BIM-based building component design method and system

Abstract

The invention relates to the technical field of building component design, and particularly discloses a BIM-based building component design method and system, a first reconnaissance data type is extracted from building reconnaissance data, and first construction site data is extracted from the building reconnaissance data according to the first reconnaissance data type, carrying out one-hot coding on the first survey data type by adopting a data analysis and extraction technology based on deep learning, and carrying out feature extraction on the construction site image; and intelligently obtaining a first construction site data estimation value according to multi-mode implicit semantic association between the first investigation data type one-hot coding feature and the construction site image semantic coding feature, and comparing the first construction site data estimation value with the first construction site data estimation value to confirm whether to trigger the data verification instruction. Data can be extracted and analyzed from multiple angles and levels, fine and complex field changes are captured, and comprehensiveness and accuracy of the data are ensured.

Description

Technical Field

[0001] This application relates to the technical field of building component design, and more specifically, to a BIM-based building component design method and system. Background Art

[0002] Building Information Modeling (BIM) is a digital technology-based method in the construction industry that enables the full life-cycle management of design, construction, and operation by integrating data from various stages of a building project. A BIM model contains not only geometric information but also multi-dimensional data such as material properties, construction progress, and cost information, providing strong support for the efficient management and collaborative work of building projects. Building survey data is the foundation of building design and construction, and the accuracy and integrity of this data directly affect the feasibility and safety of building projects.

[0003] Chinese Patent CN116822027A proposes a BIM-based building design method, device, computer equipment, and storage medium. It first collects building survey data and construction site images, then ensures the accuracy of the information through data verification to generate reliable building design data. Next, it uses this data to generate a building foundation BIM model and building component models, and synchronizes the component models into the foundation BIM model to form a model to be verified. Finally, it performs integrity verification on the model to generate the final building design BIM model.

[0004] In the patent, to determine whether data needs to be verified, it first compares the initial construction image with the image survey data to generate image difference data, and then uses this data to update the initial survey data to form data to be compared and then compares it with the construction site data. However, simple comparison can only identify obvious visual differences and may not accurately capture subtle or complex on-site changes (such as soil properties, underground pipeline layouts, etc.), easily missing important details and affecting the comprehensiveness and accuracy of the data. In addition, this simple comparison mainly relying on image features and initial survey data fails to fully utilize the deep associations between images and data. For example, the influence of building structures on the surrounding environment, etc., these associations are crucial in complex design and construction processes. The lack of consideration of these associations will lead to insufficient comprehensiveness and accuracy of data verification, increasing design risks.

[0005] Therefore, an optimized BIM-based building component design solution is needed to solve the above technical problems. Summary of the Invention

[0006] The present application provides a BIM-based building component design method and system, which can extract and analyze data from multiple perspectives and levels, capture subtle and complex on-site changes, and ensure the comprehensiveness and accuracy of the data. Therefore, this intelligent verification method not only improves the efficiency of data verification, but also can timely detect and correct potential problems, reducing design risks.

[0007] In a first aspect, there is provided a BIM-based building component design method, including: obtaining building basic data, and extracting building survey data and construction site images from the building basic data; based on the building survey data and the construction site images, determining whether to trigger a data verification instruction; in response to triggering the data verification instruction, obtaining building design data; creating a building basic BIM model based on the building design data, and obtaining a building component model; synchronizing the building component model to the building basic BIM model to form a model to be verified, and performing integrity verification on the model to be verified to obtain a building design BIM model; wherein, based on the building survey data and the construction site images, determining whether to trigger a data verification instruction includes: extracting a first survey data type from the building survey data; obtaining first construction site data according to the first survey data type, and performing one-hot encoding on the first survey data type to obtain a first survey data type one-hot encoding feature; performing feature extraction on the construction site image to obtain a construction site image semantic encoding feature, and performing multi-modal implicit semantic association of building data with the first survey data type one-hot encoding feature to obtain a first survey data type-guided construction site image semantic focusing feature; based on the first construction site data estimated value obtained by predicting the first construction site data based on the first survey data type-guided construction site image semantic focusing feature, confirming whether to trigger the data verification instruction.

[0008] Second aspect, a BIM-based building component design system is provided, which is characterized by comprising: a building basic data acquisition module for acquiring building basic data and extracting building survey data and construction site images from the building basic data; a data verification instruction trigger judgment module for judging whether to trigger a data verification instruction based on the building survey data and the construction site images; a building design data module for obtaining building design data in response to triggering the data verification instruction; a building component model module for creating a building basic BIM model based on the building design data and obtaining a building component model; a model integrity verification module to be verified for synchronizing the building component model to the building basic BIM model to form a model to be verified and performing integrity verification on the model to be verified to obtain a building design BIM model; wherein, the data verification instruction trigger judgment module includes: A first survey data type extraction unit for extracting a first survey data type from the building survey data; A first survey data type one-hot encoding unit for obtaining first construction site data according to the first survey data type and performing one-hot encoding on the first survey data type to obtain a first survey data type one-hot encoding feature; A building data multimodal implicit semantic association unit for extracting features from the construction site images to obtain a construction site image semantic encoding feature and performing building data multimodal implicit semantic association with the first survey data type one-hot encoding feature to obtain a first survey data type-guided construction site image semantic focusing feature; A first construction site data prediction unit for confirming whether to trigger the data verification instruction based on an estimated value of first construction site data obtained by predicting the first construction site data based on the first survey data type-guided construction site image semantic focusing feature.

[0009] The present application has at least the following technical effects: A BIM-based building component design method and system provided by the present application extracts the first type of survey data from the building survey data, and according to the first type of survey data, extracts the first construction site data from the building survey data, and uses deep learning-based data analysis and extraction technology to perform one-hot encoding on the first type of survey data, and extracts features from the construction site images. Thus, based on the multi-modal implicit semantic association between the one-hot encoding features of the first type of survey data and the semantic encoding features of the construction site images, an estimated value of the first construction site data is intelligently obtained, and the comparison between it and the estimated value of the first construction site data is used to confirm whether to trigger the data verification instruction. It can extract and analyze data from multiple perspectives and levels, capture subtle and complex on-site changes, and ensure the comprehensiveness and accuracy of the data. Therefore, this intelligent verification method not only improves the efficiency of data verification, but also can timely detect and correct potential problems, reducing design risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings of the embodiments of the present application will be briefly introduced below. Obviously, the drawings described below only relate to some embodiments of the present application and do not limit the present application.

[0011] Figure 1 It is a schematic flowchart of the BIM-based building component design method of the embodiment of the present application.

[0012] Figure 2 It is a schematic flowchart for determining whether to trigger a data verification instruction based on the building survey data and the construction site images in the BIM-based building component design method of the embodiment of the present application.

[0013] Figure 3 It is a schematic diagram of data flow for determining whether to trigger a data verification instruction based on the building survey data and the construction site images in the BIM-based building component design method of the embodiment of the present application.

[0014] Figure 4 It is a schematic block diagram of the BIM-based building component design system of the embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0015] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts also belong to the scope of protection of the present application.

[0016] This application proposes a BIM-based building component design method. Figure 1 It is a schematic flowchart of the BIM-based building component design method according to an embodiment of this application. As Figure 1 shown, the BIM-based building component design method includes: S1, obtaining building basic data, and extracting building survey data and construction site images from the building basic data; S2, based on the building survey data and the construction site images, determining whether to trigger a data verification instruction; S3, in response to triggering the data verification instruction, obtaining building design data; S4, creating a building basic BIM model based on the building design data, and obtaining a building component model; S5, synchronizing the building component model to the building basic BIM model to form a model to be verified, and performing integrity verification on the model to be verified to obtain a building design BIM model.

[0017] In the above BIM-based building component design method, in S1, building basic data is obtained, and building survey data and construction site images are extracted from the building basic data. It should be understood that building basic data covers site survey data (such as geology, hydrology, topography, etc.) and image materials of the construction site. These information are the prerequisite conditions for building design, and can help designers understand the actual environment and conditions of the project. For example, assume that a high-rise building is to be built in a newly developed residential area. Then, a professional survey team needs to conduct a detailed geological exploration of this land first, record information such as soil type and groundwater level, and take photos or videos of the construction site. These materials constitute a part of the building basic data.

[0018] In the above BIM-based building component design method, in S2, based on the building survey data and the construction site images, it is determined whether to trigger a data verification instruction. It should be understood that in the building design process, building survey data and construction site images are very important information sources. Building survey data usually includes information such as geology, hydrology, and topography. These data directly affect the basic design and structural safety of the building. Construction site images provide the real-time situation of the construction site, and can help the design team understand the specific environment and conditions of the site. By comparing these two types of data, the accuracy and consistency of the data can be ensured, and design mistakes caused by data errors can be prevented.

[0019] Traditional data verification methods often rely on simple visual comparison. This method can only identify obvious visual differences, but for some subtle and complex on-site changes, such as changes in soil properties and layout adjustments of underground pipelines, it may not be able to accurately capture. By combining building survey data and construction site images, advanced technologies such as deep learning can be used to identify and analyze these subtle changes to ensure the comprehensiveness and accuracy of the data.

[0020] Accordingly, in the process of determining whether to trigger a data verification instruction based on the building survey data and the construction site images, the technical concept of this application is to extract the first type of survey data from the building survey data, and according to the first type of survey data, extract the first construction site data from the building survey data, and use deep learning-based data analysis and extraction techniques to perform one-hot encoding on the first type of survey data, and extract features from the construction site images, so as to intelligently obtain an estimated value of the first construction site data based on the multi-modal implicit semantic association between the one-hot encoded features of the first type of survey data and the semantic encoded features of the construction site images, and confirm whether to trigger the data verification instruction by comparing it with the estimated value of the first construction site data. It can extract and analyze data from multiple perspectives and levels, capture subtle and complex on-site changes, and ensure the comprehensiveness and accuracy of the data. Therefore, this intelligent verification method not only improves the efficiency of data verification, but also can timely discover and correct potential problems, reducing design risks.

[0021] Figure 2 It is a schematic flowchart for determining whether to trigger a data verification instruction based on the building survey data and the construction site images in the BIM-based building component design method of this application embodiment. Figure 3 It is a schematic diagram of data flow for determining whether to trigger a data verification instruction based on the building survey data and the construction site images in the BIM-based building component design method of this application embodiment. As Figure 2 and Figure 3 shown, determining whether to trigger a data verification instruction based on the building survey data and the construction site images includes: S31, extracting the first type of survey data from the building survey data; S32, obtaining the first construction site data according to the first type of survey data, and performing one-hot encoding on the first type of survey data to obtain one-hot encoded features of the first type of survey data; S33, extracting features from the construction site images to obtain semantic encoded features of the construction site images, and performing multi-modal implicit semantic association of building data with the one-hot encoded features of the first type of survey data to obtain semantic focused features of the construction site images guided by the first type of survey data; S34, based on the semantic focused features of the construction site images guided by the first type of survey data, predicting the first construction site data to obtain an estimated value of the first construction site data, and confirming whether to trigger the data verification instruction.

[0022] In the above BIM-based building component design method, in S31, extract the first type of survey data from the building survey data. It should be understood that for a project that conducts a survey of the construction site, extract the corresponding type of survey data from the building survey data. Further, according to the type of survey data, extract the corresponding data from the building survey data as the construction site data. It can be understood that the construction site data is the survey data corresponding to each survey item when conducting a survey of the construction site.

[0023] In the above BIM-based building component design method, in S32, according to the first type of survey data, obtain the first construction site data, and perform one-hot encoding on the first type of survey data to obtain the one-hot encoding feature of the first type of survey data. It should be understood that considering that the first type of survey data contains a large amount of on-site data, and the first construction site data contains specific information about the project location such as geological conditions, soil types, groundwater levels, etc. Therefore, in order to more accurately and meticulously analyze and determine whether to trigger a data verification instruction to ensure the safety and stability of the building design, in the technical solution of this application, according to the first type of survey data, extract the first construction site data from the building survey data. Then, in order to be able to convert categorical data such as geological conditions, topography, and hydrological conditions into a numerical form that can be processed by machine learning algorithms, while eliminating the possible ordinal relationship between categories for easier subsequent understanding and analysis, in the technical solution of this application, perform one-hot encoding on the first type of survey data to obtain the one-hot encoding vector of the first type of survey data.

[0024] Optionally, in an embodiment of this application, obtaining the first construction site data according to the first type of survey data and performing one-hot encoding on the first type of survey data to obtain the one-hot encoding feature of the first type of survey data includes: extracting the first construction site data from the building survey data according to the first type of survey data; performing one-hot encoding on the first type of survey data to obtain the one-hot encoding vector of the first type of survey data as the one-hot encoding feature of the first type of survey data.

[0025] In the above BIM-based building component design method, in S33, after feature extraction is performed on the construction site image to obtain the semantic encoding features of the construction site image, multi-modal implicit semantic association of building data is carried out with the one-hot encoding features of the first survey data type to obtain the semantic focusing features of the construction site image guided by the first survey data type. It should be understood that the construction site image is an important data source reflecting the actual situation of the construction site, containing rich visual information, such as the appearance of the building, the construction progress, environmental changes, etc. Through deep learning technology, high-level semantic features can be extracted from the construction site image. These features include not only the category and location of objects, but also the context information of the environment. The extracted features are encoded into feature maps, which contain key information in the image, such as the category, location, shape, etc. of the objects. By performing multi-modal implicit semantic association on the construction site image and the survey data, a more comprehensive and accurate feature representation can be generated for data verification. This feature representation not only contains the visual information of the image, but also integrates the background information of the survey data, and can more accurately reflect the actual situation of the construction site.

[0026] Optionally, in an embodiment of the present application, after feature extraction is performed on the construction site image to obtain the semantic encoding features of the construction site image, multi-modal implicit semantic association of building data is carried out with the one-hot encoding features of the first survey data type to obtain the semantic focusing features of the construction site image guided by the first survey data type, including: inputting the construction site image into a construction site image feature extractor based on a depthwise separable convolutional neural network model to obtain a semantic encoding feature map of the construction site image as the semantic encoding features of the construction site image; performing multi-modal implicit semantic association on the one-hot encoding vector of the first survey data type and the semantic encoding feature map of the construction site image to obtain the semantic focusing features of the construction site image guided by the first survey data type.

[0027] In the above BIM-based building component design method, the construction site image is input into the construction site image feature extractor based on the depthwise separable convolutional neural network model to obtain the construction site image semantic encoding feature map as the construction site image semantic encoding feature. It should be understood that considering that the construction site image contains rich and high-level feature information, and the depthwise separable convolutional neural network model decomposes the traditional convolutional operation into depthwise convolution and pointwise convolution, significantly reducing the computational amount and model parameters while maintaining good feature extraction ability, which makes the model more efficient in processing large-scale image data to capture the context information in the image. Therefore, in order to capture and mine the high-level semantic features in the construction site image, such as the category and location of objects, in the technical solution of this application, the construction site image is input into the construction site image feature extractor based on the depthwise separable convolutional neural network model to capture and refine rich semantic feature information, obtaining the construction site image semantic encoding feature map.

[0028] In the above BIM-based building component design method, the one-hot encoded vector of the first survey data type and the construction site image semantic encoding feature map are subjected to multi-modal implicit semantic association of building data to obtain the first survey data type-guided construction site image semantic focusing feature. It should be understood that considering that the one-hot encoded vector of the first survey data type and the construction site image semantic encoding feature map each express different aspects of information in the building survey process. Specifically, the one-hot encoded vector of the first survey data type is mainly used to identify and distinguish different survey data types, providing clear category guidance for the model, while the construction site image semantic encoding feature map is a high-level abstraction of the actual situation of the construction site, providing rich visual and semantic information. In order to effectively fuse these two types of information to form a comprehensive feature representation to provide more comprehensive data support to ensure the accurate prediction of the first construction site data, in the technical solution of this application, the one-hot encoded vector of the first survey data type and the construction site image semantic encoding feature map are subjected to multi-modal implicit semantic association of building data to obtain the first survey data type-guided construction site image semantic focusing feature map as the first survey data type-guided construction site image semantic focusing feature.

[0029] Optionally, in an embodiment of the present application, performing multi-modal implicit semantic association on the one-hot encoded vector of the first survey data type and the semantic encoded feature map of the construction site image to obtain the first survey data type-guided semantic focus feature of the construction site image, including: performing feature distillation on the semantic encoded feature map of the construction site image to obtain a semantic distilled feature vector of the construction site image; inputting the one-hot encoded vector of the first survey data type and the semantic distilled feature vector of the construction site image into a multi-modal implicit association feature capture network to obtain an implicit common feature vector of survey data-construction site; inputting the implicit common feature vector of survey data-construction site into a feature preliminary screening module based on a gated unit to obtain a primary survey data-construction site feature preliminary screening weight vector; based on the primary survey data-construction site feature preliminary screening weight vector, performing preliminary feature screening on the semantic encoded feature map of the construction site image to obtain a preliminary screened cross-modal joint encoded feature map of survey data-construction site; inputting the implicit common feature vector of survey data-construction site into a feature secondary screening module based on a multi-level gated unit to obtain a secondary survey data-construction site feature preliminary screening weight vector; based on the secondary survey data-construction site feature preliminary screening weight vector, performing secondary feature screening on the preliminary screened cross-modal joint encoded feature map of survey data-construction site to obtain a first survey data type-guided semantic focus feature map of the construction site image as the first survey data type-guided semantic focus feature of the construction site image.

[0030] Optionally, in an embodiment of the present application, performing feature distillation on the semantic encoded feature map of the construction site image to obtain a semantic distilled feature vector of the construction site image, including: performing point convolution encoding on the semantic encoded feature map of the construction site image to obtain a semantic convolution encoded feature map of the construction site image; performing average pooling processing on each feature matrix of the semantic convolution encoded feature map of the construction site image along the channel dimension to obtain a semantic feature vector of the construction site image; using a sigmoid function to process the semantic feature vector of the construction site image to obtain the semantic distilled feature vector of the construction site image.

[0031] Optionally, in an embodiment of the present application, inputting the one-hot encoded vector of the first exploration data type and the semantic distilled feature vector of the construction site image into a multi-modal implicit association feature capture network to obtain an exploration data-construction site implicit common feature vector, including: concatenating the one-hot encoded vector of the first exploration data type and the semantic distilled feature vector of the construction site image to obtain an exploration data-construction site concatenated feature vector; multiplying the exploration data-construction site concatenated feature vector by an exploration data-construction site weight matrix and then performing element-wise addition with an exploration data-construction site bias vector to obtain an exploration data-construction site implicit feature vector; inputting the exploration data-construction site implicit feature vector into a tanh function to obtain the exploration data-construction site implicit common feature vector.

[0032] Optionally, in an embodiment of the present application, inputting the exploration data-construction site implicit common feature vector into a feature preliminary screening module based on a gated unit to obtain a first-level exploration data-construction site feature preliminary screening weight vector, including: multiplying the exploration data-construction site implicit common feature vector by a first weight matrix and then performing element-wise addition with a first bias vector to obtain a first exploration data-construction site weight vector; inputting the first exploration data-construction site weight vector into a sigmoid function to obtain a first exploration data-construction site activation weight vector; using a mask function to perform feature preliminary screening on the first exploration data-construction site activation weight vector to obtain the first-level exploration data-construction site feature preliminary screening weight vector, including: comparing each eigenvalue in the first exploration data-construction site activation weight vector with a preset threshold to obtain the first-level exploration data-construction site feature preliminary screening weight vector composed of multiple updated eigenvalues; wherein, in response to the eigenvalue being greater than a second preset threshold, using the original value of the eigenvalue as the updated eigenvalue; in response to the eigenvalue being greater than a first preset threshold and less than or equal to the second preset threshold, using the value obtained by dividing the eigenvalue by two as the updated eigenvalue; in response to the eigenvalue being less than or equal to the first preset threshold, using zero as the updated eigenvalue.

[0033] Optionally, in an embodiment of the present application, inputting the exploration data - implicit common feature vector of the construction site into the feature secondary screening module based on a multi-level gated unit to obtain a secondary exploration data - initial screening weight vector of the construction site features, including: multiplying the first exploration data - weight vector of the construction site by a second weight matrix and then performing element-wise addition with a second bias vector to obtain a second exploration data - weight vector of the construction site; inputting the second exploration data - weight vector of the construction site into a sigmoid function to obtain a second exploration data - activated weight vector of the construction site; using a mask function to perform feature secondary screening on the second exploration data - activated weight vector of the construction site to obtain the secondary exploration data - initial screening weight vector of the construction site features.

[0034] Optionally, in an embodiment of the present application, based on the primary exploration data - initial screening weight vector of the construction site features, performing primary feature screening on the semantic encoding feature map of the construction site image to obtain a primary screening exploration data - cross-modal joint encoding feature map of the construction site, including: performing element-wise multiplication of the feature values at each position in the primary exploration data - initial screening weight vector of the construction site features by each feature matrix along the channel dimension in the semantic encoding feature map of the construction site image to obtain an exploration data - cross-modal joint encoding weighted feature map of the construction site; inputting the exploration data - cross-modal joint encoding weighted feature map of the construction site into a tanh function to obtain the primary screening exploration data - cross-modal joint encoding feature map of the construction site.

[0035] Optionally, in an embodiment of the present application, based on the secondary exploration data - initial screening weight vector of the construction site features, performing secondary feature screening on the primary screening exploration data - cross-modal joint encoding feature map of the construction site to obtain a first exploration data type - guided semantic focus feature map of the construction site image, including: performing element-wise multiplication of the feature values at each position in the secondary exploration data - initial screening weight vector of the construction site features by each feature matrix along the channel dimension in the primary screening exploration data - cross-modal joint encoding feature map of the construction site to obtain an exploration data - cross-modal joint encoding weighted feature map of the construction site; inputting the exploration data - cross-modal joint encoding weighted feature map of the construction site into a sigmoid function to obtain the first exploration data type - guided semantic focus feature map of the construction site image.

[0036] In summary, in the embodiment of the present application, the one-hot encoded vector of the first exploration data type and the semantic encoding feature map of the construction site image are subjected to multi-modal implicit association of building data, which can be expressed by the formula: Among them, is the semantic coding feature map of the construction site image, is to perform point convolution coding on the feature map, is to perform average pooling processing on each feature matrix along the channel dimension in the feature map, is function, is the semantic distillation feature vector of the construction site image, is the one-hot encoding vector of the first type of exploration data, is the concatenation operation, and are the exploration data-construction site weight matrix and the exploration data-construction site bias vector respectively, represents matrix multiplication, is the tanh function, is the implicit common feature vector of the exploration data-construction site, and are the first weight matrix and the first bias vector respectively, is the first exploration data-construction site activation weight vector, is the masking operation, is the first-level exploration data-construction site feature initial screening weight vector, is element-wise multiplication by position, is the initial screening exploration data-construction site cross-modal joint coding feature map, and are the second weight matrix and the second bias vector respectively, is the second exploration data-construction site activation weight vector, is the second-level exploration data-construction site feature initial screening weight vector, is the preset threshold, is the first type of exploration data-guided construction site image semantic focusing feature map.

[0037] Specifically, first, perform feature distillation on the semantic encoding feature map of the construction site image to obtain a semantic distillation feature vector of the construction site image. In particular, this process utilizes the concepts of dimensionality reduction and simplification in deep learning. The aim is to remove noise and redundant information in the feature map, retain the most relevant parts, and obtain a more refined and representative feature representation. Additionally, the feature distillation technology adopted can also ensure that data from different sources can be compared and integrated in a unified semantic space, thereby enhancing effective interaction in subsequent steps. Then, in order to extract common and implicit feature representations from multi-modal data to better understand the actual situation of the construction site, input the one-hot encoded vector of the first type of survey data and the semantic distillation feature vector of the construction site image into a multi-modal implicit association feature capture network to obtain an implicit common feature vector of survey data-construction site. It should be understood that through implicit association capture, potential common patterns or structures of the two features can be identified and mined, although these patterns or structures are not directly apparent. Specifically, the multi-modal implicit association feature capture network enhances cross-modal consistent expression by introducing a collaborative attention mechanism to better reveal the internal associations between data from different sources. After that, perform a first-level weight screening on the implicit common feature vector through a feature pre-screening module based on a gated unit to use the gated unit to dynamically assign weights to each feature, highlight those features that are more important for the current task, and enable the model to focus on key information, obtaining a first-level implicit common feature vector of survey data-construction site weight vector. And based on this first-level weight vector, perform a preliminary feature screening on the semantic encoding feature map of the construction site image to reorganize the content in the original semantic encoding feature map of the construction site image through feature weighting, generating a pre-screened implicit common feature map of survey data-construction site cross-modal joint encoding that is more focused on associated characteristics, such that the initially screened features contain the feature information of both modalities and the associations between them. Next, in order to further optimize and refine the feature representation, use a feature second-level screening module based on a multi-level gated unit to process the implicit common feature vector. In particular, this architecture provides more diverse control mechanisms, can more finely adjust the information transmission process, each layer focuses on specific types of feature extraction or pattern recognition, gradually eliminates irrelevant or redundant features, retains the most core information, and finally obtains a second-level implicit common feature vector of survey data-construction site weight vector. Finally, based on this second-level weight vector, perform a second-level feature screening on the pre-screened implicit common feature map of survey data-construction site cross-modal joint encoding. After two consecutive targeted selections and reconstructions, obtain a first type of survey data-guided construction site image semantic focus feature map that is highly concentrated and contains significant association information between the two modalities.

[0038] In the above BIM-based building component design method, for S34, based on the first construction site data prediction obtained by guiding the semantic focus features of the construction site image according to the first survey data type to obtain the first construction site data estimated value, it is confirmed whether to trigger the data verification instruction. Optionally, in an embodiment of the present application, based on the first construction site data prediction obtained by guiding the semantic focus features of the construction site image according to the first survey data type to obtain the first construction site data estimated value, confirming whether to trigger the data verification instruction includes: inputting the semantic focus feature map of the construction site image guided by the first survey data type into the first construction site data prediction module based on a decoder to obtain the first construction site data estimated value; based on the comparison between the first construction site data and the first construction site data estimated value, it is confirmed whether to trigger the data verification instruction.

[0039] That is, the semantic focus feature of the construction site image guided by the first survey data type obtained by performing implicit semantic association using the one-hot encoding vector of the first survey data type and the image semantic encoding feature map of the construction site image is decoded, so as to intelligently obtain the first construction site data estimated value, and the comparison between it and the first construction site data estimated value is used to confirm whether to trigger the data verification instruction. By comparing the first construction site data and the estimated value, the accuracy of the model prediction can be verified, ensuring that the actual situation at the construction site is consistent with the expectation, so as to timely discover the deviation between the actual data and the estimated value and confirm whether to trigger the data verification instruction. In this way, data can be extracted and analyzed from multiple angles and levels, capturing subtle and complex on-site changes, ensuring the comprehensiveness and accuracy of the data. Therefore, this intelligent verification method not only improves the efficiency of data verification, but also can timely discover and correct potential problems, reducing the design risk.

[0040] When the one-hot encoding vector of the first survey data type and the image semantic encoding feature map of the construction site image respectively represent the one-hot encoding feature of the first survey data type and the image semantic encoding feature of the construction site image, during cross-modal optimization encoding based on multi-modal implicit association, the difference in feature modality and feature dimension between the one-hot encoding vector of the first survey data type and the image semantic encoding feature map of the construction site image will lead to sparse multi-modal implicit association, affecting the interactive semantic logical dependence and reducing the accuracy of the decoding result based on the interactive semantic distribution.

[0041] Preferably, inputting the semantic focus feature map of the construction site image guided by the first survey data type into the first construction site data prediction module based on a decoder to obtain the first construction site data estimated value includes: Unfolding the semantic focus feature map of the construction site image guided by the first survey data type into a semantic focus feature vector of the construction site image guided by the first survey data type; Calculate the mean parameter of the construction site image semantic features and the dispersion index of the construction site image semantic features corresponding to the construction site image semantic focus feature vector of the first survey data type; Based on the mean parameter of the construction site image semantic features, perform feature propagation coding modulation on the construction site image semantic focus feature vector of the first survey data type to obtain the construction site image semantic coding probability conditional constraint vector, expressed as: Among them, represents the construction site image semantic focus feature vector of the first survey data type, represents dot product by position, represents addition by position, is the mean parameter of the construction site image semantic features, is the conditional constraint adjustment factor, represents the construction site image semantic coding probability conditional constraint vector; Apply the dispersion index of the construction site image semantic features to the construction site image semantic coding probability conditional constraint vector, and perform a position-wise difference operation with the construction site image semantic focus feature vector of the first survey data type to obtain the construction site image semantic coding two-way fairness evaluation vector, expressed as: Among them, represents the position-wise difference operation, is the dispersion index of the construction site image semantic features, represents the construction site image semantic coding two-way fairness evaluation vector; After superimposing the bitwise reciprocal vector of the construction site image semantic coding two-way fairness evaluation vector and the construction site image semantic coding probability conditional constraint vector, perform a dot product by position with the construction site image semantic focus feature vector of the first survey data type to obtain the construction site image semantic coding group attribute transmission vector, expressed as: Among them, represents the bitwise reciprocal vector of the construction site image semantic coding two-way fairness evaluation vector represents the construction site image semantic coding group attribute transmission vector; Based on the mean parameter of the construction site image semantic features and the dispersion index of the construction site image semantic features, perform feature callback on the construction site image semantic coding group attribute transmission vector to obtain the optimized construction site image semantic focus feature vector of the first survey data type, expressed as: Among them, represents a weight hyperparameter, represents the first exploration data type of the optimization to guide the semantic focus feature vector of the construction site image; Input the first exploration data type of the optimization to guide the semantic focus feature vector of the construction site image into the first construction site data prediction module based on the decoder to obtain the first construction site data estimated value.

[0042] Accordingly, by designing and applying a two-way fairness evaluation framework based on probabilistic conditional constraints, a progressive tuning is implemented for the group attribute conduction system of the first exploration data type-guided construction site image semantic focus feature vector obtained by unfolding the first exploration data type-guided construction site image semantic focus feature map. In this process, the statistical correlation mapping system is organically integrated as a dynamic adjustment component of the hierarchical fairness compensation strategy, so as to achieve a robust expression of the first exploration data type-guided construction site image semantic focus feature vector at the distribution steady-state mapping level based on the system design of the multi-dimensional feature collaboration architecture. The intelligent adjustment function-driven feature interaction mode system constructed thereby significantly enhances the logical dependence mapping of the first exploration data type-guided construction site image semantic focus feature vector to the generated target domain, thereby improving the accuracy of the first construction site data estimated value obtained by inputting the first exploration data type-guided construction site image semantic focus feature vector into the first construction site data prediction module based on the decoder. In this way, data can be extracted and analyzed from multiple angles and levels, capturing subtle and complex on-site changes, ensuring the comprehensiveness and accuracy of the data. Therefore, this intelligent verification method not only improves the efficiency of data verification, but also can timely detect and correct potential problems, reducing design risks.

[0043] In the above BIM-based building component design method, in step S3, in response to triggering the data verification instruction, building design data is obtained. It should be understood that in the building design process, the accuracy and reliability of data are crucial. When the system triggers the data verification instruction, it means that there are significant inconsistencies or potential problems between the existing building exploration data and the construction site image. By responding to this instruction, the design team can re-verify and update the data to ensure that all information is up-to-date and accurate. This helps to avoid design mistakes caused by data errors and improve the reliability and safety of the design.

[0044] The triggering of data verification instructions usually means that potential problems or inconsistencies have been discovered. By responding to these instructions in a timely manner, the design team can quickly take action to re-check the relevant data and correct errors. This not only saves time and resources but also avoids bigger problems in subsequent design and construction processes, reducing rework and delays. Architectural design is a complex process involving various aspects of data and information. By responding to data verification instructions, the design team can ensure the scientificity and rationality of all data. For example, minor changes in geological conditions may have a significant impact on the foundation design of a building. By re-verifying this data, the design can be made more scientific, reasonable, and in line with actual requirements. During the actual architectural design and construction process, on-site conditions may change. By responding to data verification instructions, the design team can adjust the design plan in a timely manner, enhancing the flexibility and adaptability of the design. For example, if the on-site construction images show changes in the underground pipeline layout, the design team can adjust the design plan in a timely manner to avoid construction difficulties and safety hazards caused by improper design.

[0045] In the above BIM-based architectural component design method, in step S4, a building foundation BIM model is created based on the architectural design data, and an architectural component model is obtained. It should be understood that architectural design data refers to all relevant information that has been verified and updated during the architectural design process, including but not limited to geological exploration data, hydrological data, topographic data, material properties, construction progress, cost information, etc. These data are the basis for creating the building foundation BIM model and the architectural component model, ensuring the accuracy and reliability of the design. BIM (Building Information Modeling) is an architectural design method based on digital technology. By integrating data from all stages of a building project, it enables the full life-cycle management of design, construction, and operation. The building foundation BIM model not only contains geometric information but also multi-dimensional data such as material properties, construction progress, and cost information. Based on the architectural design data, the design team constructs the foundation model of the building on the BIM platform. This model not only reflects the three-dimensional structure of the building but also includes detailed material information, construction progress arrangements, cost budgets, etc. Through the BIM model, the design team can carry out multi-disciplinary collaborative design, improving design efficiency and quality. For example, suppose a high-rise residential building is to be designed. The design team will first create the foundation model of the building in the BIM software according to the verified architectural design data, including geological exploration reports, hydrological data, topographic maps, etc. This model will detail the overall layout, floor distribution, and structural framework of the building. At the same time, the model will also include information such as the material properties, construction progress arrangements, and cost budgets of each component, providing comprehensive support for subsequent design and construction.

[0046] Further, the building component model refers to the detailed models of various specific components in a building, such as walls, doors, windows, stairs, mechanical and electrical equipment, etc. These models not only include geometric shapes but also detailed data such as material properties, dimension information, installation locations, etc. Based on the created building foundation BIM model, the design team will further refine the design and generate detailed models of each building component according to the building functional areas and building design data. These models will be integrated into the foundation BIM model to form a complete building model.

[0047] In the above BIM-based building component design method, in step S5, the building component model is synchronized to the building foundation BIM model to form a model to be verified, and the integrity verification of the model to be verified is performed to obtain the building design BIM model. It should be understood that after synchronizing the building component model to the foundation BIM model, a model to be verified is formed. Performing integrity verification on this model can discover and correct potential design problems, ensuring the accuracy and reliability of the model. Among them, integrity verification is a systematic inspection process aimed at verifying whether all components in the model have been correctly installed without omission or error. The verification content includes but is not limited to the quantity, type, dimension, location, etc. of the components. The verification can be carried out by combining automated tools and manual review. Automated tools can quickly check a large amount of data, and manual review can ensure the accuracy of details. Through integrity verification, the high quality of building design and the high efficiency of construction can be ensured. A complete and accurate BIM model can provide detailed guidance for the construction team, reduce errors and rework during the construction process, and improve construction efficiency.

[0048] In summary, the BIM-based building component design method according to the embodiments of the present application is elucidated. It extracts the first type of survey data from the building survey data, and according to the first type of survey data, extracts the first construction site data from the building survey data, and uses deep learning-based data analysis and extraction techniques to perform one-hot encoding on the first type of survey data, extract features from the construction site images, and thus intelligently obtain the estimated value of the first construction site data based on the multi-modal implicit semantic association between the one-hot encoded features of the first type of survey data and the semantic encoded features of the construction site images, and compare it with the estimated value of the first construction site data to confirm whether to trigger the data verification instruction. In this way, data can be extracted and analyzed from multiple perspectives and levels, capturing subtle and complex on-site changes, ensuring the comprehensiveness and accuracy of the data. Therefore, this intelligent verification method not only improves the efficiency of data verification but also can timely discover and correct potential problems, reducing design risks.

[0049] Figure 4 It is a schematic block diagram of the BIM-based building component design system according to the embodiments of the present application. AsFigure 4 As shown in Figure 4 , the BIM-based building component design system 100 includes: a building basic data acquisition module 110, configured to acquire building basic data and extract building survey data and construction site images from the building basic data; a data verification instruction trigger judgment module 120, configured to judge whether to trigger a data verification instruction based on the building survey data and the construction site images; a building design data module 130, configured to obtain building design data in response to triggering the data verification instruction; a building component model module 140, configured to create a building basic BIM model based on the building design data and obtain a building component model; and a to-be-verified model integrity verification module 150, configured to synchronize the building component model to the building basic BIM model to form a to-be-verified model, and perform integrity verification on the to-be-verified model to obtain a building design BIM model.

[0050] Optionally, in an embodiment of the present application, the data verification instruction trigger judgment module includes: a first survey data type extraction unit, configured to extract a first survey data type from the building survey data; a first survey data type one-hot encoding unit, configured to obtain first construction site data according to the first survey data type and perform one-hot encoding on the first survey data type to obtain a first survey data type one-hot encoding feature; a building data multi-modal implicit semantic association unit, configured to perform feature extraction on the construction site images to obtain a construction site image semantic encoding feature, and perform building data multi-modal implicit semantic association with the first survey data type one-hot encoding feature to obtain a first survey data type-guided construction site image semantic focusing feature; and a first construction site data prediction unit, configured to confirm whether to trigger the data verification instruction based on a first construction site data estimated value obtained by performing first construction site data prediction based on the first survey data type-guided construction site image semantic focusing feature.

[0051] The specific operations of the various modules and units in the above BIM-based building component design system have been introduced in detail in the description of the Figures 1 to 3 BIM-based building component design method above, and therefore, the repeated description thereof will be omitted.

[0052] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above-described exemplary embodiments, and the present invention can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be embraced within the present invention. Any reference signs in the claims should not be construed as limiting the claims involved.

[0053] In addition, it is obvious that the term "comprising" does not exclude other elements or steps, and the singular does not exclude the plural. The multiple elements stated in the apparatus claims can also be implemented by one element through software or hardware.

[0054] Finally, it should be noted that the above description has been given for purposes of illustration and description. In addition, the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A BIM-based building component design method, characterized in that, Including: Obtain building basic data, and extract building survey data and construction site images from the building basic data; Based on the building survey data and the construction site images, determine whether to trigger a data verification instruction; In response to triggering the data verification instruction, obtain building design data; Create a building basic BIM model based on the building design data, and obtain a building component model; synchronize the building component model to the building basic BIM model to form a model to be verified, and perform integrity verification on the model to be verified to obtain a building design BIM model; wherein, based on the building survey data and the construction site images, determining whether to trigger a data verification instruction includes: Extract the first survey data type from the building survey data; According to the first survey data type, obtain first construction site data, and perform one-hot encoding on the first survey data type to obtain a one-hot encoding feature of the first survey data type; Extract features from the construction site images to obtain a semantic encoding feature of the construction site images, and then perform multi-modal implicit semantic association of building data with the one-hot encoding feature of the first survey data type to obtain a semantic focusing feature of the construction site images guided by the first survey data type; Based on the estimated value of the first construction site data obtained by predicting the first construction site data using the semantic focusing feature of the construction site images guided by the first survey data type, confirm whether to trigger the data verification instruction.

2. The BIM-based building component design method according to claim 1, characterized in that, According to the first survey data type, obtain first construction site data, and perform one-hot encoding on the first survey data type to obtain a one-hot encoding feature of the first survey data type, including: Extract the first construction site data from the building survey data according to the first survey data type; Perform one-hot encoding on the first survey data type to obtain a one-hot encoding vector of the first survey data type as the one-hot encoding feature of the first survey data type.

3. The BIM-based building component design method according to claim 2, characterized in that Extract features from the construction site images to obtain a semantic encoding feature of the construction site images, and then perform multi-modal implicit semantic association of building data with the one-hot encoding feature of the first survey data type to obtain a semantic focusing feature of the construction site images guided by the first survey data type, including: Input the construction site images into a construction site image feature extractor based on a depthwise separable convolutional neural network model to obtain a semantic encoding feature map of the construction site images as the semantic encoding feature of the construction site images; Perform multi-modal implicit semantic association of building data on the one-hot encoding vector of the first survey data type and the semantic encoding feature map of the construction site images to obtain the semantic focusing feature of the construction site images guided by the first survey data type.

4. The BIM-based building component design method according to claim 3, characterized in that Perform multi-modal implicit semantic association of building data on the one-hot encoding vector of the first survey data type and the semantic encoding feature map of the construction site images to obtain the semantic focusing feature of the construction site images guided by the first survey data type, including: Perform feature distillation on the semantic encoding feature map of the construction site images to obtain a semantic distilled feature vector of the construction site images; Input the one-hot encoded vector of the first type of exploration data and the semantic distilled feature vector of the construction site image into the multi-modal implicit association feature capture network to obtain the implicit common feature vector of exploration data - construction site; Input the implicit common feature vector of exploration data - construction site into the feature pre-screening module based on the gated unit to obtain the first-level implicit common feature vector of exploration data - construction site; Based on the first-level implicit common feature vector of exploration data - construction site, perform preliminary feature screening on the semantic encoded feature map of the construction site image to obtain the pre-screened cross-modal joint encoded feature map of exploration data - construction site; Input the implicit common feature vector of exploration data - construction site into the feature secondary screening module based on the multi-level gated unit to obtain the second-level implicit common feature vector of exploration data - construction site; Based on the second-level implicit common feature vector of exploration data - construction site, perform secondary feature screening on the pre-screened cross-modal joint encoded feature map of exploration data - construction site to obtain the first type of exploration data-guided semantic focus feature map of the construction site image as the first type of exploration data-guided semantic focus feature of the construction site image.

5. The BIM-based building component design method according to claim 4, characterized in that, Perform feature distillation on the semantic encoded feature map of the construction site image to obtain the semantic distilled feature vector of the construction site image, including: Perform point convolution encoding on the semantic encoded feature map of the construction site image to obtain the semantic convolution encoded feature map of the construction site image; Perform average pooling on each feature matrix along the channel dimension of the semantic convolution encoded feature map of the construction site image to obtain the semantic feature vector of the construction site image; Use the sigmoid function to process the semantic feature vector of the construction site image to obtain the semantic distilled feature vector of the construction site image.

6. The BIM-based building component design method according to claim 5, wherein, Input the one-hot encoded vector of the first type of exploration data and the semantic distilled feature vector of the construction site image into the multi-modal implicit association feature capture network to obtain the implicit common feature vector of exploration data - construction site, including: Perform concatenation processing on the one-hot encoded vector of the first type of exploration data and the semantic distilled feature vector of the construction site image to obtain the concatenated feature vector of exploration data - construction site; Multiply the concatenated feature vector of exploration data - construction site by the weight matrix of exploration data - construction site and then perform element-wise addition with the bias vector of exploration data - construction site to obtain the implicit feature vector of exploration data - construction site; Input the implicit feature vector of exploration data - construction site into the tanh function to obtain the implicit common feature vector of exploration data - construction site.

7. The BIM-based building component design method according to claim 6, wherein, Input the implicit common feature vector of exploration data - construction site into the feature pre-screening module based on the gated unit to obtain the first-level implicit common feature vector of exploration data - construction site, including: Multiply the implicit common feature vector of exploration data - construction site by the first weight matrix and then perform element-wise addition with the first bias vector to obtain the first implicit common feature vector of exploration data - construction site; Input the first implicit common feature vector of exploration data - construction site into the sigmoid function to obtain the first activation weight vector of exploration data - construction site; Using the mask function to perform preliminary feature screening on the first exploration data - construction site activation weight vector to obtain the first - level exploration data - construction site feature preliminary screening weight vector, including: comparing each eigenvalue in the first exploration data - construction site activation weight vector with a preset threshold to obtain the first - level exploration data - construction site feature preliminary screening weight vector composed of multiple updated eigenvalues; wherein, in response to the eigenvalue being greater than the second preset threshold, using the original value of the eigenvalue as the updated eigenvalue; in response to the eigenvalue being greater than the first preset threshold and less than or equal to the second preset threshold, using the value obtained by dividing the eigenvalue by two as the updated eigenvalue; in response to the eigenvalue being less than or equal to the first preset threshold, using zero as the updated eigenvalue.

8. The BIM-based building component design method according to claim 7, characterized in that Inputting the exploration data - construction site implicit common feature vector into the feature secondary screening module based on a multi - level gated unit to obtain the second - level exploration data - construction site feature preliminary screening weight vector, including: Multiplying the first exploration data - construction site weight vector by a second weight matrix and then performing element - wise addition with a second bias vector to obtain the second exploration data - construction site weight vector; Inputting the second exploration data - construction site weight vector into the sigmoid function to obtain the second exploration data - construction site activation weight vector; Using the mask function to perform feature secondary screening on the second exploration data - construction site activation weight vector to obtain the second - level exploration data - construction site feature preliminary screening weight vector.

9. The BIM-based building component design method according to claim 8, characterized in that, Based on the first construction site data estimation value obtained by guiding the construction site image semantic focusing feature based on the first exploration data type to perform the first construction site data prediction, confirming whether to trigger the data verification instruction, including: Inputting the first exploration data type - guided construction site image semantic focusing feature map into the first construction site data prediction module based on a decoder to obtain the first construction site data estimation value; Based on the comparison between the first construction site data and the first construction site data estimation value, confirming whether to trigger the data verification instruction.

10. A BIM-based building component design system, characterized in that, Including: A building foundation data acquisition module, configured to acquire building foundation data and extract building exploration data and construction site images from the building foundation data; A data verification instruction trigger judgment module, configured to judge whether to trigger a data verification instruction based on the building exploration data and the construction site images; A building design data module, configured to obtain building design data in response to triggering the data verification instruction; A building component model module, configured to create a building foundation BIM model based on the building design data and obtain a building component model; A pending verification model integrity verification module, configured to synchronize the building component model to the building foundation BIM model to form a pending verification model, and perform integrity verification on the pending verification model to obtain a building design BIM model; wherein, the data verification instruction trigger judgment module includes: A first exploration data type extraction unit, configured to extract the first exploration data type from the building exploration data; The first exploration data type one-hot encoding unit is used to obtain the first construction site data according to the first exploration data type, and perform one-hot encoding on the first exploration data type to obtain the first exploration data type one-hot encoding feature; The building data multimodal implicit semantic association unit is used to extract features from the construction site image to obtain the construction site image semantic encoding feature, and then perform building data multimodal implicit semantic association with the first exploration data type one-hot encoding feature to obtain the first exploration data type-guided construction site image semantic focusing feature; The first construction site data prediction unit is used to confirm whether to trigger the data verification instruction based on the first construction site data estimated value obtained by predicting the first construction site data based on the first exploration data type-guided construction site image semantic focusing feature.

Images