A bridge drawing selection method and system fusing quality score and geometric matching

Abstract

The application discloses a bridge drawing optimization method and system fusing quality score and geometric matching, and belongs to the fields of bridge engineering design and computer information technology, and comprises the following steps: constructing a structured bridge design drawing knowledge base, receiving target design requirements of a user, calculating engineering quantity indexes of candidate drawings, calculating drawing quality scores, multi-dimensional similarity matching and comprehensive recommendation scoring, and outputting a recommended result. The application can automatically analyze structural parameters and performance indexes in a large amount of historical drawings, can quickly screen candidate schemes according to target requirements (such as span, width and span number) input by the user, can combine multi-dimensional quality factors such as specification compliance, economy and drawing integrity to quantitatively score and sort the candidate drawings, and can finally output recommended drawings with high matching degree, high quality and direct reference or fine adjustment, so that the bridge design efficiency and quality are significantly improved.

Description

Technical Field

[0001] This invention relates to the fields of bridge engineering design and computer information technology, specifically to a method and system for selecting the best bridge drawings by integrating quality scoring and geometric matching. Background Technology

[0002] Bridge engineering is a crucial component of infrastructure construction, and its design process heavily relies on the experience of structural engineers and historical project data. In practice, to improve design efficiency, ensure structural safety, and control project costs, designers often refer to or reuse bridge design drawings from similar past projects. However, with the surge in the number of bridge projects, the increasing complexity of bridge types, and the continuous updating of design standards, traditional manual retrieval and comparison methods are no longer sufficient to meet the demands for efficient, accurate, and intelligent design.

[0003] Currently, the industry generally adopts the following methods for reusing drawings:

[0004] (1) Keyword search: Designers can perform fuzzy searches in the drawing management system using limited fields such as project name, span, and bridge width;

[0005] (2) Manual experience screening: Manual matching is performed based on the engineer's personal memory or a historical case database shared within the team;

[0006] (3) Template application: Modify and adapt using standardized drawing sets (such as general drawings and standard span simply supported beam drawings).

[0007] The above methods have obvious limitations:

[0008] Low retrieval accuracy: Keywords cannot fully reflect the structural features of drawings and are prone to missing key parameters (such as reinforcement ratio, steel content, cross-sectional shape, etc.).

[0009] Reliance on subjective experience: Different engineers have different understandings of "similarity", which leads to unstable reuse results;

[0010] Lack of quantitative evaluation: It is impossible to automatically determine the merits of candidate drawings in terms of economy, safety, and compliance with standards;

[0011] The quality of drawings varies: historical drawings may contain design flaws, missing information, or not comply with current specifications. If they are reused directly without verification, they may cause safety hazards or rework risks.

[0012] In recent years, the application of artificial intelligence (AI) technology in the engineering field has deepened. Computer vision (CV) and natural language processing (NLP) technologies have been used for extracting information from drawings, and machine learning models have begun to be applied to structural parameter prediction and optimization. However, existing technologies mostly focus on single tasks (such as rebar identification and load calculation), and have not yet formed an end-to-end bridge design drawing reuse system centered on intelligent recommendation. In particular, there is a lack of comprehensive solutions that integrate multiple dimensions such as geometric matching degree, structural performance indicators, and drawing quality scores, and achieve automatic sorting and recommendation.

[0013] To address this, a method and system for selecting the best bridge drawings that integrates quality scoring and geometric matching is proposed. Summary of the Invention

[0014] The technical problem to be solved by this invention is: how to solve the core problems of low intelligence, single evaluation dimension and unreliable recommendation results in existing drawing reuse methods, and to provide a bridge drawing selection method that integrates quality scoring and geometric matching.

[0015] The present invention solves the above-mentioned technical problems through the following technical solution, and the present invention includes the following steps:

[0016] S1: Building a knowledge base of structured bridge design drawings

[0017] Collect historical bridge design drawings, identify and extract structural parameters from the drawings, and form a structured database, i.e., a knowledge base;

[0018] S2: Receive user target design requirements

[0019] Users input the design requirements of the target bridge through a human-computer interaction interface, which generates a target parameter vector.

[0020] S3: Calculate the quantity indices of candidate drawings

[0021] Multiple candidate drawings are obtained by filtering based on the target parameter vector, and the engineering quantity index is calculated based on the extracted construction parameters.

[0022] S4: Calculate the drawing quality score

[0023] The quality of the drawings is evaluated from four dimensions: completeness, compliance with standards, rationality of design, and feedback from historical use. The evaluation is then normalized and weighted to obtain the quality score of each candidate drawing.

[0024] S5: Multidimensional Similarity Matching and Comprehensive Recommendation Scoring

[0025] Geometric similarity is scored based on span, bridge width, and number of spans. Then, the geometric similarity score and the index excellence score are considered together and assigned weights to calculate the comprehensive recommendation score for each candidate drawing.

[0026] S6: Output Recommendation Results

[0027] Sort all candidate drawings in descending order of their overall recommendation score and output the recommendation list.

[0028] Furthermore, in step S1, the metadata fields associated with the j-th drawing in the knowledge base include: bridge type, span. Bridge width , span Cross-sectional width Cross-sectional height Total area of ​​longitudinal reinforcement in the tension zone Total weight of reinforcing steel Concrete volume Material grade.

[0029] Furthermore, in step S2, the target parameter vector is specifically as follows:

[0030] ;

[0031] in: For the target span; Target bridge width; The target span.

[0032] Furthermore, in step S3, based on the target span input by the user... Perform initial screening and retain those in the knowledge base that satisfy the span requirement. The drawings were used as candidate drawings. The span tolerance set by the user.

[0033] Furthermore, in step S3, the engineering quantity indicators include reinforcement ratio, steel content, and concrete usage per unit area, and the calculation formulas are as follows:

[0034] Reinforcement ratio: ;

[0035] Steel content: ;

[0036] Concrete usage per single surface area: ;

[0037] Where i represents the i-th candidate drawing.

[0038] Furthermore, in step S4, the drawing quality is scored. The calculation formula is as follows:

[0039] ;

[0040] in, , , , The scores are the completeness score, specification compliance score, design rationality score, and historical usage feedback score for the i-th candidate drawing, respectively. , , , These are the completeness weight coefficient, the standard compliance weight coefficient, the design rationality weight coefficient, and the historical usage feedback weight coefficient.

[0041] Furthermore, in the aforementioned drawing quality assessment, the completeness score, specification compliance score, design rationality score, and historical usage feedback score are obtained in the following ways:

[0042] Integrity score It is based on the number of essential elements identified in the drawings. Total number of preset essential elements The ratio is obtained by using the following formula: ;

[0043] Standard compliance score It is calculated based on the proportion of serious violations detected in the drawings to the maximum tolerable number of violations. The calculation formula is as follows: ;in, Let be the number of check items that are satisfied in the i-th candidate drawing; This represents the total number of inspection items required by the relevant standards for the current bridge type.

[0044] Design rationality score It refers to the quantitative assessment value of whether the specific structural parameters in the bridge design drawings are within the "reasonable range of similar projects";

[0045] Historical usage feedback rating This refers to scoring a drawing based on the number of times it has been cited throughout its history. The calculation formula is as follows: ,in, These represent the total number of times the i-th candidate drawing was searched and the number of times the i-th candidate drawing was successfully referenced, respectively. This is a correction factor; This is the activation threshold.

[0046] Furthermore, the design rationality score The specific calculation process is as follows:

[0047] S431: Select a set of evaluation indicators, including reinforcement ratio. Steel content Span-to-height ratio Concrete usage per unit area The k-th evaluation index of the i-th candidate drawing is denoted as The total number of evaluation indicators for the i-th candidate drawing is K;

[0048] S432: Construct a reference sample set of "similar bridges". Based on the relevant attributes of the i-th drawing, select a subset of similar bridges from the knowledge base. Next, calculate the statistical values ​​of each evaluation indicator in this subset: mean and standard deviation; the calculation formula is as follows:

[0049] The mean of the k-th evaluation index in the bridge subset: ;

[0050] The standard deviation of the k-th evaluation index in the bridge subset: ;

[0051] in, The number of bridge subsets of the same type is less than the set threshold value. If it is less than the set threshold value, the statistical value of the entire knowledge base or the preset value will be used.

[0052] S433: Calculate the standardized outlier of each evaluation index for each evaluation index of the i-th candidate drawing. Calculate its standardized deviation relative to the mean of the same class. Standardized outlier, or normalized anomaly, is expressed as follows:

[0053] ;

[0054] S434: Take the maximum standardized outlier as the overall degree of deviation. The calculation formula is as follows:

[0055] ;

[0056] S435: The design rationality score is obtained through exponential decay fitting.

[0057] ;

[0058] in, The attenuation coefficient is the fitted value.

[0059] Furthermore, in step S5, the specific processing procedure is as follows:

[0060] S51: Calculate the normalized weighted distance Calculate the geometric similarity score of candidate drawings. :

[0061] ;

[0062] ;

[0063] in, The geometric dimensions of span, bridge width, and number of spans are weighted. These are the standard deviations of bridge spans, bridge widths, and number of spans in the knowledge base that differ from the target span, bridge width, and number of spans by less than 20%. This is the similarity attenuation coefficient; This is the maximum tolerance distance;

[0064] S52: Based on a comprehensive evaluation of various engineering quantity indicators and drawing quality scores, the candidate drawings are scored for their degree of excellence. The calculation is as follows:

[0065] ;

[0066] ;

[0067] in, This refers to the coefficient for the sub-item of the indicator's excellence. This represents the optimal value for the reinforcement ratio specified in the code. The maximum and minimum reinforcement ratios in the knowledge base; These are the maximum and minimum steel content values ​​in the knowledge base. This refers to the maximum and minimum values ​​of concrete volume per single surface area in the knowledge base.

[0068] S53: Calculate the comprehensive recommendation score for candidate drawings by taking into account both geometric similarity and the score of excellence in the indicators. :

[0069] ;

[0070] in, To match priority weights.

[0071] This invention also provides a bridge drawing selection system that integrates quality scoring and geometric matching, applied to the above-mentioned selection method, including:

[0072] The drawing parsing and knowledge base construction module is used to process multi-source heterogeneous bridge design drawings, extract structural parameters, and build a searchable knowledge base.

[0073] The target requirement input and interaction module is used to receive user design requirements and provide a human-computer interaction interface;

[0074] Engineering index calculation engine, used to calculate engineering quantity indices for candidate drawings based on extracted construction parameters;

[0075] The multi-dimensional quality and excellence rating module is used to calculate the quality rating of drawings and the excellence rating of indicators.

[0076] The matching and ranking recommendation engine is used to calculate geometric similarity, comprehensive recommendation score, and rank the comprehensive recommendation score.

[0077] The visual recommendation module is used to display recommendation results to users.

[0078] The present invention has the following advantages over the prior art:

[0079] 1. This invention has a high degree of intelligence: it automatically analyzes drawings and extracts structural parameters through AI, eliminating the need for manual review and greatly improving the efficiency of information acquisition.

[0080] 2. This invention recommends accurate and reliable results: the primary principle is to ensure geometric matching by making the span, width and number of spans as close as possible, while integrating performance indicators and quality scores to avoid misinterpretations that are "similar in form but not in essence".

[0081] 3. This invention features multi-dimensional quantitative evaluation: For the first time, it incorporates drawing completeness, standard compliance, economy, and historical feedback into a unified scoring system to achieve objective and interpretable ranking.

[0082] 4. This invention improves design quality and safety: It automatically filters out non-compliant or abnormal drawings, reducing design risks caused by reusing low-quality drawings.

[0083] 5. Support the accumulation and reuse of engineering knowledge: form an intelligent drawing knowledge base that can be iteratively optimized, and promote the transformation of design experience from "personal memory" to "system assets". Attached Figure Description

[0084] Figure 1 This is a flowchart illustrating the bridge drawing selection method that integrates quality scoring and geometric matching in an embodiment of the present invention.

[0085] Figure 2 This is a schematic diagram of the bridge drawing selection system that integrates quality scoring and geometric matching in an embodiment of the present invention. Detailed Implementation

[0086] The embodiments of the present invention are described in detail below. These embodiments are implemented based on the technical solution of the present invention, and provide detailed implementation methods and specific operation processes. However, the scope of protection of the present invention is not limited to the following embodiments.

[0087] like Figure 1 As shown, this embodiment provides a technical solution: a method for selecting the best bridge drawings by integrating quality scoring and geometric matching, comprising the following steps:

[0088] S1: Construct a knowledge base of structured bridge design drawings.

[0089] Historical bridge design drawings (including CAD, BIM, PDF, etc.) are collected. Computer vision (CV) and natural language processing (NLP) technologies are used to automatically identify and extract structural parameters from the drawings, forming a structured database. The metadata fields associated with the j-th drawing include, but are not limited to: bridge type, span... (unit: m), bridge width (unit: m), number of spans Cross-sectional width (unit: mm), cross-sectional height (unit: mm), total area of ​​longitudinal reinforcement in the tension zone (unit: mm) 2 Total weight of reinforcing steel (unit: kg), concrete volume (unit: mm) 3 ), material grade, etc.

[0090] In this step, the automatic identification and extraction of construction parameters from the drawings using computer vision (CV) and natural language processing (NLP) technologies includes the following processes:

[0091] First, the system preprocesses and semantically segments the input bridge design drawings (such as CAD vector graphics, PDFs, or scanned images). By training a dedicated deep learning model, the system automatically identifies and crops key sub-drawing areas, such as main beam cross-sections, elevation layouts, reinforcement details, quantity tables, and text descriptions. Subsequently, the system uses CV technology to perform fine-grained analysis on each sub-drawing. Through a joint model of object detection and OCR, it accurately locates and identifies key elements such as dimension lines (such as arrowed leader lines and numbers), reinforcement details, material lists, cross-section symbols, and component labels. At the same time, it uses graph neural networks to analyze the relative positions, connections, and directional logic between these elements (e.g., whether a certain dimension value points to the height of the main beam, or whether a certain row of reinforcement belongs to the edge beam reinforcement).

[0092] Based on this, the system introduces a topological semantic understanding module trained on a large number of bridge drawings. This module integrates structural engineering knowledge graphs with NLP semantic parsing capabilities. Through supervised learning and training on “annotations-components-parameter types” in historical drawings, the model can accurately understand the physical meaning of numerical values ​​in a specific engineering context.

[0093] Finally, all extracted results are organized into a structured parameter vector with engineering semantics. This process ensures the accuracy of parameter extraction, laying a highly reliable data foundation for subsequent intelligent matching and recommendation based on target needs.

[0094] S2: Receive user target design requirements.

[0095] Users input the design requirements of the target bridge through a human-computer interaction interface, which generates a target parameter vector:

[0096] ;

[0097] in: Target span (in meters); Target bridge width (in meters); The target span number (dimensionless).

[0098] S3: Calculate the key engineering quantity indicators for candidate drawings.

[0099] Candidate drawings are initially screened from the knowledge base based on user design requirements. The target span is then determined by the user input. Perform a hard initial screening to retain those in the knowledge base that meet the span requirement. The drawings were used as candidate drawings. The user-defined span tolerance quickly eliminates obviously unsuitable solutions, reducing subsequent computational load.

[0100] For candidate drawings, based on the extracted structural parameters, engineering quantity indicators such as reinforcement ratio, steel content, and concrete usage per unit area are calculated. The calculation formulas are shown below:

[0101] Reinforcement ratio: ;

[0102] Steel content: ;

[0103] Concrete usage per single surface area: .

[0104] S4: Calculate the drawing quality score.

[0105] The quality of candidate drawings is evaluated from four dimensions: completeness, compliance with standards, design rationality, and historical usage feedback. Each sub-item is normalized to the interval [0,1], and a weighted fusion is performed to obtain the drawing quality score. The calculation formula is as follows:

[0106] ;

[0107] in, , , , The scores are: completeness score, specification compliance score, design rationality score, and historical usage feedback score for the i-th drawing. , , , These are the completeness weight coefficient, the standard compliance weight coefficient, the design rationality weight coefficient, and the historical usage feedback weight coefficient.

[0108] More specifically, integrity score It is based on the number of essential elements identified in the drawings. Total number of preset essential elements The ratio is obtained by using the following formula: .

[0109] It should be noted that the number of essential elements refers to the actual number of structural parameters that the system successfully identifies and extracts after analyzing a bridge design drawing. These parameters are necessary for calculating key engineering quantity indicators (such as reinforcement ratio, steel content, and concrete usage per unit area). These elements include, but are not limited to: calculated span of the main girder, girder height, top slab thickness, bottom slab thickness, web thickness, concrete strength grade, diameter and spacing of reinforcing bars, and protective layer thickness. Each item that is reliably identified is counted as one valid essential element; items that are missing or fail to be identified are not counted.

[0110] The total number of predefined essential elements refers to the total number of items in the minimum set of parameters required to complete a full set of index calculations. This total number is fixed by the system based on structural design specifications and engineering experience. For example, for a conventional T-beam bridge, if calculating the reinforcement ratio and steel content requires 8 core parameters (such as span, beam height, rib width, top slab thickness, rebar diameter, spacing, concrete grade, and protective layer thickness), then the total number of essential elements is 8. This value varies depending on the bridge type and calculation objective, but remains fixed for the same type of recommended task, serving as the benchmark denominator for completeness assessment.

[0111] More specifically, the compliance score It is calculated based on the proportion of serious violations detected in the drawings to the maximum tolerable number of violations. The calculation formula is as follows:

[0112]

[0113] in, Let be the number of check items that are satisfied in the i-th candidate drawing; This represents the total number of standard inspection items applicable to the current bridge type.

[0114] It should be noted that the number of satisfied inspection items refers to the total number of inspection items that meet the requirements when automatically verifying the compliance of a candidate bridge design drawing. This is determined by the system based on a series of quantifiable structural clauses applicable to that bridge type in current bridge design specifications (such as the "Specifications for Design of Highway Reinforced Concrete and Prestressed Concrete Bridges and Culverts" JTG 3362), comparing each item with the extracted structural parameters. Each inspection item is judged as satisfied or not satisfied using Boolean logic; only when the parameter value clearly falls within the allowable range of the specification is it counted as a satisfied item. The total number of specification inspection items refers to the number of all quantifiable inspection clauses preset by the system for compliance verification for the bridge type corresponding to the current candidate drawing.

[0115] More specifically, design rationality score This refers to a quantitative assessment of whether key structural parameters in bridge design drawings fall within the "reasonable range for similar projects." The value ranges from [0,1]. A value closer to 1 indicates that the design parameters in the drawings conform to industry experience and statistical norms, showing no obvious anomalies or excessively conservative / aggressive tendencies. A lower value indicates parameter deviation, poor economic efficiency, or potential safety hazards. The specific calculation process is as follows:

[0116] S431: Select a set of key evaluation indicators, including reinforcement ratio. Steel content Span-to-height ratio Concrete usage per unit area And so on, the k-th indicator of the i-th drawing is denoted as The total number of indicators for the i-th drawing is K.

[0117] S432: Construct a reference sample set of "similar bridges". Based on the bridge type, span range, and other attributes of the i-th drawing, select a subset of similar bridges from the knowledge base. Further, calculate the statistical value for each indicator in this subset, including the mean and standard deviation, as shown in the following expression:

[0118] The mean of the k-th index in the bridge subset: ;

[0119] The standard deviation of the k-th index in the bridge subset: ;

[0120] in, The number of bridge subsets of the same type is used. If it is less than a certain threshold, the statistical value of the entire knowledge base or the preset value is used.

[0121] S433: Calculate the standardized outlier of each indicator. For each indicator of the i-th drawing. Calculate its standardized deviation relative to the mean of the same class. The expression is as follows:

[0122] ;

[0123] S434: Take the maximum anomaly as the overall deviation. The calculation expression is as follows:

[0124] ;

[0125] S435: The design rationality score is obtained through exponential decay fitting.

[0126] ;

[0127] in, To fit the attenuation coefficient, satisfying It can take the value 0.5.

[0128] More specifically, historical usage feedback rating This refers to scoring drawings based on the number of times they have been cited throughout history, which also serves as a reflection of the designers' satisfaction with the drawings. The calculation formula is as follows:

[0129] ;

[0130] in, These represent the total number of times the i-th candidate drawing was searched and the number of times the i-th candidate drawing was successfully referenced, respectively. As a correction factor, a minimum value can be taken, which is... ; The threshold value can be 3.

[0131] S5: Multidimensional similarity matching and comprehensive recommendation scoring.

[0132] A geometric similarity scoring method is established to evaluate the degree of closeness between candidate drawings and user target requirements in terms of basic structural dimensions. Geometric similarity scoring is based on three key geometric parameters: span, bridge width, and number of spans. Then, by comprehensively considering similarity and the excellence of the indicators and assigning weights accordingly, a comprehensive recommendation score for each candidate drawing is calculated. The specific processing procedure is as follows:

[0133] S51: Calculate the normalized weighted distance Calculate the geometric similarity score of candidate drawings. :

[0134] ;

[0135] ;

[0136] in, The geometric dimensions of span, bridge width, and number of spans are weighted. These are the standard deviations of bridge spans, bridge widths, and number of spans in the knowledge base that differ from the target span, bridge width, and number of spans by less than 20%. This is the similarity attenuation coefficient, which can take a value of 2; This is the maximum tolerance distance.

[0137] S52: Based on a comprehensive evaluation of various engineering quantity indicators and drawing quality scores, the candidate drawings are scored for their degree of excellence. The calculation is performed using the formula shown below:

[0138] ;

[0139] ;

[0140] in, This refers to the coefficient for the sub-item of the indicator's excellence. This represents the optimal value for the reinforcement ratio specified in the code. The maximum and minimum reinforcement ratios in the knowledge base; These are the maximum and minimum steel content values ​​in the knowledge base. These represent the maximum and minimum values ​​of concrete volume per single surface area in the knowledge base.

[0141] S53: Provide a comprehensive recommendation score for candidate drawings, taking into account both similarity and the excellence of the indicators. :

[0142] ;

[0143] in, To match priority weights.

[0144] S6: Sort the candidate drawings according to the comprehensive recommendation score and output the recommendation results.

[0145] All candidate drawings Sort the data in descending order and output a recommended list. For each item, provide parameters such as drawing number, thumbnail, comparison of drawing parameters with target parameters, reinforcement ratio, steel content, drawing quality score, and overall recommendation score.

[0146] like Figure 2 As shown, this invention also provides a bridge drawing selection system that integrates quality scoring and geometric matching to implement the above-mentioned method. This system can be deployed on a design institute's internal server, a private cloud platform, or integrated into mainstream BIM / CAD design environments. It includes six core functional modules: a drawing parsing and knowledge base construction module, a target requirement input and interaction module, an engineering indicator calculation engine, a multi-dimensional quality and excellence scoring module, a matching and ranking recommendation engine, and a visual recommendation and feedback closed-loop module.

[0147] (1) The function of the drawing parsing and knowledge base construction module is to automatically process multi-source heterogeneous bridge design drawings, extract structured parameters, and build a searchable knowledge base. It includes a multi-format parsing engine, a parameter extraction model, and a structured storage unit. The multi-format parsing engine is used for PDF text recognition, CAD layer analysis, BIM model reading, etc.; the parameter extraction model is used to identify information such as bridge span, bridge width, and reinforcement from drawing text or elements; the structured storage unit stores the extracted results in the database and establishes index fields to support fast filtering.

[0148] (2) The target requirement input and interaction module's function is to receive user design intent and provide a user-friendly human-computer interaction interface. Components include a web frontend, intelligent guidance components, and API interfaces; the web frontend is used to obtain user design parameters, such as... The intelligent guidance component automatically recommends typical parameter groups based on historical projects; the API interface supports integration with other design systems to automatically obtain the target parameters of the current project.

[0149] (3) The engineering index calculation engine function automatically calculates core performance indicators such as reinforcement ratio, steel content, and concrete usage per unit area for candidate drawings. It includes rule calculation units and AI-assisted estimation modules. The rule calculation units have a built-in structural formula library for calculating engineering quantity indicators.

[0150] (4) The multidimensional quality and excellence scoring module is used to calculate drawing quality scores, indicator excellence scores, etc. Components include an integrity detector, a standard compliance check engine, a rationality assessment unit, a historical feedback database, and a weighted fusion unit. It can output the drawing quality score for each drawing. and indicator excellence score .

[0151] (5) The matching and ranking recommendation engine function performs the calculation of geometric similarity and comprehensive recommendation score, and ranks the comprehensive recommendation scores. Components include dynamic normalization unit, exponential similarity calculator, comprehensive scorer, sorter, etc., which can output a ranked list of recommended drawings and detailed score breakdown information for each item.

[0152] (6) The visual recommendation and feedback closed-loop module displays recommendation results to users and collects user feedback to optimize the system. Components include a recommendation result display panel, a difference highlighting component, a user feedback collector, and a knowledge base self-updating mechanism.

[0153] The system data flow steps are as follows:

[0154] (1) Users submit design goals through the target requirements input and interaction module;

[0155] (3) The system performs an initial screening of candidate sets from the knowledge base;

[0156] (4) The engineering index calculation engine and the multi-dimensional quality and excellence scoring module process each candidate drawing in parallel;

[0157] (5) The matching and ranking recommendation engine integrates geometric and indicator scores to generate a ranked list;

[0158] (6) The visualization recommendation and feedback closed-loop module presents the results and writes user feedback back to the knowledge base to form a closed loop.

[0159] In summary, the bridge drawing selection method and system that integrates quality scoring and geometric matching described in the above embodiments can automatically analyze the structural parameters and performance indicators in massive historical drawings; can quickly filter candidate schemes based on user-input target requirements (such as span, width, and number of spans); can quantitatively score and rank candidate drawings by combining multi-dimensional quality factors such as compliance with specifications, economy, and drawing completeness; and can ultimately output highly matched, high-quality recommended drawings that can be directly referenced or fine-tuned, significantly improving the efficiency and quality of bridge design.

[0160] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A method for selecting the best bridge drawings by integrating quality scoring and geometric matching, characterized in that, Includes the following steps: S1: Building a knowledge base of structured bridge design drawings Collect historical bridge design drawings, identify and extract structural parameters from the drawings, and form a structured database, i.e., a knowledge base; S2: Receive user target design requirements Users input the design requirements of the target bridge through a human-computer interaction interface, which generates a target parameter vector. S3: Calculate the quantity indices of candidate drawings Multiple candidate drawings are obtained by filtering based on the target parameter vector, and the engineering quantity index is calculated based on the extracted construction parameters. S4: Calculate the drawing quality score The quality of the drawings is evaluated from four dimensions: completeness, compliance with standards, rationality of design, and feedback from historical use. The evaluation is then normalized and weighted to obtain the quality score of each candidate drawing. S5: Multidimensional Similarity Matching and Comprehensive Recommendation Scoring Geometric similarity is scored based on span, bridge width, and number of spans. Then, the geometric similarity score and the index excellence score are considered together and assigned weights to calculate the comprehensive recommendation score for each candidate drawing. S6: Output Recommendation Results Sort all candidate drawings in descending order of comprehensive recommendation score and output the recommendation list; In step S1, the metadata fields associated with the j-th drawing in the knowledge base include: bridge type and span. Bridge width , span Cross-sectional width Cross-sectional height Total area of ​​longitudinal reinforcement in the tension zone Total weight of reinforcing steel Concrete volume Material grade; In step S3, the engineering quantity indicators include reinforcement ratio, steel content, and concrete usage per unit area, and the calculation formulas are as follows: Reinforcement ratio: ; Steel content: ; Concrete usage per single surface area: ; Where i represents the i-th candidate drawing; In step S4, the drawing quality is scored. The calculation formula is as follows: ； in, , , , The scores are the completeness score, specification compliance score, design rationality score, and historical usage feedback score for the i-th candidate drawing, respectively. , , , These are the completeness weighting coefficient, the specification compliance weighting coefficient, the design rationality weighting coefficient, and the historical usage feedback weighting coefficient. The methods for obtaining the completeness score, specification compliance score, design rationality score, and historical usage feedback score in the drawing quality evaluation are as follows: Integrity score It is based on the number of essential elements identified in the drawings. Total number of preset essential elements The ratio is obtained by using the following formula: ; Standard compliance score It is calculated based on the proportion of serious violations detected in the drawings to the maximum tolerable number of violations. The calculation formula is as follows: ;in, Let be the number of check items that are satisfied in the i-th candidate drawing; This represents the total number of inspection items required by the relevant standards for the current bridge type. Design rationality score It refers to the quantitative assessment value of whether the specific structural parameters in the bridge design drawings are within the "reasonable range of similar projects"; Historical usage feedback rating This refers to scoring a drawing based on the number of times it has been cited throughout its history. The calculation formula is as follows: ,in, These represent the total number of times the i-th candidate drawing was searched and the number of times the i-th candidate drawing was successfully referenced, respectively. This is a correction factor; This is the activation threshold; The design rationality score The specific calculation process is as follows: S431: Select a set of evaluation indicators, including reinforcement ratio. Steel content Span-to-height ratio Concrete usage per unit area The k-th evaluation index of the i-th candidate drawing is denoted as The total number of evaluation indicators for the i-th candidate drawing is K; S432: Construct a reference sample set of "similar bridges", and select a subset of similar bridges from the knowledge base based on the relevant attributes of the i-th drawing. Next, calculate the statistical values ​​of each evaluation indicator in this subset: mean and standard deviation; the calculation formula is as follows: The mean of the k-th evaluation index in the bridge subset: ; The standard deviation of the k-th evaluation index in the bridge subset: ; in, The number of bridge subsets of the same type is less than the set threshold value. If it is less than the set threshold value, the statistical value of the entire knowledge base or the preset value will be used. S433: Calculate the standardized outlier of each evaluation index for each evaluation index of the i-th candidate drawing. Calculate its standardized deviation relative to the mean of the same class. Standardized outlier, or normalized anomaly, is expressed as follows: ； S434: Take the maximum standardized outlier as the overall degree of deviation. The calculation formula is as follows: ； S435: The design rationality score is obtained through exponential decay fitting. ； in, The fitting attenuation coefficient; In step S5, the specific processing procedure is as follows: S51: Calculate the normalized weighted distance Calculate the geometric similarity score of candidate drawings. : ； ； in, The geometric dimensions of span, bridge width, and number of spans are weighted. These are the standard deviations of bridge spans, bridge widths, and number of spans in the knowledge base that differ from the target span, bridge width, and number of spans by less than 20%. This is the similarity attenuation coefficient; This is the maximum tolerance distance; S52: Based on a comprehensive evaluation of various engineering quantity indicators and drawing quality scores, the candidate drawings are scored for their degree of excellence. The calculation is as follows: ； ； in, This refers to the coefficient for the sub-item of the indicator's excellence. This represents the optimal value for the reinforcement ratio specified in the code. The maximum and minimum reinforcement ratios in the knowledge base; These are the maximum and minimum steel content values ​​in the knowledge base. This refers to the maximum and minimum values ​​of concrete volume per single surface area in the knowledge base. S53: Calculate the comprehensive recommendation score for candidate drawings by taking into account both geometric similarity and the score of excellence in the indicators. : ； in, To match priority weights.

2. The method for selecting the best bridge drawings by integrating quality scoring and geometric matching according to claim 1, characterized in that, In step S2, the target parameter vector is specifically as follows: ； in: For the target span; Target bridge width; The target span.

3. The method for selecting the best bridge drawings by integrating quality scoring and geometric matching according to claim 2, characterized in that, In step S3, based on the target span input by the user... Perform initial screening and retain those in the knowledge base that satisfy the span requirement. The drawings were selected as candidate drawings. The span tolerance set by the user.

4. A bridge drawing selection system integrating quality scoring and geometric matching, characterized in that, The method applied to the selection method as described in any one of claims 1 to 3 includes: The drawing parsing and knowledge base construction module is used to process multi-source heterogeneous bridge design drawings, extract structural parameters, and build a searchable knowledge base. The target requirement input and interaction module is used to receive user design requirements and provide a human-computer interaction interface; Engineering index calculation engine, used to calculate engineering quantity indices for candidate drawings based on extracted construction parameters; The multi-dimensional quality and excellence rating module is used to calculate the quality rating of drawings and the excellence rating of indicators. The matching and ranking recommendation engine is used to calculate geometric similarity, comprehensive recommendation score, and rank the comprehensive recommendation score. The visual recommendation module is used to display recommendation results to users.

Images