Electrical equipment three-dimensional geometric model automatic construction system based on CAD drawing

Abstract

The invention discloses an electrical equipment three-dimensional geometric model automatic construction system based on a CAD drawing, relates to the field of digital modeling, and is used for realizing full-process software and hardware cooperation from drawing input to electrical equipment three-dimensional geometric model automatic output. A user logs in the system, a CAD drawing is imported, a drawing analysis module calls a corresponding drawing processing unit to preprocess the CAD drawing according to the format of the CAD drawing, and structured analysis data is obtained through analysis according to the preprocessed CAD drawing; the coordinate correction module receives a correction instruction of a user and performs coordinate correction on the structured analysis data; the model generation module identifies the type of the electrical equipment according to the structured analysis data, and calls a corresponding model generation unit to generate an electrical equipment model; and the data export and storage module exports or stores the electrical equipment model. According to the method, the drawings in different formats are modeled through a unified framework, the modeling efficiency is high, the precision is high, and the digital twin platform can be directly connected.

Description

Technical Field

[0001] This invention relates to the field of digital modeling technology, and in particular to an automatic system for constructing three-dimensional geometric models of electrical equipment based on CAD drawings. Background Technology

[0002] Existing 3D modeling technologies for electrical equipment largely rely on scattered software tools or manual operation, lacking integrated system devices to support full-process automation, resulting in the following structural and functional integration defects: 1. Dispersed modules without a unified architecture: In the traditional modeling process, drawing recognition requires running image processing software separately, modeling requires calling 3D design tools, and accuracy verification relies on third-party measurement programs. There is no structured connection between the various links, and manual intervention is required for data transmission, resulting in low efficiency.

[0003] 2. Lack of dedicated hardware collaboration in multi-format drawing processing: For the recognition of raster drawings (PNG / JPG) and vector drawings (DXF / DWG), existing technologies do not have dedicated parallel processing units and rely solely on general-purpose computer software. This can easily lead to data congestion and poor robustness when processing complex drawings.

[0004] 3. Lack of integrated structure between modeling and accuracy calibration: The 3D modeling module and the accuracy calibration module are independent of each other. Calibration data needs to be manually imported into the modeling tool, which cannot achieve a real-time closed loop of "modeling-calibration-optimization", resulting in the model accuracy depending on manual operation.

[0005] 4. Lack of standardized interfaces for interaction and output: The existing system lacks a unified visual interaction hardware interface and industry standard format output unit. The model cannot be directly connected to the digital twin platform, and additional format conversion equipment is required, which increases engineering costs.

[0006] Although known 3D modeling techniques may involve functions that optimize some of the above-mentioned problems, none of them have solved the problem of "multi-module integrated structure". They still have the defects of fragmented processes and insufficient hardware and software coordination. Summary of the Invention

[0007] The purpose of this invention is to provide an automatic construction system for three-dimensional geometric models of electrical equipment based on CAD drawings, addressing all or part of the problems mentioned above, so as to achieve full-process hardware and software collaboration from drawing input to automatic output of three-dimensional geometric models of electrical equipment.

[0008] The technical solution adopted in this invention is as follows: An automatic construction system for three-dimensional geometric models of electrical equipment based on CAD drawings, comprising: The user login module is configured to verify the user account and then proceed to the drawing parsing module after successful verification. The drawing parsing module is configured to: receive CAD drawings input by the user, preprocess the CAD drawings according to the format of the CAD drawings by calling the corresponding drawing processing unit, and parse the preprocessed CAD drawings to obtain structured parsing data; The coordinate correction module is configured to: receive correction instructions from the user and perform coordinate correction on the structured parsed data; The model generation module is configured to: identify the type of electrical equipment based on the structured parsing data, or based on the structured parsing data after coordinate correction, and generate an electrical equipment model by calling the corresponding model generation unit according to the type of electrical equipment; The data export and storage module is configured to: receive a user's export command to export the electrical equipment model, and / or receive a user's storage command to store the electrical equipment model in the user's account storage space.

[0009] Furthermore, the user login module is also configured as follows: User account credentials are stored using SHA-256 encryption.

[0010] Furthermore, the CAD drawings include two formats: vector graphics and raster graphics; the drawing processing unit includes a vector graphics processing unit and a raster graphics processing unit, wherein: The vector graphics processing unit preprocesses the vector graphics paper according to the following configuration: identifying the layers and entities of the vector graphics paper, matching the layers and entities according to the set regular expressions; caching the layers using the Least Recently Used (LRU) caching mechanism; and removing redundant layers based on the identified entities. The raster image processing unit preprocesses the raster image according to the following configuration: smoothing the noise of the raster image through Gaussian filtering; performing edge detection on the raster image based on the Canny operator; performing contour segmentation based on local variance adaptive threshold; repairing missing contours through morphological opening and closing operations; and establishing a pixel-geometric coordinate affine transformation mapping relationship.

[0011] Furthermore, the drawing parsing module parses the structured parsing data according to the following configuration: Perform multi-scale edge detection on preprocessed CAD drawings; The detected contour points are simplified by curvature weighting to obtain feature points; Establish a coordinate system for the feature points; The feature points are encapsulated as structured data.

[0012] Furthermore, before parsing the structured parsing data, the drawing parsing module first initializes the thread pool according to the input CAD drawing format and quantity; different CAD drawing formats are parsed in parallel using different threads.

[0013] Furthermore, the coordinate correction module performs coordinate correction on the structured parsed data according to the following configuration: The structured parsed data is visualized. Receive correction commands from users to delete, reset, or undo feature point coordinates, and update the coordinates of the corresponding feature points; Store structured parsed data with modified coordinates.

[0014] Furthermore, the coordinate correction module executes correction instructions for deleting feature point coordinates according to the following configuration: The curvature of the feature points is checked against preset curvature constraints. If the check passes, the feature point coordinates are retained; otherwise, the feature point coordinates are deleted.

[0015] Furthermore, the electrical equipment type includes rotating body type and combined body type; the model generation module includes rotating body generation unit and combined body generation unit, wherein: The rotating body generation unit generates an electrical equipment model according to the following configuration: scanning and detecting the rotation axis based on the image center axis; fitting the rotation parameters using the least squares method, converting the two-dimensional feature points into polar coordinate parameters, constructing a feature point cloud using equal-angle sampling, and generating a continuous surface by rotating 360° around the rotation axis; dividing the mesh using a ring connection strategy and converting it into a triangular mesh structure. The assembly generation unit generates an electrical equipment model according to the following configuration: extracting and simplifying multi-view contours through adaptive contour detection and Douglas-Puk algorithm, the contours including geometric feature dimensions; dynamically adjusting the stretching parameters based on multi-view projection relationships to generate basic columns through solid stretching; constructing rotation matrices through Euler-Rodriguez formula to complete multi-coordinate system spatial correction of each basic column; and fusing multi-view meshes using Boolean intersection operation.

[0016] Furthermore, the model generation module also includes a calibration unit; the calibration unit is configured to: Data extraction and calibration: Based on the dimensions marked on the CAD drawings, the extracted geometric feature dimensions are compared in real time, and data with deviations exceeding the first threshold are extracted and corrected a second time. Modeling process control: The tolerance threshold level is dynamically adjusted during the modeling of the combined body generation unit; the rotation parameters are optimized through multiple iterative fitting during the modeling of the rotated body generation unit. Output accuracy verification: Automatically measure the dimensions of electrical equipment models, evaluate the modeling accuracy, and correct electrical equipment models whose measured dimensions deviate from the labeled dimensions by more than the first threshold.

[0017] Furthermore, the data export and storage module has a built-in STEP / IGES format conversion hardware interface to interface with the digital twin platform, allowing the electrical equipment model to be directly imported into the digital twin platform.

[0018] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: 1. Modular structure improves maintainability: The system adopts a layered modular architecture, and each functional module can be installed, removed and upgraded independently. When replacing the drawing processing sub-unit or optimizing the modeling module, there is no need to reconstruct the entire system. 2. Parallel processing improves efficiency: The dedicated processing subunit for raster / vector paper works in parallel through a data buffer bus. Experiments have shown that the processing time for a single complex drawing is reduced to within 5 minutes. 3. Closed-loop structure ensures accuracy: The modeling module and the accuracy calibration module interact directly through a real-time data interface. The calibration data is automatically fed back to the modeling unit to adjust the parameters, and the model size error is stably controlled within 0.01mm. 4. Standardized interfaces enhance compatibility: Built-in STEP / IGES format conversion hardware interface and digital twin platform docking unit, the model can be directly imported into platforms such as Unity and Digital Twin Hub without the need for additional conversion equipment. Attached Figure Description

[0019] The present invention will be described by way of example and with reference to the accompanying drawings, wherein: Figure 1 This is a construction diagram of an automatic system for building 3D geometric models of electrical equipment based on CAD drawings.

[0020] Figure 2 This is the system's operation flowchart.

[0021] Figure 3 This is the user login module interface.

[0022] Figure 4 This is the flowchart of the drawing parsing module.

[0023] Figure 5 This is the input of the original raster drawing.

[0024] Figure 6 Yes Figure 5 A diagram showing the raster drawing after Gaussian filtering.

[0025] Figure 7 This is the flowchart of the coordinate correction module.

[0026] Figure 8 This is the flowchart of the model generation module.

[0027] Figure 9This is a schematic diagram of a typical solid of revolution.

[0028] Figure 10 This is a diagram of the three-dimensional geometric model of the JYZ insulator 2-ZS-110 generated by the model generation module.

[0029] Figure 11 This is a diagram of the 3D geometric model of the GIS CB3 Phase 1 components generated by the model generation module.

[0030] Figure 12 This is a demonstration diagram of Boolean modeling for a typical column.

[0031] Figure 13 Yes Figure 12 The modeling results are shown in the figure.

[0032] Figure 14 This is a grid alignment diagram of a 1000KVA transformer.

[0033] Figure 15 Yes Figure 14 The modeling results are shown in the figure.

[0034] Figure 16 This is the original 3D model of a 1000KVA transformer. Detailed Implementation

[0035] All features disclosed in this specification, or all steps in all disclosed methods or processes, may be combined in any way, except for mutually exclusive features and / or steps.

[0036] Any feature disclosed in this specification (including any appended claims and abstract) may be replaced by other equivalent or similar features, unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is merely one example of a series of equivalent or similar features.

[0037] To address the problems of fragmented modules, insufficient hardware-software collaboration, and unstructured data integration in current 3D modeling systems for constructing 3D models of electrical equipment, this application provides an automatic 3D geometric model construction system for electrical equipment based on CAD drawings. The system aims to achieve end-to-end hardware-software collaboration from drawing input to automatic output of the 3D geometric model of the electrical equipment. Based on this objective, the proposed solution aims to solve the following problems: 1. How to build a modular integrated technology system that integrates drawing recognition, 3D modeling, accuracy calibration, and visualization interaction functions to avoid efficiency losses caused by fragmented processes; 2. How to design a dual-path recognition technology for multi-format drawings to achieve efficient feature extraction from raster and vector drawings and improve the robustness of complex drawing processing; 3. How to establish a closed-loop technical mechanism for modeling and calibration, enabling direct data interaction between accuracy calibration and the modeling process without manual intervention, thus ensuring model accuracy; 4. How to set up standardized technical interfaces to enable direct connection between the model and industry standard formats (STEP / IGES) and digital twin platforms, reducing additional conversion steps.

[0038] Based on the above considerations, such as Figure 1 As shown in the figure, the automatic construction system for 3D geometric models of electrical equipment based on CAD drawings proposed in this application includes the following modules. These modules work collaboratively based on the PyQt5 framework and interact with each other via JSON data. The operational flow between the modules is as follows: Figure 2 As shown.

[0039] (1) User login module.

[0040] The user login module is used to verify the legitimacy of a user's identity and to associate the user account with the electrical equipment model constructed by that user.

[0041] Specifically, the user login module is configured to verify user accounts.

[0042] A user account includes a username and password. The user login module displays the login interface, such as... Figure 3 As shown, the system receives the username and password entered by the user and matches and verifies them against the user account stored in the backend of the user login module.

[0043] In addition, to improve the convenience of user login while ensuring the security of user accounts, the user login module also has the function of storing user account credentials, such as... Figure 3 In the settings, you can select "Remember me" to save your username and password. Furthermore, the user login module uses SHA-256 encryption to store user account credentials, meaning the password is stored as a 256-bit hash value instead of plaintext, ensuring user account security.

[0044] In addition, the user login module is configured to automatically enter the drawing parsing module after successful verification. That is, after the user account is verified, the user will automatically enter the interface for automatically building a 3D geometric model.

[0045] II. Drawing Analysis Module.

[0046] The drawing parsing module is used to parse structured analytical data for constructing three-dimensional geometric models from two-dimensional CAD drawings, including raster drawings (such as PNG, JPG, etc.) or vector drawings (such as DXF, DWG, etc.).

[0047] Specifically, the drawing parsing module is configured to: receive CAD drawings input by the user, preprocess the CAD drawings according to their format by calling the corresponding drawing processing unit, and parse the preprocessed CAD drawings to obtain structured parsing data.

[0048] As an optional implementation method, such as Figure 4 As shown, the drawing parsing module is configured with the following process to perform the parsing operation on CAD drawings: S21. After receiving CAD drawings input by the user (such as uploading, dragging and dropping, etc.), initialize the thread pool.

[0049] The thread pool contains multiple threads, each of which independently performs related operations or calculations. The threads run in parallel, thus allowing multiple CAD drawings to be parsed simultaneously when inputting multiple CAD drawings, thereby improving modeling efficiency.

[0050] As an optional implementation, in step S11, a thread pool is initialized based on the format and quantity of the input CAD drawings; different formats of CAD drawings are parsed in parallel using different threads. For example, the number of threads can be initialized based on the number of CAD drawing format categories, with each format of CAD drawing being parsed in the same thread; alternatively, the number of threads can be initialized based on the quantity of CAD drawings, with each CAD drawing being parsed in a separate thread; or, the thread pool can be initialized simultaneously based on both the CAD drawing format and quantity. The number of threads is first initialized based on the number of CAD drawing formats. If the number of CAD drawings of a certain format exceeds a certain value (e.g., 2 or 3), then that value is used as the maximum number of CAD drawings that a thread can process, and a new thread is created for each additional number of CAD drawings exceeding that value. For example, if the input quantity of raster drawings is 5, the input quantity of vector drawings is 4, and the maximum number of drawings allowed per thread is 2, then 3 threads are created for raster drawings, and 2 threads are created for vector drawings, for a total of 5 threads are initialized.

[0051] S22. Perform image preprocessing on CAD drawings.

[0052] CAD drawings come in two formats: vector graphics and raster graphics. Different formats require different preprocessing methods. As an optional implementation, the drawing processing unit includes a vector graphics processing unit and a raster graphics processing unit.

[0053] The vector graphics processing unit preprocesses vector graphics paper according to the following configuration: (1) Identify vector paper layers (such as geometric shapes, annotation text, etc.) and entities (such as lines, surfaces, and volumes), and perform layer and entity matching according to the set regular expressions.

[0054] Regular expressions define matching patterns, and the identified layers and entities are matched and filtered according to the set regular expressions to improve the accuracy and efficiency of filtering.

[0055] (2) Cache the layer using the Least Recently Used (LRU) caching mechanism.

[0056] Layers with high access frequency are typically stored first. During the filtering process, layers that have already been processed are cached to reduce redundant calculations. Furthermore, the LRU cache size can be dynamically adjusted based on the number of identified layers (e.g., linearly adjusted according to a certain ratio) to balance resource consumption and preprocessing performance.

[0057] (3) Remove redundant layers based on the identified entities.

[0058] The system scans and identifies entities, as well as redundant layers such as duplicate lines and invalid annotation text, to streamline CAD drawings while ensuring that critical information is not lost.

[0059] The raster image processing unit preprocesses raster drawings according to the following configuration: (1) Smooth the noise in the raster paper by Gaussian filtering.

[0060] like Figure 5 The image shown is the imported original raster image, which, after being filtered by the raster image processing unit, yields the following result: Figure 6 The raster drawing shown.

[0061] (2) Edge detection of raster paper based on Canny operator.

[0062] (3) Contour segmentation based on adaptive thresholding of local variance. In the target object region, the gray-level change is relatively small, and the local variance is also small. However, at the target contour boundary, the gray-level changes between the target and the background are abrupt, and the local variance is large. By calculating the local variance in the neighborhood of each pixel to determine the segmentation threshold of that pixel, it is possible to better adapt to the changes in local image characteristics and achieve more accurate contour segmentation.

[0063] (4) Repairing the missing contour through morphological opening and closing operations. In specific operation, the missing contour is repaired by erosion or dilation operations; the opening operation is to first erode and then dilate to eliminate noise; the closing operation is to first dilate and then erode to fill the fracture.

[0064] (5) Establish the pixel-geometric coordinate affine transformation mapping relationship. Constructing this mapping relationship can effectively reduce the amount of data storage and image complexity, and reduce the computational pressure of subsequent processes.

[0065] S23. Perform multi-scale edge detection on the preprocessed CAD drawings.

[0066] Multi-scale edge detection involves constructing multiple scale spaces to perform edge detection on CAD drawings, and then combining the detected contour points from each scale space. Smaller scale spaces can improve edge detection accuracy, while larger scale spaces can effectively suppress noise. Common multi-scale edge detection methods include Gaussian-Laplacian multi-scale detection and multi-scale wavelet transform.

[0067] S24. Perform curvature-weighted contour simplification on the detected contour points to obtain feature points.

[0068] As an optional implementation, the curvature-weighted contour simplification process includes the following steps: (1) Calculation of curvature at contour points.

[0069] Calculate the curvature value for each contour point. A larger curvature value indicates a more significant curvature of the contour at that feature point, which can be used as a geometric feature point for modeling the entity.

[0070] (2) Contour point weight allocation.

[0071] Weights are assigned based on the curvature values ​​of the contour points. Points with larger curvature values ​​are assigned higher weights and are more likely to be retained in subsequent simplification processes.

[0072] (3) Simplified processing of contour points.

[0073] Based on a preset tolerance threshold, a weighted strategy (such as the Douglas-Puk algorithm) is used to simplify the contour points. Contour points with low weight (below the tolerance threshold) are deleted first, while contour points with high weight (above the tolerance threshold) are retained as feature points.

[0074] S25. Establish a coordinate system for the feature points.

[0075] The CAD drawing may contain multiple components, and the coordinate systems used for feature point extraction and recognition may be different. Here, we will unify the coordinate systems.

[0076] S26. Encapsulate the feature points into structured data.

[0077] Feature point information includes feature point identifiers, coordinates, grayscale values, etc. Discrete feature points are encapsulated into structured data through a set encapsulation format to facilitate transmission and recognition.

[0078] S27. Perform validity verification on the encapsulated structured data.

[0079] The so-called validity check involves performing consistency checks, encapsulation logic checks, and integrity checks on the encapsulated structured data to ensure the correctness and reliability of the encapsulated data. If the validity check passes, the structured parsed data is output; otherwise, an error message indicating that the check failed is output.

[0080] By integrating multi-scale edge detection, curvature-weighted contour simplification, coordinate system unification, and structured data encapsulation, standardized structured data is output, providing a unified input source for subsequent coordinate correction and 3D modeling. Furthermore, the parsing process incorporates multi-threaded scheduling and data validity verification to ensure both parsing efficiency and data quality.

[0081] III. Coordinate Correction Module.

[0082] The coordinate correction module is configured to receive correction commands from users and perform coordinate correction on the structured parsed data.

[0083] like Figure 2 As shown, in one optional implementation, the coordinate correction module operates in response to the user's selection, meaning that the coordinate correction module may not be invoked, and the model generation module can directly generate the electrical equipment model based on the structured analytical data output by the drawing analysis module.

[0084] As an optional implementation method, such as Figure 7 As shown, the coordinate correction module performs coordinate correction on the structured parsed data according to the following configuration: S31. Visualize and display structured parsed data.

[0085] The structured analytical data shown mainly displays the coordinate positions of each feature point.

[0086] S32. Receive correction instructions from the user to delete, reset, or undo the coordinates of feature points, and update the coordinates of the corresponding feature points.

[0087] As an optional implementation method, such as Figure 7 As shown, in step S32, the coordinate correction module executes the correction instruction for deleting feature point coordinates according to the following configuration: the curvature of the feature point is checked according to the preset curvature constraint conditions. If the check passes, the feature point coordinates are retained; otherwise, the feature point coordinates are deleted.

[0088] In step S32, a point deletion rule with curvature constraints is introduced, along with a doubly linked list-based undo / reset stack management. This supports visual interaction of coordinate points and restoration of historical states, optimizes the structured data output from image parsing, and improves the accuracy of subsequent modeling.

[0089] S33. Store the structured parsed data after the coordinates have been modified.

[0090] IV. Model Generation Module.

[0091] The model generation module is configured to: identify the electrical equipment type based on structured analytical data (i.e., without using the coordinate correction module) or structured analytical data with coordinate correction (i.e., using the coordinate correction module), and then call the corresponding model generation unit to generate the electrical equipment model based on the electrical equipment type.

[0092] Different types of electrical equipment require different methods for constructing their 3D geometric models. As one optional implementation, electrical equipment types include revolution bodies and composite bodies. Correspondingly, the model generation module includes revolution body generation units and composite body generation units. For example... Figure 8 The diagram shows the operation flowchart of the model generation unit.

[0093] like Figure 8 As shown, the rotating body generation unit generates the electrical equipment model according to the following configuration: (1) Based on image center axis scanning detection and positioning of rotation axis.

[0094] like Figure 9 The figures shown are three-dimensional geometric models of representative solids of revolution, namely cylinder, cone, sphere and frustum.

[0095] (2) By fitting the rotation parameters using the least squares method, the two-dimensional feature points are converted into polar coordinate parameters. The feature point cloud is constructed by sampling at equal angles and rotated 360° around the rotation axis to generate a continuous surface.

[0096] (3) The mesh is divided and converted into a triangular mesh structure using a ring-connection strategy to ensure model smoothness. The ring-connection strategy involves connecting adjacent rotated domains in a ring shape, thereby ensuring mesh matching and smooth transition at rotational junctions. For example... Figure 10 The diagram shown is of the JYZ insulator 2-ZS-110 generated using this unit. Figure 11 The image shown is a schematic diagram of GIS CB3 Phase 1 components generated using this unit.

[0097] The assembly generation unit generates electrical equipment models according to the following configuration: (1) Multi-view contours, including geometric feature dimensions, are extracted and simplified by adaptive contour detection and the Douglas-Peucker algorithm.

[0098] (2) Based on the multi-view projection relationship, the stretching parameters are dynamically adjusted to generate the basic column by stretching the solid.

[0099] (3) The rotation matrix is ​​constructed by Euler-Rodriguez formula to complete the multi-coordinate system spatial correction of each basic column.

[0100] (4) Use Boolean intersection operation to merge multi-view grids.

[0101] In addition, spatial segmentation and surface fitting techniques can be combined to optimize model boundaries and topology.

[0102] For example, such as Figure 12 The image shown is a demonstration diagram of Boolean modeling of a general column in this unit. Figure 13 For the modeling results; Figure 14 The image shown is a mesh alignment diagram for a 1000KVA transformer. Figure 15 The image shows the modeling results for a 1000KVA transformer. Figure 16 The image shown is the original 3D model of a 1000KVA transformer. (Comparison) Figure 10 , Figure 11 , Figure 15 and Figure 16 It is evident that the modeling of the three-dimensional geometric model of the electrical equipment in this application maintains a high degree of consistency in terms of structural integrity and dimensions. Combined with... Figure 10 , Figure 11 The modeling effect of solids of revolution, and Figure 13 and Figure 15 The modeling results of the composite object verify the effectiveness and reliability of the proposed solution in the 3D reconstruction of composite objects and the 3D reconstruction of rotating objects.

[0103] As an optional implementation, the model generation module also includes a calibration unit. This calibration unit is configured as follows: (1) Data extraction and calibration: Based on the dimensions marked on the CAD drawings, the extracted geometric feature dimensions are compared in real time. Data with deviations exceeding the first threshold (such as 0.01mm) are extracted and corrected a second time to ensure the accuracy of the input data.

[0104] (2) Modeling process control: The tolerance threshold level (such as 0.01mm level) is dynamically adjusted in the modeling of the combined body generation unit to avoid gaps and overlaps in the model; the rotation parameters are optimized by multiple iterations of fitting in the modeling of the rotating body generation unit to reduce the extraction error of ellipticity and taper.

[0105] (3) Output accuracy verification: Automatically measure the size of electrical equipment model and evaluate the modeling accuracy. For example, use the maximum absolute error (MAE) and root mean square error (RMSE) to evaluate the modeling accuracy. Correct electrical equipment models whose measured size and labeled size deviation exceeds the first threshold to ensure that the key dimensions of the three-dimensional geometric model and the labeled dimensions of the CAD drawing have a deviation of less than 0.01 mm.

[0106] V. Data Export and Storage Module.

[0107] The data export and storage module is configured to: receive a user's export command to export the electrical equipment model, and / or receive a user's storage command to store the electrical equipment model in the user's account storage space.

[0108] As an optional implementation, the data export and storage module has a built-in STEP / IGES format conversion hardware interface to interface with digital twin platforms such as Unity and Digital Twin Hub, allowing electrical equipment models to be directly imported into the digital twin platform and reducing the loss of accuracy during format conversion.

[0109] Taking the modeling of a 1000KVA transformer and JYZ insulator 2-ZS-110 as an example, after the user logs into the system, the system operation process includes: 1. When a user imports a transformer DXF vector drawing and an insulator PNG raster drawing, the system automatically recognizes the format and initiates the corresponding processing path.

[0110] 2. The system extracts the multi-view contour of the transformer and the rotational symmetry features of the insulator, and completes coordinate mapping and parameter transformation.

[0111] 3. The transformer is modeled using Boolean operations to generate a combined structure, and the insulator is modeled using a rotating body to generate a three-dimensional curved surface, with precision calibration performed simultaneously.

[0112] 4. The optimized output is a 3D model that conforms to industry standards and can be directly used for digital twin construction or engineering simulation.

[0113] Overall, the proposed solution has the following characteristics: 1. Dual-path drawing unified recognition technology system: Innovatively construct a separate path processing scheme for raster drawings (PNG / JPG) and vector drawings (DXF / DWG). Through the combination of "Canny edge detection + dynamic threshold segmentation + morphological optimization" algorithm, the raster drawings are accurately parsed. Combined with "regular expression optimization engine + entity deep cleaning + intelligent linkage with engineering software", the vector drawings are efficiently extracted. This solves the industry pain point of the difficulty in uniformly processing multi-format drawings. Moreover, the feature mapping accuracy is guaranteed by the pixel-geometric coordinate affine transformation model.

[0114] 2. Hybrid 3D modeling adaptation technology: Differentiated modeling schemes are designed for different types of electrical components (column assemblies, rotationally symmetric components). Boolean operation modeling (multi-coordinate system spatial correction + Boolean intersection operation + mesh optimization) is organically integrated with parametric modeling of rotating bodies (automatic positioning of rotation axis + polar coordinate transformation + ring mesh generation) to ensure the topological integrity of complex structures and overcome the limitations of insufficient adaptability of single modeling methods.

[0115] 3. Full-process multi-dimensional accuracy calibration mechanism: Establish a closed-loop error control system of "data extraction - modeling process - output verification". Through real-time comparison of key dimensions, dynamic adjustment of tolerance, and MAE / RMSE accuracy evaluation, the model size error is stably controlled within 0.01mm, which solves the problem that traditional modeling accuracy depends on manual labor and error is difficult to control.

[0116] 4. Automated Integration System Architecture: Based on the PyQt5 framework, a modular collaborative system is built, integrating functions such as multi-threaded parallel processing, persistent JSON storage, exception fault tolerance, and export of industry standard formats (STEP / IGES). It realizes unmanned operation of the entire process from drawing import to model output, and supports direct connection to digital twin platforms to improve the adaptability of engineering applications.

[0117] This invention is not limited to the specific embodiments described above. The invention extends to any new feature or combination disclosed in this specification, as well as any new method or process step or combination disclosed herein.

Claims

1. An automatic construction system for three-dimensional geometric models of electrical equipment based on CAD drawings, characterized in that, include: The user login module is configured to verify the user account and then proceed to the drawing parsing module after successful verification. The drawing parsing module is configured to: receive CAD drawings input by the user, preprocess the CAD drawings according to the format of the CAD drawings by calling the corresponding drawing processing unit, and parse the preprocessed CAD drawings to obtain structured parsing data; The coordinate correction module is configured to: receive correction instructions from the user and perform coordinate correction on the structured parsed data; The model generation module is configured to: identify the type of electrical equipment based on the structured parsing data, or based on the structured parsing data after coordinate correction, and generate an electrical equipment model by calling the corresponding model generation unit according to the type of electrical equipment; The data export and storage module is configured to: receive a user's export command to export the electrical equipment model, and / or receive a user's storage command to store the electrical equipment model in the user's account storage space.

2. The automatic construction system for three-dimensional geometric models of electrical equipment based on CAD drawings as described in claim 1, characterized in that, The user login module is also configured as follows: User account credentials are stored using SHA-256 encryption.

3. The automatic construction system for three-dimensional geometric models of electrical equipment based on CAD drawings as described in claim 1, characterized in that, The CAD drawings include two formats: vector graphics and raster graphics; the drawing processing unit includes a vector graphics processing unit and a raster graphics processing unit, wherein: The vector graphics processing unit preprocesses the vector graphics paper according to the following configuration: identifying the layers and entities of the vector graphics paper, matching the layers and entities according to the set regular expressions; caching the layers using the Least Recently Used (LRU) caching mechanism; and removing redundant layers based on the identified entities. The raster image processing unit preprocesses the raster image according to the following configuration: smoothing the noise of the raster image through Gaussian filtering; performing edge detection on the raster image based on the Canny operator; performing contour segmentation based on local variance adaptive threshold; repairing missing contours through morphological opening and closing operations; and establishing a pixel-geometric coordinate affine transformation mapping relationship.

4. The automatic construction system for three-dimensional geometric models of electrical equipment based on CAD drawings as described in claim 1, characterized in that, The drawing parsing module parses the structured parsing data according to the following configuration: Perform multi-scale edge detection on preprocessed CAD drawings; The detected contour points are simplified by curvature weighting to obtain feature points; Establish a coordinate system for the feature points; The feature points are encapsulated as structured data.

5. The automatic construction system for three-dimensional geometric models of electrical equipment based on CAD drawings as described in claim 4, characterized in that, Before parsing the structured data, the drawing parsing module initializes a thread pool based on the format and quantity of the input CAD drawings; different formats of CAD drawings are parsed in parallel using different threads.

6. The automatic construction system for three-dimensional geometric models of electrical equipment based on CAD drawings as described in claim 1, characterized in that, The coordinate correction module performs coordinate correction on the structured parsed data according to the following configuration: The structured parsed data is visualized. Receive correction commands from users to delete, reset, or undo feature point coordinates, and update the coordinates of the corresponding feature points; Store structured parsed data with modified coordinates.

7. The automatic construction system for three-dimensional geometric models of electrical equipment based on CAD drawings as described in claim 6, characterized in that, The coordinate correction module executes correction instructions for deleting feature point coordinates according to the following configuration: The curvature of the feature points is checked against preset curvature constraints. If the check passes, the feature point coordinates are retained; otherwise, the feature point coordinates are deleted.

8. The automatic construction system for three-dimensional geometric models of electrical equipment based on CAD drawings as described in claim 1, characterized in that, The electrical equipment types include rotating body types and combined body types; the model generation module includes a rotating body generation unit and a combined body generation unit, wherein: The rotating body generation unit generates an electrical equipment model according to the following configuration: scanning and detecting the rotation axis based on the image center axis; fitting the rotation parameters using the least squares method, converting the two-dimensional feature points into polar coordinate parameters, constructing a feature point cloud using equal-angle sampling, and generating a continuous surface by rotating 360° around the rotation axis; dividing the mesh using a ring connection strategy and converting it into a triangular mesh structure. The assembly generation unit generates an electrical equipment model according to the following configuration: extracting and simplifying multi-view contours through adaptive contour detection and Douglas-Puk algorithm, the contours including geometric feature dimensions; dynamically adjusting the stretching parameters based on multi-view projection relationships to generate basic columns through solid stretching; constructing rotation matrices through Euler-Rodriguez formula to complete multi-coordinate system spatial correction of each basic column; and fusing multi-view meshes using Boolean intersection operation.

9. The automatic construction system for three-dimensional geometric models of electrical equipment based on CAD drawings as described in claim 8, characterized in that, The model generation module further includes a calibration unit; the calibration unit is configured to: Data extraction and calibration: Based on the dimensions marked on the CAD drawings, the extracted geometric feature dimensions are compared in real time, and data with deviations exceeding the first threshold are extracted and corrected a second time. Modeling process control: The tolerance threshold level is dynamically adjusted during the modeling of the combined body generation unit; the rotation parameters are optimized by multiple iterations of fitting during the modeling of the rotated body generation unit. Output accuracy verification: Automatically measure the dimensions of electrical equipment models, evaluate the modeling accuracy, and correct electrical equipment models whose measured dimensions deviate from the labeled dimensions by more than the first threshold.

10. The automatic construction system for three-dimensional geometric models of electrical equipment based on CAD drawings as described in claim 1, characterized in that, The data export and storage module has a built-in STEP / IGES format conversion hardware interface to interface with the digital twin platform and directly import the electrical equipment model into the digital twin platform.

Images