Substation electrical drawing intelligent recognition and digital modeling method and system and electronic equipment

Abstract

This invention discloses a method, system, and electronic device for intelligent recognition and digital modeling of substation electrical drawings, relating to the field of image processing technology. The method includes: electrical drawing input and multimodal semantic parsing; spatial localization of text and graphic elements based on small models; semantic result association based on text similarity; text-graphic element binding incorporating spatial distance and semantic constraints; inclusion relationship parsing and digital logic construction; and electrical connection recognition and digital model generation. This method achieves high-precision recognition of complex electrical drawings, effectively reducing text-graphic element mismatch problems. It can automatically identify inclusion relationships represented by solid and dashed boxes, construct a multi-level digital logic structure, and the generated digital model supports electronic drawing reconstruction, facilitating manual verification and correction, significantly improving engineering practicality. Furthermore, it is not dependent on fixed templates and is applicable to substation electrical drawings from different drawing standards and manufacturers.

Description

Technical Field

[0001] This invention relates to the field of image processing methods, and in particular to a method, system, and electronic equipment for intelligent recognition and digital modeling of substation electrical drawings. Background Technology

[0002] With the rapid development of new power systems and smart grids, digital modeling of substations has become a crucial foundation for supporting power system planning, design, operation, maintenance, and intelligent analysis. In substation systems, electrical drawings are core technical documents describing the logical relationships of protection, measurement and control, communication, and automation. They contain a large number of electrical elements, terminal symbols, text annotations, and complex electrical connections, characterized by diverse element types, dense text annotations, and complex structural hierarchies. Currently, electrical drawings still largely exist in paper or unstructured electronic form, and their digital modeling process mainly relies on manual input or semi-automatic auxiliary tools, resulting in low efficiency, susceptibility to errors, difficulty in verification, and high maintenance costs.

[0003] Current intelligent recognition and structured processing technologies for electrical drawings primarily rely on traditional small models, including targeted object detection, character recognition, and rule-based topology inference. These methods typically offer high inference efficiency, accurate spatial positioning, and well-structured output results, facilitating engineering deployment and direct data storage. However, electrical drawings contain a large amount of context-dependent implicit semantic information, such as precise interpretation of terminal functions and the hierarchical relationships of components within multi-level logical structures. Small models have limited ability to understand such deep semantics, making it difficult to independently achieve consistent inference from element recognition to semantic understanding. This often necessitates the introduction of complex expert rule bases or manual intervention for post-processing, leading to insufficient system generalization, poor adaptability to drawings from different manufacturers or drawing standards, and high rule maintenance costs. In recent years, emerging technologies such as multimodal large models have demonstrated strong potential in the joint understanding and reasoning of images and text. These models can interpret text annotations based on the overall context of the drawing and infer the potential functional logical relationships between graphic elements. However, the output of multimodal large models is usually mainly unstructured text descriptions or weakly structured intermediate results, which generally suffer from problems such as coarse spatial positioning granularity, weak geometric accuracy, repeatability, and engineering verifiability of the output results. This makes it difficult to directly integrate with digital modeling processes that require precise geometric and topological information at the primitive level. Therefore, relying solely on small models or solely on large models cannot simultaneously meet the comprehensive requirements of digital modeling of electrical drawings. There is an urgent need for a technical solution that integrates the semantic reasoning advantages of multimodal large models with the geometric positioning advantages of small models to achieve high-precision intelligent recognition and digital modeling for engineering applications. Summary of the Invention

[0004] The technical problem to be solved by this invention is how to provide a method for intelligent recognition and digital modeling of substation electrical drawings that can achieve high-precision recognition of complex electrical drawings.

[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a method for intelligent recognition and digital modeling of substation electrical drawings, comprising the following steps: S1, Electrical Drawing Input and Multimodal Semantic Analysis: The substation electrical drawings to be processed are taken as input. The image content and text annotations are jointly analyzed by a lightweight multimodal large model to analyze the electrical meaning of the text annotations and obtain the electrical elements, terminals and their semantic relationships. S2, Spatial localization of text and graphic elements based on small models: Using small models to detect and locate text annotations and electrical graphic elements in electrical drawings, and obtain their spatial coordinate information in the image; S3, Semantic result association based on text similarity: Match the output of the multimodal large model with the recognition result of the small model based on the similarity of text content, and associate spatial coordinate information with the corresponding text semantic result; S4, Introducing text-element binding with spatial distance and semantic constraints: Taking into account the spatial distance constraints between text annotations and electrical elements as well as the electrical semantic consistency constraints, high-precision binding between text annotations and electrical elements is achieved; S5, Inclusion Relationship Analysis and Digital Logic Construction: Using a multimodal large model to identify inclusion relationships represented by solid or dashed boxes, and combining the spatial location information of electrical elements to construct the logical structure of the system; S6, Electrical Connection Identification and Digital Model Generation: Constructs connection links between electrical elements through connected component analysis, forms a complete electrical topology relationship, generates a standardized digital model, and supports the restoration and verification of electronic drawings.

[0006] This invention also discloses an intelligent recognition and digital modeling system for substation electrical drawings, the system comprising: Multimodal semantic parsing module: Taking the substation electrical drawings to be processed as input, it performs joint analysis on the image content and text annotations through a lightweight multimodal large model, parses the electrical meaning of the text annotations, and obtains electrical elements, terminals and their semantic relationships; Text and graphic element spatial positioning module: used to detect and locate text annotations and electrical elements in electrical drawings using small models, and obtain their spatial coordinate information in the image; Semantic result association module: used to match the output results of the multimodal large model with the recognition results of the small model based on the similarity of text content, and associate spatial coordinate information with the corresponding text semantic results; Text-to-Electrical Element Binding Module: This module comprehensively considers the spatial distance constraints and electrical semantic consistency constraints between text annotations and electrical elements to achieve high-precision binding between text annotations and electrical elements. Relationship parsing and digital logic construction module: used to identify inclusion relationships represented by solid or dashed boxes using a multimodal large model, and to construct the logical structure of the system by combining the spatial location information of electrical elements; Electrical Wiring and Digital Model Generation Module: This module is used to construct connection links between electrical elements through connected component analysis, form a complete electrical topology, generate a standardized digital model, and support the restoration and verification of electronic drawings.

[0007] The present invention also discloses an electronic device, including a memory and a processor, wherein the memory is used to store a computer program, and the processor runs the computer program to cause the electronic device to perform the method described thereon.

[0008] The beneficial effects of adopting the above technical solution are as follows: First, the method of the present invention achieves high-precision recognition of complex electrical drawings by combining the semantic understanding capability of a multimodal large model with the high-precision spatial positioning capability of a small model. Second, by introducing a dual constraint mechanism of text similarity and spatial distance, it effectively reduces the problem of text and graphic element mismatch. Third, it can automatically identify the inclusion relationship represented by solid and dashed boxes, constructing a multi-level digital logical structure. Finally, the generated digital model supports the reconstruction of electronic drawings, facilitates manual verification and correction, significantly improves engineering practicality, and is not dependent on fixed templates, making it applicable to substation electrical drawings from different manufacturers and drawing standards. Attached Figure Description

[0009] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0010] Figure 1 This is the main flowchart of the method described in Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of the system described in Embodiment 2 of the present invention; Figure 3 This is a schematic block diagram of the electronic device described in Embodiment 3 of the present invention. Detailed Implementation

[0011] The technical solutions of the embodiments of the present invention will be clearly and completely described 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 the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0012] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0013] like Figure 1 As shown in the figure, an embodiment of the present invention discloses a method for intelligent recognition and digital modeling of substation electrical drawings, the method comprising the following steps: S1, Electrical drawing input and multimodal semantic parsing: Obtain the electrical drawing of the substation to be processed, preprocess the electrical drawing, and input it into the lightweight multimodal large model. The multimodal large model performs joint analysis on the image content and text annotations, parses the electrical meaning corresponding to the text annotations, extracts electrical elements, terminal information and their semantic relationships, and obtains the first recognition result containing text semantics, electrical attributes and element semantics.

[0014] S2, text and graphic element spatial localization based on small model: using a small model to perform target detection and localization on the text annotations and electrical graphic elements in the electrical drawings, obtaining the spatial coordinate information of each text annotation and each electrical graphic element in the image, forming a second recognition result containing text coordinates, graphic element coordinates and category information.

[0015] S3, semantic result association based on text similarity: matching the text content in the first recognition result with the text recognition result in the second recognition result, determining candidate matching relationships based on the similarity of the text content, and filtering by similarity threshold, associating the spatial coordinate information in the second recognition result with the corresponding text semantic result.

[0016] S4, Introducing text-to-graphic binding with spatial distance and semantic constraints: Based on the completion of the association between text semantics and spatial coordinates, the spatial distance constraints between text annotations and electrical graphics elements and the electrical semantic consistency constraints are comprehensively considered to determine the one-to-one or many-to-one binding relationship between text annotations and electrical graphics elements, thereby achieving high-precision text-to-graphic binding.

[0017] S5, Inclusion Relationship Analysis and Digital Logic Construction: Using the multimodal large model, the inclusion relationships represented by solid or dashed boxes in the electrical drawings are identified. Combined with the obtained spatial coordinate information of electrical elements, the hierarchical structure and logical inclusion relationships between elements are determined, and a digital model reflecting the system's logical structure is constructed.

[0018] S6, Electrical Connection Identification and Digital Model Generation: Based on the spatial connectivity between electrical elements, connected component analysis is used to construct electrical connection links between elements, forming a complete electrical topology, and generating a digital model that conforms to preset data specifications; at the same time, the digital model supports the restoration and interactive verification of electronic drawings.

[0019] The above steps will be explained in detail below with specific methods: Step S1, Electrical drawing input and multimodal semantic parsing: First, obtain the electrical drawings of the substation to be processed. The electrical drawings can be image files obtained by scanning paper drawings or unstructured electronic drawing files.

[0020] The electrical drawings Input to lightweight multimodal large model The multimodal large model is a joint representation model capable of simultaneously processing image and text modal inputs, and its output can be expressed as: in, Indicates the first The semantic representation of each text annotation Indicates the first Semantic representation of electrical primitives Indicates the semantics of a terminal or interface. This indicates the semantic relationship between text labels and electrical elements.

[0021] By leveraging the multimodal large model's ability to jointly model contextual information, the electrical meaning of text annotations can be analyzed, thereby avoiding ambiguity issues caused by relying solely on character recognition.

[0022] Step S2, spatial localization of text and primitives based on the small model: Based on the completion of semantic parsing, a small model is used. Target detection and positioning are performed on the text annotations and electrical elements in the electrical drawings.

[0023] The small model is a visual model based on object detection, and its output can be expressed as: in, Indicates the first A text-labeled spatial bounding box, Indicates the first The spatial bounding box of each electrical element These represent the corresponding category information.

[0024] The small model focuses on acquiring high-precision spatial location information, providing a geometric basis for subsequent text-primitive binding.

[0025] Based on semantic parsing, a small model is used to perform target detection and localization on text annotations and electrical primitives in the electrical drawings. Preferably, the small model employs a detection network based on a target detection framework, such as the YOLO series network, to detect text regions and primitive regions, obtaining a set of text bounding boxes. With primitive bounding box set .

[0026] Furthermore, for each text bounding box Character recognition is performed on the corresponding image region to obtain the text string on the small model side. To improve matching stability, the text string can be normalized, including: unifying case, removing spaces and delimiters, replacing synonyms with equivalents, and standardizing the mapping of common electrical symbols, resulting in a normalized text string. .

[0027] Step S3, semantic result association based on text similarity: Subsequently, the text content in the first recognition result is matched with the text recognition result in the second recognition result to associate the spatial coordinate information obtained by the small model with the output result of the large model, thereby achieving the alignment and fusion of semantic information and spatial information.

[0028] In one implementation, each text annotation output by the multimodal large model is denoted as a string. , forming a set The normalized text string obtained from the small model side is denoted as , forming a set .

[0029] To improve matching robustness, the string is converted into a token set. The string can be split into several substrings at the character level, resulting in: in, Indicates the tokenization function. Text on the side of the large model The set of tokens, Text on the side of the small model The set of tokens.

[0030] Define large model text With small model text The Jaccard similarity coefficient is: when At that time, it was believed that there was a candidate matching relationship between the two at the text level, among which The preset threshold can be set to a value in the range of 0.5–0.8, depending on the paper noise and font quality.

[0031] After completing the above matching, the spatial coordinates of the small model side are... The textual semantic results backfilled into the large model side are used to form semantic results with coordinates: This enables the association of coordinate information recognized by the small model with the output of the large model, providing a foundation for the spatial constraint calculation of subsequent "text-primitive binding".

[0032] Step S4: A comprehensive score based on spatial distance and semantic consistency, and text-to-primitive binding: Building upon the association between textual semantics and spatial coordinates, this embodiment introduces spatial distance constraints and electrical semantic consistency constraints to further improve the accuracy of textual annotations and electrical element binding, and jointly evaluates candidate binding relationships.

[0033] S4-1, Construction of candidate binding pairs: Let the set of textual semantic results obtained in step three be: in, Indicates the first The semantic content of each text annotation This indicates the corresponding text bounding box.

[0034] Let the set of electrical primitives obtained from the small model detection be: in, Indicates the first Semantic representation of electrical primitives This represents its spatial bounding box.

[0035] For each text annotation Select several electrical elements within its neighborhood. As candidate binding objects, a candidate set is formed: in, This represents a candidate search region constructed based on spatial proximity relationships.

[0036] S4-2, Formal definition of spatial distance constraints: Candidate binding pairs Define the spatial distance function between text annotations and electrical elements: The distance function can be calculated based on the distance to the center point of the bounding box.

[0037] To facilitate integration with other constraints, the spatial distance is normalized to obtain a spatial consistency score: in, This is a spatial distance scale parameter used to control the distance decay rate.

[0038] S4-3, Formal definition of electrical semantic consistency constraints: To measure text annotation With electrical elements To determine the degree of semantic matching, a semantic consistency function is defined: in, This represents a semantic consistency determination function, used to determine whether the electrical meaning described in the text is consistent with the semantics of the graphic elements. The semantic consistency function outputs a binary result: S4-4, Construction of the comprehensive scoring function: Based on the aforementioned spatial distance constraints and semantic consistency constraints, for each candidate binding pair Construct a comprehensive scoring function: in, and These are weighting coefficients used to balance the influence of spatial and semantic factors, and they satisfy... .

[0039] S4-5, Determining Binding Relationships and Resolving Conflicts: For each text annotation In its candidate set Select the electrical element with the highest overall score as the binding object: When the following conditions are met: At that time, determine the text annotation. With electrical elements There is a binding relationship between them, This is the preset binding threshold.

[0040] When multiple text labels are bound to the same electrical element, a many-to-one binding relationship is allowed.

[0041] S4-6, Output binding result: After completing the above comprehensive scoring and screening process, the final text-image binding set is obtained: The binding result serves as the basis for subsequent inclusion relationship resolution and electrical topology modeling.

[0042] Step S5, based on the inclusion relationship parsing and digital logic construction of solid / dashed boxes: After completing the text-to-graphic element binding, in order to construct a digital model that reflects the functional division and logical structure of the substation system, this embodiment further parses the inclusion relationships represented by solid or dashed boxes in the electrical drawings.

[0043] S5-1, Detection and Type Differentiation of Frame Structures: Using a multimodal large model, closed line structures in electrical drawings are identified, and a set of box structures representing logic units is extracted. in, Indicates the first The spatial area of ​​the frame structure The type identifier indicates the type of the frame structure, which includes: solid frame and dashed frame.

[0044] S5-2, Determining the spatial inclusion of graphic elements and frame structures: Let the set of electrical elements obtained in step four be: For each electrical element Determine whether it is framed. Included.

[0045] The defined space contains decision functions: in, A threshold is included to avoid false positives caused by overlapping or contacting of graphic elements. At that time, the primitive is considered It belongs to the frame structure .

[0046] S5-3, Constructing multi-level containment relationships: In actual electrical drawings, there are often nested relationships between frame structures. Therefore, a set of frame structures is needed. Execution level resolution.

[0047] If it exists: Then it is considered a frame structure Enclosed in a frame structure Inside.

[0048] Based on the above relationships, construct a hierarchical set of the frame structure: This forms a top-down logical hierarchy.

[0049] S5-4, Determining the logical attribution relationship of graphic elements: For each electrical element A graph element may be contained within multiple box structures. To determine its final logical affiliation, the following rules are preferred: if there are multiple levels of containment, the innermost box structure is selected as the direct logical parent node of the graph element; if it is only contained within a dashed box, it is marked as a logical subordinate relationship; if it is contained within a solid box, it is marked as a strong containment relationship.

[0050] Therefore, the mapping relationship from graphical primitives to logical units is constructed: S5-5, Formal Representation of Digital Logical Model: Based on the hierarchical relationship of the frame structure and the ownership relationship of the elements, a digital logical model of the system is constructed: in: Represents a set of logical units; Represents a set of electrical elements; This indicates the hierarchical relationship between logical units; It indicates the logical attribution relationship of graphic elements.

[0051] This digital logical model can explicitly represent the composition structure of each functional unit in the system, providing logical constraints for the subsequent construction of electrical topology relationships.

[0052] Step S6, Electrical link identification and topology modeling based on connected component analysis: After completing the construction of the digital logical structure, in order to further restore the real connection relationship between each electrical unit in the substation system, this embodiment identifies the electrical connections in the electrical drawings and constructs a terminal-level electrical topology model.

[0053] S6-1, Extraction and binarization of electrical connection regions: First, extract the lines representing electrical connections from the electrical drawings. These lines may be straight, broken, or curved, and may intersect, overlap, or be broken.

[0054] In one implementation, the image can be processed based on color, line width, or structural features to map the electrical connection region into a binary image: in, Pixel-level representation of electrical connections.

[0055] S6-2, Connected Component Analysis and Segment Grouping: Binary line image Perform connected component analysis to extract the set of connected components: Each connected component It represents a group of spatially continuous electrical connection pixels.

[0056] Connectivity analysis can effectively handle polylines, polylines, and local breaks, aggregating line segments belonging to the same electrical path into a unified object.

[0057] S6-3, Extraction and representation of connection endpoints: For each connected component Extract its geometric endpoint set: Among them, endpoints This indicates the start or end point of a connection within the connected domain. Endpoints can be obtained through pixel neighborhood degree determination, endpoint detection after skeletonization, or morphological analysis. The endpoints are used to represent potential connection interfaces for electrical connections.

[0058] S6-4, Spatial adsorption matching of endpoints and terminals: Let the set of terminals obtained in steps one and four be: in, Indicates the first One terminal, It indicates its spatial location.

[0059] For each endpoint of the connection Calculate the spatial distance between it and each terminal: When the following conditions are met: At that time, the endpoints of the line are considered to be With terminals There is a connection relationship, where This is the endpoint snapping threshold. Through the aforementioned endpoint-terminal snapping process, pixel-level connections are mapped to terminal-level electrical connections.

[0060] S6-5, Construction and verification of electrical connection relationships: For each connected component If its endpoint set If successfully attached to two or more terminals, an electrical connection is established between the corresponding terminals.

[0061] Define the set of terminal-level connection edges: in, Indicator terminals With terminals There is an electrical connection between them.

[0062] The connection relationships can be further validated using the logical structure obtained in step five, including: whether there are abnormal connections across unrelated logical units; whether there are floating endpoints or isolated connections; and whether there are multi-branch connections with an abnormal number of terminals. Abnormal connections can be marked as pending validation.

[0063] S6-6, Construction of terminal-level electrical topology diagram: Based on terminal set With the set of connecting edges Constructing a terminal-level electrical topology diagram .in, Represents a set of terminal nodes. This represents the electrical connections between terminals. Furthermore, terminal-level topology can be mapped to primitive device-level topology to support system analysis and visualization at different granularities.

[0064] S6-7, Standardized Digital Model Generation and Drawing Reproduction: Finally, the digital logical model obtained in step five is combined with... With terminal-level electrical topology diagram Generate a standardized digital model that conforms to preset data specifications.

[0065] The digital model supports the generation of corresponding electronic drawings and can highlight electrical connections, logic units and terminal connections in the interactive interface to facilitate engineering verification, correction and operation and maintenance applications.

[0066] Example 2 Corresponding to the method described in Embodiment 1, Embodiment 2 of the present invention also discloses an intelligent recognition and digital modeling system for substation electrical drawings, the system comprising: Multimodal semantic parsing module 101: Takes the substation electrical drawings to be processed as input, performs joint analysis on the image content and text annotations through a lightweight multimodal large model, parses the electrical meaning of the text annotations, and obtains electrical elements, terminals and their semantic relationships; Text and graphic element spatial positioning module 102: Used to detect and locate text annotations and electrical graphic elements in electrical drawings using a small model, and obtain their spatial coordinate information in the image; Semantic result association module 103: used to match the output results of the multimodal large model with the recognition results of the small model based on the similarity of text content, and associate spatial coordinate information with the corresponding text semantic results; Text-to-graphic element binding module 104: It is used to comprehensively consider the spatial distance constraints and electrical semantic consistency constraints between text annotations and electrical graphic elements to achieve high-precision binding between text annotations and electrical graphic elements; Relationship parsing and digital logic construction module 105: used to identify inclusion relationships represented by solid or dashed boxes using a multimodal large model, and to construct the logical structure of the system by combining the spatial location information of electrical elements; Electrical Wiring and Digital Model Generation Module 106: Used to construct the connection links between electrical elements through connected domain analysis, form a complete electrical topology relationship, generate a standardized digital model, and support the restoration and verification of electronic drawings.

[0067] It should be noted that the specific implementation methods of each module in the system described in this embodiment two can refer to the methods described in embodiment one, and will not be repeated here.

[0068] Example 3 In one exemplary embodiment, the present invention also provides a computer device, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 3 As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements the intelligent recognition and digital modeling method for substation electrical drawings described in Embodiment 1.

[0069] Those skilled in the art will understand that Figure 3The structures shown are merely block diagrams of some structures related to the present application and do not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than shown in the figures, or combine certain components, or have different component arrangements. In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0070] In one exemplary embodiment, the present invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0071] In one exemplary embodiment, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0072] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0073] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0074] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic units, data processing logic units, etc., and are not limited to these.

[0075] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0076] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for intelligent recognition and digital modeling of substation electrical drawings, characterized in that... Includes the following steps: S1, Electrical Drawing Input and Multimodal Semantic Analysis: The substation electrical drawings to be processed are taken as input. The image content and text annotations are jointly analyzed by a lightweight multimodal large model to analyze the electrical meaning of the text annotations and obtain the electrical elements, terminals and their semantic relationships. S2, Spatial localization of text and graphic elements based on small models: Using small models to detect and locate text annotations and electrical graphic elements in electrical drawings, and obtain their spatial coordinate information in the image; S3, Semantic result association based on text similarity: Match the output of the multimodal large model with the recognition result of the small model based on the similarity of text content, and associate spatial coordinate information with the corresponding text semantic result; S4, Introducing text-element binding with spatial distance and semantic constraints: Taking into account the spatial distance constraints between text annotations and electrical elements as well as the electrical semantic consistency constraints, high-precision binding between text annotations and electrical elements is achieved; S5, Inclusion Relationship Analysis and Digital Logic Construction: Using a multimodal large model to identify inclusion relationships represented by solid or dashed boxes, and combining the spatial location information of electrical elements to construct the logical structure of the system; S6, Electrical Connection Identification and Digital Model Generation: Constructs connection links between electrical elements through connected component analysis, forms a complete electrical topology relationship, generates a standardized digital model, and supports the restoration and verification of electronic drawings.

2. The intelligent recognition and digital modeling method for substation electrical drawings as described in claim 1, characterized in that, S1 specifically includes the following steps: First, obtain the electrical drawings of the substation to be processed. The electrical drawings can be image files obtained by scanning paper drawings or unstructured electronic drawing files; The electrical drawings Input to lightweight multimodal large model The multimodal large model is a joint representation model capable of simultaneously processing image and text modal inputs, and its output is represented as follows: in, Indicates the first The semantic representation of each text annotation Indicates the first Semantic representation of electrical primitives Indicates the semantics of a terminal or interface. This indicates the semantic relationship between text labels and electrical elements.

3. The intelligent recognition and digital modeling method for substation electrical drawings as described in claim 1, characterized in that, S2 specifically includes the following steps: Using small models Target detection and localization are performed on the text annotations and electrical elements in the electrical drawings; The small model is a visual model based on object detection, and its output is expressed as follows: in, Indicates the first A text-labeled spatial bounding box, Indicates the first The spatial bounding box of each electrical element These represent the corresponding category information; Based on semantic parsing, a small model is used to perform target detection and localization on the text annotations and electrical elements in the electrical drawings.

4. The intelligent recognition and digital modeling method for substation electrical drawings as described in claim 3, characterized in that, The small model employs the YOLO series of networks based on object detection frameworks to detect text regions and primitive regions, obtaining a set of text bounding boxes. With primitive bounding box set For each text bounding box Character recognition is performed on the corresponding image region to obtain the text string on the small model side. .

5. The intelligent recognition and digital modeling method for substation electrical drawings as described in claim 1, characterized in that, S3 specifically includes the following steps: The text content in the recognition results of the multimodal large model is matched with the text recognition results in the recognition results of the small model to associate the spatial coordinate information obtained by the small model with the output results of the large model, so as to realize the alignment and fusion of semantic information and spatial information. Each text annotation output by the multimodal large model is recorded as a string. , forming a set The normalized text string obtained from the small model side is denoted as , forming a set ; The string is split into several substrings at the character level, resulting in: in, Indicates the tokenization function. Text on the side of the large model The set of tokens, Text on the side of the small model The set of tokens; Define large model text With small model text The Jaccard similarity coefficient is: when At that time, it was believed that there was a candidate matching relationship between the two at the text level, among which The preset threshold; After completing the above matching, the spatial coordinates of the small model side are... The textual semantic results backfilled into the large model side are used to form semantic results with coordinates: 。 6. The intelligent recognition and digital modeling method for substation electrical drawings as described in claim 1, characterized in that, S4 specifically includes the following steps: S4-1, Construction of candidate binding pairs: Let the set of textual semantic results obtained in step S3 be: in, Indicates the first The semantic content of each text annotation This indicates the corresponding text bounding box; Let the set of electrical primitives obtained from the small model detection be: in, Indicates the first Semantic representation of electrical primitives Indicates its spatial bounding box; For each text annotation Select several electrical elements within its neighborhood. As candidate binding objects, a candidate set is formed: in, This indicates a candidate search region constructed based on spatial proximity relationships; S4-2, Formal definition of spatial distance constraints: Candidate binding pairs Define the spatial distance function between text annotations and electrical elements: The distance function can be calculated based on the distance to the center point of the bounding box; The spatial distance is normalized to obtain the spatial consistency score: in, This is a spatial distance scale parameter used to control the distance decay rate; S4-3, Formal definition of electrical semantic consistency constraints: To measure text annotation With electrical elements To determine the degree of semantic matching, a semantic consistency function is defined: in, This represents a semantic consistency determination function, used to determine whether the electrical meaning described by the text is consistent with the semantics of the graphic primitives. The semantic consistency function outputs a binary result: S4-4, Construction of the comprehensive scoring function: Based on the aforementioned spatial distance constraints and semantic consistency constraints, for each candidate binding pair Construct a comprehensive scoring function: in, and These are weighting coefficients used to balance the influence of spatial and semantic factors, and they satisfy... ; S4-5, Determining Binding Relationships and Resolving Conflicts: For each text annotation In its candidate set Select the electrical element with the highest overall score as the binding object: When the following conditions are met: At that time, determine the text annotation. With electrical elements There is a binding relationship between them, The preset binding threshold; S4-6, Output binding result: After completing the above comprehensive scoring and screening process, the final text-image binding set is obtained: The binding result serves as the basis for subsequent inclusion relationship resolution and electrical topology modeling.

7. The intelligent recognition and digital modeling method for substation electrical drawings as described in claim 1, characterized in that, S5 specifically includes the following steps: S5-1, Detection and Type Differentiation of Frame Structures: Using a multimodal large model, closed line structures in electrical drawings are identified, and a set of box structures representing logic units is extracted. in, Indicates the first The spatial area of ​​the frame structure A type identifier indicating the box structure, the type identifier including: solid line box and dashed line box; S5-2, Determining the spatial inclusion of graphic elements and frame structures: Let the set of electrical elements obtained in step S4 be: For each electrical element Determine whether it is framed. Included; The defined space contains decision functions: in, A threshold is included to avoid false positives caused by overlapping or contacting primitives; when At that time, the primitive is considered It belongs to the frame structure ; S5-3, Constructing multi-level containment relationships: Frame structure set Execute hierarchical resolution, if it exists: Then it is considered a frame structure Enclosed in a frame structure Inside; Based on the above relationships, construct a hierarchical set of the frame structure: This forms a top-down logical hierarchy. S5-4, Determining the logical attribution relationship of graphic elements: For each electrical element A graphic element may be contained within multiple box structures. To determine its final logical affiliation, the following rules are used: If multiple levels of containment exist, the innermost box structure is selected as the direct logical parent node of the graphic element; if it is only contained within a dashed box, it is marked as a logical dependency; if it is contained within a solid box, it is marked as a strong containment. Thus, a mapping relationship from graphic elements to logical units is constructed: S5-5, Formal Representation of Digital Logical Model: Based on the hierarchical relationship of the frame structure and the ownership relationship of the elements, a digital logical model of the system is constructed: in: Represents a set of logical units; Represents a set of electrical elements; This indicates the hierarchical relationship between logical units; It indicates the logical attribution relationship of graphic elements.

8. The intelligent recognition and digital modeling method for substation electrical drawings as described in claim 1, characterized in that, S6 specifically includes the following steps: S6-1, Extraction and binarization of electrical connection regions: First, extract the line areas representing electrical connections in the electrical drawings. These lines include straight lines, broken lines, or curves, and may intersect, overlap, or break. Image processing is performed based on color, line width, or structural features to map electrical connection regions into binary images: in, Pixel-level representation of electrical connections; S6-2, Connected Component Analysis and Segment Grouping: Binary line image Perform connected component analysis to extract the set of connected components: Each connected component Represents a group of spatially continuous electrical connection pixels; S6-3, Extraction and representation of connection endpoints: For each connected component Extract its geometric endpoint set: Among them, endpoints The endpoints represent the starting or ending points of the connection within the connected domain; the endpoints are obtained through pixel neighborhood degree determination, skeletonized endpoint detection, or morphological analysis; the endpoints are used to represent potential connection interfaces for electrical connections. S6-4, Spatial adsorption matching of endpoints and terminals: Let the terminal set obtained in steps S1 and S4 be: in, Indicates the first One terminal, Indicates its spatial location; For each endpoint of the connection Calculate the spatial distance between it and each terminal: When the following conditions are met: At that time, the endpoints of the line are considered to be With terminals There is a connection relationship, where The endpoint adsorption threshold; S6-5, Construction and verification of electrical connection relationships: For each connected component If its endpoint set If the device is successfully attached to two or more terminals, an electrical connection is established between the corresponding terminals. Define the set of terminal-level connection edges: in, Indicator terminals With terminals There is an electrical connection between them; Perform consistency checks on the connection relationships, including: whether there are abnormal connections across unrelated logical units; whether there are floating endpoints or isolated connections; whether there are multi-branch connections with an abnormal number of terminals. Abnormal connections are marked as pending verification. S6-6, Construction of terminal-level electrical topology diagram: Based on terminal set With the set of connecting edges Constructing a terminal-level electrical topology diagram ; in, Represents a set of terminal nodes. It represents the electrical connection relationships between terminals; it maps terminal-level topology to primitive device-level topology to support system analysis and visualization at different granularities; S6-7, Standardized Digital Model Generation and Drawing Reproduction: Finally, the digital logical model obtained in step S5 is combined with... With terminal-level electrical topology diagram Generate a standardized digital model that conforms to preset data specifications.

9. A system for intelligent recognition and digital modeling of substation electrical drawings, characterized in that... The system includes: Multimodal semantic parsing module: Taking the substation electrical drawings to be processed as input, it performs joint analysis on the image content and text annotations through a lightweight multimodal large model, parses the electrical meaning of the text annotations, and obtains electrical elements, terminals and their semantic relationships; Text and graphic element spatial positioning module: used to detect and locate text annotations and electrical elements in electrical drawings using small models, and obtain their spatial coordinate information in the image; Semantic result association module: used to match the output results of the multimodal large model with the recognition results of the small model based on the similarity of text content, and associate spatial coordinate information with the corresponding text semantic results; Text-to-Electrical Element Binding Module: This module comprehensively considers the spatial distance constraints and electrical semantic consistency constraints between text annotations and electrical elements to achieve high-precision binding between text annotations and electrical elements. Relationship parsing and digital logic construction module: used to identify inclusion relationships represented by solid or dashed boxes using a multimodal large model, and to construct the logical structure of the system by combining the spatial location information of electrical elements; Electrical Wiring and Digital Model Generation Module: This module is used to construct connection links between electrical elements through connected component analysis, form a complete electrical topology, generate a standardized digital model, and support the restoration and verification of electronic drawings.

10. An electronic device, characterized in that, The device includes a memory and a processor, the memory being used to store a computer program, and the processor running the computer program to cause the electronic device to perform the method as described in any one of claims 1 to 8.

Images