Artificial intelligence-based bim data processing and construction quality evaluation method

Abstract

The application discloses a BIM data processing and building construction quality evaluation method based on artificial intelligence, and comprises the following steps: acquiring data, processing BIM model data to generate standard BIM component data; performing benchmark conversion to generate BIM component quality benchmark data; performing quality field coding to generate a BIM construction quality benchmark field; performing on-site construction semantic processing to generate an on-site semantic shadow body; mapping to the BIM construction quality benchmark field, performing component-level corresponding processing, generating a construction semantic corresponding relationship; performing component-level difference calculation to generate quality deviation elements, and performing component position aggregation and quality type merging to generate a construction quality deviation distribution map; generating a quality influence conduction relationship and a construction quality influence chain; and generating a building construction quality evaluation result based on the construction quality influence chain. The application realizes quality evaluation by fusing BIM and on-site data through AI, and has the advantages of accurate positioning, traceability and efficient rectification.

Description

Technical Field

[0001] This invention relates to the field of intelligent assessment technology for building construction quality, and in particular to a method for BIM data processing and building construction quality assessment based on artificial intelligence. Background Technology

[0002] Current construction quality management is usually based on BIM models, construction acceptance specifications, on-site measurement records, and manual inspection results. The dimensions, spatial location, construction procedures, and acceptance indicators of components are checked separately, and then management personnel record and track the rectification of abnormal components.

[0003] However, existing technologies mostly remain at the level of comparing single components or single indicators, making it difficult to uniformly map on-site images, point clouds, measurement records, and acceptance texts into the BIM quality benchmark, and also making it difficult to form component-level semantic correspondences. This results in unclear sources of quality deviations, insufficient component association, and an inability to accurately analyze the transmission path of quality risks and the priority of rectification.

[0004] Therefore, how to provide AI-based BIM data processing and building construction quality assessment methods is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] One objective of this invention is to propose an AI-based method for BIM data processing and construction quality assessment. This invention integrates BIM and on-site data using AI to achieve quality assessment, and has the advantages of accurate positioning, traceability, and efficient rectification.

[0006] The BIM data processing and construction quality assessment method based on artificial intelligence according to embodiments of the present invention includes the following steps: Acquire BIM model data, project quality control documents, construction quality acceptance documents, and on-site multi-source quality data; perform component analysis, coordinate unification, attribute cleaning, and floor merging on the BIM model data to generate standard BIM component data. Based on standard BIM component data and construction quality specification data, benchmark conversion is performed on component quality elements to generate BIM component quality benchmark data. Based on BIM component quality benchmark data, quality field coding is performed on component quality elements to generate a BIM construction quality benchmark field. Perform on-site construction semantic processing on multi-source quality data to generate on-site semantic shadow volumes; Map the on-site semantic shadow volume to the BIM construction quality benchmark field, and perform component-level correspondence processing on the on-site semantic shadow volume and the BIM construction quality benchmark field based on component quality elements to generate construction semantic correspondence relationship. Based on the semantic correspondence of construction, the component-level difference calculation of the on-site semantic shadow body and the BIM construction quality benchmark field is performed to generate quality deviation elements, and the component position is aggregated and the quality type is merged to generate a construction quality deviation distribution map. Based on the construction quality deviation distribution map, the quality impact transmission relationship is generated, and the initial deviation location, affected object association, and transmission path splicing are performed on the quality impact transmission relationship to generate the construction quality impact chain; The construction quality assessment results are generated based on the construction quality impact chain.

[0007] Optionally, the acquisition of the multi-source quality data on site includes: collecting construction site images, construction site point clouds, construction measurement records, and construction acceptance texts; The generation of the construction quality specification data includes parsing the quality items in the project quality control documents and construction quality acceptance documents, identifying component types, construction procedures, acceptance indicators, allowable deviation values, and quality grade determination content, and generating construction quality specification data.

[0008] Optionally, the generation of the BIM component quality benchmark data includes: A component quality element table is generated based on standard BIM component data. The component quality element table includes component number, component type, floor location, component identity, spatial location, size boundary, and component connection relationship. A quality benchmark entry table is generated based on construction quality specification data. The quality benchmark entry table includes component type, construction procedure, acceptance index, allowable deviation and quality grade. Using component type as the associated field, the construction procedures, acceptance indicators, allowable deviations and quality grades in the quality benchmark entry table are written into the data row of the corresponding component number in the component quality element base table to generate the component quality benchmark slot table. The component identification, spatial location, dimensional boundaries, component connection relationship, construction procedure, acceptance index, allowable deviation and quality grade in the component quality benchmark slot table are sorted into slots, and the components are merged at the component level according to component number, floor location and component connection relationship to generate BIM component quality benchmark data.

[0009] Optionally, the generation of the BIM construction quality benchmark field includes: Read the component quality elements from the BIM component quality benchmark data and generate a component quality code index according to the component number and floor location; Based on the component quality coding index, spatial location and size boundaries are mapped to the same floor coordinate range to generate a mass field coordinate grid and component boundary placeholders; Write the component identity into the component boundary placeholder and bind it with the component number to generate a quality boundary coding layer; Based on the quality boundary coding layer, the component connection relationship is mapped to the component boundary occupants to generate the component connection relationship coding layer; The construction sequence is written into the corresponding component boundary occupants and the relationship between adjacent components to generate a construction sequence coding layer; Write the acceptance criteria into the corresponding construction procedures and component boundary places to generate the acceptance criterion coding layer; The component numbers are aligned, the floor positions are aligned, and the inter-layer relationships are merged for the quality field coordinate grid, quality boundary coding layer, component connection relationship coding layer, construction procedure coding layer, and acceptance index coding layer to generate a BIM construction quality reference field.

[0010] Optionally, the generation of the on-site semantic shadow body includes: Image correction, component edge enhancement and semantic segmentation are performed on the construction site images to generate on-site image component contour data, which includes component contours and image coordinate positions. The point cloud at the construction site is subjected to noise removal, point cloud registration, component point cluster division and geometric surface reconstruction to generate spatial morphology data of the point cloud at the construction site. The construction survey records are cleaned, survey items are categorized, and components are associated with them to generate on-site measured index data; the construction acceptance documents are identified by process name, process completion status, and process-related components to generate on-site construction process status data. Spatial registration is performed based on image coordinates and component spatial locations to generate on-site component spatial registration data, which is then bound to on-site measured index data and on-site construction process status data to generate on-site component semantic merging data. Based on the on-site image component outline data, on-site point cloud spatial morphology data, on-site measured index data, and on-site construction procedure status data, the on-site component semantic merging data is used to merge components and merge the relationships between adjacent components, generating an on-site semantic shadow body.

[0011] Optionally, the generation of the construction semantic correspondence includes: Perform component-level reading of the on-site semantic shadow body, and generate on-site component quality element data according to the component quality elements; Component-level reading is performed on the BIM construction quality benchmark field, and BIM component quality element data is generated according to the component quality elements. The spatial locations of the on-site component quality element data and the BIM component quality element data are matched within the same coordinate range. The dimensions are matched for boundary coincidence. The component identity is matched in conjunction with the component identity to generate the initial component correspondence relationship. The component connection relationship in the initial component correspondence is matched with the adjacent component, the construction process is matched with the process name and process status, and the acceptance index is matched with the index name. The matching results of the adjacent component matching, the process name and process status matching, and the index name matching are written into the initial component correspondence to generate the enhanced component correspondence. Based on the completeness of the corresponding component quality elements, the correspondence of enhanced components is merged at the component level to generate construction semantic correspondence.

[0012] Optionally, the generation of the construction quality deviation distribution map includes: Based on the construction semantic correspondence, the spatial location, size boundary, component connection relationship, construction procedure status and acceptance index of the corresponding component are read from the site semantic shadow body and BIM construction quality benchmark field to generate component difference calculation pairs. Map the spatial location of the on-site component and the spatial location of the BIM component in the component difference calculation pair to the same floor coordinate range, and perform boundary coincidence calculation on the component geometric dimensions and dimension boundaries to generate a geometric difference image; The differences in connection objects, connection directions and contact positions of component connections are calculated, and the results of the connection difference calculations are written into a geometric difference film to generate a relationship difference film. The differences between the on-site construction process status and the construction process are calculated, the differences between the on-site measured indicators and the acceptance indicators are calculated, and the corresponding calculation results are written into the relationship difference film to generate the quality difference film. Based on the quality difference film, the differences in spatial location, size boundary, component connection relationship, construction procedure, and acceptance index are converted into quality deviation elements. The quality deviation elements are then bound to the component spatial location, connection relationship information, construction procedure information, and component boundary range to generate quality deviation association information. Based on the spatial location of components in the quality deviation association information, aggregate the component location of quality deviation elements within the same floor coordinate range to generate a component location deviation set. According to the quality deviation type, the component position deviation set is merged by quality type to generate quality type merging information. The quality deviation elements, quality deviation association information, component spatial location and quality type merging information are distributed and bound to generate a construction quality deviation distribution map.

[0013] Optionally, the generation of the construction quality impact chain includes: Read the quality deviation elements, quality deviation association information, component spatial location and quality type merging information from the construction quality deviation distribution map, and group the data corresponding to the same component spatial location into the same deviation record to generate a set of deviation trigger points; Using the set of deviation trigger points as the starting point for transmission, the deviation of the component connection relationship from the corresponding connection object, connection direction and contact position is written into the corresponding deviation record to generate the connection transmission base map; The deviation records in the connection transmission base map are associated with construction procedures. The construction procedures and procedures corresponding to the deviations are written into the corresponding deviation records to generate the process transmission base map. Spatial range association is performed on the deviation records in the process transmission base map. The floor position, component spatial position and component boundary range corresponding to the spatial position deviation and dimension boundary deviation are written into the corresponding deviation record to generate the spatial transmission base map. Merge transmission relationships based on the same deviation record to generate quality impact transmission relationships; The set of deviation trigger points is used to locate the initial deviation, generate the initial deviation component, and merge the affected components and affected construction procedures in the quality impact transmission relationship to generate the affected object. Then, the transmission path is spliced ​​according to the component connection relationship, construction procedure transmission relationship and spatial position transmission relationship to generate the construction quality impact chain.

[0014] Optionally, the generation of the building construction quality assessment results includes: Read the initial deviation component, quality deviation element, quality type merging information, affected objects, and transmission path in the construction quality impact chain; The initial deviation component is identified as an abnormal component, and a quality deviation type is generated according to the quality type merging information; Based on the correspondence between abnormal components and quality deviation types, generate sources of quality risk; Merge the affected components among the affected objects to generate risk-affected components; The transmission paths are merged to generate risk transmission paths; The risk levels of quality deviation types, sources of quality risks, components affecting risks, and risk transmission paths are merged to generate quality risk levels. Rectification priorities are generated based on quality risk level, affected components, and risk transmission path. Abnormal components, quality deviation type, source of quality risk, affected components, risk transmission path, quality risk level, and rectification priority are then combined into the construction quality assessment results.

[0015] The beneficial effects of this invention are: This invention unifies the processing of BIM model data, project quality control documents, construction quality acceptance documents, and multi-source on-site quality data. It integrates BIM component information, construction specifications, on-site images, on-site point clouds, measurement records, and acceptance texts into a single construction quality assessment process, avoiding the fragmentation of BIM models, on-site inspection data, and acceptance documents found in existing technologies. Through quality item parsing, component parsing, coordinate unification, attribute cleaning, and floor merging, standardized BIM component data and construction quality specification data are generated, providing subsequent quality assessments with clear data sources, unified component objects, and stable quality benchmarks.

[0016] This invention further generates BIM component quality benchmark data and BIM construction quality benchmark fields through component quality element benchmark transformation and quality field coding. This allows component identity, spatial location, dimensional boundaries, component connection relationships, construction procedures, and acceptance indicators to be expressed within the same quality field. Through quality boundary coding, component connection relationship coding, construction procedure coding, and acceptance indicator coding, on-site quality data is no longer just isolated inspection results, but can form a component-level correspondence with component boundaries, connection relationships, and construction procedures in BIM, thereby improving the accuracy and interpretability of quality comparison.

[0017] This invention generates a site semantic shadow volume through on-site construction semantic processing. It merges the component outlines in the construction site images, the spatial morphology in the point cloud, the measured indicators in the measurement records, and the construction process status in the acceptance text. This allows the actual construction status on site to participate in quality assessment in the form of structured component objects. Then, the site semantic shadow volume is mapped to the BIM construction quality benchmark field to generate a construction semantic correspondence. This can solve the problem of the difficulty in accurately corresponding on-site components with BIM components, reduce manual verification errors, and improve the automation level of construction quality identification.

[0018] This invention generates quality deviation elements through component-level difference calculation, and aggregates these elements by component location and quality type to form a construction quality deviation distribution map. This map simultaneously reflects spatial location deviation, dimensional boundary deviation, component connection relationship deviation, construction sequence deviation, and acceptance index deviation. This processing method not only locates abnormal components but also identifies the type of quality element in which the abnormality occurs, elevating the quality problem from simply discovering an anomaly to pinpointing the source of the deviation.

[0019] This invention also generates quality impact transmission relationships through a construction quality deviation distribution map, and further generates a construction quality impact chain, enabling the continuous expression of quality impacts between component connections, construction procedures, and spatial locations. The final generated building construction quality assessment results include abnormal components, quality deviation types, sources of quality risks, risk-affected components, risk transmission paths, quality risk levels, and rectification priorities. This provides a clear basis for on-site quality rectification, improving the completeness of quality risk identification, the pertinence of rectification decisions, and the level of intelligence in construction quality management. Attached Figure Description

[0020] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is an overall flowchart of the artificial intelligence-based BIM data processing and building construction quality assessment method proposed in this invention. Figure 2 This is a schematic diagram illustrating the process of generating on-site semantic shadow bodies from multi-source on-site quality data in the BIM data processing and construction quality assessment method based on artificial intelligence proposed in this invention. Figure 3 This is a schematic diagram illustrating the generation of the construction quality deviation distribution map in the AI-based BIM data processing and building construction quality assessment method proposed in this invention. Detailed Implementation

[0021] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0022] refer to Figures 1-3 The method for BIM data processing and construction quality assessment based on artificial intelligence includes the following steps: Acquire BIM model data, project quality control documents, construction quality acceptance documents, and on-site multi-source quality data, including construction site images, construction site point clouds, construction measurement records, and construction acceptance texts. Perform quality item parsing on the project quality control documents and construction quality acceptance documents to generate construction quality specification data. Perform component parsing, coordinate unification, attribute cleaning, and floor merging on the BIM model data to generate standard BIM component data. Based on standard BIM component data and construction quality specification data, the component quality elements are transformed to generate BIM component quality benchmark data. The component quality elements include component identity, spatial location, size boundaries, component connection relationship, construction sequence and acceptance index. Based on BIM component quality benchmark data, the component quality elements are encoded in a quality field to generate a BIM construction quality benchmark field. The quality field encoding includes quality boundary encoding, component connection relationship encoding, construction procedure encoding, and acceptance index encoding. On-site construction semantic processing is performed on multi-source quality data to generate on-site semantic shadow volumes. The on-site construction semantic processing includes component outline recognition, spatial morphology reconstruction, structuring of measured indicators, recognition of construction process status, spatial registration, and component merging. Map the on-site semantic shadow volume to the BIM construction quality benchmark field, and perform component-level correspondence processing on the on-site semantic shadow volume and the BIM construction quality benchmark field based on component quality elements to generate construction semantic correspondence relationship. Based on the semantic correspondence of construction, component-level differences are calculated between the on-site semantic shadow body and the BIM construction quality benchmark field to generate quality deviation elements. The quality deviation elements include spatial position deviation, size boundary deviation, component connection relationship deviation, construction procedure deviation and acceptance index deviation. The component position is aggregated and the quality type is merged to generate a construction quality deviation distribution map. Based on the construction quality deviation distribution map, a quality impact transmission relationship is generated, which includes component connection relationship transmission relationship, construction procedure transmission relationship and spatial location transmission relationship. The initial deviation location, affected object association and transmission path splicing are performed on the quality impact transmission relationship to generate a construction quality impact chain. The construction quality assessment results are generated based on the construction quality impact chain. The construction quality assessment results include abnormal components, quality deviation types, sources of quality risks, risk-affected components, risk transmission paths, quality risk levels, and rectification priorities.

[0023] In this embodiment, the acquisition of multi-source quality data on site includes: collecting construction site images, construction site point clouds, construction measurement records, and construction acceptance texts; the generation of construction quality specification data includes: parsing quality items in project quality control documents and construction quality acceptance documents, identifying component types, construction procedures, acceptance indicators, allowable deviation values, and quality grade judgment content, and generating construction quality specification data. The BIM model data, project quality control documents, construction quality acceptance documents, and on-site multi-source quality data specifically include: BIM model data including the 3F floor BIM model file BIM-F3-A01, beam components BIM-B-03012, column components BIM-C-03008, and slab components BIM-S-03005; project quality control documents including QC-F3-A01; construction quality acceptance documents including AC-F3-A01; and on-site multi-source quality data including construction site images. From MG-F3-001 to IMG-F3-080, the construction site point cloud PC-F3-001, the construction measurement record MR-F3-A01, and the construction acceptance text AT-F3-A01, after parsing the quality items, construction quality specification data is generated. The construction quality specification data includes the allowable deviation value of 5mm for beam component axis deviation, 8mm for cross-sectional dimension deviation, 5mm for flatness deviation, 5mm for column component verticality deviation, and 10mm for floor slab elevation deviation.

[0024] In this embodiment, the generation of BIM component quality benchmark data includes: A component quality element table is generated based on standard BIM component data. The component quality element table includes a component number field, a component type field, a floor location field, a component identity field, a spatial location field, a size boundary field, and a component connection relationship field. The generation of the component quality element table specifically includes: generating a component quality element table based on standard BIM component data. In the component quality element table, the beam component BIM-B-03012 corresponds to the component number field BIM-B-03012, the component type field beam, the floor location field 3F, the component identity field 3F beam component, the spatial location fields X=12400mm, Y=8600mm, Z=3200mm, the dimension boundary fields length 6000mm, width 300mm, height 500mm, and the component connection relationship fields BIM-C-03008 and BIM-S-03005. A quality benchmark entry table is generated based on construction quality specification data. The quality benchmark entry table includes fields for component type, construction procedure, acceptance index, allowable deviation, and quality grade. The generation of the quality benchmark entry table specifically includes: generating the quality benchmark entry table based on construction quality specification data. The quality benchmark entry table includes fields such as beam type for beam components, formwork installation, rebar tying, concrete pouring and curing for construction procedures, axis deviation, cross-sectional dimension deviation and flatness deviation for acceptance indicators, allowable deviation fields of 5mm, 8mm and 5mm, and qualified for quality grade. Associate the component type field in the component quality element base table with the component type field in the quality benchmark entry table, and write the construction procedure field, acceptance index field, allowable deviation field and quality grade field in the quality benchmark entry table into the data row of the corresponding component number in the component quality element base table to generate the component quality benchmark slot table. The generation of the component quality benchmark slot table specifically includes: associating the component type field beam in the component quality element base table with the component type field beam in the quality benchmark entry table, writing the construction procedure field, acceptance index field, allowable deviation field and quality grade field in the quality benchmark entry table into the data row of the corresponding component number in BIM-B-03012, and generating the component quality benchmark slot table. Based on the component quality benchmark slot table, the component identity field, spatial location field, size boundary field, component connection relationship field, construction procedure field, acceptance index field, allowable deviation field and quality grade field are written into the corresponding component quality element slots to generate slotted component quality benchmark data. The generation of the slotted component quality benchmark data specifically includes: based on the component quality benchmark slot table, writing the component identity, spatial location, size boundary, component connection relationship, construction procedure, acceptance index, allowable deviation field and quality grade field corresponding to BIM-B-03012 into the component quality element slot, forming the slotted component quality benchmark data corresponding to BIM-B-03012. The quality benchmark data of slotted components are merged at the component level according to component number, floor location and component connection relationship to generate BIM component quality benchmark data; The generation of BIM component quality benchmark data specifically includes: according to the component number BIM-B-03012, the floor location 3F, and the component connection relationship BIM-C-03008 and BIM-S-03005, the quality benchmark data of the slotted components are merged at the component level to form BIM component quality benchmark data.

[0025] In this embodiment, the generation of the BIM construction quality benchmark field includes: Read the component quality elements from the BIM component quality benchmark data, and arrange them at the component level according to the component number and floor location to generate a component quality code index; The generation of the component quality code index specifically includes: creating a file for each BIM component quality benchmark data according to the component number; writing BIM-B-03012, 3F, beam, 3F beam component, X=12400mm, Y=8600mm, Z=3200mm, length 6000mm, width 300mm, height 500mm, BIM-C-03008, BIM-S-03005, formwork installation, rebar binding, concrete pouring, curing, axis deviation 5mm, cross-sectional dimension deviation 8mm, and flatness deviation 5mm into the same component file; and then using the combination of BIM-B-03012 and 3F as the index key to form the component quality code index. Based on the component quality coding index, spatial location and size boundaries are mapped to the same floor coordinate range, generating a mass field coordinate grid, and component boundary placeholders are generated in the mass field coordinate grid; The generation of the mass field coordinate grid and component boundary placeholders specifically includes: reading the floor location 3F, spatial location X=12400mm, Y=8600mm, Z=3200mm, and dimension boundary length 6000mm, width 300mm, and height 500mm from the component mass code index; uniformly mapping the 3F floor spatial location to a millimeter-level floor coordinate range to generate the mass field coordinate grid; then converting the dimension boundary of BIM-B-03012 into a closed boundary line, with the closed boundary line plane coordinates including X=12400mm to 18400mm and Y=8450mm to 8750mm; writing the component number BIM-B-03012, boundary vertex coordinates, and boundary coverage coordinates into the corresponding coordinate positions in the mass field coordinate grid to generate the component boundary placeholders; Based on the mass field coordinate grid and component boundary placeholders, the component identity is written into the corresponding component boundary placeholder, and the component boundary placeholder is bound to the component number to generate a mass boundary coding layer; The generation of the quality boundary coding layer specifically includes: reading the mass field coordinate grid and component boundary placeholders, writing the component identity 3F beam component into the component boundary placeholder corresponding to BIM-B-03012, and binding the component boundary placeholder with the component number BIM-B-03012 to generate the quality boundary coding layer. Based on the quality boundary coding layer, the component connection relationship is mapped to the corresponding component boundary occupants to generate the component connection relationship coding layer; The generation of the component connection relationship coding layer specifically includes: reading the component number BIM-B-03012, component boundary occupancy and boundary coordinates in the mass boundary coding layer; writing the connection object number, connection direction X, contact surface coordinates X=18400mm, Y=8450mm to 8750mm, Z=2950mm to 3450mm between BIM-B-03012 and BIM-C-03008 into the same connection record; and then writing the connection record into the mass field coordinate grid position between the two component boundary occupancy to form the component connection relationship coding layer. Based on the component connection relationship coding layer, the construction procedure is written into the adjacent component relationship in the corresponding component boundary occupancy and component connection relationship coding layer to generate the construction procedure coding layer. The generation of the construction process coding layer specifically includes: reading the component number BIM-B-03012, connection record and adjacent component relationship in the component connection relationship coding layer; writing the formwork installation, rebar tying, concrete pouring and curing into the corresponding component boundary occupants of BIM-B-03012; writing the planned process status of transitioning to curing after concrete pouring into the concrete pouring process record; and then attaching the node rebar tying and node concrete pouring at the beam-column connection to the connection record of BIM-B-03012 and BIM-C-03008 to form the construction process coding layer. Based on the construction process coding layer, the acceptance indicators are written into the corresponding construction process and the corresponding component boundary occupants to generate the acceptance indicator coding layer. The generation of the acceptance index coding layer specifically includes: reading the component number BIM-B-03012, floor location 3F, construction procedure concrete pouring, and component boundary occupancy from the construction procedure coding layer; writing the allowable deviation value of 5mm for axis deviation, 8mm for cross-sectional dimension deviation, and 5mm for flatness deviation into the component boundary occupancy under the corresponding construction procedure; and then binding the acceptance index with BIM-B-03012, 3F, and concrete pouring to form the acceptance index coding layer. Based on the component quality coding index, the component number alignment, floor position alignment and inter-layer relationship merging are performed on the quality field coordinate grid, quality boundary coding layer, component connection relationship coding layer, construction procedure coding layer and acceptance index coding layer to generate a BIM construction quality benchmark field. The generation of the BIM construction quality benchmark field specifically includes: according to the component number BIM-B-03012 and the floor location 3F, aligning the quality field coordinate grid, quality boundary coding layer, component connection relationship coding layer, construction procedure coding layer and acceptance index coding layer between layers, merging the component space, component boundary, connection record, procedure record and acceptance index record into the same quality field entry to form the BIM construction quality benchmark field.

[0026] In this embodiment, the generation of the on-site semantic shadow body includes: Image correction, component edge enhancement and image semantic segmentation are performed on the construction site images to generate on-site image component contour data, which includes component contours and image coordinate positions. The generation of the on-site image component outline data specifically includes: performing image correction, component edge enhancement, and image semantic segmentation processing on the construction site images IMG-F3-001 to IMG-F3-080, recording the beam component outline as IMG-B-03012, and recording the image coordinate positions as the upper left corner pixel P1=1420, 860, and the lower right corner pixel P2=2360, 940, thus forming the on-site image component outline data; The point cloud at the construction site is subjected to noise removal, point cloud registration, component point cluster division and geometric surface reconstruction to generate spatial morphological data of the point cloud at the construction site. The spatial morphological data of the point cloud at the construction site includes the three-dimensional outline of the component, the spatial position of the component and the geometric dimensions of the component. The generation of the on-site point cloud spatial morphology data specifically includes: noise removal, point cloud registration, component point cluster division, and geometric surface reconstruction of the construction site point cloud PC-F3-001; the beam component point cluster is recorded as PC-B-03012; the on-site spatial location is recorded as X=12418mm, Y=8612mm, Z=3194mm; and the on-site geometric dimensions are recorded as length 6000mm, width 311mm, and height 497mm, thus forming the on-site point cloud spatial morphology data. The construction survey records are cleaned, survey items are classified, and components are associated to generate on-site measured index data, which includes axis deviation, elevation deviation, verticality deviation, flatness deviation, and cross-sectional dimension deviation. The generation of the on-site measured index data specifically includes: cleaning the fields, classifying the measurement items, and associating the components in the construction measurement record MR-F3-A01; writing the on-site measured indexes of PC-B-03012 that are involved in the quality calculation of this beam component into the same on-site measured index record; the on-site measured index record includes the measured value of axis deviation 18mm, the measured value of cross-sectional dimension deviation 11mm, and the measured value of flatness deviation 7mm, thus forming the on-site measured index data; The construction acceptance documents are used to identify the process name, process completion status, and process-related components to generate on-site construction process status data. The generation of the on-site construction process status data specifically includes: identifying the process name, process completion status and process-related components of the construction acceptance text AT-F3-A01; writing the process name "concrete pouring" corresponding to PC-B-03012, the on-site process status "concrete pouring completed but not yet transferred to curing", and the related component number PC-B-03012 into the same process status record to form on-site construction process status data. Spatial registration is performed based on the image coordinates in the component outline data and the component spatial position in the point cloud spatial morphology data to generate on-site component spatial registration data. On-site measured index data and on-site construction process status data are then bound to the on-site component spatial registration data to generate on-site component semantic merging data. The generation of the semantic merging data of the on-site components specifically includes: spatial registration based on the image coordinate position in the on-site image component contour data and the component spatial position in the on-site point cloud spatial morphology data, corresponding IMG-B-03012 with PC-B-03012 to generate on-site component spatial registration data, and then binding the measured value of axis deviation 18mm, the measured value of cross-sectional dimension deviation 11mm, the measured value of flatness deviation 7mm and the on-site process status of concrete pouring completed but not yet transferred to curing to PC-B-03012 to form on-site component semantic merging data; Based on the on-site image component outline data, on-site point cloud spatial morphology data, on-site measured index data, and on-site construction process status data, the on-site component semantic merging data is used to merge components, and the contact positions and spatial adjacency relationships of adjacent components within the same floor coordinate range are merged to generate an on-site semantic shadow volume. The on-site semantic shadow volume includes component outline, component three-dimensional outline, component spatial position, component geometric dimensions, component connection relationship, on-site measured index data, and on-site construction process status data. The generation of the on-site semantic shadow body specifically includes: reading the on-site image component outline data IMG-B-03012, the on-site point cloud spatial morphology data PC-B-03012, the on-site measured index data, and the on-site construction process status data; writing the component outline, component three-dimensional outline, component spatial position X=12418mm, Y=8612mm, Z=3194mm, component geometric dimensions length 6000mm, width 311mm, height 497mm, measured value of axis deviation 18mm, measured value of cross-sectional dimension deviation 11mm, measured value of flatness deviation 7mm, and the on-site process status of concrete pouring completed but not yet transferred to curing into the same on-site component object; merging the contact position and spatial adjacency relationship between PC-B-03012 and PC-C-03008 to form the on-site semantic shadow body.

[0027] In this embodiment, the generation of construction semantic correspondence includes: Perform component-level reading of the on-site semantic shadow body, and generate on-site component quality element data according to the component quality elements; The generation of the on-site component quality element data specifically includes: reading PC-B-03012 from the on-site semantic shadow body, and writing the component outline IMG-B-03012, the component three-dimensional outline PC-B-03012, the component spatial position X=12418mm, Y=8612mm, Z=3194mm, the component geometric dimensions length 6000mm, width 311mm, height 497mm, the component connection relationship PC-C-03008, the on-site process status concrete pouring completed but not yet transferred to curing, the on-site measured indicators axis deviation 18mm, cross-sectional dimension deviation 11mm, and flatness deviation 7mm into the same on-site component record to form the on-site component quality element data; Component-level reading is performed on the BIM construction quality benchmark field, and BIM component quality element data is generated according to the component quality elements. The generation of BIM component quality element data specifically includes: reading BIM-B-03012 in the BIM construction quality benchmark field, and writing the component identity (3F beam component), spatial location (X=12400mm, Y=8600mm, Z=3200mm), dimension boundary length (6000mm), width (300mm), height (500mm), component connection relationship (BIM-C-03008), construction process (formwork installation, rebar binding, concrete pouring, post-concrete pouring curing), and acceptance indicators (axis deviation 5mm, cross-sectional dimension deviation 8mm, flatness deviation 5mm) into the same BIM component record to form BIM component quality element data; The spatial locations of the on-site component quality element data and the BIM component quality element data are matched within the same coordinate range. The dimensions are matched for boundary coincidence. The component identity is matched in conjunction with the component identity to generate the initial component correspondence relationship. The generation of the initial component correspondence specifically includes: mapping PC-B-03012 and BIM-B-03012 to the same 3F floor coordinate range; calculating the coordinate difference between the on-site spatial position X=12418mm, Y=8612mm, Z=3194mm of PC-B-03012 and the spatial position X=12400mm, Y=8600mm, Z=3200mm of BIM-B-03012, resulting in an X-axis coordinate difference of 18mm, a Y-axis coordinate difference of 12mm, and a Z-axis coordinate difference of -6mm, forming a position correspondence record; calculating the overlap between the on-site boundary vertex coordinate sequence and the BIM closed boundary vertex coordinate sequence, resulting in a planar boundary overlap area of ​​1.72m², a boundary offset distance of 18mm, and a boundary corresponding vertex sequence; and finally writing the position correspondence record, boundary correspondence record, and component identity correspondence result into the same component correspondence entry. The component connection relationship in the initial component correspondence is matched with the adjacent component, the construction process is matched with the process name and process status, and the acceptance index is matched with the index name. The matching results of the adjacent component matching, the process name and process status matching, and the index name matching are written into the initial component correspondence to generate the enhanced component correspondence. The generation of the enhanced component correspondence specifically includes: reading PC-B-03012 and BIM-B-03012 from the initial component correspondence; writing the on-site connection object PC-C-03008 and the BIM connection object BIM-C-03008 into the component connection correspondence result; writing the on-site process name "concrete pouring" and the on-site process status "concrete pouring completed but not yet transferred to curing" into the construction process correspondence result; writing the measured values ​​of axial deviation (18mm), cross-sectional dimension deviation (11mm), and flatness deviation (7mm) with the allowable deviation values ​​of axial deviation (5mm), cross-sectional dimension deviation (8mm), and flatness deviation (5mm) into the acceptance index correspondence result; and then writing the component connection correspondence result, construction process correspondence result, and acceptance index correspondence result into the initial component correspondence to generate the enhanced component correspondence. Based on the completeness of the corresponding component quality elements, the correspondence of enhanced components is merged at the component level to generate construction semantic correspondence. The generation of the construction semantic correspondence specifically includes: reading the component identity correspondence result, spatial position correspondence result, size boundary correspondence result, component connection relationship correspondence result, construction procedure correspondence result, and acceptance index correspondence result from the enhanced component correspondence; writing PC-B-03012 and BIM-B-03012 into the same component correspondence entry; writing the spatial position correspondence result as X-axis coordinate difference 18mm, Y-axis coordinate difference 12mm, and Z-axis coordinate difference -6mm; writing the size boundary correspondence result as the overlapping area of ​​the plane boundary 1.72m² and the boundary offset distance 18mm; writing the component connection relationship correspondence result as PC-C-03008 corresponding to BIM-C-03008; writing the construction procedure correspondence result as concrete pouring completed but not yet transferred to curing corresponding to the planned procedure status of concrete pouring completed and then transferred to curing; and writing the acceptance index correspondence result as axis deviation 18mm corresponding to axis deviation 5mm, cross-sectional dimension deviation 11mm corresponding to cross-sectional dimension deviation 8mm, and flatness deviation 7mm corresponding to flatness deviation 5mm, thus forming the construction semantic correspondence.

[0028] In this embodiment, the generation of the construction quality deviation distribution map includes: Based on the construction semantic correspondence, the spatial location, geometric dimensions, connection relationships, on-site construction process status, and on-site measured indicators of the corresponding on-site components are read from the on-site semantic shadow body. The spatial location, dimension boundaries, connection relationships, construction process, and acceptance indicators of the corresponding BIM components are read from the BIM construction quality benchmark field to generate component difference calculation pairs. The generation of the component difference calculation pair specifically includes: reading the on-site component spatial position X=12418mm, Y=8612mm, Z=3194mm corresponding to PC-B-03012 according to the construction semantic correspondence; the on-site component geometric dimensions: length 6000mm, width 311mm, height 497mm; the on-site component connection relationship PC-C-03008; the on-site process status: concrete pouring completed but not yet transferred to curing; and the on-site measured indicators: axis deviation 18mm, cross-sectional dimension deviation 11mm, and flatness. The deviation is 7mm. Then, read the spatial location corresponding to BIM-B-03012: X=12400mm, Y=8600mm, Z=3200mm, dimension boundary length 6000mm, width 300mm, height 500mm, component connection relationship BIM-C-03008, construction process after concrete pouring and subsequent curing and acceptance indicators: axis deviation 5mm, cross-sectional dimension deviation 8mm, flatness deviation 5mm. Write the above on-site component data and BIM component data into the same component difference calculation pair. The spatial positions of on-site components and BIM components in the component difference calculation pair are mapped to the same floor coordinate range, and the geometric dimensions and dimensional boundaries of the components are calculated to generate a geometric difference film. The geometric difference film is used to record spatial position differences and dimensional boundary differences. The generation of the geometric difference film specifically includes: unifying the on-site spatial positions X=12418mm, Y=8612mm, Z=3194mm of PC-B-03012 with the spatial positions X=12400mm, Y=8600mm, Z=3200mm of BIM-B-03012 to the coordinate range of the 3F floor, generating spatial position differences, including X-axis coordinate difference of 18mm, Y-axis coordinate difference of 12mm and Z-axis coordinate difference of -6mm; calculating the boundary coincidence of the on-site boundary vertex coordinate sequence of PC-B-03012 with the closed boundary vertex coordinate sequence of BIM-B-03012, generating dimensional boundary differences, including planar boundary overlap area of ​​1.72m², boundary offset distance of 18mm and boundary offset direction of X-axis offset; and then writing the spatial position differences and dimensional boundary differences into the same component difference carrier to form a geometric difference film. The differences in connection objects, connection directions and contact positions of the component connection relationships in the component difference calculation pair are calculated, and the results of the connection relationship difference calculation are written into the geometric difference film to generate a relationship difference film. The relationship difference film is used to record spatial position differences, size boundary differences and component connection relationship differences. The generation of the relationship difference film specifically includes: reading the on-site component connection relationship of PC-B-03012, which includes the connection object PC-C-03008, the connection direction X, and the contact center position X=18418mm, Y=8612mm, Z=3194mm; reading the BIM component connection relationship of BIM-B-03012, which includes the connection object BIM-C-03008, the connection direction X, and the contact center position X=18400mm, Y=8600mm, Z=3200mm; writing the corresponding difference of the connection object, the corresponding difference of the connection direction, and the coordinate difference of the contact center position into the component connection relationship difference field, wherein the coordinate difference of the contact center position includes 18mm in the X direction, 12mm in the Y direction, and -6mm in the Z direction; and then writing the component connection relationship difference field into the geometric difference film to form the relationship difference film. The process of calculating the difference between the on-site construction process status and the construction process is performed in the component difference calculation pair. The difference between the on-site measured index and the acceptance index is calculated. The results of the process status difference calculation and the index difference calculation are written into the relationship difference film to generate a quality difference film. The quality difference film is used to record the spatial position difference, size boundary difference, component connection relationship difference, construction process difference and acceptance index difference of the same component. The generation of the quality difference film specifically includes: reading the on-site construction process status of PC-B-03012, which includes the process name "concrete pouring" and the on-site process status "concrete pouring completed but not yet transferred to curing"; reading the construction process of BIM-B-03012, which includes formwork installation, rebar tying, concrete pouring, concrete pouring followed by curing, and curing; mapping the on-site construction process status to the BIM construction process in a text structure to form the construction process difference; and reading the on-site measured indicators of PC-B-03012, which include... The measured values ​​of axis deviation, cross-sectional dimension deviation, and flatness deviation are 18mm, 11mm, and 7mm respectively. The acceptance criteria of BIM-B-03012 are read. The acceptance criteria include allowable deviation values ​​of 5mm for axis deviation, 8mm for cross-sectional dimension deviation, and 5mm for flatness. The measured values ​​and allowable deviation values ​​of the same criteria are calculated to obtain the difference values ​​of axis deviation, 13mm, cross-sectional dimension deviation, and flatness deviation, 3mm, and 2mm respectively. The differences in construction procedures and acceptance criteria are then written into the relationship difference film to form the quality difference film. Based on the quality difference film, the differences in spatial location, size boundary, component connection relationship, construction procedure, and acceptance index are converted into spatial location deviation, size boundary deviation, component connection relationship deviation, construction procedure deviation, and acceptance index deviation, respectively, generating quality deviation elements. The quality deviation elements are then bound to the corresponding component spatial location, connection relationship information, construction procedure information, and component boundary range to generate quality deviation association information. The generation of the quality deviation elements and quality deviation association information specifically includes: writing the spatial position differences in the quality difference film into the spatial position deviation field, which includes an X-axis coordinate difference of 18mm, a Y-axis coordinate difference of 12mm, and a Z-axis coordinate difference of -6mm; writing the dimensional boundary differences into the dimensional boundary deviation field, which includes the overlapping area of ​​the planar boundaries of 1.72m², the boundary offset distance of 18mm, and the boundary offset direction of X-axis offset; and writing the differences in component connection relationships into the component connection relationship deviation field, which includes the contact center position coordinate difference of 18mm in the X-axis, Y-axis, and Z-axis coordinate differences of -6mm. For the Z direction (12mm) and Z direction (-6mm), the differences in construction procedures are written into the construction procedure deviation field. The construction procedure deviation field includes concrete pouring completed but not yet transferred to curing. The differences in acceptance indicators are written into the acceptance indicator deviation field. The acceptance indicator deviation field includes axis deviation difference of 13mm, cross-sectional dimension deviation difference of 3mm, and flatness deviation difference of 2mm, forming quality deviation elements. Then, the quality deviation elements are written into the same associated entry along with the component spatial position X=12400mm, Y=8600mm, Z=3200mm, connection relationship information, construction procedure information, and component boundary range, forming quality deviation associated information. Based on the spatial location of components in the quality deviation association information, aggregate the component location of quality deviation elements within the same floor coordinate range to generate a component location deviation set. The generation of the component position deviation set specifically includes: reading the component spatial positions X=12400mm, Y=8600mm, and Z=3200mm from the quality deviation association information; aggregating the spatial position deviation, dimension boundary deviation, component connection relationship deviation, construction procedure deviation, and acceptance index deviation corresponding to BIM-B-03012 to the same floor coordinate position; writing the aggregated quality deviation elements into the 3F floor component position entry; and generating the component position deviation set. Based on spatial location deviation, dimensional boundary deviation, component connection relationship deviation, construction procedure deviation, and acceptance index deviation, the component location deviation set is merged into quality types to generate quality type merging information. The quality deviation elements, quality deviation association information, component spatial location, and quality type merging information are then distributed and bound to generate a construction quality deviation distribution map. The generation of the construction quality deviation distribution map specifically includes: reading the quality deviation element corresponding to BIM-B-03012 in the component position deviation set; writing the spatial position deviation into the spatial position quality type field; writing the dimension boundary deviation into the dimension boundary quality type field; writing the component connection relationship deviation into the component connection relationship quality type field; writing the construction procedure deviation into the construction procedure quality type field; and writing the acceptance index deviation into the acceptance index quality type field, thus forming quality type merging information. The quality type merging information includes 5 quality deviation types. Then, BIM-B-03012, 3F, component spatial position X=12400mm, Y=8600mm, Z=3200mm, quality deviation element, quality deviation association information, and quality type merging information are written into the same distribution entry to form the construction quality deviation distribution map.

[0029] In this embodiment, the generation of the construction quality influence chain includes: Read the quality deviation elements, quality deviation association information, component spatial location and quality type merging information from the construction quality deviation distribution map, and group the quality deviation elements, quality deviation association information and quality type merging information corresponding to the same component spatial location into the same deviation record to generate a set of deviation trigger points; The generation of the deviation trigger point set specifically includes: reading the distribution entries in the construction quality deviation distribution map, merging the quality deviation elements, quality deviation association information, and quality type merging information corresponding to BIM-B-03012 at the coordinates X=12400mm, Y=8600mm, Z=3200mm on the 3F floor into a single deviation record. The deviation record includes the component number BIM-B-03012, floor location 3F, component spatial location, quality deviation elements, quality deviation association information, and quality type merging information. Then, the deviation records are aggregated according to the component spatial location to form the deviation trigger point set. Using the set of deviation trigger points as the starting point for transmission, the deviation of the component connection relationship from the corresponding connection object, connection direction and contact position is written into the corresponding deviation record to generate the connection transmission base map; The generation of the connection transmission base map specifically includes: reading the component connection relationship deviation and connection relationship information in the deviation record, writing the connection object BIM-C-03008, connection direction X, contact center position X=18400mm, Y=8600mm, Z=3200mm, contact center position coordinate difference X direction 18mm, Y direction 12mm, and Z direction -6mm into the same deviation record, then reading the spatial adjacency relationship between BIM-B-03012 and BIM-S-03005, writing the connection object BIM-S-03005 into the same deviation record, and arranging the deviation records according to the spatial adjacency relationship between BIM-B-03012, BIM-C-03008, and BIM-S-03005 to form the connection transmission base map; Associate the construction process with the corresponding deviation record in the connection transmission base map, write the construction process deviation and the corresponding construction process and process status into the corresponding deviation record, and generate the process transmission base map; The generation of the process transmission base map specifically includes: reading the construction process information carried by the deviation record in the connection transmission base map, writing the construction process name "concrete pouring", the process status "concrete pouring completed but not transferred to curing", and the associated component number BIM-B-03012 into the corresponding deviation record, and then attaching the construction process information to the deviation record where the connected object BIM-C-03008 is located, so that the component connection information and the construction process information are on the same transmission carrier, forming the process transmission base map; Spatial range association is performed on the corresponding deviation records in the process transmission base map. The floor position, component spatial position and component boundary range corresponding to the spatial position deviation and dimension boundary deviation are written into the corresponding deviation record to generate the spatial transmission base map. The generation of the spatial transmission base map specifically includes: reading the component spatial position and component boundary range carried by the deviation record in the process transmission base map, writing the floor position 3F, component spatial position X=12400mm, Y=8600mm, Z=3200mm, component boundary range length 6000mm, width 300mm, height 500mm and boundary coordinate sequence into the corresponding deviation record, and then arranging the deviation records according to the floor position 3F and component spatial position to align the connection information, process information and spatial information to form a spatial transmission base map; Based on the same deviation record, the transmission relationships of the connection objects, connection directions, construction procedures, procedure status, floor locations, component spatial locations, and component boundary ranges in the spatial transmission base map are merged to generate quality influence transmission relationships. The quality influence transmission relationships include component connection relationship transmission relationships, construction procedure transmission relationships, and spatial location transmission relationships. The generation of the quality impact transmission relationship specifically includes: reading the deviation record in the spatial transmission base map, merging the connection objects BIM-C-03008 and BIM-S-03005, the connection direction X and the contact center position coordinate difference X-direction 18mm, Y-direction 12mm, Z-direction -6mm into the component connection relationship transmission relationship, merging the construction process name concrete pouring, the process status concrete pouring completed but not transferred to curing, and the associated component number BIM-B-03012 into the construction process transmission relationship, merging the floor position 3F, the component spatial position X=12400mm, Y=8600mm, Z=3200mm, and the component boundary range into the spatial position transmission relationship, and then writing the component connection relationship transmission relationship, the construction process transmission relationship, and the spatial position transmission relationship into the same transmission relationship object to form the quality impact transmission relationship; The set of deviation trigger points in the spatial transmission base map is used to locate the initial deviation and generate the initial deviation component. The affected components and affected construction procedures in the quality influence transmission relationship are merged to generate the affected object. The transmission path is spliced ​​according to the component connection relationship, construction procedure transmission relationship and spatial position transmission relationship to generate the transmission path. The quality deviation element, quality type merging information, initial deviation component, affected object and transmission path are chained together to generate the construction quality influence chain. The generation of the construction quality impact chain specifically includes: reading the set of deviation trigger points and deviation records in the spatial transmission base map; merging BIM-B-03012 carrying the deviation record into the starting deviation component; reading the connection objects BIM-C-03008, BIM-S-03005, construction procedure name concrete pouring, and component spatial position X=12400mm, Y=8600mm, Z=3200mm in the quality impact transmission relationship, merging them into affected objects; and then splicing them into a transmission path according to the component connection relationship transmission relationship, construction procedure transmission relationship, and spatial position transmission relationship. The transmission path includes the component connection path from BIM-B-03012 to BIM-C-03008, the component connection path from BIM-B-03012 to BIM-S-03005, the concrete pouring procedure transmission path, and the 3F floor spatial position transmission path. Finally, the quality deviation elements, quality type merging information, starting deviation component, affected objects, and transmission path carried by the deviation record are written into the same chain record to form the construction quality impact chain.

[0030] In this embodiment, the generation of construction quality assessment results includes: Read the initial deviation component, quality deviation element, quality type merging information, affected objects, and transmission path in the construction quality impact chain; The reading of the construction quality impact chain specifically includes: reading the chain records in the construction quality impact chain. The chain records include the starting deviation component BIM-B-03012, quality deviation elements, quality type merging information, affected objects BIM-C-03008 and BIM-S-03005, and the transmission path. The quality type merging information includes spatial location quality type, dimension boundary quality type, component connection relationship quality type, construction procedure quality type, and acceptance index quality type. The initial deviation component is identified as an abnormal component, and a quality deviation type is generated according to the quality type merging information; The generation of the quality deviation type specifically includes: identifying the initial deviation component BIM-B-03012 as an abnormal component, converting the quality type merging information in the construction quality influence chain into a quality deviation type, which includes spatial location deviation, dimension boundary deviation, component connection relationship deviation, construction procedure deviation, and acceptance index deviation. Based on the correspondence between abnormal components and quality deviation types, generate sources of quality risk; The generation of the quality risk sources specifically includes: according to the correspondence between abnormal component BIM-B-03012 and quality deviation type, writing the spatial position deviation, dimension boundary deviation, component connection relationship deviation, construction procedure deviation and acceptance index deviation of BIM-B-03012 into the quality risk source record to form the quality risk source; Merge the affected components among the affected objects to generate risk-affected components; The generation of the risk impact component specifically includes: reading the affected components BIM-C-03008 and BIM-S-03005 in the affected object, and merging BIM-C-03008 and BIM-S-03005 into a risk impact component; The transmission paths are merged to generate risk transmission paths; The generation of the risk transmission path specifically includes: reading the component connection paths from BIM-B-03012 to BIM-C-03008, the component connection paths from BIM-B-03012 to BIM-S-03005, the concrete pouring process transmission path, and the 3F floor spatial location transmission path from the transmission path, and writing the four types of paths into the same path record to form a risk transmission path; The risk levels of quality deviation types, sources of quality risks, components affecting risks, and risk transmission paths are merged to generate quality risk levels. The generation of the quality risk level specifically includes: reading the quality deviation type, quality risk source, risk-affected component, and risk transmission path; writing the number of 5 quality deviation types, the number of 2 risk-affected components, the length of 4 risk transmission paths, and the spatial position deviation of the beam component as the source of quality risk into the same risk degree vector; dividing the number of 5 quality deviation types by the total number of 5 quality deviation types to get 1.00; dividing the number of 2 risk-affected components by the total number of associated components to get 0.50; dividing the length of 4 risk transmission paths by the maximum number of path segments to get 0.67; and then generating the quality risk level according to the normalized risk degree vector. Rectification priorities are generated based on quality risk level, affected components, and risk transmission path. Abnormal components, quality deviation type, source of quality risk, affected components, risk transmission path, quality risk level, and rectification priority are then combined into the construction quality assessment results. The generation of the rectification priority specifically includes: reading the quality risk level, the components affected by the risk, and the risk transmission path; using the quality risk level as the primary sorting item, and using the number of two components affected by the risk and the length of four segments of the risk transmission path as secondary sorting items to form a rectification sorting sequence. The rectification sorting sequence includes the abnormal component number BIM-B-03012, the quality risk level, the number of components affected by the risk (2), the length of four segments of the risk transmission path, and the sorting position. Then, the rectification priority is generated according to the rectification sorting sequence, and the abnormal component, quality deviation type, quality risk source, components affected by the risk, risk transmission path, quality risk level, and rectification priority are written into the same assessment result record.

[0031] Example 1: To verify the feasibility of this invention in practice, it was applied to a construction quality assessment scenario for the main structure of an office building. The construction site simultaneously contained BIM model data, quality control documents, construction quality acceptance documents, site images, site point clouds, construction measurement records, and construction acceptance texts. Due to the large number of beams, columns, and slabs within the floors, and the similar shapes of adjacent components, traditional manual verification requires repeated switching between the BIM model, measurement records, acceptance texts, and site photos. This easily leads to problems such as inaccurate correspondence between on-site components and BIM components, unclear sources of quality deviations, omissions of affected components, and difficulty in determining the rectification sequence. Especially at beam-column connections and beam-slab junctions, single-dimensional verification cannot explain whether deviations have affected adjacent components and subsequent processes.

[0032] In this scenario, the system first acquires BIM model data, quality control documents, construction quality acceptance documents, and multi-source on-site quality data. It then parses the quality control documents and construction quality acceptance documents for quality items to generate construction quality specification data. Next, it performs component parsing, coordinate unification, attribute cleaning, and floor merging on the BIM model data to generate standard BIM component data. Taking the fifth-floor beam component BIM-L-05018 as a sample, its BIM spatial location is X=9820mm, Y=7350mm, Z=15800mm, and its dimensional boundaries are length 4200mm, width 250mm, and height 450mm. Connecting components include column component BIM-C-05006 and slab component BIM-P-05011. Acceptance indicators include an allowable deviation of 5mm for axis deviation, 8mm for cross-sectional dimension deviation, and 5mm for flatness deviation. The system writes the component identity, spatial location, dimensional boundaries, component connection relationships, construction procedures, and acceptance indicators of the beam component into the BIM component quality benchmark data, and generates a BIM construction quality benchmark field through quality boundary coding, component connection relationship coding, construction procedure coding, and acceptance indicator coding.

[0033] In field applications, the system performs component contour recognition on site images, spatial morphology reconstruction on site point clouds, index structuring processing on construction measurement records, and process status recognition on construction acceptance texts, generating site semantic shadow entities. The site component corresponding to BIM-L-05018 is denoted as PC-L-05018, with site spatial locations of X=9831mm, Y=7366mm, and Z=15794mm. Its site geometric dimensions are length 4210mm, width 258mm, and height 447mm. The measured site indicators include axis deviation of 12mm, cross-sectional dimension deviation of 9mm, and flatness deviation of 6mm. The site process status is that the formwork has not been closed after the rebar binding is completed. The system maps the on-site semantic shadow to the BIM construction quality benchmark field, generates the construction semantic correspondence, and performs component-level difference calculations on the on-site components and BIM components, resulting in an X-axis coordinate difference of 11mm, a Y-axis coordinate difference of 16mm, a Z-axis coordinate difference of -6mm, an axis deviation difference of 7mm, a cross-sectional dimension deviation difference of 1mm, and a flatness deviation difference of 1mm. At the same time, it generates construction sequence deviation and component connection relationship deviation.

[0034] The system further aggregates and categorizes spatial location deviations, dimensional boundary deviations, component connection relationship deviations, construction sequence deviations, and acceptance index deviations by component location and quality type, forming a construction quality deviation distribution map. Based on this map, the system uses beam component BIM-L-05018 as the starting deviation component, column component BIM-C-05006, and slab component BIM-P-05011 as affected components, generating component connection relationship transmission relationships, construction sequence transmission relationships, and spatial location transmission relationships, ultimately forming a construction quality impact chain. The construction quality assessment results record abnormal component BIM-L-05018, with quality deviation types including spatial location deviation, dimensional boundary deviation, component connection relationship deviation, construction sequence deviation, and acceptance index deviation. The source of quality risk is the inconsistency between the spatial positioning and the process status of the beam component. The risk transmission paths include beam-column connection paths, beam-slab connection paths, and process transmission paths. Its rectification priority is ranked among the highest of the same batch of components.

[0035] In the sample data processing for the fourth quarter of 2025, 96 on-site images, 2 sets of on-site point clouds, 24 construction measurement records, and 12 construction acceptance documents from the five-story main structure construction area were selected, covering a total of 32 beam, column, and slab components. When using manual verification, the average time to complete data correspondence, deviation recording, and document archiving for a single component was 5.8 minutes; after adopting this invention, this time was reduced to 2.1 minutes. Manual verification required an average of 3 re-checks of the BIM model, measurement records, and acceptance documents for each quality deviation found; after adopting this invention, this was reduced to 1. Manual verification revealed 7 instances of missing associated records for affected components; after adopting this invention, this was reduced to 1. The average time to manually generate a rectification sequence was 8.6 minutes; after adopting this invention, this was reduced to 3.0 minutes. Manually completing the review of components in the same batch required 2 quality personnel working together for half a working day; after adopting this invention, the review and confirmation could be completed by 1 quality personnel in approximately 2 hours. The above sample data shows that the present invention can reduce the number of manual data switching, shorten the component correspondence and quality recording time, reduce the number of missing associations of affected components, and improve the efficiency of generating rectification sequence, thereby making the construction quality assessment results more complete and intuitive, and the subsequent rectification arrangements more targeted.

[0036] Table 1. Comparison of Performance Samples in the Entire BIM Construction Quality Assessment Process

[0037] In terms of processing time, the traditional manual verification method takes 185.6 minutes to complete the corresponding deviation records and data archiving for 32 components, the BIM-assisted manual verification method takes 147.2 minutes, the BIM and point cloud geometric comparison method takes 108.8 minutes, and the present invention takes 67.2 minutes. Compared with the traditional manual verification method, the present invention reduces time by 118.4 minutes; compared with the BIM and point cloud geometric comparison method, the present invention still reduces time by 41.6 minutes. This result shows that the present invention does not simply reduce manual input steps, but rather, through the corresponding processing of the BIM construction quality benchmark field and the on-site semantic shadow body, it centralizes on-site images, on-site point clouds, measurement records, and acceptance texts under the same component object, reducing the time spent by humans switching, searching, and confirming between different data.

[0038] In terms of the number of data reviews, traditional manual verification methods require an average of 3.0 reviews per quality deviation, BIM-assisted manual verification requires 2.4 reviews, BIM-point cloud geometric comparison requires 1.9 reviews, and this invention requires only 1.0 review. This metric aligns with the actual work patterns in on-site quality management: manual methods require multiple reviews of BIM models, measurement records, and acceptance documents; while geometric comparison reduces the model positioning step, it still requires manual review of process documents and acceptance items. This invention, by generating the on-site semantic shadow body, binds the component outline, point cloud morphology, measured indicators, and process status. Subsequent mapping to the BIM construction quality benchmark field directly establishes a construction semantic correspondence, thus further reducing the number of reviews.

[0039] From the perspective of the average residual value of corresponding component positions, the traditional manual verification method is 18.2mm, the BIM-assisted manual verification method is 14.7mm, the BIM and point cloud geometric comparison method is 9.6mm, and the present invention is 5.4mm. This set of data is reasonable in terms of engineering common sense, because manual visual inspection and table verification are easily affected by the similar appearance of components, shooting angle, and on-site occlusion. The BIM and point cloud geometric comparison method can improve spatial positioning, but it is insufficient in terms of mutual constraints between image contours, measurement indicators, and process text. The present invention writes multi-source data into the on-site semantic shadow volume and performs component-level semantic correspondence in the BIM construction quality benchmark field, which can reduce mismatches between adjacent beams, columns, and slabs with similar geometric coordinates, thus resulting in smaller position residuals.

[0040] In terms of the number of quality deviations missed, the traditional manual verification method missed 6 instances, the BIM-assisted manual verification method missed 4 instances, the BIM and point cloud geometric comparison method missed 3 instances, and the present invention missed 1 instance. This result shows that the present invention has more complete coverage of quality deviation types. The traditional manual verification method is more likely to focus on explicit indicators such as axis, dimension and flatness, while the BIM and point cloud geometric comparison method focuses more on spatial position and geometric boundary. However, the present invention generates spatial position deviation, dimensional boundary deviation, component connection relationship deviation, construction procedure deviation and acceptance index deviation simultaneously, which can classify geometric problems, relationship problems, procedure problems and acceptance index problems into the same construction quality deviation distribution map, thus significantly reducing the number of omissions.

[0041] In terms of the number of missed correlations for affected components, the traditional manual verification method missed 7 correlations, the BIM-assisted manual verification method missed 5 correlations, the BIM and point cloud geometric comparison method missed 4 correlations, and the present invention missed 1 correlation. This improvement mainly comes from the generation of the construction quality impact chain; the traditional method usually records problems around the abnormal component itself, and it is difficult to continue to track the impact on beam-column connections, beam-slab junctions, and subsequent processes; the present invention generates component connection relationship transmission relationship, construction process transmission relationship, and spatial location transmission relationship based on the construction quality deviation distribution map, and coherently expresses the initial deviation component, affected components, and transmission path, so that quality risk is transformed from single-point recording to chain-like traceability results.

[0042] In terms of the average time taken to form the rectification sequence, the traditional manual verification method takes 8.6 minutes, the BIM-assisted manual verification method takes 6.9 minutes, the BIM and point cloud geometric comparison method takes 5.1 minutes, and the present invention takes 3.0 minutes. The reason for this reduction is that the construction quality assessment results output by the present invention not only include abnormal components and quality deviation types, but also the source of quality risk, the components affected by the risk, the risk transmission path, the quality risk level, and the rectification priority. Quality personnel no longer need to separately analyze the impact range of each abnormal component, nor do they need to check adjacent components and process records one by one, thus shortening the time for forming the rectification sequence.

[0043] As can be seen from the data in Table 1, the present invention has a stable improvement in component matching speed, number of data review times, position matching accuracy, deviation record integrity, impact on component association and rectification sequence generation, and there are no calculation contradictions among the various data.

[0044] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for BIM data processing and construction quality assessment based on artificial intelligence, characterized in that: Includes the following steps: Acquire BIM model data, project quality control documents, construction quality acceptance documents, and on-site multi-source quality data; perform component analysis, coordinate unification, attribute cleaning, and floor merging on the BIM model data to generate standard BIM component data. Based on standard BIM component data and construction quality specification data, benchmark conversion is performed on component quality elements to generate BIM component quality benchmark data. Based on BIM component quality benchmark data, quality field coding is performed on component quality elements to generate a BIM construction quality benchmark field. Perform on-site construction semantic processing on multi-source quality data to generate on-site semantic shadow volumes; Map the on-site semantic shadow volume to the BIM construction quality benchmark field, and perform component-level correspondence processing on the on-site semantic shadow volume and the BIM construction quality benchmark field based on component quality elements to generate construction semantic correspondence relationship. Based on the semantic correspondence of construction, the component-level difference calculation of the on-site semantic shadow body and the BIM construction quality benchmark field is performed to generate quality deviation elements, and the component position is aggregated and the quality type is merged to generate a construction quality deviation distribution map. Based on the construction quality deviation distribution map, the quality impact transmission relationship is generated, and the initial deviation location, affected object association, and transmission path splicing are performed on the quality impact transmission relationship to generate the construction quality impact chain; The construction quality assessment results are generated based on the construction quality impact chain.

2. The method for BIM data processing and construction quality assessment based on artificial intelligence according to claim 1, characterized in that, The acquisition of the multi-source quality data on site includes: collecting construction site images, construction site point clouds, construction measurement records, and construction acceptance texts; The generation of the construction quality specification data includes parsing the quality items in the project quality control documents and construction quality acceptance documents, identifying component types, construction procedures, acceptance indicators, allowable deviation values, and quality grade determination content, and generating construction quality specification data.

3. The method for BIM data processing and construction quality assessment based on artificial intelligence according to claim 1, characterized in that, The generation of the BIM component quality benchmark data includes: A component quality element table is generated based on standard BIM component data. The component quality element table includes component number, component type, floor location, component identity, spatial location, size boundary, and component connection relationship. A quality benchmark entry table is generated based on construction quality specification data. The quality benchmark entry table includes component type, construction procedure, acceptance index, allowable deviation and quality grade. Using component type as the associated field, the construction procedures, acceptance indicators, allowable deviations and quality grades in the quality benchmark entry table are written into the data row of the corresponding component number in the component quality element base table to generate the component quality benchmark slot table. The component identification, spatial location, dimensional boundaries, component connection relationship, construction procedure, acceptance index, allowable deviation and quality grade in the component quality benchmark slot table are sorted into slots, and the components are merged at the component level according to component number, floor location and component connection relationship to generate BIM component quality benchmark data.

4. The method for BIM data processing and construction quality assessment based on artificial intelligence according to claim 1, characterized in that, The generation of the BIM construction quality benchmark field includes: Read the component quality elements from the BIM component quality benchmark data and generate a component quality code index according to the component number and floor location; Based on the component quality coding index, spatial location and size boundaries are mapped to the same floor coordinate range to generate a mass field coordinate grid and component boundary placeholders; Write the component identity into the component boundary placeholder and bind it with the component number to generate a quality boundary coding layer; Based on the quality boundary coding layer, the component connection relationship is mapped to the component boundary occupants to generate the component connection relationship coding layer; The construction sequence is written into the corresponding component boundary occupants and the relationship between adjacent components to generate a construction sequence coding layer; Write the acceptance criteria into the corresponding construction procedures and component boundary places to generate the acceptance criterion coding layer; The component numbers are aligned, the floor positions are aligned, and the inter-layer relationships are merged for the quality field coordinate grid, quality boundary coding layer, component connection relationship coding layer, construction procedure coding layer, and acceptance index coding layer to generate a BIM construction quality reference field.

5. The method for BIM data processing and construction quality assessment based on artificial intelligence according to claim 1, characterized in that, The generation of the on-site semantic shadow body includes: Image correction, component edge enhancement and semantic segmentation are performed on the construction site images to generate on-site image component contour data, which includes component contours and image coordinate positions. The point cloud at the construction site is subjected to noise removal, point cloud registration, component point cluster division and geometric surface reconstruction to generate spatial morphology data of the point cloud at the construction site. The construction survey records are cleaned, survey items are categorized, and components are associated with them to generate on-site measured index data; the construction acceptance documents are identified by process name, process completion status, and process-related components to generate on-site construction process status data. Spatial registration is performed based on image coordinates and component spatial locations to generate on-site component spatial registration data, which is then bound to on-site measured index data and on-site construction process status data to generate on-site component semantic merging data. Based on the on-site image component outline data, on-site point cloud spatial morphology data, on-site measured index data, and on-site construction procedure status data, the on-site component semantic merging data is used to merge components and merge the relationships between adjacent components, generating an on-site semantic shadow body.

6. The method for BIM data processing and construction quality assessment based on artificial intelligence according to claim 1, characterized in that, The generation of the construction semantic correspondence includes: Perform component-level reading of the on-site semantic shadow body, and generate on-site component quality element data according to the component quality elements; Component-level reading is performed on the BIM construction quality benchmark field, and BIM component quality element data is generated according to the component quality elements. The spatial locations of the on-site component quality element data and the BIM component quality element data are matched within the same coordinate range. The dimensions are matched for boundary coincidence. The component identity is matched in conjunction with the component identity to generate the initial component correspondence relationship. The component connection relationship in the initial component correspondence is matched with the adjacent component, the construction process is matched with the process name and process status, and the acceptance index is matched with the index name. The matching results of the adjacent component matching, the process name and process status matching, and the index name matching are written into the initial component correspondence to generate the enhanced component correspondence. Based on the completeness of the corresponding component quality elements, the correspondence of enhanced components is merged at the component level to generate construction semantic correspondence.

7. The method for BIM data processing and construction quality assessment based on artificial intelligence according to claim 1, characterized in that, The generation of the construction quality deviation distribution map includes: Based on the construction semantic correspondence, the spatial location, size boundary, component connection relationship, construction procedure status and acceptance index of the corresponding component are read from the site semantic shadow body and BIM construction quality benchmark field to generate component difference calculation pairs. Map the spatial location of the on-site component and the spatial location of the BIM component in the component difference calculation pair to the same floor coordinate range, and perform boundary coincidence calculation on the component geometric dimensions and dimension boundaries to generate a geometric difference image; The differences in connection objects, connection directions and contact positions of component connections are calculated, and the results of the connection difference calculations are written into a geometric difference film to generate a relationship difference film. The differences between the on-site construction process status and the construction process are calculated, the differences between the on-site measured indicators and the acceptance indicators are calculated, and the corresponding calculation results are written into the relationship difference film to generate the quality difference film. Based on the quality difference film, the differences in spatial location, size boundary, component connection relationship, construction procedure, and acceptance index are converted into quality deviation elements. The quality deviation elements are then bound to the component spatial location, connection relationship information, construction procedure information, and component boundary range to generate quality deviation association information. Based on the spatial location of components in the quality deviation association information, aggregate the component location of quality deviation elements within the same floor coordinate range to generate a component location deviation set. According to the quality deviation type, the component position deviation set is merged by quality type to generate quality type merging information. The quality deviation elements, quality deviation association information, component spatial location and quality type merging information are distributed and bound to generate a construction quality deviation distribution map.

8. The method for BIM data processing and construction quality assessment based on artificial intelligence according to claim 1, characterized in that, The generation of the construction quality impact chain includes: Read the quality deviation elements, quality deviation association information, component spatial location and quality type merging information from the construction quality deviation distribution map, and group the data corresponding to the same component spatial location into the same deviation record to generate a set of deviation trigger points; Using the set of deviation trigger points as the starting point for transmission, the deviation of the component connection relationship from the corresponding connection object, connection direction and contact position is written into the corresponding deviation record to generate the connection transmission base map; The deviation records in the connection transmission base map are associated with construction procedures. The construction procedures and procedures corresponding to the deviations are written into the corresponding deviation records to generate the process transmission base map. Spatial range association is performed on the deviation records in the process transmission base map. The floor position, component spatial position and component boundary range corresponding to the spatial position deviation and dimension boundary deviation are written into the corresponding deviation record to generate the spatial transmission base map. Merge transmission relationships based on the same deviation record to generate quality impact transmission relationships; The set of deviation trigger points is used to locate the initial deviation, generate the initial deviation component, and merge the affected components and affected construction procedures in the quality impact transmission relationship to generate the affected object. Then, the transmission path is spliced ​​according to the component connection relationship, construction procedure transmission relationship and spatial position transmission relationship to generate the construction quality impact chain.

9. The method for BIM data processing and construction quality assessment based on artificial intelligence according to claim 1, characterized in that, The generation of the construction quality assessment results includes: Read the initial deviation component, quality deviation element, quality type merging information, affected objects, and transmission path in the construction quality impact chain; The initial deviation component is identified as an abnormal component, and a quality deviation type is generated according to the quality type merging information; Based on the correspondence between abnormal components and quality deviation types, generate sources of quality risk; Merge the affected components among the affected objects to generate risk-affected components; The transmission paths are merged to generate risk transmission paths; The risk levels of quality deviation types, sources of quality risks, components affecting risks, and risk transmission paths are merged to generate quality risk levels. Rectification priorities are generated based on quality risk level, affected components, and risk transmission path. Abnormal components, quality deviation type, source of quality risk, affected components, risk transmission path, quality risk level, and rectification priority are then combined into the construction quality assessment results.

Images