A collaborative design method and system for bridge steel structures based on BIM technology

Abstract

This invention relates to the field of Building Information Modeling (BIM) and intelligent design technology, and provides a method and system for collaborative design of bridge steel structures based on BIM technology. The method includes: acquiring multiple bridge steel structure BIM models and their design parameter vectors, structural topology diagrams, and collaborative design data vectors from a BIM database; constructing a structural feature extraction model by training a BIM structural encoder model, calculating the similarity between models, and retrieving similar models; establishing a BIM data fusion model, training it by combining the design parameter vectors of the target model and the collaborative design data vectors of its similar models to obtain a collaborative design prediction model; for the model to be collaboratively designed, first retrieving similar models, and then generating collaborative design data through the prediction model based on its design parameters and the collaborative data of similar models. This invention can achieve intelligent collaborative optimization of bridge steel structure design, improve design efficiency and accuracy, and reduce manual intervention and repetitive work.

Description

Technical Field

[0001] This invention relates to the field of building information modeling and intelligent design technology, and more specifically, to a collaborative design method and system for bridge steel structures based on BIM technology. Background Technology

[0002] In the Architecture, Engineering, and Construction (AEC) industry, the design of bridge steel structures is a complex and multidisciplinary process. Traditional bridge steel structure design relies primarily on two-dimensional drawings and manual coordination, which is not only inefficient but also prone to inconsistencies and communication breakdowns. With the advent of Building Information Modeling (BIM) technology, designers can integrate the geometric information, spatial relationships, geographic information, and component attribute information of the bridge steel structure into a three-dimensional digital model, thereby improving design efficiency and quality. However, current BIM technology still has some limitations in collaborative design of bridge steel structures. Existing BIM collaborative design methods mainly focus on information sharing and communication between different disciplines, but for the collaborative design of complex structures like bridge steel structures, especially in parametric collaboration and intelligent model matching during the design phase, effective technical support is lacking. Existing collaborative design systems often require designers to manually compare models and adjust parameters, which is not only time-consuming and labor-intensive but also prone to errors. Furthermore, for projects like bridge steel structures that require multidisciplinary collaborative design, existing technologies struggle to quickly evaluate and optimize different design schemes, leading to extended design cycles and increased costs.

[0003] In implementing the embodiments of the present invention, the inventors discovered at least the following problems or defects in the prior art: First, existing BIM collaborative design methods cannot effectively extract features and calculate similarity for the complex topological structure of bridge steel structures, making it difficult to achieve intelligent model matching and collaborative design. Second, existing technologies lack effective data fusion and prediction mechanisms when processing collaborative design data, failing to fully utilize existing design experience and data to generate new collaborative design schemes. Finally, existing BIM collaborative design systems still need improvement in user interaction and automation, failing to meet the demands of modern bridge steel structure design for efficient collaboration and rapid iteration. Summary of the Invention

[0004] This invention provides a collaborative design method and system for bridge steel structures based on BIM technology.

[0005] In a first aspect of the present invention, a collaborative design method for bridge steel structures based on BIM technology is provided, comprising:

[0006] Obtain multiple completed bridge steel structure BIM models from the BIM engineering database, and obtain the BIM design parameter vectors, BIM structural topology diagrams, and collaborative design data vectors for each model.

[0007] A BIM structure encoder model is established, and the BIM structure encoder model is trained based on the multiple BIM structure topology maps to obtain a structural feature extraction model.

[0008] Based on the structural feature extraction model, the similarity index between each pair of bridge steel structure BIM models is calculated, and bridge steel structure BIM models that meet the conditions are obtained as similar models.

[0009] Establish a BIM data fusion model and train the BIM data fusion model to obtain a collaborative design prediction model;

[0010] Obtain the BIM model of the bridge steel structure to be collaboratively designed. Based on the BIM model to be collaboratively designed, use the similar model to input the BIM design parameter vector of the BIM model to be collaboratively designed and the collaborative design data vector of several similar models corresponding to the model as inputs into the collaborative design prediction model for processing, and output the collaborative design data of the BIM model to be collaboratively designed.

[0011] Furthermore, each of the bridge steel structure BIM models includes: a BIM design parameter module, a BIM structural topology module, and a collaborative design data module; wherein, the BIM design parameter module includes: multiple structural components contained in the model, the BIM parametric attribute values ​​of each component, and the BIM annotation information contained in the model.

[0012] Furthermore, obtaining the BIM design parameter vector includes:

[0013] The BIM design parameter module is standardized using a BIM data preprocessing tool.

[0014] The standardized design parameters are converted into the BIM design parameter vector using the BIM parameter encoding method.

[0015] Furthermore, obtaining the collaborative design data vector includes:

[0016] The collaborative design data module is processed using a BIM information cleaning tool to standardize the data.

[0017] Obtain the BIM collaborative data template of the engineering category to which the bridge steel structure belongs, and use the sequence alignment algorithm to calculate the matching degree between the standardized collaborative data and the template;

[0018] When the matching degree reaches or exceeds a set threshold, the redundant part of the normalized collaborative data that matches the template is removed;

[0019] The collaborative data, after removing redundant parts, is converted into the collaborative design data vector using BIM data encoding.

[0020] Furthermore, the processing of the BIM design parameter module or the collaborative design data module using BIM data preprocessing tools or BIM information cleaning tools includes:

[0021] Remove unstructured BIM labels and format control symbols;

[0022] Clear temporary markers from the design phase;

[0023] Filter statements containing special engineering symbols, unit identifiers, or component codes;

[0024] Tolerance annotations and version identifiers in unified parameters;

[0025] Remove temporary change parameters from the construction detailing phase;

[0026] Clear the component attribute parameters added after the design loop is closed;

[0027] Remove redundant labels from component categories.

[0028] Furthermore, generating the BIM structural topology diagram includes:

[0029] The initial BIM model and the revised BIM model are respectively parsed into parametric topologies;

[0030] The BIM change detection algorithm is used to calculate the topological node mapping relationship between the initial and revised BIM models;

[0031] Based on the mapping relationship, the change operation type of the BIM model is determined, and topological feature nodes are generated according to the change type. The topological feature nodes are combined into a three-dimensional topological diagram according to the hierarchical structure of the initial BIM model to form a BIM structural topological diagram. The change operation types include geometric adjustment, component removal, component addition, position offset, and topology preservation.

[0032] Further, the establishment of the BIM structure encoder model, and the training of the BIM structure encoder model based on the multiple BIM structure topology maps to obtain a structural feature extraction model, includes:

[0033] The encoder uses a BIM graph neural network to aggregate the geometric attributes of each node in the structural topology graph from bottom to top to form an encoding tensor.

[0034] The decoder uses a BIM model to reconstruct the network, restoring the encoded tensor to a three-dimensional topological structure from top to bottom.

[0035] Based on the multiple BIM structure topology maps, a topology reconstruction training strategy is used to enable the three-dimensional topology to completely reproduce the input BIM structure topology map.

[0036] Furthermore, the calculation of the pairwise similarity index between the bridge steel structure BIM models includes:

[0037] The BIM structure encoder model is used to encode two BIM models to obtain two corresponding feature tensors. The structural feature similarity of the two feature tensors is calculated, and the similarity index of the two BIM models is determined based on the numerical value of the structural feature similarity.

[0038] Furthermore, training the BIM data fusion model includes:

[0039] The parameter encoder uses a BIM feature encoder to encode the BIM design parameter vector of each BIM model into a first feature vector;

[0040] The collaborative data encoder uses a BIM context encoder to encode the collaborative design data vectors of several similar models corresponding to each BIM model into a second feature vector;

[0041] The second feature vector is weighted according to the corresponding similarity index, and then tensor-concatenated with the first feature vector through the BIM feature fusion layer to form a fused feature vector;

[0042] The predictive decoder uses a BIM data generation network, combined with an attention mechanism and a gated recurrent unit, to decode the fused feature vector into collaborative design data for the BIM model to be generated.

[0043] In a second aspect of the invention, a collaborative design system for bridge steel structures based on BIM technology is provided, comprising:

[0044] The BIM data integration module is used to obtain bridge steel structure BIM models from the BIM engineering database, and to obtain BIM design parameter vectors, BIM structural topology diagrams and collaborative design data vectors for each model.

[0045] The BIM feature comparison module is used to establish a BIM structure encoder model, train the BIM structure encoder model based on multiple BIM structure topology maps to obtain a structural feature extraction model; and calculate the similarity index between each pair of bridge steel structure BIM models based on the structural feature extraction model to obtain bridge steel structure BIM models that meet the conditions as similar models.

[0046] The BIM model training module is used to establish a BIM data fusion model and train the BIM data fusion model to obtain a collaborative design prediction model.

[0047] The BIM collaborative generation module is used to obtain the BIM model of the bridge steel structure to be collaboratively designed. Based on the BIM model to be collaboratively designed, the module uses the BIM design parameter vector of the BIM model to be collaboratively designed and the collaborative design data vector of several similar models corresponding to the model as inputs to the collaborative design prediction model for processing, and outputs the collaborative design data of the BIM model to be collaboratively designed.

[0048] The embodiments of the present invention have at least the following beneficial effects: The BIM-based collaborative design method and system for bridge steel structures of the present invention can improve the efficiency and accuracy of collaborative design for bridge steel structures. By establishing a BIM structural encoder model and training a structural feature extraction model, accurate feature extraction and similarity calculation of the topological structure of the bridge steel structure BIM model can be performed, thereby quickly retrieving several models with the highest similarity to the model to be designed, providing valuable references for designers, reducing repetitive design work, and shortening the design cycle. Simultaneously, the construction of the BIM data fusion model and the generation of the collaborative design prediction model can organically combine the design parameters of the model to be collaboratively designed with the collaborative design data of similar models, using advanced algorithms to generate high-quality collaborative design data, further improving the scientific nature and reliability of the design and reducing design risks.

[0049] Furthermore, this invention enables efficient processing and standardized management of BIM design parameters and collaborative design data. Through BIM data preprocessing and cleaning tools, design parameters and collaborative design data can be standardized and regulated, removing redundant information and temporary tags to ensure data accuracy and consistency, providing a high-quality data foundation for subsequent model training and collaborative design. Overall, this invention provides an intelligent and automated solution for the collaborative design of bridge steel structures, contributing to the technological advancement and development of the bridge steel structure design industry. Attached Figure Description

[0050] The above and other objects, features, and advantages of exemplary embodiments of the present invention will become readily apparent from the following detailed description taken in conjunction with the accompanying drawings. Several embodiments of the invention are illustrated in the drawings by way of example and not limitation, wherein:

[0051] Figure 1 A flowchart illustrating a collaborative design method for bridge steel structures based on BIM technology, provided in an embodiment of the present invention.

[0052] Figure 2 This is a schematic diagram of a collaborative design system for bridge steel structures based on BIM technology, provided in an embodiment of the present invention.

[0053] Figure 3 A schematic diagram of the structure of an electronic device according to an embodiment of the present invention is shown. Detailed Implementation

[0054] The principles and spirit of the invention will now be described with reference to several exemplary embodiments. It should be understood that these embodiments are provided merely to enable those skilled in the art to better understand and implement the invention, and are not intended to limit the scope of the invention in any way. Rather, these embodiments are provided to make the invention more thorough and complete, and to fully convey the scope of the invention to those skilled in the art.

[0055] Those skilled in the art will understand that embodiments of the present invention can be implemented as a system, apparatus, device, method, or computer program product. Therefore, the present invention can be specifically implemented in the following forms: entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0056] It should be noted that the number of any elements in the accompanying drawings is for illustrative purposes only and not as a limitation, and any naming is for distinction only and has no limiting meaning.

[0057] The following is for reference. Figure 1 , Figure 1 This is a flowchart illustrating a collaborative design method for bridge steel structures based on BIM technology, provided in an embodiment of the present invention. Figure 1 As shown, a collaborative design method for bridge steel structures based on BIM technology includes:

[0058] S1. Obtain multiple completed bridge steel structure BIM models from the BIM engineering database, and obtain the BIM design parameter vector, BIM structural topology diagram and collaborative design data vector for each model.

[0059] S2. Establish a BIM structure encoder model, and train the BIM structure encoder model based on the multiple BIM structure topology maps to obtain a structural feature extraction model.

[0060] S3. Based on the structural feature extraction model, calculate the similarity index between each pair of bridge steel structure BIM models, and obtain the bridge steel structure BIM models that meet the conditions as similar models.

[0061] S4. Establish a BIM data fusion model and train the BIM data fusion model to obtain a collaborative design prediction model;

[0062] S5. Obtain the BIM model of the bridge steel structure to be collaboratively designed. Based on the BIM model to be collaboratively designed, use the similar model to input the BIM design parameter vector of the BIM model to be collaboratively designed and the collaborative design data vector of several similar models corresponding to the model as inputs to the collaborative design prediction model for processing, and output the collaborative design data of the BIM model to be collaboratively designed.

[0063] It should be noted that this invention proposes a collaborative design method for bridge steel structures based on BIM technology. BIM, or Building Information Modeling, is a digital technology that integrates information throughout the entire lifecycle of a building project. It optimizes the design, construction, and operation processes by creating and using three-dimensional models. In this invention, multiple completed bridge steel structure BIM models are obtained from a BIM engineering database. This step aims to collect a sufficient number of design samples for subsequent model training and similarity analysis. The BIM model contains rich information, such as BIM design parameter vectors, BIM structural topology diagrams, and collaborative design data vectors. The BIM design parameter vectors refer to the set of parametric attribute values ​​for each structural component in the model. These parametric attribute values ​​can include the component's dimensions, material properties, connection methods, etc. The BIM structural topology diagram describes the geometric structure and spatial relationships of the bridge steel structure, reflecting the connections and layout between components. The collaborative design data vectors record data generated during the collaborative design process, such as design change records and collaborative operation records. This data helps in understanding and optimizing the collaborative design process.

[0064] The encoder model encodes information from the BIM structural topology map, extracting encoded tensors that characterize structural features. The decoder then reconstructs the encoded tensors into a three-dimensional topological structure, achieving a complete reproduction of the input BIM structural topology map. During the training of the BIM structural encoder model, a topology reconstruction training strategy is employed, using a large number of BIM structural topology map samples to enable the model to learn the feature representations of different structural topologies. The similarity index between pairwise bridge steel structure BIM models is calculated by encoding two BIM models using the BIM structural encoder model to obtain two corresponding feature tensors, and then calculating the structural feature similarity between these two feature tensors. The similarity index reflects the degree of similarity between two BIM models in terms of structural features; a higher value indicates greater similarity. In this invention, by setting a threshold, several similar models whose similarity to the BIM model to be collaboratively designed meets the set threshold can be selected. These models will serve as reference models for subsequent collaborative design predictions.

[0065] For the construction of the BIM data fusion model, the parametric encoder uses a BIM feature encoder, which encodes the BIM design parameter vectors of each bridge steel structure BIM model into a first feature vector. The collaborative data encoder uses a BIM context encoder to encode the collaborative design data vectors of several similar models corresponding to the model into a second feature vector. When fusing these two feature vectors, the second feature vector is weighted according to the corresponding similarity index, and then tensor-concatenated with the first feature vector through the BIM feature fusion layer to form a fused feature vector. The predictive decoder uses a BIM data generation network, combined with an attention mechanism and a gated recurrent unit, to decode the fused feature vector into collaborative design data for the BIM model to be generated. In practical applications, the input parameters of the BIM data fusion model can include specific values ​​of the BIM design parameter vectors, such as component dimensions and material properties, as well as information such as design change records in the collaborative design data vectors. Through this model construction and data processing method, high-quality collaborative design data can be effectively generated, providing support for the collaborative design of bridge steel structures.

[0066] The collaborative design prediction model analyzes and processes the input BIM design parameter vector and collaborative design data vector, ultimately outputting collaborative design data suitable for the BIM model to be collaboratively designed. First, using the aforementioned structural feature extraction model, feature extraction and similarity calculation are performed on the BIM model to be collaboratively designed. By comparing it with historical BIM models in the database, several similar models whose similarity to the model to be collaboratively designed meets a set threshold are retrieved. These similar models have high structural similarity to the model to be designed and can provide valuable references for the current design. The similarity threshold is a pre-set parameter used to measure the degree of similarity between the model to be collaboratively designed and historical models. Only historical models with a similarity reaching or exceeding this threshold are selected as similar models for subsequent collaborative design data generation.

[0067] Next, the BIM design parameter vector of the BIM model to be collaboratively designed will be used as part of the input. This vector represents the parametric attributes of each structural component in the current model, such as dimensions, material type, and connection method, and is a digital expression of the model's design intent and structural characteristics. Simultaneously, the collaborative design data vectors of several similar models retrieved will be used as another part of the input. These collaborative design data vectors record information such as the collaborative strategies used in the design of similar structures, design change records, and details of collaboration between multiple disciplines.

[0068] The collaborative design prediction model processes the two sets of input data mentioned above. Internally, the model may contain multiple processing layers, such as a parametric encoder, a collaborative data encoder, a feature fusion layer, and a predictive decoder. The parametric encoder extracts features from the BIM design parameter vectors, while the collaborative data encoder extracts features from the collaborative design data vectors of similar models. The feature fusion layer integrates the extracted features to form a fused feature vector. Finally, the predictive decoder decodes the fused feature vector to generate collaborative design data for the BIM model to be collaboratively designed.

[0069] In some embodiments, each bridge steel structure BIM model includes: a BIM design parameter module, a BIM structural topology module, and a collaborative design data module; wherein, the BIM design parameter module includes: multiple structural components contained in the model, BIM parametric attribute values ​​of each component, and BIM annotation information contained in the model.

[0070] It should be noted that each bridge steel structure BIM model mentioned in this invention includes a BIM design parameter module, a BIM structural topology module, and a collaborative design data module. The BIM design parameter module is one of the core parts of the model, containing detailed information about each structural component. This information includes not only the geometric dimensions of the components, such as length, width, and height, but also the BIM parametric attribute values ​​of each component, such as material type, load-bearing capacity, and connection method. In addition, the model also includes BIM annotation information, which records design intent, construction requirements, or other important design specifications, providing additional reference for designers and construction personnel. The BIM structural topology module mainly describes the geometric structure and spatial relationships of the bridge steel structure, displaying the connection methods and overall layout between components in the form of a topology diagram. The collaborative design data module records various data generated during the collaborative design process, which is crucial for understanding design changes, optimizing the design process, and improving design efficiency.

[0071] Specifically, the BIM parametric attribute values ​​in the BIM design parameter module are key information for component design. These attribute values ​​can include material properties of the components, such as the strength grade and yield strength of the steel; geometric properties, such as the cross-sectional shape and dimensions of the components; and connection properties, such as welding methods and bolt connection specifications. These parametric attribute values ​​are stored in the BIM model in a standardized format, facilitating querying and modification by designers. The BIM structural topology module visualizes the geometric structure and spatial relationships of the bridge steel structure through a topology diagram. Nodes in the topology diagram represent components, and edges represent the connections between components. This diagram structure clearly shows the layout and connection methods of the entire bridge steel structure. The collaborative design data module records various data during the collaborative design process, such as design change records and collaborative operation records. This data is stored in time-series format, allowing designers to trace every step and change in the design process.

[0072] Preferably, parametric modeling tools can be used to better construct and manage BIM design parameter modules. These tools allow designers to quickly generate and modify component models by inputting parametric attribute values. For example, designers can automatically generate corresponding component models by inputting parameters such as the length, width, and material type of a component, and can easily adjust these parameters to meet different design requirements. In the construction of the BIM structural topology module, professional BIM modeling software such as Revit or ArchiCAD can be used. These software programs provide powerful topology modeling functions, automatically generating topology diagrams based on design parameters and supporting the editing and modification of the topology. In the management of the collaborative design data module, BIM collaboration platforms such as BIM360 can be utilized. These platforms can record various operations and changes during the design process in real time and provide version control functions to ensure the integrity and consistency of design data. Through these specific technical means, the construction efficiency and quality of bridge steel structure BIM models can be effectively improved, providing a solid foundation for subsequent collaborative design.

[0073] In some embodiments, obtaining the BIM design parameter vector includes:

[0074] The BIM design parameter module is standardized using a BIM data preprocessing tool.

[0075] The standardized design parameters are converted into the BIM design parameter vector using the BIM parameter encoding method.

[0076] It should be noted that the process of obtaining BIM design parameter vectors mentioned in this invention involves standardizing and encoding the information in the BIM design parameter module to facilitate subsequent model training and collaborative design. The BIM data preprocessing tool is a software tool specifically designed for processing BIM data; it can clean and standardize the data in the BIM design parameter module. The BIM parameter encoding method converts the standardized design parameters into a vector form suitable for model processing. The purpose of this process is to ensure that the data format input into the model is uniform and accurate, thereby improving the model's training effect and the reliability of collaborative design.

[0077] Specifically, the data in the BIM design parameter module typically includes various types of parameters, such as geometric parameters of components (length, width, height, etc.), material parameters (strength, modulus of elasticity, etc.), and connection parameters (connection type, connection dimensions, etc.). The role of BIM data preprocessing tools is to standardize these parameters, for example, converting dimensional parameters in different units to the same unit, such as millimeters, and standardizing material parameter descriptions to standard material property codes. BIM parameter encoding methods convert these standardized parameters into numerical vectors using specific encoding rules, such as one-hot encoding or embedding, to transform discrete parameter values ​​into continuous numerical vectors. These vectors can then be used as input to machine learning or deep learning models for subsequent feature extraction and collaborative design prediction.

[0078] Preferably, BIM data preprocessing tools can be integrated into BIM modeling software, such as adding preprocessing plugins to Revit or ArchiCAD. This allows designers to directly standardize parameters during the design process. During parameter encoding, appropriate encoding methods can be selected for different parameter types. For example, normalization can be directly used for continuous geometric parameters; for discrete material type parameters, embedding encoding can be used, mapping each material type to a low-dimensional vector space. Furthermore, to improve encoding efficiency and accuracy, pre-trained encoder models can be utilized. These models have been trained on large amounts of BIM data and can automatically learn the optimal encoding method for parameters. Through these specific technical means, the generation process of BIM design parameter vectors can be ensured to be efficient and accurate, providing high-quality data support for subsequent collaborative design.

[0079] In some embodiments, obtaining the collaborative design data vector includes:

[0080] The collaborative design data module is processed using a BIM information cleaning tool to standardize the data.

[0081] Obtain the BIM collaborative data template of the engineering category to which the bridge steel structure belongs, and use the sequence alignment algorithm to calculate the matching degree between the standardized collaborative data and the template;

[0082] When the matching degree reaches or exceeds a set threshold, the redundant part of the normalized collaborative data that matches the template is removed;

[0083] The collaborative data, after removing redundant parts, is converted into the collaborative design data vector using BIM data encoding.

[0084] It should be noted that the process of obtaining collaborative design data vectors mentioned in this invention involves standardizing and encoding the information in the collaborative design data module for subsequent model training and collaborative design. The core of this process is to ensure the consistency and accuracy of the collaborative design data while removing redundant information and improving data usability. The BIM information cleaning tool is a software tool specifically designed for processing BIM data; it can clean and standardize the data in the collaborative design data module. The sequence alignment algorithm is an algorithm used to calculate the matching degree between the standardized collaborative data and the template, thereby identifying and removing redundant parts of the data. The BIM data encoding method converts the processed collaborative data into a vector form suitable for model processing, facilitating subsequent collaborative design prediction.

[0085] Specifically, the data in the collaborative design data module typically includes design change records, collaborative operation records, design review comments, and other information. During the generation process, this data may contain inconsistencies in format and redundant information, requiring processing using BIM information cleaning tools. The main functions of these cleaning tools include deleting unstructured BIM tags, removing temporary markers from the design phase, and filtering statements containing special engineering symbols to ensure data accuracy and consistency. Sequence alignment algorithms are used to compare the normalized collaborative data with predefined BIM collaborative data templates and calculate their matching degree. When the matching degree reaches or exceeds a set threshold, it indicates a high degree of consistency between the data and the template; at this point, redundant parts matching the template can be removed. BIM data encoding methods convert the processed collaborative data into numerical vectors. For example, embedded encoding or one-hot encoding can be used to convert discrete data values ​​into continuous numerical vectors for use as model input.

[0086] Preferably, BIM information cleaning tools can be integrated into BIM collaborative design platforms, such as adding cleaning plugins to platforms like BIM360 or Navisworks. This allows designers to directly clean and standardize data during the collaborative design process. In the application of sequence alignment algorithms, specific matching thresholds can be set, such as 80% or higher, to ensure data accuracy and consistency. For BIM data encoding methods, appropriate encoding methods can be selected based on the characteristics of the collaborative design data. For example, textual embedding encoding can be used for textual data in design change records, converting the text into a low-dimensional vector representation; time-series encoding can be used for time-series data in collaborative operation records, combining time information and operation type information for encoding. Through these specific technical means, the generation process of collaborative design data vectors can be ensured to be efficient and accurate, providing high-quality data support for subsequent collaborative design.

[0087] In some embodiments, processing the BIM design parameter module or the collaborative design data module using a BIM data preprocessing tool or a BIM information cleaning tool includes:

[0088] Remove unstructured BIM labels and format control symbols;

[0089] Clear temporary markers from the design phase;

[0090] Filter statements containing special engineering symbols, unit identifiers, or component codes;

[0091] Tolerance annotations and version identifiers in unified parameters;

[0092] Remove temporary change parameters from the construction detailing phase;

[0093] Clear the component attribute parameters added after the design loop is closed;

[0094] Remove redundant labels from component categories.

[0095] It should be noted that the process of processing BIM design parameter modules or collaborative design data mentioned in this invention is to ensure the accuracy and consistency of the data, thereby improving the efficiency and quality of subsequent collaborative design. Through BIM data preprocessing tools or BIM information cleaning tools, a series of cleaning and standardization operations can be performed on the data, removing redundant information and unnecessary labels, making the data more suitable for model training and collaborative design. The application of these tools and methods can effectively reduce noise in the data, improve data usability, and provide a solid foundation for subsequent model building and prediction.

[0096] Specifically, the functions of BIM data preprocessing tools and BIM information cleaning tools include: deleting unstructured BIM labels and format control symbols, which may be used for temporary marking or format adjustments during the design process but have no practical meaning in the final collaborative design; removing temporary marking information from the design phase, which is usually used for annotation and reminders during the design process but does not need to be retained in the final model; filtering statements containing special engineering symbols, unit identifiers, or component codes, which may be used for specific engineering annotations during the design process but may cause confusion in collaborative design; unifying tolerance annotations and version identifiers in parameters to ensure that all parameters are annotated in a consistent way, facilitating subsequent processing and comparison; removing temporary change parameters from the construction detailing stage, which may be added during the construction stage but are not needed in the design stage; removing component attribute parameters added after design closure, which may be added after the design is completed but may not be needed in collaborative design; and removing redundant component classification labels to ensure that each component's classification label is unique and accurate, avoiding duplication and confusion.

[0097] Preferably, BIM data preprocessing tools can be integrated into BIM modeling software, such as adding a preprocessing plugin to Revit or ArchiCAD. This allows designers to directly standardize parameters during the design process. In practice, tool parameters can be set, such as specifying the types of unstructured labels to be deleted and setting a uniform format for tolerance annotations. For BIM information cleaning tools, they can be integrated into BIM collaborative design platforms, such as adding a cleaning plugin to BIM360 or Navisworks, to clean data in real time during collaborative design. During data processing, specific cleaning rules can be set, such as filtering specific engineering symbols and removing specific temporary markers. These specific technical means ensure efficient and accurate data processing, providing high-quality data support for subsequent collaborative design.

[0098] In some embodiments, generating the BIM structural topology diagram includes:

[0099] The initial BIM model and the revised BIM model are respectively parsed into parametric topologies;

[0100] The BIM change detection algorithm is used to calculate the topological node mapping relationship between the initial and revised BIM models;

[0101] Based on the mapping relationship, the change operation type of the BIM model is determined, and topological feature nodes are generated according to the change type. The topological feature nodes are combined into a three-dimensional topological diagram according to the hierarchical structure of the initial BIM model to form a BIM structural topological diagram. The change operation types include geometric adjustment, component removal, component addition, position offset, and topology preservation.

[0102] It should be noted that the process of generating a BIM structural topology map mentioned in this invention aims to present the geometric structure and spatial relationships of the bridge steel structure in a clear and operable form. This process involves parsing the initial BIM model and the revised BIM model into parametric topologies, and calculating the topology node mapping relationships between them using a BIM change detection algorithm. The determination of the topology node mapping relationship is based on the change operation type, including geometric adjustments, component removal, component addition, position offsets, and topology preservation. These change operation types reflect the modifications and optimizations of the BIM model during the design process. Through these steps, a three-dimensional topology map can be generated, which can fully reflect the structural features and change history of the BIM model, providing an important reference for subsequent collaborative design.

[0103] Specifically, the initial BIM model refers to the model created in the early stages of design, while the revised BIM model refers to the model modified during the design process. Parsing these two models into parametric topologies means extracting the geometric parameters and spatial relationships of each component in the model and representing them in a parametric form, facilitating subsequent processing and analysis. The BIM change detection algorithm is used to compare the differences between two BIM models. It determines which parts of the model have changed by calculating the topological node mapping relationship between the initial and revised models. Change operation types are specific descriptions of model changes. For example, geometric adjustment refers to a change in the shape or size of a component; component removal refers to the deletion of a component from the model; component addition refers to the addition of a new component to the model; position offset refers to the movement of a component's position; and topology preservation indicates that the model's topology has not changed. These change operation types provide designers with a clear way to understand and record the model's change history.

[0104] Preferably, the process of generating a BIM structural topology diagram can be further refined into the following steps: First, using BIM modeling software, such as Revit or ArchiCAD, the initial BIM model and the revised BIM model are exported as an intermediate format of parametric topology, such as IFC format. Then, a specialized BIM change detection tool, such as Solibri Model Checker, is used to calculate the topological node mapping relationship between the two models. During the calculation process, algorithm parameters can be set, such as defining a threshold for geometric adjustments to determine when a component's shape or size change exceeds this threshold and is considered a geometric adjustment operation. For each change operation type, detailed change information can be recorded; for example, for a component removal operation, the identifier of the removed component and the removal time are recorded. Finally, based on these change operation types and mapping relationships, a BIM visualization tool, such as Navisworks, is used to combine the changed topology into a 3D topology diagram and integrate it with the hierarchical structure of the initial BIM model to form a complete BIM structural topology diagram. Through these specific technical means, the generation process of the BIM structural topology diagram can be ensured to be efficient and accurate, providing clear and detailed structural information for subsequent collaborative design.

[0105] In some embodiments, establishing a BIM structure encoder model and training the BIM structure encoder model based on the plurality of BIM structure topology maps to obtain a structural feature extraction model includes:

[0106] The encoder uses a BIM graph neural network to aggregate the geometric attributes of each node in the structural topology graph from bottom to top to form an encoding tensor.

[0107] The decoder uses a BIM model to reconstruct the network, restoring the encoded tensor to a three-dimensional topological structure from top to bottom.

[0108] Based on the multiple BIM structure topology maps, a topology reconstruction training strategy is used to enable the three-dimensional topology to completely reproduce the input BIM structure topology map.

[0109] It should be noted that the process of establishing a BIM structural encoder model and training a structural feature extraction model mentioned in this invention is to extract effective information that can characterize the features of the bridge steel structure from the BIM structural topology map. The encoder model aggregates the geometric attributes of nodes in the topology map to form an encoding tensor, while the decoder restores the encoding tensor to a three-dimensional topology. The core of this process lies in using a topology reconstruction training strategy to enable the model to completely reproduce the input BIM structural topology map, thereby providing accurate feature representations for subsequent similarity calculations and collaborative design. The BIM graph neural network is a neural network specifically designed for processing graph structure data. It can effectively capture the relationships and geometric attributes between nodes, while the BIM model reconstruction network is used to restore the extracted features to the original three-dimensional structure, ensuring the accuracy and completeness of feature extraction.

[0110] Specifically, the BIM structure encoder model consists of an encoder and a decoder. The encoder employs a BIM graph neural network, whose function is to perform bottom-up feature aggregation of the geometric attributes of each node in the BIM structure topology map. These geometric attributes include information such as the node's position, shape, and size; this information is integrated into an encoded tensor through the graph neural network's aggregation operation. The decoder employs a BIM model reconstruction network, whose task is to reconstruct the 3D topology from the encoded tensor from top to bottom. The topology reconstruction training strategy involves training the model with a large number of BIM structure topology map samples, enabling the decoder to accurately reconstruct the encoded tensor into a topology consistent with the input topology. Figure 1 The training process involves optimizing model parameters to minimize the difference between the input and reconstructed topological graphs.

[0111] Preferably, the process of constructing a BIM structure encoder model can be further refined into the following steps: First, collect a large number of BIM structure topology map samples, which should cover different types of bridge steel structures to ensure the model's generalization ability. Then, use a BIM graph neural network to encode the topology map of each sample. During the encoding process, the number of network layers and the number of nodes per layer can be set; for example, a multi-layer graph convolutional network can be used, and the number of nodes per layer can be adjusted according to the complexity of the topology map. Next, decode the encoded tensor through a BIM model reconstruction network to reconstruct the three-dimensional topology. During training, mean squared error (MSE) can be used as the loss function, and the model parameters can be optimized through backpropagation to minimize the difference between the input topology map and the reconstructed topology map.

[0112] Furthermore, to improve the training efficiency of the model, techniques such as batch normalization and early stopping mechanisms can be employed. These specific technical means ensure that the BIM structure encoder model can effectively extract structural features and provide high-quality feature representations for subsequent similarity calculations and collaborative design.

[0113] In some embodiments, calculating the pairwise similarity index between the bridge steel structure BIM models includes:

[0114] The BIM structure encoder model is used to encode two BIM models to obtain two corresponding feature tensors. The structural feature similarity of the two feature tensors is calculated, and the similarity index of the two BIM models is determined based on the numerical value of the structural feature similarity.

[0115] It should be noted that the process of calculating the pairwise similarity index between bridge steel structure BIM models mentioned in this invention is to quantify the structural similarity between different BIM models. This process involves encoding two BIM models using a BIM structural encoder model to obtain two corresponding feature tensors, and then calculating the structural feature similarity between these two feature tensors. The numerical value of the similarity index reflects the degree of similarity between two BIM models in terms of structural features; the higher the value, the more similar the two models are. This similarity calculation method provides an important basis for subsequent model retrieval and collaborative design, helping designers quickly find reference models similar to the model to be designed, thereby improving design efficiency and quality.

[0116] Specifically, the BIM structure encoder model encodes the BIM structural topology map, converting complex geometric structures and spatial relationships into feature tensors. A feature tensor is a multi-dimensional array that can compactly represent the structural features of the model. The calculation of structural feature similarity is accomplished by comparing two feature tensors. This calculation process can employ various similarity measurement methods, such as cosine similarity and Euclidean distance. Cosine similarity measures the similarity between two vectors by calculating the cosine of the angle between them; the closer the value is to 1, the higher the similarity. Euclidean distance measures the difference between two vectors by calculating the straight-line distance between them; the smaller the distance, the higher the similarity. In practical applications, an appropriate similarity measurement method can be selected based on specific design requirements.

[0117] The process of calculating similarity metrics can be further divided into the following steps: First, each BIM model is encoded using a trained BIM structure encoder model to obtain the feature tensor of each model. During encoding, topological data of the model, such as node coordinates and edge connections, can be input. Then, a suitable similarity metric, such as cosine similarity, is selected. When calculating the cosine similarity of two feature tensors, the dot product of the two tensors can be calculated first, then the norm of each tensor can be calculated separately, and finally the dot product can be divided by the product of the two norms. This yields a similarity value between 0 and 1. To improve computational efficiency, matrix operations can be used to calculate the similarity between multiple models simultaneously. Furthermore, a similarity threshold can be set according to actual needs, such as 0.8, to filter out reference models with a similarity higher than this threshold to the model to be designed. These specific technical means ensure that the similarity calculation process is efficient and accurate, providing a reliable reference for subsequent collaborative design.

[0118] In some embodiments, the BIM data fusion model includes:

[0119] The parameter encoder uses a BIM feature encoder to encode the BIM design parameter vector of each BIM model into a first feature vector;

[0120] The collaborative data encoder uses a BIM context encoder to encode the collaborative design data vectors of several similar models corresponding to each BIM model into a second feature vector;

[0121] The second feature vector is weighted according to the corresponding similarity index, and then tensor-concatenated with the first feature vector through the BIM feature fusion layer to form a fused feature vector;

[0122] The predictive decoder uses a BIM data generation network, combined with an attention mechanism and a gated recurrent unit, to decode the fused feature vector into collaborative design data for the BIM model to be generated.

[0123] The BIM data fusion model is a system used to integrate BIM design parameters and collaborative design data. Its purpose is to generate high-quality collaborative design data, thereby providing strong support for the collaborative design of bridge steel structures. This model utilizes multiple modules, including a parametric encoder, a collaborative data encoder, feature weighting, feature fusion, and a predictive decoder, to effectively integrate BIM design parameter vectors and collaborative design data vectors. It also employs attention mechanisms and gated loop units to improve the accuracy and efficiency of data fusion. This data fusion method can fully leverage existing design experience and data to generate collaborative design schemes that meet a set threshold of similarity to the model to be designed, thus improving design efficiency and quality.

[0124] Specifically, the parameter encoder in the BIM data fusion model employs a BIM feature encoder, which encodes the BIM design parameter vector of each bridge steel structure BIM model into a first feature vector. The BIM design parameter vector includes information such as the geometric parameters, material properties, and connection methods of each structural component in the model. The collaborative data encoder uses a BIM context encoder to encode the collaborative design data vectors of several similar models corresponding to each BIM model into a second feature vector. The collaborative design data vector contains information such as design change records and collaborative operation records. In the feature fusion stage, the second feature vector is weighted according to the corresponding similarity index, and then tensor-concatenated with the first feature vector through the BIM feature fusion layer to form a fused feature vector. The similarity index reflects the structural similarity between the reference model and the model to be designed. The predictive decoder uses a BIM data generation network, combined with an attention mechanism and a gated recurrent unit, to decode the fused feature vector into collaborative design data for the generated BIM model. The attention mechanism is used to highlight important features, while the gated recurrent unit is used to process sequential data, ensuring good spatiotemporal consistency of the generated collaborative design data.

[0125] The process of constructing a BIM data fusion model can include the following steps: First, define the structure and parameters of the BIM feature encoder. For example, a Multilayer Perceptron (MLP) or a Convolutional Neural Network (CNN) can be used as the architecture of the feature encoder, with input parameters being specific values ​​in the BIM design parameter vector, such as the length, width, and material strength of components. For the collaborative data encoder, a Recurrent Neural Network (RNN) or a Long Short-Term Memory (LSTM) can be used to process the sequential information in the collaborative design data vector, with input parameters being time-series data of design change records. During feature weighting, the second feature vector can be weighted according to the similarity index; for example, the higher the similarity, the greater the corresponding weight. The feature fusion layer can use a simple concatenation operation to merge the weighted second feature vector with the first feature vector into a fused feature vector. In the predictive decoder, parameters of the attention mechanism can be set, such as the number of attention heads and the dimension of the hidden layers, to highlight important feature information. Simultaneously, the parameters of the gated recurrent unit, such as the dimension of the hidden states and the type of gating function, also need to be adjusted according to the specific data characteristics. These specific technical means can ensure that the BIM data fusion model can effectively generate high-quality collaborative design data, providing efficient and accurate support for the collaborative design of bridge steel structures.

[0126] The various embodiments of the present invention have the following beneficial effects: The present invention can improve the intelligence level of bridge steel structure design. It automatically extracts structural features from historical models through a BIM structural encoder model and retrieves the most relevant reference cases based on similarity, enabling designers to quickly obtain validated optimization solutions. The BIM data fusion model can intelligently integrate current design parameters with collaborative data from similar models, generating the optimal collaborative design solution through feature weighting and attention mechanisms, effectively avoiding repetitive design work and improving design quality.

[0127] This method fully leverages historical project data for knowledge reuse, ensuring comparability between different projects through parameter standardization and topological feature extraction. 3D topological reconstruction based on graph neural networks accurately captures structural change characteristics, while sequence alignment algorithms and redundancy removal optimize the quality of collaborative data. The resulting collaborative design prediction model automatically generates collaborative data that conforms to engineering specifications, significantly shortening the design cycle and reducing the risk of human error, providing intelligent decision support for bridge steel structure design.

[0128] like Figure 2 As shown in some embodiments, a collaborative design system for bridge steel structures based on BIM technology is provided. The system includes:

[0129] BIM data integration module 201 is used to obtain bridge steel structure BIM models from BIM engineering databases, and obtain BIM design parameter vectors, BIM structural topology diagrams and collaborative design data vectors for each model.

[0130] BIM feature comparison module 202 is used to establish a BIM structure encoder model, train the BIM structure encoder model based on multiple BIM structure topology maps to obtain a structural feature extraction model; and calculate the similarity index between each pair of bridge steel structure BIM models based on the structural feature extraction model, and obtain bridge steel structure BIM models that meet the conditions as similar models.

[0131] BIM model training module 203 is used to establish a BIM data fusion model and train the BIM data fusion model to obtain a collaborative design prediction model.

[0132] BIM collaborative generation module 204 is used to obtain the bridge steel structure BIM model to be collaboratively designed. Based on the BIM model to be collaboratively designed, the BIM design parameter vector of the BIM model to be collaboratively designed and the collaborative design data vector of several similar models corresponding to the model are used as inputs to the collaborative design prediction model for processing, and the collaborative design data of the BIM model to be collaboratively designed are output.

[0133] It is understandable that the modules and references recorded in this BIM-based collaborative design system for bridge steel structures are... Figure 1 The steps described in the BIM-based collaborative design method for bridge steel structures correspond to those in the BIM-based collaborative design method for bridge steel structures. Therefore, the operations, features, and beneficial effects described above for the BIM-based collaborative design method for bridge steel structures also apply to the BIM-based collaborative design system for bridge steel structures and its constituent modules, and will not be repeated here.

[0134] The following is for reference. Figure 3 The diagram illustrates a structural schematic of an electronic device 300 suitable for implementing some embodiments of the present invention. The electronic devices in some embodiments of the present invention may include, but are not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), in-vehicle terminals (e.g., in-vehicle navigation terminals), and fixed terminals such as digital TVs and desktop computers. Figure 3 The terminal device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of the present invention.

[0135] like Figure 3 As shown, the electronic device 300 may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 301, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 302 or a program loaded from a storage device 308 into a random access memory (RAM) 303. The RAM 303 also stores various programs and data required for the operation of the electronic device 300. The processing unit 301, ROM 302, and RAM 303 are interconnected via a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0136] Typically, the following devices can be connected to I / O interface 305: input devices 306 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 307 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 308 including, for example, magnetic tapes, hard disks, etc.; and communication devices 309. Communication device 309 allows electronic device 300 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 3 An electronic device 300 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively. Figure 3 Each box shown can represent a device or multiple devices as needed.

[0137] Furthermore, the storage medium in the embodiments of this application stores program instructions capable of implementing all the above methods. These program instructions can be stored in the storage medium in the form of a software product, including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks, or terminal devices such as computers, servers, mobile phones, and tablets.

[0138] The above description is merely a selection of preferred embodiments of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention as described in the embodiments is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of the present invention.

Claims

1. A bridge steel structure collaborative design method based on BIM technology, characterized in that, The application comprises the following steps: Obtain a plurality of completed bridge steel structure BIM models from a BIM engineering database, obtain BIM design parameter vectors, BIM structure topology graphs, and collaborative design data vectors of each model; Establish a BIM structure encoder model, train the BIM structure encoder model based on the plurality of BIM structure topology graphs to obtain a structure feature extraction model; Based on the structure feature extraction model, calculate the similarity index between each two bridge steel structure BIM models, and obtain bridge steel structure BIM models meeting the conditions as similar models; Establish a BIM data fusion model, train the BIM data fusion model to obtain a collaborative design prediction model; Obtain a bridge steel structure BIM model to be collaboratively designed, based on the BIM model to be collaboratively designed, use the similar model to input the BIM design parameter vector of the BIM model to be collaboratively designed and the collaborative design data vector of a plurality of similar models corresponding to the model into the collaborative design prediction model for processing, and output the collaborative design data of the BIM model to be collaboratively designed; The calculation of the similarity index between each two bridge steel structure BIM models comprises the following steps: Encode two BIM models by using the BIM structure encoder model to obtain corresponding two feature tensors, calculate the structure feature similarity of the two feature tensors, and determine the similarity index of the two BIM models based on the numerical value of the structure feature similarity; Obtaining bridge steel structure BIM models meeting the conditions as similar models is achieved by setting a threshold value, which can filter out a plurality of similar models with a similarity to the BIM model to be collaboratively designed meeting the set threshold value, and these models will be used as reference models for subsequent collaborative design prediction.

2. The method of claim 1, wherein the method further comprises: Each bridge steel structure BIM model comprises a BIM design parameter module, a BIM structure topology module, and a collaborative design data module; wherein the BIM design parameter module comprises a plurality of structural members contained in the model, BIM parameterized attribute values of each member, and BIM annotation information contained in the model.

3. The method of claim 2, wherein the method further comprises: The BIM design parameter vector is obtained by: standardizing the BIM design parameter module by using a BIM data preprocessing tool; and converting the standardized design parameters into the BIM design parameter vector by using a BIM parameter encoding method.

4. The method of claim 3, wherein the method further comprises: The collaborative design data vector is obtained by: standardizing the collaborative design data module by using a BIM information cleaning tool; obtaining a BIM collaborative data template of the engineering category to which the bridge steel structure belongs, calculating the matching degree of the standardized collaborative data and the template by using a sequence alignment algorithm; removing the redundant part of the standardized collaborative data matched with the template when the matching degree reaches or exceeds a set threshold value; and converting the collaborative data after removing the redundant part into the collaborative design data vector by using a BIM data encoding method.

5. The method of claim 4, wherein the method further comprises: The processing of the BIM design parameter module or the collaborative design data module by using the BIM data preprocessing tool or the BIM information cleaning tool comprises: deleting unstructured BIM tags and format control symbols; clearing temporary mark information in the design stage; filtering sentences containing special engineering symbols, unit identifiers or component codes; unifying tolerance annotations and version identifiers in parameters; removing temporary change parameters in the construction development stage; clearing component attribute parameters supplemented after design closure; removing redundant component classification tags.

6. The method of claim 1, wherein the bridge steel structure collaborative design method is characterized by, generating the BIM structure topology graph comprises: parsing the initial BIM model and the revised BIM model into parameterized topology structures respectively; calculating the topology node mapping relationship of the initial and revised BIM models by using a BIM change detection algorithm; based on the mapping relationship, determining the change operation type of the BIM model, generating a topology feature node according to the change type, combining the topology feature node into a three-dimensional topology graph according to the hierarchical structure of the initial BIM model, and forming a BIM structure topology graph; wherein the change operation type includes geometric adjustment, component removal, component addition, position offset and topology preservation.

7. The method of claim 1, wherein the method further comprises: The BIM structure encoder model is trained based on the plurality of BIM structure topology graphs to obtain a structure feature extraction model, comprising: the encoder uses a BIM graph neural network to aggregate the geometric properties of each node in the structure topology graph from bottom to top to form an encoding tensor; the decoder uses a BIM model reconstruction network to restore the three-dimensional topology structure from top to bottom based on the encoding tensor; based on the plurality of BIM structure topology graphs, the three-dimensional topology structure can completely reproduce the input BIM structure topology graph through a topology reconstruction training strategy.

8. The method of claim 1, wherein the bridge steel structure collaborative design method is characterized by, Training the BIM data fusion model comprises: the parameter encoder uses a BIM feature encoder to encode the BIM design parameter vector of each bridge steel structure BIM model into a first feature vector; the collaborative data encoder uses a BIM context encoder to encode the collaborative design data vector of a plurality of similar models corresponding to each BIM model into a second feature vector; the second feature vector is weighted according to the corresponding similarity index, and the first feature vector is spliced into a fusion feature vector through a BIM feature fusion layer; the prediction decoder uses a BIM data generation network combined with an attention mechanism and a gated recurrent unit to decode the fusion feature vector into collaborative design data of the to-be-generated BIM model.

9. A bridge steel structure collaborative design system based on BIM technology, characterized in that, comprises: a BIM data integration module for obtaining bridge steel structure BIM models from a BIM engineering database, obtaining BIM design parameter vectors, BIM structure topology graphs and collaborative design data vectors for each model based on each model; a BIM feature comparison module for establishing a BIM structure encoder model, training the BIM structure encoder model based on a plurality of BIM structure topology graphs to obtain a structure feature extraction model; based on the structure feature extraction model, calculating the similarity index between the bridge steel structure BIM models, and obtaining the bridge steel structure BIM models that meet the conditions as similar models; The BIM model training module is configured to establish a BIM data fusion model, and train the BIM data fusion model to obtain a collaborative design prediction model. The BIM collaborative generation module is configured to obtain a bridge steel structure BIM model to be collaboratively designed, and based on the BIM model to be collaboratively designed, utilize the similar model to input a BIM design parameter vector of the BIM model to be collaboratively designed and collaborative design data vectors of a plurality of similar models corresponding to the model as inputs into the collaborative design prediction model for processing, and output collaborative design data of the BIM model to be collaboratively designed. The calculation of the similarity index between the bridge steel structure BIM models includes: utilizing the BIM structure encoder model to encode two BIM models to obtain corresponding two feature tensors, calculating the structural feature similarity of the two feature tensors, and determining the similarity index of the two BIM models based on the numerical value of the structural feature similarity. The bridge steel structure BIM model satisfying the condition is obtained as a similar model by setting a threshold value, and a plurality of similar models satisfying the set threshold value in similarity with the BIM model to be collaboratively designed can be screened out, and these models will be used as reference models for subsequent collaborative design prediction.

Images