BIM-based steel bar automated processing system and method

Through the BIM-based reinforcement automation processing system, real-time data analysis and model updates are used to use clustering algorithms and multi-layer convolutional neural network models, the problem of insufficient real-time data processing and model update capabilities of traditional systems is solved, and efficient and accurate steel bar processing and intelligent management.

CN118628069BActive Publication Date: 2024-11-12CCCC THIRD HIGHWAY ENG CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411102725.1
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-13
Publication Date
2024-11-12
Estimated Expiration
2044-08-13

AI Technical Summary

Technical Problem

Traditional steel bar processing systems lack real-time data processing and model update capabilities, and cannot accurately reflect the actual processing conditions, resulting in low processing accuracy and efficiency, and limited capabilities in abnormal detection and data analysis, making it difficult to meet the modern construction industry's efficient, accurate and intelligent Reinforcement processing requirements.

Method used

BIM-based reinforcement automation processing system, through data collection module and BIM model building module, data analysis and model update are used to use clustering algorithms and multi-layer convolutional neural network models to build parameterized reinforcement model and relationship model between features to realize real-time Performance metric evaluation and abnormal detection and correction.

Benefits of technology

It improves the real-time adaptability and accuracy of steel bar processing, reduces manual intervention, can quickly evaluate the performance of steel bars, identify potential problems in advance, and optimizes the processing process, improving the intelligence level and production efficiency of the system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118628069B_ABST
    Figure CN118628069B_ABST
Patent Text Reader

Abstract

The present invention relates to the technical field of steel bar automatic processing and analysis, and specifically to a steel bar automatic processing system and method based on BIM. In the present invention, historical and real-time steel bar processing data are collected through a data collection module, intelligent processing is performed using a BIM model construction module, a clustering algorithm and a multi-layer convolutional neural network model are applied to extract high-level features, and a parameterized steel bar model and a feature relationship model are constructed to form a complete BIM model; the BIM model application module extracts high-level features and updates the parameterized steel bar model according to the real-time steel bar processing data, and analyzes performance indicators accordingly. If an abnormality is found, the feature relationship model is used to verify and correct the data; if it is normal, a processing instruction is directly generated; through intelligent data processing, dynamic model updating and automatic exception processing, the accuracy, efficiency and reliability of the system are significantly improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of automated processing and analysis of steel bars, and in particular to a BIM-based automated processing system and method for steel bars. Background Art

[0002] Traditional rebar processing systems face many technical challenges in meeting the complex and ever-changing demands of construction projects.

[0003] First, the system lacks real-time data processing and model updating capabilities, which makes it difficult to adapt to project changes in a timely manner, affecting processing accuracy and efficiency.

[0004] Secondly, performance index evaluation often relies on static models, which cannot accurately reflect the actual processing situation and increase the difficulty of quality control.

[0005] Furthermore, traditional systems are inefficient in anomaly detection and problem diagnosis, often requiring a lot of manual intervention, which prolongs the problem-solving time.

[0006] In addition, traditional methods have limited capabilities in feature extraction and data analysis, making it difficult to mine valuable information from massive historical data, which restricts the system's optimization and learning capabilities. At the same time, due to the lack of intelligent prediction and analysis functions, the system is difficult to identify potential problems in advance and make corresponding adjustments, increasing production risks.

[0007] Finally, the traditional system is not adaptable enough when dealing with new steel bar designs or processing methods, and it is difficult to quickly respond to the development and changes of industry technology. These technical problems have led to obvious deficiencies in the efficiency, accuracy, reliability and intelligence level of the traditional steel bar processing system, making it difficult to meet the needs of the modern construction industry for efficient, accurate and intelligent steel bar processing. Summary of the invention

[0008] The purpose of the present invention is to provide a BIM-based steel bar automated processing system and method to solve the problems raised in the above background technology.

[0009] To achieve the above object, one of the objects of the present invention is to provide a BIM-based steel bar automated processing system, comprising a data collection module, a BIM model construction module and a BIM model application module, wherein:

[0010] The data collection module collects historical steel bar processing data and real-time steel bar processing data;

[0011] The BIM model construction module uses a clustering algorithm to identify clustering results based on historical steel bar processing data, and the clustering results include steel bar categories and corresponding features of the categories; based on the clustering results, a multi-layer convolutional neural network model is used to extract historical high-level features; based on the historical high-level features, a parameterized steel bar model and a feature relationship model are constructed, and the parameterized steel bar model and the feature relationship model are used together as a BIM model, wherein the parameterized steel bar model includes a parameter value range, a constraint relationship between parameters, and a performance analysis function defined based on an architectural rule library;

[0012] The BIM model application module uses a multi-layer convolutional neural network model to extract real-time advanced features based on real-time steel bar processing data; and uses the real-time advanced features to update the parameterized steel bar model in the BIM model, extracts the corresponding performance indicators based on the performance analysis function in the updated parameterized steel bar model, and performs performance indicator abnormality analysis, wherein:

[0013] If the performance index analysis is abnormal, the feature relationship model in the BIM model is used to verify the real-time steel bar processing data, obtain the abnormal key features, and intelligently correct them according to the predefined constraint relationships and value ranges in the updated parametric steel bar model; the parametric steel bar model is updated with the corrected real-time steel bar processing data; and this process is repeated until the performance index analysis is normal;

[0014] If the performance index analysis is normal, the updated parameterized steel bar model and the feature relationship model are used to generate steel bar processing instructions.

[0015] As a further improvement of the present technical solution, the historical steel bar processing data collected by the data collection module includes but is not limited to a steel bar basic parameter set, a material property set, a processing parameter set and a construction project.

[0016] As a further improvement of the technical solution, the BIM model construction module includes a steel bar feature and category identification unit, which identifies clustering results using a clustering algorithm based on historical steel bar processing data. The clustering results include steel bar categories and corresponding features of the categories, specifically including:

[0017] Preprocess the historical rebar processing data, including cleaning and standardization, to convert various types of data in the historical rebar processing data into a format that can be processed by the clustering algorithm;

[0018] K cluster centers are preset, where K represents different steel bar feature categories; the preprocessed historical steel bar processing data is feature vectorized, and each construction project is represented as a point in a multidimensional space;

[0019] Calculate the Euclidean distance from each data point in the historical steel bar processing data to each cluster center, and assign the data point to the nearest cluster center; recalculate the center point of each cluster until the cluster center reaches the preset number of iterations; output the clustering results after iteration, where the clustering results are steel bar categories and corresponding features of the categories, specifically including the steel bar categories to which each construction project belongs and the features of each category.

[0020] As a further improvement of the technical solution, the BIM model construction module includes an advanced feature extraction unit. The advanced feature extraction unit designs a multi-layer convolutional neural network structure based on the clustering results. The network includes multiple convolutional layers, pooling layers and fully connected layers. The convolutional layer uses filters of different sizes to capture local features in the steel bar drawings. The pooling layer reduces the data dimension by downsampling and extracts the most significant features. The steel bar category in the clustering results is used as the initial label to train the multi-layer convolutional neural network model.

[0021] During the training process, the network parameters are continuously adjusted through the back-propagation algorithm to minimize the difference between the predicted results and the initial labels; after the training is completed, the multi-layer convolutional neural network model extracts the historical rebar processing data and the real-time rebar processing data into historical high-level features and real-time high-level features respectively.

[0022] As a further improvement of the present technical solution, the BIM model construction module includes a parameterized steel bar model construction unit, which uses the historical advanced features extracted in the advanced feature extraction unit, and the basic parameter set of steel bars, material property set and processing parameter set in the historical steel bar processing data as inputs for constructing a parameterized steel bar model, thereby constructing a parameterized steel bar model, specifically including:

[0023] Mapping historical advanced features to specific basic parameters of steel bars, material property sets and processing parameters, and establishing the connection between historical advanced features and actual steel bar properties;

[0024] For each mapped parameter, define its value range and default value;

[0025] Creating a geometric model generation function, wherein the geometric model generation function generates a three-dimensional geometric model of the steel bar according to given parameters;

[0026] According to the architectural rule base, establish the parameter value range and the constraint relationship between parameters;

[0027] Create an update function that updates the geometry and ensures that all constraints are satisfied when parameters change.

[0028] Add a performance analysis function that can calculate various performance indicators of steel bars based on current parameters.

[0029] As a further improvement of the present technical solution, the BIM model construction module includes a feature relationship model construction unit, which uses the historical high-level features extracted in the high-level feature extraction unit and the data features and feature values ​​in the historical steel bar processing data as inputs for constructing the feature relationship model, thereby constructing the feature relationship model, specifically including:

[0030] Calculate the correlation matrix between features to reflect the strength of the linear relationship between each feature;

[0031] Based on the correlation analysis results, a feature graph is created, where nodes represent features and edges represent the strength of the relationship between features. Connections are established between features only when the correlation exceeds a preset correlation threshold.

[0032] The importance score of each feature is calculated using the feature importance of random forest, and each feature is assigned an importance score to reflect the degree of influence of the feature on the prediction result;

[0033] Build a prediction model for each feature. Each model uses all other features as input to predict the value of the feature. The prediction model is used to capture the nonlinear relationship between features.

[0034] Define a causal inference function to infer the causal structure between features;

[0035] Create a feature relationship query function that receives two features as input and returns the direct correlation between them, the shortest path in the feature graph, the prediction accuracy, and the causal relationship; where prediction accuracy = 1-(root mean square error / average value of the feature), where the root mean square error represents the average deviation between the predicted value and the actual observed value in the process of building the prediction model. The root mean square error is obtained by calculating the error between each predicted value and the actual value, averaging all the errors, and taking the square root; the average value of the feature represents the arithmetic mean of all predicted values ​​of the feature in the data set;

[0036] Added model update function to update the correlation matrix, feature map, feature importance score, and prediction model when there are real-time advanced feature inputs.

[0037] As a further improvement of the technical solution, the BIM model application module includes a performance index extraction unit, which updates the parameterized steel bar model in the BIM model according to the real-time steel bar processing data, and extracts the corresponding performance index according to the updated parameterized steel bar model and the feature relationship model, specifically including:

[0038] Use the advanced feature extraction unit to process real-time steel bar processing data and extract real-time advanced features;

[0039] Using the update function of the parametric steel bar model, the parametric steel bar model is updated according to the mapped new parameters. During the update process, all parameters are ensured to be within the value range defined during construction; the constraint relationship in the architectural rule library is applied to ensure that the updated model still meets all constraint conditions; using the geometric model generation function, the three-dimensional geometric model of the steel bar is regenerated according to the updated parameters;

[0040] Based on the updated parameterized steel bar model, using the performance analysis function in the model, calculating various performance indicators of the updated model as actual performance indicators, the performance indicators including but not limited to tensile strength, fatigue life and weight;

[0041] The prediction model in the inter-feature relationship model is used to predict the expected performance indicators corresponding to the real-time advanced features. Since a prediction model is built for each feature in the inter-feature relationship model, the model uses all other features as input to predict the value of the feature as the prediction performance indicator.

[0042] As a further improvement of the technical solution, the BIM model application module includes a performance indicator abnormality analysis unit, which performs abnormal analysis on the performance indicator, evaluates the degree of difference by comparing the actual performance indicator with the predicted performance indicator, sets a difference threshold, and determines the degree of difference according to the difference threshold, wherein:

[0043] If the percentage difference between the actual performance index and the predicted performance index is within or outside the difference threshold range, the degree of difference is abnormal, indicating that the performance index analysis is abnormal. The feature relationship model in the BIM model is used to verify the real-time steel bar processing data, obtain the abnormal key features, and correct them, including:

[0044] Extract predictive performance indicators of abnormal degree of difference;

[0045] For each abnormal performance indicator, the feature relationship query function in the feature relationship model is used to analyze its relationship with other features; this function receives two features as input, returns the query result, and extracts the prediction accuracy in the query result;

[0046] According to the prediction accuracy in the query results, the abnormal key features of the real-time steel bar processing data are determined, and by analyzing the prediction accuracy of each feature for the abnormal performance index, the key feature with the highest prediction accuracy is selected as the abnormal key feature;

[0047] Correct the abnormal key features. Use the constraint relationship and value range in the parametric steel bar model to check whether the abnormal key features exceed the predefined reasonable range. If they exceed, adjust them to the nearest reasonable value for correction. Update the parametric steel bar model with the corrected real-time key features until the performance index analysis is normal.

[0048] If the percentage difference between the actual performance index and the predicted performance index is within the difference threshold range, the degree of difference is normal, indicating that the performance index analysis is normal, and the updated parameterized steel bar model and the feature relationship model are used to generate steel bar processing instructions.

[0049] The second object of the present invention is to provide a method for an automated steel bar processing system based on BIM, comprising the following method steps:

[0050] S1. Collect historical steel bar processing data and real-time steel bar processing data;

[0051] S2. Based on the historical steel bar processing data, a clustering algorithm is used to identify clustering results, where the clustering results include steel bar categories and corresponding features of the categories; based on the clustering results, a multi-layer convolutional neural network model is used to extract historical high-level features; based on the historical high-level features, a parameterized steel bar model and a feature relationship model are constructed, and the parameterized steel bar model and the feature relationship model are used together as a BIM model;

[0052] S3. Based on the real-time steel bar processing data, a multi-layer convolutional neural network model is used to extract real-time advanced features. The real-time advanced features are used to update the parametric steel bar model in the BIM model. Based on the updated parametric steel bar model, the corresponding performance indicators are extracted and the performance indicator abnormality analysis is performed, where:

[0053] If the performance index analysis is abnormal, the feature relationship model in the BIM model is used to verify the real-time steel bar processing data, obtain the abnormal key features, and correct them; the corrected real-time steel bar processing data is used to update the parametric steel bar model;

[0054] If the performance index analysis is normal, the updated parameterized steel bar model and the feature relationship model are used to generate steel bar processing instructions.

[0055] Compared with the prior art, the present invention has the following beneficial effects:

[0056] 1. The multiple applications of the parametric steel bar model in the BIM-based steel bar automatic processing system and method have produced significant beneficial effects. First, by using real-time advanced features to update the parametric steel bar model in the BIM model, the system can adapt to the specific needs and changes of the current construction project in real time. This dynamic update mechanism ensures that the model always reflects the latest steel bar processing situation, improving the accuracy and practicality of the model. Secondly, the corresponding performance indicators are extracted according to the updated parametric steel bar model, so that the system can quickly evaluate the various performances of the current steel bar design, such as tensile strength, fatigue life and weight. This step provides a reliable data basis for subsequent performance indicator analysis. Thirdly, during the performance indicator analysis process, if an abnormal situation is found, the system will use the constraint relationship and value range of the parametric steel bar model to correct the abnormal key features. This self-correction mechanism not only improves the reliability of the system, but also reduces the need for human intervention. Finally, the corrected real-time steel bar processing data will be used again to update the parametric steel bar model to form a closed-loop optimization process. This multiple application method ensures the continuous optimization of the steel bar processing process, greatly improves the processing accuracy and efficiency, and reduces material waste and production costs.

[0057] 2. The multiple applications of the feature relationship model in the BIM-based steel bar automatic processing system and method play a key role in this system method and bring significant beneficial effects; first, the performance index extraction unit uses the prediction model in the feature relationship model to predict the corresponding expected performance index based on the real-time steel bar processing data. This prediction ability enables the system to have an accurate estimate of the steel bar performance before actual processing, so as to discover potential problems in advance and make adjustments; secondly, in the process of abnormal performance index analysis, the system analyzes the relationship between abnormal performance indicators and other features through the feature relationship query function in the feature relationship model; this relationship analysis helps the system quickly locate the key factors that may cause abnormalities, greatly shortening the problem diagnosis time; thirdly, by analyzing the prediction accuracy of each feature in the feature relationship model for abnormal performance indicators, the system can intelligently select the key features with the highest prediction accuracy as abnormal key features; this method not only improves the accuracy of abnormal detection, but also optimizes the efficiency of the correction process. Finally, the continuous update mechanism of the feature relationship model ensures that the model can adapt to new data and changing production environments, and maintain the long-term effectiveness of the model. This multiple application method significantly improves the intelligence level of the system and makes the steel bar processing process more accurate, efficient and reliable.

[0058] 3. The multiple applications of the multi-layer convolutional neural network model in the BIM-based steel bar automatic processing system and method in this system method have brought significant technical breakthroughs and benefit improvements. First, by using the multi-layer convolutional neural network model to extract advanced features from historical steel bar processing data, the system can automatically identify and extract key features from massive historical data; this deep learning method greatly improves the efficiency and accuracy of feature extraction, and lays a solid data foundation for the construction of subsequent parametric steel bar models and feature relationship models; secondly, the same multi-layer convolutional neural network model is applied to the advanced feature extraction of real-time steel bar processing data; this real-time processing capability enables the system to quickly respond to changes in the current production environment and provide timely decision support; in addition, the clustering algorithm performs preliminary classification and feature extraction on historical steel bar processing data, provides a structured data basis for the multi-layer convolutional neural network model, reduces data complexity, and provides meaningful labels and feature sets for neural network training; this hierarchical learning strategy significantly improves the learning efficiency of the model and the accuracy of feature extraction, improves the adaptability and generalization ability of the system, and can better cope with new steel bar design or processing methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is a schematic diagram of the overall module unit of the present invention;

[0060] Figure 2 It is a schematic diagram of the overall method flow of the present invention.

[0061] In the figure: 100, data collection module; 200, BIM model construction module; 201, steel bar feature and category recognition unit; 202, advanced feature extraction unit; 203, parameterized steel bar model construction unit; 204, feature relationship model construction unit; 300, BIM model application module; 301, performance indicator extraction unit; 302, performance indicator abnormality analysis unit. DETAILED DESCRIPTION

[0062] The following will be combined with the accompanying drawings in the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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.

[0063] Unless otherwise specified, the terms (including technical terms) used herein have the commonly understood meanings to those skilled in the art. In addition, it is understood that the terms defined in commonly used dictionaries should be understood to have the same meanings as those in the context of the relevant fields, and should not be understood as idealized or overly formal meanings.

[0064] According to the document, the following are some definitions of terms:

[0065] Historical rebar processing data: refers to various information collected and accumulated in past construction projects, including but not limited to basic rebar parameter sets, material property sets, processing parameter sets, and construction projects;

[0066] Real-time rebar processing data: refers to various information collected in real time from the current construction project or the rebar being processed, such as real-time basic rebar parameters, real-time material attribute sets, and real-time processing parameters;

[0067] Clustering results: The steel bar categories and corresponding features obtained by the clustering algorithm, including the steel bar categories to which each construction project belongs and the typical features of each category;

[0068] Historical high-level features: complex features such as spatial relationships, structural layout, and material combinations extracted from historical rebar processing data through a multi-layer convolutional neural network model;

[0069] Real-time advanced features: complex features extracted from real-time rebar processing data through a multi-layer convolutional neural network model;

[0070] Parametric reinforcement model: A model built based on historical advanced features and basic reinforcement parameters, including geometric model generation functions, parameter constraint relationships, and performance analysis functions;

[0071] Feature relationship model: a model that reflects the relationship between various features, including correlation matrix, feature graph, feature importance score and prediction model;

[0072] BIM model: BIM model stands for Building Information Model, which is a comprehensive model composed of a parametric reinforcement model and a feature relationship model;

[0073] Performance indicators: including actual performance indicators and predicted performance indicators, such as tensile strength, fatigue life and weight;

[0074] Abnormal key features: When analyzing abnormal performance indicators, the features that have the greatest impact on abnormal performance indicators are identified through the feature relationship model.

[0075] Next, see Figure 1-Figure 2 One of the purposes of this embodiment is to provide a BIM-based steel bar automated processing system, which includes a data collection module 100, a BIM model construction module 200 and a BIM model application module 300.

[0076] The data collection module 100 collects historical steel bar processing data and real-time steel bar processing data from a multi-source data interface, wherein the historical steel bar processing data is collected from a public database, and the historical steel bar processing data refers to various information collected and accumulated in past construction projects, including but not limited to basic steel bar parameter sets (such as length, diameter, bending angle, bending radius, etc.), processing types (straight steel bars, bent steel bars, stirrups, etc.), processing parameters (bending angle, bending radius, etc.), material attribute sets (such as steel bar grade, tensile strength, elastic modulus, etc.) and construction projects, etc.; and is used to provide basic data support for subsequent parametric templates and feature relationship models;

[0077] Real-time rebar processing data is collected using a sensor network or based on relevant data input by the user from the front end. Real-time rebar processing data refers to various information collected in real time from the current construction project or the rebar being processed, such as real-time basic rebar parameters, real-time material property sets, and real-time processing parameters.

[0078] The steel bar feature and category identification unit 201 in the BIM model construction module 200 uses a clustering algorithm to identify clustering results based on historical steel bar processing data. The clustering results include steel bar categories and corresponding features of the categories, specifically including:

[0079] Preprocess the historical rebar processing data, including cleaning and standardization, to convert various types of data in the historical rebar processing data into a format that can be processed by the clustering algorithm;

[0080] K cluster centers are preset, where K represents different steel bar feature categories (such as straight steel bars, simple hooks, complex bends, and stirrups). The pre-processed historical steel bar processing data is feature vectorized, and each construction project is represented as a point in a multidimensional space.

[0081] Calculate the Euclidean distance from each data point in the historical steel bar processing data to each cluster center, and assign the data point to the nearest cluster center; and recalculate the center point of each cluster (that is, the average position of all data points in this class); this assignment-recalculation process continues until the cluster center reaches the preset number of iterations; in each iteration, calculate the variance within the cluster and the distance between clusters to evaluate the clustering quality; finally, output the clustering results after iteration, where the clustering results are steel bar categories and corresponding features of the categories, including the steel bar categories to which each construction project belongs and the typical features of each category (that is, cluster centers); for example, different steel bar categories (such as straight steel bar categories, simple hook categories, complex bend categories, and stirrup categories), each steel bar category has a corresponding feature set (such as the number of bends, total length, and bending angle distribution, etc.).

[0082] Through the above process, the steel bar feature and category recognition unit 201 can automatically extract key steel bar categories and category corresponding features from a large amount of historical steel bar processing data, laying the foundation for subsequent historical high-level feature extraction and parametric template construction.

[0083] The advanced feature extraction unit 202 in the BIM model construction module 200 designs a multi-layer convolutional neural network structure based on the clustering results. The network includes multiple convolutional layers, pooling layers, and fully connected layers. The convolutional layers use filters of different sizes to capture local features in the steel bar drawings, such as edges, textures, and shapes. The pooling layer reduces the data dimension by downsampling and extracts the most significant features. The steel bar categories in the clustering results are used as initial labels to train the multi-layer convolutional neural network model.

[0084] During the training process, the network parameters are continuously adjusted through the back-propagation algorithm to minimize the difference between the predicted results and the initial labels; after the training is completed, the multi-layer convolutional neural network model extracts historical rebar processing data and real-time rebar processing data into historical high-level features and real-time high-level features, such as complex spatial relationships, structural layouts, and material combinations.

[0085] The parameterized steel bar model construction unit 203 in the BIM model construction module 200 uses the historical advanced features extracted by the advanced feature extraction unit 202, and the steel bar basic parameter set, material attribute set and processing parameter set in the historical steel bar processing data as inputs for constructing the parameterized steel bar model, thereby constructing the parameterized steel bar model, specifically including:

[0086] Mapping the historical advanced features to specific reinforcement basic parameters, material property sets and processing parameters. This step establishes the connection between the historical advanced features and the actual reinforcement properties.

[0087] For each mapped parameter, define its value range and default value, which ensures that the model parameters vary within a reasonable range;

[0088] Creating a geometric model generation function, the geometric model generation function generates a three-dimensional geometric model of the steel bar according to given parameters, including but not limited to creating straight segments, curved segments, etc.;

[0089] According to the architectural rule base, establish the parameter value range and the constraint relationship between parameters, such as ensuring that the bending radius is not less than half of the steel bar diameter, or the total length does not exceed a certain maximum value;

[0090] Create an update function that updates the geometry and ensures that all constraints are satisfied when parameters change.

[0091] Added performance analysis function, which can calculate various performance indicators of steel bars, such as tensile strength, fatigue life and weight, based on current parameters;

[0092] Through the above description, a parametric steel bar model is constructed.

[0093] The feature relationship model construction unit 204 in the BIM model construction module 200 uses the historical high-level features extracted by the high-level feature extraction unit 202, as well as the data features and feature values ​​in the historical steel bar processing data, as inputs for constructing the feature relationship model, thereby constructing the feature relationship model, specifically including:

[0094] Calculate the correlation matrix between features to reflect the strength of the linear relationship between each feature;

[0095] Based on the correlation analysis results, a feature graph is created, where nodes represent features and edges represent the strength of the relationship between features. Connections are established between features only when the correlation exceeds a preset correlation threshold.

[0096] Use the feature importance of random forest to calculate the importance score of each feature. This step will assign an importance score to each feature, reflecting the degree of influence of the feature on the prediction result;

[0097] Build a prediction model for each feature. Each model uses all other features as input to predict the value of the feature. This prediction model is used to capture the nonlinear relationship between features and helps to better understand the mutual influence between features.

[0098] Define a causal inference function that can analyze the possible causal relationship between features and infer the causal structure between features;

[0099] Create a feature relationship query function that can receive two features as input and return the direct correlation between them, the shortest path in the feature graph, the prediction accuracy, and possible causal relationships; where prediction accuracy = 1-(root mean square error / average value of the feature), where the root mean square error represents the average deviation between the predicted value and the actual observed value in the process of building the prediction model. The root mean square error is obtained by calculating the error between each predicted value and the actual value, averaging all the errors, and taking the square root; the average value of the feature represents the arithmetic mean of all predicted values ​​of the feature in the data set;

[0100] Add a model update function to update the correlation matrix, feature map, feature importance score, and prediction model when new data arrives.

[0101] The inter-feature relationship model construction unit 204 uses the parameterized steel bar model and the inter-feature relationship model together as a BIM model.

[0102] The performance index extraction unit 301 in the BIM model application module 300 updates the parameterized steel bar model in the BIM model according to the real-time steel bar processing data, and extracts the corresponding performance index according to the updated parameterized steel bar model and the feature relationship model, specifically including:

[0103] Using the advanced feature extraction unit 202 to process the real-time steel bar processing data, extracting real-time advanced features;

[0104] Using the update function of the parametric steel bar model, the parametric steel bar model is updated according to the mapped new parameters. During the update process, all parameters are ensured to be within the value range defined during construction; the constraint relationship in the architectural rule library is applied to ensure that the updated model still meets all constraint conditions; using the geometric model generation function, the three-dimensional geometric model of the steel bar is regenerated according to the updated parameters;

[0105] Based on the updated parameterized steel bar model, using the performance analysis function in the model, various performance indicators of the updated model are calculated as actual performance indicators, and the performance indicators include but are not limited to tensile strength, fatigue life and weight;

[0106] The prediction model in the inter-feature relationship model is used to predict the expected performance indicators corresponding to the real-time advanced features. Since a prediction model is built for each feature in the inter-feature relationship model, which uses all other features as input to predict the value of the feature, the performance indicator is actually also regarded as a feature. Therefore, the prediction model in the inter-feature relationship model is used to predict the performance indicators of real-time steel bar processing data as the predicted performance indicators.

[0107] The performance indicator abnormality analysis unit 302 in the BIM model application module 300 performs abnormality analysis on the performance indicator, evaluates the degree of difference by comparing the actual performance indicator with the predicted performance indicator, sets a difference threshold (e.g., ±5%), and determines the degree of difference according to the difference threshold, wherein:

[0108] If the percentage difference between the actual performance index and the predicted performance index is within or outside the difference threshold range, the degree of difference is abnormal, which means that the performance index analysis is abnormal. The real-time steel bar processing data is verified by using the feature relationship model in the BIM model to obtain the abnormal key features and correct them. The parametric steel bar model is updated with the corrected real-time steel bar processing data, including:

[0109] Extracting predictive performance indicators with abnormal degree of difference, for example, tensile strength and fatigue life are identified as abnormal predictive performance indicators;

[0110] For each abnormal performance indicator, the feature relationship query function in the feature relationship model is used to analyze the relationship between it and other features. For example, for tensile strength, the relationships that may be queried include tensile strength and carbon content, tensile strength and heat treatment temperature, tensile strength and diameter, and tensile strength and cold working degree. This function receives two features as input, returns the query result, and extracts the prediction accuracy in the query result.

[0111] According to the prediction accuracy in the query results, the abnormal key features of the real-time steel bar processing data are determined. By analyzing the prediction accuracy of each feature for the abnormal performance index, the key feature with the highest prediction accuracy is selected as the abnormal key feature, because the higher the prediction accuracy, the greater the probability that the key feature will affect the abnormal performance index;

[0112] Correct the abnormal key features. Use the constraint relationship and value range in the parametric steel bar model to check whether the abnormal key features exceed the predefined reasonable range. If they exceed, adjust them to the nearest reasonable value for correction. Update the parametric steel bar model with the corrected real-time key features until the performance index analysis is normal.

[0113] If the percentage difference between the actual performance index and the predicted performance index is within the difference threshold, the degree of difference is normal, which means that the performance index analysis is normal. The updated parameterized steel bar model and the feature relationship model are used to generate steel bar processing instructions.

[0114] The second purpose of this embodiment is to provide a method for an automated steel bar processing system based on BIM, comprising the following method steps:

[0115] S1. Collect historical steel bar processing data and real-time steel bar processing data;

[0116] S2. Based on the historical steel bar processing data, a clustering algorithm is used to identify clustering results, where the clustering results include steel bar categories and corresponding features of the categories; based on the clustering results, a multi-layer convolutional neural network model is used to extract historical high-level features; based on the historical high-level features, a parameterized steel bar model and a feature relationship model are constructed, and the parameterized steel bar model and the feature relationship model are used together as a BIM model, where the parameterized steel bar model includes a parameter value range, a constraint relationship between parameters, and a performance analysis function defined based on an architectural rule library;

[0117] S3. According to the real-time steel bar processing data, a multi-layer convolutional neural network model is used to extract real-time advanced features; and the parametric steel bar model in the BIM model is updated using the real-time advanced features. According to the performance analysis function in the updated parametric steel bar model, the corresponding performance indicators are extracted, and the performance indicator abnormality analysis is performed, wherein:

[0118] If the performance index analysis is abnormal, the feature relationship model in the BIM model is used to verify the real-time steel bar processing data, obtain the abnormal key features, and intelligently correct them according to the predefined constraint relationships and value ranges in the updated parametric steel bar model; the parametric steel bar model is updated with the corrected real-time steel bar processing data; and this process is repeated until the performance index analysis is normal;

[0119] If the performance index analysis is normal, the updated parameterized steel bar model and the feature relationship model are used to generate steel bar processing instructions.

[0120] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and the above embodiments and descriptions are only preferred examples of the present invention, and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements all fall within the scope of the present invention to be protected.

Claims

1. The BIM-based steel bar automated processing system is characterized by: It comprises a data collection module (100), a BIM model construction module (200) and a BIM model application module (300), wherein: The data collection module (100) collects historical steel bar processing data and real-time steel bar processing data; The BIM model construction module (200) identifies clustering results using a clustering algorithm based on historical steel bar processing data, where the clustering results include steel bar categories and features corresponding to the categories; According to the clustering results, a multi-layer convolutional neural network model is used to extract historical high-level features; According to the historical high-level features, a parametric steel bar model and a feature relationship model are constructed, and the parametric steel bar model and the feature relationship model are used together as a BIM model, wherein the parametric steel bar model includes a parameter value range, a constraint relationship between parameters, and a performance analysis function defined based on an architectural rule library; The BIM model application module (300) extracts real-time advanced features using a multi-layer convolutional neural network model based on real-time steel bar processing data; The real-time advanced features are used to update the parametric reinforcement model in the BIM model. According to the performance analysis function in the updated parametric reinforcement model, the corresponding performance indicators are extracted and the performance indicator abnormality analysis is performed, where: If the performance index analysis is abnormal, the feature relationship model in the BIM model is used to verify the real-time steel bar processing data, obtain the abnormal key features, and intelligently correct them according to the predefined constraint relationships and value ranges in the updated parametric steel bar model; the parametric steel bar model is updated with the corrected real-time steel bar processing data; and this process is repeated until the performance index analysis is normal; If the performance index analysis is normal, the updated parameterized steel bar model and the feature relationship model are used to generate steel bar processing instructions.

2. The BIM-based steel bar automated processing system according to claim 1, characterized in that: The historical steel bar processing data collected by the data collection module (100) includes a steel bar basic parameter set, a material property set, a processing parameter set and a construction project.

3. The BIM-based steel bar automated processing system according to claim 1, characterized in that: The BIM model construction module (200) comprises a steel bar feature and category identification unit (201), wherein the steel bar feature and category identification unit (201) identifies clustering results using a clustering algorithm based on historical steel bar processing data, wherein the clustering results include steel bar categories and features corresponding to the categories, specifically including: Preprocess the historical rebar processing data, including cleaning and standardization, to convert various types of data in the historical rebar processing data into a format that can be processed by the clustering algorithm; K cluster centers are preset, where K represents different steel bar feature categories; the preprocessed historical steel bar processing data is feature vectorized, and each construction project is represented as a point in a multidimensional space; Calculate the Euclidean distance from each data point in the historical steel bar processing data to each cluster center, and assign the data point to the nearest cluster center; recalculate the center point of each cluster until the cluster center reaches the preset number of iterations; output the clustering results after iteration, where the clustering results are steel bar categories and corresponding features of the categories, specifically including the steel bar categories to which each construction project belongs and the features of each category.

4. The BIM-based steel bar automated processing system according to claim 3 is characterized in that: The BIM model construction module (200) comprises an advanced feature extraction unit (202), wherein the advanced feature extraction unit (202) designs a multi-layer convolutional neural network structure based on the clustering result, wherein the network comprises a plurality of convolutional layers, pooling layers and fully connected layers; The convolutional layer uses filters of different sizes to capture local features in the reinforcement drawings; The pooling layer reduces the data dimension by downsampling and extracts the most significant features; Use the steel bar categories in the clustering results as initial labels to train a multi-layer convolutional neural network model; During the training process, the network parameters are continuously adjusted through the back-propagation algorithm to minimize the difference between the predicted results and the initial labels; After training, the multi-layer convolutional neural network model extracts historical rebar processing data and real-time rebar processing data into historical high-level features and real-time high-level features, respectively.

5. The BIM-based steel bar automated processing system according to claim 4, characterized in that: The BIM model construction module (200) comprises a parameterized steel bar model construction unit (203), wherein the parameterized steel bar model construction unit (203) uses the historical high-level features extracted by the high-level feature extraction unit (202), and the steel bar basic parameter set, material property set and processing parameter set in the historical steel bar processing data as inputs for constructing a parameterized steel bar model, thereby constructing the parameterized steel bar model, specifically comprising: Mapping historical advanced features to specific basic parameters of steel bars, material property sets and processing parameters, and establishing the connection between historical advanced features and actual steel bar properties; For each mapped parameter, define its value range and default value; Creating a geometric model generation function, wherein the geometric model generation function generates a three-dimensional geometric model of the steel bar according to given parameters; According to the architectural rule base, establish the parameter value range and the constraint relationship between parameters; Create an update function that updates the geometry and ensures that all constraints are satisfied when parameters change. Add a performance analysis function that can calculate various performance indicators of steel bars based on current parameters.

6. The BIM-based steel bar automated processing system according to claim 5, characterized in that: The BIM model construction module (200) comprises an inter-feature relationship model construction unit (204), wherein the inter-feature relationship model construction unit (204) uses the historical high-level features extracted by the high-level feature extraction unit (202) and the data features and feature values ​​in the historical steel bar processing data as inputs for constructing an inter-feature relationship model, thereby constructing the inter-feature relationship model, specifically comprising: Calculate the correlation matrix between features to reflect the strength of the linear relationship between each feature; Based on the correlation analysis results, a feature graph is created, where nodes represent features and edges represent the strength of the relationship between features. Connections are established between features only when the correlation exceeds a preset correlation threshold. The importance score of each feature is calculated using the feature importance of random forest, and each feature is assigned an importance score to reflect the degree of influence of the feature on the prediction result; Build a prediction model for each feature. Each model uses all other features as input to predict the value of the feature. The prediction model is used to capture the nonlinear relationship between features. Define a causal inference function to infer the causal structure between features; Create a feature relationship query function that receives two features as input and returns the direct correlation between them, the shortest path in the feature graph, the prediction accuracy, and the causal relationship; where prediction accuracy = 1-(root mean square error / average value of the feature), where the root mean square error represents the average deviation between the predicted value and the actual observed value in the process of building the prediction model. The root mean square error is obtained by calculating the error between each predicted value and the actual value, averaging all the errors, and taking the square root; the average value of the feature represents the arithmetic mean of all predicted values ​​of the feature in the data set; Added model update functions to update the correlation matrix, feature graph, feature importance scores, and prediction model when there are real-time advanced feature inputs.

7. The BIM-based steel bar automated processing system according to claim 6, characterized in that: The BIM model application module (300) comprises a performance index extraction unit (301), wherein the performance index extraction unit (301) updates the parameterized steel bar model in the BIM model according to the real-time steel bar processing data, and extracts corresponding performance indexes according to the updated parameterized steel bar model and the feature relationship model, specifically comprising: Using an advanced feature extraction unit (202) to process real-time steel bar processing data and extract real-time advanced features; Using the update function of the parametric steel bar model, the parametric steel bar model is updated according to the mapped new parameters. During the update process, all parameters are ensured to be within the value range defined during construction; the constraint relationship in the architectural rule library is applied to ensure that the updated model still meets all constraint conditions; using the geometric model generation function, the three-dimensional geometric model of the steel bar is regenerated according to the updated parameters; Based on the updated parameterized steel bar model, using the performance analysis function in the model, calculating various performance indicators of the updated model as actual performance indicators, the performance indicators including but not limited to tensile strength, fatigue life and weight; The prediction model in the inter-feature relationship model is used to predict the expected performance indicators corresponding to the real-time advanced features. Since a prediction model is built for each feature in the inter-feature relationship model, the model uses all other features as input to predict the value of the feature as the prediction performance indicator.

8. The BIM-based steel bar automated processing system according to claim 7, characterized in that: The BIM model application module (300) comprises a performance indicator abnormality analysis unit (302), wherein the performance indicator abnormality analysis unit (302) performs abnormality analysis on the performance indicator, evaluates the degree of difference by comparing the actual performance indicator with the predicted performance indicator, sets a difference threshold, and determines the degree of difference according to the difference threshold, wherein: If the percentage difference between the actual performance index and the predicted performance index is within or outside the difference threshold range, the degree of difference is abnormal, indicating that the performance index analysis is abnormal. The feature relationship model in the BIM model is used to verify the real-time steel bar processing data, obtain the abnormal key features, and correct them, including: Extract predictive performance indicators of abnormal degree of difference; For each abnormal performance indicator, the feature relationship query function in the feature relationship model is used to analyze its relationship with other features; this function receives two features as input, returns the query result, and extracts the prediction accuracy in the query result; According to the prediction accuracy in the query results, the abnormal key features of the real-time steel bar processing data are determined, and by analyzing the prediction accuracy of each feature for the abnormal performance index, the key feature with the highest prediction accuracy is selected as the abnormal key feature; Correct the abnormal key features. Use the constraint relationship and value range in the parametric steel bar model to check whether the abnormal key features exceed the predefined reasonable range. If they exceed, adjust them to the nearest reasonable value for correction. Update the parametric steel bar model with the corrected real-time key features until the performance index analysis is normal. If the percentage difference between the actual performance index and the predicted performance index is within the difference threshold range, the degree of difference is normal, indicating that the performance index analysis is normal, and the updated parameterized steel bar model and the feature relationship model are used to generate steel bar processing instructions.

9. A method for using the BIM-based steel bar automated processing system according to any one of claims 1 to 8, characterized in that: The method comprises the following steps: S1. Collect historical steel bar processing data and real-time steel bar processing data; S2. Based on the historical steel bar processing data, a clustering algorithm is used to identify clustering results, where the clustering results include steel bar categories and corresponding features of the categories; According to the clustering results, a multi-layer convolutional neural network model is used to extract historical high-level features; according to the historical high-level features, a parameterized steel bar model and a feature relationship model are constructed, and the parameterized steel bar model and the feature relationship model are used together as a BIM model, wherein the parameterized steel bar model includes a parameter value range, a constraint relationship between parameters, and a performance analysis function defined based on an architectural rule library; S3, based on the real-time steel bar processing data, a multi-layer convolutional neural network model is used to extract real-time advanced features; The real-time advanced features are used to update the parametric reinforcement model in the BIM model. According to the performance analysis function in the updated parametric reinforcement model, the corresponding performance indicators are extracted and the performance indicator analysis is performed, where: If the performance index analysis is abnormal, the feature relationship model in the BIM model is used to verify the real-time steel bar processing data, obtain the abnormal key features, and intelligently correct them according to the predefined constraint relationships and value ranges in the updated parametric steel bar model; the parametric steel bar model is updated with the corrected real-time steel bar processing data; and this process is repeated until the performance index analysis is normal; If the performance index analysis is normal, the updated parameterized steel bar model and the feature relationship model are used to generate steel bar processing instructions.