CN-122021232-A - Intelligent examination method and system for building construction diagram

CN122021232ACN 122021232 ACN122021232 ACN 122021232ACN-122021232-A

Abstract

The invention discloses an intelligent examination method and system for a building construction diagram, and belongs to the technical field of building construction. The technical problems that the traditional manual inspection mode is low in efficiency and easy to make mistakes, and the existing automatic system is difficult to process heterogeneous drawing data and understand complex building specifications are solved. The technical scheme is characterized in that the method comprises the steps of receiving original drawing files generated by different drawing software through a data interface, analyzing the original drawing files into standardized structured data objects, fusing two-dimensional geometric and text information based on the data objects, reconstructing the data objects into a light three-dimensional building information model with semantic attributes, converting building standard treatises into structured calculation logic rule nodes through natural language processing to form a standard knowledge graph, finally carrying out dynamic matching reasoning on the three-dimensional model and the knowledge graph, automatically identifying violations through parameterization calculation and logic judgment, and generating a visual inspection report. The invention is mainly used for realizing the automation, the intellectualization and the accurate examination of the building construction drawing.

Inventors

KONG DONGDONG
JIANG XUEYI
LIU YUFEI
LU KEXIN

Assignees

北京建院工程咨询有限公司

Dates

Publication Date: 20260512
Application Date: 20251201

Claims (10)

1. The intelligent examination method for the building construction drawing is characterized by comprising the following steps of: S1, standardized input and analysis of heterogeneous drawing data, namely receiving original building construction drawing electronic files generated by different computer aided design software through a data interface, calling a format analysis module, and uniformly analyzing and converting the original building construction drawing electronic files into standardized structured data objects containing geometric vector information, layer attribute information and text annotation information, wherein the geometric vector information comprises coordinates and topological relations of points, lines, planes and entities, the layer attribute information comprises layer names, lines and colors of components, and the text annotation information comprises dimension annotations, text descriptions and related geometric objects; S2, multi-source information fusion and three-dimensional building information model reconstruction, namely fusing two-dimensional geometric primitives and text labeling information through a component recognition rule base based on standardized structured data objects to generate three-dimensional building components with semantic attributes, and then aggregating the three-dimensional building components into a lightweight three-dimensional building information model; S3, knowledge mapping treatment of computer-readable standard treaty, namely pre-storing national and industry mandatory standard treaty related to a building construction drawing, and converting the national and industry mandatory standard treaty into a structured computer-executable calculation logic rule node through a natural language treatment technology to form a standard knowledge map; and S4, intelligent examination and conflict detection based on dynamic rule matching, namely dynamically matching and reasoning the lightweight three-dimensional building information model and the standard knowledge graph, carrying out parameterization calculation and logic judgment on the lightweight three-dimensional building information model according to calculation logic rule nodes, generating examination conflict records when detecting violations, and carrying out visual labeling and report generation.
2. The intelligent inspection method of building construction drawing according to claim 1, wherein in step S2, fusing two-dimensional geometric primitives with text labeling information through a component recognition rule base, generating a three-dimensional building component with semantic attributes specifically comprises: s211, component type identification, namely, based on the layer name in the layer attribute information, combining the primitive types in the geometric vector information with spatial distribution, performing matching through a predefined component identification rule base, and primarily identifying candidate building components and types thereof, wherein the types comprise walls, beams, columns, plates, doors and windows; S212, three-dimensional geometry generation, namely, calling a corresponding three-dimensional generation algorithm according to the type of the identified candidate building components, namely, stretching the wall and column components based on two-dimensional closed outlines of the wall and column components, wherein the stretching height is determined according to a height value or a default standard value in bound text marking information; And S213, reconstructing the topological relation, namely automatically calculating and recording the spatial topological relation among the components based on the spatial positions of all the generated three-dimensional building components, wherein the spatial topological relation comprises a connection relation, a supporting relation and a spatial inclusion relation.
3. The intelligent inspection method of building construction drawing according to claim 2, wherein in step S2, the aggregation into a lightweight three-dimensional building information model specifically comprises: S221, injecting semantic information, namely performing one-to-one or one-to-many association binding on the size, the material and the fire grade attribute related to the component in the text marking information and the three-dimensional building component generated in the step S212 to finish the injection of the semantic information; S222, model light-weight encapsulation, namely adopting an incremental light-weight algorithm to remove intermediate data and historical operation records in the three-dimensional modeling process, reserving final geometric data, semantic attribute data of building components and topological relation data among the components, and encapsulating to form a light-weight three-dimensional building information model, wherein the format of the light-weight three-dimensional building information model is independent of any specific computer aided design software environment.
4. The intelligent inspection method for building construction drawings according to claim 1, wherein the step S4 specifically comprises: S410, triggering inspection rules, namely traversing each building component entity and semantic attribute thereof in the lightweight three-dimensional building information model, and automatically matching and activating associated calculation logic rule nodes from the standard knowledge graph according to the component types; S420, parameterization calculation and logic judgment, namely extracting needed geometric parameters and semantic parameters from the lightweight three-dimensional building information model in real time according to activated calculation logic rule nodes, and carrying out parameterization mathematical calculation and space relation logic judgment; S430, generating and visually marking inspection results, namely comparing the parameterized calculation and logic judgment results with threshold values or conditions in rule nodes, if the rule is violated, generating an inspection conflict record containing the identification of the violating component, the violated specific specification, the violated parameter calculation value and the standard requirement value, and carrying out visual highlighting marking of the three-dimensional space position on all the inspection conflict records in a lightweight three-dimensional building information model, and generating a structured inspection report.
5. The intelligent inspection method of building construction drawing according to claim 1, wherein before dynamically matching and reasoning the lightweight three-dimensional building information model and the standard knowledge graph in step S4, further comprising dynamically screening out a target component set or target area from the lightweight three-dimensional building information model based on an inspection range selection instruction input by a user, and automatically expanding a target inspection domain based on a component topology relationship, wherein the dynamic matching and reasoning is performed only for the target component set or target area.
6. The intelligent inspection method for building construction drawing according to claim 1, wherein in step S4, after generating the inspection conflict record, the intelligent classification and classification step for the inspection conflict is further included, specifically: Automatically grading each audit conflict record according to a predefined conflict severity grade rule base, wherein the conflict severity grade at least comprises a major risk grade, a general violation grade and a suggested optimization grade; Meanwhile, according to the professional field of the rule and the treaty violated by the conflict record, the rule and the treaty are automatically classified, and the professional field at least comprises a fireproof safety class, a structural safety class, an energy-saving environment-friendly class and an ergonomic class; In the generated structured examination report and three-dimensional visual annotation, the examination conflict records are ordered, screened and differentially displayed according to the conflict severity level and the professional field classification.
7. The intelligent inspection method of building construction drawing according to claim 1, further comprising a model optimization feedback loop independent of the core inspection flow, the model optimization feedback loop being executed periodically or triggered, specifically comprising: after the user confirms or corrects the checking conflict record generated in the step S4, storing the finally confirmed conflict case, including the corresponding model component data, the triggered calculation logic rule node and the final judgment result, as a labeling sample into a history conflict sample library; the self-adaptive adjustment of the rule confidence coefficient is that based on a historical conflict sample library, the trigger frequency and the false alarm rate of each calculation logic rule node are counted, when the false alarm rate of a certain rule is continuously higher than a threshold value of ten percent, the system automatically reduces the confidence coefficient weight of the rule when the rule is matched and triggered, or gives an uncertainty prompt to a user when the rule is triggered; and (3) mining new rule patterns, namely performing association rule mining on cases in a historical conflict sample library, and submitting the potential new rule patterns to an administrator for auditing as optimization suggestions when two or more types of component violations are found to coexist frequently in the samples and the combination patterns are not covered by the existing standard knowledge patterns, so as to be used for expanding or correcting the standard knowledge patterns.
8. The intelligent inspection method of building construction drawings according to claim 7, further comprising an incremental inspection mode, which is triggered when the updating of a standard knowledge graph or the local revision of a lightweight three-dimensional building information model is detected, specifically comprising: Analyzing a change influence domain, namely analyzing newly added or modified calculation logic rule nodes when the standard knowledge graph is updated, and determining a building component type set of the associated influence; Dynamically constructing a target examination domain, namely performing union operation on the analysis-obtained influence component type set and the revised component set to generate a target examination domain; and (3) carrying out dynamic rule matching and reasoning, parameterized calculation and logic judgment on the components in the target examination domain, and generating an examination conflict record.
9. The intelligent inspection method of building construction drawing according to claim 1, wherein in step S4, a corresponding inspection logic traceability description is automatically generated while each inspection conflict record is generated, and the generation process of the inspection logic traceability description comprises: Automatically recording original specification treaty content of the computing logic rule node triggering the conflict; automatically extracting key model parameters and numerical values thereof used in the judgment, wherein the key model parameters at least comprise geometric parameters and semantic parameters extracted from the lightweight three-dimensional building information model in real time; based on the rule nodes and the key model parameters, automatically combining to generate a natural language description text containing the offence component identification, the violated rule treaty, the specific parameter values used, the logic judgment process and the final conclusion; And storing the audit logic traceability description and the corresponding audit conflict record in a correlated way, and outputting the audit logic traceability description and the corresponding audit conflict record to the structured audit report and the three-dimensional visual interface.
10. The system of the intelligent inspection method for the building construction map according to claims 1 to 9, comprising: The data analysis module is configured to receive the original building construction drawing electronic files generated by different computer aided design software through the data interface and analyze and convert the original building construction drawing electronic files into standardized structured data objects; The three-dimensional model reconstruction module is configured to construct a lightweight three-dimensional building information model fusing geometric and semantic information through a component recognition and three-dimensional generation algorithm based on the standardized structured data object; The standard knowledge graph module is configured to store and manage the structural calculation logic rule nodes obtained through conversion by the natural language processing technology and form a standard knowledge graph; the intelligent examination engine module is configured to dynamically match and infer the lightweight three-dimensional building information model and the standard knowledge graph, perform parameterization calculation and logic judgment and generate an examination conflict record; The inspection range selection module is configured to receive a user instruction, dynamically screen a target component set or a target area from the lightweight three-dimensional building information model, and allow the intelligent inspection engine module to conduct directional inspection; The conflict processing module is configured to automatically classify the severity level and the professional field of the review conflict record generated by the intelligent review engine module; The model optimization feedback module is configured to construct a historical conflict sample library, adaptively adjust rule confidence and mine a potential new rule mode based on the historical conflict sample library so as to optimize the standard knowledge graph module; an incremental review control module configured to automatically analyze the change impact domain and construct a target review domain upon detection of a model revision or specification update to trigger an incremental review mode; the logic traceability generation module is configured to automatically generate a natural language description containing the original specification treaty, key parameters and a judging process for each audit conflict record.

Description

Intelligent examination method and system for building construction diagram Technical Field The invention relates to the technical field of building construction. More particularly, the invention relates to an intelligent examination method and system for a building construction drawing. Background In the field of inspection of building construction drawings, the prior art mainly relies on manual inspection or automated auxiliary inspection based on two-dimensional computer-aided design drawings. The methods face difficulties in processing heterogeneous drawing data generated by different design software, and because of the difference between data formats and structures, the information analysis process is complex, and automatic data extraction and integration are difficult to realize. Further, the country and industry building codes under which the examination is based are mostly in the form of natural language text. Such specification treatises lack of structural semantics in computer systems that can be directly understood and executed, making it difficult for automated inspection systems to accurately understand and apply the complete logic and constraints of the specification, often still requiring the inspector to manually interpret and convert into operational instructions. In addition, the existing automatic inspection method is mostly operated on the level of a two-dimensional drawing, and it is difficult to accurately calculate geometric relationships and parameters in a three-dimensional space, such as real space distances between components or volumes of complex areas. Meanwhile, the methods generally carry out global batch examination on the drawings, can not flexibly focus on specific components or local areas according to actual demands, and when large projects or local design iterations are faced, a great amount of computation resources are consumed in the examination process, and the response efficiency is low. How to realize effective analysis of heterogeneous data, calculation and understanding of natural language specifications, and accurate and efficient space and range control of the examination process is a technical problem to be solved in the field. Disclosure of Invention It is an object of the present invention to solve at least the above problems and to provide at least the advantages to be described later. The invention also aims to provide an intelligent examination method and system for the building construction drawing, which solve the technical problems that the traditional manual examination mode is low in efficiency and easy to make mistakes, and the traditional automatic system is difficult to process heterogeneous drawing data and understand complex building specifications. And the automation, the intellectualization and the accurate examination of the building construction diagram are realized. To achieve these objects and other advantages and in accordance with the purpose of the invention, there is provided a construction map intelligent inspection method comprising the steps of: S1, standardized input and analysis of heterogeneous drawing data, namely receiving original building construction drawing electronic files generated by different computer aided design software through a data interface, calling a format analysis module, and uniformly analyzing and converting the original building construction drawing electronic files into standardized structured data objects containing geometric vector information, layer attribute information and text annotation information, wherein the geometric vector information comprises coordinates and topological relations of points, lines, planes and entities, the layer attribute information comprises layer names, lines and colors of components, and the text annotation information comprises dimension annotations, text descriptions and related geometric objects; S2, multi-source information fusion and three-dimensional building information model reconstruction, namely fusing two-dimensional geometric primitives and text labeling information through a component recognition rule base based on standardized structured data objects to generate three-dimensional building components with semantic attributes, and then aggregating the three-dimensional building components into a lightweight three-dimensional building information model; S3, knowledge mapping treatment of computer-readable standard treaty, namely pre-storing national and industry mandatory standard treaty related to a building construction drawing, and converting the national and industry mandatory standard treaty into a structured computer-executable calculation logic rule node through a natural language treatment technology to form a standard knowledge map; and S4, intelligent examination and conflict detection based on dynamic rule matching, namely dynamically matching and reasoning the lightweight three-dimensional building information model and the standard knowledge graph, c