical parameters and operating rules of power grid equipment, and construct a digital twin mechanism model of power grid equipment that conforms to actual operating rules; S4. Conduct data-driven modeling, based on historical and real-time measurement data, and use deep learning algorithms to mine high-dimensional data mapping features to optimize the nonlinear representation capability of the digital twin mechanism model.

2. The power grid digital twin modeling method based on multi-paradigm fusion according to claim 1, characterized in that, S1, standardizing multi-source heterogeneous power grid data, establishing unified data specifications based on the SG-CIM model, and constructing a high-quality modeling data foundation includes: S11. Initiate the multi-source data automatic aggregation process to receive structured, semi-structured, and unstructured power grid data from different external systems; S12. Based on the SG-CIM model, extract the power grid equipment classes, core attributes and cross-equipment relationships, establish a precise mapping between multi-source heterogeneous data and model elements, clarify the data fields, formats, encoding and transmission protocol specifications, and unify the data format to a structured format that conforms to the IEC61970 standard. S13. Perform preprocessing operations on the aggregated heterogeneous data, use intelligent cleaning algorithms to remove duplicate data and correct erroneous data, fill missing values ​​through interpolation, eliminate data noise using smoothing techniques, and use semantic web technology to achieve semantic alignment of heterogeneous data and eliminate semantic differences between different data sources. S14. Set four quality indicators for data integrity, accuracy, consistency, and timeliness. Use a dual-drive mode of expert review and intelligent verification to conduct full-cycle quantitative evaluation of the preprocessed data. After evaluation, a standardized dataset is formed to build the data base for modeling.

3. The power grid digital twin modeling method based on multi-paradigm fusion according to claim 2, characterized in that, S11, initiating the multi-source data automatic aggregation process, receiving structured, semi-structured, and unstructured power grid data from different external systems, includes: Obtain equipment diagram data from the power grid resource business platform through services; Obtain electrical and non-electrical measurement data of the equipment from the real-time measurement center through services; Obtain geographic environment and meteorological monitoring data from enterprise-level meteorological centers through services; Obtain defect, fault, power outage, and maintenance record data from the PMS3.0 business system.

4. The power grid digital twin modeling method based on multi-paradigm fusion according to claim 2, characterized in that, S12 involves extracting power grid equipment classes, core attributes, and cross-equipment relationships based on the SG-CIM model, establishing a precise mapping between multi-source heterogeneous data and model elements, clarifying data fields, formats, encoding, and transmission protocol specifications, and unifying the data format to a structured format conforming to the IEC 61970 standard, including: Based on the core architecture of the SG-CIM model, we extract the core attributes of various power grid equipment categories, including switches, transformers, capacitors, and reactors, and sort out the cross-equipment relationships. Establish a mapping relationship between heterogeneous data and SG-CIM model elements; Establish a unified data standard system; All types of heterogeneous structured data are uniformly converted into the CIM XML structured format that conforms to the IEC61970 standard.

5. The power grid digital twin modeling method based on multi-paradigm fusion according to claim 1, characterized in that, S2 involves conducting geometric modeling, using a combination of GIM file parsing and 3D modeling technology to accurately reconstruct the geometric shape, spatial layout, and component relationships of power grid equipment. S21. Parse the GIM standard file and extract the attribute parameters, hierarchical structure relationships, and geometric parameters of the power grid equipment. S22. Based on the extracted primitive geometric parameters, the model is reconstructed and assembled in the 3D model development platform to generate the basic 3D white model of the power grid equipment. S23. Based on the hierarchical structure relationship, perform model assembly and spatial registration, establish the association mapping between components, and form a three-dimensional model with accurate geometric shape and spatial layout.

6. The power grid digital twin modeling method based on multi-paradigm fusion according to claim 5, characterized in that, S21, parsing the GIM standard file and extracting the attribute parameters, hierarchical structure relationships, and geometric parameters of power grid equipment, includes: Load and parse the GIM compressed file package, establish a complete five-level engineering hierarchy tree from system level to component level, generate a GIM structure tree and device list described in JSON format, and clarify the parent-child dependency and spatial membership relationship between devices; Parse the device property file, extract the Chinese and English property names and values ​​line by line, and map the extracted properties to the SG-CIM standard; The primitives are extracted and transformed step by step from the local coordinate system to the global engineering coordinate system, correcting coordinate offsets caused by projection transformation or acquisition accuracy; the final output is a primitive parameter set containing geometric type, vertex coordinates, line width, and transformation matrix information.

7. The power grid digital twin modeling method based on multi-paradigm fusion according to claim 5, characterized in that, S22, based on the extracted primitive geometric parameters, involves model reconstruction and assembly on a 3D model development platform to generate a basic 3D white model of the power grid equipment, including: Input the geometric parameters of the primitives into the 3D model development platform to generate the corresponding 3D geometry. The geometry is refined and assembled into component-level 3D white models; Based on the hierarchical structure, the assembled component-level models are placed in the correct spatial positions determined by the global coordinate system of the project, establishing parent-child relationships between components, and finally generating a three-dimensional white model of the overall foundation of the power grid facility.

8. The power grid digital twin modeling method based on multi-paradigm fusion according to claim 5, characterized in that, Step S23, which involves assembling and spatially registering the model based on the hierarchical structure, establishing the association mapping between components, and forming a three-dimensional model with accurate geometric shape and spatial layout, includes: In a 3D scene, the generated component-level 3D white model is spatially positioned to form the overall 3D structural framework of the substation or line corridor. Establish relationships between components and add logical connectors; establish parent-child relationships between models; and add logical connectors representing the connection relationships of conductors and busbars based on the power grid topology. The starting and ending points of the lines in the model are mapped and associated with the nodes in the physical power grid to generate a 3D model that interacts with the physical power grid in real time.

9. The power grid digital twin modeling method based on multi-paradigm fusion according to claim 1, characterized in that, S3, implementing information mechanism modeling, embedding physical parameters and operating rules of power grid equipment, and constructing a digital twin mechanism model of power grid equipment that conforms to actual operating laws includes: S31. Based on the physical characteristics of power grid equipment, embed the mechanism equations of multi-physics field behavior, and integrate electrical and non-electric measurement data to form a multi-scale physical parameter database for the equipment; S32. Based on the internal operating rules of the equipment and the response behavior to external disturbances, construct a differential algebraic equation model or finite state machine model of the equipment in transient and steady states to characterize the dynamic functional behavior of the equipment. S33. Assemble and couple the multiphysics model, behavior model and rule model of a single device, and form a digital twin full-space model that supports real-time simulation and offline deduction through model verification, validation and confirmation mechanisms.

10. The power grid digital twin modeling method based on multi-paradigm fusion according to claim 9, characterized in that, S31, based on the physical characteristics of power grid equipment, embeds the mechanism equations of multi-physics behavior, integrates electrical and non-electrical measurement data, and forms a multi-scale physical parameter database for the equipment, including: The core mechanism equations of the multiphysics field of power grid equipment are selected and spatiotemporally discretized. Integrate static parameters and dynamic measurement data, and correct model parameters through parameter identification and data assimilation; The physical parameters are organized and stored in a structured manner according to their physical meaning and spatiotemporal scale, forming a multi-scale physical parameter database.

11. The power grid digital twin modeling method based on multi-paradigm fusion according to claim 9, characterized in that, S32, based on the internal operating rules of the equipment and the response behavior to external disturbances, constructs a differential algebraic equation model or finite state machine model of the equipment in transient and steady states, characterizing the dynamic functional behavior of the equipment, including: For the dynamic characteristics of power grid equipment, for continuously changing physical quantities, a differential-algebraic equation model describing their transient and steady-state behavior is constructed based on Kirchhoff's laws and the law of electromagnetic induction; for discrete logical operation behavior, a finite state machine model based on state, event, and action as core elements is constructed. Set initial conditions, boundary conditions, and equation coefficients related to the physical parameters of the device for the differential-algebraic equation model; define all possible state sets, event sets that trigger state transitions, and action sets to be executed during state transitions for the finite state machine model; and encode the internal operating rules of the device into state transition logic. An interactive interface is established between the differential algebraic equation model and the finite state machine model, so that the state switching signal output by the finite state machine model can be used as the triggering condition for changes in the structure or parameters of the differential algebraic equation model. At the same time, the physical quantities calculated by the differential algebraic equation model are used as the event inputs of the finite state machine model, thereby realizing an integrated representation of the dynamic behavior of the device.

12. The power grid digital twin modeling method based on multi-paradigm fusion according to claim 9, characterized in that, S33, assembling and coupling the multiphysics model, behavior model, and rule model of a single device, and forming a digital twin full-space model that supports real-time simulation and offline derivation through model verification and confirmation mechanisms, includes: Define a unified data exchange format and calling interface to enable data interaction and logical association between different models; The accuracy of the model is quantified and its reliability is ensured through model verification and validation. The assembled and coupled models form a digital twin full-space model that supports real-time simulation and offline deduction, reflecting the complete information and dynamic characteristics of the power grid from equipment components to the system level.

13. The power grid digital twin modeling method based on multi-paradigm fusion according to claim 1, characterized in that, S4 involves data-driven modeling, using deep learning algorithms to mine high-dimensional data mapping features based on historical and real-time measurement data, and optimizing the model's nonlinear representation capabilities, including: S41. Integrate historical operating data and real-time measurement data of power grid equipment and preprocess them to construct a high-dimensional feature space. Then, use the chi-square test to screen the feature subset that is strongly correlated with the equipment status to form a training sample set. S42. Design a deep learning architecture, input preprocessed feature data, optimize model parameters through backpropagation algorithm; introduce physical constraint terms into the loss function to improve model performance; S43. Evaluate model metrics through cross-validation, deploy an online learning mechanism to update model parameters, and generate a high-fidelity data-driven model.

14. A power grid digital twin modeling system based on multi-paradigm fusion, characterized in that, The system includes: The data governance and standardization module is used to aggregate, clean and standardize multi-source heterogeneous data of the power grid, establish unified data specifications based on the SG-CIM model and build a high-quality modeling data foundation. The 3D geometric modeling engine module is used to perform geometric modeling, accurately restoring the geometric shape, spatial layout, and component relationships of power grid equipment through GIM file parsing and 3D reconstruction technology. The multiphysics mechanism modeling module is used to implement information mechanism modeling, embed the physical parameters and operating rules of power grid equipment, and construct a digital twin mechanism model of power grid equipment that conforms to the actual operating law. The multi-paradigm fusion and model management module is used for data-driven modeling. Based on historical and real-time measurement data, it uses deep learning algorithms to mine high-dimensional data mapping features and optimize the model's nonlinear representation capabilities.

15. An electronic device, comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the power grid digital twin modeling method based on multi-paradigm fusion according to any one of claims 1-13.

16. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the power grid digital twin modeling method based on multi-paradigm fusion as claimed in any one of claims 1-13.

Images