Method, system and equipment for analyzing project progress based on BIM model and AI video, and medium

Abstract

The invention discloses a method, a system, equipment and a medium for analyzing project progress based on a BIM model and an AI video, and relates to the technical field of constructional engineering management, and the method comprises the steps of collecting regional image data, and executing semantic segmentation and environment modeling. And constructing a multi-scale spatial feature network and a dynamic skeleton modeling engine, inputting the three-dimensional environment diagram, the structure recognizable region diagram and the obstacle probability distribution diagram into the multi-scale spatial feature network and the dynamic skeleton modeling engine, and outputting execution structure recognition and defect anomaly labeling. And triggering an engineering response mechanism based on the risk event scoring function and remotely and dynamically adjusting the construction path. According to the method, through a mode of combining semantic segmentation and three-dimensional environment modeling, high-precision structure identification of the construction site image is realized, and the problems of unclear structure identification and fuzzy component boundary in traditional video monitoring are effectively solved.

Description

Technical Field

[0001] The present invention relates to the technical field of construction project management, and specifically to a method, system, device, and medium for engineering progress based on BIM models and AI video analysis. Background Art

[0002] With the continuous development of building information modeling technology, its application in the field of construction engineering has become increasingly widespread. Especially in the whole-process management of engineering planning, design, construction, and operation and maintenance, BIM models have become an important means to achieve visualization, collaboration, and informatization management. At the same time, the maturity of artificial intelligence and computer vision technology has also provided a technical foundation for the automatic monitoring and intelligent analysis of the engineering site, making it possible to transform from traditional manual inspections to intelligent, automated, and data-driven monitoring methods.

[0003] In the construction stage of engineering projects, engineering progress management, as one of the core contents of project management, plays a key role in ensuring the completion of projects on schedule, rationally allocating resources, and controlling costs. The current mainstream engineering progress management methods mostly rely on progress management software such as Project combined with manual photo-taking and manual reporting to confirm engineering nodes and compare plans. However, such methods have obvious limitations: on the one hand, there are problems such as strong subjectivity of information, lagging updates, and difficulty in obtaining on-site data in manual recording of engineering progress, seriously affecting the real-time nature and accuracy of project control; on the other hand, although some engineering projects have tried to conduct progress plan and visualization comparison through BIM models, this process still highly depends on manual maintenance of BIM progress status and lacks automatic data collection and dynamic perception capabilities.

[0004] Existing engineering progress management methods have the problems that progress data acquisition mainly relies on manual collection and manual entry, lacking intelligence and real-time nature, it is difficult for AI recognition results to accurately match BIM model components, lacking a semantic alignment mechanism at the component level, lacking the engineering response ability based on structure recognition and risk scoring mechanisms, and being unable to make dynamic adjustments and optimization decisions according to the recognition results. Summary of the Invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] Therefore, the technical problems solved by the present invention are: existing engineering progress management and monitoring technical methods have problems such as data collection relying on manual work, low progress recognition accuracy, and untimely information update, and how to automatically identify the construction structure state based on on-site video data and achieve accurate matching with the BIM model, so as to dynamically analyze the actual construction progress and make path optimization adjustments.

[0007] To solve the above technical problems, the present invention provides the following technical solutions: A method for analyzing the engineering progress based on a BIM model and AI video, including collecting regional image data, performing semantic segmentation and environmental modeling.

[0008] Construct a multi-scale spatial feature network and a dynamic skeleton modeling engine.

[0009] Input the three-dimensional environmental map, the structure recognizable area map, and the obstacle probability distribution map into the multi-scale spatial feature network and the dynamic skeleton modeling engine, and output the execution of structure recognition and defect anomaly annotation.

[0010] Trigger the engineering response mechanism based on the risk event scoring function and remotely dynamically adjust the construction path.

[0011] The three-dimensional environmental map includes generating a three-dimensional environmental map of the structural boundary information, the occlusion area position, and the construction path by environmental modeling.

[0012] The structure recognizable area map and the obstacle probability distribution map include inputting the collected regional image data into the constructed deep convolutional neural network model to obtain the structure recognizable area map and the obstacle probability distribution map.

[0013] Constructing a multi-scale spatial feature network and a dynamic skeleton modeling engine includes identifying target components and structural unit nodes through a multi-scale residual graph neural network and a skeleton fitting engine and outputting a structure map.

[0014] As a preferred solution of the method for analyzing the engineering progress based on a BIM model and AI video according to the present invention, wherein: the collecting of the regional image data includes using an infrared thermal imaging device, a visible light high-definition camera, and a lidar sensing component mounted on the construction site to perform area coverage scanning on the target building structure at set time intervals, and performing unified time series numbering and spatial registration operations on the multi-modal images and point cloud data collected in the scanned area.

[0015] The spatial registration operation includes using a sparse feature matching algorithm based on laser SLAM to perform rough alignment on the structural consistency between image frames, performing fine-grained registration on the basis of the point cloud, and binding each frame of data to the building digital map model through the timestamp and positioning information to form a perception frame cluster.

[0016] As a preferred solution of the method for analyzing the engineering progress based on a BIM model and AI video according to the present invention, wherein: the performing of semantic segmentation and environmental modeling includes using a deep convolutional neural network model to perform multi-channel feature extraction on the collected image data, constructing a semantic segmentation map, respectively labeling the structure, personnel, equipment, and abnormal areas as semantic category layers, and using a strategy based on spatial consistency and gray voting for misclassification correction.

[0017] Based on the segmentation map, by modeling the environment, generate a three-dimensional environmental map with structural boundary information, the positions of occluded areas, and the passage paths.

[0018] Generating the three-dimensional environmental map includes constructing the three-dimensional environmental map using an edge surface fitting method based on dense point cloud interpolation.

[0019] Introduce voxel sparsification operations to compress the data volume, and output the structure recognizable area map and the obstacle probability distribution map.

[0020] As a preferred solution of the method for analyzing the engineering progress based on the BIM model and AI video described in the present invention, wherein: the construction of the multi-scale spatial feature network and the dynamic skeleton modeling engine includes performing multi-source processing on the three-dimensional environmental map, the structure recognizable area map, and the obstacle probability distribution map, and fusing the displacement change data in the time series on the three-dimensional environmental map, the structure recognizable area map, and the obstacle probability distribution map after multi-source fusion processing to construct a dynamic graph structure, and identifying the target components and the structural unit nodes.

[0021] Identifying the target components and the structural unit nodes includes adopting a feature aggregation mechanism based on the K-hop neighborhood relationship, taking each candidate node as the center, propagating the surrounding topology and spatial features through multi-order graph convolution operations, outputting the embedding vector of each node, and classifying the nodes into beam, slab, and column structural unit types by setting a classification discriminant head.

[0022] Use a structural line model based on skeleton fitting to establish a skeleton connection map of the building edge line and the key nodes, use an adaptive curvature enhancement strategy to correct the node sparse area, and enhance the continuity and integrity of the overall structure recognition through a graph regularization algorithm, and output the structure spectrum.

[0023] Establishing the skeleton connection map of the building edge line and the key nodes includes selecting the key points with high boundary response intensity and significant curvature change in the structural feature map as the skeleton fitting nodes, generating a preliminary skeleton map through an initial connection method based on the spatial minimum spanning tree, and constructing an optimized skeleton connection relationship by combining the direction gradient and the normal vector consistency between the nodes.

[0024] As a preferred solution of the method for analyzing the engineering progress based on the BIM model and AI video described in the present invention, wherein: the defect anomaly annotation includes extracting the visual feature distribution and the thermal anomaly layer in the potential anomaly area according to the optimized skeleton connection relationship, and performing matching calculations with the known defect templates in the anomaly knowledge base to obtain the defect anomaly annotation result.

[0025] The known defect templates include crack propagation patterns, equipment detachment forms, and heat concentration anomalies.

[0026] The matching includes implementing matching calculations using an abnormal matching algorithm that combines cosine distance and spatial clustering for judgment.

[0027] As a preferred solution of the method for engineering progress based on BIM model and AI video analysis according to the present invention, wherein: the engineering response mechanism triggered by the risk event scoring function includes constructing a risk event vector sequence according to the defect abnormal annotation results, taking the continuous abnormal data window during the inspection as the input, using the LSTM-Attention double-layer network structure to predict the risk evolution trend in the future time period, and performing risk quantification assessment on the current state to produce an actual engineering progress analysis report.

[0028] As a preferred solution of the method for engineering progress based on BIM model and AI video analysis according to the present invention, wherein: the remote dynamic adjustment of the construction path includes recalculating the optimal construction path according to the current building structure map, the defect position coordinates, the wind field, and the occlusion.

[0029] Recalculating the optimal construction path includes using a multi-objective A improvement algorithm based on heuristic guidance.

[0030] Another object of the present invention is to provide a system for engineering progress based on BIM model and AI video analysis, which can realize high-precision recognition of the construction structure and generation of semantic maps by constructing a multi-scale graph neural network and a dynamic skeleton modeling engine, and solves the problems that the current BIM+AI combined progress management technology relies on manual progress updates and lacks component-level recognition and risk response capabilities.

[0031] As a preferred solution of the system for engineering progress based on BIM model and AI video analysis according to the present invention, wherein: it includes a data acquisition module, a defect annotation module, and an engineering adjustment module.

[0032] The data acquisition module is used to collect regional image data and perform semantic segmentation and environmental modeling.

[0033] The defect annotation module is used to construct a multi-scale spatial feature network and a dynamic skeleton modeling engine, input the three-dimensional environmental map, the structure recognizable area map, and the obstacle probability distribution map into the multi-scale spatial feature network and the dynamic skeleton modeling engine, and output to perform structure recognition and defect abnormal annotation.

[0034] The engineering adjustment module is used to trigger the engineering response mechanism based on the risk event scoring function and remotely dynamically adjust the construction path.

[0035] A computer device includes a memory and a processor, the memory stores a computer program, and the processor executes the computer program to implement the steps of the method for engineering progress based on BIM model and AI video analysis.

[0036] A computer-readable storage medium stores a computer program thereon, and when the computer program is executed by a processor, it implements the steps of a method for engineering progress based on a BIM model and AI video analysis.

[0037] Advantages of the present invention: The method for engineering progress based on a BIM model and AI video analysis provided by the present invention combines semantic segmentation with three-dimensional environmental modeling to achieve high-precision structure recognition of construction site images, effectively solving the problems of unclear structure recognition and blurred component boundaries in traditional video surveillance. By constructing a multi-scale spatial feature network and a dynamic skeleton modeling engine, a structure atlas is formed and the component-level defect abnormal areas are accurately marked, realizing automatic recognition of the construction structure and precise positioning of defects, overcoming the problem that the construction status cannot be automatically recognized in the existing BIM model progress management. Using the LSTM-Attention network structure to predict the trend and quantitatively evaluate risk events, automatically triggering the engineering response mechanism, realizing an intelligent closed-loop from structure recognition to risk decision-making, and solving the problem of delayed discovery of progress anomalies. Introducing a multi-objective improvement algorithm combined with environmental dynamic factors to calculate the optimal construction path, improving the flexibility and real-time performance of construction management path adjustment, avoiding path redundancy and safety risks brought by manual adjustment. The present invention has achieved better results in terms of structure recognition accuracy, construction progress response efficiency, and intelligent management ability of the engineering site. Description of the Drawings

[0038] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for description in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings based on these drawings without creative efforts.

[0039] Figure 1 It is the overall flowchart of the method for engineering progress based on a BIM model and AI video analysis provided by the first embodiment of the present invention. Specific Embodiments

[0040] To make the above objects, features, and advantages of the present invention more obvious and understandable, the specific embodiments of the present invention will be described in detail below with reference to the drawings of the specification. Obviously, the described embodiments are some embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0041] Example 1, referring to Figure 1 , which is an embodiment of the present invention, provides a method for engineering progress based on a BIM model and AI video analysis, including: S1: Collect image data of the area, and perform semantic segmentation and environmental modeling.

[0042] Using the infrared thermal imaging device, visible light high-definition camera, and lidar sensing component mounted on the construction site, perform area coverage scanning on the target building structure at set time intervals, and perform unified time series numbering and spatial registration operations on the multi-modal images and point cloud data collected within the scanned area.

[0043] A preferred solution for performing unified time series numbering and spatial registration operations is: , where , , represent the coordinate values of the scanning points on the construction site in three-dimensional space, represents the abscissa of the pixel point of the construction feature point, represents the ordinate of the pixel point of the construction feature point, represents the focal length of the construction equipment camera in the direction, represents the focal length of the construction equipment camera in the direction, , represent the coordinate positions of the principal point of the construction equipment camera on the image, represents the depth data corresponding to the construction feature point.

[0044] The spatial registration operation includes using a sparse feature matching algorithm based on laser SLAM to perform rough alignment to maintain the consistency of the image frame structure, and performing fine-grained registration based on the point cloud.

[0045] A preferred solution for performing fine-grained registration based on the point cloud is: , where represents the mean square error of the registration between the construction point cloud and the BIM component anchor point, represents the number of pairs of feature points participating in the registration in the construction point cloud, represents the th scanning point coordinate in the construction point cloud, represents the BIM component node anchor point coordinate corresponding to in the target point cloud, represents the registration rotation matrix at the construction site, represents the registration translation vector at the construction site, represents the Euclidean norm.

[0046] Bind each frame of data to the building digital map model through timestamps and positioning information to form a perception frame cluster.

[0047] Use a deep convolutional neural network model to extract multi-channel features from the collected image data and construct a semantic segmentation map.

[0048] A preferred solution for constructing the semantic segmentation map is: , where, represents the pixel gray value after normalizing the construction site image, represents the gray value of the pixel point in the original construction image, represents the minimum gray value in the construction image, represents the maximum gray value in the construction image.

[0049] Label the structure, personnel, equipment, and abnormal areas as semantic category layers respectively, and adopt a strategy based on spatial consistency and gray voting to correct misclassifications.

[0050] Based on the segmentation map, generate a three-dimensional environment map by modeling the environment, including the structural boundary information, the position of the occluded area, and the construction path.

[0051] Generating the three-dimensional environment map includes using an edge surface fitting method based on dense point cloud interpolation for constructing the three-dimensional environment map.

[0052] Introduce a voxel sparsification operation to compress the data volume and output a structural recognizable area map and an obstacle probability distribution map.

[0053] A preferred solution for outputting the three-dimensional environment map, the structural recognizable area map, and the obstacle probability distribution map is: , where, represents the structural strength of the construction area, represents the construction image at the gradient value in the direction, represents the gradient value of the image in the

[0054] Use the obtained structural strength of the construction area as the input for the three-dimensional modeling of the construction environment and the occlusion recognition link.

[0055] S2: Construct a multi-scale spatial feature network and a dynamic skeleton modeling engine.

[0056] Use a structural line model based on skeleton fitting to establish a skeleton connection map of the building edge line and key nodes, use an adaptive curvature enhancement strategy to correct the sparse node area, and enhance the continuity and integrity of the overall structure recognition through a graph regularization algorithm, and output a structural map.

[0057] A preferred solution for establishing the skeleton connection diagram between the building edge line and the key nodes is as follows: , wherein, represents the angle between the connection vector between construction nodes and the local normal vector, represents the construction node and 's connection vector, represents the normal vector of the construction node corresponding to the connection line direction, represents the dot product operation, represents the Euclidean norm.

[0058] Establishing the skeleton connection diagram between the building edge line and the key nodes includes selecting the key points with high boundary response intensity and significant curvature change in the structural feature map as the skeleton fitting nodes, generating a preliminary skeleton diagram through the initial connection method based on the spatial minimum spanning tree, and constructing an optimized skeleton connection relationship by combining the direction gradient and normal vector consistency between nodes.

[0059] S3: Input the three-dimensional environmental map, the structural recognizable area map, and the obstacle probability distribution map into the multi-scale spatial feature network and the dynamic skeleton modeling engine, and output the execution of structural recognition and defect anomaly annotation.

[0060] Perform multi-source processing on the three-dimensional environmental map, the structural recognizable area map, and the obstacle probability distribution map, fuse the three-dimensional environmental map, the structural recognizable area map, and the obstacle probability distribution map after multi-source fusion processing, and fuse the displacement change data in the time series to construct a dynamic graph structure to identify the target components and structural unit nodes.

[0061] Identifying the target components and structural unit nodes includes adopting a feature aggregation mechanism based on the K-hop neighborhood relationship, taking each candidate node as the center, propagating the surrounding topology and spatial features through multi-order graph convolution operations, outputting the embedding vector of each node, and classifying the nodes into beam, slab, and column structural unit types by setting a classification discriminant head.

[0062] A preferred solution for the discriminant head to classify the nodes into beam, slab, and column structural unit types is as follows: , wherein, represents the embedding vector of the th construction node after the th layer, represents the learnable weight matrix of the construction drawing spectrum of the <o:p>< / o:p> th layer, represents the activation function, represents the construction node feature aggregation function of the th layer, represents the construction node The adjacent node of The embedding representation of the layer, Indicates construction node The set of neighbors.

[0063] The structural characteristics of each construction node are propagated through the characteristics of adjacent nodes, enhancing the discrimination ability of local components.

[0064] According to the optimized skeleton connection relationship, the visual feature distribution and thermal anomaly layer in the potential abnormal area are extracted, and matching calculations are performed against the known defect templates in the anomaly knowledge base to obtain the defect anomaly labeling results.

[0065] An optimal solution for defect anomaly marking is: , in, represents two feature vectors for comparison, express The cosine similarity between .

[0066] Known defect templates include crack propagation patterns, equipment shedding patterns, and thermal concentration anomalies.

[0067] Matching includes using an abnormal matching algorithm based on the joint judgment of cosine distance and spatial clustering to realize matching calculation.

[0068] S3: Trigger the engineering response mechanism based on the risk event scoring function and remotely and dynamically adjust the construction path.

[0069] Based on the defect anomaly labeling results, a risk event vector sequence is constructed. Continuous abnormal data windows during the inspection process are used as input. The LSTM-Attention two-layer network structure is used to predict the risk evolution trend in the future time period, and a risk quantitative assessment of the current status is performed to produce an analysis report on the actual project progress.

[0070] A preferred solution for risk quantitative assessment is: , , , , in, Indicates the The risk scoring results of the construction nodes are: represents the nonlinear feature adjustment factor, represents the Gaussian penalty factor, represents the status difference factor of the construction node, Indicates the Weighting coefficient of construction indicators represents the value of the th construction node under the th type of construction indicator represents the mean value of the th type of construction indicator in the global scope represents the standard deviation of the th type of construction indicator represents a small constant to avoid division by zero represents the penalty coefficient of the th type of construction indicator represents the ideal reference value of the th type of construction indicator represents the th construction node under the th type of construction indicator represents the total number of nodes

[0071] Recalculate the optimal construction path according to the current building structure map, defect location coordinates, wind field, and occlusion

[0072] Recalculating the optimal construction path includes using a multi-objective improved algorithm based on heuristic guidance

[0073] A preferred solution of the multi-objective improved algorithm based on heuristic guidance is as follows , wherein represents the total cost of the construction node represents the actual cost from the starting point to the construction node represents the heuristic estimated cost from the construction node to the end point

[0074] Embodiment 2 is an embodiment of the present invention, providing an engineering progress system based on BIM model and AI video analysis, including a data acquisition module, a defect annotation module, and an engineering adjustment module

[0075] Among them: The data acquisition module is used to collect regional image data and perform semantic segmentation and environmental modeling

[0076] It should also be noted that the data acquisition module, as the front-end perception unit of the system, collects multi-modal raw data of the construction site, constructs a multi-source synchronous data frame through spatial registration and temporal numbering, performs image semantic segmentation, generates multi-category semantic layers, reconstructs and fits point clouds, outputs a three-dimensional environmental map, and performs voxel sparsification processing to generate a structure recognizable area map and an obstacle probability distribution map

[0077] ​​The processed structure diagram, obstacle diagram, and semantic layer are used as the input data for the defect annotation module, providing the multi-source basis required for constructing the graph neural network and skeleton structure modeling.

[0078] Among them: The defect annotation module is used to construct a multi-scale spatial feature network and a dynamic skeleton modeling engine. The three-dimensional environmental map, structure recognizable area map, and obstacle probability distribution map are input into the multi-scale spatial feature network and the dynamic skeleton modeling engine, and the output is for performing structure recognition and defect anomaly annotation.

[0079] It should also be noted that the defect annotation module undertakes the core recognition and analysis functions of the system, and outputs the structure unit node atlas under the multi-scale spatial feature network, the structure contour map based on skeleton connection, the defect anomaly annotation results generated by the fusion of the heat layer and image features, the embedding vectors, status indicators, and node type labels of each node component.

[0080] The results are used as the input for the risk event modeling of the project adjustment module, which makes intelligent adjustments to the construction path and operation plan by analyzing the abnormal distribution and risk trend.

[0081] Among them: The project adjustment module is used to trigger the project response mechanism based on the risk event scoring function and remotely dynamically adjust the construction path.

[0082] It should also be noted that after the project adjustment module completes the risk scoring and path recalculation, it feeds back the adjusted optimal construction path, risk level spatial distribution map, node priority processing order, and re-sampling request for the specified area to the data acquisition module, realizing the dynamic optimization control of the acquisition range and frequency, ensuring that the subsequent collected data can perform higher-frequency and more refined scanning and sampling for the risk area, and forming a closed loop of data - analysis - feedback.

Claims

1. A method for analyzing project progress based on BIM models and AI videos, characterized in that, including: Collecting regional image data and performing semantic segmentation and environmental modeling; Constructing a multi-scale spatial feature network and a dynamic skeleton modeling engine; Inputting the 3D environmental map, the structurally recognizable area map, and the obstacle probability distribution map into the multi-scale spatial feature network and the dynamic skeleton modeling engine, and outputting the execution of structural recognition and defect anomaly annotation; Triggering an engineering response mechanism based on a risk event scoring function and remotely dynamically adjusting the construction path; The 3D environmental map includes generating a 3D environmental map of structural boundary information, occlusion area positions, and construction paths through environmental modeling; The structurally recognizable area map and the obstacle probability distribution map include inputting the collected regional image data into a pre-constructed deep convolutional neural network model to obtain the structurally recognizable area map and the obstacle probability distribution map; Constructing a multi-scale spatial feature network and a dynamic skeleton modeling engine includes identifying target components and structural unit nodes through a multi-scale residual graph neural network and a skeleton fitting engine and outputting a structural map.

2. The method for engineering progress based on BIM model and AI video analysis according to claim 1, wherein: The collection of regional image data includes Using an infrared thermal imaging device, a visible light high-definition camera, and a lidar sensing component mounted at the construction site to perform area coverage scanning of the target building structure at set time intervals, and performing unified time series numbering and spatial registration operations on the multi-modal images and point cloud data collected within the scanned area; The spatial registration operation includes adopting a sparse feature matching algorithm based on laser SLAM to perform rough alignment on the structural consistency between image frames, performing fine-grained registration based on the point cloud, and binding each frame of data to the building digital map model through timestamps and positioning information to form a perception frame cluster.

3. The method for engineering progress based on BIM model and AI video analysis according to claim 1 or 2, characterized in that: The execution of semantic segmentation and environmental modeling includes Using a deep convolutional neural network model to perform multi-channel feature extraction on the collected image data, constructing a semantic segmentation map, labeling structures, personnel, equipment, and abnormal areas as semantic category layers respectively, and adopting a strategy based on spatial consistency and gray voting for misclassification correction; Generating a 3D environmental map based on environmental modeling on the segmentation map, generating a 3D environmental map of structural boundary information, occlusion area positions, and construction paths; Generating a 3D environmental map includes that the 3D environmental map construction adopts an edge surface fitting method based on dense point cloud interpolation; Introducing a voxel sparsification operation to compress the data volume and outputting the structurally recognizable area map and the obstacle probability distribution map.

4. The method for analyzing the engineering progress based on the BIM model and AI video as claimed in claim 3, wherein: The construction of the multi-scale spatial feature network and the dynamic skeleton modeling engine includes Performing multi-source processing on the 3D environmental map, the structurally recognizable area map, and the obstacle probability distribution map, fusing the 3D environmental map, the structurally recognizable area map, and the obstacle probability distribution map after multi-source fusion processing, fusing displacement change data in the time series to construct a dynamic graph structure, and identifying target components and structural unit nodes; Identifying target components and structural unit nodes includes adopting a feature aggregation mechanism based on K-hop neighborhood relationships, taking each candidate node as the center, propagating surrounding topology and spatial features through multi-order graph convolution operations, outputting the embedding vector of each node, and classifying the nodes into beam, slab, and column structural unit types through a set classification discriminator head. Using a structural line model based on skeleton fitting, a skeleton connection graph of the building edge line and key nodes is established. An adaptive curvature enhancement strategy is used to correct the node sparse area, and the graph regularization algorithm is used to enhance the continuity and integrity of the overall structure recognition, and the structure graph is output; Establishing a skeleton connection graph of the building edge line and key nodes includes selecting key points with high boundary response intensity and significant curvature change in the structural feature map as skeleton fitting nodes, generating a preliminary skeleton graph through an initial connection method based on the spatial minimum spanning tree, and combining the direction gradient and normal vector consistency between nodes to construct an optimized skeleton connection relationship.

5. The method for analyzing the engineering progress based on the BIM model and AI video as claimed in any one of claims 1, 2 or 4, characterized in that: The defect anomaly annotation includes, According to the optimized skeleton connection relationship, extract the visual feature distribution and thermal anomaly layer in the potential anomaly area, and perform matching calculations by comparing with the known defect templates in the anomaly knowledge base to obtain the defect anomaly annotation result; The known defect templates include crack propagation patterns, equipment detachment forms, and thermal concentration anomalies; The matching includes implementing the matching calculation using an anomaly matching algorithm based on the joint judgment of cosine distance and spatial clustering.

6. The method for engineering progress based on BIM model and AI video analysis according to claim 5, wherein: The triggering of the engineering response mechanism based on the risk event scoring function includes, Construct a risk event vector sequence according to the defect anomaly annotation result, use the continuous anomaly data window during the inspection process as the input, utilize the LSTM-Attention double-layer network structure to predict the risk evolution trend in the future time period, and conduct a risk quantitative assessment of the current state to produce an analysis report on the actual engineering progress.

7. The method for engineering progress based on BIM model and AI video analysis according to any one of claims 1, 2, 4 or 6, characterized in that: The remote dynamic adjustment of the construction path includes, According to the current building structure map, defect location coordinates, wind field, and occlusion, recalculate the optimal construction path; Recalculating the optimal construction path includes using a multi-objective improved algorithm based on heuristic guidance.

8. The engineering progress system based on BIM model and AI video analysis is characterized in that: Including a data acquisition module, a defect annotation module, and an engineering adjustment module; The data acquisition module is used to collect regional image data and perform semantic segmentation and environmental modeling; The defect annotation module is used to construct a multi-scale spatial feature network and a dynamic skeleton modeling engine, input the three-dimensional environmental map, the structural recognizable area map, and the obstacle probability distribution map into the multi-scale spatial feature network and the dynamic skeleton modeling engine, and output the execution of structure recognition and defect anomaly annotation; The engineering adjustment module is used to trigger the engineering response mechanism based on the risk event scoring function and remotely dynamically adjust the construction path.

9. A computer device, comprising a memory and a processor, the memory storing a computer program, characterized in that, When the processor executes the computer program, it realizes the steps of the method for analyzing the engineering progress based on the BIM model and AI video analysis according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it realizes the steps of the method for analyzing the engineering progress based on the BIM model and AI video analysis according to any one of claims 1 to 7.

Images