CN-122020773-A - MIC prefabricated part mold turning method based on BIM and CAD fusion

CN122020773ACN 122020773 ACN122020773 ACN 122020773ACN-122020773-A

Abstract

The invention relates to the technical field of building information models and digital construction, and discloses a MIC prefabricated part turnover method based on BIM and CAD fusion, which comprises the following steps of firstly carrying out geometric cleaning and semantic feature extraction on two-dimensional CAD data of a MIC part, and determining semantic types and initial geometric parameters of the graphic elements; according to the semantic type and the initial geometric parameters, calculating connection probability among the components, constructing a MIC component space topology association map describing the internal logic relationship of the MIC component, and calculating visual saliency characteristic points of the components as invisible anchor points. By constructing multidimensional geometric feature vectors and carrying out semantic judgment by using a weighted scoring function, the method gets rid of the strong dependence of the traditional model-turning technology on CAD layer names, deeply analyzes the parallelism, closure and relative inclusion relationship of the primitives, the system can accurately distinguish structural parts from electromechanical pipelines according to the geometric essence of the graph, and the robustness and accuracy of the semantic recognition of the MIC structural parts are remarkably improved.

Inventors

CHENG YONGWEI
SHI FENGMING
SHI YE
Shi Jiaopeng

Assignees

广东蓝杰科技有限公司

Dates

Publication Date: 20260512
Application Date: 20251222

Claims (10)

1. The MIC prefabricated part turnover method based on BIM and CAD fusion is characterized by comprising the following steps of: S1, performing geometric cleaning and semantic feature extraction on two-dimensional CAD data of an MIC component, and determining semantic types and initial geometric parameters of the primitives; S2, calculating connection probability among components according to the semantic type and the initial geometric parameters, and constructing a MIC component space topology association map for describing the internal logic relationship of the MIC component; S3, calculating visual saliency characteristic points of a component to serve as an invisible anchor point, and packaging data of the invisible anchor point and the MIC component space topology association map into a parameterized BIM model to generate a BIM model fused with topology semantics; s4, in the field application stage, utilizing an invisible anchor point in the BIM model to carry out space positioning, and carrying out augmented reality AR dynamic rendering on the BIM model according to the MIC component space topology association map; S5, acquiring field actual measurement point cloud data, calculating geometric deviation between the field actual measurement point cloud data and the BIM model, reversely correcting key parameters of a semantic feature extraction stage according to the geometric deviation, and regenerating the BIM model by utilizing the corrected key parameters.
2. The method for turning over MIC prefabricated parts based on BIM and CAD fusion according to claim 1, wherein the geometric cleaning of the two-dimensional CAD data of the MIC parts comprises: traversing all geometric primitives in the two-dimensional CAD data, and setting a noise threshold value and a closure judgment threshold value; If the geometric length of the geometric primitive is smaller than the noise threshold value, the geometric primitive is regarded as noise to be removed; And if the Euclidean distance between the two geometric primitive endpoints is smaller than the closing judgment threshold value, vertex merging operation is carried out, and a closed geometric path is constructed.
3. The MIC prefabricated component turnover method based on BIM and CAD fusion according to claim 1, wherein determining the semantic type of the primitive comprises: constructing a multidimensional geometrical feature vector of the graphic primitive, wherein the components of the multidimensional geometrical feature vector at least comprise parallelism, closeness, length-width ratio and relative inclusion relation of the graphic primitive; Calculating a semantic score of the primitive by using a weighted scoring function, wherein the semantic score is the sum of products of normalized numerical values of the characteristic components and corresponding weight coefficients; and presetting a classification threshold set, comparing the semantic score with the classification threshold set, and judging that the semantic type of the primitive belongs to a structural member, an electromechanical pipeline or other members.
4. The method for turning over MIC prefabricated parts based on BIM and CAD fusion according to claim 1, wherein the constructing a spatial topological association map of MIC parts describing internal logic relations of the MIC parts comprises: selecting a node identified as a structural member and a node of an electromechanical pipeline, and calculating the connection probability of the structural member and the electromechanical pipeline in logical association, wherein the connection probability is jointly determined by the minimum Euclidean distance of a three-dimensional bounding box and the overlapping state of two-dimensional projection; setting a topology threshold, and if the calculated connection probability is larger than the topology threshold, establishing a connection edge in the map to form the MIC component space topology association map.
5. The MIC prefabricated component turnover method based on BIM and CAD fusion according to claim 1, wherein the calculating the visual saliency feature point of the component as the invisible anchor point includes: extracting geometric corner points as candidate points aiming at the identified rigid structure chart elements; calculating visual saliency scores of the candidate points, wherein the visual saliency scores are jointly determined by local curvature modulus, line density gradient and construction shielding factors, and the construction shielding factors are preset based on the installation procedure of the MIC component; And selecting a plurality of points with the visual saliency scores ranked at the front as the invisible anchor points, and recording absolute coordinates and visual descriptors of the invisible anchor points.
6. The method for turning over MIC prefabricated parts based on BIM and CAD fusion according to claim 1, wherein the encapsulating the data of the hidden anchor points and the MIC part spatial topology correlation map into a parameterized BIM model includes: invoking a preset MIC parameterized family library according to the semantic type, and mapping the extracted initial geometric parameters to family parameters to generate a three-dimensional entity; And writing the generated topology ID of the MIC component space topology association map and the calculated visual descriptor of the invisible anchor point into an extended attribute field of a corresponding BIM component to realize data ID association.
7. The method for turning over MIC prefabricated parts based on BIM and CAD fusion according to claim 1, wherein the spatially locating with the hidden anchor points in the BIM model includes: Collecting a field image through AR equipment, and extracting field feature points; matching the field feature points with visual descriptors of the pre-implanted invisible anchor points in the BIM model; And establishing a transformation matrix of the world coordinate system and the model coordinate system based on the matching result.
8. The MIC prefabricated component turning method based on BIM and CAD fusion according to claim 1, wherein the performing augmented reality AR dynamic rendering on the BIM model according to the MIC component space topology association map includes: when the current component node is selected, traversing the rest component nodes in the MIC component space topology association map; calculating the shortest path hop count from the rest component nodes to the current component node; Setting a rendering level depth threshold, rendering only the components with the shortest path hops less than or equal to the rendering level depth threshold, and hiding the components without logical association.
9. The MIC prefabricated component turnover method based on BIM and CAD fusion according to claim 1, wherein the calculating the geometric deviation of the on-site real-time point cloud data and the BIM model includes: acquiring local sparse point cloud on site as the cloud data of the real-time points on site, and calculating the cloud data of the real-time points on site by using an iterative nearest point algorithm to register with an optimal rigid body transformation matrix of a BIM grid model; and calculating an average deviation vector of the key feature surface, and judging that the geometric deviation exists if the modular length of the average deviation vector exceeds a preset tolerance.
10. The MIC prefabricated component turning method based on BIM and CAD fusion according to claim 9, wherein the reversely correcting the key parameters of the semantic feature extraction stage according to the geometric deviation includes: determining parameters affecting the geometry of the component as said key parameters; Calculating the component of the average deviation vector in the normal vector direction of the component surface; And updating the key parameters by utilizing the components and a preset learning rate, and feeding the updated key parameters back to the semantic feature extraction step to carry out semantic judgment again.

Description

MIC prefabricated part mold turning method based on BIM and CAD fusion Technical Field The invention relates to the technical field of building information models and digital construction, in particular to a MIC prefabricated part turnover method based on BIM and CAD fusion. Background The modular integrated architecture (MIC) is a highly integrated industrial form of architecture, and requires high-precision prefabricated assembly of structures, decorations and electromechanical pipelines in factories, which makes the conversion precision and efficiency from two-dimensional design drawings to three-dimensional information models (BIM) a key link for determining success and failure of projects, and although the automatic mold turning technology based on CAD drawings is currently applied in the field of general architecture, the prior art has significant limitations when facing the special scene of MIC with high density and high integration. In the existing CAD model turning technology, the mainstream scheme is mostly dependent on layer naming or simple geometric rules to carry out primitive identification and three-dimensional conversion, the mode is excessively dependent on the rigor of an upstream design end drawing specification, once a CAD drawing is provided with layer confusion, linear non-specification or non-standard primitives, the model turning software cannot accurately identify component semantics, so that a large number of erroneous geometric entities exist in a generated BIM model, and more importantly, the existing model turning technology usually regards each component as an independent geometric object to carry out discretization treatment, and depth analysis on physical connection logic and space topological relation among the components is lacking, so that the generated model is just stacking of geometric shapes and cannot reflect logic association between complex electromechanical pipelines and structural frameworks in a MIC module, and the application value of the model in subsequent deepening design and construction guidance is directly limited. In addition, in the application link of virtual-real comparison and aesthetic drawing of a construction site, the existing technical flow is usually in a splitting state, the generation of a BIM model and the positioning data applied by site Augmented Reality (AR) are often disjointed, the site AR positioning mainly depends on two-dimension code marks or GPS signals, but under the condition of complex electromagnetic environment and vision shielding in an MIC module, the traditional positioning means are extremely easy to fail or have insufficient precision, meanwhile, due to lack of preprocessing of a model logic relationship, mobile terminal equipment often has to render full data when loading a high-precision MIC model, dynamic local loading is difficult to carry out according to the connecting logic of a component, rendering is serious, perspective test receiving requirements of hidden engineering cannot be met, in addition, the current model turning flow presents unidirectional data flow characteristics, model deviation found by site actual measurement is usually only manually corrected, a closed-loop mechanism for feeding back the site data to a front-end model turning algorithm is lacking, so that the model turning algorithm cannot automatically optimize identification parameters through learning history deviation, and the problem of inaccuracy of identification of a nonstandard component is difficult to fundamentally solve. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a MIC prefabricated component turning method based on BIM and CAD fusion, which solves the problems that the existing MIC component turning technology is seriously dependent on CAD layer specifications, so that the semantic recognition rate is low, the generated model lacks internal logic topological association, the model data cannot support the on-site non-marking point AR to be positioned and dynamically rendered with high precision, and the on-site actual measurement data-based reverse self-correction mechanism of turning parameters is lacking. The invention aims at realizing the technical scheme that the MIC prefabricated part turnover method based on BIM and CAD fusion comprises the following steps: S1, performing geometric cleaning and semantic feature extraction on two-dimensional CAD data of an MIC component, and determining semantic types and initial geometric parameters of the primitives; S2, calculating connection probability among components according to the semantic type and the initial geometric parameters, and constructing a MIC component space topology association map for describing the internal logic relationship of the MIC component; S3, calculating visual saliency characteristic points of a component to serve as an invisible anchor point, and packaging data of the invisible anchor point and the MIC component spa