Electrical engineering drawing wiring relation intelligent identification system based on multi-modal fusion

Abstract

The invention discloses an electrical engineering drawing wiring relation intelligent identification system based on multi-modal fusion, and the system comprises a file management module which is used for supporting the opening, browsing, exporting and temporary caching of a PDF file, creating an engineering file to store an identification progress, manually correcting a record and parameter configuration, and achieving breakpoint continuous processing; and cross-drawing wiring relation linkage analysis is supported. The drawing preprocessing module comprises an image optimization sub-module and a layer separation sub-module, the image optimization sub-module is used for extracting an original vector path and filtering redundant graphic elements for the vector version PDF and carrying out denoising and binarization processing for the scanning version PDF, and the layer separation sub-module is used for separating layers based on graphic attributes.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of intelligent processing of electrical engineering drawings, and more specifically relates to an electrical engineering drawing wiring relationship intelligent identification system based on multi-modal fusion. BACKGROUND

[0002] In electrical engineering design and operation, wiring relationship identification and conversion are core links (such as the link relationship of "power supply → wire → terminal → load" needs to be clear in the distribution cabinet principle diagram). Traditional technologies mainly rely on manual processing, general OCR tools or simple graphic recognition scripts, which have significant defects:

[0003] 1. Pure manual processing (mainstream traditional method)

[0004] Technical principle: engineers manually review PDF drawings, identify wires, terminal numbers and parameters by eye, and manually draw wiring tables or enter Excel.

[0005] Core shortcomings:

[0006] Low efficiency: a complex drawing (such as a distribution cabinet principle diagram containing hundreds of wires) takes weeks, and large projects (factory power distribution systems) take even longer;

[0007] High error rate: manual checking of "wire-terminal-parameter" is prone to omissions due to visual fatigue (such as misreading number tube digits or confusing wire diameters), with an error rate of 5%-10%;

[0008] Difficult to trace: modification records are scattered in Excel or paper documents, and when the drawing is changed, the whole process needs to be checked again, with high maintenance cost.

[0009] 2. General OCR tools

[0010] Technical principle: only extract text information (such as terminal number "XT1:1" and wire diameter "1.5mm²") from PDF, but cannot recognize graphics (wires and components), and cannot establish the association between text and graphics.

[0011] Core shortcomings:

[0012] "Text island" problem: the extracted parameters (such as "1.5mm²") cannot be matched to specific wires, and manual matching is still required, so the efficiency improvement is limited;

[0013] Graphics are completely ignored: cannot recognize wire direction and branch relationship, cannot build wiring topology, and cannot help with the core problem of "which wire connects which two terminals".

[0014] Simple graphic recognition tools (such as straight line detection scripts based on OpenCV)

[0015] Technical principle: Use Hough transform algorithm to detect straight lines (conductors) in PDF images, but lack targeted optimization and do not combine the professional features of electrical engineering drawings.

[0016] Core shortcomings:

[0017] High misidentification rate: The lines of the frame line, grid line and component symbol (such as the rectangular frame of the contactor coil) are misidentified as conductors, which need to be manually deleted one by one;

[0018] No topology logic: Only isolated line segments can be detected, and the start / end point and branch point (such as T-shaped and cross-shaped connections) of the conductor cannot be identified, and the output result has no actual engineering significance.

[0019] Overall, the core bottleneck of the prior art is that it fails to build a complete technical link around the "graph topology - text parameter - professional rule" trinity features of electrical engineering drawings, or only handles a single element (such as text or graphics), or lacks targeted optimization and interactive correction capabilities. This results in their inability to meet the engineering demand of "efficient and accurate extraction of wiring relationship from PDF drawings", which is the core problem that the new system solves through multi-module cooperation (preprocessing → graph recognition → text association → interactive correction). SUMMARY

[0020] In view of the shortcomings of the prior art, the present application provides an intelligent identification system for wiring relationship of electrical engineering drawings based on multi-modal fusion to solve the problems raised in the above background art.

[0021] To achieve the above purpose, the present application is implemented by the following technical scheme: an intelligent identification system for wiring relationship of electrical engineering drawings based on multi-modal fusion, comprising:

[0022] A file management module for supporting opening, browsing, exporting and temporary caching of PDF files, creating engineering files to save identification progress, manual correction records and parameter configurations, realizing breakpoint processing, and supporting cross-drawing wiring relationship linkage analysis;

[0023] A drawing preprocessing module including an image optimization submodule and a layer separation submodule, wherein the image optimization submodule is used to extract original vector paths and filter redundant graphical elements for vector version PDF, and to perform denoising and binarization processing for scanned version PDF, and the layer separation submodule is used to separate layers based on graphical attributes;

[0024] A graphic recognition module, including a wire recognition and topology analysis submodule and a component symbol positioning submodule, wherein the wire recognition and topology analysis submodule is used to extract wires, construct a wire topology graph, and process complex scenes through graphic algorithms, and the component symbol positioning submodule is used to identify common electrical component symbols and mark their pin connection points;

[0025] A text and parameter extraction module, including an identifier recognition submodule and a parameter extraction and matching submodule, wherein the identifier recognition submodule is used to extract text information and bind identifiers to corresponding components or wires, and the parameter extraction and matching submodule is used to extract wire parameters and distinguish parameter types based on text semantic analysis;

[0026] A wiring logic sorting module for automatically sorting complete loops and generating link relationships based on the wire topology graph and identifiers;

[0027] An interaction and correction module, including a visual editing submodule and a manual intervention tool submodule, wherein the visual editing submodule provides wire manual marking tools (such as supplementing unrecognized wires and deleting misrecognized lines) and identifier manual binding functions (dragging text to corresponding graphics) based on a PyQt5 graphical interface; real-time highlights of the currently recognized wire topology and parameter matching results are provided, and switching views (original drawing / topology graph / parameter annotation graph) is supported; the manual intervention tool submodule provides a parameter editing panel (manual modification of wire diameter, color, etc., supporting batch replacement, such as uniforming the color of all wires in a loop to "black"); topology adjustment (manual addition / deletion of nodes, adjustment of wire connection relationships, correction of automatically recognized errors) is supported;

[0028] A result output and export module, including a structured result display submodule and a multi-format export submodule, wherein the structured result display submodule displays the sorted results in the form of a wiring table (including loop number, starting point, ending point, wire segment number, wire diameter, color, number tube, etc., supporting sorting and filtering, such as sorting by wire diameter); a topology diagram (a simplified loop diagram, with key nodes and parameters marked, intuitively presenting the wiring logic) is generated; the multi-format export submodule supports exporting wiring relationship data to Excel tables, CSV files, JSON structured data, and BIM compatible formats (such as IFC or specific XML), meeting the needs of production, operation and maintenance, and design departments in multiple scenarios;

[0029] Each module works together to achieve the recognition and conversion of the wiring relationship of electrical engineering drawings.

[0030] As an optional scheme of the present application, the file management module comprises: a PDF file operation sub-module, configured to support opening, browsing, exporting and temporary caching of a PDF file, and adapted to large-size electrical engineering drawings; and a project management sub-module, configured to create an engineering file (.dct), save current identification progress, manual correction records and parameter configurations, realize breakpoint continuous processing, and support linkage analysis of wiring relationships across drawings.

[0031] As an optional scheme of the present application, in the drawing preprocessing module: the image optimization sub-module is configured to extract original vector paths (including lines and curves) for a vector version of the PDF, filter redundant graphic elements (such as background grids and decorative frames), and retain core contents such as wires and components; and perform denoising and binarization processing on a scanned version of the PDF.

[0032] The layer separation sub-module is configured to separate layers based on graphic attributes (such as color, line width and line type), for example, classify thick solid lines as main circuit wires, thin solid lines as control circuits, and dashed lines as shielding lines.

[0033] As an optional scheme of the present application, in the graphic recognition module: the wire recognition and topology analysis sub-module is configured to extract wires by using graphic algorithms (including line fitting and endpoint matching), distinguish wires from non-wire graphics (such as component frames), identify intersection points and branch points (such as T-shaped and cross-shaped connections), construct a wire topology graph based on a graph theory model (nodes are connection points / component pinouts, and edges are wires), and automatically complete broken wires by solving wire interruption or overlap problems through connected component analysis.

[0034] The component symbol positioning sub-module is configured to identify common electrical component symbols (such as terminal blocks, contactors, relays and fuses), position the symbols based on contour features (such as rectangular terminals and circular buttons), and mark pin connection points (as nodes of the wire topology).

[0035] As an optional scheme of the present application, in the text and parameter extraction module: the identifier recognition sub-module is configured to obtain text information (such as terminal numbers and equipment numbers) by using vector text extraction (to obtain accurate coordinates and content) and OCR (to optimize engineering character set recognition for scanned drawings), and bind identifiers (such as terminal numbers and wire number tubes) to corresponding components or wires by using a "nearest matching" algorithm (based on the distance between the coordinates of the text and the graphics).

[0036] The parameter extraction and matching sub-module is configured to extract wire parameters (such as wire diameter, color and number tube), distinguish parameter types based on text semantic analysis (keyword matching), and support parameter rule customization (such as user-specified wire diameter format).

[0037] As an optional solution of the present application, the wiring logic combing module is used to automatically comb a complete loop from a starting component (such as a power supply) to an ending component (such as a load) based on the wire topology graph and the identifier, and generate a link relationship containing "starting number-wire parameter-intermediate node-ending number".

[0038] As an optional solution of the present application, the system realizes the automatic processing of electrical engineering drawings through the cooperative process of each module, specifically including:

[0039] 1. The PDF drawing is loaded and the engineering file is created through the file management module.

[0040] 2. The image optimization and layer separation of the drawing are performed through the drawing preprocessing module.

[0041] 3. The wire topology graph is extracted and the component pin is located through the graphic recognition module.

[0042] 4. The identifier is bound and the parameter is extracted through the text and parameter extraction module.

[0043] 5. The complete loop link relationship is generated through the wiring logic combing module.

[0044] As an optional solution of the present application, the system is suitable for electrical engineering drawings mixed with vector and scanned versions, and the quality of the basic data is optimized through the noise removal, binarization and layer separation of the drawing preprocessing module, and the wire interruption or overlap problem is solved through the connected domain analysis of the graphic recognition module.

[0045] As an optional solution of the present application, the project management module supports breakpoint continuous processing and cross-drawing linkage analysis, and the modification record is saved in the engineering file, which is convenient for quick tracing and adjustment when the drawing is changed later.

[0046] As an optional solution of the present application, the parameter extraction and matching sub-module supports user-defined parameter rules (such as wire diameter format and color identification rule) to adapt to the personalized needs of different projects.

[0047] The present application provides an electrical engineering drawing wiring relationship intelligent identification system based on multi-modal fusion, which has the following beneficial effects:

[0048] 1. Significant improvement in recognition accuracy and efficiency: The drawing preprocessing module filters out redundant information through "image optimization (denoising / binarization)" and "layer separation (classification by professional attributes)", and the graphic recognition module accurately extracts conductors using a targeted algorithm (line segment fitting + endpoint matching) based on layered graphics. Compared with traditional OCR tools (high misrecognition rate) and simple graphic recognition scripts (unable to handle complex scenarios), the conductor recognition accuracy is improved to more than 95% (actual measurement data), and the time to process a complex drawing is reduced from weeks to minutes.

[0049] 2. Complete topology and parameter binding: Through "conductor topology analysis (construction of graph theory model) + identifier space correlation (nearest matching) + parameter rule customization", the accurate binding of conductor network topology graph, terminal number / code pipe parameters, and specific conductors is achieved, outputting complete link relationships containing "start point-conductor parameter-node-end point" (such as "power supply→1.5 mm² yellow conductor→terminal XT1:1→contactor→load"). This solves the problems of "text island" (parameters without corresponding conductors) and "no topology logic" (only isolated line segments are detected) in traditional technology.

[0050] 3. Adaptability to complex scenarios: The drawing preprocessing module supports both vector and scanned PDF, optimizing the quality of basic data through denoising and tilt correction. The connected component analysis of the graphic recognition module can automatically complete conductors obscured or overlapped by text, and the component symbol positioning module accurately identifies electrical component pins based on contour features, covering complex scenarios such as frequent conductor crossing / obscuring and inconsistent new and old formats in electrical drawings.

[0051] 4. Enhanced engineering practicability: The project management module supports breakpoint continuation processing and cross-drawing linkage analysis, and modification records are saved in the engineering file (.dct), facilitating quick tracing and adjustment during later drawing changes. The parameter rule customization function allows users to adjust the matching format of line diameter / color according to enterprise specifications, adapting to the individual needs of different projects. BRIEF DESCRIPTION OF DRAWINGS

[0052] Figure 1 The system module flowchart of the present application is shown in the figure. DETAILED DESCRIPTION

[0053] The embodiments of the present application will be further described in detail below in conjunction with the drawings and examples. The following examples are used to illustrate the present application, but cannot be used to limit the scope of the present application.

[0054] In the description of the present application, unless otherwise specified, the meaning of "a plurality of" is two or more; the orientation or positional relationship indicated by the terms "upper", "lower", "left", "right", "inner", "outer", "front end", "rear end", "head", "tail" and the like is based on the orientation or positional relationship shown in the drawings, and is only for the convenience of describing the present application and simplifying the description, and does not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. In addition, the terms "first", "second", "third" and the like are only for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0055] In the description of the present application, it should be noted that, unless otherwise specified and limited, the terms "connected", "connected" should be understood broadly, for example, it can be fixedly connected, or it can be detachably connected, or integrally connected; it can be mechanically connected, or it can be electrically connected; it can be directly connected, or it can be indirectly connected through an intermediate medium. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0056] Please refer to Figure 1 The present application provides a technical solution: an electrical engineering drawing wiring relationship intelligent identification system based on multi-modal fusion, the system includes the following core function modules and connection relationships:

[0057] 1. File management module, including PDF file operation submodule and project management submodule, wherein the PDF file operation submodule: supports opening and browsing of PDF files, provides basic viewing functions such as drawing scaling and local magnification (adapted to large size electrical engineering drawings), supports drawing export (such as saving as picture format) and temporary caching (improve repeated opening efficiency).

[0058] The project management submodule supports creating engineering files (.dct), saving the current identification progress, manual correction records and parameter configuration, and realizes breakpoint processing; supports cross-drawing wiring relationship linkage analysis (such as unified construction of conductor network of multiple associated drawings).

[0059] 2. Drawing preprocessing module, including image optimization submodule and layer separation submodule, wherein the image optimization submodule: for vector version PDF, extracts original vector path (line, curve), filters redundant graphic elements (such as background grid, decorative border), and retains core content such as conductor and component; for scanned version PDF, improve image clarity through denoising and binarization.

[0060] Layer separation module: separate layers based on graphic attributes (color, line width, line type) (e.g. mark thick solid lines as main circuit wires, thin solid lines as control circuits, and dashed lines as shielding lines) to simplify subsequent recognition logic.

[0061] 3. Graphic recognition module, including wire recognition and topology analysis submodule, component symbol positioning submodule, wire recognition and topology analysis submodule's wire extraction function specifically: distinguish wires from non-wire graphics (such as component borders) through graphic algorithms (combined with line fitting-merging continuous short straight lines, endpoint matching-recognizing wire start / end points);

[0062] Topology relationship construction function specifically: identify intersection points and branch points (T-shaped, cross-shaped connections), and construct wire network topology graph based on graph theory model (node = connection point / component pin, edge = wire);

[0063] Complex scene processing: solve wire interruption (e.g. blocked by text) and overlapping (e.g. parallel wires) problems through connected component analysis, and automatically complete broken wires.

[0064] Component symbol positioning submodule: identify common electrical component symbols (such as terminal blocks, contactors, relays, fuses), locate their positions based on contour features (such as rectangular terminals, circular buttons), and mark pin connection points (as nodes of wire topology).

[0065] 4. Text and parameter extraction module, including identifier recognition submodule and parameter extraction and matching submodule, wherein the function of the identifier recognition submodule is mainly:

[0066] 1) Text extraction: support vector text extraction (precise coordinates + content) and OCR (for scanned drawings, optimize engineering character set recognition), obtain text information in the drawing (such as terminal number "XT1:1", equipment number "KM1");

[0067] 2) Spatial correlation: bind identifiers to corresponding components or wires through "nearest matching" algorithm (calculate the coordinate distance between text and graphics) (such as terminal number binding to terminal pin, wire number tube binding to wire segment).

[0068] And the function of the parameter extraction and matching submodule is: extract wire parameters such as wire diameter (e.g. "1.5mm²"), color (e.g. "yellow green"), and number tube (e.g. "W123"), distinguish parameter types based on text semantic analysis (keyword matching); support parameter rule customization (e.g. user preset "wire diameter format = number + mm²"), improve matching accuracy.

[0069] 5. The wiring logic parsing module is specifically a loop relationship analysis submodule. The main function of the loop relationship analysis submodule is to automatically parse complete loops (from the starting component such as a power supply to the ending component such as a load) based on the conductor topology graph and identifiers, and generate a link relationship of "starting number-conductor parameter-intermediate node-ending number" (such as "power supply L1→1.5 mm² yellow conductor→terminal XT1:1→contactor KM1→load M1").

[0070] 6. The interaction and correction module includes a visual editing submodule and a manual intervention tool submodule. The visual editing submodule is based on a PyQt5 graphical interface and provides a conductor manual marking tool (such as supplementing unrecognized conductors and deleting misidentified lines), an identifier manual binding function (dragging text to corresponding graphics), real-time highlighting of currently recognized conductor topology and parameter matching results, and support for switching views (original drawing / topology graph / parameter annotation graph). The manual intervention tool submodule provides a parameter editing panel (manual modification of line diameter, color, etc., with support for batch replacement, such as uniform "black" for all conductors in a loop), topology adjustment (manual addition / deletion of nodes, adjustment of conductor connection relationships, and correction of automatically recognized errors), and other functions.

[0071] 7. The result output and export module includes a structured result display submodule and a multi-format export submodule. The structured result display submodule displays the parsed results in the form of a wiring table (including loop number, starting point, ending point, conductor segment number, line diameter, color, number tube, etc., with support for sorting and filtering, such as sorting by line diameter), generates a topology diagram (a simplified loop diagram with key nodes and parameters labeled, providing an intuitive presentation of wiring logic), and supports exporting wiring relationship data to Excel tables, CSV files, JSON structured data, and BIM-compatible formats (such as IFC or specific XML), meeting the needs of production, operation, and design departments in various scenarios.

[0072] Specific usage and effects of the embodiment:

[0073] Step 1: Start the system and import the target drawing (file management module)

[0074] Operation method:

[0075] (1) The user opens the system client (supports Windows / Linux operating systems), logs in after entering the main interface;

[0076] (2) Click "New Project" or "Open Existing Project" on the main interface, and select the electrical engineering PDF drawing file to be processed (supports single or multiple associated drawings, such as main distribution cabinet drawings and sub-distribution cabinet drawings);

[0077] (3) The system automatically calls the file management module to complete the drawing loading through the PDF file operation submodule. Users can perform scaling, local enlargement and other operations through the built-in browsing tool (adapt to large-size electrical drawing detail viewing).

[0078] If it is a new project, the system pops up a prompt box to require creating a project file (the default format is.dct). After the user confirms, the project management submodule initializes the project environment, generates a unique project ID and saves the initial state (including identifying progress, parameter configuration template, etc.). If it is an existing project, the historical project file is directly loaded to restore the progress and configuration of the last processing.

[0079] Step 2: Drawing preprocessing (drawing preprocessing module)

[0080] Operation mode:

[0081] (1) The system automatically triggers the drawing preprocessing module without user manual intervention, but users can view the preprocessing progress and intermediate results (such as optimized drawing preview) through the interface;

[0082] (2) The image optimization submodule performs differential processing according to the drawing type (vector version / scanned version):

[0083] If it is a vector PDF (such as a vector drawing exported by CAD), the original vector path (including line, curve and other core graphic elements) is extracted, and redundant information (such as background grid, decorative border, non-electrical related annotation box) is filtered;

[0084] If it is a scanned PDF (such as a scanned copy of paper drawings), first, denoising is performed (remove noise and blurred edges generated during scanning), and then binarization (convert the image to a clear black and white format) is performed to improve the distinction between the lead and the background;

[0085] The layer separation submodule separates the preprocessed graphics based on graphic attributes (color, line width, line type): for example, thick solid lines (usually representing main circuit leads) are separated into one layer, thin solid lines (control circuit leads) are separated into one layer, and dashed lines (shielding lines or auxiliary lines) are separated into one layer. Different colored leads (such as red fire wire and blue zero wire) can also be independently layered.

[0086] Step 3: Graphic and component recognition (graphic recognition module)

[0087] Operation mode:

[0088] (1) The system calls the graphic recognition module to sequentially perform lead recognition and component positioning;

[0089] (2) The lead recognition and topology analysis submodule processes the preprocessed layered graphics through graphic algorithms:

[0090] First, the line segment fitting technique is used to merge consecutive short straight line segments (such as slightly disconnected wire segments), and then the end point matching algorithm is used to identify the starting and ending points of the wire (such as the connection points of the wire and the device pins);

[0091] Based on the graph theory model, the wire topology graph is constructed - the wire intersection points and branch points (such as T-shaped connections: one wire branches out into two branches; cross-shaped connections: two wires intersect and conduct) are defined as "nodes", and the wires themselves are defined as "edges", and finally a complete wire network logical structure is formed;

[0092] For complex scenes (such as wires interrupted by text annotations, and parallel wires overlapping), the connected component analysis technique is used to automatically complete the broken wire segments (such as inferring the wire direction of the occluded part by the continuity of adjacent pixels);

[0093] The component symbol positioning sub-module identifies common electrical component symbols (such as terminal blocks, contactors, relays, fuses) in the drawing based on the pre-trained electrical component contour feature library (such as the rectangular contour of the terminal block, the circular / square coil contour of the contactor, the contact structure of the relay, etc.), and accurately locates their pin connection points (as key nodes in the wire topology graph).

[0094] Step 4: Text and parameter extraction and binding (text and parameter extraction module)

[0095] Operation mode:

[0096] (1) The system calls the text and parameter extraction module to synchronously process the text information and parameter annotations in the drawing;

[0097] (2) The identifier recognition sub-module extracts text information through two techniques:

[0098] For vector PDF (containing directly extractable vector text), get the accurate coordinates and content of the text (such as terminal number "XT1:1", device number "KM1");

[0099] For scanned PDF (requires OCR recognition), optimize the engineering character set recognition (for common combinations of numbers, letters, and symbols in the electrical field, such as "1.5mm²" "W123"), and improve the text recognition accuracy;

[0100] Through the "nearest matching" algorithm (calculate the coordinate distance between the text and the graphical elements, such as the text closest to the terminal pin and meeting the naming rules is automatically bound to the pin), the identifiers (such as terminal numbers, wire number tubes) are bound to the corresponding device pins or wire segments;

[0101] The parameter extraction and matching sub-module extracts wire-related parameters (such as wire diameter "1.5 mm²", color "yellow-green", and number tube "W123") from the text, distinguishes parameter types based on text semantic analysis (keyword matching: such as "mm²" for wire diameter and "yellow-green" for ground wire color), and supports users to customize rules through a parameter configuration interface (for example, presetting "wire diameter format = number + mm²" and "color keyword = red / yellow / green") to further improve matching accuracy.

[0102] Step 5: Wiring logic sorting and result output (wiring logic sorting module)

[0103] Operation method:

[0104] (1) The system calls the wiring logic sorting module to automatically generate complete wiring relationships based on the constructed wire topology graph, identifier binding information, and parameter data.

[0105] (2) The module starts from the starting point component (such as the power supply inlet end L1 / L2 / L3), traces along the "edges" (wires) of the wire topology graph, sequentially passes through intermediate nodes (such as terminal blocks and contactor contacts), and finally reaches the end point component (such as the load motor M1), generating a standardized link relationship description (for example: "power supply L1 → 1.5 mm² yellow wire → terminal XT1:1 → contactor KM1 main contact → 2.5 mm² blue wire → load M1").

[0106] The final result is output in two forms: a visual interface and structured data.

[0107] Visual interface: displays the topology graph of the wire network (nodes are component pins, and lines are wires, with parameters such as wire diameter / color labeled), and users can interactively view the link details of any loop.

[0108] Structured data: exported as an Excel table or JSON file, containing fields such as "starting point number, wire parameters (wire diameter / color), intermediate nodes, and end point number", which can be directly used for production wiring tables, operation and maintenance manuals, or BIM system integration.

[0109] Step 6: Semi-automatic interactive correction (interactive and correction module)

[0110] Operation method:

[0111] (1) Users enter the interactive and correction module (the core is the semi-automatic correction function) through the main interface, and the system automatically loads the initial wiring logic results and corresponding original drawings / topology graphs / parameter labeling diagrams.

[0112] The visual editing sub-module (based on the PyQt5 graphical interface) provides the following tools:

[0113] 1) Manual wire marking tool: Users can use the mouse to click or draw tools to add key wires that the system does not recognize (such as auxiliary wires that are obscured by complex backgrounds), or select misidentified lines (such as misidentifying component borders as wires) and delete them;

[0114] 2) Manual Identifier Binding Function: Supports drag-and-drop operation—drag text identifiers (such as terminal number "XT2:3") that are not correctly bound in the drawing to the corresponding component pins or wire nodes to achieve precise association;

[0115] 3) Real-time view switching: Users can switch between viewing the original drawing (original PDF image), topology diagram (logical structure of the conductor network), or parameter annotation diagram (binding relationship between conductors and parameters) to help judge and correct accuracy; the system highlights the currently identified conductor topology (green for correct connection, yellow for pending verification) and parameter matching results in real time (blue for successful parameter type matching, red for conflict).

[0116] The manual intervention tools submodule provides the following functions:

[0117] 1) Parameter editing panel: Users can directly modify wire parameters (such as changing the wire diameter from "1.5mm²" to "2.5mm²" or changing the color from "yellow-green" to "yellow"), or uniformly adjust the properties of all wires in a certain circuit through the batch replacement function (such as unifying the color of all wires in a certain circuit to "black").

[0118] 2) Topology adjustment function: Supports manual addition / deletion of nodes (such as adding a connection point or deleting misidentified branch nodes), adjustment of wire connection relationship (such as adjusting the misconnected wire from terminal XT1:1 to the correct XT2:3), and correction of automatic identification error logic (such as misidentifying a T-type connection as a straight connection).

[0119] Step 7: Structured Result Output and Export (Result Output and Export Module)

[0120] Operation method:

[0121] (1) After the user confirms the correction, the system will proceed to the result output and export module, where the system provides the following output options:

[0122] The structured results display submodule presents the final sorting results in the form of a wiring table, including fields such as loop number, start point, end point, number of conductor segments, wire diameter, color, and tube number. It supports sorting and filtering by conditions such as wire diameter, color, and start point (e.g., filtering out all conductor loops with a wire diameter greater than "2.5mm²"). It generates a topology diagram (simplified loop diagram), annotates key nodes (such as power supply, load, and terminal block) and core parameters (such as wire diameter and color), and intuitively presents the wiring logic.

[0123] Multi-format export sub-module, supports export to the following formats:

[0124] Excel table (for production department to make wiring table, including complete fields and filtering function);

[0125] CSV file (compatible with other data analysis tools);

[0126] JSON structured data (for system integration or secondary development);

[0127] BIM compatible format (such as IFC or specific XML, for building design department to integrate into BIM model).

[0128] Users can choose to export all data or filtered specific loop data according to their needs (such as exporting only corrected link relationships).

[0129] The apparatus embodiments described above are merely illustrative, wherein the units illustrated as separate components can or can not be physically separated, and the components illustrated as units can or can not be physical units, i.e., they can be located in one place, or distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the embodiments according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0130] Finally, it should be noted that: the above examples are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing examples, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A multi-modal fusion-based electrical engineering drawing connection relationship intelligent identification system, characterized in that, The application comprises the following technical solutions: A file management module is used to support opening, browsing, exporting and temporary caching of PDF files, to create project files to save identification progress, manual correction records and parameter configurations, to realize breakpoint continuation processing, and to support cross-drawing linkage relationship analysis; A drawing preprocessing module comprises an image optimization submodule and a layer separation submodule, wherein the image optimization submodule is used to extract original vector paths and filter redundant graphic elements for vector version PDF, and to perform denoising and binarization processing for scanned version PDF, and the layer separation submodule is used to separate layers based on graphic attributes; A graphic recognition module comprises a conductor recognition and topology analysis submodule and a component symbol positioning submodule, wherein the conductor recognition and topology analysis submodule is used to extract conductors, construct conductor topology graphs and process complex scenarios through graphic algorithms, and the component symbol positioning submodule is used to identify common electrical component symbols and mark their pin connection points; A text and parameter extraction module comprises an identifier recognition submodule and a parameter extraction and matching submodule, wherein the identifier recognition submodule is used to extract text information and bind identifiers to corresponding components or conductors, and the parameter extraction and matching submodule is used to extract conductor parameters and distinguish parameter types based on text semantic analysis; A wiring logic sorting module is used to automatically sort complete loops and generate link relationships based on conductor topology graphs and identifiers; An interaction and correction module comprises a visual editing submodule and a manual intervention tool submodule, wherein the visual editing submodule provides a conductor manual marking tool (such as supplementing unrecognized conductors and deleting misrecognized lines), an identifier manual binding function (dragging text to corresponding graphics) based on a PyQt5 graphic interface, real-time highlight display of currently recognized conductor topology and parameter matching results, and support for switching views (original drawing / topology graph / parameter annotation drawing); the manual intervention tool submodule provides a parameter editing panel (manual modification of line diameter, color and other information, support for batch replacement, such as uniform "black" color for all conductors in a loop), support for topology adjustment (manual addition / deletion of nodes, adjustment of conductor connection relationship, correction of automatically recognized error logic) A result output and export module comprises a structured result display submodule and a multi-format export submodule, wherein the structured result display submodule displays sorting results in the form of a wiring table (including loop number, starting point, ending point, conductor segment number, line diameter, color, number tube and other fields, support for sorting and filtering, such as sorting by line diameter), generates a topology schematic diagram (a simplified loop diagram, key nodes and parameters are marked, and wiring logic is intuitively presented), and the multi-format export submodule supports exporting wiring relationship data into an Excel table, a CSV file, JSON structured data and BIM compatible formats (such as IFC or specific XML), meeting the needs of production, operation and maintenance and design departments in multiple scenarios; Wherein, the modules work together to realize the wiring relationship recognition and conversion of electrical engineering drawings.

2. The multi-modal fusion-based electrical engineering drawing connection relationship intelligent identification system according to claim 1, characterized in that, The file management module comprises a PDF file operation sub-module for supporting opening, browsing, exporting and temporary caching of a PDF file, and adapting to large-size electrical engineering drawings; and a project management sub-module for creating an engineering file (.dct), saving current identification progress, manual correction records and parameter configurations, realizing breakpoint continuous processing, and supporting linkage analysis of wiring relationships across drawings. 3.The multi-modal fusion based electrical engineering drawing connection relationship intelligent recognition system according to claim 1, characterized in that, In the drawing preprocessing module: The image optimization sub-module is used for extracting original vector paths (including lines and curves) for a vector version of the PDF, filtering redundant graphic elements (such as background grids and decorative borders), and retaining core contents such as wires and components; and is used for denoising and binarization processing of a scanned version of the PDF; The layer separation sub-module is used for separating layers based on graphic attributes (such as color, line width and line type) (for example, thick solid lines are classified as main circuit wires, thin solid lines are classified as control circuits, and dashed lines are classified as shielding lines).

4. The multi-modal fusion-based electrical engineering drawing connection relationship intelligent identification system according to claim 1, characterized in that, In the graphic recognition module: The wire recognition and topology analysis sub-module is used for extracting wires through graphic algorithms (including line fitting and endpoint matching), distinguishing wires from non-wire graphics (such as component borders), identifying intersection points and branch points (such as T-shaped and cross-shaped connections), constructing a wire topology graph based on a graph theory model (nodes are connection points / component pins, and edges are wires), and automatically completing broken wires by solving wire interruption or overlap problems through connected domain analysis; The component symbol positioning sub-module is used for identifying common electrical component symbols (such as terminal blocks, contactors, relays and fuses), positioning their locations based on contour features (such as rectangular terminals and circular buttons), and marking pin connection points (as nodes of the wire topology).

5. The multi-modal fusion-based electrical engineering drawing connection relationship intelligent identification system according to claim 1, characterized in that, In the text and parameter extraction module: The identifier recognition sub-module is used for obtaining text information (such as terminal numbers and equipment numbers) through vector text extraction (to obtain accurate coordinates and content) and OCR (to optimize engineering character set recognition for scanned drawings), and binding identifiers (such as terminal numbers and wire number tubes) to corresponding components or wires through a "nearest matching" algorithm (based on the distance between the coordinates of the text and the graphics). The parameter extraction and matching sub-module is used for extracting wire parameters (such as wire diameter, color and number tube), distinguishing parameter types based on text semantic analysis (keyword matching), and supporting parameter rule customization (such as user-specified wire diameter format).

6. The multi-modal fusion-based electrical engineering drawing connection relationship intelligent identification system according to claim 1, characterized in that, The wiring logic sorting module is used for automatically sorting complete circuits from a starting component (such as a power supply) to an ending component (such as a load) based on the wire topology graph and the identifiers, and generating link relationships containing "starting number-wire parameter-intermediate node-ending number".

7. The system according to any one of claims 1-6, characterized in that, The system realizes automatic processing of electrical engineering drawings through the collaborative processes of the modules, specifically including:

1. loading a PDF drawing and creating an engineering file through the file management module; 2. performing image optimization and layer separation on the drawing through the drawing preprocessing module; 3. extracting a wire topology graph and positioning component pins through the graphic recognition module; 4. binding identifiers and extracting parameters through the text and parameter extraction module; 5. Generate complete loop link relationship through wiring logic combing module.

8. The system according to any of claims 1-6, characterized in that, The system is suitable for electrical engineering drawings mixed with vector and scanned versions. The quality of basic data is optimized through denoising, binarization and layer separation of the drawing preprocessing module, and the problem of conductor interruption or overlap is solved through connected domain analysis of the graph recognition module.

9. The system according to any of claims 1-6, characterized in that, The project management module supports breakpoint continuous processing and cross-drawing linkage analysis, and the modification records are saved in the engineering file, facilitating quick tracing and adjustment during later drawing changes.

10. The system of any of claims 1-6, wherein, The parameter extraction and matching sub-module supports user-defined parameter rules (such as line diameter format, color identification rule) to adapt to the individual needs of different projects.

Images