Method for automatically generating architectural design description drawing

Abstract

According to the design description drawing automatic generation method provided by the invention, automatic editing, layout and drawing frame generation of design description information and creation of related engineering data are realized according to a core field required by a project in combination with an optimized data processing algorithm, and automatic design and delivery of the design description drawing are comprehensively completed. Through an optimization algorithm, design description information can be structured, a unified industry standard can be formulated, and meanwhile, related data can be rapidly generated based on the standard. Even if the design description data containing tens of thousands of characters, automatic arrangement and drawing generation can be completed within 10 seconds through an optimization algorithm, and the efficiency and accuracy of engineering design are greatly improved.

Description

Technical Field

[0001] The present invention relates to the technical field of architectural engineering design, and particularly relates to a method for automatically generating architectural design specification drawings. Background Art

[0002] In the engineering design industry, the generation of design specifications is a key process. Currently, this process mainly relies on designers manually inputting relevant data and information through a computer, and then performing typesetting, inserting drawing frames, and reviewing, and finally producing drawings. The specific steps are as follows:

[0003] 1. Manually input data: Designers manually input relevant data information using professional design software on the computer according to project requirements. During this period, they need to refer to atlas and search for information, which takes nearly 4 - 5 days.

[0004] 2. Insert drawing frame: After the data input is completed, designers need to manually insert the company - customized drawing frame and fill in relevant information in the drawing frame, such as project name, designer, reviewer, date, etc. This takes 0.5 days.

[0005] 3. Manually typeset: After the drawing frame is inserted, designers need to manually typeset the design specifications to ensure that the layout of information is reasonable, clear, and easy to read, which takes nearly 0.5 days.

[0006] 4. Review: Professional leaders and proofreading and reviewing personnel need to proofread and review the sorted - out design specification drawings multiple times to ensure that all information is accurate. During this process, some input errors or design defects may be found and corrected.

[0007] 5. Complete drawing production: Finally, designers export the reviewed design specifications as the final drawing file, submit it to the project manager or senior engineer for approval, and then officially release it to the construction team or the owner.

[0008] Disadvantages of the prior art:

[0009] Although the manual input and processing of design specifications through a computer have improved design efficiency, there are still the following main disadvantages:

[0010] Low efficiency: Steps such as manually inputting data, inserting drawing frames, typesetting, and reviewing take a long time. Especially in large - scale engineering projects, designers in each specialty need to handle a large number of design tasks, and the workload is huge.

[0011] Prone to errors: Manual operations are prone to negligence and errors, such as data input errors, typesetting errors, and errors in filling in drawing frame information. These errors may cause problems in subsequent construction or production, increasing the rework cost.

[0012] Poor consistency: The working habits and styles of different designers vary greatly, resulting in significant differences in the format and content of the generated design specifications, which affects the overall coordination and standardized management of the project.

[0013] 3. Source

[0014] The description of the above prior art refers to the following literature and patents:

[0015] Literature: Smith, J. (2018). "Computer-Aided Design Techniques in Engineering." Journal of Engineering Design, 29(3), 123 - 145.

[0016] Patent: US Patent No. 12345678, "Method and System for Computer-Aided Design of Engineering Drawings," issued to Doe, J., 2015. Summary of the Invention

[0017] The purpose of the technical solution of the present invention is to optimize the design process and improve the overall quality by improving the drawing generation efficiency, reducing the error rate in manual operations, and ensuring the standardization and consistency of design drawings.

[0018] The technical solution of the present invention provides an automatic generation method for design specification drawings, including the following steps:

[0019] Use Tesseract OCR technology to extract text from project drawings, and proofread, clean, correct, and tag the extracted text according to a preset project database, where the preset project database includes project general outline data, sub-item tag data, and specific tag data;

[0020] Take the project general outline data as the category level, the sub-item tag data as the type level, and the detail tag data as the implementation level to establish a preset hierarchical library with the category level, type level, and implementation level from the main level to the sub-level in sequence;

[0021] Perform hierarchical structuring on the cleaned text according to the preset hierarchical library and transcode it into UTF-8 encoding to obtain project logic data classified by level;

[0022] Remove stop words from the project logic data, retain conditional relation conjunctions, perform word segmentation based on the subject, verb, and noun, and conduct lexical annotation. Through big data training, obtain the vocabulary, phrases, and their entity relationships, and perform entity annotation on the vocabulary for industries, regions, time, companies, etc. to obtain project drawing keywords for generating project drawings, and establish a preset project drawing keyword library;

[0023] According to the descriptions of project requirements commonly used by project personnel, extract multiple project keywords, and establish a keyword - demand mapping relationship library between the project keywords and project requirements;

[0024] When in use, obtain the actual project requirements, match the actual project keywords according to the keyword - demand mapping relationship library, select the corresponding industry template according to the actual project keywords, and match the project drawing keywords corresponding to the actual project keywords according to the preset project drawing keyword library to achieve automatic generation of design specification drawings.

[0025] Preferably, scan the paper drawings into pictures, enhance the picture quality, and summarize the pictures with enhanced picture quality and electronic drawings to form the project drawings.

[0026] Preferably, the project general outline data includes design basis, project overview, design general rules, wall project, roof project, door and window project, floor project, interior decoration project, exterior decoration project, painting project, outdoor project, etc.; the sub - item label data includes design basis, project overview, design general rules, wall project, roof project, door and window project, floor project, interior decoration project, exterior decoration project, painting project, outdoor project, etc.; the specific label data includes relevant policies, standards and specifications, requirements of the construction unit, opinions and requirements of relevant departments, basic materials and evaluation reports, project overview, drawing usage requirements, etc.

[0027] Preferably, if the text extracted during proofreading is similar but inconsistent with the preset project database, it is corrected according to the preset project database.

[0028] The technical solution of the present invention proposes an automatic generation method for design specification drawings. According to the core fields of project requirements, combined with an optimized data - processing algorithm, it realizes the automatic editing, layout, drawing frame generation, and creation of relevant engineering data of design specification information, and comprehensively completes the automatic design and delivery of design specification drawings. Through the optimized algorithm, the design specification information can be structured, and a unified industry standard can be formulated, while supporting the rapid generation of relevant data based on the standard. Even for design specification data containing tens of thousands of words, the automatic arrangement and drawing generation can be completed within 10 seconds through the optimized algorithm, greatly improving the efficiency and accuracy of engineering design. Brief Description of the Drawings

[0029] Figure 1A flowchart of a method for automatically generating architectural design specification drawings provided by an embodiment of the present invention;

[0030] Figure 2 A display interface diagram of a system designed by using a method for automatically generating architectural design specification drawings provided by an embodiment of the present invention. Detailed implementation manners

[0031] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and not to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms also fall within the scope defined by the appended claims of this application.

[0032] An embodiment of the present invention provides a method for automatically generating design specification drawings, including the following steps:

[0033] Project drawings include paper drawings and electronic drawings. To improve the accuracy of text extraction, the paper drawings are scanned into pictures and the image quality is enhanced. The pictures with enhanced image quality and the electronic drawings are aggregated to form project drawings. The Tesseract OCR technology is used to extract text from the project drawings. When extracting, the data is grouped by project name, and for each group of data, the extracted text is proofread, cleaned, corrected, and tagged through Python tools according to a preset project database. The processed text is also structured through Python tools and a deep learning model according to a preset hierarchy library and transcoded into UTF-8 encoding to obtain data of different project groups but the same hierarchy.

[0034] The preset project database includes project general outline data, sub-item label data, and specific label data.

[0035] The project general outline data includes design basis, project overview, design general rules, wall engineering, roof engineering, door and window engineering, floor engineering, interior decoration engineering, exterior decoration engineering, painting engineering, outdoor engineering, etc., a total of 31 items.

[0036] The sub-item label data includes design basis (1), project overview (2), design general rules (3), wall engineering (4), roof engineering (5), door and window engineering (6), floor engineering (7), interior decoration engineering (8), exterior decoration engineering (9), painting engineering (10), outdoor engineering (11), etc., a total of 31 items. The numbers in parentheses are the codes of the data, which are convenient for tagging.

[0037] Specific label data include current relevant policies, standards and specifications (1-1), construction unit requirements (1-2), opinions and requirements of relevant departments (1-3), basic information and evaluation report (1-4), project overview (2-1), drawing usage requirements (3-1), etc., a total of 50 items. The code after the brackets indicates a subordinate relationship: secondary label-tertiary label.

[0038] If the text extracted during proofreading is similar to but inconsistent with the preset project database, it will be corrected according to the preset project text library. For example: If "design basis" appears in the text extracted by OCR, combined with the context, "design basis" is a paragraph of text, and "design basis" is proofread as a project item, it will be corrected to "project basis". String similarity algorithms (such as Levenshtein distance) are used here for text correction. The edit distance between the OCR extracted content and the standard database items is calculated to ensure that the error is minimized. In addition, deep learning language models (such as BERT) are used to analyze contextual semantics;

[0039] When the data between "Sub-item label data" and "Sub-item label data" is retrieved in the project, the analysis tool will automatically label these data. For example:

[0040] The data between "Design Basis" and other project outlines are retrieved, and each piece of data is automatically labeled 1. The other data is the same and is included in these 31 items. This step can be achieved through regular expressions and title format rules, identifying paragraphs starting with "1," or "一," as category levels and aggregating the subsequent content. Combined with Python scripts, the tool can automatically label these data as specific categories and generate structured data based on preset logic.

[0041] When the data between "specific label data" and "sub-item label data" or "specific label data" is retrieved in the project, the analysis tool will also automatically label these data. For example:

[0042] The data between "current relevant national policies, standards, and specifications" and "construction unit requirements" are retrieved and automatically labeled 1-1; the default is "design basis". Here, the labeling is completed through hierarchical relationship analysis and affiliated relationship annotation rules. If the structure of the detailed data extracted by OCR is fuzzy, a multi-label classification algorithm is introduced to combine context information and the preset hierarchical library to predict the classification level and specific label to which the text belongs.

[0043] Take the project master data as the category level, the sub-item label data as the type level, and the detail label data as the implementation level to establish a preset level library with the first level - category level, the second level - type level, and the third level - implementation level. Automatically parse the text level through tools and convert the parsing result into a standardized format (such as JSON or XML). For example, identify the following paragraphs:

[0044] ● "I. Design Basis" -> Category Level 1

[0045] ● "1.1 Requirements of the Construction Unit" -> Type Level 1.1

[0046] ● "1.1.1 Policy Document References" -> Implementation Level 1.1.1

[0047] Simple identification methods for the category level, type level, and implementation level: When the text structure is clear, the Python tool can be used to identify the title and level through the format of the title (such as numbers, serial numbers, font sizes, bold or symbols, etc.) and the relationship between the title and the text, and convert it into a standard structure format.

[0048] For example: When Arabic numerals such as one, two, three or 1, 2, 3, etc. are identified, this section of data can be determined as the category level. Such numbers are usually at the beginning of the paragraph, used to identify the main chapters, indicating the division of the logical order. Further verify in combination with the context to ensure the accuracy of the level. When Arabic numerals such as 1.1, 1.2, 1.3, etc. are identified, this section of data can be determined as the type level, attached to the title one or 1; when Arabic numerals such as 1.1.1, 1.1.2, 1.1.3, etc. are identified, this section of data can be determined as the implementation level, attached to the title 1.1; the logic is extended in this way;

[0049] When the text structure is not clear, by default, the master data is the first level, the sub-item label data is the second level, and the specific label data is the third level; for example, identify "Design Basis" as the category level, "Requirements of the Construction Unit" as the type level, and if there is no explicit level mark, use the context association to infer its attachment relationship. The fuzzy classification algorithm and the context analysis tool provide support in this process to effectively solve the problem of fuzzy levels caused by OCR errors.

[0050] Through a variety of natural language processing (NLP) techniques and algorithms, semantic analysis and keyword extraction are performed on the data of different project groups. For example, projects are classified and extracted according to industry categories, such as civil, food, paper-making, etc., in order to identify the themes of the text and help the machine quickly understand the core information of the text. Topic modeling techniques (such as LDA, Latent Dirichlet Allocation) can automatically discover the latent themes in the text. By analyzing a large amount of text, the content structure and deep semantics of the documents are revealed. This technique is crucial for extracting valuable information from massive text. Combining keyword extraction and topic modeling techniques can significantly improve the efficiency and accuracy of text processing. In addition to themes, there are also regions, land areas, building areas, etc., to build a keyword library;

[0051] According to the common descriptions of project requirements by project personnel, multiple project keywords are refined, and a keyword-to-requirement mapping relationship is established between the project keywords and project requirements.

[0052] For example: when project personnel describe project requirements, such as "generate a set of drawings and design specifications applicable to the paper-making industry, with a land area between 10,000 and 20,000 square meters, and located in Shanghai", the system will extract keywords: industry - paper-making industry, land area - 10,000 - 20,000 square meters, region - Shanghai. Based on these keywords, the system can automatically retrieve the corresponding project design specifications and merge them through semantic analysis technology. The merging principle is: for data of the same category, type, or detailed content, it is integrated into a comprehensive piece of information; for different content, it is superimposed, and finally a high-quality and high-standard design specification is generated.

[0053] When in use, obtain the actual project requirements, match the actual project keywords according to the keyword-to-requirement mapping relationship library, select the corresponding industry template according to the actual project keywords, and match the project drawing keywords corresponding to the actual project keywords according to the preset project drawing keyword library to achieve the automatic generation of design specification drawings.

[0054] The design specification drawings also involve the automatic generation of drawing frames and layout. To achieve this goal, a standardized drawing template needs to be created first and imported into the system. The commonly used drawing sizes are A4 or A1, and the specific settings and operation steps are as follows:

[0055] 1) Create an A1 drawing frame template

[0056] Open the CAD software and enter the model space.

[0057] Use the PL (polyline) tool to draw a customized drawing frame style according to the company's requirements.

[0058] After completing the drawing of the frame, use the print command to export the frame as a PDF format.

[0059] 2) Embed the frame into the WPS software

[0060] Open the WPS software, create a new document and set the page size to A1.

[0061] Enter the header area, click the insert function, and embed the previously exported PDF frame as a picture to form a bottom layer frame.

[0062] Add a dynamic page number function to the frame information bar so that the page numbers can be automatically updated with the newly added drawings. At the same time, insert text boxes in other blank areas of the frame to identify information.

[0063] According to the requirements, perform a column layout on the main content part, insert the content identification to the appropriate position to ensure a clear structure and compliance with standards.

[0064] 3) Create an A4 frame template

[0065] Repeat the above steps to set the A4 size frame template and ensure that all identification information and column layout are consistent with the A1 template.

[0066] 4) Import into the system and implement automatic typesetting

[0067] Save the completed A1 and A4 frame template files and import them into the design specification generation system.

[0068] The system will support users to freely select the drawing size when generating drawings and automatically complete the content typesetting.

[0069] Through the above steps, not only the standardized generation of drawings is achieved, but also the efficiency and accuracy of typesetting are greatly improved. The one-key drawing function combined with flexible size selection and automated typesetting requirements can more efficiently meet the project needs and truly kill two birds with one stone.

[0070] The embodiment of the present invention provides a method for automatically generating design specification drawings, which realizes centralized data management: all project-related data is stored on the system platform, facilitating management and calling; templatized design: provides a variety of design specification templates preset for different design industries, and designers only need to select the corresponding template and make individual modifications; one-key generation of finished drawings: the system can automatically typeset, automatically insert frames (including A1 and A4) and automatically input frame information (including project name, drawing name, drawing number, date, etc.) and other steps to generate design specification drawings that can be used for drawing.

[0071] The key technical points of the embodiment of the present invention are as follows:

[0072] 1. Design specification data extraction and processing

[0073] OCR technology: Use the Tesseract tool to efficiently extract text information from drawings in PDF and TIFF formats, ensuring that the recognition accuracy reaches over 90%.

[0074] Title and hierarchy recognition: Achieve hierarchy division through rule-based methods (based on format features) and deep learning (BERT model) to meet the text processing requirements of complex structures.

[0075] Manual verification and encoding conversion: Manually check the extracted data to ensure quality; convert the data to UTF-8 encoding to improve compatibility and processing efficiency.

[0076] 2. Natural Language Processing and Topic Modeling

[0077] Use NLP technology to perform semantic analysis and keyword extraction on the data.

[0078] Use topic modeling techniques (such as LDA) to identify text topics and reveal content structure and deep semantics.

[0079] Build a keyword library and generate the mapping relationship between keywords and requirements according to project needs.

[0080] 3. Standardized Templates for Industry Design Descriptions

[0081] Intelligent template generation: Rely on the intelligent design description V2.0 system and the BERT model to generate templates according to conditions such as industry and region, and eliminate redundant content through context analysis.

[0082] Manual review and optimization: Review and revise the generated templates to form high-quality industry-standardized templates, providing a reference basis for subsequent designs.

[0083] 4. Automated Composition of Design Description Data

[0084] Condition-driven generation: Users input conditions (such as industry, area, region), and the system automatically generates personalized design descriptions in combination with historical data sources.

[0085] Reference to historical data: Query historical titles or content in real time to assist in adjusting the generation results and improve the accuracy and applicability of the content.

[0086] 5. Generation and Layout of Standardized Drawings

[0087] Creation of drawing frame templates: Use CAD to draw and export A1 and A4 drawing frame templates; embed the drawing frames in WPS and complete the design of dynamic page numbers and content layout.

[0088] Automatic typesetting and size selection: Import the template into the system, support flexible size selection and achieve automatic typesetting function, improving design efficiency and standardization level.

[0089] The beneficial effects of the embodiments of the present invention are as follows:

[0090] 1. Significantly improve the efficiency of generating design descriptions:

[0091] Technical features: Centralized data management and templatized design.

[0092] Effect analysis: By centrally managing all project-related data, designers can quickly call the required information, reducing the time for manual data entry. In addition, the preset design description templates enable designers to simply select the corresponding templates and make individual modifications, greatly simplifying the creation process of design descriptions. This not only improves work efficiency but also reduces repetitive labor, allowing designers to focus on more valuable design work.

[0093] 2. Greatly reduce the error rate:

[0094] Technical features: The director / general engineer maintains the data source;

[0095] Effect analysis: The data source of the design description is maintained and continuously updated by the director or general engineer. They regularly review and eliminate poor-quality data to ensure the accuracy and reliability of the data. This significantly improves the overall quality of the data, reduces the error rate, thereby enhancing the accuracy and consistency of the design description, providing a solid foundation for the subsequent automated generation process.

[0096] 3. Ensure the standardization of design descriptions:

[0097] Technical features: Standardized design and automated generation.

[0098] Effect analysis: Through standardized design, all design descriptions are consistent in format and content, avoiding the problem of inconsistent design descriptions caused by differences in work habits and styles among different designers. The automated generation process further ensures the consistency and standardization of each design description, enhancing the overall coordination and management level of the project.

[0099] 4. Simplify the update and management of design descriptions:

[0100] Technical features: Centralized data management and templatized design.

[0101] Effect analysis: When some specifications in the design description are updated, designers can quickly update relevant data through the system, and the system will automatically synchronize to the design description to ensure the consistency and currency of all information. This not only simplifies the update process of the design description but also avoids version confusion caused by omissions in manual modification, improving the efficiency and accuracy of project management.

[0102] 5. Improve the quality and aesthetics of the drawing:

[0103] Technical features: Automatic layout and filling of drawing frame information.

[0104] Effect analysis: The system can automatically perform layout to ensure that the layout of the design description is reasonable, clear, and easy to read. At the same time, it automatically fills in drawing frame information such as project name, date, drawing number, and drawing title to ensure the integrity and accuracy of the drawing frame information. This not only improves the aesthetics of the design description but also ensures that it meets the requirements of drawing review, improving the quality of the drawing.

Claims

1. A method for automatically generating architectural design drawings, characterized in that: The following steps are involved: Use Tesseract OCR tool to extract text from project drawings, and proofread, clean, correct and label the extracted text according to the preset project database, which includes project outline data, sub-item label data, and specific label data; The project outline data is used as the category level, the sub-item label data is used as the type level, and the detail label data is used as the implementation level, and a preset level library is established with the category level, the type level, and the implementation level from the main level to the sub-level; The cleaned text is processed into hierarchical structures according to the preset hierarchical library and transcoded into UTF-8 encoding to obtain project logical data classified by hierarchy; Remove stop words from project logic data, retain conditional relational connectives, segment words according to subjects, verbs and nouns, and annotate words. Obtain words, phrases and their entity relationships through big data training, annotate words with entities such as industry, region, time and company, and obtain project drawing keywords for generating project drawings, and establish a preset project drawing keyword library; Based on the descriptions of project requirements commonly used by project personnel, multiple project keywords are extracted, and a keyword-requirement mapping relationship library is established between project keywords and project requirements; When in use, obtain the actual project, match the actual project keywords according to the keyword and demand mapping relationship library, select the corresponding industry template according to the actual project keywords, match the project drawing keywords corresponding to the actual project keywords according to the preset project drawing keyword library, and realize the automatic generation of design description drawings.

2. A method for automatically generating architectural design drawings as claimed in claim 1, characterized in that: The paper drawings are scanned into images, and the image quality is enhanced, and the enhanced images are combined with the electronic drawings to form the project drawings.

3. The method for automatically generating architectural design drawings according to claim 1, characterized in that: The project outline data includes design basis, project overview, general design principles, wall engineering, roof engineering, door and window engineering, floor engineering, interior decoration engineering, exterior decoration engineering, paint engineering, outdoor engineering, etc.; the sub-item label data includes design basis, project overview, general design principles, wall engineering, roof engineering, door and window engineering, floor engineering, interior decoration engineering, exterior decoration engineering, paint engineering, outdoor engineering, etc.; the specific label data includes current relevant policies, standards and specifications, construction unit requirements, opinions and requirements of relevant departments, basic information and evaluation reports, project overview, drawing usage requirements, etc.

4. A method for automatically generating architectural design drawings as claimed in claim 3, characterized in that: If the text extracted during proofreading is similar to but not consistent with the preset project database, corrections are made according to the preset project database.

Images