Road and bridge visual construction operation and maintenance system and method based on BIM (Building Information Modeling) technology

Abstract

The invention provides a road and bridge visual construction operation and maintenance system and method based on a BIM technology, and relates to the field of intelligent analysis. According to the method, design data are extracted from design software, construction data are recorded in real time in the construction process, stress data are obtained by arranging a stress sensor, and nondestructive testing is carried out to obtain a stress distribution image. A three-dimensional model is constructed by using a BIM technology, stress features are extracted from the BIM model mapping stress distribution through a 3D convolutional neural network, and a contrast loss function is constructed to enhance the distinguishing between a normal area and an abnormal area. A potential abnormal region is identified by applying an energy-based graph optimization method, and a three-dimensional region of stress abnormal distribution is determined through a clustering algorithm. Real-time updating of data is ensured based on the cloud computing technology, the precision of decision support is improved by adopting advanced data processing and machine learning technologies, and the efficiency and safety of road and bridge construction and operation and maintenance are improved.

Description

Technical Field

[0001] The present application relates to the field of intelligent analysis, and more specifically, to a road and bridge visual construction and operation and maintenance system and method based on BIM technology. Background Art

[0002] Road and bridge engineering is an important part of the infrastructure construction field and plays a key role in urban transportation. With the rapid development of the economy and the acceleration of urbanization, the demand for road and bridge engineering and transportation is also growing, which puts higher requirements on the construction, construction and operation and maintenance of roads and bridges. The quality and efficiency of road and bridge construction and operation and maintenance management are directly related to the service life, maintenance cost and environmental impact of roads and bridges. In the traditional road and bridge construction and operation and maintenance system, road and bridge construction information is often scattered in different departments and systems, lacking an effective information integration and sharing mechanism, resulting in low decision-making efficiency and accuracy. It mainly relies on manual inspections and regular maintenance, which are subjective and limited. It is difficult to detect hidden parts or minor road and bridge anomalies in time, resulting in safety hazards in road and bridge engineering.

[0003] The document with the prior art publication number CN118296570A provides a road and bridge visual construction and operation and maintenance system and method based on BIM technology. By combining the data of BIM model and Internet of Things equipment, the relevant data of design, construction, operation and maintenance are integrated, the information islands are reduced, and the efficiency of information sharing and decision-making is improved; by constructing a stress measurement point distribution space matrix based on position data and stress values, and combining with artificial intelligence algorithms, the stress distribution state of the road and bridge is monitored in real time, the safety hazards in the construction process are discovered in time, and preventive maintenance is provided; real-time monitoring and data analysis are realized through Internet of Things equipment and BIM technology, the construction progress and operation and maintenance efficiency are improved, and the dependence on manual inspections is reduced.

[0004] However, although the existing technology has improved the efficiency and safety of road and bridge construction and operation and maintenance to a certain extent, it still has some shortcomings. It relies on a large number of sensors and IoT devices to obtain real-time data, and the collected stress data is discrete measurement point data. It is impossible to collect complete stress data for the bridge, and the warning and processing capabilities for abnormal situations are limited, and it is impossible to provide more accurate decision support.

[0005] In view of this, we proposed a road and bridge visual construction and operation and maintenance system and method based on BIM technology. Summary of the invention

[0006] The present invention provides a road and bridge visualized construction operation and maintenance system and method based on BIM technology, and the technical scheme includes the following aspects.

[0007] According to one aspect of an embodiment of the present application, a road and bridge visual construction and operation and maintenance system based on BIM technology is provided, and the system includes the following modules.

[0008] The data acquisition module is used to collect data, which includes design data, construction data and stress data. The design data of the bridge is extracted from the design software, the construction data during the construction process is recorded in real time, stress sensors are arranged at key structural parts of the bridge to obtain stress data, non-destructive testing is performed regularly, and stress distribution images are obtained.

[0009] The data preprocessing module cleans the collected data and removes noise and outliers.

[0010] The BIM model building module uses BIM technology to build a preliminary BIM model of the road and bridge according to the design data and construction data, and maps the stress distribution image to the preliminary BIM model using the spatial interpolation method to obtain the BIM model.

[0011] The feature extraction and analysis module applies a 3D convolutional neural network to extract stress features from the stress distribution BIM model and uses a contrast loss function to enhance the distinction between stress features in normal and abnormal areas.

[0012] The 3D graph construction and energy optimization module constructs a 3D graph based on the features extracted by the 3D convolutional neural network, applies the energy-based graph optimization method, optimizes node connectivity, and identifies potential abnormal areas.

[0013] The anomaly detection and clustering module analyzes the minimum energy and node connectivity, sets thresholds to identify abnormal areas, and uses clustering algorithms to determine the three-dimensional areas identified as abnormal stress distribution.

[0014] According to one aspect of the embodiment of the present application, a data acquisition module for a visual construction and operation system of a road and bridge based on BIM technology is provided. The data acquisition module uses an API or a data export tool to extract design data from the design software, including the geometric dimensions, load conditions, and material properties of the bridge, and stores the extracted data in a database to form a design data set; records the construction data of the construction site, inputs data through a mobile device, and the construction site management personnel update the progress, material consumption, and worker operation records in real time during the construction process, and stores the construction data in the database to form a construction data set; determines the installation position of the stress sensor according to the design and construction stage of the bridge, configures the sensor sampling frequency and monitoring parameters, and collects stress data; formulates a non-destructive testing plan, and uses ultrasonic testing for non-destructive testing in each testing cycle, and records the test results to generate a stress distribution image. All the collected data are uniformly stored in a centralized database, and cloud computing technology is used to ensure the real-time transmission and update of various types of data, which is convenient for subsequent analysis and decision-making. A data synchronization mechanism is used to ensure that the design data, construction data, and stress data can be updated in real time in the database to ensure data consistency and timeliness.

[0015] According to one aspect of an embodiment of the present application, a road and bridge visualization construction and operation and maintenance system based on BIM technology is provided, which uses the statistical method IQR to identify and remove outliers in collected data, IQR=Q3-Q1; wherein Q1 and Q3 are the first quartile and the third quartile of the collected data, respectively, and the thresholds are set as LowerBound=Q1-1.5×IQR, UpperBound=Q3+1.5×IQR, and the collected data below the lower bound LowerBound and above the upper bound UpperBound are regarded as outliers and removed; in the case of missing collected data, linear interpolation is used to fill it, y=y0+[(y1-y0)(x-x0)] / (x1-x0), wherein x0 and x1 are independent variables of known data points, y0 and y1 are corresponding dependent variables, and x is the independent variable to be interpolated.

[0016] According to one aspect of the embodiment of the present application, a BIM model building module of a road and bridge visual construction and operation and maintenance system based on BIM technology is provided. The BIM model building module receives bridge design data, construction data and stress distribution images, uses the design data to build a preliminary BIM model, and V BIM =∑ i=1 n V i , where V BIM is the volume of the overall preliminary BIM model, V iis the volume of the i-th building element; the stress distribution image is processed to ensure consistency with the spatial relationship of the preliminary BIM model, and the stress data is mapped to the corresponding area in the preliminary BIM model using a spatial interpolation method, S mapped =BIL(S image ,S BIM ), where S mapped To map the stress data onto the preliminary BIM model, S image is the stress distribution data in the stress distribution image, S BIM is the data of the corresponding area in the preliminary BIM model, and BIL is the bilinear interpolation method; combined with stress data for verification, S corrected =S mapped +k(S sensor -S mapped ), where S corrected is the corrected stress data, S sensor is the stress data, k is the weight coefficient, which controls the degree of correction; the corrected stress data is spatially interpolated to make a smooth transition, S final =BIL(S corrected ,S BIM ), where S final Corrected stress data for final mapping; Enhanced stress state display in BIM model through color coding, C stress =f1(S final ), where C stress is the color value of the stress state, and f1 is the mapping function, which converts the stress value into color coding.

[0017] According to one aspect of the embodiment of the present application, a feature extraction and analysis module of a road and bridge visual construction and operation and maintenance system based on BIM technology is provided. The feature extraction and analysis module uses a 3D convolutional neural network to extract stress features. The feature extraction and analysis module receives a BIM model M that maps a stress distribution image. BIM , contains the spatial distribution of stress data, which is expressed as a three-dimensional structure. The stress features of the stress data in the BIM model are extracted by using a 3D convolutional neural network. F=3D_CNN(M BIM ), where F is the extracted stress eigenvector.

[0018] Construct a contrast loss function, which is used to enhance the stress feature distinction between normal areas and abnormal areas. The loss function L is, L=1 / N∑ i=1 N (P i D(F i ,F j ) 2 +(1-P i )max(0,mD(Fi ,F j )) 2 ), where N is the sample size; P i is the label of sample i, the normal area is 1, and the abnormal area is 0; D(F i ,F j ) is the stress characteristic F i and F j The distance between them, using the Euclidean distance, D(F i ,F j )=[∑ k=1 d (F i (k) -F j (k) ) 2 ] 1 / 2 , where d is the dimension of the stress feature; m is the set threshold, which represents the minimum distance between the normal area and the abnormal area. During the training process, the parameters of the stress feature extraction model are adjusted by minimizing the contrast loss function L. Using the optimization algorithm, θ new =θ old -η▽L, where θ is the model parameter, η is the learning rate, and ▽L is the gradient of the loss function with respect to the parameter.

[0019] According to one aspect of the embodiment of the present application, a three-dimensional graph construction and energy optimization module for a road and bridge visual construction and operation system based on BIM technology is provided. The three-dimensional graph construction and energy optimization module constructs a three-dimensional graph based on the stress features extracted by a 3D convolutional neural network, and further optimizes the connectivity of the three-dimensional graph through an energy-based graph optimization method. By constructing a three-dimensional graph, the stress features are connected to the nodes and edges in the physical space. Each node v i Represents an area in the BIM model, and the corresponding stress characteristic is F i , three-dimensional coordinates (x i ,y i ,z i ) corresponds to the physical location in the BIM model, v i =(F i ,(x i ,y i ,z i ), where F i is the stress feature extracted by 3D convolutional neural network, (x i ,y i ,z i ) is the location of the node in the BIM model. Each edge e ij Connect two nodes v i and v j , the edge weight is defined as the geometric distance d(vi ,v j ) and stress characteristic distance D(F i ,F j ), w ij =αd(v i ,v j )+βD(F i ,F j ), where d(v i ,v j )=[(x i -x j ) 2 +(y i -y j ) 2 +(z i -z j ) 2 ] 1 / 2 is the geometric distance; D(F i ,F j ) is the Euclidean distance of the stress characteristic vector; α and β are weighting coefficients used to adjust the importance of geometric distance and stress characteristic distance. After the three-dimensional graph is constructed, the graph G=(V,E) contains a node set V and an edge set E.

[0020] According to one aspect of the embodiment of the present application, the three-dimensional graph construction and energy optimization module is based on energy-based graph optimization; the energy function is used to quantify the overall optimization goal of the three-dimensional graph, which is to minimize the connection energy between nodes, that is, to optimize the edge weights between nodes. The total energy EE is defined as the sum of the energies of all edges, EE=∑ (i,j)∈E w ij L(v i ,v j ), where L(v i ,v j ) is the connection loss function, which takes the similarity or difference between nodes; w ij is node v i and v j The edge weights between them are optimized by minimizing the energy function EE to identify potential abnormal regions, and the abnormal regions and normal regions are segmented based on the energy minimization algorithm.

[0021] According to one aspect of the embodiment of the present application, an anomaly detection and clustering module of a road and bridge visual construction and operation system based on BIM technology is provided. The anomaly detection and clustering module finds the edge set that minimizes the energy function EE through graph optimization. Low energy areas represent areas with high stress feature similarity and good connectivity, while high energy areas represent areas with abnormal stress distribution. After graph optimization, the optimized nodes are clustered by a clustering algorithm, and nodes with similar stress features are clustered into the same class, C={C1,C2,…,C k} is the clustering result, where C i Represents the node set of the i-th category, each category C i Each corresponds to a three-dimensional area, representing normal or abnormal stress distribution. By setting a threshold T, nodes with energy levels higher than the set threshold T are marked as abnormal. The threshold is set to, T=μ E +λσ E , where μ E and σ E are the mean and standard deviation of energy E, respectively. λ is the adjustment coefficient. If the energy EE of a certain area exceeds T, the area is judged as a stress abnormality area.

[0022] According to one aspect of the embodiment of the present application, a visualization and decision support module for a road and bridge visual construction and operation and maintenance system based on BIM technology is provided. The visualization and decision support module visualizes the anomaly detection results and the BIM model to help engineers intuitively identify potential structural problems and make decisions. After identifying the potential abnormal area, the abnormal area is combined with the three-dimensional BIM model of the bridge and visualized, that is, the coordinates, nodes and features of the identified abnormal area are superimposed on the BIM model. These nodes represent the distribution area of ​​stress anomalies. Specifically, through mapping technology, the nodes of the abnormal area correspond to the three-dimensional structure of the BIM model, and the abnormal area is marked with color coding. visual =f2(BIM model ,A anomalies ), where V visual Is the visualization result, BIM model is the BIM model of the bridge. anomaliesIt is the stress abnormal area obtained from the three-dimensional graph and clustering module, and f2 is the mapping function. In order to facilitate user operation and analysis, multi-dimensional display functions are provided, including 2D and 3D switching. Users can switch between 2D cross-sections and 3D overall models to obtain stress distribution views from different angles; zoom in and out, providing a zoom function to facilitate viewing detailed stress information in a key area; time series analysis, users can view stress changes in a certain period of time, understand stress trends at different time points, and discover potential structural changes. Automatically generate a stress analysis report, including stress values ​​and abnormal distribution areas in each area, to help managers understand the structural status of the entire bridge.

[0023] According to one aspect of an embodiment of the present application, a method for visual construction and operation of roads and bridges based on BIM technology is provided, and the method includes the following steps.

[0024] S1. Extract the design data of the bridge from the design software, record the construction data in real time during the construction process, arrange stress sensors at key structural parts of the bridge to obtain stress data, conduct non-destructive testing regularly, and obtain stress distribution images.

[0025] S2. Clean the design data, construction data and stress data to remove noise and outliers.

[0026] S3. Use BIM technology to build a BIM model of the road and bridge, including design and construction information, and map the stress distribution image into the BIM model using the spatial interpolation method.

[0027] S4. Apply 3D convolutional neural network to extract stress features from the stress distribution BIM model, and use contrast loss function to enhance the distinction between stress features in normal and abnormal areas.

[0028] S5. Construct a three-dimensional graph based on the stress features extracted by the 3D convolutional neural network, apply the energy-based graph optimization method to optimize the node connectivity and identify potential abnormal areas.

[0029] S6. By analyzing the minimum energy value and node connectivity, the abnormal area is identified, and the threshold is set to use the clustering algorithm to determine the three-dimensional area identified as abnormal stress distribution.

[0030] S7. Overlay the design data and construction process information on the BIM model to provide a visual display, and overlay the identified abnormal areas on the BIM model to provide an intuitive display.

[0031] The present application provides a road and bridge visual construction and operation and maintenance system and method based on BIM technology, which has the following advantages.

[0032] 1. This application improves the integrity and accuracy of the data. By using the three-dimensional model constructed using BIM technology, it achieves comprehensive coverage of the bridge structure, thereby overcoming the limitations of traditional data based on discrete measurement points, more accurately monitoring and analyzing the stress distribution state of the bridge, and providing more reliable data support for decision-making.

[0033] 2. This application strengthens the abnormal warning and processing capabilities. By applying 3D convolutional neural networks and energy-based graph optimization methods, the system can more effectively identify potential abnormal areas and accurately locate and classify abnormal areas through clustering algorithms. The accuracy of abnormal detection is improved, and safety hazards in the construction process can be discovered and handled in a timely manner, thereby avoiding possible accidents.

[0034] 3. This application optimizes the construction and operation and maintenance efficiency by superimposing design data, construction process information and real-time monitoring data on the BIM model to achieve visual display, so that construction and operation and maintenance personnel can intuitively understand the health status and construction progress of the bridge, which helps to improve work efficiency and reduce unnecessary on-site inspections and maintenance work, thereby reducing overall costs.

[0035] 4. This application provides intuitive display and decision support for construction and operation and maintenance personnel through visualization and decision support modules. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 A structural diagram of a road and bridge visual construction and operation and maintenance system based on BIM technology is provided for one embodiment of the present application.

[0037] Figure 2 A flow chart of a method for visual construction and operation of roads and bridges based on BIM technology is provided for one embodiment of the present application. DETAILED DESCRIPTION

[0038] In order to make the objectives, technical solutions and advantages of the present application clearer, the implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

[0039] Please refer to Figure 1 An embodiment of the present application provides a road and bridge visual construction and operation and maintenance system based on BIM technology, including the following modules.

[0040] The data acquisition module is used to collect data, including design data, construction data and stress data. The design data of the bridge is extracted from the design software, the construction data during the construction process is recorded in real time, stress sensors are arranged at key structural parts of the bridge to obtain stress data, non-destructive testing is performed regularly, and stress distribution images are obtained.

[0041] The data preprocessing module cleans the collected data and removes noise and outliers.

[0042] The BIM model building module uses BIM technology to build a preliminary BIM model of the road and bridge according to the design data and construction data, and maps the stress distribution image to the preliminary BIM model using the spatial interpolation method to obtain the BIM model.

[0043] The feature extraction and analysis module applies a 3D convolutional neural network to extract stress features from the stress distribution BIM model and uses a contrast loss function to enhance the distinction between stress features in normal and abnormal areas.

[0044] The 3D graph construction and energy optimization module constructs a 3D graph based on the features extracted by the 3D convolutional neural network, applies the energy-based graph optimization method, optimizes node connectivity, and identifies potential abnormal areas.

[0045] The anomaly detection and clustering module analyzes the minimum energy and node connectivity, sets thresholds to identify abnormal areas, and uses clustering algorithms to determine the three-dimensional areas identified as abnormal stress distribution.

[0046] The visualization and decision support module overlays the design data and construction process information on the BIM model to provide a visual display, and overlays the identified abnormal areas on the BIM model to provide an intuitive display.

[0047] As an optional solution of this application, the data acquisition module uses an API or data export tool to extract design data from the design software, including the geometric dimensions, load conditions, and material properties of the bridge, such as concrete strength and steel grade, and stores the extracted data in a database to form a design data set; record the construction data of the construction site, input data through mobile devices such as tablets and smart phones, and the construction site management personnel update the progress, material consumption and worker operation records in real time during the construction process, and store the construction data in the database to form a construction data set; according to the design and construction stage of the bridge, determine the installation location of the stress sensor, such as piers, main beams, and connection points, configure the sensor sampling frequency and monitoring parameters, and collect stress data; formulate a non-destructive testing plan, use ultrasonic testing equipment for non-destructive testing in each testing cycle, and record the test results to generate a stress distribution image. All collected data are uniformly stored in a centralized database, using cloud computing technology to ensure the real-time transmission and update of various types of data, facilitating subsequent analysis and decision-making, and using a data synchronization mechanism to ensure that the design data, construction data, and stress data can be updated in real time in the database to ensure data consistency and timeliness.

[0048] As an optional solution of the present application, the data preprocessing module uses the statistical method IQR to identify and remove outliers, IQR = Q3-Q1, where Q1 and Q3 are the first quartile and third quartile of the data, respectively, and the threshold is set to LowerBound = Q1-1.5×IQR, UpperBound = Q3+1.5×IQR, and the data below the lower bound LowerBound and above the upper bound UpperBound are regarded as outliers and removed. In the case of missing data, linear interpolation is used to fill in to ensure the continuity of the data, y = y0 + [(y1-y0)(x-x0)] / (x1-x0), where x0 and x1 are independent variables of known data points, y0 and y1 are the corresponding dependent variables, and x is the independent variable to be interpolated.

[0049] As an optional solution of the present application, the BIM model building module receives the bridge design data, construction stage data and stress distribution images. The design data is used to build a preliminary 3D BIM model, specifically, defining the geometry, size and position of building elements such as piers, main beams and bridge decks, V BIM =∑ i=1 n V i , where V BIM is the volume of the entire BIM model, V i is the volume of the i-th building element. The stress distribution image obtained by nondestructive testing is mapped to the BIM model. Specifically, the stress distribution image is processed to ensure that the spatial relationship is consistent with the BIM model, and a spatial interpolation method such as bilinear interpolation (BIL) is used to map the stress data to the corresponding area in the BIM model. mapped =BIL(S image ,S BIM ), where S mapped is the stress data mapped to the BIM model, S image is the stress distribution image, S BIM is the data of the corresponding area in the BIM model. It is verified in combination with stress data to improve the accuracy of the mapping. Specifically, the stress data is corrected, S corrected =S mapped +k(S sensor -S mapped ), where S corrected is the corrected stress data, S sensor is the stress data, k is the weight coefficient, which controls the degree of correction; the corrected stress data is spatially interpolated to make a smooth transition, S final =BIL(S corrected ,S BIM ), where S finalThe final mapped and corrected stress data. By color coding, the stress state display in the BIM model is enhanced. stress =f1(S final ), where C stress is the color value of the stress state, and f1 is the mapping function, which converts the stress value into color coding.

[0050] As an optional solution of the present application, the feature extraction and analysis module receives the BIM model M mapped with the stress distribution image. BIM , contains the spatial distribution of stress data, which is expressed as a three-dimensional structure. The stress features of the stress data in the BIM model are extracted by using a 3D convolutional neural network. F=3D_CNN(M BIM ), where F is the extracted stress feature vector. Construct a contrast loss function, which is used to enhance the stress feature distinction between normal areas and abnormal areas. Set the loss function L as follows: L=1 / N∑ i=1 N (P i D(F i ,F j ) 2 +(1-P i )max(0,mD(F i ,F j )) 2 ), where N is the sample size; P i is the label of sample i, the normal area is 1, and the abnormal area is 0; D(F i ,F j ) is the stress characteristic F i and F j The distance between them, using the Euclidean distance, D(F i ,F j )=[∑ k=1 d (F i (k) -F j (k) ) 2 ] 1 / 2 , where d is the dimension of the feature; m is the set threshold, which represents the minimum distance between the normal area and the abnormal area. During the training process, the parameters of the stress feature extraction model are adjusted by minimizing the contrast loss function L, using optimization algorithms such as Adam, θ new =θ old -η▽L, where θ is the model parameter, η is the learning rate, and ▽L is the gradient of the loss function with respect to the parameter.

[0051] As an optional solution of this application, the 3D graph construction and energy optimization module constructs a 3D graph based on the extracted features and further optimizes the connectivity of the 3D graph through an energy-based graph optimization method. By constructing a 3D graph, the stress features are connected to the nodes and edges in the physical space. Each node v i Represents an area in the BIM model, and the corresponding stress characteristic is F i , three-dimensional coordinates (x i ,y i ,z i ) corresponds to the physical location in the BIM model, v i =(F i ,(x i ,y i ,z i ), where F i is the stress feature extracted by 3D convolutional neural network, (x i ,y i ,z i ) is the location of the node in the BIM model. Each edge e ij Connect two nodes v i and v j , the edge weight is defined as the geometric distance d(v i ,v j ) and characteristic distance D(F i ,F j ), w ij =αd(v i ,v j )+βD(F i ,F j ), where d(v i ,v j )=[(x i -x j ) 2 +(y i -y j ) 2 +(z i -z j ) 2 ] 1 / 2 is the geometric distance; D(F i ,F j ) is the Euclidean distance of the feature vector; α and β are weighting coefficients used to adjust the importance of geometric distance and feature distance. After the three-dimensional graph is constructed, the graph G=(V,E) contains the node set V and the edge set E. The energy function is used to quantify the overall optimization goal of the three-dimensional graph, which is to minimize the connection energy between nodes, that is, to optimize the edge weights between nodes. The total energy EE is defined as the sum of the energies of all edges, EE=∑ (i,j)∈E w ijL(v i ,v j ), where L(v i ,v j ) is the connection loss function, which takes the similarity or difference between nodes; w ij is node v i and v j The edge weights between them are calculated by minimizing the energy function E to optimize the connectivity of the graph to identify potential abnormal areas. Algorithms based on energy minimization, such as the minimum cut algorithm, are used to separate abnormal areas from normal areas.

[0052] As an optional solution of this application, the anomaly detection and clustering module finds the edge set that minimizes the energy function EE through graph optimization. Low energy areas represent areas with high feature similarity and good connectivity, while high energy areas represent areas with abnormal stress distribution. After graph optimization, the optimized nodes are clustered using a clustering algorithm such as K-means to cluster nodes with similar stress characteristics into the same class, C={C1,C2,…,C k} is the clustering result, where C i Represents the node set of the i-th category, each category C i Each corresponds to a three-dimensional area, representing normal or abnormal stress distribution. By setting a threshold T, nodes with energy levels higher than the set threshold T are marked as abnormal. The threshold is set to, T=μ E +λσ E , where μ E and σ E are the mean and standard deviation of energy E, respectively. λ is the adjustment coefficient. If the energy E of a certain area exceeds T, the area is judged as a stress abnormality area.

[0053] As an optional solution of this application, the visualization and decision support module visualizes the anomaly detection results and the BIM model to help engineers intuitively identify potential structural problems and make decisions. After identifying the potential abnormal area, the abnormal area is combined with the three-dimensional BIM model of the bridge and visualized, that is, the coordinates, nodes and features of the identified abnormal area are superimposed on the BIM model. These nodes represent the distribution area of ​​stress anomalies. Specifically, through mapping technology, the nodes of the abnormal area correspond to the three-dimensional structure of the BIM model, and the abnormal area is marked with color coding. visual =f2(BIM model ,A anomalies ), where V visual Is the visualization result, BIM model is the BIM model of the bridge. anomaliesIt is the stress abnormal area obtained from the three-dimensional graph and clustering module, and f2 is the mapping function. In order to facilitate user operation and analysis, multi-dimensional display functions are provided, including 2D and 3D switching. Users can switch between 2D cross-sections and 3D overall models to obtain stress distribution views from different angles; zoom in and out, providing a zoom function to facilitate viewing detailed stress information; time series analysis, users can view stress changes in a certain period of time, understand stress trends at different time points, and discover potential structural changes. Automatically generate a stress analysis report, including stress values ​​and abnormal distribution areas in each area, to help managers understand the structural status of the entire bridge.

[0054] Please refer to Figure 2 An embodiment of the present application provides a road and bridge visual construction and operation method based on BIM technology, which includes the following steps.

[0055] S1. Extract the design data of the bridge from the design software, record the construction data in real time during the construction process, arrange stress sensors at key structural parts of the bridge to obtain stress data, conduct non-destructive testing regularly, and obtain stress distribution images.

[0056] S2. Clean the design data, construction data and stress data to remove noise and outliers.

[0057] S3. Use BIM technology to build a BIM model of the road and bridge, including design and construction information, and map the stress distribution image into the BIM model using the spatial interpolation method.

[0058] S4. Apply 3D convolutional neural network to extract stress features from the stress distribution BIM model, and use contrast loss function to enhance the distinction between stress features in normal and abnormal areas.

[0059] S5. Construct a three-dimensional graph based on the stress features extracted by the 3D convolutional neural network, apply the energy-based graph optimization method to optimize the node connectivity and identify potential abnormal areas.

[0060] S6. By analyzing the minimum energy value and node connectivity, the abnormal area is identified, and the threshold is set to use the clustering algorithm to determine the three-dimensional area identified as abnormal stress distribution.

[0061] S7. Overlay the design data and construction process information on the BIM model to provide a visual display, and overlay the identified abnormal areas on the BIM model to provide an intuitive display.

[0062] In this embodiment, step S1 realizes data collection and real-time update of the whole process from design, construction to monitoring by integrating mobile devices, stress sensors, ultrasonic detection equipment, cloud computing, and data synchronization mechanisms. This not only improves the efficiency and accuracy of data collection, but also ensures the unified storage and consistency of data, providing strong support for project management, quality control, bridge safety monitoring, etc. At the same time, cloud computing and data synchronization mechanisms ensure the real-time, sharing, and efficiency of data, which helps to make timely analysis and decision-making.

[0063] In this embodiment, step S2 identifies and removes outliers through IQR, and linear interpolation fills in missing values, which can significantly improve the quality of the data and ensure the accuracy and completeness of the data. Improve the reliability of the data: After removing outliers, the data can better reflect the actual situation and avoid interference of erroneous data on subsequent analysis. Enhance the continuity of the data: Use linear interpolation to fill in missing data to ensure the integrity of the data set, avoid data gaps or discontinuities caused by missing values, and ensure the stability of the analysis results. Improve the accuracy of the analysis: By cleaning and repairing the data, deviations and errors are reduced, providing a more accurate and reliable basis for subsequent data analysis, modeling, and decision-making.

[0064] In this embodiment, step S3 can achieve efficient fusion of stress data and structural geometric information by mapping the stress distribution image into the BIM model. The spatial interpolation method can infer the stress distribution of the missing area in the entire model space based on the collected stress data, and then provide accurate estimates for the stress conditions at each location. This method can improve the accuracy of stress analysis, allowing engineers to accurately locate the stress conditions of specific areas or structural components, promptly discover excessive stress or structural weaknesses, and provide data support for subsequent maintenance and reinforcement.

[0065] In this embodiment, step S4 combines a 3D convolutional neural network with a contrast loss function. This solution can automatically identify key abnormal stress features in large-scale three-dimensional stress distribution data. Compared with traditional manual detection or rule-based methods, this method has higher efficiency and accuracy. This solution reduces the workload of manual analysis, automatically identifies areas where problems may exist, and enables structural health monitoring to achieve real-time and automated BIL. It improves the accuracy of bridge monitoring, reduces omissions and false alarms, and helps engineers make more reliable decisions.

[0066] In this embodiment, step S5 extracts stress features through the BIL3D convolutional neural network BIL, constructs a three-dimensional graph, and optimizes node connectivity in combination with the BIL energy-based graph optimization method BIL. The solution has the following advantages: Accurately extract stress features and effectively capture spatial relationships. Optimize graph structure and accurately identify abnormal areas. Enhance anomaly detection and improve the sensitivity of structural health monitoring. Provide efficient decision support, which helps to carry out maintenance and repair scientifically and timely. Automation and intelligence reduce manual intervention and improve efficiency. Support monitoring and management of multiple structures and have good scalability.

[0067] In this embodiment, step S6 identifies the three-dimensional area of ​​abnormal stress distribution by analyzing the BIL energy minimum value BIL and the BIL node connectivity, combined with the BIL clustering algorithm BIL. The solution has the following advantages: Efficient and accurate identification of abnormal stress areas, avoiding misjudgment and omission of traditional methods. Automated abnormal area detection, reducing manual intervention and improving efficiency. Accurately analyze the spatial and multidimensional characteristics of stress distribution to help understand complex stress states. Optimize maintenance decisions and resource allocation to improve the economy and effectiveness of maintenance work. Enhance adaptability to complex structures and support health monitoring of large-scale and complex bridges. Support intelligent and real-time monitoring to provide forward-looking structural health warnings.

Claims

1. A visual construction and maintenance system for roads and bridges based on BIM technology, characterized in that: The system includes the following modules: A data acquisition module is used to collect data, including design data, construction data and stress data. The design data of the bridge is extracted from the design software, the construction data during the construction process is recorded in real time, stress sensors are arranged at key structural parts of the bridge to obtain stress data, non-destructive testing is performed regularly, and stress distribution images are obtained; Data preprocessing module cleans the collected data and removes noise and outliers; The BIM model building module uses BIM technology to build a preliminary BIM model of the road and bridge based on the design data and construction data, and maps the stress distribution image to the preliminary BIM model using the spatial interpolation method; Get the BIM model, Feature extraction and analysis module, which applies 3D convolutional neural network to extract stress features from the stress distribution BIM model and uses contrast loss function to enhance the distinction between stress features in normal and abnormal areas; The 3D graph construction and energy optimization module constructs a 3D graph based on the features extracted by the 3D convolutional neural network, applies an energy-based graph optimization method, optimizes node connectivity, and identifies potential abnormal areas; The anomaly detection and clustering module analyzes the minimum energy and node connectivity, sets thresholds to identify abnormal areas, and uses clustering algorithms to determine the three-dimensional areas identified as abnormal stress distribution.

2. According to the BIM technology-based road and bridge visual construction and operation and maintenance system according to claim 1, it is characterized in that: Data acquisition module: The data acquisition module uses an API or a data export tool to extract design data from the design software and stores the extracted data in a database to form a design data set; Record construction data at the construction site, input data through mobile devices, store the construction data in the database, and form a construction data set; Determine the installation location of stress sensors according to the design and construction stage of the bridge, install sensors to transmit monitoring sensor data in real time, configure sensor sampling frequency and monitoring parameters, and collect stress data; Develop a non-destructive testing plan, use ultrasonic testing to perform non-destructive testing in each testing cycle, record the test results, and generate stress distribution images.

3. According to the BIM technology-based road and bridge visual construction and operation and maintenance system according to claim 1, it is characterized in that: Data preprocessing module: Use the statistical method IQR to identify and remove outliers in the collected data, IQR = Q3-Q1; Among them, Q1 and Q3 are the first quartile and the third quartile of the collected data respectively. The threshold is set as LowerBound=Q1-1.5×IQR, UpperBound=Q3+1.5×IQR. The collected data below the lower bound LowerBound and above the upper bound UpperBound are regarded as outliers and removed; In the case of missing collected data, linear interpolation is used to fill it in, y=y0+[(y1-y0)(x-x0)] / (x1-x0), where x0 and x1 are independent variables with known data points, y0 and y1 are the corresponding dependent variables, and x is the independent variable to be interpolated.

4. According to the BIM technology-based road and bridge visual construction and operation and maintenance system according to claim 1, it is characterized in that: BIM model building modules: The BIM model building module receives bridge design data, construction data and stress distribution images, and uses the design data to build a preliminary BIM model. BIM =∑ i=1 n V i , where V BIM is the volume of the overall preliminary BIM model, V i is the volume of the i-th building element; The stress distribution image is processed to ensure consistency with the spatial relationship of the preliminary BIM model, and the stress data is mapped to the corresponding areas in the preliminary BIM model using spatial interpolation methods. mapped =BIL(S image ,S BIM ), where S mapped To map the stress data onto the preliminary BIM model, S image is the stress distribution data in the stress distribution image, S BIM is the data of the corresponding area in the preliminary BIM model, and BIL is the bilinear interpolation method; Combined with stress data for verification, S corrected =S mapped +k(S sensor -S mapped ), where S corrected is the corrected stress data, S sensor is stress data, k is weight coefficient, which controls the degree of correction; The corrected stress data is spatially interpolated to smooth the transition, S final =BIL(S corrected ,S BIM ), where S final Corrected stress data for final mapping; Enhance the display of stress states in BIM models through color coding, stress =f1(S final ), where C stress is the color value of the stress state, and f1 is the mapping function, which converts the stress value into color coding.

5. According to the BIM technology-based road and bridge visual construction and operation and maintenance system according to claim 1, it is characterized in that: The feature extraction and analysis module uses 3D convolutional neural networks to extract stress features: The feature extraction and analysis module receives the BIM model M mapped with the stress distribution image. BIM , contains the spatial distribution of stress data, showing a three-dimensional structure; By using 3D convolutional neural network to extract stress features from stress data in BIM model, F=3D_CNN(M BIM ), where F is the extracted stress eigenvector.

6. According to the BIM technology-based road and bridge visual construction and operation and maintenance system of claim 5, it is characterized in that: The feature extraction and analysis module uses a contrast loss function: Constructing a contrast loss function, which is used to enhance the distinction of stress characteristics between normal areas and abnormal areas; The loss function L is, L=1 / N∑ i=1 N (P i D(F i ,F j ) 2 +(1-P i )max(0,mD(F i ,F j )) 2 ), where N is the sample size, P i is the label of sample i, the normal area is 1 and the abnormal area is 0; D(F i ,F j ) is the stress characteristic F i and F j The distance between them, using the Euclidean distance, D(F i ,F j )=[∑ k=1 d (F i (k) -F j (k) ) 2 ] 1 / 2 , where d is the dimension of the stress feature; m is the set threshold, indicating the minimum distance between the normal area and the abnormal area; During the training process, the parameters of the stress feature extraction model are adjusted by minimizing the contrast loss function L, using an optimization algorithm, θ new =θ old -η▽L, where θ is the model parameter, η is the learning rate, and ▽L is the gradient of the loss function with respect to the parameter.

7. According to the BIM technology-based road and bridge visual construction and operation and maintenance system according to claim 1, it is characterized in that: 3D graph construction and energy optimization module constructs 3D graph: Construct a 3D graph based on the stress features extracted by the 3D convolutional neural network, and connect the stress features with the nodes and edges in the physical space; Each node v i Represents an area in the BIM model, and the corresponding stress characteristic is F i , three-dimensional coordinates (x i ,y i ,z i ) corresponds to the physical location in the BIM model, v i =(F i ,(x i ,y i ,z i )), where (x i ,y i ,z i ) is the location of the node in the BIM model; Each edge e ij Connect two nodes v i and v j , the edge weight is defined as the geometric distance d(v i ,v j ) and stress characteristic distance D(F i ,F j ), w ij =αd(v i ,v j )+βD(F i ,F j ), where d(v i ,v j )=[(x i -x j ) 2 +(y i -y j ) 2 +(z i -z j ) 2 ] 1 / 2 is the geometric distance, D(F i ,F j ) is the Euclidean distance of the stress characteristic vector, α and β are weighting coefficients used to adjust the importance of geometric distance and stress characteristic distance; After the three-dimensional graph is constructed, the graph G=(V,E) contains the node set V and the edge set E.

8. According to claim 7, a visual construction and operation system for roads and bridges based on BIM technology is characterized in that: Dimensional graph construction and energy optimization module Energy-based graph optimization: The energy function is used to quantify the overall optimization goal of the three-dimensional graph, which is to minimize the connection energy between nodes, that is, to optimize the edge weights between nodes; Define the total energy EE as the sum of the energies of all edges, EE=∑ (i,j)∈E w ij L(v i ,v j ), where L(v i ,v j ) is the connection loss function, which takes the similarity or difference between nodes, w ij is node v i and v j The edge weights between The connectivity of the three-dimensional graph is optimized by minimizing the energy function EE to identify potential abnormal areas, and the abnormal and normal areas are segmented based on the energy minimization algorithm.

9. According to the BIM technology-based road and bridge visual construction and operation and maintenance system of claim 1, it is characterized by: Anomaly detection and clustering module: Through graph optimization, the edge set that minimizes the energy function EE is found. Low energy areas represent areas with high feature similarity and good connectivity, while high energy areas represent areas with abnormal stress distribution. After graph optimization, the optimized nodes are clustered by clustering algorithm, and nodes with similar stress characteristics are clustered into the same class, C={C1,C2,…,C k } is the clustering result, where C i Represents the node set of the i-th category, each category C i Each corresponds to a three-dimensional region, representing normal or abnormal stress distribution; By setting a threshold T, nodes with energy levels higher than the set threshold T are marked as abnormal. The threshold is set to, T = μ E +λσ E , where μ E and σ E are the mean and standard deviation of energy E, respectively, and λ is the adjustment coefficient; If the energy EE of a certain area exceeds T, the area is judged as a stress abnormality area.

10. The visual construction and operation system for roads and bridges based on BIM technology according to claim 1 is characterized in that: Also includes: Visualization and decision support module: Visualize the anomaly detection results and BIM models; After identifying potential abnormal areas, the nodes in the abnormal areas are mapped to the 3D structure of the BIM model through mapping technology, and the abnormal areas are marked with color coding. visual =f2(BIM model ,A anomalies ), where V visual Is the visualization result, BIM model is the BIM model of the bridge. anomalies is the stress anomaly area obtained from the three-dimensional graph and clustering module, and f2 is the mapping function; Provide multi-dimensional display functions, including 2D and 3D switching, zooming in and out, and time series analysis; Automatically generate stress analysis reports, including stress values ​​in each area and abnormal distribution areas.

11. A road and bridge visualized construction and operation method based on BIM technology, applied to a road and bridge visualized construction and operation system based on BIM technology as claimed in any one of claims 1 to 10, characterized in that: include: S1. Extract the design data of the bridge from the design software, record the construction data in real time during the construction process, arrange stress sensors at key structural parts of the bridge to obtain stress data, conduct non-destructive testing regularly, and obtain stress distribution images; S2. Clean the design data, construction data and stress data to remove noise and outliers; S3. Use BIM technology to build a BIM model of the road and bridge, including design and construction information, and map the stress distribution image to the BIM model using spatial interpolation method; S4. Apply 3D convolutional neural network to extract stress features from stress distribution BIM model, using contrast loss function to enhance the distinction between stress features of normal and abnormal areas; S5. Construct a three-dimensional graph based on the stress features extracted by the 3D convolutional neural network, apply the energy-based graph optimization method to optimize node connectivity and identify potential abnormal areas; S6, identifying abnormal areas by analyzing energy minimum values ​​and node connectivity, setting thresholds and using clustering algorithms to determine three-dimensional areas identified as abnormal stress distribution; S7. Overlay the design data and construction process information on the BIM model to provide a visual display, and overlay the identified abnormal areas on the BIM model to provide an intuitive display.

Images