WO2025124458A1 - 一种建筑信息模型搜索方法、装置及存储介质 - Google Patents

一种建筑信息模型搜索方法、装置及存储介质 Download PDF

Info

Publication number
WO2025124458A1
WO2025124458A1 PCT/CN2024/138681 CN2024138681W WO2025124458A1 WO 2025124458 A1 WO2025124458 A1 WO 2025124458A1 CN 2024138681 W CN2024138681 W CN 2024138681W WO 2025124458 A1 WO2025124458 A1 WO 2025124458A1
Authority
WO
WIPO (PCT)
Prior art keywords
search
features
model
building information
building
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
PCT/CN2024/138681
Other languages
English (en)
French (fr)
Inventor
林佳瑞
宋盛禹
潘鹏
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tsinghua University
Original Assignee
Tsinghua University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Tsinghua University filed Critical Tsinghua University
Priority to US19/142,948 priority Critical patent/US20260119729A1/en
Publication of WO2025124458A1 publication Critical patent/WO2025124458A1/zh
Anticipated expiration legal-status Critical
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/10—Geometric CAD
    • G06F30/13—Architectural design, e.g. computer-aided architectural design [CAAD] related to design of buildings, bridges, landscapes, production plants or roads
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/33—Querying
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/33—Querying
    • G06F16/335—Filtering based on additional data, e.g. user or group profiles
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/35—Clustering; Classification
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/30—Semantic analysis
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T17/00—Three-dimensional [3D] modelling for computer graphics
    • G06T17/10—Constructive solid geometry [CSG] using solid primitives, e.g. cylinders, cubes

Definitions

  • the present application relates to the field of building information technology, and in particular to a building information model search method and device.
  • BIM Building Information Modeling
  • the purpose of the present application is to provide a building information model search method and device, which can realize the search of building information models at the overall building level or the multi-component combination level.
  • the present application provides a building information model search method, comprising:
  • search intent information includes search intent semantic features, search intent topological features, and search intent geometric features
  • the building information model as the search result corresponding to the search text is determined, and the search result is recommended to the user.
  • the multimodal feature extraction of the building information model at the multi-component combination level in the model library to be searched includes:
  • the semantic information of the attributes of the components at each level in the multi-component combination level building information model is extracted, and then the semantic information of the attributes of the components at each level is summarized and counted to obtain the semantic features of the attributes of the multi-component combination level building information model.
  • the components at each level include: a building space, a wall that may be included in the building space, and a door or a window that may be included in the wall.
  • the method extracts multimodal features from a building information model at a multi-component combination level in a model library to be searched, and further includes:
  • the spatial adjacency relationship between various building spaces is determined as the topological feature of the building information model at the multi-component combination level.
  • the spatial adjacency relationship includes three relationships: non-adjacent, adjacent and non-connected, and connected.
  • the multimodal feature extraction of the multi-component combination-level building information model in the model library to be searched further includes:
  • the plane outline information of the building information model is extracted as the geometric features of the building information model at the multi-component combination level.
  • parsing the search text to obtain search intent information corresponding to the search text includes:
  • search intent semantic features, search intent topological features and search intent geometric features of the search intent information are obtained.
  • the deep embedding learning-based method of calculating the comprehensive similarity between the search intent information and the multimodal features of each building information model in the model library to be searched includes:
  • the extracted semantic features, topological features and geometric features of the building information model, as well as the search intent semantic features, search intent topological features and search intent geometric features of the search intent are embedded into a unified vectorized representation for similarity calculation.
  • the similarity calculation is a weighted cosine similarity calculation.
  • the present application provides a building information model search device, comprising:
  • a feature extraction module is used to extract multimodal features from the building information models at the multi-component combination level in the model library to be searched, and obtain multimodal features corresponding to each building information model, wherein the multimodal features include semantic features, topological features and geometric features;
  • a parsing module used to obtain a search text input by a user, parse the search text, and obtain search intent information corresponding to the search text, wherein the search intent information includes search intent semantic features, search intent topological features, and search intent geometric features;
  • a similarity calculation module used for calculating the comprehensive similarity between the search intention information and the multimodal features of each building information model in the model library to be searched based on deep embedding learning;
  • the recommendation module is used to determine the building information model as the search result corresponding to the search text according to the sorting result of the comprehensive similarity between the search intention information and the multimodal features of each building information model in the model library to be searched, and visually display the search result to the user.
  • the present application has the following advantages: (1) it realizes the semantic-topological-geometric multimodal feature search of the building-level BIM model; (2) it achieves good search results and significantly improves the accuracy; (3) it optimizes the algorithm operation efficiency and ensures the search speed.
  • FIG1 is a schematic diagram of a flow chart of a building information model search method in an embodiment of the present application
  • FIG2 is a flowchart of a semantic feature extraction program based on IFC model information parsing
  • Figure 3 is an example of extracting property information of a building BIM model based on IFC
  • Figure 4 shows how to handle the topological connection between spatial units such as rooms or courtyards.
  • FIG5 is a flowchart of a topological connectivity feature extraction program based on adjacency relationships
  • FIG6 is a schematic diagram of a method for extracting building shape features based on geometric contour data
  • FIG7 is a schematic diagram of a method for extracting plane contour features of each room on each floor of a building BIM model
  • FIG8 is a flow chart of an algorithm for extracting plane contour features of a building BIM model
  • Figure 9 is a search intent parsing process based on text segmentation and regular expressions
  • FIG10 is a flowchart of topological feature embedding of a BIM model based on spatial adjacency and connectivity features
  • FIG11 is a flowchart of a BIM model shape embedding procedure based on contour and plan features
  • FIG12 is a schematic diagram of a ResNet50 planar shape feature embedding model framework
  • Figure 13 is the architectural BIM model search process based on comprehensive similarity sorting.
  • the technical solution of this application provides a building information model search method and device accordingly, the method comprising: extracting multimodal features of building information models at the multi-component combination level in the model library to be searched, obtaining multimodal features corresponding to each building information model, the multimodal features including semantic features, topological features and geometric features; obtaining a search text input by a user, parsing the search text, obtaining search intent information corresponding to the search text, the search intent information including search intent semantic features, search intent topological features and search intent geometric features; calculating the comprehensive similarity between the search intent information and the multimodal features of each building information model in the model library to be searched based on deep embedding learning; determining the building information model as the search result corresponding to the search text according to the ranking result of the comprehensive similarity between the search intent information and the multimodal features of each building information model in the model library to be searched,
  • the present application provides a building information model search method, the method comprising:
  • search intent information includes search intent semantic features, search intent topological features, and search intent geometric features
  • the model search technology of the prior art can only realize the search of a single component, while the purpose of this application is to realize the search of a model of a multi-component combination.
  • a multi-component combination refers to the combination of components at multiple levels.
  • a typical subset of the model of a multi-component combination is the building information model at the level of the overall building.
  • the overall building for example, can be a single-family villa in a rural environment, which includes multi-story building spaces, each floor of which can include several different types of rooms, and each room contains different components.
  • the building space can also include open-air courtyards and other spaces.
  • the subsequent embodiments can be described by taking the search of the model at the level of the overall building as an example.
  • S1 to S4 of the method of the present application implement two core contents: S1 implements the multimodal feature extraction of the overall building-level BIM model, and S2 to S4 implement BIM model similarity retrieval based on deep embedding learning.
  • Core content 1 Multimodal feature extraction of BIM models at the overall building level (or other multi-component combination level).
  • Multimodal features including semantic features, topological features, and geometric features.
  • the first step to search the BIM model at the overall building level is to extract information such as attribute text, topology, geometry, etc. in the BIM model.
  • the IFC standard is an open source BIM standard formulated by the buildingSMART organization, and various open source parsing tools can be used to extract information from BIM models in IFC format.
  • This application uses Python scripts for data parsing for building BIM models created using Revit software and exported to IFC4 model format. Specifically, using the IfcOpenShell tool, the "open” and "by_type” methods are used to traverse and obtain components of specified types in the model.
  • the geometric shape of the building plane, the topological connection relationship between the rooms, the attribute information of components such as walls, doors and windows, etc. can be extracted from the BIM model, thereby forming an automatic extraction algorithm for multiple types of features of the BIM model.
  • the semantic information of the attributes of the components at each level in the building information model at the overall building level or the multi-component combination level is extracted, and then the semantic information of the attributes of the components at each level is summarized and counted to obtain the semantic features of the attributes of the building information model at the overall building level or the multi-component combination level, as shown in Figure 2. Specifically, it includes:
  • IfcSpace class that describes the information of each space in the apartment, obtain the room name from the basic LongName property, obtain the specific floor information of the room from the name of the room's associated component, find the property named "area" from RelatingPropertyDefinition to obtain the area information, obtain the room's location information from IfcSpace.Representation.Representations.SweptArea, obtain the room's outline information from BoundedBy, and obtain the partition information between rooms from the ifcRelSpaceBoundary property of BoundedBy.
  • IfcWindow class For each wall, parse the IfcWindow class associated with the wall to obtain the basic information of all windows, and obtain all the property information of the windows through the RelatingPropertyDefinition related to IfcWindow; parse IfcDoor to obtain the basic information of each door, and then read the RelatingPropertyDefinition of IfcDoor to obtain the property information of the door.
  • the property information of furniture and appliances in the room can be parsed by parsing IfcFurniture in IfcSpace.
  • the attribute information of the BIM model is summarized into a specific form of Python attribute information dictionary according to the rules, such as " ⁇ Province: Beijing, Area: 219, Cost: 600000, Number of floors: 3, Number of rooms: 10, Number of bedrooms: 4, Number of bathrooms: 3, Number of kitchens: 1... ⁇ ", and then the information of a single house can be processed into a feature vector according to the agreed parameter method for subsequent apartment type query and matching.
  • the above overall process is shown in Figure 3.
  • the spatial adjacency relationship between each building space is determined according to the attributes of the components at each level, which serves as the topological features of the building information model at the overall building level or the multi-component combination level.
  • the topological connectivity characteristics of the building BIM model refer to the relative position relationship and connection method between the rooms in the building.
  • This application divides the topological relationship of the rooms into two types: whether they are adjacent and whether they are connected, as shown in Figure 4, that is, there are only three possibilities for the relationship between any two rooms: non-adjacent, adjacent and non-connected, and connected.
  • virtual room separators Virtual Room Separator
  • This application distinguishes whether the room separation is a physical separation by extracting the PhysicalOrVirtualBoundary attribute in IfcRelSpaceBoundary of IfcSpace.
  • the algorithm flow is shown in Figure 5, and the specific steps are as follows
  • IfcSpace attribute information we can obtain the spatial information of each room.
  • the local coordinate position and direction of the room from Representation.SweptArea.Position and the relative coordinates of the contour points, we can calculate the absolute coordinate position of each contour point in each room in the global coordinate system of the entire apartment.
  • the dictionary is converted into a Networkx topology map for storage.
  • the plane outline information of the building information model is extracted as the geometric features of the building information model at the overall building level or the multi-component combination level.
  • the geometric shape features of a house involve two-dimensional and three-dimensional information, and it is difficult to directly extract information and abstract features by directly parsing the BIM model file. Due to the complexity of three-dimensional geometric features, this application focuses on analyzing the two-dimensional geometric shape features of the building BIM model.
  • the geometric features of a house can be determined by the spatial layout and plane outline information of the building. The former can be directly reflected by the floor plan, while the latter can be obtained by extracting a list of coordinates representing the outline of the building. They both contain rich geometric information, so this application comprehensively considers these two types of information in the task of extracting geometric features, as shown in Figure 6, and the specific steps are as follows.
  • the first-floor plan of the BIM model is captured from the Revit software as the overall shape data of the apartment.
  • the floor plan information of the apartment can be obtained.
  • IfcOpenShell is used to extract the local coordinates of Representation.SweptArea.Position of each room in the apartment to form a room outline information list.
  • Core content 2 BIM model similarity retrieval based on deep embedding learning.
  • the search from text to BIM model requires the calculation of the similarity between the search text and the BIM model. Therefore, this part first uses the text segmentation and regular expression methods of natural language processing to parse the search text to extract the search intent contained in the natural text. Secondly, using deep learning and special engineering tools, the multimodal features and search intent of the BIM model are respectively embedded in the search text and converted into a unified feature vector. Thirdly, a weighted comprehensive similarity calculation is performed based on the feature vector, and intelligent search and recommendation of the BIM model is realized based on the similarity sorting. The following is a detailed introduction of the steps in three parts.
  • the purpose of extracting search intent is to obtain the semantic attributes, room topological relationships, geometric descriptions and other information of the required BIM model from the text information, so as to query the target model based on this information.
  • the process is shown in Figure 9, and the specific steps are as follows.
  • Descriptive nouns and common expressions related to the housing and real estate fields were collected and compiled to create a domain word list for architectural BIM model search, which helps in text segmentation and defines the smallest unit for proper noun subdivision.
  • Jieba is a commonly used open source Chinese segmentation application package in Python.
  • the search text is segmented based on the domain vocabulary and Jieba.
  • Jieba's paddlepaddle mode is used to segment the search sentence into segments with word attributes and word order based on the domain vocabulary, so as to facilitate subsequent attribute and conjunction matching.
  • the topological connectivity features of the room are extracted. For example, “living room connected to kitchen” can extract the keywords “living room” and “kitchen” representing the room name, as well as the word “connected” representing its connectivity, thereby extracting semantic information related to topological connectivity, and finally summarizing it into a connectivity dictionary and NetworkX network diagram.
  • fuzzy shape descriptions such as "square apartment” and “long and narrow bedroom” are extracted from the search text to generate a corresponding contour feature list, including parameters such as the aspect ratio of the circumscribed rectangle, and finally summarized into a shape attribute dictionary.
  • the semantic information of each attribute is directly organized into a vector consisting of text and data in a unified format for embedding semantic features.
  • the topological information is stored in the form of a Networkx topological graph, and the topological feature embedding is implemented using the graph kernel method.
  • the floor plan image is embedded through a convolutional neural network, the contour information is embedded through the feature parameters of the contour, and the geometric shape description in the search text is also embedded as the feature parameters of the house contour.
  • the specific methods include:
  • the multimodal feature extraction method of the BIM model in section 5.2.1 can extract the semantic attributes of the BIM model as " ⁇ Province: Jiangsu, Area: 149 square meters, Cost: 100,000 US dollars, Number of floors: 2, Number of bedrooms: 3, Number of bathrooms: 2, Number of living rooms: 1, Number of kitchens: 1... ⁇ ", and the search intent extraction method in this section (1) can also generate an attribute dictionary of the corresponding form.
  • Word2Vec is a type of neural network model commonly used in natural language processing tasks to convert words into word basis vectors. This application uses manually defined conversion rules and combines the Word2Vec neural network method to embed the attribute dictionary into the semantic feature vector.
  • the processing method refers to Table 2, and the specific steps are as follows:
  • the embedding is relatively simple. After the data attributes are weighted and normalized, different types of data are embedded according to their type characteristics by manually defining rules.
  • the text is usually a string attribute, which needs to be embedded as a string vector and then embedded through the Word2Vec model.
  • semantic information is divided into text information and data information according to rules, embedded as string word vectors and data vectors respectively, and then weighted and combined as feature vectors, with the corresponding weights manually determined by experts.
  • Graph kernel is an effective method for embedding graph features in graph neural networks. Therefore, the Deepwalk deep random walk method based on the graph kernel is selected to embed topological features. This method is unsupervised, has better transferability, and is more suitable for the graph data in this application.
  • the program flow is shown in Figure 10, and the specific embedding method is as follows.
  • the Networkx library is a commonly used open source Python library for graph data format storage and sharing, while Grakelze is a commonly used open source Python library for graph kernel machine learning tasks in graph convolution. This is implemented based on NetworkX and Grakel. Based on the node list, node attribute list, and adjacency matrix of the Networkx topology graph, the Grakel.graph method is used to convert the Networkx topology graph into Grakel graph network data.
  • the same Graph Kernel is fixed, and based on the Grakel.GraphKernel method, the Grakel graph network is embedded as a topological feature vector, and then weighted normalization is performed to obtain the BIM model topological feature vector that measures the similarity of the overall interconnection relationship of the apartment types.
  • the geometric shape information of the BIM model has been divided into contour data information and floor plan data information for processing, so the two types of features are embedded separately to complement each other.
  • the geometric contour extracted from the BIM model not only contains the overall outer contour information of the apartment, but also contains the position information of each type of room in the apartment on the plane. Therefore, the shape description feature can be directly constructed based on the entire plane contour information of the apartment, and the apartment shape can be vectorized.
  • the feature vector of the apartment shape is extracted from the apartment floor plan, thereby getting rid of the influence of noise and transformation of a single contour feature, and vectorizing the overall plane layout shape of the apartment.
  • the embedding process is shown in Figure 11, and the specific embedding method is as follows.
  • the OpenCV computer vision toolkit is used to extract features from the previously extracted BIM model contour information.
  • the contour information is used to calculate the indicators describing the shape of the apartment and its internal rooms, including the aspect ratio of the outer enveloping rectangle, the area ratio of the inner area of the contour to the outer enveloping rectangle, the coordinates of the centroid of the apartment, the coordinates of each room information, etc.
  • the data is then normalized to form the apartment shape feature description indicators that describe the geometric contour features, as shown in Table 3. The details are as follows:
  • the number of fitting polygon contour points is fixed to 30, and the outer contour of the apartment is fitted into a 30-gon polygon, so that the outer contour coordinates are normalized to form a fixed-length contour point vector, thereby realizing the embedding of the outer contour information.
  • the cv2.boundingRect method is used to calculate the circumscribed rectangle of the outline.
  • the aspect ratio of the circumscribed rectangle is used to measure the squareness of the plane shape of the overall building BIM model and is embedded in the form of a floating-point number.
  • the shape features are embedded by fitting ellipses.
  • the cv2.fitEllipse method is used to obtain the ellipse closest to the building outline, and the major and minor axis vector features of the ellipse are extracted and embedded in the form of fixed-length vectors.
  • subdivided spaces such as rooms and courtyards inside the building BIM model
  • their squareness can be measured by the average aspect ratio of the circumscribed rectangle of such rooms in the apartment.
  • the average aspect ratio information of each bedroom, living room, kitchen, and courtyard is calculated separately using the cv2.boundingRect method to form the aspect ratio features of the four types of subdivided spaces, which are embedded in the form of floating-point numbers.
  • this application introduces feature engineering technology based on convolutional neural networks, and uses a pre-trained ResNet50 model to extract the overall geometric features of the plane, as follows.
  • the ResNet50 neural network was selected, and the input and output layer structure of the neural network was adjusted.
  • the input was unified into a three-channel image in the format of 224 ⁇ 224, and the output was unified into a 2048-dimensional image shape feature vector.
  • the neural network structure is shown in Figure 12.
  • the widely used ImageNet is used to pre-train the ResNet50 network structure.
  • the CheckPoint weight generation model pre-trained based on the ImageNet dataset is run and the fully connected layer data is extracted to output the 2048-dimensional image shape feature vector of the plane image.
  • the above-mentioned floor plan feature vector is weighted and combined with the contour feature index of the building BIM model to form a shape feature vector that integrates the two types of data, the apartment outline and the floor plan shape feature.
  • the weights of the two types of feature embedding are manually determined by experts and manually adjusted three times according to the results.
  • character matching ratio is used as a measure of similarity
  • floating-point and integer quantized elements such as bedroom area and number
  • relative difference is used to measure similarity
  • the similarity is also measured by taking the cosine similarity of the sub-vectors.
  • the cv2.matchShapes method of OpenCV is introduced to compare the similarity of the contours by the relative difference of the Hu moment of the contours.
  • the initial weights are first determined manually by experts and then sorted in descending order according to the comprehensive similarity to form a list of BIM model search results.
  • the weight of the comprehensive similarity is adjusted through feedback from experts who manually evaluate the search results. After more than 10 rounds of manual comparison and adjustment, the comprehensive similarity results meet the evaluation of the corresponding experts.
  • the semantic-topological-geometric multimodal feature search of BIM models at the overall building level or multi-component combination level is realized, supporting a large model scale and a wide search range.
  • IfcOpenShell the semantic features, topological connectivity features, and shape features of the overall building BIM model are extracted and deeply embedded.
  • the model scale applicable to the BIM model search algorithm is expanded to the overall building level.
  • the similarity search of the three types of semantic-topological-geometric features from text to building BIM models is realized for the first time.
  • a feature extraction module is used to extract multimodal features from the building information models at the overall building level or the multi-component combination level in the model library to be searched, and obtain multimodal features corresponding to each building information model, wherein the multimodal features include semantic features, topological features and geometric features;
  • a parsing module used to obtain a search text input by a user, parse the search text, and obtain search intent information corresponding to the search text, wherein the search intent information includes search intent semantic features, search intent topological features, and search intent geometric features;
  • a similarity calculation module used for calculating the comprehensive similarity between the search intention information and the multimodal features of each building information model in the model library to be searched based on deep embedding learning;
  • the recommendation module is used to determine the building information model as the search result corresponding to the search text according to the sorting result of the comprehensive similarity between the search intention information and the multimodal features of each building information model in the model library to be searched, and recommend the search result to the user.
  • the embodiment of the present application also provides a computer-readable storage medium, the storage medium includes a stored program, wherein when the program is running, the device where the storage medium is located is controlled to execute the above method.
  • the specific implementation process will not be repeated here.
  • the present application also provides a computer device.
  • the computer device of this embodiment includes: a processor, a memory, and a computer program stored in the memory and executable on the processor.
  • the processor executes the computer program, the aforementioned method in the embodiment is implemented. To avoid repetition, it is not described one by one here.
  • the computer program is executed by the processor, the functions of each model/unit in the device in the embodiment are implemented. To avoid repetition, it is not described one by one here.
  • the computer device may be a computing device such as a desktop computer, a notebook, a PDA, a server, a cloud server, etc.
  • the computer device may include, but is not limited to, a processor and a memory.
  • the computer device may include more or fewer components than shown in the figure, or may combine certain components, or may include different components.
  • the computer device may also include input and output devices, network access devices, buses, etc.
  • the processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.
  • CPU central processing unit
  • DSP digital signal processors
  • ASIC application-specific integrated circuits
  • FPGA field-programmable gate arrays
  • a general-purpose processor may be a microprocessor or any conventional processor, etc.
  • the memory can be an internal storage unit of a computer device, such as a hard disk or memory of a computer device.
  • the memory can also be an external storage device of a computer device, such as a plug-in hard disk equipped on a computer device, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc.
  • SMC Smart Media Card
  • SD Secure Digital
  • the memory can also include both an internal storage unit of a computer device and an external storage device.
  • the memory is used to store computer programs and other programs and data required by the computer device.
  • the memory can also be used to temporarily store data that has been output or is about to be output.
  • the disclosed systems, devices and methods can be implemented in other ways.
  • the device embodiments described above are only schematic.
  • the division of the above units is only a logical function division. There may be other division methods in actual implementation.
  • multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed.
  • the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.
  • the above-mentioned integrated unit implemented in the form of a software functional unit can be stored in a computer-readable storage medium.
  • the above-mentioned software functional unit is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) or a processor (Processor) to perform some steps of the above-mentioned methods in various embodiments of the present application.
  • the aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Geometry (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Computer Hardware Design (AREA)
  • Artificial Intelligence (AREA)
  • Computational Linguistics (AREA)
  • Databases & Information Systems (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Software Systems (AREA)
  • Structural Engineering (AREA)
  • Mathematical Analysis (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Optimization (AREA)
  • Civil Engineering (AREA)
  • Computational Mathematics (AREA)
  • Architecture (AREA)
  • Medical Informatics (AREA)
  • Evolutionary Biology (AREA)
  • Health & Medical Sciences (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computer Graphics (AREA)
  • General Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

本申请涉及一种建筑信息模型搜索方法，包括：对待搜索的模型库中的建筑信息模型进行多模态特征提取，得到各个建筑信息模型对应的多模态特征，包括语义特征、拓扑特征和几何特征；获取用户输入的搜索文本，对所述搜索文本进行解析，得到所述搜索文本对应的搜索意图信息，所述搜索意图信息包括搜索意图语义特征、搜索意图拓扑特征和搜索意图几何特征；基于深度嵌入学习，计算所述搜索意图信息与待搜索的模型库中的各个建筑信息模型的多模态特征的综合相似度；根据所述搜索意图信息与待搜索的模型库中的各个建筑信息模型的多模态特征的综合相似度的排序结果，确定作为所述搜索文本对应的搜索结果的建筑信息模型，并向用户推荐所述搜索结果。

Description

一种建筑信息模型搜索方法、装置及存储介质 技术领域
本申请涉及建筑信息技术领域，特别是关于一种建筑信息模型搜索方法及装置。
背景技术
建筑信息模型(Building Information Modeling，BIM)技术实现了直观的设计信息三维可视化表达，为跨专业协同设计、技术交底、建设项目的全过程工程管理提供了高效的解决方案。
然而，在研究中发现，现有研究缺乏对复杂建筑BIM模型的高效搜索手段。即现有的BIM模型搜索方法只能搜索模型中的建筑构件(构件例如是墙、门、窗、梁等等)，缺乏考虑整体建筑模型特征的能力(整体建筑例如是多层建筑物，包括多层、若干房间的所有构件)。
发明内容
针对上述问题，本申请的目的是提供一种建筑信息模型搜索方法及装置，能够实现整体建筑级别或多构件组合级别的建筑信息模型的搜索。
为实现上述目的，本申请采取以下技术方案：
第一方面，本申请提供一种建筑信息模型搜索方法，包括：
对待搜索的模型库中的多构件组合级别的建筑信息模型进行多模态特征提取，得到各个建筑信息模型对应的多模态特征，所述多模态特征包括语义特征、拓扑特征和几何特征；
获取用户输入的搜索文本，对所述搜索文本进行解析，得到所述搜索文本对应的搜索意图信息，所述搜索意图信息包括搜索意图语义特征、搜索意图拓扑特征和搜索意图几何特征；
基于深度嵌入学习，计算所述搜索意图信息与待搜索的模型库中的各个建筑信息模型的多模态特征的综合相似度；
根据所述搜索意图信息与待搜索的模型库中的各个建筑信息模型的多模态特征的综合相似度的排序结果，确定作为所述搜索文本对应的搜索结果的建筑信息模型，并向用户推荐所述搜索结果。
在本申请的一种实现方式中，所述对待搜索的模型库中的多构件组合级别的建筑信息模型进行多模态特征提取，包括：
提取多构件组合级别的建筑信息模型中各个级别的组件的属性的语义信息，再对各个级别的组件的属性的语义信息进行汇总统计，得到多构件组合级别的建筑信息模型的属性的语义特征。
在本申请的一种实现方式中，所述各个级别的组件，包括：建筑空间，可包含于建筑空间中的墙体，以及可包含于墙体中的门或窗。
在本申请的一种实现方式中，所述方对待搜索的模型库中的多构件组合级别的建筑信息模型进行多模态特征提取，还包括：
根据各个级别的组件的属性，确定各个建筑空间之间的空间邻接关系，作为多构件组合级别的建筑信息模型的拓扑特征。
在本申请的一种实现方式中，所述空间邻接关系包括：不相邻、相邻且不相连，以及相连三种关系。
在本申请的一种实现方式中，所述对待搜索的模型库中的多构件组合级别的建筑信息模型进行多模态特征提取，还包括：
提取建筑信息模型的平面轮廓信息，作为多构件组合级别的建筑信息模型的几何特征。
在本申请的一种实现方式中，所述对所述搜索文本进行解析，得到所述搜索文本对应的搜索意图信息，包括：
基于自然语言处理的文本分割以及正则表达式进行解析，得到搜索意图信息的搜索意图语义特征、搜索意图拓扑特征和搜索意图几何特征。
在本申请的一种实现方式中，所述基于深度嵌入学习，计算所述搜索意图信息与待搜索的模型库中的各个建筑信息模型的多模态特征的综合相似度，包括：
根据提取的建筑信息模型的语义特征、拓扑特征和几何特征，以及搜索意图的搜索意图语义特征、搜索意图拓扑特征和搜索意图几何特征，嵌入到统一的矢量化表示中进行相似度计算。
在本申请的一种实现方式中，所述相似度计算为加权余弦相似度计算。
第二方面，本申请提供一种建筑信息模型搜索装置，包括：
特征提取模块，用于对待搜索的模型库中的多构件组合级别的建筑信息模型进行多模态特征提取，得到各个建筑信息模型对应的多模态特征，所述多模态特征包括语义特征、拓扑特征和几何特征；
解析模块，用于获取用户输入的搜索文本，对所述搜索文本进行解析，得到所述搜索文本对应的搜索意图信息，所述搜索意图信息包括搜索意图语义特征、搜索意图拓扑特征和搜索意图几何特征；
相似度计算模块，用于基于深度嵌入学习，计算所述搜索意图信息与待搜索的模型库中的各个建筑信息模型的多模态特征的综合相似度；
推荐模块，用于根据所述搜索意图信息与待搜索的模型库中的各个建筑信息模型的多模态特征的综合相似度的排序结果，确定作为所述搜索文本对应的搜索结果的建筑信息模型，并向用户可视化展示所述搜索结果。
本申请由于采取以上技术方案，其具有以下优点：(1)实现了建筑级BIM模型的语义-拓扑-几何多模态特征搜索；(2)实现了良好的搜索效果，准确性显著提升；(3)优化了算法运行效率，保证了搜索速度。
附图说明
图1是本申请实施例中建筑信息模型搜索方法的流程示意图；
图2是基于IFC的模型信息解析的语义特征抽取程序流程图；
图3是基于IFC的建筑BIM模型属性信息提取示例；
图4是房间或庭院等空间单元之间的拓扑联通关系的处理方式
图5是基于邻接关系的拓扑联通特征抽取程序流程图；
图6是基于几何轮廓数据的建筑形状特征抽取方法示意；
图7是建筑BIM模型每层各房间平面轮廓特征的抽取方法示意；
图8是建筑BIM模型平面轮廓特征的抽取算法流程；
图9是基于文本分割与正则表达式的搜索意图解析流程；
图10是基于空间邻接联通特征的BIM模型拓扑特征嵌入流程图；
图11是基于轮廓与平面图特征的BIM模型形状嵌入程序流程图；
图12是ResNet50平面图形状特征嵌入模型框架示意图；
图13是基于综合相似度排序的建筑BIM模型搜索流程。
具体实施方式
为使本申请实施例的目的、技术方案和优点更加清楚，下面将结合本申请实施例的附图，对本申请实施例的技术方案进行清楚、完整地描述。显然，所描述的实施例是本申请的一部分实施例，而不是全部的实施例。基于所描述的本申请的实施例，本领域普通技术人员所获得的所有其他实施例，都属于本申请保护的范围。
针对现有技术亟需提供多构件组合级别(典型的，如整体建筑级别)的BIM模型的搜索方法的问题。本申请技术方案相应提供建筑信息模型搜索方法及装置，所述方法，包括：对待搜索的模型库中的多构件组合级别的建筑信息模型进行多模态特征提取，得到各个建筑信息模型对应的多模态特征，所述多模态特征包括语义特征、拓扑特征和几何特征；获取用户输入的搜索文本，对所述搜索文本进行解析，得到所述搜索文本对应的搜索意图信息，所述搜索意图信息包括搜索意图语义特征、搜索意图拓扑特征和搜索意图几何特征；基于深度嵌入学习，计算所述搜索意图信息与待搜索的模型库中的各个建筑信息模型的多模态特征的综合相似度；根据所述搜索意图信息与待搜索的模型库中的各个建筑信息模型的多模态特征的综合相似度的排序结果，确定作为所述搜索文本对应的搜索结果的建筑信息模型，并向用户推荐所述搜索结果。本方案，实现多构件组合级别的建筑信息模型的搜索。
请参阅本申请实施例的更多附图，在本申请的更多详细的实施例中进一步说明本申请提供的方法。
如图1，本申请提供一种建筑信息模型搜索方法，所述方法，包括：
S1，对待搜索的模型库中的多构件组合级别的建筑信息模型进行多模态特征提取，得到各个建筑信息模型对应的多模态特征，所述多模态特征包括语义特征、拓扑特征和几何特征；
S2，获取用户输入的搜索文本，对所述搜索文本进行解析，得到所述搜索文本对应的搜索意图信息，所述搜索意图信息包括搜索意图语义特征、搜索意图拓扑特征和搜索意图几何特征；
S3，基于深度嵌入学习，计算所述搜索意图信息与待搜索的模型库中的各个建筑信息模型的多模态特征的综合相似度；
S4，根据所述搜索意图信息与待搜索的模型库中的各个建筑信息模型的多模态特征的综合相似度的排序结果，确定作为所述搜索文本对应的搜索结果的建筑信息模型，并向用户推荐所述搜索结果。
下面在更多的详细实施例中，说明上述方法的具体原理和流程。
现有技术的模型搜索技术，只能实现单种构件的搜索，而本申请的目的是实现多构件组合的模型的搜索。多构件组合，指的是多个级别的构件进行组合。多构件组合的模型的一个典型的子集，就是整体建筑级别的建筑信息模型。在应用场景中，整体建筑，例如可以是乡村环境中的独栋别墅，其包含多层建筑空间，每层空间又可以包括若干不同类型的房间，各个房间中又包含不同的构件。建筑空间还可以包括露天的庭院等空间。为了便于说明，后续实施例，可以以整体建筑级别的模型的搜索为例来进行阐述。
本申请的方法的上述流程S1至S4实现了2个核心内容：其中S1实现了整体建筑级别BIM模型的多模态特征提取，而S2至S4实现了基于深度嵌入学习的BIM模型相似度检索。
下面分别叙述上述的2个核心内容。
核心内容1：整体建筑级别(或其它多构件组合级别的)BIM模型的多模态特征提取。
多模态特征，包括语义特征、拓扑特征和几何特征。
由于现有搜索算法及相关深度学习模型无法直接处理RVT、IFC等建筑BIM模型的文件格式。因此对整体建筑级别的BIM模型做搜索的第一个步骤便是对BIM模型中存在的属性文本、拓扑、几何等信息进行信息抽取。IFC标准是由buildingSMART组织制定的开源BIM标准，可采用各类开源解析工具对IFC格式的BIM模型进行信息抽取。本申请针对采用Revit软件创建，并导出为IFC4模型格式的建筑BIM模型，采用Python脚本进行数据解析。具体的，利用IfcOpenShell工具，使用“open”和“by_type”方法遍历获取模型中指定类型的组件，可从BIM模型中提取出建筑平面的几何形状、各房间之间的拓扑联通关系、墙壁门窗等组件的属性信息等，从而形成BIM模型多类特征的自动抽取算法。
对于整体建筑级别或多构件组合级别的BIM模型的语义特征，提取整体建筑级别或多构件组合级别的建筑信息模型中各个级别的组件的属性的语义信息，再对各个级别的组件的属性的语义信息进行汇总统计，得到整体建筑级别或多构件组合级别的建筑信息模型的属性的语义特征，如图2所示。具体包括：
1)提取建筑空间(Space)属性。
对描述户型每个空间信息的IfcSpace类进行解析，从基础的LongName属性中获取房间名称，从该房间关联构件的名称中获取该房间所处的具体楼层信息，从RelatingPropertyDefinition中查找到名为“面积”的属性即可获得面积信息，从IfcSpace.Representation.Representations.SweptArea中获取该房间的位置信息，从BoundedBy获取房间的轮廓信息，同时从BoundedBy的ifcRelSpaceBoundary属性可获取房间之间的分隔信息。
2)提取空间中墙体的分隔属性。
对每个房间中的IfcWall进行提取，获取房间内的墙体名称和类型信息，同时从IfcWall的RelatingPropertyDefinition中分别查找名为“宽度”，“长度”，“无连接高度”，“底部约束”，“尺寸标注”的属性，获取该墙体的全部属性信息。
3)提取墙体中门窗构件联通属性。
对于各个墙体，解析墙体关联的IfcWindow类获取全部窗户的基础信息，通过IfcWindow相关的RelatingPropertyDefinition获取窗户的全部属性信息；解析IfcDoor获取各扇门的基础信息，再读取IfcDoor的RelatingPropertyDefinition获取门的属性信息。
4)必要时可提取的其他属性信息。
必要时，可通过解析IfcSpace内的IfcFurniture来解析房间内的家具电器属性信息。
5)文本处理与汇总整理。
对属性文本进行预处理，删去无效信息，并进行同义词替换，规范文本表达，创建将上述属性语义信息存储到roomSpaceDic，wallDic，doorDic，windowDic，furDic等多个字典，实现相应子类到IFC原始文件数据的快速索引。
6)生成属性语义特征向量。
在抽取各类属性信息后，按照规则将BIM模型的属性信息汇总为特定形式的python属性信息词典，例如“{省：北京，面积：219，造价：600000，层数：3，房间数量：10，卧室数量：4，卫生间数量：3，厨房数量：1……}”，进而可将单个房屋的信息按照约定参数的方法处理为特征向量，用于后续的户型查询与匹配。上述整体流程如图3所示。
对于整体建筑级别或多构件组合级别的BIM模型的拓扑特征，根据各个级别的组件的属性，确定各个建筑空间之间的空间邻接关系，作为整体建筑级别或多构件组合级别的建筑信息模型的拓扑特征。
建筑BIM模型的拓扑联通特征指的是建筑中各个房间之间的相对位置关系和连接方式。本申请将房间的拓扑关系分为是否相邻以及是否相连两个类型进行处理，如图4所示，即任意两房间关系仅有不相邻、相邻且不相连、相连三个可能性。在BIM模型中除了墙体可以分隔房间以外，虚拟房间分割(Virtual Room Separator)同样可以分隔房间，本申请通过抽取IfcSpace的IfcRelSpaceBoundary中的PhysicalOrVirtualBoundary属性来区分房间分隔是否为物理分隔。算法流程如图5所示，具体步骤如下
1)处理房间IfcSpace与墙体IfcWall的空间关系，提取相邻关系。
1.1)抽取各房间的空间位置
通过抽取IfcSpace属性信息，获取各个房间的空间信息。提取Representation.SweptArea.Position的房间局部坐标位置和方向，同时提取轮廓点的相对坐标，可以在整个户型的全局坐标系中计算出每个房间每个轮廓点的绝对坐标位置。
1.2)抽取房间空间轮廓中各墙体的空间位置轮廓。
通过抽取IfcWall，获取各个房间的分隔信息。基于各个房间IfcSpace.Boundedby可提取出该房间轮廓上的IfcWall墙体，在IfcWall的ObjectPlacement.RelativePlacement中通过坐标转换的方式计算出墙体每个边界点的绝对坐标。
1.3)房间相邻关系判定。
若两个IfcSpace房间含有共同的IfcWall边界，且存在轮廓点连线坐标重合的情况，则可以认为两个IfcSpace房间存在相邻关系，将相邻关系存入adjacentDic词典。
2)从相邻的房间中，处理门IfcDoor与房间和墙体的关联关系，抽取相连信息。
2.1)抽取门构件的从属关系以及对应的房间分隔信息。
提取所有IfcDoor的名称与全局编号，再从各个IfcDoor的ProvidesBoundaries中提取出对应墙壁分隔的房间信息。
2.2)房间相连关系判定
若IfcDoor的ProvidesBoundaries中具备两个独立的IfcSpace房间，同时这两个房间的RelativePlacement中可以判定他们存在共同的IfcWall，且该IfcDoor位于此IfcWall中，则可认为该IfcDoor的ProvidesBoundaries中连接的两个IfcSpace房间具备相连关系，将相连关系存入accessDic词典。
3)验证联通关系，整理并存储。
在获取了相邻和相连关系词典后，可以验证所有的相连关系都满足相邻，且墙体与门的厚度应符合规定值，排除信息提取错误与建模错误的影响，最后将词典转化为Networkx拓扑图的形式进行储存。
进一步的，提取建筑信息模型的平面轮廓信息，作为整体建筑级别或多构件组合级别的建筑信息模型的几何特征。
具体的，与属性信息和拓扑关系不同，房屋的几何形状特征涉及二维与三维两类信息，较难直接通过直接解析BIM模型文件进行信息抽取和特征抽象。由于三维几何特征的复杂性，本申请侧重于分析建筑BIM模型的二维几何形状特征。通常，房屋(本申请的整体建筑级别的BIM模型的一个实例)的几何特征可以由建筑的空间布局和平面轮廓信息来确定。前者可以直接由平面图反映，而后者可以通过提取表示建筑物轮廓的坐标列表来获得。它们都包含丰富的几何信息，因此本申请在提取几何特征的任务中综合考虑了这两种类型的信息，如图6所示，具体步骤如下。
1)建筑BIM模型线框透视平面图抽取
1.1)建筑BIM模型线框平面图截取
从Revit软件中截取BIM模型的1层平面图作为户型整体的形状数据，通过直接采用建筑BIM模型的俯视视角，固定户型的坐标方位，截取线框格式的平面图，可以获得户型的平面图信息。
1.2)建筑BIM模型线框平面图图像预处理
对平面图进行图像预处理，固定建筑朝向，通过缩放的方式对尺寸进行归一化存储。
2)建筑BIM模型平面轮廓特征抽取
2.1)建筑BIM模型房间轮廓信息抽取
对于建筑BIM模型中每个房间的几何轮廓信息，采用IfcOpenShell抽取户型每个房间的Representation.SweptArea.Position的局部坐标，形成房间轮廓信息列表。
2.2)建筑BIM模型整体轮廓信息换算
通过每个房间的轮廓信息列表以及局部坐标系信息，反算出整体轮廓点的全局坐标，并形成建筑BIM模型的整体平面轮廓数据列表，构造出整体建筑的外轮廓数据，如图7所示。
2.3)轮廓特征归一化存储
将整体建筑的外轮廓点坐标列表拟合为固定轮廓点数的外轮廓多边形，从而将外轮廓点信息用定长向量进行统一存储，上述整体算法流程如图8所示。
核心内容2：基于深度嵌入学习的BIM模型相似度检索。
从建筑BIM模型中提取出属性、拓扑与几何的多模态特征后，文本到BIM模型的搜索需要搜索文本与BIM模型之间的相似度计算。因此，此部分首先使用自然语言处理的文本分割与正则表达式手段来解析搜索文本，以提取自然文本中蕴含的搜索意图。其次，利用深度学习和特工程工具，将BIM模型的多模态特征和搜索意图分别嵌入到搜索文本中，并转化为统一的特征向量。再次，基于特征向量进行加权综合相似度计算，基于相似度排序实现BIM模型的智能搜索和推荐。以下分三部分介绍详细步骤。
(1)基于文本分割与正则表达式的搜索意图解析
提取搜索意图的目的是从文本信息中获得所需BIM模型的语义属性、房间拓扑关系、几何描述等信息，从而依靠这些信息查询目标模型，其流程如图9，其具体步骤如下所示。
1)定义建筑BIM模型领域词表
收集与住房和地产领域相关的描述性名词和常用表达，编译来创建用于建筑BIM模型搜索的领域单词列表，这有助于文本分割并定义专有名词细分的最小单位。
2)基于Jieba库GRU神经网络模式进行领域文本分词
Jieba为常用的Python语言开源中文分割应用程序包，此处基于领域词表和Jieba对搜索文本进行分割，利用Jieba的paddlepaddle模式，基于门控循环网络，根据领域单词表，将搜索语句分割成具有单词属性和单词顺序的分段，以便于后续的属性和连词匹配。
3)基于正则表达式的搜索意图解析
通过手动定义的正则表达式来提取搜索意图中的语义、拓扑和几何特征的描述，部分正则表达式的样式如表1所示。具体如以下步骤所示。
表1
3.1)基于正则表达式的搜索意图语义特征解析
对于所需BIM模型的属性和语义信息，提取搜索语句中的各种描述性子句，并根据句法结构匹配其中的名词、动词、介词等关键词，例如当出现“area”和“is”时，提取“is”后的数字作为房屋的面积进行搜索，最后将相应属性和语义汇总为词典。
3.2)基于正则表达式的搜索意图拓扑特征解析
通过与3.1)类似的方法，提取房间的拓扑连通性特征，例如“客厅连厨房”可以提取代表房间名称的关键词“客厅”和“厨房”，以及表征其连通性的词“连通”，从而提取与拓扑连通性相关的语义信息，最后汇总为连通性词典与NetworkX网络图。
3.3)基于正则表达式的搜索意图几何特征解析
对于形状特征，从检索文本中抽取“户型方正”，“卧室狭长”类似的模糊形状描述，生成对应轮廓特征列表，包含外接矩形长细比等参数，最后汇总为形状属性词典。
4)搜索意图解析汇总
最后将正则表达式提取出的各类特征词典进行汇总，以便于后续搜索。
(2)基于深度嵌入学习的多模态特征向量化表示
在提取BIM模型的语义、拓扑和几何特征信息以及搜索意图的语义、拓扑和几何特征后，需要将它们嵌入到统一的矢量化表示中进行相似性计算。总体上，每个属性的语义信息直接组织成一个由统一格式的文本和数据组成的向量，用于嵌入语义特征。拓扑信息以Networkx拓扑图的形式存储，并使用图核方法实现拓扑特征嵌入。对于几何特征的嵌入，平面图图像通过卷积神经网络进行嵌入，轮廓信息通过轮廓的特征参数进行嵌入，搜索文本中的几何形状描述也作为房屋轮廓的特征参数进行嵌入。具体方法包括：
1)基于人工规则与Word2Vec神经网络结合的语义特征的嵌入
5.2.1部分的BIM模型多模态特征提取方法可以提取BIM模型的语义属性为“{省：江苏，面积：149平方米，成本：100000美元，楼层数：2，卧室数：3，浴室数：2、客厅数：1，厨房数：1…}”，本部分(1)中的搜索意图提取方法也可以生成相应形式的属性字典。Word2Vec是自然语言处理任务中常用的一类将词转化为词基向量的神经网络模型。本申请使用手动定义的转换规则并结合Word2Vec神经网络的方法属性字典嵌入到语义特征向量中，处理方法参照表2，具体步骤如下所示：
表2
1.1)同近义词标准化处理
具体来说，对于可能在语义词典中广泛存在的近同义词，如“浴室”、“盥洗室”、“卫生”、“厕所”等，通过手动定义近同义词列表并将其描述统一为同一描述性词典来进行同义词替换。
1.2)基于人工规则的属性数据特征嵌入
对于属性中的数据信息，其可以直接通过数值差异进行详细性比较，因此嵌入相对简单。数据属性进行加权和归一化后，采用人工定义规则的方式，对不同类型数据依照其类型特征进行嵌入。
1.3)基于Word2Vec的属性语义特征嵌入
对于属性中的语义信息，其文本通常为字符串属性，需要嵌入为字符串向量再通过Word2Vec模型进行嵌入。具体的，Word2Vec模型是由Gensim(一个流行的用于自然语言处理的Python库)包实现。在1.1)中定义的同义词列表的基础上，构建了一个相应的文本语料库作为训练集，通过该语料库，使用训练参数“vector_size＝100，window＝5，min_count＝1，worker＝4”来训练Word2Vec模型。
1.4)混合数据与语义的特征嵌入
其他类型的语义信息按照规则划分为文本信息和数据信息，分别作为字符串词向量和数据向量嵌入，然后作为特征向量进行加权和组合，相应的权重由专家手动确定。
2)基于图神经网络的拓扑特征嵌入
此前从BIM模型与搜索文本中提取出的空间拓扑关系均可转化为NetworkX拓扑图进行存储。图核是图神经网络中一种有效的图特征嵌入的方法，因此选择了基于图核的Deepwalk深度随机行走方法来嵌入拓扑特征，这种方法是无监督的，具有更好的迁移性，更适合于本申请中的图数据。其程序流程如图10所示，具体嵌入方法如下。
2.1)基于Networkx与Grakel将拓扑图转化为网络图
Networkx库是图数据格式存储与共享的常用开源python库，Grakelze则是图卷积中图内核机器学习任务的常用开源python库。此处基于NetworkX和Grakel进行实现。基于Networkx拓扑图的节点列表、节点属性列表和邻接矩阵，使用Grakel.graph方法将Networkx拓扑图形转换为Grakel图网络数据。
2.2)基于随机游走核的网络图特征向量化
通过随机游走的形式，固定相同的Graph Kernel，基于Grakel.GraphKernel方法，将Grakel图网络嵌入为拓扑特征向量，再进行加权归一化处理，得到衡量户型整体户型联通关系相似程度的BIM模型拓扑特征向量。
3)基于轮廓视觉特征与卷积神经网络的几何形状特征嵌入
此前已经将BIM模型的几何形状信息分为轮廓数据信息与平面图数据信息进行处理，因此分别嵌入两类特征，起到互相补充的作用。其中从BIM模型中抽取的几何轮廓不仅包含总体的户型外轮廓信息，还包含户型中每一类房间在平面上的位置信息，因此可直接基于户型全部平面轮廓信息构造形状描述特征，对户型形状进行向量化。同时基于卷积神经网络，从户型平面图中提取户型形状的特征向量，从而脱离单一轮廓特征的噪声与变换的影响，对户型的整体平面布局形状进行向量化。其嵌入流程如图11所示，具体嵌入方法如下。
3.1)基于轮廓视觉特征的几何轮廓特征抽取
采用OpenCV计算机视觉工具包，对此前提取的BIM模型轮廓信息进行特征抽取。通过轮廓信息计算对户型与其内部房间形状描述的指标，包括外包络矩形的长宽比、轮廓内面积与外包络矩形的面积比、户型形心坐标、各房间信息坐标等，再将数据归一化形成描述几何轮廓特征的户型形状特征描述指标，如表3所示。具体如下：
表3
3.1.1)建筑BIM模型外轮廓坐标拟合嵌入
通过cv2.approxPolyDP拟合多边形方法，固定拟合多边形轮廓点数为30，将户型的外轮廓拟合为30边形，从而将外轮廓坐标归一化，形成定长的轮廓点向量，实现外轮廓信息的嵌入。
3.1.2)建筑BIM模型轮廓长宽比特征嵌入
采用cv2.boundingRect方法计算轮廓的外接矩形，通过外接矩形的长宽比衡量整体建筑BIM模型平面形状的方正程度，以浮点数的形式嵌入。
3.2.3)建筑BIM模型轮廓图像据特征嵌入
通过cv2.moments方法计算户型轮廓的图像矩，从而从图像矩中表征建筑BIM模型的质心、惯性矩、三阶矩等特征。
3.2.4)建筑BIM模型你和椭圆特征嵌入
同时考虑到实际存在形状比较狭长的户型，采用拟合椭圆对其形状特征进行嵌入，通过cv2.fitEllipse方法得出与建筑轮廓最接近的椭圆，提取椭圆的长短轴向量特征，以定长向量的形式嵌入。
3.2.5)建筑BIM模型内部细分空间形状特征嵌入
对于建筑BIM模型内部的房间和院落等细分空间的形状，则可通过户型中该类房间的平均外接矩形长宽比衡量其方正程度，通过cv2.boundingRect方法分别计算各卧室、客厅、厨房、院落的平均长宽比信息，形成四类细分空间的长宽比特征，以浮点数的形式嵌入。
3.2)基于卷积神经网络的平面图数据信息嵌入
对于建筑BIM模型平面图的特征抽取，本申请引入了基于卷积神经网络的特征工程技术，用预训练的ResNet50模型来抽取平面整体的几何特征，具体如下。
3.2.1)卷积神经网络结构选型
在卷积神经网络模型的选择与构建上，选用ResNet50神经网络，调整神经网络的输入输出层结构，将输入统一为224×224格式的三通道图片，输出统一为2048维度的图片形状特征向量，其神经网络结构如图12所示。
3.2.2)基于预训练模型的平面图数据嵌入
选用广泛运用的ImageNet对ResNet50网络结构进行预训练，基于ImageNet数据集预训练好的CheckPoint权重生成模型，运行并提取全连接层数据，即可输出平面图的2048维度的图片形状特征向量。
3.3)建筑BIM模型轮廓与平面图特征嵌入集成
将上述平面图特征向量与通过建筑BIM模型轮廓特征指标进行加权结合，即形成了综合户型轮廓与平面图形状特征两类数据的形状特征向量。两类特征嵌入的权重通过专家人工确定，并根据效果进行人工3轮调整。
(3)基于综合相似度排序的建筑BIM模型搜索
此前已实现了BIM模型语义、拓扑‘几何多模态特征和搜索意图信息的统一矢量化嵌入方法。为了进行从文本到BIM模型的搜索，本小节根据BIM模型的特征向量和搜索意图进行综合相似度计算，并通过加权余弦相似度排序实现对BIM模型的检索，其总体流程如图13所示，具体方法如下。
3.1)各相似性评估指标的确定
通过特征向量的加权余弦相似度，可以很容易地比较BIM模型与模型、文本与BIM模型之间的相似度，然后通过相似度排序实现BIM模型的智能检索。然而，特征向量的元素具有不同的组成、大小和含义，因此有必要使用适当的相似性评估指标来计算向量的不同元素的相似性，几类典型相似性指标处理方法可参照表4所示，具体如下：
表4
3.1.1)字符串元素的相似性指标
对于字符串元素，例如位置、房间的名称等，字符匹配率被用作相似性的度量；对于浮点和整数量化元素，如卧室面积和数量，使用相对差异来衡量相似性。
3.1.2)子向量元素的相似性指标
对于具有特定意义的子向量元素，如拓扑图的词向量和拟合矩形特征、形状的像矩等，还通过取子向量的余弦相似度来度量相似度。
3.1.3)坐标列表元素的相似性指标
数据列表对于形状特征中的拟合外轮廓坐标列表，引入OpenCV的cv2.matchShapes方法，通过轮廓的Hu矩的相对差异来比较轮廓的相似性。
3.2)各详细性指标集成权重的确定
除了相似性指标，计算综合相似性的另一个主要重点是确定每类特征的权重。考虑到搜索数据集的大小，本研究使用专家手动反馈调节来建立权重，具体如下。
3.2.1)确定初始权重
初始权重首先由专家手动确定，然后根据综合相似度按降序排序，形成BIM模型搜索结果列表。
3.2.2)基于人工反馈的权重调整
综合相似度的权重通过专家手动评估搜索结果的方式进行反馈调整。经过10轮以上的人工比对调整，综合相似度结果满足相应专家评价。
3.2.3)得出最终权重并形成相似度排序算法
语义特征权重值的百分比约为88％，拓扑特征权重为5％，几何特征权重为7％。通过加权特征向量余弦相似度排序，形成相应的BIM图库搜索排序算法，具有计算速度快的优点，可以通过调整权重来优化搜索效果。
综上，本申请的方法，实现了如下的效果：
(1)实现了整体建筑级别或多构件组合级别BIM模型的语义-拓扑-几何多模态特征搜索，支持搜索的模型规模大，搜索范围广。基于IfcOpenShell实现了整体建筑BIM模型的语义特征、拓扑联通特征与形状特征的抽取与深度嵌入，首次将BIM模型搜索算法适用的模型规模拓展到了整体建筑级别，同时首次实现了文本到建筑BIM模型语义-拓扑-几何三类特征的相似性搜索。
(2)保证(1)的前提下，实现了良好的搜索效果，准确性显著提升。引入深度嵌入式学习，嵌入“语义-拓扑-几何”多模态特征开发了基于综合相似度的建筑级BIM图集智能检索算法。相应智能搜索方法在典型工程项目中已得到良好的应用验证。就搜索准确性与搜索排序效果而言，在测试集中算法的mNDCG1和mNDCG5指标均达到90％左右，实现了与专家人工检索几乎相同的检索质量和检索效果，相比传统仅考虑语义的检索算法搜索效果显著提升了6.5％以上，让好的图集不仅找得到，更找得准、找得全、找得好。
(3)保证(1)和(2)的搜索质量前提下，优化了算法运行效率，保证了搜索速度。在测试集中，在保证检索效果的前提下，三名专家检索50个搜索语句的平均用时为1740秒，而本申请可实现与专家检索类似的检索质量，同时算法可以在普通民用笔记本上流畅运行，同时算法仅用时为2.23秒(硬件环境为M1 Pro，16GB内存)，检索效率相对人工检索有着数百倍的效率提升，总耗时缩短近3个数量级。因此，在保证检索质量的应用环境下，本申请可显著减少人工劳动，使专家与设计者可以更专注于难度更高的人工工作。
本申请另一方面提供一种建筑信息模型搜索装置，包括：
特征提取模块，用于对待搜索的模型库中的整体建筑级别或多构件组合级别的建筑信息模型进行多模态特征提取，得到各个建筑信息模型对应的多模态特征，所述多模态特征包括语义特征、拓扑特征和几何特征；
解析模块，用于获取用户输入的搜索文本，对所述搜索文本进行解析，得到所述搜索文本对应的搜索意图信息，所述搜索意图信息包括搜索意图语义特征、搜索意图拓扑特征和搜索意图几何特征；
相似度计算模块，用于基于深度嵌入学习，计算所述搜索意图信息与待搜索的模型库中的各个建筑信息模型的多模态特征的综合相似度；
推荐模块，用于根据所述搜索意图信息与待搜索的模型库中的各个建筑信息模型的多模态特征的综合相似度的排序结果，确定作为所述搜索文本对应的搜索结果的建筑信息模型，并向用户推荐所述搜索结果。
本申请实施例还提供了一种计算机可读存储介质，存储介质包括存储的程序，其中，在程序运行时控制存储介质所在设备执行上述的方法。其具体的实现过程，在此不再重复赘述。
本申请实施例还提供一种计算机设备。该实施例的计算机设备包括：处理器、存储器以及存储在存储器中并可在处理器上运行的计算机程序，处理器执行计算机程序时实现实施例中的前述方法，为避免重复，此处不一一赘述。或者，该计算机程序被处理器执行时实现实施例中装置中各模型中/单元的功能，为避免重复，此处不一一赘述。
计算机设备可以是桌上型计算机、笔记本、掌上电脑、服务器及云端服务器等计算设备。计算机设备可包括，但不仅限于，处理器、存储器。本领域技术人员可以理解，可以包括比图示更多或更少的部件，或者组合某些部件，或者不同的部件，例如计算机设备还可以包括输入输出设备、网络接入设备、总线等。
所称处理器可以是中央处理单元(Central Processing Unit，CPU)，还可以是其他通用处理器、数字信号处理器(Digital Signal Processor，DSP)、专用集成电路(Application Specific Integrated Circuit，ASIC)、现场可编程门阵列(Field-Programmable Gate Array，FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件等。通用处理器可以是微处理器或者该处理器也可以是任何常规的处理器等。
存储器可以是计算机设备的内部存储单元，例如计算机设备的硬盘或内存。存储器也可以是计算机设备的外部存储设备，例如计算机设备上配备的插接式硬盘，智能存储卡(Smart Media Card,SMC)，安全数字(Secure Digital,SD)卡，闪存卡(Flash Card)等。进一步地，存储器还可以既包括计算机设备的内部存储单元也包括外部存储设备。存储器用于存储计算机程序以及计算机设备所需的其他程序和数据。存储器还可以用于暂时地存储已经输出或者将要输出的数据。
所属领域的技术人员可以清楚地了解到，为描述的方便和简洁，上述描述的系统，装置和单元的具体工作过程，可以参考前述方法实施例中的对应过程，在此不再赘述。
在本申请所提供的几个实施例中，应该理解到，所揭露的系统，装置和方法，可以通过其它的方式实现。例如，以上所描述的装置实施例仅仅是示意性的，例如，上述单元的划分，仅仅为一种逻辑功能划分，实际实现时可以有另外的划分方式，例如，多个单元或组件可以结合或者可以集成到另一个系统，或一些特征可以忽略，或不执行。另一点，所显示或讨论的相互之间的耦合或直接耦合或通信连接可以是通过一些接口，装置或单元的间接耦合或通信连接，可以是电性，机械或其它的形式。
上述以软件功能单元的形式实现的集成的单元，可以存储在一个计算机可读取存储介质中。上述软件功能单元存储在一个存储介质中，包括若干指令用以使得一台计算机装置(可以是个人计算机，服务器，或者网络装置等)或处理器(Processor)执行本申请各个实施例上述方法的部分步骤。而前述的存储介质包括：U盘、移动硬盘、只读存储器(Read-Only Memory，ROM)、随机存取存储器(Random Access Memory，RAM)、磁碟或者光盘等各种可以存储程序代码的介质。
以上上述仅为本申请的较佳实施例而已，并不用以限制本申请，凡在本申请的精神和原则之内，所做的任何修改、等同替换、改进等，均应包含在本申请保护的范围之内。

Claims (11)

  1. 一种建筑信息模型搜索方法，其特征在于，所述方法，包括：
    对待搜索的模型库中的多构件组合级别的建筑信息模型进行多模态特征提取，得到各个建筑信息模型对应的多模态特征，所述多模态特征包括语义特征、拓扑特征和几何特征；
    获取用户输入的搜索文本，对所述搜索文本进行解析，得到所述搜索文本对应的搜索意图信息，所述搜索意图信息包括搜索意图语义特征、搜索意图拓扑特征和搜索意图几何特征；
    基于深度嵌入学习，计算所述搜索意图信息与待搜索的模型库中的各个建筑信息模型的多模态特征的综合相似度；
    根据所述搜索意图信息与待搜索的模型库中的各个建筑信息模型的多模态特征的综合相似度的排序结果，确定作为所述搜索文本对应的搜索结果的建筑信息模型，并向用户推荐所述搜索结果。
  2. 根据权利要求1所述的建筑信息模型搜索方法，其特征在于，所述对待搜索的模型库中的多构件组合级别的建筑信息模型进行多模态特征提取，包括：
    提取多构件组合级别的建筑信息模型中各个级别的组件的属性的语义信息，再对各个级别的组件的属性的语义信息进行汇总统计，得到多构件组合级别的建筑信息模型的属性的语义特征。
  3. 根据权利要求2所述的建筑信息模型搜索方法，其特征在于，所述各个级别的组件，包括：建筑空间，可包含于建筑空间中的墙体，以及可包含于墙体中的门或窗等。
  4. 根据权利要求3所述的建筑信息模型搜索方法，其特征在于，所述方法对待搜索的模型库中的多构件组合级别的建筑信息模型进行多模态特征提取，还包括：
    根据各个级别的组件的属性，确定各个建筑空间之间的空间邻接关系，作为多构件组合级别的建筑信息模型的拓扑特征。
  5. 根据权利要求4所述的建筑信息模型搜索方法，其特征在于，所述空间邻接关系包括：不相邻、相邻且不相连，以及相连三种关系。
  6. 根据权利要求2所述的建筑信息模型搜索方法，其特征在于，所述对待搜索的模型库中的多构件组合级别的建筑信息模型进行多模态特征提取，还包括：
    提取建筑信息模型的平面轮廓信息，作为多构件组合级别的建筑信息模型的几何特征。
  7. 根据权利要求1所述的建筑信息模型搜索方法，其特征在于，所述对所述搜索文本进行解析，得到所述搜索文本对应的搜索意图信息，包括：
    基于自然语言处理的文本分割以及正则表达式进行解析，得到搜索意图信息的搜索意图语义特征、搜索意图拓扑特征和搜索意图几何特征。
  8. 根据权利要求7所述的建筑信息模型搜索方法，其特征在于，所述基于深度嵌入学习，计算所述搜索意图信息与待搜索的模型库中的各个建筑信息模型的多模态特征的综合相似度，包括：
    根据提取的建筑信息模型的语义特征、拓扑特征和几何特征，以及搜索意图的搜索意图语义特征、搜索意图拓扑特征和搜索意图几何特征，嵌入到统一的矢量化表示中进行相似度计算。
  9. 根据权利要求8所述的建筑信息模型搜索方法，其特征在于，所述相似度计算为加权余弦相似度计算。
  10. 一种建筑信息模型搜索装置，其特征在于，包括：
    特征提取模块，用于对待搜索的模型库中的多构件组合级别的建筑信息模型进行多模态特征提取，得到各个建筑信息模型对应的多模态特征，所述多模态特征包括语义特征、拓扑特征和几何特征；
    解析模块，用于获取用户输入的搜索文本，对所述搜索文本进行解析，得到所述搜索文本对应的搜索意图信息，所述搜索意图信息包括搜索意图语义特征、搜索意图拓扑特征和搜索意图几何特征；
    相似度计算模块，用于基于深度嵌入学习，计算所述搜索意图信息与待搜索的模型库中的各个建筑信息模型的多模态特征的综合相似度；
    推荐模块，用于根据所述搜索意图信息与待搜索的模型库中的各个建筑信息模型的多模态特征的综合相似度的排序结果，确定作为所述搜索文本对应的搜索结果的建筑信息模型，并向用户可视化展示所述搜索结果。
  11. 一种计算机可读存储介质，其特征在于，存储有计算机程序，所述计算机程序被处理器执行，控制所述处理器所在设备实现如下的建筑信息模型搜索方法：
    对待搜索的模型库中的多构件组合级别的建筑信息模型进行多模态特征提取，得到各个建筑信息模型对应的多模态特征，所述多模态特征包括语义特征、拓扑特征和几何特征；
    获取用户输入的搜索文本，对所述搜索文本进行解析，得到所述搜索文本对应的搜索意图信息，所述搜索意图信息包括搜索意图语义特征、搜索意图拓扑特征和搜索意图几何特征；
    基于深度嵌入学习，计算所述搜索意图信息与待搜索的模型库中的各个建筑信息模型的多模态特征的综合相似度；
    根据所述搜索意图信息与待搜索的模型库中的各个建筑信息模型的多模态特征的综合相似度的排序结果，确定作为所述搜索文本对应的搜索结果的建筑信息模型，并向用户推荐所述搜索结果。
PCT/CN2024/138681 2023-12-15 2024-12-12 一种建筑信息模型搜索方法、装置及存储介质 Pending WO2025124458A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US19/142,948 US20260119729A1 (en) 2023-12-15 2024-12-12 Building information modeling search method, apparatus, and storage medium

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202311730705.4 2023-12-15
CN202311730705.4A CN117786044B (zh) 2023-12-15 2023-12-15 一种建筑信息模型搜索方法及装置

Publications (1)

Publication Number Publication Date
WO2025124458A1 true WO2025124458A1 (zh) 2025-06-19

Family

ID=90390152

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2024/138681 Pending WO2025124458A1 (zh) 2023-12-15 2024-12-12 一种建筑信息模型搜索方法、装置及存储介质

Country Status (3)

Country Link
US (1) US20260119729A1 (zh)
CN (1) CN117786044B (zh)
WO (1) WO2025124458A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN120995567A (zh) * 2025-09-10 2025-11-21 济南城建动能转换开发建设集团有限公司 一种基于智能加载的大体量bim模型优化系统
US12632663B1 (en) * 2025-09-12 2026-05-19 AtomBeam Technologies Inc. Persistent cognitive machine with an advanced distributed thought cache

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117786044B (zh) * 2023-12-15 2024-08-13 清华大学 一种建筑信息模型搜索方法及装置
CN118114127B (zh) * 2024-04-30 2024-07-02 四川能投建工集团有限公司 基于bim的建筑模型数据分析方法及系统
CN119939687A (zh) * 2024-12-02 2025-05-06 华为云计算技术有限公司 工业模型生成方法和工业模型生成装置
CN121029906A (zh) * 2025-10-28 2025-11-28 中国地质科学院地质力学研究所 基于ai的地质gis模型快速检索方法及系统

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110218777A1 (en) * 2010-03-03 2011-09-08 Honeywell International Inc. System and method for generating a building information model
CN110968713A (zh) * 2019-11-25 2020-04-07 盈嘉互联（北京）科技有限公司 Bim多尺度信息检索方法及装置
CN116304225A (zh) * 2023-03-22 2023-06-23 联洋国融(上海)科技有限公司 一种基于mas系统的构件检索及评价方法
CN116451317A (zh) * 2023-03-30 2023-07-18 北京构力科技有限公司 一种建筑运维方法、装置、设备及存储介质
CN117786044A (zh) * 2023-12-15 2024-03-29 清华大学 一种建筑信息模型搜索方法及装置

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2016079792A1 (ja) * 2014-11-17 2016-05-26 株式会社 東芝 検索装置および検索方法
CN111797201A (zh) * 2020-06-23 2020-10-20 中民筑友建设科技集团有限公司 Bim模型的获取方法、装置、设备及计算机可读存储介质

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20110218777A1 (en) * 2010-03-03 2011-09-08 Honeywell International Inc. System and method for generating a building information model
CN110968713A (zh) * 2019-11-25 2020-04-07 盈嘉互联（北京）科技有限公司 Bim多尺度信息检索方法及装置
CN116304225A (zh) * 2023-03-22 2023-06-23 联洋国融(上海)科技有限公司 一种基于mas系统的构件检索及评价方法
CN116451317A (zh) * 2023-03-30 2023-07-18 北京构力科技有限公司 一种建筑运维方法、装置、设备及存储介质
CN117786044A (zh) * 2023-12-15 2024-03-29 清华大学 一种建筑信息模型搜索方法及装置

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN120995567A (zh) * 2025-09-10 2025-11-21 济南城建动能转换开发建设集团有限公司 一种基于智能加载的大体量bim模型优化系统
US12632663B1 (en) * 2025-09-12 2026-05-19 AtomBeam Technologies Inc. Persistent cognitive machine with an advanced distributed thought cache

Also Published As

Publication number Publication date
CN117786044A (zh) 2024-03-29
CN117786044B (zh) 2024-08-13
US20260119729A1 (en) 2026-04-30

Similar Documents

Publication Publication Date Title
US20260119729A1 (en) Building information modeling search method, apparatus, and storage medium
KR102564144B1 (ko) 텍스트 관련도를 확정하기 위한 방법, 장치, 기기 및 매체
CN106682059B (zh) 根据图像的结构化的知识建模和提取
US9613024B1 (en) System and methods for creating datasets representing words and objects
CN106682060B (zh) 根据图像的结构化的知识建模、提取和局部化
Li et al. BIMSeek++: Retrieving BIM components using similarity measurement of attributes
WO2021139262A1 (zh) 文献主题词聚合方法、装置、计算机设备及可读存储介质
CN108399163A (zh) 结合词聚合与词组合语义特征的文本相似性度量方法
CN114550195A (zh) 用于从建筑图纸中提取楼层平面图元素的机器学习技术
CN106547739A (zh) 一种文本语义相似度分析方法
CN112365939B (zh) 一种基于医疗健康大数据的数据治理方法及系统
CN111325235B (zh) 面向多语种的通用地名语义相似度计算方法及其应用
WO2023124160A1 (zh) 一种自动生成三维房屋布局的方法、系统、装置及介质
Lu et al. Organizational graph generation for structured architectural floor plan dataset
US20090094010A1 (en) Three dimensional surface and solid query processing
CN108416008A (zh) 一种基于自然语言处理的bim产品数据库语义检索方法
Wang et al. Query answering system for building information modeling using BERT NN Algorithm and NLG
CN108959314A (zh) 一种语义检索方法和装置
CN118917323B (zh) 一种结合大模型的金融文档检测方法及系统
Iyer et al. A multi-scale hierarchical 3D shape representation for similar shape retrieval
CN112347289A (zh) 一种图像管理方法及终端
CN110457484B (zh) 一种基于图的逻辑表达方法、系统、介质和设备
CN119862878B (zh) 基于多关系深度检索文本匹配的人名消歧方法
CN116756267A (zh) 事件时序关系识别方法、装置、设备及介质
Huang et al. Intelligent BIM Searching via Deep Embedding of Geometric, Semantic, and Topological Features

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 24902875

Country of ref document: EP

Kind code of ref document: A1