A multi-dimensional review method and system for construction drawings

Abstract

The present invention discloses a multi-dimensional verification method and system for construction drawings, which relates to the field of construction engineering technology. The method includes: establishing an association between two-dimensional construction drawings and building information models, outputting a linkage construction drawing block, and mapping the connection data of the two-dimensional construction drawings and the linkage two-dimensional drawings according to the model positioning identification information based on the linkage construction drawing block; identifying the multi-dimensional verification features of the two-dimensional construction drawings, establishing a target drawing identification feature cluster, performing feature anomaly matching according to the connection data, obtaining anomaly matching results, performing verification evaluation according to the anomaly matching results, and outputting the verification evaluation results. The present invention solves the technical problems of low verification efficiency, single verification dimension, and poor accuracy in the prior art, and achieves the technical effect of improving the comprehensiveness and accuracy of the verification and improving the verification efficiency.

Description

Technical Field

[0001] The present invention relates to the field of building engineering technology, and in particular to a multi-dimensional verification method and system for construction drawings. Background Art

[0002] In fields such as construction engineering and industrial manufacturing, construction drawings are the key basis for ensuring the smooth progress of projects. With the continuous development of technology, the complexity of construction drawings is also increasing. In the current drawing review technology, there are still many challenges. On the one hand, the existing review often relies on manual operation, which is extremely inefficient; on the other hand, although some software and systems are used in drawing review, there are deficiencies in data accuracy and review dimensions.

[0003] The existing technology has technical problems such as low verification efficiency, single verification dimension and poor accuracy. Summary of the invention

[0004] The present application provides a multi-dimensional review method and system for construction drawings, which is used to solve the technical problems in the prior art such as low review efficiency, single review dimension and poor accuracy.

[0005] In view of the above problems, the present application provides a multi-dimensional review method and system for construction drawings.

[0006] In a first aspect of the present application, a multi-dimensional review method for construction drawings is provided, the method comprising:

[0007] Establish an association between a two-dimensional construction drawing and a building information model, wherein the building information model is a three-dimensional building model of a target building, wherein the two-dimensional construction drawing has model positioning identification information; use the model positioning identification information of the two-dimensional construction drawing as an index, search for display information in the building information model, and output a linkage construction drawing block, wherein the linkage construction drawing block contains all linkage two-dimensional drawings that have a position connection relationship with the two-dimensional construction drawing, and all linkage two-dimensional drawings have corresponding model positioning identification information; based on the linkage construction drawing block, map and associate the connection data of the two-dimensional construction drawing and the linkage two-dimensional drawing according to the model positioning identification information; identify the multi-dimensional verification features of the two-dimensional construction drawing, establish a target drawing recognition feature cluster, and each target drawing recognition feature cluster corresponds to a verification dimension; according to the connection data, perform feature anomaly matching with the verification features in the target drawing recognition feature cluster to obtain anomaly matching results, and establish a mapping association between the anomaly matching results and the verification dimension; perform verification evaluation according to the evaluation objectives of the verification dimension according to the anomaly matching results, and output the verification evaluation results.

[0008] A second aspect of the present application provides a multi-dimensional review system for construction drawings, the system comprising:

[0009] A module for establishing an association between two-dimensional construction drawings and building information models, wherein the module is used to establish an association between two-dimensional construction drawings and building information models, wherein the building information model is a three-dimensional building model of a target building, wherein the two-dimensional construction drawings have model location identification information; a linkage construction drawing block output module, wherein the linkage construction drawing block output module is used to search for display information in the building information model using the model location identification information of the two-dimensional construction drawings as an index, and output a linkage construction drawing block, wherein the linkage construction drawing block contains all linkage two-dimensional drawings that have a position connection relationship with the two-dimensional construction drawings, and all linkage two-dimensional drawings have corresponding model location identification information; a model location identification information mapping module, wherein the model location identification information mapping module is used to map the linkage construction drawing blocks based on the linkage construction drawing blocks. A construction drawing block, which maps and associates the connection data of the two-dimensional construction drawing with the linked two-dimensional drawing according to the model positioning identification information; a target drawing recognition feature cluster establishment module, which is used to identify the multi-dimensional verification features of the two-dimensional construction drawing and establish a target drawing recognition feature cluster, each target drawing recognition feature cluster corresponding to an verification dimension; an abnormal matching result acquisition module, which is used to perform feature abnormality matching with the verification features in the target drawing recognition feature cluster according to the connection data, obtain abnormal matching results, and establish a mapping association between the abnormal matching results and the verification dimension; an verification evaluation result output module, which is used to perform verification evaluation according to the evaluation objectives of the verification dimension based on the abnormal matching results, and output the verification evaluation results.

[0010] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0011] Establish the association between the two-dimensional construction drawings and the building information model, use the model positioning identification information of the two-dimensional construction drawings as an index, search the display information in the building information model, output the linkage construction drawing block, and based on the linkage construction drawing block, map the connection data of the two-dimensional construction drawings and the linkage two-dimensional drawings according to the model positioning identification information; identify the multi-dimensional inspection features of the two-dimensional construction drawings, and establish the target drawing identification feature cluster; according to the connection data, perform feature anomaly matching with the inspection features in the target drawing identification feature cluster to obtain the anomaly matching result, and establish the mapping association between the anomaly matching result and the inspection dimension; perform inspection and evaluation according to the evaluation target of the inspection dimension according to the anomaly matching result, and output the inspection and evaluation result. The technical effect of improving the comprehensiveness and accuracy of inspection and improving the efficiency of inspection and verification is achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0013] Figure 1 A schematic diagram of a multi-dimensional review method for construction drawings provided in an embodiment of the present application;

[0014] Figure 2 A schematic diagram of a process for identifying multi-dimensional verification features of a two-dimensional construction drawing in a multi-dimensional verification method for construction drawings provided in an embodiment of the present application;

[0015] Figure 3 A schematic diagram of the structure of a multi-dimensional review system for construction drawings provided in an embodiment of the present application.

[0016] Explanation of the reference numerals: module 10 for establishing the association between two-dimensional construction drawings and building information models, module 20 for outputting linked construction drawing blocks, module 30 for mapping model positioning identification information, module 40 for establishing target drawing recognition feature clusters, module 50 for acquiring abnormal matching results, and module 60 for outputting inspection and evaluation results. DETAILED DESCRIPTION

[0017] The present application provides a multi-dimensional review method and system for construction drawings, which is used to solve the technical problems in the prior art such as low review efficiency, single review dimension and poor accuracy.

[0018] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0019] Embodiment 1

[0020] like Figure 1 As shown, the present application provides a multi-dimensional review method for construction drawings, the method comprising:

[0021] Step S100: establishing an association between a two-dimensional construction drawing and a building information model, wherein the building information model is a three-dimensional building model of a target building, wherein the two-dimensional construction drawing has model location identification information.

[0022] Specifically, establishing the association between two-dimensional construction drawings and building information models means building a clear correspondence between two-dimensional construction drawings of different dimensions and three-dimensional building information models, so that they can refer to and collaborate with each other. The building information model adopts the BIM model, which comprehensively and in detail covers various information of the target building, such as building structure, component parameters, materials, etc. It is a highly refined three-dimensional digital model of the target building. The two-dimensional construction drawings have model positioning identification information in the form of specific symbols, codes, coordinate annotations, etc., and their function is to accurately indicate the corresponding position or element of a part or element on the two-dimensional drawing in the three-dimensional building information model. For example, there may be a specific number next to a line segment on the two-dimensional construction drawing, and the same number is also on the corresponding three-dimensional component in the building information model. Through this number, the association between them can be established. Through such an association, designers and other relevant parties can more intuitively view the two-dimensional and three-dimensional information at the same time, deepen their understanding of the building structure and layout, reduce the difficulty and deviation of understanding, and more clearly find the inconsistencies between the two-dimensional drawings and the three-dimensional model, correct design errors in time, and reduce changes and rework during construction.

[0023] Step S200: Using the model positioning identification information of the two-dimensional construction drawing as an index, searching for display information in the building information model, and outputting a linked construction drawing block, wherein the linked construction drawing block includes all linked two-dimensional drawings that have a positional connection relationship with the two-dimensional construction drawing, and all linked two-dimensional drawings have corresponding model positioning identification information.

[0024] Specifically, the model location identification information of the two-dimensional construction drawings is used as an index, which means that these model location identification information is a unique label or code, such as a specific number, symbol combination or other uniquely identifiable elements. These index information are used as clues for accurate retrieval in a large amount of information; using this index to search in the building information model, which is a three-dimensional database containing all the details of the building. Through precise search operations, the display information corresponding to the index can be quickly and accurately found, including the building components, spatial layout and other specific contents. The output linkage construction drawing block is a sorted and associated collection, which contains other two-dimensional construction drawings that are directly connected to the original two-dimensional construction drawings in position. All linked two-dimensional drawings have corresponding model location identification information, which means that each linked two-dimensional drawing has its own unique model location identification information, and can establish a clear connection with the building information model and other related drawings through its own identification information. For example, a linked two-dimensional drawing has a specific number "ABC123", and the corresponding part in the building information model also has the same number identification, which ensures the accurate connection between the two, so that when viewing and using these drawings, you can more conveniently and quickly obtain comprehensive and accurate information, thereby greatly improving the work efficiency and quality in the design and construction processes.

[0025] Step S300: Based on the linked construction drawing block, the connection data of the two-dimensional construction drawing and the linked two-dimensional drawing are associated according to the model positioning identification information mapping.

[0026] Specifically, the linked construction drawing block has been screened and integrated through the previous steps, and contains a series of two-dimensional drawings that have a close positional connection relationship with the original two-dimensional construction drawings. According to the model positioning identification information mapping, the model positioning identification information here is like a precise pointer or coordinate point. Each identification information is unique and can accurately point out the position and identity of a specific two-dimensional construction drawing in the entire building information model. When we operate based on these identification information, we can accurately find the corresponding ones among many drawings and establish the connection between them. For example, the identification information of a two-dimensional construction drawing is "X001", and the identification information of another linked two-dimensional drawing that is connected to it may be "X001-A". Through the comparison and mapping of this identification information, we can clearly know that there is a close connection between them. The connection data associating the two-dimensional construction drawings with the linked two-dimensional drawings refers to the connection data associating the two-dimensional construction drawings with the linked two-dimensional drawings. The connection data includes detailed information such as relative position, connection angle, size coordination, etc. Through these connection data, you can immediately understand its specific relationship and connection method with other drawings, so that users can see the specific connection between the two-dimensional construction drawings and the linked two-dimensional drawings more clearly and intuitively, thereby improving the efficiency and accuracy of the review.

[0027] Step S400: Identify the multi-dimensional verification features of the two-dimensional construction drawings, and establish target drawing recognition feature clusters, each target drawing recognition feature cluster corresponding to a verification dimension.

[0028] Specifically, identifying the multi-dimensional inspection features of the two-dimensional construction drawings is to conduct a detailed and in-depth analysis of the two-dimensional construction drawings to find out the various features related to the inspection, such as checking the dimensioning of the building structure to ensure that the size of each part meets the design requirements during actual construction and there will be no size mismatch; paying attention to material specifications to clarify what materials should be used in construction to ensure building quality and performance; and construction requirements, including key information such as construction technology and processes. Establishing a target drawing recognition feature cluster is to classify and organize the various inspection features identified, and group features of similar nature or belonging to the same inspection category. For example, all dimensioning and shape features related to the building structure are grouped into one cluster, representing the structural dimension; features related to material selection and use are put together to form a cluster of the material dimension; and features related to construction safety form a cluster of the safety dimension, etc. Each target drawing identification feature cluster corresponds to an inspection dimension, which clarifies that each cluster has a specific inspection direction. The cluster of the structural integrity dimension is mainly used to inspect whether the building structure is complete and stable and whether it can withstand the expected load; the compliance dimension focuses on checking whether the drawings comply with relevant specifications, standards and regulatory requirements; the safety dimension focuses on evaluating whether the safety risks in the construction process and building use are effectively controlled. Through such division, the inspection work is more professional and refined, and the two-dimensional construction drawings can be comprehensively evaluated and reviewed from different angles to ensure that the drawings meet the requirements in all aspects and improve the efficiency and accuracy of the inspection.

[0029] Step S500: According to the connection data, feature anomaly matching is performed using the inspection features in the target drawing identification feature cluster to obtain anomaly matching results, and a mapping association is established between the anomaly matching results and the inspection dimension.

[0030] Specifically, the connection data uses the previously determined association data between the two-dimensional construction drawings and the linked two-dimensional drawings. These connection data provide important clues and references, making the subsequent inspection feature matching more accurate and targeted. The inspection features in the target drawing recognition feature cluster are used for feature anomaly matching. The various specific inspection features in the established target drawing recognition feature cluster, such as size, specification, requirements, etc., are compared and analyzed one by one with the preset standards or normal conditions to find out whether there are any irregularities. Obtaining the abnormal matching results is to extract the places that are found to be inconsistent with the normal situation and form a special result set. These anomalies may include dimensional errors, material mismatches, unmet construction requirements, etc. Establishing a mapping association between abnormal matching results and inspection dimensions is to correspond the abnormal matching results with specific inspection dimensions, and clarify whether these anomalies are problems with structural integrity, compliance, or safety. Doing so can clearly show in which specific inspection category the problem lies, making it easier to solve and deal with it in a targeted manner. It can also facilitate the classification, statistics and analysis of different types of problems to better evaluate the quality status of the drawings. Through such mapping and association, the entire inspection work can be more orderly and efficient.

[0031] Step S600: Performing verification and evaluation according to the evaluation objectives of the verification dimension based on the abnormal matching results, and outputting the verification and evaluation results.

[0032] Specifically, based on the abnormal matching results obtained above, the inspection and evaluation work is carried out in accordance with the evaluation objectives set for each inspection dimension. That is to say, for different inspection dimensions, such as structural integrity, compliance, safety, etc., according to their respective standards and requirements, the abnormal matching results are analyzed and evaluated to determine the status of the drawings under this dimension, whether they meet the requirements, what problems or risks exist, etc. Finally, the results of these evaluations are summarized, sorted and output to form a comprehensive evaluation conclusion on the entire two-dimensional construction drawings in various inspection dimensions, so that relevant personnel can understand the quality level of the drawings and the aspects that need to be improved. For example, in the structural integrity inspection dimension, the abnormal matching results are used to evaluate whether the structure is stable and has no defects; in the compliance inspection dimension, the drawings are judged based on the abnormalities to determine whether they comply with relevant laws and regulations and standards, etc., and then these evaluations are output in the form of specific results, such as giving qualified, unqualified or specific scores.

[0033] In a possible implementation, step S300 further includes:

[0034] Step S310: Identify the connection relationship between the two-dimensional construction drawing and the linked two-dimensional drawing, the connection relationship including the spatial structure connection relationship, the pipeline connection relationship, and the construction procedure connection relationship; Step S320: Perform feature analysis on each connection relationship respectively to determine the connection extension annotation feature; Step S330: Take the connection extension annotation feature as the target, perform feature search based on the connection relationship from the two-dimensional construction drawing, and obtain the connection data.

[0035] Specifically, it is necessary to carefully identify the connection relationship between the 2D construction drawings and the linked 2D drawings. The connection relationship here is mainly divided into three types. Spatial structure connection relationship, such as the mutual position and support relationship between different building structures; pipeline connection relationship, such as the connection direction of various pipelines such as water, electricity, and gas; construction procedure connection relationship, involving the order of construction and the relationship between different processes. This step aims to comprehensively sort out the various ways of connection between the two.

[0036] To analyze the characteristics of each connection relationship separately means to conduct in-depth and detailed analysis one by one for the previously identified spatial structure connection relationship, pipeline connection relationship, construction procedure connection relationship, etc. For the spatial structure connection relationship, the characteristics such as the shape, size, angle, etc. of the specific structure will be analyzed; for the pipeline connection relationship, the characteristics such as the diameter, material, and connection method of the pipeline will be analyzed; for the construction procedure connection relationship, the sequence requirements of each process, key nodes and other characteristics will be analyzed. Through this detailed analysis process, the connection extension annotation features are determined. These annotation features are representative and key characteristics extracted from the connection relationship, which can further clarify and refine the specific performance and requirements of the connection relationship. For example, a certain extension annotation feature determined in the spatial structure connection relationship may indicate that a specific connection method or angle is required between two structures to ensure stability; the annotation feature determined in the pipeline connection relationship may show that a certain section of the pipeline requires a specific turning radius to avoid excessive resistance;

[0037] The goal of connecting the extension annotation features is to take these already determined annotation features of great significance as the key objects for searching and analyzing; based on the connection relationship, the feature search is carried out from the two-dimensional construction drawings, which is based on the previously clear spatial structure connection relationship, pipeline connection relationship, construction procedure connection relationship, etc., and starts from the specific carrier of the two-dimensional construction drawings to find the content related to these annotation features in a targeted manner. In this search process, the various elements and details on the drawings are carefully compared to see whether they meet or reflect these annotation features. Through such a comprehensive and detailed search, the connection data is finally obtained. These connection data contain specific information related to the connection extension annotation features and the corresponding connection relationships, such as which parts have specific connection methods, which pipeline directions and characteristics meet the requirements, and which links in the construction process are related. These connection data play a key role in further understanding and processing the relationship between the two-dimensional construction drawings and the linked two-dimensional drawings, as well as in subsequent analysis, evaluation, decision-making, etc., and provide detailed data support for a deeper understanding of the situation of the drawings.

[0038] In one possible implementation, Figure 2 , step S400 further includes:

[0039] Step S410: Configure multi-dimensional inspection targets according to construction requirements; Step S420: Analyze and locate standardized annotations and differentiated annotations according to the multi-dimensional inspection targets; Step S430: Identify the two-dimensional construction drawings according to the standardized annotations to obtain standardized inspection features; Step S440: Establish a conversion mapping relationship for the differentiated annotations, and the conversion mapping relationship is used to expand the differentiated annotations and establish a unified identification conversion relationship for the differentiated annotations; Step S450: Map and identify the two-dimensional construction drawings based on the conversion mapping relationship to obtain differentiated inspection features, and the differentiated inspection features have a conversion relationship identifier; Step S460: Utilize the standardized inspection features and the differentiated inspection features of each dimension to form a multi-dimensional inspection feature, and the multi-dimensional inspection feature includes the inspection features of all dimensions.

[0040] Specifically, according to the construction requirements, it means to operate closely according to the relevant specifications, standards, technical indicators and project-specific requirements in the construction field. These construction requirements cover many aspects, such as structural stability and safety requirements, requirements for building function realization, construction technology and quality standards, etc. Configuring multi-dimensional inspection objectives reflects comprehensiveness and systematicness. Multi-dimensionality means that the inspection objectives are not limited to a single aspect, but are considered from multiple angles. These dimensions can include but are not limited to: structural strength inspection objectives, process compliance inspection objectives, function realization inspection objectives, etc. These objectives will serve as guidance and evaluation criteria for subsequent inspection work.

[0041] Analyzing and locating standardized annotations is to find annotations from drawings that follow generally accepted standards, specifications and conventions. These annotations are usually widely accepted and followed in the industry, representing general requirements and characteristics, such as common dimension annotations, material annotations, process annotations, etc., presented in standard formats and symbols. Differentiated annotations are annotations that are different from the usual, unique or project-specific, and may be due to special design requirements, specific technical requirements or differences from conventional practices.

[0042] Identify the two-dimensional construction drawings and use image recognition technology and other means to find and identify the specific elements or features corresponding to these standardized annotations in the two-dimensional construction drawings, and obtain standardized inspection features. These inspection features represent the specific aspects of the two-dimensional construction drawings that meet the standardization requirements, such as the size of a specific area meets standard specifications and the marking of a certain material meets conventional standards.

[0043] Establish the conversion mapping relationship of differentiated annotations. Through the study of different differentiated annotations, find out the association or correspondence between them. This relationship may be based on functions, logic, semantics, etc. For example, it is found that although some differentiated annotations are different in form, they actually point to the same type of information or operation. Through this analysis and the establishment of associations, a systematic conversion mapping relationship is formed. Some simple differentiated annotations should be expanded into more detailed and comprehensive information expressions, and a unified identification conversion relationship of differentiated annotations should be established to form a unified and standardized relationship system, so that no matter what kind of differentiated annotation is encountered, the corresponding standardized or understandable form can be found through this conversion mapping relationship, thereby achieving unified identification and conversion, eliminating the difficulties in understanding and processing caused by differentiated annotations, and improving the efficiency and accuracy of inspection work.

[0044] Using the established conversion mapping relationship, the differentiated annotations in the drawings are converted and identified according to the corresponding rules. Through this mapping process, information with specific meaning and value can be extracted from the differentiated annotations. This information constitutes the differentiated verification features, and these differentiated verification features also have conversion relationship identifiers, which help to clarify their source and conversion basis, and facilitate subsequent tracing and analysis.

[0045] Utilizing the standardized verification features and differentiated verification features of each dimension means integrating the standardized features and differentiated features obtained from different angles, classifying and combining them according to the corresponding dimensions, and forming verification features unique to each dimension. For example, in the structural dimension, it includes standardized verification features and differentiated verification features related to structural strength; in the process dimension, there is a corresponding feature set in terms of process. The multi-dimensional verification features include the final verification feature set that covers information from all relevant dimensions. The purpose of this is to more comprehensively and accurately evaluate and verify two-dimensional construction drawings. By integrating the verification features of each dimension, the drawings can be analyzed and judged from multiple angles, thereby improving the accuracy and reliability of the verification.

[0046] In a possible implementation, step S440 further includes:

[0047] Step S441: construct a differential annotation database, which includes abnormal case identification and drawing staff presets; step S442: use the differential annotation database for supervised training to obtain a conversion mapping relationship module; step S443: obtain a project construction sketch, which is a construction drawing of the same building as the current target construction drawing; step S444: use the project construction sketch to verify the conversion mapping relationship module to obtain loss characteristics; step S445: perform incremental training on the conversion mapping relationship module according to the loss characteristics to determine the conversion mapping relationship.

[0048] Specifically, building a differentiated annotation database is to create a system or data set specifically for storing and managing information related to differentiated annotation. The abnormal case identification part is to collect and organize special annotation cases that deviate from the norm in past drawings. For example, some annotation anomalies caused by misunderstandings, mistakes or special circumstances, such as incorrect dimension annotations, annotations with missing key information, etc. These abnormal cases can serve as an important reference for subsequent differentiated annotation analysis and processing, helping the system to better identify and respond to similar problems that may arise. The drafting staff preset focuses on recording the unique annotation styles, habits or possible personal characteristics of different drafting staff. Since different draftsmen have different ways of expression and preferences, incorporating this information into the database can take these individual differences into account in subsequent analysis and processing, so as to more accurately understand and process the annotation information. Such a database construction provides an important foundation and reference for the establishment of subsequent analysis, learning and conversion mapping relationships, which helps to improve the processing capabilities and adaptability of differentiated annotations.

[0049] The differentially labeled data in the database are divided into a training set and a validation set. The training set is used for the actual training of the model. Through a large amount of data, the model learns the patterns and rules in the data. The validation set is used to evaluate the performance of the model during the training process to help determine whether the model is overfitting or underfitting, so as to adjust the training strategy in time. Select a suitable machine learning model, such as a neural network. The neural network has a strong representation and learning ability and can handle complex data relationships. According to the characteristics of the problem and the nature of the data, different types of neural networks, such as multi-layer perceptrons and convolutional neural networks, are selected to better meet the processing needs of differentially labeled data. For each sample in the training set (including information such as abnormal cases and drawing staff presets), it is used as input, and the corresponding correct conversion mapping relationship is used as the target output. The learning goal of the model is clarified. The input sample contains various specific information. The model needs to extract features from this information and learn the association with the correct conversion mapping relationship, so that the model can gradually learn to predict the correct conversion mapping relationship based on the input information. The parameters of the model are continuously adjusted through the back propagation algorithm to minimize the error between the prediction result and the target output. The back propagation algorithm is a key algorithm in neural network training. According to the error between the prediction result and the target output, the gradient is calculated in reverse and the parameters of the model are updated. Through continuous iteration of this process, the model gradually optimizes the parameters so that the prediction result is closer and closer to the target output, thereby improving the accuracy and performance of the model. During the training process, the validation set is used to evaluate the performance of the model and make appropriate adjustments. After such a training process, the conversion mapping relationship module is finally obtained. This module has the ability to accurately output the corresponding conversion mapping relationship based on the input differentiated annotation information, and can convert complex and diverse differentiated annotations into clear, unambiguous forms that conform to specific rules or standards.

[0050] A project construction sketch is obtained through a specific channel or method, such as from a relevant database, file storage system, design team or other source, which is the same building as the current target construction drawing, and the construction sketch is used to verify and optimize the conversion mapping relationship module.

[0051] When the project construction sketch is input into the conversion mapping relationship module, the module will process the sketch according to its internal algorithms and parameters and try to generate the corresponding mapping results. However, due to the complexity of the actual situation and the possible imperfections of the module, there will be differences between the generated results and the ideal accurate mapping. These differences are quantified and analyzed to form loss characteristics. Loss characteristics can include various indicators, such as the deviation value between the predicted results and the actual situation in certain key parts, the degree of inaccuracy of conversion in different areas, etc. Through these loss characteristics, we can clearly understand the specific performance and shortcomings of the module when processing this specific construction sketch.

[0052] Based on the obtained loss characteristics, the conversion mapping relationship module can be incrementally trained in a targeted manner, and local adjustments and optimizations can be made according to the problems revealed by the loss characteristics. For example, in the incremental training of the conversion mapping relationship module, the model parameters, such as weights and biases, are initialized using stochastic gradient descent (SGD), and then the iterative process is started. In each iteration, one or a small number of samples (called a small batch) are randomly selected from the new project construction sketch data set, and the loss function values ​​corresponding to these samples are calculated according to the current model parameters. By deriving the loss function, the gradient of the model parameters is obtained. This gradient represents the direction in which the loss function rises fastest under the current parameter value, and then the parameters are updated according to the pre-set learning rate in the opposite direction of the gradient. The learning rate determines the magnitude of each parameter adjustment. By repeating this process continuously, the model parameters are gradually adjusted in the direction that reduces the loss function. Stochastic gradient descent SGD will select a small number of construction sketch samples each time, calculate the loss under the current model, and then fine-tune the weights and biases of the neural network according to the gradient and learning rate, in the hope of obtaining better mapping results when processing the next batch of samples. As the number of iterations increases, the model will gradually converge to a better parameter configuration, thereby achieving a more accurate conversion mapping relationship. Through this incremental training method, the performance of the module in processing specific construction sketches can be quickly improved while minimizing the impact on the existing good learning results. With continuous incremental training and optimization, the module's conversion mapping relationship will gradually become more accurate and reliable, and ultimately determine a conversion mapping relationship that better meets actual needs, so that the module can better cope with various construction drawings and sketch processing tasks.

[0053] In a possible implementation, step S500 further includes:

[0054] Step S510: construct an abnormal feature set, wherein the abnormal feature set has a dimension label; step S520: establish an identification path between the abnormal feature set and a target drawing identification feature cluster according to the dimension label; step S530: traverse the target drawing identification feature cluster using the abnormal feature set based on the identification path to obtain the abnormal matching result.

[0055] Specifically, the abnormal feature set is a collection of various features that may indicate abnormal situations. These features may include specific shapes, line patterns, color distributions, and other elements that are different from normal situations. Each feature is assigned a dimension label, which is used to clarify the specific dimension or attribute category to which the feature belongs, so as to facilitate subsequent accurate classification and processing.

[0056] According to the dimension labels, the identification path of the abnormal feature set and the target drawing recognition feature cluster is established. The identification path here is equivalent to an association mechanism or navigation method. Through the correspondence between the dimension labels, it is determined how to find the features or areas associated with the target drawing recognition feature cluster from the specific features in the abnormal feature set, and the logical connection between the abnormal feature set and the target drawing is established.

[0057] Each feature in the abnormal feature set is searched and matched in the target drawing recognition feature cluster according to the recognition path. In this process, each feature is carefully analyzed to see if there is any abnormal matching between it and the feature in the target drawing. If it is found that an abnormal feature is significantly different from the corresponding feature in the target drawing or does not meet the expected association, it is recorded and the abnormal matching result is finally obtained. This abnormal matching result can provide specific information about the abnormal situation in the target drawing to help further analysis and decision-making. For example, it can be determined whether the target drawing needs to be modified, adjusted or other corresponding measures need to be taken to solve possible problems based on the abnormal matching results.

[0058] In a possible implementation, step S510 further includes:

[0059] Step S511: extract abnormal case samples and cluster and segment them into multi-dimensional verification cases;

[0060] Step S512: extracting abnormal information of the audit cases in each dimension, performing feature comparison on the abnormal information, and determining the difference features;

[0061] Step S513: Obtain the abnormal feature set according to the difference features.

[0062] Specifically, from a large amount of case data, use specific algorithms or rules (such as standard deviation method, isolation forest algorithm, etc.) to accurately find those case samples with abnormal conditions. These abnormalities may be reflected in the data's numerical abnormalities, pattern abnormalities, trend abnormalities, etc. Then, use cluster analysis, a data processing technology, to classify and segment these abnormal case samples according to multiple different dimensions, such as time, space, attributes, behavior, etc., to form multiple inspection case groups with specific characteristics.

[0063] For each multi-dimensional verification case obtained through clustering, we deeply explore the specific abnormal information contained therein, which are specific numerical deviations, specific event sequences, unique behavior patterns, etc. These abnormal information are compared one by one to analyze their similarities and differences in characteristics. Through this comparison, we can clearly find those characteristics with significant differences, which represent the unique properties between different verification cases.

[0064] Based on the determined difference features, they are integrated into a set, which is the abnormal feature set. This abnormal feature set comprehensively covers the key abnormal features extracted from the inspection cases of various dimensions. When we encounter new situations or data, we can compare and match them with this set to quickly determine whether there are similar abnormal features, so as to achieve effective identification and early warning of abnormal situations.

[0065] In a possible implementation, step S513 further includes:

[0066] Step S5131: Obtain the precision constraint conditions of the construction, wherein the precision constraint conditions include multi-dimensional constraint conditions; Step S5132: Obtain new abnormal features of the construction, wherein the new abnormal features are new abnormal features customized by the construction users or identified by the self-check module; Step S5133: Use the multi-dimensional constraint conditions to correct the difference features in the abnormal feature set, and add the new abnormal features to the abnormal feature set.

[0067] Specifically, precision constraints are a series of regulations to ensure that the construction quality meets the requirements. Multi-dimensional constraints mean setting standards from multiple aspects, including size, vertical, flatness, material strength, etc. These constraints in different dimensions together constitute a comprehensive precision requirement system to ensure that the construction results meet the design and use requirements.

[0068] The newly added abnormal features customized by construction users are some situations that are not in line with the norm or expectations that construction users perceive at the construction site based on their own professional knowledge and experience. For example, construction users may find that the ground in a certain area has uneven settlement. Although this situation has not been clearly listed as an abnormality before, it is customized as a new abnormality based on actual observation and judgment; and the new abnormalities identified by the self-check module are usually obtained with the help of various advanced monitoring technologies and equipment. For example, using image recognition technology, new cracks or flaws are found on the surface of the building. The self-check module can more objectively and accurately detect some abnormalities that may be overlooked manually, and complement the user-defined abnormal features to jointly improve the grasp and management of abnormal situations in construction.

[0069] The use of multi-dimensional constraints to correct the difference features in the abnormal feature set is to re-examine the features that are considered to be different based on the previously determined accuracy requirements covering multiple aspects, such as size, verticality, flatness and other constraints. It may be found that the features that were previously considered abnormal are actually within a reasonable range under these detailed and comprehensive constraints, so they need to be adjusted to change their abnormal identification. Adding new abnormal features to the abnormal feature set ensures that new abnormal situations, whether defined by the user or identified by the self-check module, can be accurately recorded and summarized in the abnormal feature set. This allows the abnormal feature set to always accurately reflect the actual situation on the construction site without missing any new problems. For example, under the measurement of multi-dimensional constraints, a slightly uneven wall that was originally considered abnormal may be corrected to normal, and a newly discovered crack at a beam-column node is promptly added to the abnormal feature set for subsequent targeted processing and follow-up. Through such operations, the management and response to abnormal situations in construction can be made more scientific, accurate and effective.

[0070] In a possible implementation, the multi-dimensional review method for construction drawings further includes:

[0071] Based on the linked construction drawing block, a parallel verification channel is constructed, and the parallel verification channel is fitted with a multi-dimensional recognition module; the multi-dimensional recognition module is used to perform parallel recognition on the linked construction drawings respectively, and a linked anomaly recognition result is output, and the linked anomaly recognition result includes anomaly matching results of multiple linked construction drawings.

[0072] Specifically, a channel capable of conducting inspections at the same time is created based on the linked construction drawing blocks. This channel is not single-directional or single-functional, but has the ability to process in parallel. The parallel inspection channel is fitted with a multi-dimensional recognition module, which can examine the construction drawings from multiple different angles, including but not limited to the accuracy of geometric shapes, the precision of dimensions, the integrity of annotation information, and the compliance with process requirements. By fitting the multi-dimensional recognition module with the parallel inspection channel, the inspection work can be carried out from multiple levels at the same time, greatly improving the efficiency and comprehensiveness of the inspection.

[0073] By using a multi-dimensional recognition module, the identification work is carried out simultaneously on those interrelated linkage construction drawings; the parallel identification of the linkage construction drawings is carried out independently for each linkage construction drawing, but in parallel at the same time. In this way, all relevant drawings can be quickly scanned comprehensively. The output linkage abnormality identification result is a summary and presentation of the abnormal conditions of the entire linkage construction drawing set, among which, the abnormal matching results including multiple linkage construction drawings mean that this result reflects the abnormal conditions on each drawing, as well as the mutual relationship and matching of these abnormalities between multiple drawings. These abnormal matching results can more clearly understand the overall problems existing in the linkage construction drawings, and accurately point out the existing symptoms and mutual relationships.

[0074] In summary, the embodiments of the present application have at least the following technical effects:

[0075] The present application provides a multi-dimensional review method for construction drawings, which establishes an association between two-dimensional construction drawings and building information models, uses the model location identification information of the two-dimensional construction drawings as an index, searches for display information in the building information model, outputs a linked construction drawing block, and maps the connection data of the two-dimensional construction drawings and the linked two-dimensional drawings based on the model location identification information based on the linked construction drawing block; identifies the multi-dimensional review features of the two-dimensional construction drawings, establishes a target drawing identification feature cluster, performs feature anomaly matching with the review features in the target drawing identification feature cluster based on the connection data, obtains anomaly matching results, performs review and evaluation according to the evaluation target of the review dimension based on the anomaly matching results, and outputs the review and evaluation results. The technical effect of improving the comprehensiveness and accuracy of the review and verification and improving the review and verification efficiency is achieved.

[0076] Embodiment 2

[0077] Based on the same inventive concept as the multi-dimensional inspection method for construction drawings in the aforementioned embodiment, Figure 3 As shown, the present application provides a multi-dimensional review system for construction drawings, and the system and method embodiments in the present application embodiments are based on the same inventive concept. The system includes:

[0078] A module 10 for establishing an association between two-dimensional construction drawings and building information models is used to establish an association between two-dimensional construction drawings and building information models, wherein the building information model is a three-dimensional building model of a target building, wherein the two-dimensional construction drawings have model positioning identification information.

[0079] The linked construction drawing block output module 20 is used to search the display information in the building information model with the model positioning identification information of the two-dimensional construction drawing as an index, and output the linked construction drawing block. The linked construction drawing block includes all linked two-dimensional drawings that have a positional connection relationship with the two-dimensional construction drawing, and all linked two-dimensional drawings have corresponding model positioning identification information.

[0080] The model location identification information mapping module 30 is used to map and associate the connection data between the two-dimensional construction drawing and the linked two-dimensional drawing based on the linked construction drawing block according to the model location identification information.

[0081] The target drawing recognition feature cluster establishment module 40 is used to identify the multi-dimensional verification features of the two-dimensional construction drawings and establish target drawing recognition feature clusters, each of which corresponds to a verification dimension.

[0082] The abnormal matching result acquisition module 50 is used to perform feature abnormality matching based on the connection data and the inspection features in the target drawing identification feature cluster to obtain abnormal matching results and establish a mapping association between the abnormal matching results and the inspection dimensions.

[0083] The verification and evaluation result output module 60 is used to perform verification and evaluation according to the evaluation objectives of the verification dimension based on the abnormal matching results, and output the verification and evaluation results.

[0084] Furthermore, the model location identification information mapping module 30 also includes:

[0085] A connection relationship identification unit is used to identify the connection relationship between the two-dimensional construction drawing and the linked two-dimensional drawing, wherein the connection relationship includes a spatial structure connection relationship, a pipeline connection relationship, and a construction procedure connection relationship.

[0086] The connection extensibility annotation feature determination unit is used to perform feature analysis on each connection relationship respectively to determine the connection extensibility annotation feature.

[0087] A connection data acquisition unit is used to perform feature search based on the connection relationship from the two-dimensional construction drawing with the connection extensibility annotation feature as the target to obtain the connection data.

[0088] Furthermore, the target drawing recognition feature cluster building module 40 further includes:

[0089] A multi-dimensional verification target configuration unit is used to configure multi-dimensional verification targets according to construction requirements.

[0090] A labeling and parsing unit is used to parse and locate standardized labels and differentiated labels according to the multi-dimensional verification target.

[0091] A standardized inspection feature acquisition unit is used to identify the two-dimensional construction drawings according to the standardized annotations to obtain standardized inspection features.

[0092] A differential annotation conversion mapping relationship establishing unit is used to establish a conversion mapping relationship of the differential annotation, and the conversion mapping relationship is used to expand the differential annotation and establish a unified identification conversion relationship of the differential annotation.

[0093] A differentiated inspection feature acquisition unit is used to map and identify the two-dimensional construction drawing based on the conversion mapping relationship to obtain a differentiated inspection feature, wherein the differentiated inspection feature has a conversion relationship identifier.

[0094] A multi-dimensional verification feature building unit, wherein the multi-dimensional verification feature building unit is used to use the standardized verification features and the differentiated verification features of each dimension to build a multi-dimensional verification feature, wherein the multi-dimensional verification feature includes the verification features of all dimensions.

[0095] Furthermore, the differential annotation conversion mapping relationship establishing unit also includes:

[0096] A differential annotation database construction unit, wherein the differential annotation database construction unit is used to construct a differential annotation database, wherein the differential annotation database includes abnormal case identification and drawing staff presets.

[0097] A conversion mapping relationship module acquisition unit is used to perform supervised training using the differential annotation database to acquire a conversion mapping relationship module.

[0098] A project construction sketch acquisition unit, wherein the project construction sketch acquisition unit is used to acquire a project construction sketch, wherein the project construction sketch and the current target construction drawing are construction drawings of the same building.

[0099] A loss feature acquisition unit, wherein the loss feature acquisition unit is used to verify the conversion mapping relationship module using the project construction sketch to obtain loss features.

[0100] A conversion mapping relationship determination unit is used to perform incremental training on the conversion mapping relationship module according to the loss feature to determine the conversion mapping relationship.

[0101] Furthermore, the abnormal matching result acquisition module 50 also includes:

[0102] An abnormal feature set construction unit is used to construct an abnormal feature set, and the abnormal feature set has a dimension label.

[0103] An identification path unit, wherein the identification path unit is used to establish an identification path between the abnormal feature set and the target drawing identification feature cluster according to the dimension label.

[0104] An abnormal matching result acquisition unit is used to traverse the target drawing recognition feature cluster using the abnormal feature set based on the recognition path to obtain the abnormal matching result.

[0105] Furthermore, the abnormal feature set construction unit also includes:

[0106] The abnormal case sample extraction unit is used to extract abnormal case samples and cluster and segment them into multi-dimensional verification cases.

[0107] The difference feature determination unit is used to extract abnormal information of the inspection cases in each dimension, perform feature comparison on the abnormal information, and determine the difference features.

[0108] An abnormal feature set acquisition unit is used to obtain the abnormal feature set according to the difference feature.

[0109] Furthermore, the abnormal feature set acquisition unit also includes:

[0110] The precision constraint condition acquisition unit is used to obtain the precision constraint conditions of the building construction, and the precision constraint conditions include multi-dimensional constraint conditions.

[0111] A new abnormal feature acquisition unit is used to obtain new abnormal features of construction. The new abnormal features are new abnormalities that are customized by construction users or identified by a self-check module.

[0112] A difference feature correction unit is used to correct the difference features in the abnormal feature set by using the multi-dimensional constraint conditions, and add the newly added abnormal features to the abnormal feature set.

[0113] Furthermore, the multi-dimensional review system for construction drawings also includes:

[0114] A parallel verification channel construction module is used to construct a parallel verification channel based on the linked construction drawing block, and the parallel verification channel fits a multi-dimensional recognition module.

[0115] A linkage anomaly identification result output module is used to use the multi-dimensional identification module to perform parallel identification on the linkage construction drawings respectively, and output linkage anomaly identification results, wherein the linkage anomaly identification results include anomaly matching results of multiple linkage construction drawings.

[0116] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0117] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

[0118] This specification and the drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A multi-dimensional inspection method for construction drawings, characterized in that: The method comprises: Establishing an association between a two-dimensional construction drawing and a building information model, wherein the building information model is a three-dimensional building model of a target building, wherein the two-dimensional construction drawing has model location identification information; Using the model location identification information of the two-dimensional construction drawing as an index, searching for display information in the building information model, and outputting a linked construction drawing block, wherein the linked construction drawing block includes all linked two-dimensional drawings that have a position connection relationship with the two-dimensional construction drawing, and all linked two-dimensional drawings have corresponding model location identification information; Based on the linked construction drawing block, mapping and associating the connection data of the two-dimensional construction drawing and the linked two-dimensional drawing according to the model positioning identification information; Identify the multi-dimensional verification features of the two-dimensional construction drawings, and establish target drawing recognition feature clusters, each target drawing recognition feature cluster corresponding to a verification dimension; According to the connection data, feature anomaly matching is performed using the inspection features in the target drawing recognition feature cluster to obtain an anomaly matching result, and a mapping association is established between the anomaly matching result and the inspection dimension; Performing verification and evaluation according to the evaluation objectives of the verification dimension based on the abnormal matching results, and outputting the verification and evaluation results; Mapping and associating the connection data of the two-dimensional construction drawing and the linkage two-dimensional drawing according to the model positioning identification information includes: Identify the connection relationship between the two-dimensional construction drawing and the linkage two-dimensional drawing, wherein the connection relationship includes a spatial structure connection relationship, a pipeline connection relationship, and a construction procedure connection relationship; Perform feature analysis on each connection relationship to determine the connection extension annotation features; Taking the connection extension annotation feature as a target, performing feature search from the two-dimensional construction drawing based on the connection relationship to obtain the connection data; Identify the multi-dimensional verification features of the 2D construction drawings, including: Configure multi-dimensional inspection objectives according to construction requirements; According to the multi-dimensional verification objectives, analyze and locate standardized annotations and differentiated annotations; Identify the two-dimensional construction drawing according to the standardized annotation to obtain standardized inspection features; Establishing a conversion mapping relationship of the differential annotation, wherein the conversion mapping relationship is used to expand the differential annotation and establish a unified identification conversion relationship of the differential annotation; Based on the conversion mapping relationship, mapping and identifying the two-dimensional construction drawing is performed to obtain a differentiated inspection feature, where the differentiated inspection feature has a conversion relationship identifier; The standardized verification features and the differentiated verification features of each dimension are used to construct multi-dimensional verification features, and the multi-dimensional verification features include verification features of all dimensions.

2. The multi-dimensional inspection method for construction drawings according to claim 1, characterized in that: Establishing the conversion mapping relationship of the differential annotation includes: Constructing a differentiated annotation database, the differentiated annotation database including abnormal case identification and drawing staff preset; Using the differential annotation database to perform supervised training to obtain a conversion mapping relationship module; Obtain a project construction sketch, where the project construction sketch and the current target construction drawing are construction drawings of the same building; Using the project construction sketch to verify the conversion mapping relationship module to obtain loss characteristics; Incrementally train the conversion mapping relationship module according to the loss feature to determine the conversion mapping relationship.

3. The multi-dimensional inspection method for construction drawings according to claim 1, characterized in that: Perform feature anomaly matching with the verification features in the target drawing recognition feature cluster to obtain anomaly matching results, including: Constructing an abnormal feature set, wherein the abnormal feature set has a dimension label; Establishing an identification path of the abnormal feature set and the target drawing identification feature cluster according to the dimension label; Based on the identification path, the target drawing identification feature cluster is traversed using the abnormal feature set to obtain the abnormal matching result.

4. The multi-dimensional inspection method for construction drawings according to claim 3, characterized in that: The constructing of the abnormal feature set includes: Extract abnormal case samples and cluster and segment them into multi-dimensional verification cases; Extract abnormal information of audit cases in each dimension, compare the characteristics of abnormal information, and determine the difference characteristics; The abnormal feature set is obtained according to the difference features.

5. The multi-dimensional inspection method for construction drawings according to claim 4, characterized in that: Obtaining the abnormal feature set also includes: Obtaining precision constraints for building construction, wherein the precision constraints include multi-dimensional constraints; Obtaining newly added abnormal features of the construction, where the newly added abnormal features are newly added abnormal features that are defined by the construction user or identified by the self-checking module; The multi-dimensional constraint condition is used to modify the difference features in the abnormal feature set, and the newly added abnormal features are added to the abnormal feature set.

6. The multi-dimensional inspection method for construction drawings according to claim 1, characterized in that: The multi-dimensional verification method for construction drawings also includes: Based on the linked construction drawing blocks, a parallel verification channel is constructed, and the parallel verification channel is fitted with a multi-dimensional recognition module; The multi-dimensional recognition module is used to perform parallel recognition on the linked construction drawings respectively, and a linked anomaly recognition result is output, wherein the linked anomaly recognition result includes anomaly matching results of multiple linked construction drawings.

7. A multi-dimensional verification system for construction drawings, used to implement the multi-dimensional verification method for construction drawings as claimed in any one of claims 1 to 6, characterized in that: The system comprises: A module for establishing an association between a two-dimensional construction drawing and a building information model, wherein the module is used to establish an association between a two-dimensional construction drawing and a building information model, wherein the building information model is a three-dimensional building model of a target building, wherein the two-dimensional construction drawing has model location identification information; A linked construction drawing block output module, the linked construction drawing block output module is used to use the model positioning identification information of the two-dimensional construction drawing as an index, search the display information in the building information model, and output a linked construction drawing block, wherein the linked construction drawing block includes all linked two-dimensional drawings that have a position connection relationship with the two-dimensional construction drawing, and all linked two-dimensional drawings have corresponding model positioning identification information; A model positioning identification information mapping module, the model positioning identification information mapping module is used to map and associate connection data between the two-dimensional construction drawing and the linked two-dimensional drawing based on the linked construction drawing block according to the model positioning identification information; A target drawing recognition feature cluster establishment module, the target drawing recognition feature cluster establishment module is used to identify the multi-dimensional verification features of the two-dimensional construction drawings and establish target drawing recognition feature clusters, each target drawing recognition feature cluster corresponding to a verification dimension; An abnormal matching result acquisition module, the abnormal matching result acquisition module is used to perform feature abnormal matching with the verification feature in the target drawing recognition feature cluster according to the connection data, obtain an abnormal matching result, and establish a mapping association between the abnormal matching result and the verification dimension; The verification and evaluation result output module is used to perform verification and evaluation according to the evaluation objectives of the verification dimension based on the abnormal matching results, and output the verification and evaluation results.

Images