Graph structure representation and model error recognition method and device of BIM model

Abstract

The invention relates to the technical field of BIM intelligent design and deep learning, in particular to a graph structure representation and model error recognition method and device of a BIM model, and the method comprises the steps: extracting semantic features, spatial features and topological features among various types of components according to the component information of different types of components in the BIM model; generating corresponding graph structure representation based on the semantic features, the spatial features and the topological features, and constructing a training data set by using the graph structure representation; and training the pre-training graph neural network model by using the training data set to obtain a trained training model, and carrying out transfer learning by using the trained training model to identify an error identification target of the BIM model by using a transfer learning model. Therefore, the problems that in the related technology, specific complex semantics, space and topological features of the BIM bring challenges which are difficult to neglect, and the BIM cannot be perfectly suitable for a pre-training and transfer learning model of the features of the BIM are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of BIM (Building Information Modeling) intelligent design and deep learning, and particularly relates to a graph structure representation of a BIM model and a model error identification method and device. BACKGROUND

[0002] As a digital representation of architectural characteristics, BIM integrates rich engineering data, including semantic features, spatial layout, material properties and other related information. For example, BIM model data often contains rich practical experience of engineers, such as standard ranges of door and window sizes, and spatial and connection relationships between various components. Therefore, it is crucial to extract, learn and reuse relevant design knowledge and patterns from these data to improve the level of architectural design, construction and maintenance.

[0003] In related technologies, structured BIM model data can be trained by machine learning to extract design features of the BIM model, deep learning can be used to learn and drive labeled data to extract corresponding features, and pre-training and transfer models can be used for self-supervised learning of unlabeled data to mine the internal features and context information of the data, and then transfer to specific tasks using a small amount of labeled data.

[0004] However, in related technologies, although significant progress has been made in text and visual data processing, the complex semantics, spatial and topological features unique to BIM models still pose a challenge that cannot be ignored, and the pre-training and transfer learning model for BIM model features cannot be perfectly adapted, which needs to be improved. SUMMARY

[0005] The present application provides a graph structure representation of a BIM model and a model error identification method and device to solve the problem that the complex semantics, spatial and topological features unique to BIM models in related technologies pose a challenge that cannot be ignored, and the pre-training and transfer learning model for BIM model features cannot be perfectly adapted.

[0006] The first aspect embodiment of the present application provides a BIM model graph structure representation and model error identification method, which is applied to a model training stage, wherein the method comprises the following steps: extracting semantic features, spatial features and topological features between various components according to component information of different types of components in a BIM model; generating a graph structure representation of the BIM model based on the semantic features, spatial features and topological features, and constructing a training data set suitable for a pre-trained graph neural network model using the graph structure representation; training the pre-trained graph neural network model using the training data set to obtain a trained training model, and performing transfer learning using the trained training model to obtain a transfer learning model, so as to identify an error identification target of the BIM model using the transfer learning model.

[0007] Optionally, in an embodiment of the present application, the semantic features, spatial features and topological features between various components are extracted according to component information of different types of components in a BIM model, comprising: extracting the semantic features based on component feature information of each component; extracting the spatial features based on position information of the each component and local spatial relationships between adjacent components; and extracting the topological features based on nesting relationships, connection relationships between different components, and vertical relationships between the different components and the ground.

[0008] Optionally, in an embodiment of the present application, the graph structure representation of the BIM model is generated based on the semantic features, spatial features and topological features, comprising: converting the different types of components into graph nodes in the graph structure representation, and obtaining node attributes of the graph nodes based on the semantic features; obtaining graph edges in the graph structure representation based on the spatial features or the topological features, and obtaining spatial edge attributes of the corresponding graph edges based on local spatial relationships between adjacent components; obtaining topological edge attributes of the corresponding graph edges based on the nesting relationships, connection relationships and vertical relationships in the topological features; and obtaining the graph structure representation based on at least one of the graph nodes, the node attributes, the graph edges, the spatial edge attributes and the topological edge attributes.

[0009] Optionally, in an embodiment of the present application, the training of the pre-trained graph neural network model using the training data set to obtain a trained training model comprises: generating a corresponding masking data set using the training data set; inputting the masking data set into an encoder of the pre-trained graph neural network model to generate first high-dimensional representations of graph nodes and / or second high-dimensional representations of edge attributes in the masking data set using the encoder; inputting the first high-dimensional representations and / or the second high-dimensional representations into a decoder of the pre-trained graph neural network model to reconstruct the masked graph nodes and the masked graph edges in the masking data set using the decoder to obtain reconstructed graph nodes and reconstructed graph edges; and obtaining the trained training model based on the training data set, the masking data set, the reconstructed graph nodes, and the reconstructed graph edges.

[0010] The second aspect embodiment of the present application provides a BIM model graph structure representation and model error identification method, applied to a model transfer learning stage, wherein the method comprises the following steps: extracting semantic features, spatial features, and topological features between various types of components according to component information of different types of components in a target BIM model; generating a target graph structure representation of the target BIM model based on the semantic features, spatial features, and topological features; inputting the target graph structure representation into a pre-trained transfer learning model to identify an error target identification result of the target BIM model, wherein the pre-trained transfer learning model is obtained by transfer learning of a trained training model.

[0011] Optionally, in an embodiment of the present application, before inputting the target graph structure representation into the pre-trained transfer learning model, the method further comprises: extracting semantic features, spatial features, and topological features between various types of components according to component information of different types of components in a BIM model containing an error identification target; generating an error graph structure representation of the BIM model containing the error identification target based on the semantic features, spatial features, and topological features, and constructing a transfer data set suitable for the pre-trained transfer learning model using the error graph structure representation; and training the pre-trained transfer learning model using the transfer data set until a preset transfer condition is met to obtain a trained transfer learning model.

[0012] The third aspect of the present application provides a BIM model graph structure representation and model error identification device, which is applied to a model training stage, wherein the device comprises: a first extraction module configured to extract semantic features, spatial features and topological features between various components according to component information of different types of components in a BIM model; a first construction module configured to generate a graph structure representation of the BIM model based on the semantic features, the spatial features and the topological features, and to construct a training data set suitable for a pre-trained graph neural network model by using the graph structure representation; and a first training module configured to train the pre-trained graph neural network model by using the training data set to obtain a trained training model, and to perform transfer learning by using the trained training model to obtain a transfer learning model, so as to identify an error recognition target of the BIM model by using the transfer learning model.

[0013] Optionally, in an embodiment of the present application, the first extraction module comprises: a first extraction unit configured to extract the semantic features based on component feature information of each component; a second extraction unit configured to extract the spatial features based on position information of the each component and local spatial relationships between adjacent components; and a third extraction unit configured to extract the topological features based on nesting relationships, connection relationships between different components, and vertical relationships between the different components and the ground.

[0014] Optionally, in an embodiment of the present application, the first construction module comprises: a first generation unit configured to convert the different types of components into graph nodes in the graph structure representation, and to obtain node attributes of the graph nodes based on the semantic features; a second generation unit configured to obtain graph edges in the graph structure representation based on the spatial features or the topological features, and to obtain spatial edge attributes of the spatial features corresponding graph edges based on local spatial relationships between adjacent components; an acquisition unit configured to acquire topological edge attributes of the topological features corresponding graph edges based on the nesting relationships, the connection relationships and the vertical relationships in the topological features; and a third generation unit configured to obtain the graph structure representation based on at least one of the graph nodes, the node attributes, the graph edges, the spatial edge attributes and the topological edge attributes.

[0015] Optionally, in an embodiment of the present application, the first training module comprises: a fourth generation unit configured to generate a corresponding masking data set by using the training data set; a fifth generation unit configured to input the masking data set into an encoder of the pre-trained graph neural network model to generate first high-dimensional representations of graph nodes and / or second high-dimensional representations of edge attributes in the masking data set by using the encoder; a sixth generation unit configured to input the first high-dimensional representations and / or the second high-dimensional representations into a decoder of the pre-trained graph neural network model to reconstruct the masked graph nodes and the masked graph edges in the masking data set by using the decoder to obtain reconstructed graph nodes and reconstructed graph edges; and a seventh generation unit configured to obtain the trained training model based on the training data set, the masking data set, the reconstructed graph nodes, and the reconstructed graph edges.

[0016] The fourth aspect embodiment of the present application provides a BIM model graph structure representation and model error identification device, which is applied to a model transfer learning stage. The device comprises: a second extraction module configured to extract semantic features, spatial features, and topological features between various types of components according to component information of different types of components in a target BIM model; a generation module configured to generate a target graph structure representation of the target BIM model based on the semantic features, the spatial features, and the topological features; and an identification module configured to input the target graph structure representation into a pre-trained transfer learning model to identify an error target identification result of the target BIM model, wherein the pre-trained transfer learning model is obtained by transfer learning of a trained training model.

[0017] Optionally, in an embodiment of the present application, the device further comprises: a third extraction module configured to extract semantic features, spatial features, and topological features between various types of components according to component information of different types of components in a BIM model containing an error identification target before the target graph structure representation is input into the pre-trained transfer learning model; a second construction module configured to generate an error graph structure representation of the BIM model containing the error identification target based on the semantic features, the spatial features, and the topological features, and to construct a transfer data set suitable for the pre-trained transfer learning model by using the error graph structure representation; and a second training module configured to train the pre-trained transfer learning model by using the transfer data set until a preset transfer condition is met to obtain a trained transfer learning model.

[0018] The fifth aspect embodiment of the present application provides an electronic device, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor. The processor executes the program to implement the BIM model graph structure representation and model error identification method as described in the above embodiments.

[0019] The sixth aspect of the present application provides a computer readable storage medium storing a computer program, which, when executed by a processor, implements the BIM model graph structure representation and model error identification method as above.

[0020] The seventh aspect of the present application provides a computer program product comprising a computer program, which, when executed, implements the BIM model graph structure representation and model error identification method as above.

[0021] The embodiments of the present application can extract semantic features, spatial features and topological features between various types of components according to the component information of different types of components in the BIM model, and then generate a graph structure representation of the BIM model, and use the graph structure representation to construct a training data set suitable for a pre-trained graph neural network model, thereby training the pre-trained graph neural network model, and using the trained training model for transfer learning to obtain a transfer learning model, and then identifying the error identification target of the BIM model. Through BIM component feature extraction, graph structure generation, pre-training and transfer learning, the accuracy, generalization ability and training efficiency of error identification can be significantly improved, providing an intelligent and automated solution for BIM quality control, and promoting the digital transformation of the construction industry. Thus, the problems of the related art, such as the complex semantic, spatial and topological features specific to the BIM model, which bring about challenges that cannot be ignored, and the pre-training and transfer learning model that cannot be perfectly adapted to the features of the BIM model, are solved.

[0022] Additional aspects and advantages of the present application will be in part apparent and in part pointed out hereinafter. BRIEF DESCRIPTION OF DRAWINGS

[0023] The above and / or additional aspects and advantages of the present application will become apparent and be readily appreciated from the following description, including the accompanying drawings, wherein:

[0024] Figure 1 A flowchart of a BIM model graph structure representation and model error identification method according to an embodiment of the present application is provided.

[0025] Figure 2 A schematic diagram of local spatial relationships between components according to an embodiment of the present application is provided.

[0026] Figure 3 A schematic diagram of topological relationships between components according to an embodiment of the present application is provided.

[0027] Figure 4 A partial flowchart of generating a graph structure representation according to an embodiment of the present application is provided.

[0028] Figure 5A schematic diagram of a GraphMAE2 graph neural network architecture provided according to an embodiment of the present application;

[0029] Figure 6 A schematic diagram of an improved information transfer mechanism provided according to an embodiment of the present application;

[0030] Figure 7 A block schematic diagram of a graph structure representation of a BIM model and a model error identification apparatus provided according to an embodiment of the present application;

[0031] Figure 8 A flowchart of a graph structure representation of a BIM model and a model error identification method provided according to yet another embodiment of the present application;

[0032] Figure 9 A flowchart of working principle of a graph structure representation of a BIM model and a model error identification method provided according to an embodiment of the present application;

[0033] Figure 10 A schematic diagram of three commonly existing error identification targets in actual engineering provided according to an embodiment of the present application;

[0034] Figure 11 A block schematic diagram of a graph structure representation of a BIM model and a model error identification apparatus provided according to yet another embodiment of the present application;

[0035] Figure 12 A structural schematic diagram of an electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0036] Embodiments of the present application are described in detail below with reference to the accompanying drawings. Examples of the embodiments are shown in the drawings, in which the same or similar components are denoted by the same or similar reference numerals throughout. The embodiments described below by reference to the drawings are exemplary and are intended to explain the present application, and should not be understood as limiting the present application.

[0037] The BIM model graph structure representation and model error identification method and device of the embodiments of the present application are described below with reference to the accompanying drawings. In view of the challenges brought by the complex semantic, spatial and topological characteristics specific to the BIM model mentioned in the background art, and the problem that the pre-training and transfer learning model cannot be perfectly adapted to the characteristics of the BIM model, the present application provides a BIM model graph structure representation and model error identification method. In this method, the semantic features, spatial features and topological features between various components can be extracted according to the component information of different types of components in the BIM model, and then the graph structure representation of the BIM model is generated. A training data set suitable for the pre-training graph neural network model is constructed using the graph structure representation, thereby training the pre-training graph neural network model, and performing transfer learning using the trained model to obtain a transfer learning model, and then identifying the error identification target of the BIM model. Through BIM component feature extraction, graph structure generation, pre-training and transfer learning, the accuracy, generalization ability and training efficiency of error identification can be significantly improved, providing an intelligent and automated solution for BIM quality control and promoting the digital transformation of the construction industry. Thus, the problems of related art, such as the challenges brought by the complex semantic, spatial and topological characteristics specific to the BIM model, and the inability of the pre-training and transfer learning model to perfectly adapt to the characteristics of the BIM model, are solved.

[0038] Specifically, Figure 1 A flowchart of a BIM model graph structure representation and model error identification method according to an embodiment of the present application is provided.

[0039] As Figure 1 shown, the BIM model graph structure representation and model error identification method is applied to the model training stage, wherein the method comprises the following steps:

[0040] In step S101, the semantic features, spatial features and topological features between various components are extracted according to the component information of different types of components in the BIM model.

[0041] It can be understood that the embodiments of the present application can realize the unified vectorization representation of the BIM model: the "semantic-spatial-topological" multi-dimensional features between various components in the BIM model are extracted and calculated according to the component information of different types of components in the BIM model.

[0042] Further, in the embodiments of the present application, all components can be represented as nodes of different types according to corresponding types, and the extracted semantic features (such as geometric properties, material information, etc., which are not limited in the present application) are represented as node features by means of numerical values, One-Hot (one-hot encoding) or multi-lingual large-scale embedding model coding, etc.; the spatial features and topological features between components can be represented as edges connecting corresponding nodes, and the corresponding spatial relationships (such as the shortest distance between two components, the included angle, etc., which are not limited in the present application) and topological relationships (such as connection relationships, embedding relationships, etc., which are not limited in the present application) are represented as corresponding edge attributes by means of numerical values or One-Hot coding, etc.; the specific representation methods can be set by those skilled in the art according to actual conditions, and the present application is not limited in this regard.

[0043] As a possible implementation manner, the embodiments of the present application can extract semantic features, spatial features and topological features between various components according to the component information of different types of components in the BIM model.

[0044] Optionally, in an embodiment of the present application, the semantic features, spatial features and topological features between various components are extracted according to the component information of different types of components in the BIM model, including: extracting semantic features based on the component feature information of each component; extracting spatial features based on the position information of each component and the local spatial relationship between adjacent components; and extracting topological features based on the nesting relationship, connection relationship between different components, and the vertical relationship between different components and the ground.

[0045] In some embodiments, the embodiments of the present application can extract the component feature information of different components based on the IFC (Industry Foundation Classes) file corresponding to the BIM model, and then obtain the semantic features of various components in the BIM model.

[0046] In the embodiments of the present application, the semantic features can include the shape, size features, structural purpose, family and family type of the components extracted from the IFC file, which are not limited in the present application, and can also include the features of each component itself, such as basic information, geometric shape, material properties, other related attributes, etc., which are not limited in the present application.

[0047] Further, in the embodiments of the present application, the shape of the component can include, but is not limited to, a cube, a cylinder, and other irregular shapes, etc., and the present application does not make specific limitations; the size characteristics can include, but are not limited to, three dimensions of length, width and height, etc., and the present application does not make specific limitations, and it should be noted that the embodiments of the present application can use the cross-sectional radius to represent the width and height for the cylindrical component (for example, a pipe, which is not limited by the present application), and can select three main size parameters that can represent the geometric characteristics of the component for the irregularly shaped component (for example, some connectors, which are not limited by the present application), and the present application does not make specific limitations; the structural purpose can be used to represent the functional positioning of the component, and can include, but is not limited to, structural components and non-structural components, etc., and the present application does not make specific limitations; the family and the family type can be used to represent the functional purpose of the object and the details of the geometric size, material, etc., and the present application does not make specific limitations; the basic information can include, but is not limited to, component ID (Identifier, identifier), name, family name, etc., and the present application does not make specific limitations, which are used to uniquely identify the component; the geometric shape can include, but is not limited to, the bounding box and the cross-sectional shape, etc., and the present application does not make specific limitations; the material attribute can describe the composition and characteristics of the component, etc., and the present application does not make specific limitations. Other related attributes can include, but are not limited to, durability, thermal conductivity, compressive strength, etc., and the present application does not make specific limitations.

[0048] In some embodiments, the embodiments of the present application can extract spatial features based on the position information of each component and the local spatial relationship between adjacent components.

[0049] In the embodiments of the present application, the position information representing the position of the component can be extracted from the IFC file corresponding to the BIM model, that is, the local coordinate system and the reference coordinate system, and then the local spatial relationship between the component and the remaining adjacent components within a range of 0.5 m is calculated to provide additional geometric and contextual information.

[0050] For example, the embodiments of the present application combine Figure 2 As shown, the local spatial relationship between different components is given, and the local spatial relationship can include, but is not limited to, the spatial relationship, the included angle, the shortest distance vector coordinates, the shortest distance, and the included angle between the center line forming plane and the horizontal plane, etc., and the present application does not make specific limitations.

[0051] Further, in the embodiments of the present application, the spatial relationship can be understood as the positional relationship of two components in space, determined by the center point or center line, which can include but is not limited to: (1) different surfaces; (2) coplanar but not parallel; (3) coplanar and parallel; (4) point to line; (5) point to point, etc., which are not specifically limited in the present application. Among them, (1), (2), (3) are applicable to the case where both components are positioned by center line; (4) is applicable to the case where one component is positioned by center point and the other component is positioned by center line; (5) is applicable to the case where both components are positioned by center point.

[0052] The included angle can be understood as the included angle between the center lines or points of the two components. If one component is positioned by the center point, the angle can be considered as 0°.

[0053] The shortest distance vector coordinates can correspond to the coordinates of the shortest distance vector between the center points or center lines of the two components.

[0054] The shortest distance can represent the shortest distance between the two components, and the negative distance represents the overlap or collision, and the positive distance represents the shortest distance between the components.

[0055] The included angle between the center line constituting plane and the horizontal plane can be understood as the included angle between the plane or line constituted by the two components and the horizontal plane. For the spatial relationship of (2)-(5), the included angle between the center line constituting plane and the horizontal plane is the included angle between the plane or line constituted by the center point or line and the horizontal plane. For the spatial relationship of (1), the included angle is 0°. According to the geometric principle, it is obvious that the five kinds of calculated spatial relationships can comprehensively describe the relative positional relationship between the two components.

[0056] In some embodiments, the embodiments of the present application can extract topological features based on the nesting relationship between different components, the connection relationship, and the vertical relationship between different components and the ground.

[0057] Among them, the embodiments of the present application can extract the nesting relationship, the connection relationship contained in the IFC file corresponding to the BIM model, and calculate the vertical relationship between the components and the ground, and then extract the corresponding topological features.

[0058] For example, the embodiments of the present application combine Figure 3 As shown, the topological relationship between different components is given, which can include but is not limited to: nesting relationship, connection relationship, and vertical relationship between components and ground (contact floor).

[0059] Among them, the nesting relationship can be understood as the construction situation that one component is completely nested in another component.

[0060] The connection relationship can be understood as the existence of direct physical connection between two components or the continuity of components in design semantics.

[0061] The contact floor can represent a contact relationship between the component and the ground, and the topological relationship is extracted by calculating the positional relationship between the component and the floor boundary box, so as to more intuitively represent the functional layout and component distribution of different building floors.

[0062] In step S102, a graph structure representation of the BIM model is generated based on the semantic features, spatial features and topological features, and a training data set suitable for the pre-trained graph neural network model is constructed using the graph structure representation.

[0063] In actual execution, the embodiments of the present application can obtain a graph structure representation of the BIM model according to the semantic features, spatial features and topological features, and then construct a training data set suitable for the pre-trained graph neural network model using the graph structure representation.

[0064] Optionally, in an embodiment of the present application, generating a graph structure representation of the BIM model based on the semantic features, spatial features and topological features includes: converting different types of components into graph nodes in the graph structure representation, and obtaining node attributes of the graph nodes based on the semantic features; obtaining graph edges in the graph structure representation based on the spatial features or the topological features, and obtaining spatial edge attributes of the spatial features corresponding to the graph edges based on the local spatial relationship between adjacent components; obtaining topological edge attributes of the topological features corresponding to the graph edges based on the nesting relationship, the connection relationship and the vertical relationship in the topological features; and obtaining the graph structure representation based on at least one of the graph nodes, the node attributes, the graph edges, the spatial edge attributes and the topological edge attributes.

[0065] In some embodiments, the embodiments of the present application can represent all different types of components as different types of graph nodes, and represent the extracted semantic features as node attributes of the corresponding graph nodes.

[0066] For example, the embodiments of the present application can regard all components in the BIM model as graph nodes in the graph structure representation; and based on the division of component types in IFC, these graph nodes are divided into different types, such as walls, beams, pipes, etc., and the present application does not make specific limitations, and the extracted semantic features are converted into node attributes of the corresponding graph nodes.

[0067] In some embodiments, the embodiments of the present application can represent the spatial relationship and the topological relationship between the components obtained by extraction and calculation as graph edges connecting the corresponding nodes, and then obtain the corresponding spatial edge attributes and topological edge attributes.

[0068] Exemplarily, the embodiment of the present application can further classify the graph edges representing the topological features according to the nesting relationship, the connection relationship and the vertical relationship in the topological features, so as to obtain the topological edge attribute of the graph edge corresponding to the topological features; the spatial features between the components are regarded as the fourth type of graph edges in the graph structure representation, and the local spatial relationship between the adjacent components is taken as the spatial edge attribute of the graph edge corresponding to the spatial features; when there is a topological feature between two components, the spatial feature is ignored, and it is ensured that the two component nodes are connected by only one edge.

[0069] In addition, it should be noted that the embodiment of the present application can use various embedding technologies to vectorize the node features, and align and fill the missing edge attributes, so as to realize the graph-based representation of the BIM model and form the training data set of the pre-trained graph neural network model.

[0070] Exemplarily, the embodiment of the present application can encode the extracted non-numerical node attributes to convert them into a computer-readable form; for some components that lack specific node and edge attributes or whose values cannot be extracted, a 0 placeholder is used to maintain consistency; all node and edge attributes are normalized, and the flowchart is as shown in Figure 4

[0071] Step S401: Encode the extracted non-numerical node attributes to convert them into a computer-readable form.

[0072] In the embodiment of the present application, the encoding method can include but is not limited to One-hot encoding, Numeric encoding and M3E (Moka Massive Mixed Embedding) encoding, and the present application does not make specific limitations. The One-hot encoding method can be understood as a method of representing a classification variable as a sparse vector, in which only the position corresponding to the category is 1, and the other positions are 0. The M3E encoding method can be understood as a method in the XX library, which embeds the text into a fixed-dimensional (for example, 64 dimensions, which are not limited in the present application) vector representation, facilitating the training and inference of the model.

[0073] Further, the encoding method of each attribute provided by the embodiment of the present application is shown in Table 1. Table 1 is a schematic table of the attribute encoding method according to an embodiment of the present application.

[0074] Table 1

[0075]

[0076] Step S402: For some components that lack specific node and edge attributes or whose values cannot be extracted, a 0 placeholder is used to maintain consistency.

[0077] ​Step S403: normalize all node and edge attributes.

[0078] In the embodiments of the present application, the normalization process is based on the drawings of each floor, and the relevant attribute values are scaled to the range of -1 to 1 under the premise of keeping the 0 value unchanged, and the calculation formula can be but not limited to expressed as:

[0079]

[0080] wherein X i represents the i-th attribute value of all components in the same floor drawing; X' i represents the normalized data.

[0081] In step S103, the pre-trained graph neural network model is trained using the training data set to obtain a trained training model, and the trained training model is used for transfer learning to obtain a transfer learning model, so as to identify the error recognition target of the BIM model using the transfer learning model.

[0082] It can be understood that the embodiments of the present application can learn the rich design experience and rules implied in the BIM model by using the pre-trained graph neural network model, and train the pre-trained graph neural network model using the training data set, thereby obtaining the trained training model.

[0083] Pre-training is a strategy for training deep learning models, and its core is to use large-scale data sets to preliminarily train the model, so that the model learns general feature representation. This process is similar to the basic learning stage before humans learn new knowledge, through extensive reading, observation and experience accumulation.

[0084] In some embodiments, the embodiments of the present application can use the trained training model for transfer learning, thereby obtaining the corresponding transfer learning model, and using the transfer learning model to identify the error recognition target of the BIM model.

[0085] Optionally, in an embodiment of the present application, training the pre-trained graph neural network model using the training data set to obtain a trained training model comprises: generating a corresponding masked data set using the training data set; inputting the masked data set into the encoder of the pre-trained graph neural network model to generate first high-dimensional representations of graph nodes and / or second high-dimensional representations of edge attributes in the masked data set using the encoder; inputting the first high-dimensional representations and / or the second high-dimensional representations into the decoder of the pre-trained graph neural network model to reconstruct the masked graph nodes and the masked graph edges in the masked data set using the decoder to obtain reconstructed graph nodes and reconstructed graph edges; and obtaining the trained training model based on the training data set, the masked data set, the reconstructed graph nodes and the reconstructed graph edges.

[0086] It can be understood that in the embodiments of the present application, the pre-trained graph neural network model can include two parts of an encoder and a decoder, wherein the task of the encoder is to construct a first high-dimensional representation of the graph nodes containing global features and a second high-dimensional representation of the edge attributes from the input graph in which part of the nodes and edges are masked, by capturing the local structure of the graph, i.e., the features of the unmasked nodes and edges and their neighborhoods; the decoder then reconstructs and recovers the masked nodes and edge attributes in the input graph according to the first high-dimensional representation and the second high-dimensional representation, and then obtains the reconstructed graph nodes and the reconstructed graph edges.

[0087] In some embodiments, the pre-trained graph neural network model can be trained by using a training data set, and then a trained model is obtained.

[0088] For example, the training data set can be input into the pre-trained graph neural network model, the pre-training problem can be converted into a node and edge feature decoding and reconstruction problem by masking part of the node and edge features in the input graph, the model can be prompted to learn from local information to recover the entire graph structure, and then a trained graph neural network model is obtained. The architecture of the pre-trained graph neural network model is selected as the GraphMAE2 graph neural network architecture, and a schematic diagram thereof is shown in Figure 5 The architecture includes two parts of an encoder and a decoder, and the purpose of pre-training is to obtain a trained encoder by inputting the reconstructed graph nodes and the reconstructed graph edges. The main task of the encoder is to construct a first high-dimensional representation of the graph nodes containing global features and a second high-dimensional representation of the edge attributes from the input graph in which part of the nodes and edges are masked, by capturing the local structure of the graph, i.e., the features of the unmasked nodes and their neighborhoods; the decoder then reconstructs and recovers the masked nodes and edge attributes in the input graph according to the first high-dimensional representation and the second high-dimensional representation, and then obtains the reconstructed graph nodes and the reconstructed graph edges. The decoder includes two decoding strategies: (1) a multi-view random re-masking strategy to reduce overfitting of the input features; and (2) a latent representation prediction to obtain more information targets. The specific settings can be made by a person skilled in the art according to the actual situation, and the present application does not make specific limitations.

[0089] Further, in the embodiments of the present application, the multi-view random re-mask strategy can be understood as randomly re-masking the embedded high-dimensional features during decoding, i.e., multiple times, randomly masking different nodes, and reconstructing the input features from the hidden encoding of the unmasked neighbor nodes using a shared decoder. The introduction of randomness is as a regularization, thereby reducing the sensitivity to perturbations in the input features and reducing overfitting. After hyperparameter tuning, the embodiments of the present application use a large masking rate of 50% during encoding and decoding, the masked node attributes are randomly filled, 3 random re-mask views are generated during decoding for decoding, and then the scaled cosine error is used to measure the reconstruction error, and the errors of the 3 views are summed for training, the calculation process can be but not limited to represented as:

[0090]

[0091] wherein V represents the input mask node, K represents the number of random re-masking during decoding, which can be 3, the present application does not make specific limitations, x i represents the i-th input feature, represents the i-th row of the predicted feature matrix Z (j) under the j-th mask view, cos(θ ij ) represents the included angle between two vectors x i and , T represents transposition, v i represents one of the input nodes.

[0092] The core of latent representation prediction is to design an additional information prediction task while minimizing the direct impact of input features on this task, therefore, the embodiments of the present application can choose to predict in the representation space rather than the input feature space. This part involves three networks, target generator, encoder and MLP (Multilayer Perceptron, Multilayer Perceptron) projector. The role of the target generator is to generate latent prediction targets from the unmasked graph, which shares the same network architecture with the encoder and MLP projector, but uses different weights. At the same time, the embodiments of the present application can project the encoding results of the encoder to the representation space using the MLP projector to obtain the latent prediction The encoder is trained by minimizing the distance between and , the calculation process can be but not limited to represented as

[0093]

[0094] ζ←τζ+(1-τ)ξ,

[0095] wherein N represents the number of nodes, is a matrix the i-th row of matrix the i-th row of matrix the i-th row of matrix denotes the angle between the two vectors, and the learnable weight ζ in the goal generator is updated by exponentially moving average on the learnable weights ξ of the encoder and the MLP projector and using a weight decay parameter τ.

[0096] Further, the embodiments of the present application can obtain a final loss function, and the expression thereof can be but is not limited to:

[0097]

[0098] It should be noted that in the embodiments of the present application, the encoder and the decoder are both 2-layer graph attention networks.

[0099] Figure 5 The GraphMAE2 graph neural network architecture shown in the figure is only applicable to homogeneous graphs and is not applicable to heterogeneous networks, therefore, the embodiments of the present application can improve the network information transfer mechanism of the architecture to make it applicable to heterogeneous networks. Figure 6 is a schematic diagram of the improved information transfer mechanism, as Figure 6 shown, the embodiments of the present application can process neighbor information according to node types, and aggregate the features of semantic, spatial and topological nodes respectively by introducing a type-aware attention mechanism.

[0100] Further, the embodiments of the present application can input a training data set into the improved GraphMAE2 graph neural network shown in Figure 6 by masking part of the node and edge features in the input graph, the training problem is converted into a node feature decoding reconstruction problem, prompting the model to learn from local information to recover the entire graph structure, and then obtaining the trained training model.

[0101] The graph representation and error identification method for BIM models proposed in this application can extract semantic, spatial, and topological features between different types of components based on component information in the BIM model, thereby generating a graph representation of the BIM model. This graph representation is then used to construct a training dataset suitable for a pre-trained graph neural network model, which is then used to train the pre-trained model. The trained model is then used for transfer learning to obtain a transfer learning model, which is used to identify error targets in the BIM model. Through BIM component feature extraction, graph generation, pre-training, and transfer learning, the accuracy, generalization ability, and training efficiency of error identification can be significantly improved, providing an intelligent and automated solution for BIM quality control and promoting the digital transformation of the construction industry. This solves the problem in related technologies where the complex semantic, spatial, and topological features unique to BIM models pose significant challenges, making it difficult to perfectly apply pre-training and transfer learning models to BIM model features.

[0102] Next, referring to the accompanying drawings, we describe the graphical representation of the BIM model and the model error identification device proposed according to the embodiments of this application.

[0103] Figure 7 This is a block diagram of the graphical representation of a BIM model and the model error identification device provided in the embodiments of this application.

[0104] like Figure 7 As shown, the BIM model graph structure representation and model error identification device 10 is applied in the model training stage. The BIM model graph structure representation and model error identification device 10 includes: a first extraction module 100, a first construction module 200 and a first training module 300.

[0105] The first extraction module 100 is used to extract semantic features, spatial features and topological features between different types of components based on the component information of different types of components in the BIM model.

[0106] The first building module 200 is used to generate a graph structure representation of the BIM model based on semantic features, spatial features and topological features, and to use the graph structure representation to build a training dataset suitable for pre-trained graph neural network models.

[0107] The first training module 300 is used to train a pre-trained graph neural network model using the training dataset to obtain a trained model, and to perform transfer learning using the trained model to obtain a transfer learning model, which is then used to identify incorrectly identified targets in the BIM model.

[0108] Optionally, in one embodiment of this application, the first extraction module 100 includes: a first extraction unit, a second extraction unit, and a third extraction unit.

[0109] The first extraction unit is configured to extract semantic features based on the component feature information of each component.

[0110] The second extraction unit is configured to extract spatial features based on the position information of each component and the local spatial relationship between adjacent components.

[0111] The third extraction unit is configured to extract topological features based on the nesting relationship, the connection relationship between different components, and the vertical relationship between different components and the ground.

[0112] Optionally, in an embodiment of the present application, the first construction module 200 comprises a first generation unit, a second generation unit, an acquisition unit and a third generation unit.

[0113] The first generation unit is configured to convert different types of components into graph nodes in a graph structure representation, and obtain node attributes of the graph nodes based on the semantic features.

[0114] The second generation unit is configured to obtain graph edges in the graph structure representation based on the spatial features or the topological features, and obtain spatial edge attributes of the spatial features corresponding to the graph edges based on the local spatial relationship between adjacent components.

[0115] The acquisition unit is configured to acquire topological edge attributes of the topological features corresponding to the graph edges based on the nesting relationship, the connection relationship and the vertical relationship in the topological features.

[0116] The third generation unit is configured to obtain the graph structure representation based on at least one of the graph nodes, the node attributes, the graph edges, the spatial edge attributes and the topological edge attributes.

[0117] Optionally, in an embodiment of the present application, the first training module 300 comprises a fourth generation unit, a fifth generation unit, a sixth generation unit and a seventh generation unit.

[0118] The fourth generation unit is configured to generate a corresponding masking data set by using a training data set.

[0119] The fifth generation unit is configured to input the masking data set into an encoder of a pre-trained graph neural network model, so as to generate first high-dimensional representations of graph nodes and / or second high-dimensional representations of edge attributes in the masking data set by using the encoder.

[0120] The sixth generation unit is configured to input the first high-dimensional representations and / or the second high-dimensional representations into a decoder of the pre-trained graph neural network model, so as to reconstruct the masked graph nodes and the masked graph edges in the masking data set by using the decoder, to obtain reconstructed graph nodes and reconstructed graph edges.

[0121] The seventh generating unit is configured to obtain the trained training model based on the training data set, the masking data set, the reconstructed graph node and the reconstructed graph edge.

[0122] It should be noted that the aforementioned explanation and description of the embodiment of the BIM model graph structure representation and model error identification method also applies to the BIM model graph structure representation and model error identification device 10 of this embodiment, which will not be described here.

[0123] The BIM model graph structure representation and model error identification device 10 according to the embodiment of the present application can extract semantic features, spatial features and topological features between various types of components according to component information of different types of components in the BIM model, and then generate a graph structure representation of the BIM model, and use the graph structure representation to construct a training data set suitable for a pre-trained graph neural network model, thereby training the pre-trained graph neural network model, and using the trained training model for transfer learning to obtain a transfer learning model, and then identifying the error identification target of the BIM model. Through BIM component feature extraction, graph structure generation, pre-training and transfer learning, the accuracy, generalization ability and training efficiency of error identification can be significantly improved, providing an intelligent and automated solution for BIM quality control, and promoting the digital transformation of the construction industry. Thus, the problems of the related art, such as the complex semantic, spatial and topological features specific to the BIM model, which bring about challenges that cannot be ignored, and the pre-training and transfer learning model that cannot be perfectly adapted to the features of the BIM model, are solved.

[0124] The above embodiment describes the model training phase, and the following embodiment describes the model transfer learning phase.

[0125] Figure 8 A flowchart of a BIM model graph structure representation and model error identification method according to yet another embodiment of the present application is provided.

[0126] As shown in Figure 8 The BIM model graph structure representation and model error identification method is applied to the model transfer learning phase, and the method comprises the following steps:

[0127] In step S801, semantic features, spatial features and topological features between various types of components are extracted according to component information of different types of components in the target BIM model.

[0128] In step S802, a target graph structure representation of the target BIM model is generated based on the semantic features, spatial features and topological features.

[0129] In step S803, the target graph structure representation is input into the pre-trained transfer learning model to identify the error target recognition result of the target BIM model, wherein the pre-trained transfer learning model is obtained by transfer learning of the trained training model.

[0130] Optionally, in an embodiment of the present application, before the target graph structure representation is input into the pre-trained transfer learning model, it further includes: extracting semantic features, spatial features and topological features between various types of components according to component information of different types of components in the BIM model containing the error recognition target; generating an error graph structure representation of the BIM model containing the error recognition target based on the semantic features, spatial features and topological features, and constructing a transfer dataset suitable for the pre-trained transfer learning model using the error graph structure representation; training the pre-trained transfer learning model using the transfer dataset until a preset transfer condition is met to obtain the trained transfer learning model.

[0131] In some embodiments, the embodiments of the present application can represent the target BIM model as a corresponding target graph structure representation, annotate a small amount of BIM model data represented as an error graph structure representation according to an error recognition target, thereby forming a transfer dataset suitable for a pre-trained transfer learning model, and fine-tune the pre-trained transfer learning model using the transfer learning dataset until a certain transfer condition is met, such as inputting the target BIM model excluding the error graph structure representation into the trained transfer learning model, which can realize automatic error recognition of the BIM model under few-shot transfer. The certain transfer condition can be set by those skilled in the art according to actual conditions, and the present application does not make specific limitations.

[0132] Further, the content of the embodiments of the present application for generating an error graph structure representation of the BIM model containing the error recognition target and constructing a transfer dataset suitable for the pre-trained transfer learning model can be: converting the error recognition task in the BIM model containing the error recognition target into a node and edge classification problem; annotating the correct node or edge as 1 and the error node or edge as 0 according to the error recognition target for a small amount of to-be-recognized data, thereby forming a transfer dataset for transfer learning.

[0133] In addition, the content of the embodiments of the present application for training the pre-trained transfer learning model using the transfer dataset can be: only retaining the encoder in the pre-trained transfer learning model for feature extraction, and connecting a new graph attention network as a classifier after the encoder; keeping the encoder unchanged during training and only training and optimizing the classifier, thereby greatly improving the calculation accuracy and supporting unified recognition of various BIM model errors, which is not achieved by existing methods.

[0134] The working principle of the BIM model graph structure representation and model error identification method according to the embodiment of the application will be explained below in combination with a specific embodiment.

[0135] wherein, Figure 9 The working principle of the BIM model graph structure representation and model error identification method according to the embodiment of the application is shown in the flowchart.

[0136] Step S901: Characterize the target BIM model as a corresponding target graph structure representation.

[0137] According to the embodiment of the application, the semantic features, spatial features and topological features between various types of components can be extracted according to the component information of different types of components in the target BIM model, and then the corresponding target graph structure representation can be obtained by using the semantic features, spatial features and topological features.

[0138] Step S902: Generate an error graph structure representation of the BIM model containing the error identification target, thereby obtaining a corresponding migration dataset.

[0139] According to the embodiment of the application, three error identification targets commonly existing in actual projects are selected: Figure 10 (1) Semantic error: For example, part of the wall is incorrectly created using the beam family, which causes a chain reaction in subsequent cost estimation, operation and maintenance, etc., resulting in deviation of cost estimation and increase of maintenance difficulty; (2) Attribute value error: For example, the height of part of the door does not meet the relevant requirements. According to the industry standard, the height of the door should not be less than 200 cm, and should not exceed 240 cm; (3) Topological relationship error: For example, the connection relationship between part of the mechanical and electrical pipeline components and their connecting pieces in distance or attribute is missing, which usually occurs when modifying the drawing to move the component. The position of the mechanical and electrical pipeline component or its corresponding connecting piece has changed, thereby causing a gap between them, or even if they are connected in geometry, the connection in the attribute may disappear due to relative movement. The specific selection method can be set by a person skilled in the art according to the actual situation, and the application does not make specific limitations.

[0140] Further, according to the three types of error identification targets, the application can label the correct nodes or edges as 1 and the incorrect nodes or edges as 0, thereby forming a migration dataset for transfer learning.

[0141] Step S903: Fine-tune the pre-trained transfer learning model using the transfer learning dataset.

[0142] Among them, the embodiment of the application only retains the encoder in the pre-training transfer learning model for feature extraction, and connects a new graph attention network with the same architecture after the encoder as a classifier; the encoder is kept unchanged during training, and only the classifier is trained and optimized. For Figure 10 The model after transfer learning can divide the nodes and edges into 4 categories, representing correct, error 1, error 2 and error 3, respectively.

[0143] Step S904: input the target graph structure representation into the pre-trained transfer learning model to identify the error target recognition result of the target BIM model.

[0144] According to the graph structure representation and model error identification method of the BIM model provided in the embodiment of the application, the semantic features, spatial features and topological features between various types of components can be extracted according to the component information of different types of components in the target BIM model, and then the target graph structure representation of the target BIM model is generated, and the target graph structure representation is input into the pre-trained transfer learning model to identify the error target recognition result of the target BIM model. Through BIM component feature extraction, graph structure generation, pre-training and transfer learning, the accuracy, generalization ability and training efficiency of error identification can be significantly improved, providing an intelligent and automated solution for BIM quality control, and promoting the digital transformation of the construction industry. Thus, the problems in the related art that the complex semantic, spatial and topological features of the BIM model bring about challenges that cannot be ignored, and the pre-training and transfer learning model cannot be perfectly applied to the features of the BIM model are solved.

[0145] Secondly, the graph structure representation and model error identification device of the BIM model according to the embodiment of the application is described with reference to the accompanying drawings.

[0146] Figure 11 The block schematic diagram of the graph structure representation and model error identification device of the BIM model provided for another embodiment of the application.

[0147] As Figure 11 shown, the graph structure representation and model error identification device 20 of the BIM model is applied in the model transfer learning stage, wherein the graph structure representation and model error identification device 20 of the BIM model comprises a second extraction module 400, a generation module 500 and an identification module 600.

[0148] The second extraction module 400 is configured to extract semantic features, spatial features and topological features between various types of components according to component information of different types of components in the target BIM model.

[0149] The generation module 500 is configured to generate a target graph structure representation of the target BIM model based on the semantic features, spatial features and topological features.

[0150] The identification module 600 is configured to input the target graph structure representation into a pre-trained transfer learning model to identify an error target identification result of the target BIM model, wherein the pre-trained transfer learning model is obtained by transfer learning of the trained training model.

[0151] Optionally, in an embodiment of the present application, the method further comprises a third extraction module, a second construction module and a second training module.

[0152] The third extraction module is configured to extract semantic features, spatial features and topological features between different types of components according to component information of the BIM model containing the error identification target before the target graph structure representation is input into the pre-trained transfer learning model.

[0153] The second construction module is configured to generate an error graph structure representation of the BIM model containing the error identification target based on the semantic features, the spatial features and the topological features, and to construct a transfer dataset suitable for the pre-trained transfer learning model using the error graph structure representation.

[0154] The second training module is configured to train the pre-trained transfer learning model using the transfer dataset until a preset transfer condition is met to obtain the trained transfer learning model.

[0155] It should be noted that the foregoing explanation and description of the embodiments of the graph structure representation of the BIM model and the method for identifying model errors are also applicable to the graph structure representation of the BIM model and the device 20 for identifying model errors of the embodiments, which will not be described here.

[0156] The device 20 for identifying model errors of the BIM model according to the embodiments of the present application can extract semantic features, spatial features and topological features between different types of components according to component information of the target BIM model, and then generate a target graph structure representation of the target BIM model, and input the target graph structure representation into a pre-trained transfer learning model to identify an error target identification result of the target BIM model. Through BIM component feature extraction, graph structure generation, pre-training and transfer learning, the accuracy, generalization ability and training efficiency of error identification can be significantly improved, providing an intelligent and automated solution for BIM quality control, and promoting the digital transformation of the construction industry. Thus, the problems of the related art, such as the complex semantic, spatial and topological features of the BIM model, which bring about challenges that cannot be ignored, and the pre-training and transfer learning model that cannot be perfectly adapted to the features of the BIM model, are solved.

[0157] Figure 12 A structural schematic diagram of an electronic device according to an embodiment of the present application is provided. The electronic device can include:

[0158] The memory 1201, the processor 1202 and the computer program stored in the memory 1201 and capable of running on the processor 1202.

[0159] The processor 1202 implements the graph structure representation of the BIM model and the model error identification method provided in the above embodiments when executing the program.

[0160] Further, the electronic device further comprises:

[0161] The communication interface 1203 is used for communication between the memory 1201 and the processor 1202.

[0162] The memory 1201 is used to store the computer program capable of running on the processor 1202.

[0163] The memory 1201 can include a high-speed RAM memory, and can also include a non-volatile memory, for example, at least one disk memory.

[0164] If the memory 1201, the processor 1202 and the communication interface 1203 are independently implemented, the communication interface 1203, the memory 1201 and the processor 1202 can be connected to each other through a bus and complete communication between each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, Figure 12 Only one thick line is used in the figure, but it does not mean that there is only one bus or only one type of bus.

[0165] Optionally, in a specific implementation, if the memory 1201, the processor 1202 and the communication interface 1203 are integrated on a chip, the memory 1201, the processor 1202 and the communication interface 1203 can complete communication between each other through an internal interface.

[0166] The processor 1202 can be a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0167] The embodiment of the present application further provides a computer readable storage medium, which stores a computer program, and the program is executed by a processor to realize the BIM model graph structure representation and model error identification method.

[0168] The embodiment of the present application further provides a computer program product, which comprises a computer program, and the program is executed to realize the BIM model graph structure representation and model error identification method.

[0169] In the description of the specification, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the specification, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any appropriate manner in any one or N embodiments or examples. In addition, the person skilled in the art can combine and combine the different embodiments or examples described in the specification and the features of the different embodiments or examples without contradiction.

[0170] In addition, the terms "first", "second" are only for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include at least one of the features. In the description of the present application, the meaning of "N" is at least two, for example, two, three, etc., unless otherwise specifically limited.

[0171] Any process or method descriptions in flow charts or otherwise described herein can be understood as representing code modules, segments, or portions of code that include one or more executable instructions for performing a step, a function of the custom logic or process, and that the scope of the preferred embodiments of the present application encompasses additional implementation in which the functions are performed in a different order, in substantially simultaneous fashion, or as part of a concurrent process, as will be understood by those skilled in the art.

[0172] The logic and / or steps represented in the flowcharts and / or described herein, for example, can be considered as a list of instructions to implement a logical function, and can be embodied in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, processor- containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions, or a combination thereof. For purposes of this specification, a "computer-readable medium" can be any apparatus that can contain, store, communicate, propagate, or transport the program for use by or in connection with the instruction execution system, apparatus, or device. The computer-readable medium can be a product of the manufacturing and / or processing. More specific examples (a non-exhaustive list) of the computer-readable medium include the following: an electronic connection having one or N wires (electronic devices), a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or Flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). In addition, the computer-readable medium can even be paper or another suitable medium upon which the program can be printed, because the program can be electronically captured, via the optical scan of the paper or other medium, then compiled, interpreted, or otherwise processed in a suitable manner, if necessary, and then stored in the computer memory.

[0173] It should be understood that aspects of the application can be implemented in hardware, software, firmware, or combinations thereof. In the above embodiments, the N steps or methods can be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. If implemented in hardware and in another embodiment, the implementation can be accomplished with any or a combination of the following technologies, which are all well known in the art: a discrete logic circuit(s) having logic gates for implementing logic functions on data signals, an application specific integrated circuit having appropriate combinational logic gates, a programmable gate array(s) (PGA), a field programmable gate array (FPGA), etc.

[0174] Those of skill in the art would understand that the steps carried out by the above-mentioned embodiments can be implemented by programs instructing the relevant hardware to complete all or part of the steps, and the programs can be stored in a computer-readable storage medium. When the programs are executed, they include one or a combination of the steps of the method embodiments.

[0175] In addition, each of the functional units in the various embodiments of the present application can be integrated in one processing module, or each of the units can be physically present separately, or two or more units can be integrated in one module. The integrated module can be implemented in the form of hardware or in the form of a software functional module. When the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer readable storage medium.

[0176] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A graphical representation of a BIM model and a method for identifying model errors, characterized in that, Applied to the model training phase, the method includes the following steps: Based on the component information of different types of components in the Building Information Model (BIM) model, semantic features, spatial features, and topological features among various components are extracted; Based on the semantic features, spatial features, and topological features, a graph structure representation of the BIM model is generated, and the graph structure representation is used to construct a training dataset suitable for a pre-trained graph neural network model. The pre-trained graph neural network model is trained using the training dataset to obtain a trained model, and the trained model is then used for transfer learning to obtain a transfer learning model, which is then used to identify incorrectly identified targets in the BIM model.

2. The method according to claim 1, characterized in that, The extraction of semantic features, spatial features, and topological features among different types of components based on component information in the BIM model includes: Based on the component feature information of each component, the semantic features are extracted; Based on the location information of each component and the local spatial relationship between adjacent components, the spatial features are extracted; The topological features are extracted based on the nesting and connection relationships between different components, as well as the vertical relationship between the different components and the ground.

3. The method according to claim 1, characterized in that, The process of generating a graph structure representation of the BIM model based on the semantic features, spatial features, and topological features includes: The different types of components are converted into graph nodes in the graph structure representation, and the node attributes of the graph nodes are obtained based on the semantic features. Based on the spatial features or the topological features, the graph edges in the graph structure representation are obtained, and based on the local spatial relationships between adjacent components, the spatial edge attributes of the graph edges corresponding to the spatial features are obtained. Based on the nesting, connection, and vertical relationships in the topological features, the topological edge attributes of the graph edges corresponding to the topological features are obtained; The graph structure representation is obtained based on at least one of the graph nodes, the node attributes, the graph edges, the spatial edge attributes, and the topological edge attributes.

4. The method according to claim 1, characterized in that, The step of training the pre-trained graphical neural network model using the training dataset to obtain the trained model includes: The corresponding occlusion dataset is generated using the training dataset; The occlusion dataset is input into the encoder of the pre-trained graph neural network model to generate a first high-dimensional representation of the graph nodes and / or a second high-dimensional representation of the edge attributes in the occlusion dataset using the encoder. The first high-dimensional representation and / or the second high-dimensional representation are input into the decoder of the pre-trained graph neural network model to reconstruct the occluded graph nodes and occluded graph edges in the occluded dataset using the decoder, so as to obtain the reconstructed graph nodes and reconstructed graph edges. The trained model is obtained based on the training dataset, the occlusion dataset, the reconstructed graph nodes, and the reconstructed graph edges.

5. A graphical representation of a BIM model and a method for identifying model errors, characterized in that, The BIM model graph structure representation and model error identification method as described in any one of claims 1-4 is applied to the model transfer learning stage, wherein the method includes the following steps: Based on the component information of different types of components in the target BIM model, extract the semantic features, spatial features and topological features between various types of components; Based on the semantic features, spatial features, and topological features, a target graph structure representation of the target BIM model is generated. The target graph structure representation is input into a pre-trained transfer learning model to identify erroneous target identification results of the target BIM model, wherein the pre-trained transfer learning model is obtained by transfer learning from the trained model.

6. The method according to claim 5, characterized in that, Before inputting the target graph structure representation into a pre-trained transfer learning model, the method further includes: Based on the component information of different types of components in the BIM model containing the erroneous identification target, semantic features, spatial features, and topological features between various types of components are extracted; Based on the semantic features, spatial features, and topological features, an error graph structure representation of the BIM model containing the misidentified target is generated, and the error graph structure representation is used to construct a transfer dataset suitable for a pre-trained transfer learning model. The pre-trained transfer learning model is trained using the transfer dataset until the preset transfer conditions are met, so as to obtain the trained transfer learning model.

7. A device for graphical representation of BIM models and error identification of models, characterized in that, Applied to the model training phase, wherein the apparatus includes: The first extraction module is used to extract semantic features, spatial features and topological features between different types of components based on the component information of different types of components in the BIM model; The first construction module is used to generate a graph structure representation of the BIM model based on the semantic features, spatial features and topological features, and to use the graph structure representation to construct a training dataset suitable for a pre-trained graph neural network model. The first training module is used to train the pre-trained graph neural network model using the training dataset to obtain a trained model, and to perform transfer learning using the trained model to obtain a transfer learning model, so as to use the transfer learning model to identify the misidentified targets of the BIM model.

8. The apparatus according to claim 7, characterized in that, The first extraction module includes: The first extraction unit is used to extract the semantic features based on the component feature information of each component; The second extraction unit is used to extract the spatial features based on the position information of each component and the local spatial relationship between adjacent components; The third extraction unit is used to extract the topological features based on the nesting relationship and connection relationship between different components, as well as the vertical relationship between the different components and the ground.

9. The apparatus according to claim 7, characterized in that, The first building module includes: The first generation unit is used to convert the different types of components into graph nodes in the graph structure representation, and to obtain the node attributes of the graph nodes based on the semantic features; The second generation unit is used to obtain graph edges in the graph structure representation based on the spatial features or the topological features, and to obtain the spatial edge attributes of the graph edges corresponding to the spatial features based on the local spatial relationships between adjacent components. The acquisition unit, based on the nesting, connection, and vertical relationships in the topological features, acquires the topological edge attributes of the graph edges corresponding to the topological features; The third generation unit is used to obtain the graph structure representation based on at least one of the graph nodes, the node attributes, the graph edges, the spatial edge attributes, and the topological edge attributes.

10. The apparatus according to claim 7, characterized in that, The first training module includes: The fourth generation unit is used to generate a corresponding occlusion dataset using the training dataset; The fifth generation unit is used to input the occlusion dataset into the encoder of the pre-trained graph neural network model, so as to use the encoder to generate a first high-dimensional representation of the graph nodes and / or a second high-dimensional representation of the edge attributes in the occlusion dataset. The sixth generation unit is used to input the first high-dimensional representation and / or the second high-dimensional representation into the decoder of the pre-trained graph neural network model, so as to use the decoder to reconstruct the occluded graph nodes and occluded graph edges in the occluded dataset, so as to obtain the reconstructed graph nodes and reconstructed graph edges. The seventh generation unit is used to obtain the trained model based on the training dataset, the occlusion dataset, the reconstructed graph nodes, and the reconstructed graph edges.

11. A graphical representation of a BIM model and a device for identifying model errors, characterized in that, Applied to the model transfer learning stage, wherein the apparatus includes: The second extraction module is used to extract semantic features, spatial features and topological features between different types of components based on the component information of different types of components in the target BIM model; The generation module is used to generate a target graph structure representation of the target BIM model based on the semantic features, spatial features, and topological features. The identification module is used to input the target graph structure representation into a pre-trained transfer learning model to identify erroneous target identification results of the target BIM model, wherein the pre-trained transfer learning model is obtained by transfer learning from the trained model.

12. The apparatus according to claim 11, characterized in that, Also includes: The third extraction module is used to extract semantic features, spatial features and topological features among different types of components based on component information of different types of components in the BIM model containing misidentified targets before inputting the target graph structure representation into the pre-trained transfer learning model. The second construction module is used to generate an error graph structure representation of the BIM model containing the misidentified target based on the semantic features, spatial features and topological features, and to construct a transfer dataset suitable for the pre-trained transfer learning model using the error graph structure representation. The second training module is used to train the pre-trained transfer learning model using the transfer dataset until the preset transfer conditions are met, so as to obtain the trained transfer learning model.

13. An electronic device, characterized in that, include: The system includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the graphical representation of a BIM model and the method for identifying model errors as described in any one of claims 1-4 or any one of claims 5-6.

14. A computer-readable storage medium having a computer program stored thereon, characterized in that, The program is executed by the processor to implement the graphical representation and model error identification method of the BIM model as described in any one of claims 1-4 or the graphical representation and model error identification method of the BIM model as described in any one of claims 5-6.

15. A computer program product, characterized in that, Includes a computer program, which, when executed, is used to implement the graphical representation and model error identification method for BIM models as described in any one of claims 1-4 or the graphical representation and model error identification method for BIM models as described in any one of claims 5-6.

Images