Adaptive merging method and system based on multiple single BIM models

Abstract

The embodiment of the invention discloses a self-adaptive merging method and system based on multiple single BIM models, and the method comprises the steps: carrying out the projection of the single BIM models from three dimensions to two dimensions, obtaining a two-dimensional model diagram on a two-dimensional plane, and extracting feature intersection points and feature edges from the two-dimensional model diagram; on the basis, combining vertexes and edges in the single BIM models and feature intersection points and feature edges corresponding to the two-dimensional model graph to form a model feature vector matrix, calculating model feature values, and obtaining a model feature value set of the multiple single BIM models; and carrying out clustering analysis on the model characteristic value set to obtain a fitted and combined target model, and carrying out combination to realize light weight of the model. And moreover, the single BIM models are automatically processed in the whole process, manual operation of designers is not needed, and the model merging efficiency is improved.

Description

Technical Field

[0001] The present application relates to the field of BIM technology, and in particular to a method and system for adaptive merging based on multiple monomer BIM models. Background Art

[0002] Building Information Modeling (BIM), a core technology in the field of digital construction, integrates 3D geometric data with non-geometric attribute information, establishing a digital foundation for building management throughout its entire lifecycle. Currently, BIM model scenarios for large-scale construction projects contain over a million components, and the file size of a single BIM model can reach gigabytes. For example, a single BIM model of a high-rise building or a subway station, comprising floor / wall / column structures, electromechanical equipment, and finishing materials, generates billions of data nodes, encompassing multi-dimensional information such as geometric coordinates and material properties, resulting in a massive amount of model data.

[0003] Therefore, lightweighting single-unit BIM models is a key technology in this field. Traditional model lightweighting solutions mainly rely on geometric simplification, data compression, and deletion of non-essential attributes. Although these methods reduce data volume, they often sacrifice model accuracy and analytical capabilities, making it difficult to meet user application requirements for single-unit BIM models.

[0004] Therefore, how to achieve efficient model merging while retaining key model information and realize the lightweighting of multiple single BIM models has become one of the key technical bottlenecks in the BIM field. Summary of the Invention

[0005] In order to solve the technical problem of lightweighting the above-mentioned single BIM model, the embodiments of the present application provide an adaptive merging method and system based on multiple single BIM models.

[0006] According to one aspect of an embodiment of the present application, a method for adaptively merging multiple BIM models is disclosed, the method comprising: Step 1: Set the spatial coordinate system of the BIM model, import the monomer BIM model, and project the monomer BIM model from three dimensions to two dimensions onto the two-dimensional plane corresponding to the spatial coordinate system to obtain a two-dimensional model diagram; Step 2: Obtain characteristic intersection points in the two-dimensional model image, and characteristic edges connected to the characteristic intersection points; the characteristic intersection points are intersection points formed by projecting the spatial line structure of the single BIM model onto the two-dimensional plane; Step 3: Based on the vertices and edges in the monomer BIM model and the characteristic intersections and characteristic edges corresponding to the two-dimensional model graph, a model feature vector matrix is ​​generated according to the order of the two-dimensional model graph obtained by projecting the monomer BIM model; Step 4: Calculate the model eigenvalues ​​according to the model eigenvector matrix; Step 5: Perform cluster analysis based on the model eigenvalues ​​to obtain a fitted target model, and merge the target models.

[0007] According to one aspect of an embodiment of the present application, an adaptive merging system based on multiple single BIM models is disclosed, comprising: A two-dimensional model drawing acquisition module sets the spatial coordinate system of the BIM model, imports a single BIM model, and projects the single BIM model from three dimensions to two dimensions onto a two-dimensional plane corresponding to the spatial coordinate system to obtain a two-dimensional model drawing; A feature acquisition module is configured to acquire feature intersections in the two-dimensional model image and feature edges connected to the feature intersections; the feature intersections are intersections formed by projecting the spatial line structure of the monomer BIM model onto the two-dimensional plane; A model feature vector matrix generation module generates a model feature vector matrix based on the vertices and edges in the monomer BIM model and the feature intersections and feature edges corresponding to the two-dimensional model graph; and according to the order of the two-dimensional model graph obtained by projecting the monomer BIM model; Model eigenvalue calculation module: calculates the model eigenvalue according to the model eigenvector matrix; Model merging module: performing cluster analysis on the model feature values ​​to obtain fitted target models, and merging the target models.

[0008] Using the aforementioned adaptive merging method and system for multiple individual BIM models, dimensionality reduction processing is performed on the individual BIM models to obtain a model eigenvector matrix consisting of vertices, edges, characteristic intersections, and characteristic edges, and the model eigenvalues ​​are calculated. Based on the set of model eigenvalues ​​for the multiple individual BIM models, cluster analysis is performed to obtain a fitted target model, which is then merged. This solution achieves efficient merging of individual BIM models without deleting or simplifying any key information, such as structural and attribute data, achieving the goal of lightweighting the models.

[0009] Furthermore, the spatial structure of the BIM model is abstracted and represented using a 3×4 model feature vector matrix consisting of three dimensional feature vectors. This eliminates the need for complex mesh-level calculations and improves processing efficiency. Furthermore, this model feature vector matrix requires only tens of bytes, making it much faster and more efficient to merge separate BIM models than the original model data, which takes up megabytes or gigabytes. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] The accompanying drawings are incorporated into and constitute a part of the specification, illustrating embodiments consistent with the present application and, together with the specification, serving to explain the principles of the present application. It is obvious that the drawings described below are merely some embodiments of the present application, and a person of ordinary skill in the art can derive other drawings based on these drawings without inventive effort. In the drawings: Figure 1 The figure is a flowchart of a method for adaptively merging multiple BIM models according to an exemplary embodiment.

[0011] Figure 2 It is a three-dimensional space schematic diagram of a library BIM model scene shown as an exemplary embodiment.

[0012] Figure 3 FIG. 1 is a projection of a chair on an XY two-dimensional plane shown in an exemplary embodiment.

[0013] Figure 4 yes Figure 3 The two-dimensional model diagram of the chair projected in one direction and its characteristic intersections and edges.

[0014] Figure 5 yes Figure 3 The two-dimensional model diagram of the chair projected from above in one direction, as well as its characteristic intersections and edges.

[0015] Figure 6 yes Figure 3 The two-dimensional model diagram of the chair projected from the side in one direction, as well as its characteristic intersections and edges.

[0016] Figure 7 It is a partial stereoscopic diagram of a library BIM model scene shown in an exemplary embodiment.

[0017] Figure 8 yes Figure 7 Schematic diagram of the BIM model structure of the chair and table projected in the XZ plane direction.

[0018] Figure 9 The figure is a schematic diagram showing a source file of a three-dimensional model of a library according to an exemplary embodiment.

[0019] Figure 10 The figure is a comparison diagram of an exemplary embodiment showing a three-dimensional model of a library after model merging.

[0020] Figure 11 The figure is a flowchart of a method for processing row weights corresponding to a model feature vector matrix, which is shown as an exemplary embodiment. DETAILED DESCRIPTION

[0021] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. When the following description refers to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all embodiments applicable to the present application. Rather, they are merely examples of apparatus and methods applicable to certain aspects of the present application, as detailed in the appended claims.

[0022] The flowcharts shown in the accompanying drawings are for illustrative purposes only and do not necessarily include all contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps may be decomposed, while others may be integrated or partially integrated. Therefore, the actual execution order may vary depending on the actual situation.

[0023] It should be noted that the term "plurality" used in this application refers to two or more. "And / or" describes the relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A alone, A and B together, or B alone. The character " / " generally indicates that the associated objects are in an "or" relationship.

[0024] It should be noted that in the specific implementation of this application, when user-related data is involved, when the embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use, and processing of relevant data must comply with the relevant laws, regulations, and standards of the relevant countries and regions. At the same time, the formulas involved in the embodiments of this application can be flexibly adjusted, such as adding or reducing corresponding parameters.

[0025] Building Information Modeling (BIM), one of the core technologies for spatial digitization, digitally transforms models of the physical world to construct numerous individual BIM models in formats such as IFC, Revit, OBJ, STL, or proprietary formats. These individual BIM models are a combination of 3D geometric data and non-geometric attribute information.

[0026] Three-dimensional geometric data consists of points, lines, surfaces, and volumes. Points represent location information, lines represent paths, boundaries, or connections, surfaces represent the surface or extent of an object, and volumes describe its volume and internal structure. Furthermore, topological relationships describe spatial relationships between geometric objects, such as connectivity, adjacency, and containment. For example, whether two lines intersect or a point is inside a polygon. Dimensions and constraints describe the length, width, and height of an object, as well as the dimensional constraints between geometric elements.

[0027] Non-geometric attribute information describes the characteristics, properties, status, classification, etc. of the object, such as ID / code, name, type, status, etc.

[0028] When faced with multiple individual BIM models, if the 3D geometric data of each individual BIM model is stored separately, the amount of data that needs to be stored is extremely large. In certain specific scenarios, such as different versions of BIM scenarios, different engineers may draw the same model with varying degrees of difference, resulting in "duplicate" storage of essentially the same BIM model, making the BIM scenario data enormous. These BIM models require merging. In other scenarios, for example, some BIM models may have partial differences, but they represent the same type of geometric model, with generally consistent structural data, differing only in brand or color. In scenarios where "precision" is not a high requirement, such as BI visualization screens, these geometric model objects can also be merged.

[0029] Based on this, combined Figure 1 This is a flowchart of the adaptive merging method for multiple single BIM models of this solution, which includes: Step 1: Set the spatial coordinate system of the BIM model, import the monomer BIM model, and project the monomer BIM model from three dimensions to two dimensions onto the two-dimensional plane corresponding to the spatial coordinate system to obtain a two-dimensional model diagram.

[0030] Specific, combined Figure 2 This is a three-dimensional schematic diagram of a library's BIM model scene. Within the BIM model scene, a large number of various BIM models are set up based on a spatial coordinate system. In a BIM system, the construction and management of models within the BIM model scene typically rely on a spatial coordinate system. Whether importing an existing BIM model or creating or editing a new BIM model on a modeling platform, it is based on a unified three-dimensional spatial coordinate system as a reference framework. The BIM model scene in this solution uses a polar coordinate system. Of course, different coordinate systems can be used based on different technical needs, such as cylindrical coordinate systems, spherical coordinate systems, etc.

[0031] Combine Figure 3 , which is the projection of a chair on the XY two-dimensional plane in one embodiment. The spatial geometric BIM model in this three-dimensional spatial coordinate system contains information related to the model, such as geometry, attributes and other information, and simulates the monomer BIM model of the entity in the physical world through digital simulation. Based on this, the position, orientation and size of the monomer BIM model are defined in the three-dimensional spatial coordinate system, and the monomer BIM model is projected onto a specific two-dimensional plane (such as projection on the XY, YZ, XZ plane) to obtain a two-dimensional model diagram of the monomer BIM model on the two-dimensional plane. The two-dimensional model diagram contains the geometric contour information of the monomer BIM model projected on the two-dimensional plane, such as the number of geometric points, lines, and surfaces and other two-dimensional plane feature information, thereby realizing the simplified expression of the monomer BIM model information.

[0032] Step 2: Obtain the characteristic intersection points in the two-dimensional model diagram and the characteristic edges connected to the characteristic intersection points; the characteristic intersection points are the intersection points formed by projecting the spatial line structure of the single BIM model onto the two-dimensional plane.

[0033] Specific, combined Figure 4-Figure 6 The figure shows a two-dimensional model diagram of a chair projected in three directions, as well as characteristic intersections and characteristic edges (due to the large number of characteristic points / edges, only some are identified for clarity and illustration). In a BIM model scenario, when a single BIM model's spatial line structures that do not intersect in the three-dimensional space are projected onto a two-dimensional plane, visual intersections are generated in the projected two-dimensional plane. These visual intersections are not real connection points of solid geometry, but rather focal points formed by the projection of non-intersecting lines in space onto a two-dimensional plane. They are introduced "image intersections," or characteristic intersections. Correspondingly, characteristic edges are connected to the characteristic intersections. Through the geometric edge relationships of the single BIM model in space, a new feature data set is mapped onto a two-dimensional plane to reflect the spatial geometric complexity of the single BIM model. This not only reduces the dimensionality and difficulty of the geometric space single BIM model, but also facilitates the acquisition and statistics of characteristic data, providing support for the subsequent merging of single BIM models.

[0034] Furthermore, the feature intersections include direct feature intersections formed by projecting vertices of a single BIM model into a two-dimensional model diagram, and projected feature intersections formed by projecting edges of a single BIM model into a two-dimensional model diagram.

[0035] Specifically, for the spatial geometric complexity of different monomer BIM models, the projection feature intersection is a projection feature intersection that directly reflects the edge of the monomer BIM model projected into the two-dimensional model diagram. This feature is used to reflect the geometric complexity of its space and is a high-value feature data. On the other hand, the direct feature intersection is a feature inherent to the monomer BIM model, such as vertices, edges and other features. In certain cases, the inherent features of the monomer BIM model are projected into the two-dimensional model diagram, and its vertices coincide with a certain point or an edge, making the direct feature intersection less valuable than the projection feature intersection. Correspondingly, the direct feature intersection has a weaker ability to reflect the spatial geometric complexity of the monomer BIM model.

[0036] Calculate the intersection ratio of the projection feature intersection point to the feature intersection point, and the inverse of the intersection ratio is the quantity weight of the projection feature intersection point.

[0037] Specifically, because projected feature intersections and direct feature intersections differ in their ability to express the spatial geometric complexity of a single BIM model, they are weighted to highlight their contribution to spatial geometric complexity. Based on this, the feature intersections are equal to the sum of the direct and projected feature intersections. The proportion of projected feature intersections to the total number of intersections is calculated, and the corresponding inverse is the weight assigned to the projected feature intersections, which is used to highlight the spatial geometric complexity of the single BIM model.

[0038] Update the number of feature intersections. The number of feature intersections is the product of the number weight and the projected feature intersection, plus the direct feature intersection.

[0039] Specifically, based on the above-mentioned direct feature intersections and the weighted projection feature intersections, the number of new feature intersections is obtained, and the dimensionality of the complex geometric space model is converted into a two-dimensional model diagram. The weights of the projection feature intersections are highlighted through differentiated weighting, which facilitates the differentiated statistics of feature data and provides support for the subsequent merging of single BIM models.

[0040] Step 3: Based on the vertices and edges in the monomer BIM model, as well as the characteristic intersections and characteristic edges corresponding to the two-dimensional model diagram; according to the order of the two-dimensional model diagram obtained by projecting the monomer BIM model, generate the model feature vector matrix.

[0041] Specifically, the monomer BIM model is projected from three directional perspectives (such as projecting on the XY, YZ, and XZ planes) to obtain the corresponding two-dimensional plane, and the features in each projected two-dimensional model diagram are extracted, which are vertices, edges, feature intersections, and feature edges in order. The data of vertices and edges can be directly extracted from the attribute information of the monomer BIM model, and the data of feature intersections and feature edges are obtained through the above step 2. It should be noted that the order composed of vertices, edges, feature intersections, and feature edges can be set independently. The order of the two-dimensional model diagrams obtained by projecting the monomer BIM model is kept consistent as a whole, that is, the model feature vector matrix obtained by each monomer BIM model is guaranteed to be consistent, thereby forming the "feature vector matrix" of the model for similarity analysis.

[0042] Using the "Eigenvector Matrix" application example, a single BIM model is projected along the XY, YZ, and XZ planes to obtain corresponding vertices, edges, feature intersections, and feature edges. For example, the following are obtained: XY plane projection: vertices V1 = 18, edges E1 = 26, feature intersections I1 = 32, and feature edges S1 = 43; YZ plane projection: vertices V2 = 12, edges E2 = 18, feature intersections I2 = 39, and feature edges S2 = 51; XZ plane projection: vertices V3 = 20, edges E3 = 28, feature intersections I3 = 29, and feature edges S3 = 38. The following eigenvector data is obtained: Projection plane Number of vertices (V) Number of edges (E) Number of characteristic intersections (I) Number of characteristic edges (S) XY (1) 18 26 32 43 YZ (2) 12 18 39 51 XZ (3) 20 28 29 38 Based on the obtained feature data, it is organized into a 3×4 model feature vector matrix, namely:

[0043] Based on this expression, the eigenvectors of the three plane directions XY, YZ, and XZ are used to form the model eigenvector matrix. It can be seen that the eigenvectors are the characteristic data of points and edges. In other words, the expression of the model eigenvector matrix for a single BIM model is not limited to the size of the model, but focuses on the spatial structure of the model.

[0044] In one embodiment, if the spatial structures of different monomer BIM models are consistent, but the sizes are inconsistent, that is, the spatial proportions are different, then the edge of the monomer BIM model can be selected and the length of the edge can be recorded. At the same time, the length of the edge of the fitted target model can be obtained. By comparing the lengths of the two to obtain the scaling ratio, the size of the original model in space can be restored during the model restoration process.

[0045] Step 4: Calculate the model eigenvalues ​​based on the model eigenvector matrix.

[0046] The obtained model eigenvector matrix—the number of vertices, edges, characteristic intersections, and characteristic edges obtained by projecting them onto the XY, YZ, and XZ planes—constructs corresponding eigenvectors. This eigenvector then forms a corresponding matrix. This matrix is ​​mathematically transformed and the corresponding numerical value calculated is the "model eigenvalue," used to characterize a single BIM model. This abstraction transforms complex 3D models into measurable structural "digital signatures" for structural analysis and merging, serving as a lightweight BIM model.

[0047] Furthermore, the present invention numerically calculates the eigenvalues ​​of the model by the eigenvector matrix, specifically including: Step 41: The model feature vector matrix is ​​an M×N matrix, where M represents the three faces formed by the spatial coordinate system, and N represents the vertices, edges, feature intersections, and feature edges, with a total of 4 columns. Step 42: Convert the model feature vector matrix into a 1×(M×N) matrix; Step 43: Calculate the value of the 1×(M×N) matrix as the model eigenvalue.

[0048] The following is an explanation based on a specific application example: The M×N (i.e. 3×4) model feature vector matrix f=

[0049] Convert to a 1×12 matrix F=[18, 26, 32, 43, 12, 18, 39, 51, 20, 28, 29, 38] By calculating the Euclidean norm of the matrix (i.e. the modulus or length of the vector), which is the model eigenvalue of the matrix: ‖F‖= ≈109.23 (save two decimal places), this value is used to characterize the single BIM model.

[0050] Step 5: Perform cluster analysis based on the model eigenvalues ​​to obtain the fitted target model and merge the target models.

[0051] Specifically, the model eigenvalues ​​of the imported individual BIM models are calculated. That is, the individual BIM models of all targets in the BIM model space are calculated according to the above steps, and the corresponding model eigenvalues ​​are obtained, thereby forming a set of model eigenvalues. Cluster analysis is performed on the set of model eigenvalues, such as k-means, k-means++, bi-kmeans, kernelk-means, and other clustering analysis methods. This can be used to obtain the target models to be merged. Merging the target models can reduce the number of models and achieve lightweighting of multiple individual BIM models.

[0052] The adaptive merging method of the above-mentioned multiple monomer BIM models is adopted. By projecting the monomer BIM model from three dimensions to two dimensions, a two-dimensional model diagram on a two-dimensional plane is obtained, and characteristic intersections are extracted from the two-dimensional model diagram. Based on this, the vertices and edges in the monomer BIM model, as well as the characteristic intersections and characteristic edges corresponding to the two-dimensional model diagram are combined to form a model characteristic vector matrix, and the model eigenvalues ​​are calculated, and a set of model eigenvalues ​​of multiple monomer BIM models is obtained. Cluster analysis is performed on the set of model eigenvalues ​​to obtain a fitted target model and merge it. It can be seen that this solution does not need to delete or simplify any key information such as structural data and attribute data, and can achieve efficient merging of monomer BIM models to achieve the goal of lightweighting the model.

[0053] On this basis, each individual BIM model is simplified by reducing its dimensionality from 3D to 2D, and the spatial structure of the BIM model is abstracted. This is represented using a 3×4 model eigenvector matrix consisting of only three dimensional eigenvectors, eliminating the need for complex mesh-level calculations and improving processing efficiency. Furthermore, this model eigenvector matrix requires only tens of bytes, making the merging of individual BIM models much faster and more efficient than the original model data in megabytes or gigabytes.

[0054] Each individual BIM model is represented by its eigenvalue (i.e., vector norm mapping), visualized as a "point on a number axis." The more complex the individual BIM model, the larger the eigenvalue obtained, and the further away from the origin its representation on the number axis is, facilitating subsequent cluster analysis of the model. This solution abstracts complex 3D models into measurable structural "digital signatures," which are used to classify and merge individual BIM models, facilitating lightweight processing of these models with high accuracy and efficiency.

[0055] The conversion of individual BIM models into vector distance representation is simple and intuitive. Furthermore, threshold ranges can be set to adjust redundancy, thereby increasing the scope for lightweighting individual BIM models. For example, models with high BIM model similarity can be merged, reducing not only the number of individual BIM models but also the number of model objects, making the BIM model space even more lightweight. This solution offers significantly faster and more efficient comparisons compared to the high-cost mesh comparison of the original model.

[0056] By adopting this method, the single BIM model can be processed in an automated manner throughout the entire process, without relying on manual operations by designers, thereby improving the efficiency of model merging.

[0057] Combined with specific cases and Figure 7-Figure 8 The technical logic of this solution is explained below. Figure 7 A partial stereogram of a BIM model scene. Figure 8 for Figure 7 A schematic diagram of the BIM model structure of a chair and table in a partial stereogram projected in the XZ plane. Using the solution of this application, individual BIM models in a BIM model scene, such as a chair or table, are projected onto a two-dimensional plane to obtain a two-dimensional model diagram. The corresponding feature intersections and feature edges are then extracted and combined with the vertices and edges of the individual BIM models to form a model feature vector matrix, and the model eigenvalues ​​are calculated. Based on this, the target model is obtained according to the model eigenvalues ​​and the models are merged to achieve lightweighting.

[0058] Further, combined Figure 9-10 The technical effects of this solution are explained. Figure 9 This is a schematic diagram of a library 3D model source file, in which a large number of single BIM models of chairs and tables are set in the 3D BIM model scene. Figure 10This diagram shows a comparison of a library's 3D model after merging multiple individual BIM models. By merging a large number of individual BIM models for chairs and tables, the number of individual BIM models was significantly reduced, and the BIM model scene size was approximately 55% of the original file size, with virtually no changes to the displayed BIM model scene. This demonstrates that this solution offers significant advantages for large 3D BIM scenes with multiple individual BIM models.

[0059] In one embodiment, the adaptive merging method based on multiple individual BIM models further includes: Step 11: Obtain the attribute information of the single BIM model; Step 12: Determine the type of the monomer BIM model according to the type information in the attribute information; Step 13: Project the same type of single BIM model from three dimensions to two dimensions onto the two-dimensional plane corresponding to the spatial coordinate system, and obtain a two-dimensional model diagram.

[0060] Specifically, for the imported monomer BIM model, its format includes IFC, Revit or custom structured model format, and contains three-dimensional geometric data and non-geometric attribute information. The type information is obtained from the attribute information of the monomer BIM model, such as extracting the model type information through the fields in the attribute information. The information will record the type of the model, such as tables, chairs, doors and windows, etc. The attribute information of the monomer BIM model is pre-classified, and on this basis, the monomer BIM model is further projected from three dimensions to two dimensions to obtain the corresponding two-dimensional model diagram, which provides support for subsequent model feature recognition. Therefore, querying the same or similar monomer BIM models in models of the same type further improves the efficiency of model recognition and merging.

[0061] In one embodiment, combining Figure 11 Flowchart of the adaptive merging method based on multiple monomer BIM models, step 3 also includes: Determine whether the minimum bounding box of a single BIM model intersects with the minimum bounding box of any BIM model; If so, obtain the intersection of the surfaces of the monomer BIM models and project them onto the intersection surface in the corresponding two-dimensional plane, and calculate the projected area of ​​the intersection surface on the corresponding two-dimensional plane; Calculate the ratio of the projected area of ​​the intersecting surface to the area of ​​the single BIM model in the corresponding two-dimensional plane. The inverse of the area ratio is the weight of the corresponding row of the model eigenvector matrix.

[0062] Specifically, for spatial models, there are a large number of single BIM models, some of which overlap with each other. The overlapping parts are also key "features", that is, the introduction of projection weights can enhance the expression ability of "spatial distribution semantics" and is used to identify whether models are mergeable.

[0063] Based on this, for overlapping BIM models, the overlapping portion is calculated using the AABB minimum bounding box method. The projected area of ​​the overlapping portion in a specific direction is calculated, and the ratio of this overlapping area to the projected area of ​​the current BIM model in that specific direction is calculated. The inverse of the area ratio is used as the weight of the eigenvector in that direction. The meaning of the BIM model space expressed by this is: the higher the overlap rate between overlapping BIM models, which can be understood as the "spatial coverage" of the BIM models in a certain direction, the lower the corresponding weight, and the more similar the two models are.

[0064] To further illustrate this solution, for a model A and a model B whose AABBs overlap in space, the weighting algorithm on the projection plane XY is as follows:

[0065] in, : The projection area of ​​the AABB overlap area of ​​model A and model B on the XY plane; : The projected area of ​​model A on the XY plane; : Weights for the rows of XY in the model eigenvector matrix.

[0066] Similarly, the weights of the corresponding rows of YZ and XZ in the model feature vector matrix can also be obtained.

[0067] In other embodiments, based on the above method steps, the method further includes: if not, the weight of the corresponding row of the model eigenvector matrix of the corresponding two-dimensional plane is 1.

[0068] Specifically, if there is no overlap between model A and model B, that is, no intersection, the expression models are arranged independently, and the weight of the corresponding row of their eigenvector matrix is ​​1.

[0069] In one embodiment, the cluster analysis of this solution uses an optimized K-means algorithm, including: Acquire attribute information based on the single BIM model, classify one or more dimensions in the attribute information, such as geometric dimensions, physical properties, functional properties, or management information, and determine the K value based on the maximum number of classifications; Based on the determined K value, K initial center point models are randomly selected; Calculate the distance between each BIM model and the K initial center point models, obtain the BIM model with the closest distance to the initial center point model, and cluster them into the same category; Calculate the average value of the same type of single BIM models, determine the new initial center point model based on the average value, and repeatedly calculate the distance of the new initial center point model until convergence to obtain the target center point model; The monomer BIM models of the same type are set as the target models for fitting and merging, and the target models are merged into the target center point model.

[0070] Specifically, by extracting the attribute information of the monomer BIM, such as geometric dimensions, physical performance, functional attributes or management information, the K value is determined according to the number of categories. Based on the K value, K center point models are randomly selected, and the distance between the monomer BIM model and the K initial center point models is calculated respectively. In this process, a redundant space can be set to cluster the monomer BIM models that are close to or within the redundant range into the same category. Then, by calculating the average value of the monomer BIM models of the same category, the initial center point model is re-determined, and the target center point model is obtained by convergence. Finally, the target model of the fitting is determined, and the target model is merged into the target center point model to realize the automatic merging of the models.

[0071] This solution vectorizes the attributes and performs K-Means clustering to identify similar monomer BIM models, and based on this, the monomer BIM models can be merged.

[0072] In other embodiments, if the target center point model corresponding to the K value does not converge, the K value is sequentially reduced by the inverse elbow method, including: Calculate the sum of squares of the errors between the single BIM model and the initial center point model; Determine the inflection point of the characteristic curve based on the sum of squared errors; When the inflection point of the characteristic curve is smooth or converges within the set range; The value corresponding to the inflection point of the characteristic curve is the K value.

[0073] Specifically, this solution innovatively combines the "sum of squared errors" (SSE) with the "elbow method." Using the "inverse" elbow method, the K value is gradually reduced, gradually converging, and finding the inflection point, rather than the traditional "elbow method" which starts from zero and increases the value to find the corresponding inflection point. The "SSE" refers to the distance between the individual BIM models and the initial center point model, which is the positional deviation obtained, that is, the sum of the squares of the three-dimensional coordinate deviations.

[0074] This algorithm determines the maximum K value based on the number of BIM model attribute categories. However, in reality, BIM model attribute categories, such as chairs with backrests, rolling chairs, and armchairs, are considered three categories. However, the "chair" referred to by these three categories includes all three types, meaning they are essentially the same category. This means that if there are errors in these attribute classifications, the selection of the K value will also be incorrect. To quickly find the appropriate K value, starting from the K values ​​set for all model categories and then searching downward until convergence is achieved, this improves computational efficiency and quickly obtains the appropriate K value.

[0075] In one embodiment, based on this solution, surface textures of individual BIM models are also considered. If the textures are identical or similar, the corresponding individual BIM models are also identical or similar and are also objects of merging. Based on this, the method further includes: Get the texture mapped by the single BIM model and the UV coordinates of the texture; Extract the vertex coordinates of the single BIM model and the target UV coordinates of the mapping corresponding to the vertex coordinates, and obtain the target pixel value corresponding to the target UV coordinates; Determine whether the target pixel values ​​between the individual BIM models are equal. If so, project the individual BIM models onto a two-dimensional plane and obtain the BIM model outline. Determine whether the outlines of the individual BIM models are the same. If so, merge the BIM models corresponding to the BIM model outlines.

[0076] Specifically, in the BIM model scene, there are a large number of monomer BIM models, such as doors, windows, walls, floors and other monomer BIM models. At this time, the monomer BIM model is a white model, that is, a BIM model with only line structure. In order to express the geometric model space in three-dimensional visualization, the monomer BIM model needs to be mapped to express the surface material and visual characteristics of the monomer BIM model. The loading of the texture into the monomer BIM model requires reading the image resources and the precise mapping relationship with the surface geometry of the monomer BIM model, that is, the UV coordinates define how the two-dimensional texture is unfolded on the surface of the three-dimensional monomer BIM model. Therefore, for the texture Figure 1 The structures or outlines of the monomer BIM models are consistent or highly similar. Extracting the textures mapped by the monomer BIM models and the corresponding UV coordinate information is the basic condition for classifying and subsequently merging the monomer BIM models.

[0077] Extract the geometric vertices in the monomer BIM model and the UV coordinates corresponding to the geometric vertices after the monomer BIM model is 2D unfolded and the texture. The UV coordinates of the texture point to the corresponding pixel values, that is, the target pixel values ​​extracted at the target position of the texture.

[0078] For multiple individual BIM models, the target pixel values ​​of their corresponding maps are obtained, and the target pixel values ​​between the BIM models are compared to determine whether the target pixel values ​​are equal. If so, the individual BIM model is projected onto a two-dimensional plane along a specific direction, and the outline of the individual BIM model is obtained. If not, the target pixel values ​​of other individual BIM models are compared until the comparison is complete. In this embodiment, the step of determining whether the target pixel values ​​are equal is set to a certain threshold. If the error ratio between the target pixel values ​​is within 1% or another ratio, it can be adaptively adjusted based on the accuracy of the individual BIM model merging, and the set of individual BIM models that are suitable for merging can be quickly determined.

[0079] To compare the contours of different BIM models, the point sets of the BIM model contours can be sequentially connected, and point-to-point comparison or distance comparison can be performed between the contours of the individual BIM models to determine whether the two individual BIM models are identical. This method includes scaled individual BIM models, and a certain error range can also be set. That is, if the hit rate or distance value of the point-to-point comparison falls within the preset range, the individual BIM models are considered to be the same.

[0080] In the BIM model scenario, the visualization deployment process of a single BIM model is usually large in size, and there are a large number of models with repeated textures but slight geometric differences. Based on this, the solution of this embodiment provides a "visual approximation" initial filtering based on the same target pixel values ​​at corresponding vertex positions of the textures, accurately delineating the fitted and merged single BIM models. Furthermore, the delineated single BIM models are projected onto a two-dimensional plane to obtain the outline of the single BIM model, and the single BIM models with the same geometric structure are determined, achieving a significant lightweight effect.

[0081] In one embodiment, based on the above-mentioned method for adaptively merging multiple individual BIM models, an adaptive merging system based on multiple individual BIM models is proposed, including: The 2D model acquisition module sets the spatial coordinate system of the BIM model, imports the monomer BIM model, and projects the monomer BIM model from 3D to 2D onto the 2D plane corresponding to the spatial coordinate system to obtain the 2D model diagram; The feature acquisition module obtains the feature intersections in the two-dimensional model diagram and the feature edges connected to the feature intersections; the feature intersections are the intersections formed by projecting the spatial line structure of the single BIM model onto the two-dimensional plane; The model feature vector matrix generation module generates the model feature vector matrix based on the vertices and edges in the monomer BIM model and the feature intersections and feature edges corresponding to the two-dimensional model diagram; and the order of the two-dimensional model diagram obtained by projecting the monomer BIM model; Model eigenvalue calculation module: calculates the model eigenvalue according to the model eigenvector matrix; Model merging module: Perform cluster analysis on the model eigenvalues ​​to obtain the fitted target model and merge the target models.

[0082] Based on the adaptive merging system of multiple monomer BIM models, a two-dimensional model diagram is obtained through the two-dimensional model diagram acquisition module; then, the feature acquisition module is used to extract the feature intersections and feature edges in the two-dimensional model diagram; then, combined with the model feature vector matrix generation module, a model feature vector matrix is ​​generated according to the vertices, edges, feature intersections, and feature edges in the monomer BIM model; then, the model eigenvalue calculation module is used to calculate the model eigenvalue according to the model eigenvector matrix; finally, the model merging module performs cluster analysis on the model eigenvalues ​​to obtain the fitted and merged target model, and merges the target models to achieve model lightweighting, and also achieve lightweighting of BIM model scenes containing multiple monomer BIM models.

[0083] The above content is only a preferred exemplary embodiment of the present application and is not intended to limit the implementation scheme of the present application. Ordinary technicians in this field can easily make corresponding changes or modifications based on the main ideas and spirit of the present application. Therefore, the scope of protection of the present application shall be based on the scope of protection required by the claims.

Claims

1. An adaptive merging method based on multiple monomer BIM models, characterized in that: include: Step 1: Set the spatial coordinate system of the BIM model, import the monomer BIM model, and project the monomer BIM model from three dimensions to two dimensions onto the two-dimensional plane corresponding to the spatial coordinate system to obtain a two-dimensional model diagram; Step 2: Obtain characteristic intersection points in the two-dimensional model image, and characteristic edges connected to the characteristic intersection points; the characteristic intersection points are intersection points formed by projecting the spatial line structure of the single BIM model onto the two-dimensional plane; Step 3: Based on the vertices and edges in the monomer BIM model and the characteristic intersections and characteristic edges corresponding to the two-dimensional model graph, a model feature vector matrix is ​​generated according to the order of the two-dimensional model graph obtained by projecting the monomer BIM model; Step 4: Calculate the model eigenvalues ​​according to the model eigenvector matrix; Step 5: Perform cluster analysis based on the model eigenvalues ​​to obtain a fitted target model, and merge the target models.

2. The adaptive merging method based on multiple monomer BIM models according to claim 1, characterized in that: The step 1 comprises: Step 11: Obtaining attribute information of the monomer BIM model; Step 12: Determine the type of the monomer BIM model according to the type information in the attribute information; Step 13: Project the monomer BIM model of the same type from three dimensions to two dimensions onto the two-dimensional plane corresponding to the spatial coordinate system, and obtain a two-dimensional model diagram.

3. The adaptive merging method based on multiple monomer BIM models according to claim 1, characterized in that: The characteristic intersections include direct characteristic intersections formed by projecting the vertices of the monomer BIM model onto the two-dimensional model graph, and projected characteristic intersections formed by projecting the edges of the monomer BIM model onto the two-dimensional model graph; Calculating an intersection ratio between the projection feature intersection and the feature intersection, where the reciprocal of the intersection ratio is the quantity weight of the projection feature intersection; The number of the feature intersections is updated, where the number of the feature intersections is the product of the number weight and the projected feature intersection, plus the direct feature intersection.

4. The adaptive merging method based on multiple monomer BIM models according to claim 1, characterized in that: The step 3 further comprises: Determine whether the minimum bounding box of the single BIM model intersects with the minimum bounding box of any BIM model; If so, obtain the intersection of the monomer BIM models and project it onto the corresponding intersection in the two-dimensional plane, and calculate the projected area of ​​the intersection on the corresponding two-dimensional plane; The ratio of the projected area of ​​the intersecting surface to the area of ​​the monomer BIM model on the corresponding two-dimensional plane is calculated, and the reciprocal of the area ratio is the weight of the corresponding row of the model eigenvector matrix.

5. The adaptive merging method based on multiple monomer BIM models according to claim 4 is characterized in that: If not, the weight of the corresponding row of the model eigenvector matrix for the corresponding two-dimensional plane is 1.

6. The adaptive merging method based on multiple monomer BIM models according to claim 1, characterized in that: The step 4 comprises: Step 41: The model feature vector matrix is ​​an M×N matrix, where M represents the three faces formed by the spatial coordinate system, and N represents the vertices, edges, feature intersections, and feature edges, with a total of 4 columns. Step 42: Convert the model feature vector matrix into a 1×(M×N) matrix; Step 43: Calculate the value of the 1×(M×N) matrix as the model eigenvalue.

7. The adaptive merging method based on multiple monomer BIM models according to claim 1 is characterized in that: The cluster analysis adopts the optimized K-means algorithm, including: Acquire attribute information based on the monomer BIM model, classify one or more dimensions in the attribute information, including geometric dimensions, physical properties, functional properties, or management information, and determine the maximum value of the number of classifications as the K value; Based on the determined K value, randomly select K initial center point models; Calculating the distances between the monomer BIM model and the K initial center point models respectively, obtaining the monomer BIM model closest to the initial center point model, and clustering them into the same category; Calculating an average value of the monomer BIM models of the same type, determining a new initial center point model based on the average value, and repeatedly calculating the distance of the new initial center point model until convergence to obtain a target center point model; The monomer BIM models of the same type are set as target models for fitting and merging, and the target models are merged into the target center point model.

8. The adaptive merging method based on multiple monomer BIM models according to claim 7, characterized in that: If the target center point model corresponding to the K value does not converge, the K value is sequentially reduced by the inverse elbow method, including: Calculating the sum of squares of errors between the monomer BIM model and the initial center point model; Determining the inflection point of the characteristic curve according to the sum of squared errors; When the inflection point of the characteristic curve is smooth or converges within a set range; The value corresponding to the inflection point of the characteristic curve is the K value.

9. The adaptive merging method based on multiple monomer BIM models according to claim 1, characterized in that: The two-dimensional plane is formed by any two mutually perpendicular coordinate axes in the spatial coordinate system.

10. An adaptive merging system based on multiple individual BIM models, characterized by: include: A two-dimensional model drawing acquisition module sets the spatial coordinate system of the BIM model, imports a single BIM model, and projects the single BIM model from three dimensions to two dimensions onto a two-dimensional plane corresponding to the spatial coordinate system to obtain a two-dimensional model drawing; A feature acquisition module is configured to acquire feature intersections in the two-dimensional model image and feature edges connected to the feature intersections; the feature intersections are intersections formed by projecting the spatial line structure of the monomer BIM model onto the two-dimensional plane; A model feature vector matrix generation module generates a model feature vector matrix based on the vertices and edges in the monomer BIM model and the feature intersections and feature edges corresponding to the two-dimensional model graph; and according to the order of the two-dimensional model graph obtained by projecting the monomer BIM model; Model eigenvalue calculation module: calculates the model eigenvalue according to the model eigenvector matrix; Model merging module: performing cluster analysis on the model feature values ​​to obtain fitted target models, and merging the target models.

Images