BIM data automatic classification and tagging method based on intelligent rules

Abstract

The invention discloses a BIM (Building Information Modeling) data automatic classification and tagging method based on an intelligent rule. The method comprises the following steps: S1, data preprocessing; s2, feature extraction; s3, intelligent classification and tagging are carried out; s4, verifying and optimizing a result; and S5, user interaction and report generation. According to the method, efficient processing of BIM data is realized by integrating a multi-dimensional feature fusion technology and an advanced deep learning algorithm. BIM data generally comprises various complex information such as huge three-dimensional models, design drawings, material lists and the like, and a traditional method is often low in efficiency and prone to making mistakes when processing the data. The feature fusion technology adopted by the invention can effectively integrate information from different dimensions, and automatically learn and extract key features through a deep learning model, thereby greatly improving the accuracy and efficiency of data processing.

Description

Technical Field

[0001] The present invention relates to the technical field of building information modeling (BIM), and in particular to an automatic classification and labeling method for BIM data based on intelligent rules. Background Art

[0002] With the rapid development of the construction industry and the deepening of digital transformation, Building Information Modeling (BIM) technology, as one of its core driving forces, is gradually becoming a key force in promoting industry change. BIM technology integrates the entire life cycle information of construction projects, including data from various stages such as design, construction, and operation and maintenance, providing project participants with a comprehensive, accurate, and visual information exchange platform. However, with the widespread popularity of BIM applications, the amount of data in BIM projects has exploded, and the data types are complex and diverse, including but not limited to CAD drawings, 3D models, document materials, sensor data, etc. The efficient management and utilization of these data has become a major challenge for the in-depth application of BIM technology.

[0003] The classification and labeling of traditional BIM data mainly rely on manual operations, which is not only time-consuming and labor-intensive, but also prone to errors. Especially when faced with large-scale, multi-source data, its efficiency and accuracy are even more difficult to guarantee. In addition, due to the different BIM data standards and data formats of different projects and stages, the difficulty of data integration and utilization is further increased. Therefore, how to achieve automatic, intelligent classification and labeling of BIM data has become an important issue that needs to be solved in the development of BIM technology.

[0004] In this context, the present invention proposes an automatic classification and labeling method for BIM data based on intelligent rules, aiming to build an efficient and accurate BIM data automatic processing system by integrating advanced data processing technology, machine learning algorithms and rule engines. The proposal of this method is not only an innovation of the traditional BIM data management method, but also a powerful promotion for the in-depth application and popularization of BIM technology. Summary of the invention

[0005] The purpose of the present invention is to provide a method for automatic classification and labeling of BIM data based on intelligent rules to solve the technical problems existing in the prior art.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions: The present invention provides a method for automatic classification and labeling of BIM data based on intelligent rules, comprising the following steps: S1. Data preprocessing, including: S11. Data collection and integration: Collect BIM data from various data sources and integrate them into a unified database through API interface or file import to provide a basis for subsequent processing; S12. Data cleaning: Improve the quality and availability of data and provide a reliable data foundation for subsequent model building and application; S13, Data enhancement: For image data, use rotation, scaling, cropping or flipping operations to increase sample diversity; for text data, enrich text expression through synonym replacement and sentence reorganization; S2. Feature extraction, including: S21. Text feature extraction: For text data, natural language processing technology is used to extract features; S22, Image feature extraction: For image data, image processing technology and deep learning algorithms are used to extract features; S23, attribute feature extraction: For attribute information in a structured form, it is directly used in the classification and labeling process; by parsing the attribute data table in the BIM model or the attribute record in the database, the key attribute features are extracted and integrated with the features of other dimensions; S24, feature fusion: feature fusion is performed in series, parallel or mixed mode; S3, intelligent classification and labeling, including: S31, ensemble learning model; S32, adaptive rule engine; S33, Result optimization and feedback: For obviously wrong classifications or labels, mark them and prompt users to make manual corrections; collect user feedback and correction results, and update and optimize the parameters and rules in the integrated learning model and rule base based on user feedback and correction results to improve the performance of subsequent classification and labeling; S4. Result verification and optimization, including: S41, Cross-validation: The classification results are verified by cross-validation. Cross-validation divides the data set into a training set and a test set, trains the model on the training set, and verifies the model performance on the test set. S42. Performance evaluation: Use performance indicators such as accuracy, precision or recall to evaluate classification and labeling results; S43, continuous optimization: Optimization directions include adjusting feature extraction algorithms, replacing base classifiers, and optimizing rules in the rule base to continuously improve the accuracy and robustness of classification and labeling; S5. User interaction and report generation.

[0007] Further, step S12 is specifically as follows: The collected data are cleaned by removing duplicate data, processing missing values, and correcting erroneous data; missing values ​​are processed by mean filling, median filling, mode filling, or model-based prediction filling methods; outliers are identified and processed by statistical or machine learning-based methods; data format conversion and standardization are performed to ensure data consistency and comparability.

[0008] Further, step S21 is specifically as follows: First, the text is segmented to remove stop words and punctuation marks. Then, the algorithm is used to convert the text into a vector representation, and finally, the key features are extracted. These features will be used in the subsequent classification and labeling process.

[0009] Further, step S22 is specifically as follows: First, the image is preprocessed, including denoising and contrast enhancement operations; then the edge, texture, and shape features of the image are extracted using a deep learning model, the convolutional neural network; finally, the features are further processed and reduced in dimension through a pooling layer and a fully connected layer structure; these image features will be fused with features from other dimensions to improve classification accuracy.

[0010] Further, step S31 is specifically as follows: Step 1: After feature extraction and fusion, an ensemble learning model is used to classify and label BIM data. Ensemble learning improves the overall classification performance by combining the prediction results of multiple base classifiers. Step 2: Select a decision tree or random forest machine learning algorithm as the base classifier; these base classifiers will independently learn the feature representation of BIM data during training and output their own prediction results; Step 3: The prediction results of multiple base classifiers are integrated by voting and weighted averaging to obtain the final classification and labeling results.

[0011] Further, step S32 is specifically as follows: Step 1: Use the integrated learning model to preliminarily classify and label BIM data; the adaptive rule engine automatically adjusts and optimizes the rules in the rule base according to the dynamic changes of data and user feedback; the rules in the rule base are static rules based on expert experience, or dynamic rules automatically generated by the system through machine learning algorithms; Step 2: Compare and verify the preliminary prediction results with the rules in the rule base. In the classification and labeling process, the ensemble learning model is first used to make preliminary predictions, and then further verified and adjusted in combination with the rules in the rule base. Step 3: If the prediction result conflicts or is inconsistent with the rules in the rule base, the system will make a decision based on the priority and confidence of the rules and select the most reasonable classification and labeling result.

[0012] Further, step S5 is specifically as follows: Step 1: Intelligent interactive interface settings: The interface design follows the user-centered design principle, and designs an interactive interface that meets user expectations by investigating user needs and habits; Step 2: Real-time feedback and progress monitoring: During the classification and labeling process, real-time feedback and progress monitoring functions are provided; users can check the progress of the current task, the amount of data processed, and the remaining time information at any time; at the same time, possible problems or abnormal situations will be promptly fed back during the classification and labeling process, and corresponding prompts and suggestions will be given; Step 3: Report generation and export: Data aggregation: First, collect the result data of classification and labeling tasks, including classification results, label information, and performance indicators; Template selection: Users can select a suitable report template from the preset template library or use a custom template; Content filling: fill the collected data into the selected template to generate a complete report; Preview and edit: Users can preview the report content in the system and make necessary edits and modifications; Export and share: After editing, users can choose to export the report as PDF, Excel or Word format files and share or archive them via email or cloud storage; Step 4: Customize the report template: Provides a report template editor that allows users to customize report templates through a graphical interface; users can select different layout styles, color schemes, font styles, and add charts, pictures and multimedia elements to enrich the report content; after editing, users can save customized templates to the template library for subsequent reuse.

[0013] By adopting the above technical solution, the present invention has the following beneficial effects: The present invention realizes efficient processing of BIM data by integrating multi-dimensional feature fusion technology and advanced deep learning algorithms. BIM data usually contains a variety of complex information such as large three-dimensional models, design drawings, and material lists. Traditional methods are often inefficient and error-prone when processing these data. The feature fusion technology used in the present invention can effectively integrate information from different dimensions, automatically learn and extract key features through deep learning models, and greatly improve the accuracy and efficiency of data processing. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0015] Figure 1 A flow chart of the method for automatic classification and labeling of BIM data based on intelligent rules provided by the present invention; Figure 2 A flow chart of the integrated learning model provided by the present invention; Figure 3 A flow chart of the adaptive rule engine provided by the present invention; Figure 4 A flow chart of result optimization and feedback provided by the present invention. DETAILED DESCRIPTION

[0016] The technical solution of the present invention will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] The specific implementation of the present invention is described in detail below in conjunction with the accompanying drawings. It should be understood that the specific implementation described here is only used to illustrate and explain the present invention, and is not used to limit the present invention.

[0018] Combination Figure 1 As shown, this embodiment provides a method for automatic classification and labeling of BIM data based on intelligent rules, which includes the following steps: Step 1: Data preprocessing: 1.1 Data collection and integration: Step 1: BIM data needs to be collected from various data sources (such as design software, construction management system, operation and maintenance platform, etc.). This data may include design drawings, construction drawings, model files, document materials, images and videos in various formats (such as DWG, RVT, PDF, JPG, etc.) and types.

[0019] Step 2: Integrate these data into a unified database through API interface, file import, etc. to provide a basis for subsequent processing.

[0020] 1.2 Data cleaning (Data cleaning improves the quality and availability of data and provides a reliable data foundation for subsequent model construction and application.) Data cleaning is a key step in data preprocessing.

[0021] Step 1: Use a variety of technical means to clean the collected data, including removing duplicate data, processing missing values, correcting erroneous data, etc.

[0022] Step 2: For missing values, the system may use mean filling, median filling, mode filling, or model-based prediction filling methods; Step 3: For outliers, the system can use statistical or machine learning-based methods to identify and process them.

[0023] Step 4: Convert and standardize the data to ensure its consistency and comparability.

[0024] 1.3 Data Augmentation In order to improve the generalization ability of the classification model, the system will also enhance the data.

[0025] Step 1: For image data, the system may use operations such as rotation, scaling, cropping, and flipping to increase the diversity of samples; Step 2: For text data, the system may enrich the text expression through synonym replacement, sentence reorganization, etc. Data enhancement helps improve the accuracy of the classification model.

[0026] Step 2: Feature extraction: 2.1 Text Feature Extraction Step 1: For text data (such as text descriptions in design drawings, descriptions in documents, etc.), natural language processing technology is used to extract features. (Analyze and extract key information from text data.) Step 2: First, the text is segmented to remove irrelevant information such as stop words and punctuation marks, and then the algorithm is used to convert the text into a vector representation, and finally the key features are extracted. These features will be used in the subsequent classification and labeling process.

[0027] 2.2 Image feature extraction Step 1: For image data (such as graphics in design drawings, photos of construction sites, etc.), image processing technology and deep learning algorithms are used for feature extraction.

[0028] Step 2: First, preprocess the image, including denoising, contrast enhancement, etc. Step 3: Then use deep learning models such as convolutional neural networks (CNN) to extract image features such as edges, textures, and shapes; (Convolutional neural networks (CNNs) can automatically learn complex features in images, reduce manual intervention, and improve the accuracy and efficiency of feature extraction.) Step 4: Finally, the features are further processed and dimensionally reduced through structures such as pooling layers and fully connected layers. These image features will be fused with features from other dimensions to improve classification accuracy.

[0029] 2.3 Attribute feature extraction Step 1: BIM data contains a large amount of attribute information (such as the material, size, location, etc. of the components). This attribute information usually exists in a structured form and is directly used in the classification and labeling process.

[0030] Step 2: Extract key attribute features by parsing the attribute data table in the BIM model or the attribute records in the database, and integrate them with the features of other dimensions.

[0031] 2.4 Feature Fusion Step 1: After extracting features of dimensions such as text, image, and attributes, these features are integrated using feature fusion technology. Feature fusion can be performed in series, parallel, or hybrid. The series method directly concatenates feature vectors of different dimensions into a longer vector; the parallel method maintains the independence of features of different dimensions and fuses them through weighted summation and other methods; the hybrid method combines the advantages of series and parallel, taking into account the correlation between features while retaining some independence.

[0032] Step 2: Through feature fusion, the system can capture the essential information of BIM data more comprehensively and provide richer and more accurate feature representation for subsequent classification and labeling processes.

[0033] Combination Figure 2-4 As shown, step 3: intelligent classification and labeling 3.1 Ensemble learning model (improved accuracy, strong robustness, strong adaptability) Step 1: After feature extraction and fusion, an ensemble learning model is used to classify and label BIM data. Ensemble learning improves the overall classification performance by combining the prediction results of multiple base classifiers.

[0034] Step 2: Select multiple machine learning algorithms such as decision trees and random forests as base classifiers. These base classifiers will independently learn the feature representation of BIM data during the training process and output their own prediction results.

[0035] Step 3: Fusion the prediction results of multiple base classifiers through voting, weighted averaging, etc. to obtain the final classification and labeling results.

[0036] 3.2 Adaptive Rules Engine Step 1: Use the integrated learning model to preliminarily classify and label BIM data. The adaptive rule engine automatically adjusts and optimizes the rules in the rule base according to the dynamic changes of data and user feedback. The rules in the rule base can be static rules based on expert experience or dynamic rules automatically generated by the system through machine learning algorithms.

[0037] Step 2: Compare and verify the preliminary prediction results with the rules in the rule base. During the classification and labeling process, the system will first use the ensemble learning model to make preliminary predictions, and then further verify and adjust them in combination with the rules in the rule base.

[0038] Step 3: If the prediction result conflicts or is inconsistent with the rules in the rule base, the system will make a decision based on the priority and confidence of the rules and select the most reasonable classification and labeling result.

[0039] 3.3 Result optimization and feedback Step 1: In order to improve the accuracy of classification and labeling, the system also designs a result optimization and feedback mechanism. After obtaining the preliminary classification and labeling results, the system will further optimize these results. For example, in the case of fuzzy classification boundaries or label conflicts, the system will use cluster analysis, graph network and other methods for further analysis and judgment; Step 2: For obviously incorrect classifications or labels, the system will mark them and prompt the user to make manual corrections. The system will also collect user feedback and correction results.

[0040] Step 3: Based on user feedback and correction results, update and optimize the parameters and rules in the integrated learning model and rule base to improve the performance of subsequent classification and labeling.

[0041] Step 4: Result verification and optimization: 4.1 Cross-Validation Step 1: After classification and labeling are completed, the system needs to verify the results to ensure the accuracy and reliability of the classification. The system uses the cross-validation method to verify the classification results. Cross-validation divides the data set into a training set and a test set, trains the model on the training set, and verifies the model performance on the test set.

[0042] Step 2: Evaluate the classification performance of the model by comparing the prediction results on the test set with the actual labels. If the classification performance does not meet the requirements, the system needs to go back and adjust the parameters and settings in the feature extraction, model training or rule optimization steps.

[0043] 4.2 Performance Evaluation Step 1: The system uses a variety of performance indicators to evaluate the classification and labeling results, including accuracy, precision, recall, etc. These indicators can fully reflect the performance of the classification model and help users understand the advantages and disadvantages of the model.

[0044] Step 2: The system also provides visualization tools to display classification results and performance indicators, helping users to more intuitively understand the classification and labeling process and results.

[0045] 4.3 Continuous Optimization Step 1: Based on the result verification and performance evaluation, the system will continue to optimize according to the evaluation results and user feedback. The optimization direction may include adjusting the feature extraction algorithm, replacing the base classifier, optimizing the rules in the rule base, etc.

[0046] Step 2: Through continuous optimization, the system can continuously improve the accuracy and robustness of classification and labeling to better meet user needs.

[0047] Step 5: User interaction and report generation: Step 1: Intelligent interactive interface settings The interface design follows the user-centered design principle. By investigating user needs and habits, an interactive interface that meets user expectations is designed. The interface layout is reasonable and the functional areas are clearly divided. Users can upload data by dragging files to the designated area and start classification and labeling tasks by clicking buttons. The system also has an adaptive function that can automatically adjust the interface layout according to the screen size and resolution of the user's device to ensure a good operating experience on various devices.

[0048] Step 2: Real-time feedback and progress monitoring During the classification and labeling process, the system provides real-time feedback and progress monitoring functions. Users can view the progress of the current task, the amount of data processed, the remaining time and other information at any time. At the same time, the system will also promptly feedback possible problems or abnormal situations during the classification and labeling process, and give corresponding prompts and suggestions. These real-time feedback and progress monitoring functions help users better understand the progress of tasks and the quality of results.

[0049] Step 3: Report generation and export Data aggregation: The system first collects the result data of classification and labeling tasks, including classification results, label information, performance indicators, etc.

[0050] Template selection: Users can select a suitable report template from the preset template library or use a custom template.

[0051] Content filling: The system fills the collected data into the selected template to generate a complete report.

[0052] Preview and Edit: Users can preview the report content in the system and make necessary edits and modifications.

[0053] Export and share: After editing, users can choose to export the report into PDF, Excel, Word or other formats, and share or archive it via email, cloud storage, etc.

[0054] Step 4: Customize the report template The system provides a report template editor that allows users to customize report templates through a graphical interface. Users can choose different layout styles, color schemes, font styles, etc., and add multimedia elements such as charts and pictures to enrich the report content. After editing, users can save the customized template to the template library for subsequent reuse.

[0055] The advantages or beneficial effects of the present invention are as follows: Technical advantages: Efficient data processing capabilities: The present invention realizes efficient processing of BIM data by integrating multi-dimensional feature fusion technology and advanced deep learning algorithms. BIM data usually contains a variety of complex information such as large three-dimensional models, design drawings, and material lists. Traditional methods are often inefficient and error-prone when processing these data. The feature fusion technology used in the present invention can effectively integrate information from different dimensions, automatically learn and extract key features through deep learning models, and greatly improve the accuracy and efficiency of data processing. In addition, the system also adopts a distributed processing architecture to distribute data processing tasks to multiple computing nodes for parallel processing, further improving the processing speed and meeting the real-time requirements of construction projects.

[0056] Accurate automatic classification and labeling: Traditional BIM data classification and labeling work mainly relies on manual operations, which is not only time-consuming and labor-intensive, but also prone to errors. The present invention realizes accurate and automated classification and labeling of BIM data through the combination of ensemble learning and adaptive rule engines. The ensemble learning model improves the accuracy and robustness of classification by combining the prediction results of multiple base classifiers; while the adaptive rule engine automatically adjusts and optimizes the rules in the rule base according to the dynamic changes of data and user feedback, ensuring the timeliness and accuracy of classification and labeling. This automated processing method not only reduces manual intervention and reduces the error rate, but also greatly improves work efficiency, providing strong support for the full life cycle management of construction projects.

[0057] Strong scalability and compatibility: With the continuous development of BIM technology and the introduction of new standards, the system needs to have good scalability and compatibility to cope with new challenges that may arise in the future. The present invention adopts a modular design concept to independently encapsulate different functional modules for easy subsequent expansion and upgrading. At the same time, by defining a unified data interface and protocol, the system can easily be compatible with BIM data of different versions and formats, ensuring seamless docking and sharing of data. This powerful scalability and compatibility enables the present invention to keep up with the development of BIM technology and constantly adapt to new application scenarios and needs.

[0058] Intelligent interaction and real-time feedback: The present invention designs an intelligent interactive interface and real-time feedback mechanism, providing users with a convenient and intuitive operating experience and timely feedback on task progress and results. Users can complete operations such as uploading, processing, classifying and labeling BIM data through simple operations, and timely understand task progress and results through the real-time feedback mechanism provided by the system. This intelligent interactive method not only improves the user's work efficiency, but also enhances the interactivity between the user and the system, helping users to better understand and use BIM data.

[0059] Economic benefits: Reduce labor costs: Traditional BIM data classification and labeling work requires a lot of human resources, which not only increases the operating costs of enterprises, but also limits the improvement of work efficiency. However, the present invention greatly reduces the dependence on manual labor and reduces the expenditure of labor costs by realizing the automatic classification and labeling of BIM data. Enterprises can invest more resources in other more valuable business areas, thereby improving the overall economic benefits.

[0060] Improve project efficiency: In the entire life cycle of a construction project, the processing efficiency of BIM data directly affects the overall progress of the project. The present invention speeds up the progress of various links such as design, construction, operation and maintenance of the project by improving the processing speed and accuracy of BIM data. This not only helps to shorten the construction cycle of the project and reduce time costs, but also helps to improve the overall quality and benefits of the project. At the same time, through the real-time feedback mechanism, enterprises can promptly discover and solve problems in the project, avoiding project risks caused by data processing delays.

[0061] Promote optimal use of resources: BIM data contains various information of construction projects, such as material lists, equipment configuration, construction progress, etc. Through the automated classification and labeling technology of the present invention, enterprises can obtain this information more conveniently, and conduct in-depth analysis and mining. This helps enterprises better understand the resource requirements and consumption of the project, so as to formulate a reasonable resource allocation plan and achieve optimal utilization of resources. For example, enterprises can predict material requirements based on the material list, purchase and reserve in advance; optimize construction plans based on equipment configuration, improve equipment utilization efficiency, etc. These measures can help reduce the operating costs of the project and improve economic benefits.

[0062] Social benefits: Improving the level of intelligence in the construction industry: The implementation of this invention will promote the widespread application and in-depth development of BIM technology in the construction industry and improve the level of intelligence in the construction industry. By realizing the automatic classification and labeling of BIM data, enterprises can manage and use BIM data more efficiently and improve the efficiency of project design, construction and operation and maintenance. This will help promote the digital transformation and intelligent upgrading of the construction industry and promote the sustainable development of the construction industry.

[0063] Promote resource conservation and environmental protection: The construction industry is an industry with large resource consumption and environmental impact. Through the implementation of the present invention, enterprises can more accurately grasp the resource demand and consumption of the project, so as to formulate a reasonable resource allocation plan and reduce the waste and loss of resources. At the same time, by optimizing construction plans and equipment configuration and other measures, enterprises can reduce energy consumption and emissions and reduce the impact on the environment. This helps to promote the green development and sustainable development of the construction industry, which is in line with the strategic requirements of the country's ecological civilization construction.

[0064] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for automatic classification and labeling of BIM data based on intelligent rules, characterized in that: The steps include: S1. Data preprocessing, including: S11. Data collection and integration: Collect BIM data from various data sources and integrate them into a unified database through API interface or file import to provide a basis for subsequent processing; S12. Data cleaning: Improve the quality and availability of data and provide a reliable data foundation for subsequent model building and application; S13, Data enhancement: For image data, use rotation, scaling, cropping or flipping operations to increase sample diversity; for text data, enrich text expression through synonym replacement and sentence reorganization; S2. Feature extraction, including: S21. Text feature extraction: For text data, natural language processing technology is used to extract features; S22, Image feature extraction: For image data, image processing technology and deep learning algorithms are used to extract features; S23, attribute feature extraction: For attribute information in a structured form, it is directly used in the classification and labeling process; by parsing the attribute data table in the BIM model or the attribute record in the database, the key attribute features are extracted and integrated with the features of other dimensions; S24, feature fusion: feature fusion is performed in series, parallel or mixed mode; S3, intelligent classification and labeling, including: S31, ensemble learning model; S32, adaptive rule engine; S33, Result optimization and feedback: For obviously wrong classifications or labels, mark them and prompt users to make manual corrections; collect user feedback and correction results, and update and optimize the parameters and rules in the integrated learning model and rule base based on user feedback and correction results to improve the performance of subsequent classification and labeling; S4. Result verification and optimization, including: S41, Cross-validation: The classification results are verified by cross-validation. Cross-validation divides the data set into a training set and a test set, trains the model on the training set, and verifies the model performance on the test set. S42. Performance evaluation: Use performance indicators such as accuracy, precision or recall to evaluate classification and labeling results; S43, continuous optimization: Optimization directions include adjusting feature extraction algorithms, replacing base classifiers, and optimizing rules in the rule base to continuously improve the accuracy and robustness of classification and labeling; S5. User interaction and report generation.

2. The method for automatic classification and labeling of BIM data based on intelligent rules according to claim 1 is characterized in that: Step S12 is specifically as follows: Clean the collected data by removing duplicate data, processing missing values, and correcting erroneous data; for missing values, use mean filling, median filling, mode filling, or model-based prediction filling methods to process; for outliers, use statistical methods or machine learning-based methods to identify and process them; The data were formatted and standardized to ensure consistency and comparability.

3. The method for automatic classification and labeling of BIM data based on intelligent rules according to claim 1 is characterized in that: Step S21 is specifically as follows: First, the text is segmented to remove stop words and punctuation marks. Then, the algorithm is used to convert the text into a vector representation, and finally, the key features are extracted. These features will be used in the subsequent classification and labeling process.

4. The method for automatic classification and labeling of BIM data based on intelligent rules according to claim 1 is characterized in that: Step S22 is specifically as follows: First, the image is preprocessed, including denoising and contrast enhancement operations; then the edge, texture, and shape features of the image are extracted using a deep learning model, the convolutional neural network; finally, the features are further processed and reduced in dimension through a pooling layer and a fully connected layer structure; these image features will be fused with features from other dimensions to improve classification accuracy.

5. The method for automatic classification and labeling of BIM data based on intelligent rules according to claim 1, characterized in that: Step S31 is specifically as follows: Step 1: After feature extraction and fusion, an ensemble learning model is used to classify and label BIM data. Ensemble learning improves the overall classification performance by combining the prediction results of multiple base classifiers. Step 2: Select a decision tree or random forest machine learning algorithm as the base classifier; these base classifiers will independently learn the feature representation of BIM data during training and output their own prediction results; Step 3: The prediction results of multiple base classifiers are integrated by voting and weighted averaging to obtain the final classification and labeling results.

6. The method for automatic classification and labeling of BIM data based on intelligent rules according to claim 1, characterized in that: Step S32 is specifically as follows: Step 1: Use the integrated learning model to preliminarily classify and label BIM data; the adaptive rule engine automatically adjusts and optimizes the rules in the rule base according to the dynamic changes of data and user feedback; the rules in the rule base are static rules based on expert experience, or dynamic rules automatically generated by the system through machine learning algorithms; Step 2: Compare and verify the preliminary prediction results with the rules in the rule base. In the classification and labeling process, the ensemble learning model is first used to make preliminary predictions, and then further verified and adjusted in combination with the rules in the rule base. Step 3: If the prediction result conflicts or is inconsistent with the rules in the rule base, the system will make a decision based on the priority and confidence of the rules and select the most reasonable classification and labeling result.

7. The method for automatic classification and labeling of BIM data based on intelligent rules according to claim 1, characterized in that: Step S5 is specifically as follows: Step 1: Intelligent interactive interface settings: The interface design follows the user-centered design principle, and designs an interactive interface that meets user expectations by investigating user needs and habits; Step 2: Real-time feedback and progress monitoring: During the classification and labeling process, real-time feedback and progress monitoring functions are provided; users can check the progress of the current task, the amount of data processed, and the remaining time information at any time; at the same time, possible problems or abnormal situations will be promptly fed back during the classification and labeling process, and corresponding prompts and suggestions will be given; Step 3: Report generation and export: Data aggregation: First, collect the result data of classification and labeling tasks, including classification results, label information, and performance indicators; Template selection: Users can select a suitable report template from the preset template library or use a custom template; Content filling: fill the collected data into the selected template to generate a complete report; Preview and edit: Users can preview the report content in the system and make necessary edits and modifications; Export and share: After editing, users can choose to export the report as PDF, Excel or Word format files and share or archive them via email or cloud storage; Step 4: Customize the report template: Provides a report template editor that allows users to customize report templates through a graphical interface; users can select different layout styles, color schemes, font styles, and add charts, pictures and multimedia elements to enrich the report content; After editing is completed, the user saves the customized template to the template library for subsequent reuse.

Images