Drawing quality evaluation management system and method based on big data

Abstract

The invention belongs to the technical field of big data, and discloses a drawing quality evaluation management system and method based on big data. Comprising the following steps: extracting meta-information of a drawing, and extracting a category identifier by utilizing file name data and professional data in the meta-information to obtain a category of the drawing; carrying out marking; key information of the drawing is obtained, and a designer risk factor, a standard conformity risk factor and a version difference risk factor are calculated; calculating an integrity evaluation value of the drawing; dividing the drawings into defect drawings and standard drawings by utilizing the integrity evaluation values of the drawings; dividing a defect drawing into a first-level defect and a second-level defect; for a first-level defect, starting a multi-dimensional feature missing rate quality evaluation method to calculate a quality score of the defect drawing; for the standard drawing, calculating a quality score of the standard drawing according to the basic quality score and the proportion matching degree; and accurate data support is provided for drawing design and examination.

Description

Technical Field

[0001] This invention relates to the field of big data technology, specifically to a big data-based drawing quality assessment management system and method. Background Technology

[0002] With the development of big data technology and artificial intelligence, the drawing quality assessment and management system has become an intelligent and efficient automated tool that effectively improves the accuracy and efficiency of drawing quality assessment, and ensures the seamless connection of continuous monitoring of design quality, standardized execution and version management.

[0003] However, existing drawing quality assessment and management systems rely on manual classification and version management, which is inefficient and prone to errors, especially when dealing with a large number of drawings. Existing methods use manual inspection of each part of the drawing to check whether it meets the requirements, which can easily overlook hidden design defects such as incorrect annotations, missing dimensions, and non-standard symbols. Furthermore, for drawing quality scoring, existing methods rely on manual review and scoring or logical judgment based on simple rules, without considering which stage the drawing is at, thus failing to achieve targeted and stage-specific drawing quality scoring.

[0004] In view of this, the present invention proposes a big data-based drawing quality assessment management system and method to solve the above problems. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: a drawing quality assessment and management system based on big data, comprising:

[0006] The Metadata Analysis module extracts metadata from drawings and uses file name data and professional data from the metadata to extract category identifiers and obtain the category of the drawings.

[0007] The core module of the graph-based risk control module sets up a full-domain knowledge graph based on the key information of the drawings, queries the comprehensive data of the knowledge graph to obtain the designer's risk factors and standard compliance risk factors; uses the iterative difference risk assessment method to obtain the version difference risk factors; and generates the integrity assessment value of the drawings; and uses the integrity assessment value of the drawings to classify the drawings into defective drawings and standard drawings.

[0008] Difference Conservation Analysis Module: For defective drawings, based on dimensions and annotations, they are classified into Level 1 defects and Level 2 defects; among them, a multi-dimensional feature missing rate quality assessment method is used to calculate the quality score of the defective drawings and define them as Level 1 defects;

[0009] The intelligent scoring module evaluates the basic quality score of standardized drawings based on a bimodal feature method; calculates the proportional matching degree of the standardized drawings; and calculates the quality score of the standardized drawings based on the basic quality score and the proportional matching degree.

[0010] Furthermore, the specific method for extracting the category identifier of the drawing using the file name data and professional data in the metadata includes:

[0011] Set up a multi-source access interface component to extract the metadata of the drawings, including file name data, professional data, and version data; extract category identifiers from the file name data and professional data in the metadata; set up JL preset categories, and match the category identifiers extracted from the file name data and professional data in the drawings with the JL preset categories to obtain the category of the drawings.

[0012] Furthermore, the specific methods for classifying drawings into defective drawings and standard drawings using the integrity assessment value include:

[0013] For drawings, key information about the drawings is obtained; based on the key information of the drawings and key information of historical drawings, a full-domain knowledge graph is constructed; for the full-domain knowledge graph, a graph query language is used to query comprehensive data; comprehensive data includes the number of design drawing categories, the number of versions, the total number of drawings by the designer, the number of basic dimensions, the average number of dimensions, the number of symbols, and the average number of symbols; and the designer's risk factor, standard compliance risk factor, and version difference risk factor are obtained; and the completeness assessment value is obtained; based on the completeness assessment value, the drawings are divided into defective drawings and standard drawings.

[0014] Furthermore, the specific methods for querying comprehensive data from the knowledge graph to obtain the designer risk factor and standard compliance risk factor include:

[0015] The designer risk factor is obtained based on the category rate and version rate. The number of basic dimensions and symbols in the drawing is matched with the comprehensive data in the global knowledge graph. If the current drawing deviates from the global knowledge graph in terms of the number of basic dimensions and symbols, there is a compliance risk in terms of basic dimensions and symbols. The compliance risk factor is obtained based on the degree of deviation. If the current drawing does not deviate from the global knowledge graph in terms of the number of basic dimensions and symbols, the compliance risk factor is set to Aa.

[0016] Furthermore, the specific methods for obtaining version difference risk factors using the iterative difference risk assessment method include:

[0017] Based on the version corresponding to the current defect drawing, extract the line segment data and shape data from the defect drawing, and obtain the initial differences between the line segment data and shape data of the current defect drawing version and the line segment data and shape data of the defect drawings of the previous Nn versions; determine whether the initial differences are evolutionary, correction, or deviation; obtain the version difference risk factor based on the evolutionary, correction, or deviation of the previous Nn versions.

[0018] Furthermore, the specific methods for classifying defective drawings into Level 1 and Level 2 defects based on dimensions and annotations include:

[0019] For defective drawings, dimensions and annotations are extracted. Drawings with missing annotations are defined as Level 1 defects. Drawings with missing dimensions or missing dimensions and annotations are defined as Level 2 defects. For Level 1 defects, a multi-dimensional feature missing rate quality assessment method is used to calculate the quality score of the defective drawings. For Level 2 defective drawings, a yellow mark is made and an early warning is issued.

[0020] Furthermore, the specific method for calculating the quality score of defective drawings by initiating a multi-dimensional feature missing rate quality assessment method includes:

[0021] For defective drawings, a bimodal feature method is used to identify the stages of the defective drawings and set a basic quality score; the annotation missing rate is obtained based on the total number of dimension types, the total number of symbol types, the number of missing dimension annotations, and the number of missing symbol annotations; and the quality score of the defective drawings is calculated based on the basic quality score and the annotation missing rate.

[0022] Furthermore, the specific methods for using a bimodal feature method to identify defective drawings at different stages and setting a basic quality score include:

[0023] Step a1: Extract key text from the defective drawings;

[0024] Step a2: Set the key field table for the drawing; if the key text does not match the key field table for the drawing, proceed to step a3; if the match is successful, set the basic quality score for the conceptual sketch, construction drawing, and as-built drawing.

[0025] Step a3: Divide the defect drawing into Jj units evenly, determine the spatial distribution of dimensions in the unit (clustered, discrete, or linear arrangement); identify the correlation of symbols; and record whether there are any anomalies in the unit (such as overlapping marks, fracture dimensions, and isolated symbols).

[0026] The drawing stages are defined based on the spatial distribution of the dimensions of the intermediate Fn units, the correlation of the symbol annotations, and the probability of anomalies.

[0027] Furthermore, the specific method for calculating the quality score of the standard drawings based on the basic quality score and the proportional matching degree includes:

[0028] A bimodal feature method is used to determine the basic quality score of the standard drawings; a standard scale range for category drawings is set, and the standard scale range corresponding to the category of the standard drawings is retrieved based on the category of the standard drawings; the scale of the standard drawings is obtained, and the scale matching degree is obtained; the quality score of the standard drawings is calculated based on the basic quality score and the scale matching degree.

[0029] A big data-based drawing quality assessment and management method, applied to the aforementioned big data-based drawing quality assessment and management system, includes:

[0030] Step SS1: Extract the metadata of the drawing and use the file name data and professional data in the metadata to extract the category identifier to obtain the category of the drawing;

[0031] Step SS2: Obtain key information of the drawings, set up a full-domain knowledge graph based on the key information of the drawings, and query the comprehensive data of the knowledge graph to obtain the designer risk factor and standard compliance risk factor; use the iterative difference risk assessment method to obtain the version difference risk factor; calculate the integrity assessment value of the drawings based on the designer risk factor, standard compliance risk factor, and version difference risk factor; use the integrity assessment value of the drawings to classify the drawings into defective drawings and standard drawings;

[0032] Step SS3: For defective drawings, extract dimensions and annotations, and classify them into Level 1 and Level 2 defects; among them, calculate the quality score of defective drawings by activating the multi-dimensional feature missing rate quality assessment method, and define it as a Level 1 defect;

[0033] Step SS4: For the standard drawings, evaluate the basic quality score of the standard drawings based on the bimodal feature method; calculate the scale matching degree of the standard drawings; and calculate the quality score of the standard drawings based on the basic quality score and the scale matching degree.

[0034] The technical effects and advantages of the big data-based drawing quality assessment management system and method of this invention are as follows:

[0035] This invention automatically extracts the metadata of drawings and combines it with file name and professional data metadata for intelligent classification. In this way, the system can quickly and accurately identify the category of drawings and automatically manage version data, avoiding errors and delays that may occur in manual classification, and greatly reducing the tediousness and errors of manual operation.

[0036] By leveraging a comprehensive knowledge graph and integrated data analysis, risk factors for designers, compliance with standards, and version differences can be assessed. This enables designers to gain a clearer understanding of potential risks during drawing version updates and the design phase, and to make timely adjustments.

[0037] The iterative difference risk assessment method can effectively assess the differences between drawing versions and provide version difference risk factors. This method helps to discover potential inconsistencies or update problems between drawing versions and avoid risks caused by small differences between different versions.

[0038] For defective drawings, the system can automatically extract drawing dimensions and annotation data, and accurately score defective drawings using a multi-dimensional feature missing rate quality assessment method. At the same time, defective drawings are defined as level one or level two defects and can be automatically marked with warnings. Through automated defect diagnosis, problems missed during manual inspection can be avoided, thus improving the accuracy of the assessment.

[0039] The standardized drawing evaluation module assesses the basic quality score of drawings using a bimodal feature method and combines it with scale matching degree to perform quality scoring; it can effectively identify the quality of standardized drawings and judge the compliance of drawings according to standard scale, ensuring that drawings meet industry or project specification requirements. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of a big data-based drawing quality assessment and management system according to the present invention;

[0041] Figure 2 This is a flowchart of the core module of the graph-based risk control system of this invention;

[0042] Figure 3 This is a schematic diagram of a big data-based drawing quality assessment and management method according to the present invention. Detailed Implementation

[0043] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0044] Example 1

[0045] Please see Figure 1 and Figure 2 As shown in this embodiment, a drawing quality assessment and management system based on big data includes:

[0046] The Metadata Analysis module extracts metadata from drawings and uses file name data and professional data from the metadata to extract category identifiers and obtain the category of the drawings.

[0047] The core module of the graph-based risk control module sets up a full-domain knowledge graph based on the key information of the drawings, queries the comprehensive data of the knowledge graph to obtain the designer's risk factors and standard compliance risk factors; uses the iterative difference risk assessment method to obtain the version difference risk factors; and generates the integrity assessment value of the drawings; and uses the integrity assessment value of the drawings to classify the drawings into defective drawings and standard drawings.

[0048] Difference Conservation Analysis Module: For defective drawings, based on dimensions and annotations, they are classified into Level 1 defects and Level 2 defects; among them, a multi-dimensional feature missing rate quality assessment method is used to calculate the quality score of the defective drawings and define them as Level 1 defects;

[0049] The intelligent scoring module evaluates the basic quality score of standardized drawings based on a bimodal feature method; calculates the proportional matching degree of the standardized drawings; and calculates the quality score of the standardized drawings based on the basic quality score and the proportional matching degree.

[0050] For example, a company undertook a commercial complex project, and the design firm provided a large number of electrical drawings; the drawings designed by the designers were scattered across local computers, cloud storage, and third-party platforms (such as Glodon), making them difficult to manage uniformly; multiple versions of the same drawing existed, making it difficult for the construction team to determine the latest version; and the non-standard naming of the drawings made it difficult for the construction team to find them.

[0051] The specific methods for extracting category identifiers from file name data and professional data in metadata to obtain the category of drawings include:

[0052] Set up a multi-source access interface to input drawings from local, cloud, or third-party platforms into the multi-source access interface component, and use OCR technology to extract the metadata of the drawings. The metadata includes file name data, drawing number data, professional data, and version data.

[0053] The category identifier is extracted from the file name data and professional data in the metadata and matched with JL preset categories to obtain the category of the drawing;

[0054] Specifically, for drawings of the same category, the system checks whether the drawing file name data and professional data are consistent. If they are consistent, the system checks the version data and marks the drawing corresponding to the latest version data (e.g., marking the file name of the drawing in red).

[0055] For example, OCR technology is used to extract the identifiers from the file name data and professional data, and a preliminary match is performed with the identifiers in the JL preset categories, and then stored; the identifiers "single line" and "system" are extracted from the file name data and professional data. The identifiers in the single line diagram category of the circuit breaker are single line, system, and some others. Then the drawing corresponding to the file name data and professional data is the electrical single line category drawing.

[0056] Ten drawings were categorized, with five being electrical single-line drawings. For these five drawings, the `fuzzywuzzy.fuzz.ratio` function was used to check if the file name data and professional data matched. If they didn't match, they were five different electrical single-line drawings; if they matched, they were the same electrical single-line drawing, just different versions. The version with the largest number was then identified as the latest version (e.g., V1, V2, V3, V4, here the largest number is V4), and the latest version was marked.

[0057] The system employs a multi-access interface configuration to enable comprehensive collection and unified processing of drawing data from multiple sources. This configuration comprises three sub-modules: a local access interface, a cloud access interface, and a third-party platform access interface. The local access interface is used to import drawing files in batches from user's local terminal or internal enterprise servers. The cloud access interface supports integration with mainstream cloud storage services (such as Alibaba Cloud, Tencent Cloud, and AWS) to remotely retrieve drawings from the cloud. The third-party platform access interface connects to data sources from external platforms such as BIM systems, CAD platforms, and design collaboration systems via APIs or data middleware. All multi-access interfaces are uniformly encapsulated into a standardized data receiving module, ensuring that all input drawings, after entering the system, can undergo OCR recognition and metadata extraction through a unified data structure, guaranteeing the breadth and compatibility of data access and the consistency of subsequent processing.

[0058] The JL preset categories are set up to automate the classification of drawings. Each category corresponds to a specific professional field or purpose of the drawing, such as electrical schematic diagrams, single-line electrical diagrams, power distribution plans, grounding system diagrams, and distribution box system diagrams. Each preset category consists of a set of identifiers, which are derived through analysis of historical drawing file name data, professional data, and human expert annotation, possessing strong industry representativeness and coverage. During operation, the file name data and professional data are extracted from the drawing's metadata. After recognizing the text using OCR technology, identifiers are extracted and initially matched against the identifier sets in the JL preset categories. If the category identifier in the drawing matches the identifiers of the JL preset categories, the drawing is finally classified into the most suitable category. The preset categories can be customized, expanded, or finely adjusted according to the actual needs of the project during system deployment, exhibiting good scalability and industry adaptability.

[0059] By uniformly collecting drawings from local, cloud, or third-party platforms through multi-source access interfaces and extracting metadata using OCR, the system can automatically identify the category and version of drawings, reducing the workload of manual processing. Furthermore, by comparing file name data and professional data of similar drawings, the system can automatically filter out the latest version and mark it prominently (e.g., in red), helping users quickly locate the most accurate and up-to-date drawings and avoid design or construction errors caused by using outdated drawings, thereby improving the intelligence, accuracy, and efficiency of drawing management.

[0060] Specific methods for classifying drawings into defective drawings and standard drawings using completeness assessment values ​​include:

[0061] For drawings, key information about the drawings is obtained, including drawing type, designer basic information, design date, views, basic dimensions, layer information, number of basic dimensions, and number of symbols; the drawing type is obtained by matching with JL preset categories;

[0062] If the drawing is a CAD drawing, use a specialized library, such as the AutoCAD API, to parse the DWG file and extract the designer's basic information, design date, views, basic dimensions, layer information, number of basic dimensions, and number of symbols. If the drawing is a PDF drawing, use a PDF parsing library, such as PyMuPDF or pdfminer, to extract the designer's basic information, design date, views, basic dimensions, layer information, number of basic dimensions, and number of symbols. If the drawing is an image drawing (such as a hand-drawn sketch or scanned drawing), use image processing technology (such as OpenCV) combined with a deep learning model to extract the designer's basic information, design date, views, basic dimensions, layer information, number of basic dimensions, and number of symbols.

[0063] Based on the key information of the drawings and the key information of historical drawings, a global knowledge graph is constructed. The key information of the drawings is used as the nodes of the global knowledge graph, and edges are constructed according to the semantic relationships and logical connections between the key information to represent the connection relationships between the nodes. For example, the drawing was created by a certain designer, and drawing V3 is a subsequent version of drawing V2.

[0064] For the full-domain knowledge graph, use a graph query language (such as Cypher) to query comprehensive data; comprehensive data includes the number of design drawing categories, the number of versions, the total number of designer drawings, the number of basic dimensions, the number of average dimensions, the number of symbols, and the average number of symbols;

[0065] Use comprehensive data to calculate the designer's risk factor and the standard compliance risk factor of the drawings;

[0066] The version difference risk factor is calculated based on the iterative difference risk assessment method.

[0067] Completeness assessment values ​​are calculated for the designer's risk factors, standard compliance risk factors, and version difference risk factors. Standard completeness assessment values ​​are then set based on empirical methods. The completeness assessment values ​​are compared with the standard completeness assessment values. If the completeness assessment value is greater than the standard completeness assessment value, the drawing is considered a standard drawing. If the completeness assessment value is less than or equal to the standard completeness assessment value, the drawing is considered a defective drawing.

[0068] Dividing drawings into standard drawings and defective drawings facilitates automated identification and risk warning of drawing quality. This method comprehensively considers multiple factors such as designer background, drawing content structure, standard compliance, and version evolution. It utilizes a full-domain knowledge graph for semantic modeling and data linkage analysis to ensure more comprehensive and accurate evaluation results. Different types of drawings are processed using specialized parsing techniques to extract key information, improving the adaptability and accuracy of data acquisition. Based on graph reasoning and differentiated risk factor calculation, it can effectively identify potential problems such as design anomalies, standard deviations, or version inconsistencies. Finally, by comparing the integrity assessment value with empirical standards, defective drawings are automatically identified, thereby reducing manual review costs and improving the intelligence level of drawing quality management and the overall project security.

[0069] By calculating the designer risk factor, standard compliance risk factor, and version difference risk factor, the quality and potential risks of drawings can be comprehensively assessed from three key dimensions: "people, drawing content, and drawing evolution." The designer risk factor reveals the designer's professional focus and stability, helping to identify high-risk design behaviors. The standard compliance risk factor measures the drawings' adherence to industry standards in terms of dimensions and symbol usage, ensuring clear and accurate design expression. The version difference risk factor reflects the extent of changes in the drawings during their evolution, helping to determine whether there are issues of design instability or frequent rework. The synergistic effect of these three factors enables a multi-faceted and dynamic assessment of drawing quality, thereby improving the intelligence level of design review, reducing project risks, and enhancing the reliability and efficiency of project delivery.

[0070] Specific ways to query comprehensive data from a knowledge graph to obtain designer risk factors and standard compliance risk factors include:

[0071] The category rate is calculated based on the number of design drawing categories and the total number of design drawings; the version rate is calculated based on the number of versions and the total number of design drawings; and the designer's risk factor is calculated based on the category rate and version rate.

[0072] The drawing is matched with the comprehensive data in the global knowledge graph based on the number of basic dimensions and symbols. If the current drawing deviates from the global knowledge graph in terms of the number of basic dimensions and symbols, there is a compliance risk in terms of basic dimensions and symbols. The compliance risk factor is obtained based on the degree of deviation. If the current drawing does not deviate from the global knowledge graph in terms of the number of basic dimensions and symbols, the compliance risk factor is set to Aa.

[0073] The degree of deviation is the ratio of the number of deviations to the total number of deviations, and the compliance risk factor can be calculated using the weighted average method.

[0074] By leveraging comprehensive data from a full-domain knowledge graph, designer risk factors are calculated through the rate of designer drawing categories and version rates. The compliance risk factors are dynamically assessed based on the degree of deviation from basic dimensions and the number of symbols. The advantage lies in its ability to combine rich historical and global information to achieve multi-dimensional and quantitative risk analysis of designer behavior and drawing quality, thereby improving the accuracy and objectivity of risk identification. At the same time, the weighted average method is used to comprehensively evaluate the degree of deviation, making the calculation of risk factors more scientific and reasonable, which is conducive to precise risk management and decision support.

[0075] Specific methods for obtaining version difference risk factors using the iterative difference risk assessment method include:

[0076] Based on the version corresponding to the current defect drawing, extract the line segment data and shape data from the defect drawing, and calculate the initial differences between the line segment data and shape data in the current version of the defect drawing and the line segment data and shape data in the previous Nn versions of the defect drawing.

[0077] The initial differences are classified as evolving, corrective, or offset; the version difference risk factor is obtained based on the evolving, corrective, or offset nature of the first Nn versions.

[0078] If the probability of evolutionary type is 40%, the probability of correction type is 30%, and the probability of deviation type is 30% for the first Nn versions, then the version difference risk factor can be obtained by weighted average method.

[0079] The evolution type involves adding or optimizing line segments or shapes, the correction type involves removing redundancy, and the offset type involves changes in position. Python + Shapely / OpenCV tools are used to determine whether a type is evolution, correction, or offset.

[0080] If there is only one version of the current defect drawing, then the version difference risk factor is defined as 0;

[0081] Using Python in conjunction with Shapely and OpenCV to determine evolutionary, corrective, and offset changes between drawing versions has the advantage of providing powerful geometric analysis and image processing capabilities. It can efficiently and accurately extract and compare the spatial features of line segments and shapes, and achieve automated multi-version difference identification.

[0082] For defective drawings, the specific methods for classifying defects into Level 1 and Level 2 based on dimensions and annotations include:

[0083] For the defective drawings, extract the dimensions and annotations, including dimension annotations and symbol annotations;

[0084] A drawing with missing annotations is defined as a Level 1 defect; a drawing with missing dimensions or missing dimensions and annotations is defined as a Level 2 defect.

[0085] For Level 1 defects, a multi-dimensional feature missing rate quality assessment method is used to calculate the quality score of the defective drawings; for Level 2 defective drawings, they are marked in yellow and an early warning is issued.

[0086] Dimensions can be extracted using the AutoCAD API, PyMuPDF parsing tools, or image processing techniques. Dimensions are text on the drawing that shows the actual size of an object or component. For CAD or PDF drawings, they can be extracted directly, while for image drawings, optical character recognition is used. Symbol annotations indicate the identifiers of different objects or components in the drawing. Deep learning models (such as convolutional neural networks) are used to classify symbols and detect symbol annotations on the drawing.

[0087] For example, in an electrical drawing, if the symbol for an electrical device is missing, but the location and type of the device can still be inferred from other lines and markings, then this omission has little impact on the use of the drawing and is considered a minor defect.

[0088] If a structural drawing lacks dimension annotations for local details (such as decorative details), but the main structural dimensions still exist and are clear, it will not have a significant impact on the overall building construction.

[0089] Dimensions are the core information in drawings; missing dimensions will directly affect the accuracy and safety of construction. Any missing dimensions related to structure, load-bearing capacity, equipment installation, etc., will directly affect the construction process. Especially in engineering construction, all construction activities rely on the dimensions in the drawings for precise operation.

[0090] Defects caused by missing dimensions are difficult to compensate for through speculation or other means; missing dimensions mean the lack of key physical quantities and spatial information, which can lead to the inability to correctly locate components, structures or equipment; if the missing dimensions are critical dimensions (such as structural load-bearing dimensions, foundation dimensions, important passage dimensions, etc.), even a small amount of missing dimensions may cause errors in the entire construction process.

[0091] Missing dimensions often require revising drawings or redesigning, and the repair process is complex and has a significant impact. Unlike missing annotations, missing dimensions usually require more corrections and reconfirmation. The repair process involves re-examining and verifying the drawings, which may delay the entire project schedule and indicate a serious defect.

[0092] By classifying defective drawings into Level 1 and Level 2 defects based on the degree of missing annotations and dimensions, not only is refined management of drawing issues achieved, but the targeting and response efficiency of quality assessment are also improved. Level 1 defects mainly involve the omission of annotation information, which may affect understanding but does not directly affect dimensional accuracy. Therefore, assessing quality through multi-dimensional feature missing rate helps to quantitatively determine the severity of the problem. Level 2 defects involve missing dimensions or both dimensions and annotations, which are critical defects and require immediate attention through yellow marking and warning mechanisms. This classification mechanism enables intelligent identification and layered response of defective drawings, improving the efficiency and accuracy of drawing review and reducing construction or manufacturing risks caused by serious defects flowing into subsequent stages.

[0093] The specific methods for calculating the quality score of defective drawings by initiating a multi-dimensional feature missing rate quality assessment method include:

[0094] For defective drawings, a bimodal feature method is used to identify the stages of the defective drawings and set a basic quality score;

[0095] Use data extraction tools (such as CAD tools, computer vision tools, etc.) to count the total number of dimension types, total number of symbol types, number of missing dimension annotations, and number of missing symbol annotations in defective drawings; and calculate the annotation missing rate based on the total number of dimension types, total number of symbol types, number of missing dimension annotations, and number of missing symbol annotations, using the formula: D = (number of missing dimension annotations + number of missing symbol annotations) / (total number of dimension types + total number of symbol types), where D is the standard missing rate;

[0096] The quality score of the defective drawings is calculated based on the basic quality score and the missing annotation rate.

[0097] The purpose of calculating quality scores for Level 1 defective drawings is to refine the quantitative understanding of missing annotation issues, thereby achieving more precise quality control and resource allocation. By identifying defective drawings through a bimodal feature method, a reasonable basic quality score is set, and combined with the missing dimension and symbol annotation rates, the actual impact of the drawings is comprehensively assessed, avoiding a "one-size-fits-all" approach. This scoring mechanism not only helps to distinguish between minor and serious defects, but also provides data support for subsequent designer performance evaluation and drawing quality tracking, thereby improving the intelligence and refinement of the entire drawing review and management process.

[0098] For defective drawings, a bimodal feature method is used to identify defects at different stages, and the specific methods for setting basic quality scores include:

[0099] Step a1: Extract key text from the defective drawings;

[0100] Step a2: Set the key field table for the drawings. The key field table for the drawings includes the conceptual sketch, construction drawing, as-built drawing, and their corresponding fields. If the key text does not match the key field table for the drawings, proceed to step a3. If the match is successful, set the basic quality score of the conceptual sketch to D1; set the basic quality score of the construction drawing to D2; and set the basic quality score of the as-built drawing to D3.

[0101] Step a3: Divide the defect drawing into Jj units evenly, determine the spatial distribution of dimensions in the units (clustered, discrete, or linear arrangement; use sklearn.cluster analysis tool to determine whether it is clustered, discrete, or linear arrangement); identify the correlation of symbol annotations (use CGAL or OpenCV to identify); and record whether there are any anomalies in the units (such as overlapping marks, fracture dimensions, and isolated symbols; use the Overkill command to determine).

[0102] The drawing stages are defined based on the spatial distribution of the dimensions of the intermediate Fn units, the correlation of the symbol annotations, and the probability of anomalies.

[0103] Specifically, if more than 2 / 3 of the Fn elements have clustered dimensional standards, and the symbol annotations of more than 2 / 3 of the Fn elements are associated, and more than 2 / 3 of the Fn elements are not anomalous, then it is an as-built drawing; if 1 / 3 to 2 / 3 of the Fn elements have clustered dimensional standards, and the symbol annotations of 1 / 3 to 2 / 3 of the Fn elements are associated, and the symbol annotations of 1 / 3 to 2 / 3 of the Fn elements are not anomalous, then it is a construction drawing; all other cases are conceptual sketches.

[0104] The basic quality of the conceptual sketch is divided into D1; the basic quality of the construction drawing is divided into D2; and the basic quality of the as-built drawing is divided into D3.

[0105] The total area of ​​the defective drawing is calculated through image processing, and the defective drawing is evenly divided into Jj units to obtain the area of ​​each unit.

[0106] By combining textual and image features in a dual-modal recognition method, the design stage of defective drawings can be accurately determined, thereby providing a reasonable basic quality score for subsequent quality assessment.

[0107] The specific methods for calculating the quality score of standard drawings based on the basic quality score and the proportional matching degree include:

[0108] Step b1: Use the bimodal feature method to determine the basic quality score of the standard drawings;

[0109] Step b2: Set the standard scale range for category drawings. Based on the category of the standard drawings, retrieve the standard scale range corresponding to the category of the standard drawings; and obtain the scale of the standard drawings. If the scale of the standard drawings is within the standard scale range of the category drawings, it means that it conforms to the standard scale range and the scale matching degree is E1. If the scale of the standard drawings is not within the standard scale range of the category drawings, it means that it does not conform to the standard scale range and the scale matching degree is E2.

[0110] Step b3: Calculate the quality score of the standard drawings based on the basic quality score and scale matching degree of the standard drawings;

[0111] For example, set the standard scale range for different types of drawings. For instance, the standard scale range for electrical layout drawings is 1:50 to 1:500; for cable laying drawings, it's 1:10 to 1:200; and for grounding system drawings, it's 1:10 to 1:500. Given a standard drawing, by setting JL preset categories, the category identifiers extracted from the file name and professional data in the standard drawing are matched with these JL preset categories to determine the category of the standard drawing: Electrical Layout. If the scale of the standard drawing is 1:100, then the scale of the standard drawing is within the standard range for Electrical Layout drawings, and the scale matching degree E1 is set to 1. If the scale of the standard drawing is 1:20, then the scale of the standard drawing is not within the standard range for Electrical Layout drawings, and the scale matching degree E2 is set to 0.

[0112] Based on the basic quality score and scale matching degree of the standard drawings, the quality score of the standard drawings can be calculated using the weighted average method.

[0113] In this embodiment, the system automatically extracts the metadata of drawings and performs intelligent classification by combining the file name and professional data metadata. In this way, the system can quickly and accurately identify the category of drawings and automatically manage version data, avoiding errors and delays that may occur in manual classification, and greatly reducing the tediousness and errors of manual operation.

[0114] By leveraging a comprehensive knowledge graph and integrated data analysis, risk factors for designers, compliance with standards, and version differences can be assessed. This enables designers to gain a clearer understanding of potential risks during drawing version updates and the design phase, and to make timely adjustments.

[0115] The iterative difference risk assessment method can effectively assess the differences between drawing versions and provide version difference risk factors. This method helps to discover potential inconsistencies or update problems between drawing versions and avoid risks caused by small differences between different versions.

[0116] For defective drawings, the system can automatically extract drawing dimensions and annotation data, and accurately score defective drawings using a multi-dimensional feature missing rate quality assessment method. At the same time, defective drawings are defined as level one or level two defects and can be automatically marked with warnings. Through automated defect diagnosis, problems missed during manual inspection can be avoided, thus improving the accuracy of the assessment.

[0117] The standardized drawing evaluation module assesses the basic quality score of drawings using a bimodal feature method and combines it with scale matching degree to perform quality scoring; it can effectively identify the quality of standardized drawings and judge the compliance of drawings according to standard scale, ensuring that drawings meet industry or project specification requirements.

[0118] Example 2

[0119] Please see Figure 3 As shown, for parts not described in detail in this embodiment, please refer to the description in Embodiment 1. This embodiment provides a big data-based drawing quality assessment and management method, including:

[0120] Step SS1: Extract the metadata of the drawing and use the file name data and professional data in the metadata to extract the category identifier to obtain the category of the drawing;

[0121] Step SS2: Obtain key information of the drawings, set up a full-domain knowledge graph based on the key information of the drawings, and query the comprehensive data of the knowledge graph to obtain the designer risk factor and standard compliance risk factor; use the iterative difference risk assessment method to obtain the version difference risk factor; calculate the integrity assessment value of the drawings based on the designer risk factor, standard compliance risk factor, and version difference risk factor; use the integrity assessment value of the drawings to classify the drawings into defective drawings and standard drawings;

[0122] Step SS3: For defective drawings, extract dimensions and annotations, and classify them into Level 1 and Level 2 defects; among them, calculate the quality score of defective drawings by activating the multi-dimensional feature missing rate quality assessment method, and define it as a Level 1 defect;

[0123] Step SS4: For the standard drawings, evaluate the basic quality score of the standard drawings based on the bimodal feature method; calculate the scale matching degree of the standard drawings; and calculate the quality score of the standard drawings based on the basic quality score and the scale matching degree.

[0124] Example 3

[0125] This embodiment discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the operation mode of the big data-based drawing quality assessment management system and method described above.

[0126] Since the electronic device described in this embodiment is the electronic device used to implement the big data-based drawing quality assessment management system and method described in this application embodiment, those skilled in the art can understand the specific implementation method and various variations of the electronic device in this embodiment based on the big data-based drawing quality assessment management system and method described in this application embodiment. Therefore, how the electronic device implements the method in this application embodiment will not be described in detail here. Any electronic device used by those skilled in the art to implement the big data-based drawing quality assessment management system and method described in this application embodiment falls within the scope of protection of this application.

[0127] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0128] The above description is merely a preferred embodiment of the present invention, and the scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for users of ordinary technical skills, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A drawing quality assessment and management system based on big data, characterized in that, include: The Metadata Analysis module extracts metadata from drawings and uses file name data and professional data from the metadata to extract category identifiers and obtain the category of the drawings. The core module of graph-based risk control: It sets up a full-domain knowledge graph based on the key information of the drawings, queries the comprehensive data of the knowledge graph to obtain the designer's risk factors and standard compliance risk factors; and uses the iterative difference risk assessment method to obtain the version difference risk factors. And generate a completeness assessment value for the drawings; The integrity assessment value of the drawings is used to classify them into defective drawings and standard drawings; Difference Conservation Analysis Module: For defective drawings, based on dimensions and annotations, they are classified into Level 1 defects and Level 2 defects; among them, a multi-dimensional feature missing rate quality assessment method is used to calculate the quality score of the defective drawings and define them as Level 1 defects; The intelligent scoring module evaluates the basic quality score of standardized drawings based on a bimodal feature method; calculates the proportional matching degree of the standardized drawings; and calculates the quality score of the standardized drawings based on the basic quality score and the proportional matching degree.

2. The drawing quality assessment and management system based on big data according to claim 1, characterized in that, The specific methods for extracting category identifiers from file name data and professional data in metadata to obtain the category of drawings include: Set up a multi-source access interface component to extract the metadata of the drawings, including file name data, professional data, and version data; extract category identifiers from the file name data and professional data in the metadata; set up JL preset categories, and match the category identifiers extracted from the file name data and professional data in the drawings with the JL preset categories to obtain the category of the drawings.

3. The drawing quality assessment and management system based on big data according to claim 2, characterized in that, The specific methods for classifying drawings into defective drawings and standard drawings using the integrity assessment value include: For drawings, key information about the drawings is obtained; based on the key information of the drawings and key information of historical drawings, a full-domain knowledge graph is constructed; for the full-domain knowledge graph, a graph query language is used to query comprehensive data; comprehensive data includes the number of design drawing categories, the number of versions, the total number of drawings by the designer, the number of basic dimensions, the average number of dimensions, the number of symbols, and the average number of symbols; and the designer's risk factor, standard compliance risk factor, and version difference risk factor are obtained; and the completeness assessment value is obtained; based on the completeness assessment value, the drawings are divided into defective drawings and standard drawings.

4. The drawing quality assessment and management system based on big data according to claim 3, characterized in that, The specific methods for querying comprehensive data from the knowledge graph to obtain the designer risk factor and standard compliance risk factor include: The designer risk factor is obtained based on the category rate and version rate. The number of basic dimensions and symbols in the drawing is matched with the comprehensive data in the global knowledge graph. If the current drawing deviates from the global knowledge graph in terms of the number of basic dimensions and symbols, there is a compliance risk in terms of basic dimensions and symbols. The compliance risk factor is obtained based on the degree of deviation. If the current drawing does not deviate from the global knowledge graph in terms of the number of basic dimensions and symbols, the compliance risk factor is set to Aa.

5. The drawing quality assessment and management system based on big data according to claim 4, characterized in that, The specific methods for obtaining version difference risk factors using the iterative difference risk assessment method include: Based on the version corresponding to the current defect drawing, extract the line segment data and shape data from the defect drawing, and obtain the initial differences between the line segment data and shape data of the current defect drawing version and the line segment data and shape data of the defect drawings of the previous Nn versions; determine whether the initial differences are evolutionary, correction, or deviation; obtain the version difference risk factor based on the evolutionary, correction, or deviation of the previous Nn versions.

6. The drawing quality assessment and management system based on big data according to claim 5, characterized in that, The specific methods for classifying defective drawings into Level 1 and Level 2 defects based on dimensions and annotations include: For defective drawings, dimensions and annotations are extracted. Drawings with missing annotations are defined as Level 1 defects. Drawings with missing dimensions or missing dimensions and annotations are defined as Level 2 defects. For Level 1 defects, a multi-dimensional feature missing rate quality assessment method is used to calculate the quality score of the defective drawings. For Level 2 defective drawings, a yellow mark is made and an early warning is issued.

7. The drawing quality assessment and management system based on big data according to claim 6, characterized in that, The specific method for calculating the quality score of defective drawings by initiating a multi-dimensional feature missing rate quality assessment method includes: For defective drawings, a bimodal feature method is used to identify the stages of the defective drawings and set a basic quality score; the annotation missing rate is obtained based on the total number of dimension types, the total number of symbol types, the number of missing dimension annotations, and the number of missing symbol annotations; and the quality score of the defective drawings is calculated based on the basic quality score and the annotation missing rate.

8. The drawing quality assessment and management system based on big data according to claim 7, characterized in that, The specific methods for using a bimodal feature method to identify defective drawings at different stages and to set a basic quality score include: Step a1: Extract key text from the defective drawings; Step a2: Set the key field table for the drawing; if the key text does not match the key field table for the drawing, proceed to step a3; if the match is successful, set the basic quality score for the conceptual sketch, construction drawing, and as-built drawing. Step a3: Divide the defective drawing into Jj units evenly, determine the spatial distribution of dimensions in the units, identify the correlation of symbol annotations, and record whether there are any abnormalities in the units; The drawing stages are defined based on the spatial distribution of the dimensions of the intermediate Fn units, the correlation of the symbol annotations, and the probability of anomalies.

9. A drawing quality assessment and management system based on big data according to claim 8, characterized in that, The specific methods for calculating the quality score of standard drawings based on the basic quality score and the proportional matching degree include: A bimodal feature method is used to determine the basic quality score of the standard drawings; a standard scale range for category drawings is set, and the standard scale range corresponding to the category of the standard drawings is retrieved based on the category of the standard drawings; the scale of the standard drawings is obtained, and the scale matching degree is obtained; the quality score of the standard drawings is calculated based on the basic quality score and the scale matching degree.

10. A big data-based drawing quality assessment and management method, applied to the big data-based drawing quality assessment and management system described in any one of claims 1 to 9, characterized in that, include: Step SS1: Extract the metadata of the drawing and use the file name data and professional data in the metadata to extract the category identifier to obtain the category of the drawing; Step SS2: Obtain key information from the drawings, set up a full-domain knowledge graph based on the key information from the drawings, and query the comprehensive data of the knowledge graph to obtain the designer's risk factor and the standard compliance risk factor. The version difference risk factor is obtained using the iterative difference risk assessment method. The integrity assessment value of the drawings is calculated based on the designer's risk factor, the standard compliance risk factor, and the version difference risk factor. The integrity assessment value of the drawings is used to classify them into defective drawings and standard drawings; Step SS3: For defective drawings, extract dimensions and annotations, and classify them into Level 1 and Level 2 defects; among them, calculate the quality score of defective drawings by activating the multi-dimensional feature missing rate quality assessment method, and define it as a Level 1 defect; Step SS4: For the standard drawings, evaluate the basic quality score of the standard drawings based on the bimodal feature method; calculate the scale matching degree of the standard drawings; and calculate the quality score of the standard drawings based on the basic quality score and the scale matching degree.

Images