A method and apparatus for comparing BIM components

Abstract

This invention discloses a method and apparatus for comparing BIM components, belonging to the field of Building Information Modeling (BIM). The comparison method includes: converting two BIM components to be compared into BREP expressions, obtaining control points, and establishing a four-dimensional coordinate system; calculating the metric tensor and inertia tensor of the two BIM components based on the four-dimensional coordinates of the control points, constructing an undirected graph with faces as nodes, metric tensors and inertia tensors as node attributes, common edges in the BREP expressions as graph edges, and the angle between the normal vectors of adjacent faces as edge attributes; obtaining geometric similarity by matching the undirected graphs corresponding to the two BIM components; sorting the attribute lists of the two BIM components alphabetically and calculating attribute similarity based on semantic comparison; and calculating a comprehensive similarity based on geometric and attribute similarity to determine whether the two BIM components are similar. The comprehensive calculation of geometric and attribute similarity improves the accuracy of BIM component comparison results and uses fewer calculation points, thus improving efficiency.

Description

Technical Field

[0001] This invention belongs to the field of Building Information Modeling (BIM), and more specifically, relates to a method and apparatus for comparing BIM components. Background Technology

[0002] In the past decade or so, the emergence of the Building Information Modeling (BIM) concept in the construction industry has played a significant role in improving the informatization level of my country's construction industry and has assisted in the comprehensive optimization of building information management models. The comparison of BIM components is a relatively basic and practical tool.

[0003] Currently, the concept of full lifecycle management of Building Information Modeling (BIM) is gradually being promoted, emphasizing "one model throughout," that is, using a unified BIM model at all stages of a project. However, the export standards of Industry Foundation Classes (IFC) between different software are not yet fully compatible, which leads to obstacles in information exchange and model sharing. Furthermore, BIM models often contain a large number of repetitive components, and the reuse of geometric information largely depends on the software used and its export mechanism. Therefore, how to effectively retrieve and identify identical components has become a key challenge for geometric reuse. Solving this problem not only affects the model's storage efficiency but also directly relates to the depth and breadth of BIM application throughout the project lifecycle.

[0004] Existing technologies include some BIM component comparisons that obtain positional similarity by performing coordinate registration on the components; and by selecting random points on the triangular faces of the components and calculating their distances to the component's center to obtain a geometric distribution, thereby calculating similarity. However, such comparisons do not consider semantic changes in the components. Furthermore, while the method of calculating geometric distribution by selecting random points yields higher accuracy with more calculations, it is extremely time-consuming. Summary of the Invention

[0005] In view of the shortcomings of related technologies, the purpose of this invention is to provide a comparison method and apparatus for BIM components, which aims to solve the problems that existing comparison methods do not simultaneously consider geometric similarity and attribute similarity, and that the calculation points selected when calculating similarity are too numerous, resulting in low comparison efficiency and low comparison accuracy.

[0006] To achieve the above objectives, in a first aspect, the present invention provides a method for comparing BIM components, comprising:

[0007] S1. Convert the two BIM components to be compared into BREP expressions. Based on the geometry corresponding to the two BREP expressions, obtain the control points and establish a four-dimensional coordinate system for each control point.

[0008] S2. Calculate the metric tensor and inertia tensor of the two BIM components based on the four-dimensional coordinates of the control points, and construct an undirected graph with faces as nodes, metric tensors and inertia tensors as node attributes, common edges in the BREP expression as edges of the graph, and the angle between the normal vectors of adjacent faces as edge attributes.

[0009] S3. Obtain geometric similarity by matching the undirected graphs corresponding to two BIM components;

[0010] S4. Sort the attribute lists of the two BIM components alphabetically and calculate the attribute similarity based on semantic comparison.

[0011] S5. Calculate the comprehensive similarity based on the geometric similarity and attribute similarity. When the comprehensive similarity is greater than the preset similarity, determine that the two BIM components to be compared are similar.

[0012] Optionally, converting the two BIM components to be compared into BREP expressions includes:

[0013] If the BIM component is an extruded body, the points of the geometry are obtained by analyzing the extruded body, and the set of points is organized in order to generate lines and surfaces in the BREP representation.

[0014] If the BIM component is CSG, the shape of the geometry is generated by parsing the geometric primitives and Boolean operations in the CSG representation, and the geometry is converted into BREP representation.

[0015] Optionally, the step of obtaining control points based on the geometry corresponding to the two BREP expressions and establishing a four-dimensional coordinate system for each control point includes:

[0016] Control points are selected from the faces of the geometry corresponding to the BREP expression, and the weight coefficients of each control point are obtained through principal component analysis. The established four-dimensional coordinate system is as follows:

[0017] A = (X A ,Y A Z A ,ω A )

[0018] Among them, X A Y A Z A Let ω be the coordinate value of the control point in the natural coordinate system. A These are the weight parameters for the control points.

[0019] Optionally, the step of calculating the metric tensor and inertia tensor of the two BIM components based on the four-dimensional coordinates of the control points includes:

[0020] Calculate the vector axes of the edges formed by the control points based on their four-dimensional coordinates:

[0021]

[0022] in, Let i be the four-dimensional coordinates of the i-th control point. Let p be the vector axis from the i-th control point to the (i+1)-th control point, where 1 ≤ i ≤ p;

[0023] Arrange the vector axes of all edges in the geometry in order to obtain vector bundles for different faces, and then decompose them:

[0024]

[0025] in, V is a vector bundle over surface α. i x for exist Components in the direction, V i y for exist Components in the direction, V i z for exist Components in the direction, V i ω for exist Components in direction; These correspond to x, y, z, and weight parameters, respectively.

[0026] The metric tensor is calculated by multiplying the outer product of the vector bundles of the surface by itself. The calculation formula is as follows:

[0027]

[0028] in, Let α be the metric tensor of the surface α;

[0029] Calculate the centroid matrix of the two BIM components, simplify the eigenvalues ​​of the centroid matrix, and obtain the inertia tensor of all surfaces.

[0030] Optionally, S3 specifically includes:

[0031] S31. Determine the matching order in an undirected graph using a breadth-first search algorithm;

[0032] S32. Traverse the metric tensors of all faces of the two BIM components according to the matching order, and determine the two faces with equivalent metric tensors as similar faces.

[0033] S33. Traverse the inertia tensors between all faces of the two BIM components according to the matching order, and determine the two faces with equivalent inertia tensors as faces with the same size.

[0034] S34. Determine the set of identical faces by taking the intersection of the set of similar faces and the set of faces with the same size, and calculate the geometric similarity:

[0035]

[0036] Where, sim G N represents the geometric similarity of the components. Gsim N represents the number of identical faces. G This represents the number of faces in a BIM component.

[0037] Optionally, S4 specifically includes:

[0038] The attribute sets of two BIM components are sorted, and natural language processing is used to analyze the content of synonymous attributes. Synonymous attributes with the same content are identified as the same attributes, and the attribute similarity is calculated.

[0039] Formula for calculating attribute similarity:

[0040]

[0041] Where, sim A For the attribute similarity of components, N Asim N represents the number of items with the same attribute. A The number of component attributes.

[0042] Optionally, the formula for calculating the comprehensive similarity is:

[0043] sim=θ*sim G +(1-θ)*sim A

[0044] Where sim is the overall similarity of the components, sim G Let θ represent the geometric similarity of the components, and θ be the weighting coefficient of the geometric similarity.

[0045] In a second aspect, the present invention also provides a BIM component comparison apparatus for performing the comparison method as described in any one of the first aspects, comprising:

[0046] The coordinate system establishment module is used to convert the two BIM components to be compared into BREP expressions, obtain the control points based on the geometry corresponding to the two BREP expressions, and establish a four-dimensional coordinate system for each control point.

[0047] The metric tensor and inertia tensor calculation module is used to calculate the metric tensor and inertia tensor of two BIM components based on the four-dimensional coordinates of the control points. It constructs an undirected graph with faces as nodes, metric tensors and inertia tensors as node attributes, common edges in the BREP expression as edges of the graph, and the angle between the normal vectors of adjacent faces as edge attributes.

[0048] The geometric similarity calculation module is used to obtain the geometric similarity by matching the undirected graphs corresponding to two BIM components;

[0049] The attribute similarity calculation module is used to sort the attribute lists of two BIM components and calculate the attribute similarity based on semantic comparison.

[0050] The similarity judgment module is used to calculate the comprehensive similarity based on the geometric similarity and attribute similarity. When the comprehensive similarity is greater than the preset similarity, the two BIM components to be compared are determined to be similar.

[0051] Compared with the prior art, the above-described technical solutions conceived in this invention can achieve the following beneficial effects:

[0052] 1. This invention provides a method for comparing BIM components. Existing technologies randomly acquire multiple calculation points in the geometric entity for similarity calculation. This solution converts the geometry of two different BIM components into BREP (Body Shape Array), which can accommodate all geometric shapes, enabling parametric comparison of BIM components and making the comparison method universal. It determines the shape similarity of faces in the BREP expression by calculating the metric tensor. The metric tensor is a coordinate-independent matrix expression, allowing the comparison method to compare geometric similarity without relying on a coordinate system. It determines dimensional consistency by calculating the inertia tensor of two similar faces, providing a more accurate comparison of whether components are geometrically consistent. Since control points and coordinate values ​​are known in the BREP expression, fewer calculation points are used compared to existing technologies, improving comparison efficiency. By sorting the attribute list and comparing the attribute set content, attribute comparison is achieved. Semantic comparison is used to calculate attribute similarity, further improving the accuracy of the comparison. The comprehensive calculation of geometric and attribute similarity makes the comparison method for BIM components more comprehensive and the results more accurate.

[0053] 2. BREP representation requires a comparison order, and there are cases where two graphic surfaces are the same but their topological relationships are inconsistent. The present invention provides a comparison method for BIM components, which avoids this situation by constructing an undirected graph. Attached Figure Description

[0054] Figure 1 This is a flowchart illustrating a method for comparing BIM components provided by the present invention.

[0055] Figure 2 It is the calculation of edge attributes in an undirected graph;

[0056] Figure 3 It is a rotated copy of a single complex geometric model;

[0057] Figure 4 It is a model drawing of a residential building;

[0058] Figure 5 It is a model drawing of a standard floor of a residential building;

[0059] Figure 6 The images show the search results, where (a) shows the search results for standard floor windows, (b) shows the search results for standard floor beams, and (c) shows the search results for standard floor doors. Detailed Implementation

[0060] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention. Furthermore, the technical features involved in the various embodiments of this invention described below can be combined with each other as long as they do not conflict with each other.

[0061] The following description, in conjunction with a preferred embodiment, illustrates the content involved in the above embodiments.

[0062] Example 1

[0063] like Figure 1 As shown, the present invention provides a method for comparing BIM components, including:

[0064] S1. Convert the two BIM components to be compared into BREP expressions. Based on the geometry corresponding to the two BREP expressions, obtain the control points and establish a four-dimensional coordinate system for each control point.

[0065] S2. Calculate the metric tensor and inertia tensor of the two BIM components based on the four-dimensional coordinates of the control points, and construct an undirected graph with faces as nodes, metric tensors and inertia tensors as node attributes, common edges in the BREP expression as edges of the graph, and the angle between the normal vectors of adjacent faces as edge attributes.

[0066] S3. Obtain geometric similarity by matching the undirected graphs corresponding to two BIM components;

[0067] S4. Sort the attribute lists of the two BIM components alphabetically and calculate the attribute similarity based on semantic comparison.

[0068] S5. Calculate the comprehensive similarity based on the geometric similarity and attribute similarity. When the comprehensive similarity is greater than the preset similarity, determine that the two BIM components to be compared are similar.

[0069] The geometric expressions of two BIM components to be compared are converted into BREP format. Geometric similarity is calculated through BREP geometric comparison, and attribute similarity is calculated through semantic comparison. By comprehensively calculating the geometric and attribute similarity of the BIM components, the similarity of the BIM components can be accurately determined. Based on parametric comparison, the scope of BIM geometric comparison is expanded, making BIM model comparison faster and the results more accurate.

[0070] The three most commonly used geometric representations of BIM components include extruded bodies, CSG, and BREP. Among them, BREP can express almost all geometric representations. Therefore, specifically, converting the two BIM components to be compared into BREP expressions includes:

[0071] If a BIM component is an extruded body, its representation consists of the extruded bottom surface, the extrusion direction, and the extrusion length. Converting the extruded body to a BREP representation involves: analyzing the extruded body to obtain the set of extruded bottom surface points; calculating the set of extruded top surface points according to the extrusion direction and length; and then organizing the sets of bottom and top surface points sequentially to generate lines and surfaces in the BREP representation.

[0072] If the BIM component is a CSG (Content Generated Character Set), the CSG representation is relatively complex and is obtained by finding the intersection of several geometric shapes. The CSG is then converted to a BREP (Body-Represented Programming) representation: by parsing the geometric primitives and Boolean operations in the CSG representation, the shape of the geometric shape is generated, and then the geometric shape is converted to a BREP representation.

[0073] After the geometry is converted, it is stored in an IFC file as a BREP representation for subsequent use, avoiding multiple conversions.

[0074] Optionally, the step of obtaining control points based on the geometric shapes corresponding to the two BREP expressions and establishing a four-dimensional coordinate system for each control point includes:

[0075] Control points are selected from the surfaces of the geometry corresponding to the BREP expression, and the weight coefficients of each control point are obtained through principal component analysis. The established four-dimensional coordinate system is as follows:

[0076] A = (X A ,Y A Z A ,ω A )

[0077] Among them, X A Y A ZA Let ω be the coordinate value of the control point in the natural coordinate system. A These are the weight parameters for the control points.

[0078] In BREP representation, components are composed of points arranged in sequence to form surfaces. Taking a surface as an example: select a surface in BREP representation, obtain control points from it, and establish a four-dimensional coordinate system. BREP representation includes the natural coordinates of the control points.

[0079] Optionally, the step of calculating the metric tensor and inertia tensor of the two BIM components based on the four-dimensional coordinates of the control points includes:

[0080] Calculate the vector axes of the edges formed by the control points based on their four-dimensional coordinates:

[0081]

[0082] in, Let i be the four-dimensional coordinates of the i-th control point. Let p be the vector axis from the i-th control point to the (i+1)-th control point, where 1 ≤ i ≤ p;

[0083] Arrange the vector axes of all edges in the geometry in order to obtain vector bundles for different faces, and then decompose them:

[0084]

[0085] in, V is a vector bundle over surface α. i x for exist Components in the direction, V i y for exist Components in the direction, V i z for exist Components in the direction, V i ω for exist Components in direction; These correspond to x, y, z, and weight parameters, respectively, and are used to transform a vector group into a matrix. For example... All are unit vectors, corresponding to (1, 0, 0, 0), (0, 1, 0, 0), (0, 0, 1, 0), and (0, 0, 0, 1), respectively.

[0086] The metric tensor is calculated by multiplying the outer product of the vector bundles of the surface by itself. The calculation formula is as follows:

[0087]

[0088] in, Let α be the metric tensor of the surface α;

[0089] Calculate the centroid matrix of the two BIM components, simplify the eigenvalues ​​of the centroid matrix, and obtain the inertia tensor of all surfaces.

[0090] The formula for calculating the centroid matrix is ​​as follows:

[0091]

[0092] in, Let m be the barycenter matrix of surface α. i Let represent the mass distribution at the i-th control point. For a surface with uniform mass distribution, the value is 1.

[0093] The eigenvalues ​​of the simplified barycenter matrix are obtained as follows:

[0094]

[0095] in, Let α be the inertial tensor of surface α. The centroid matrix eigenvalues.

[0096] After calculating the metric tensor and inertia tensor, with faces as nodes and the metric and inertia tensors as node attributes, the common edges in the BREP expression are the edges of the graph, and the angle between the normal vectors of adjacent faces is the edge attribute (see reference). Figure 2 Construct an undirected graph.

[0097] Optionally, S3 specifically includes:

[0098] S31. Determine the matching order in an undirected graph using a breadth-first search algorithm;

[0099] S32. Traverse the metric tensors of all faces of the two BIM components according to the matching order, and determine the two faces with equivalent metric tensors as similar faces.

[0100] S33. Traverse the inertia tensors between all faces of the two BIM components according to the matching order, and determine the two faces with equivalent inertia tensors as faces with the same size.

[0101] According to the matching order, all faces of the two BIM components are traversed, and the metric tensors of the two faces are compared to see if they are equivalent. The two faces with equivalent metric tensors are identified as similar faces. The consistency of dimensions is determined by the inertia tensor. The inertia tensors between the two faces are compared, and the two faces corresponding to equivalent inertia tensors are identified as faces with the same dimensions.

[0102] S34. The intersection of the set of similar faces and the set of faces with the same size is determined as the set of similar faces.

[0103] When the metric tensor and inertia tensor of two surfaces are equivalent, it can be determined that the two corresponding surfaces are the same, that is, the intersection of similar surfaces and surfaces with the same size is taken.

[0104] Calculate geometric similarity:

[0105]

[0106] Where, sim G N represents the geometric similarity of the components. Gsim N represents the number of identical faces. G The number of component surfaces.

[0107] Optionally, S4 specifically includes:

[0108] The attribute sets of two BIM components are sorted, and natural language processing is used to analyze the content of synonymous attributes. Synonymous attributes with the same content are identified as the same attributes, and the attribute similarity is calculated.

[0109] Formula for calculating attribute similarity:

[0110]

[0111] Where, sim A For the attribute similarity of components, N Asim N represents the number of items with the same attribute. A The number of component attributes.

[0112] This solution utilizes natural language processing to analyze synonymous attributes and classify some synonymous attribute expressions as identical. When efficiency requirements are lower but accuracy requirements are higher, deep learning-based natural language processing can be used to identify synonymous semantics, improving accuracy while reducing efficiency. Furthermore, if users do not require high precision in comparing BIM components, a simplified attribute comparison method can be used. For example, after sorting the attribute set, Boolean judgments can be performed to classify identical attributes, and attribute similarity can be obtained through the proportion of identical attributes.

[0113] Optionally, the formula for calculating the comprehensive similarity is:

[0114] sim=θ*sim G +(1-θ)*sim A

[0115] Where sim is the overall similarity of the components, sim GLet θ represent the geometric similarity of the components, and let θ be the weighting coefficient of the geometric similarity. The weighting coefficient is set according to the user's focus. If the user is more concerned about the geometric similarity between BIM components, the weighting coefficient will be increased. If the user is more concerned about the attribute similarity between BIM components, the weighting coefficient will be decreased.

[0116] Furthermore, after calculating the comprehensive similarity of BIM components, it can be used for component retrieval or comparison of the overall BIM model. The similarity can be customized according to the user's needs. If it is necessary to retrieve identical BIM components, the preset similarity is set to 100%; if it is necessary to retrieve geometrically identical BIM components, the geometric similarity weight coefficient is increased, and the semantic similarity is reduced (for geometric reuse); if it is necessary to retrieve components with the same function, the semantic similarity weight coefficient can be increased (e.g., retrieving fire-related components to assist in drawing review).

[0117] This invention converts the geometry of BIM components to be compared into a BREP (Body Shape Array), and determines the shape similarity of surfaces in the BREP expression by calculating a metric tensor. The metric tensor is a coordinate-independent matrix expression, allowing the comparison method to compare geometric similarities without relying on a coordinate system. It determines dimensional consistency by calculating the inertia tensor of two similar surfaces, providing a more accurate comparison of whether components are geometrically identical. It achieves attribute comparison by sorting the attribute list and comparing the attribute set contents. The comprehensive calculation of geometric and attribute similarity makes the comparison method for BIM components more comprehensive. This addresses the technical problems of existing comparison methods that do not simultaneously consider geometric and attribute similarity, and that require a large number of calculation points when calculating similarity, resulting in low comparison efficiency. Based on parameterized comparison, it achieves a more universal and efficient BIM component comparison method.

[0118] To verify the effectiveness of the proposed BIM geometric component comparison method based on geometric tensor and graph matching, several experiments were designed and conducted to compare the geometric similarity and attribute similarity of different types of BIM geometric components.

[0119] In one specific embodiment, Experiment 1 is a comparative verification of a single complex geometric model. This experiment focuses on the geometric comparison of a single complex geometric component. An architectural window containing numerous geometric details was selected as the experimental object. This window has 392 faces and 758 edges, as shown below. Figure 3 As shown, the window is copied and rotated twice, and an IFC file is exported. Then, its BREP representation is generated and its geometric tensor is calculated. A graph matching algorithm is used to evaluate the geometric similarity and perform geometric reuse. The purpose of this experiment is to verify the accuracy and robustness of the proposed method when handling complex geometric components.

[0120] After comparing the geometric structure and reusability of the building windows in Experiment 1, it can be seen that the proposed geometric comparison method based on geometric tensor and graph matching can accurately identify identical geometric components without misjudgment when dealing with complex geometric components. It can ignore the displacement, rotation, and other transformations of geometric components, focusing only on the essential characteristics of the geometric shape, thus maintaining consistently high accuracy even with increased geometric complexity.

[0121] Specific data shows that, without geometric reuse, the IFC file size in Experiment 1 was 188KB, while after geometric reuse, the file size was reduced to 66KB, a reduction of approximately 65%. This demonstrates that the proposed geometric reuse technique can significantly reduce storage requirements, especially in complex geometric components. Geometric reuse reduces redundant geometric data and optimizes file storage and loading efficiency.

[0122] In another specific embodiment, Experiment 2 compares and reuses the geometric models of the standard floor, top floor, and bottom floor. The experiment uses a 24-story residential building model as shown below. Figure 4 As shown, the model has a total of 24,433 geometric components; its standard layer has 1,232 geometric components of different shapes and complexities, such as... Figure 5 As shown in the figure. The experiment performed geometric comparison and reuse analysis on all geometric models of the standard layer, checked the number of repeated components in a single standard layer, and geometrically reused the repeated components to reduce data redundancy.

[0123] In Experiment 2, a geometric comparison and reuse analysis was conducted on the standard floor geometric model of a 24-story residential building. Several identical components were artificially included in the display of the results for this standard floor. The experimental results show that the method presented in this paper can retrieve 100% of the duplicate components and perform geometric reuse of these components. The results are shown in Table 1 and... Figure 6 As shown, geometric reuse technology significantly reduces storage space requirements. Specifically, data from Experiment 2 indicates that without geometric reuse, the file size was 15.1 MB, while after reuse, the file size was reduced to 12.3 MB, a reduction of approximately 18.5%.

[0124] Table 1. Results of Retrieval and Reuse of Artificially Set Standard Layer Repetitive Components

[0125]

[0126] In Experiment 1, the time taken to perform the same type retrieval and reuse of a window with 392 faces and 758 edges was 41 seconds. In Experiment 2, the time taken to perform component retrieval and reuse of all geometric components in the standard layer was 4721 seconds.

[0127] Experiments 1 and 2 together verified the effectiveness of the BIM component comparison method proposed in this scheme when dealing with multi-standard-story building projects. Especially in large-scale BIM models, geometric reuse can greatly reduce the storage of redundant data and improve the overall model processing efficiency.

[0128] Example 2

[0129] The present invention also provides a BIM component comparison device for performing the comparison method as described in any one of the embodiments, comprising:

[0130] The coordinate system establishment module is used to convert the two BIM components to be compared into BREP expressions, obtain the control points based on the geometry corresponding to the two BREP expressions, and establish a four-dimensional coordinate system for each control point.

[0131] The metric tensor and inertia tensor calculation module is used to calculate the metric tensor and inertia tensor of two BIM components based on the four-dimensional coordinates of the control points. It constructs an undirected graph with faces as nodes and metric tensors and inertia tensors as node attributes. In the BREP expression, the common edge is the edge of the graph, and the approximate angle between adjacent faces is the edge attribute.

[0132] The geometric similarity calculation module is used to obtain the geometric similarity by matching the undirected graphs corresponding to two BIM components;

[0133] The attribute similarity calculation module is used to sort the attribute lists of two BIM components and calculate the attribute similarity based on semantic comparison.

[0134] The similarity judgment module is used to calculate the comprehensive similarity based on the geometric similarity and attribute similarity. When the comprehensive similarity is greater than the preset similarity, the two BIM components to be compared are determined to be similar.

[0135] The BIM component comparison device provided in this embodiment of the invention is used to execute the BIM component comparison method provided in any embodiment of the invention, and has corresponding functional modules and beneficial effects.

[0136] Those skilled in the art will readily understand that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for comparing BIM components, characterized in that, include: S1. Convert the two BIM components to be compared into BREP expressions, select control points from the faces of the geometry corresponding to the BREP expressions, obtain the weight coefficients of each control point through principal component analysis, and establish a four-dimensional coordinate system for each control point; the four-dimensional coordinate system is as follows: Among them, X A Y A Z A Let ω be the coordinate value of the control point in the natural coordinate system. A The weight parameters for the control points; S2. Calculate the metric tensor and inertia tensor of the two BIM components based on the four-dimensional coordinates of the control points, and construct an undirected graph with faces as nodes, metric tensors and inertia tensors as node attributes, common edges in the BREP expression as edges of the graph, and the angle between the normal vectors of adjacent faces as edge attributes. S3. Obtain geometric similarity by matching the undirected graphs corresponding to two BIM components; S4. Sort the attribute lists of the two BIM components alphabetically and calculate the attribute similarity based on semantic comparison. S5. Calculate the comprehensive similarity based on the geometric similarity and attribute similarity. When the comprehensive similarity is greater than the preset similarity, determine that the two BIM components to be compared are similar.

2. The comparison method as described in claim 1, characterized in that, The process of converting the two BIM components to be compared into BREP expressions includes: If the BIM component is an extruded body, the points of the geometry are obtained by analyzing the extruded body, and the set of points is organized in order to generate lines and surfaces in the BREP representation. If the BIM component is CSG, the shape of the geometry is generated by parsing the geometric primitives and Boolean operations in the CSG representation, and the geometry is converted into BREP representation.

3. The comparison method as described in claim 1, characterized in that, The calculation of the metric tensor and inertia tensor of the two BIM components based on the four-dimensional coordinates of the control points includes: Calculate the vector axes of the edges formed by the control points based on their four-dimensional coordinates: in, For the first i The four-dimensional coordinates of each control point For the first i From the first control point to the... i +1 control point vector axis, 1≤ i ≤ p ; Arrange the vector axes of all edges in the geometry in order to obtain vector bundles for different faces, and then decompose them: in, Let α be a vector bundle. for exist Components in direction, for exist Components in direction, for exist Components in direction, for exist Components in direction; The metric tensor is calculated by multiplying the outer product of the vector bundles of the surface by itself. The calculation formula is as follows: in, Let α be the metric tensor of the surface α; Calculate the centroid matrix of the two BIM components, simplify the eigenvalues ​​of the centroid matrix, and obtain the inertia tensor of all surfaces.

4. The comparison method as described in claim 1, characterized in that, S3 specifically includes: S31. Determine the matching order in an undirected graph using a breadth-first search algorithm; S32. Traverse the metric tensors of all faces of the two BIM components according to the matching order, and determine the two faces with equivalent metric tensors as similar faces. S33. Traverse the inertia tensors between all faces of the two BIM components according to the matching order, and determine the two faces with equivalent inertia tensors as faces with the same size. S34. Determine the set of identical faces by taking the intersection of the set of similar faces and the set of faces with the same size, and calculate the geometric similarity: in, For the geometric similarity of components, The number of identical faces. This represents the number of faces in a BIM component.

5. The method as described in claim 4, characterized in that, S4 specifically includes: The attribute sets of two BIM components are sorted, and natural language processing is used to analyze the content of synonymous attributes. Synonymous attributes with the same content are identified as the same attributes, and the attribute similarity is calculated. Formula for calculating attribute similarity: in, For the similarity of the attributes of the components, The number of items with the same attribute. The number of component attributes.

6. The method as described in claim 5, characterized in that, The formula for calculating the overall similarity is: in, For the overall similarity of components, For the geometric similarity of components, represents the weighting coefficient for geometric similarity.

7. A comparison device for BIM components, used to perform the comparison method as described in any one of claims 1-6, characterized in that, include: The coordinate system establishment module is used to convert the two BIM components to be compared into BREP expressions, select control points from the faces in the geometry corresponding to the BREP expressions, and obtain the weight coefficients of each control point through principal component analysis. The established four-dimensional coordinate system is as follows: Among them, X A Y A Z A Let ω be the coordinate value of the control point in the natural coordinate system. A The weight parameters for the control points; The metric tensor and inertia tensor calculation module is used to calculate the metric tensor and inertia tensor of two BIM components based on the four-dimensional coordinates of the control points. It constructs an undirected graph with faces as nodes, metric tensors and inertia tensors as node attributes, common edges in the BREP expression as edges of the graph, and the angle between the normal vectors of adjacent faces as edge attributes. The geometric similarity calculation module is used to obtain the geometric similarity by matching the undirected graphs corresponding to two BIM components; The attribute similarity calculation module is used to sort the attribute lists of two BIM components and calculate the attribute similarity based on semantic comparison. The similarity judgment module is used to calculate the comprehensive similarity based on the geometric similarity and attribute similarity. When the comprehensive similarity is greater than the preset similarity, the two BIM components to be compared are determined to be similar.

Images