Fabricated building construction management system and method based on BIM and GNN model

Abstract

The embodiment of the invention discloses an assembly type building construction management system based on BIM and GNN models, the system comprises a data collection module, a data processing module, a data application module and a data visualization module, the data collection module is used for collecting multi-source heterogeneous data of the whole period of building construction, the multi-source heterogeneous data comprises a building information model, a data processing module, a data application module and a data visualization module, and the data processing module is used for processing the multi-source heterogeneous data in the whole period of building construction. BIM (Building Information Modeling) data, sensor data, video stream data and external data; the data processing module is used for processing the multi-source heterogeneous data to obtain a data processing result which comprises a fused data set, standardized graph structure data and structured decision suggestions; the data application module is used for adjusting resource allocation according to the data processing result and generating an optimization instruction which comprises a structured instruction set and a structured scheme set; and the data visualization module is used for converting the optimization instruction and the structured decision suggestion and displaying a conversion result in a visual interface. According to the invention, the integration level, the intelligence and the cost-benefit balance degree of fabricated building construction management can be improved.

Description

Technical Field

[0001] The present application relates to the field of building informationization and intelligent technology, and is related to but not limited to an assembly building construction management system and method based on BIM and GNN models. Background Art

[0002] Driven by the dual trends of global building industrialization and digital transformation, prefabricated buildings, with their advantages of high efficiency, environmental protection, and controlled quality, have become a core direction for the transformation and upgrading of the construction industry. However, with the rapid development of the industry, multiple challenges exist in the construction management of prefabricated buildings. First, traditional management models rely on manual experience and static plans, making it difficult to cope with the dynamic coordination requirements of prefabricated component production, logistics and transportation, and on-site assembly. Second, factors such as design changes, weather disruptions, and equipment failures can easily lead to construction delays. Quality control methods such as manual inspections and post-inspections have a certain lag, which can easily lead to structural safety risks. Furthermore, problems such as overstocked or insufficient prefabricated component inventory and conflicting construction routes further exacerbate resource waste and cost pressures.

[0003] Existing large-scale prefabricated building construction management systems primarily include 3D modeling, full-lifecycle data integration, and refined construction management. 3D modeling utilizes a 3D information model of prefabricated components to achieve visual simulation. In the full-lifecycle data integration phase, BIM serves as a data sharing platform, covering the entire design, production, construction, and operation and maintenance phases, supporting construction simulation, schedule management, quality control, and cost management. Refined construction management utilizes the association of component attribute parameters (such as size, material, and location) to achieve refined control of the construction process and reduce resource waste. In practical applications, existing prefabricated building construction management systems suffer from deficiencies such as insufficient inter-system coordination due to low technical integration, insufficient decision support due to limited intelligence, insufficient visualization and interactivity, and difficulty balancing costs and benefits.

[0004] Therefore, there is an urgent need for a more complete prefabricated building construction management system to make up for the shortcomings of the existing system in terms of integration, intelligence, visualization, interactivity and cost-effectiveness, so as to provide the prefabricated building industry with a more optimized, replicable and scalable intelligent solution. Summary of the Invention

[0005] The embodiments of the present application provide a prefabricated building construction management system and method based on BIM and GNN models.

[0006] The technical solution of the embodiment of the present application is implemented as follows: In a first aspect, an embodiment of the present application provides a prefabricated building construction management system based on BIM and GNN models, the system comprising a data acquisition module, a data processing module, a data application module, and a data visualization module, wherein: A data acquisition module is used to collect multi-source heterogeneous data throughout the entire construction cycle, wherein the multi-source heterogeneous data includes BIM data, sensor data, video stream data, and external data. A data processing module is used to process the multi-source heterogeneous data to obtain data processing results, wherein the data processing results include fused data sets, standardized graph structure data, and structured decision suggestions, wherein the structured decision suggestions include construction progress prediction results, decision suggestions, and resource optimization suggestions. A data application module is used to adjust resource allocation according to the data processing results and generate optimization instructions, wherein the optimization instructions include a structured instruction set and a structured solution set. A data visualization module is used to convert the optimization instructions and the structured decision suggestions, and display the conversion results in a visualization interface, wherein the conversion results include a construction progress overview, a construction progress Gantt chart, a resource distribution heat map, and a resource allocation solution.

[0007] The technical solution provided by this application collects multi-source heterogeneous data of the entire construction cycle through a data acquisition module. The multi-source heterogeneous data includes BIM data, sensor data, video stream data and external data, and realizes real-time and comprehensive monitoring of detailed information of the entire construction process, thereby completing accurate simulation of the construction plan and automatically triggering emergency plans when dangerous situations are detected, avoiding potential safety accidents and greatly improving the level of safety management in construction; the multi-source heterogeneous data is processed by a data processing module to obtain data processing results, which include fused data sets, standardized graph structure data and structured decision suggestions. The structured decision suggestions include construction progress prediction results, decision suggestions and resource optimization suggestions, thereby achieving accurate prediction of construction progress and risks, and based on Dynamically adjust the construction plan based on predictions to reduce delays and ensure the project is completed on time; adjust resource allocation based on data processing results in the data application module and generate optimization instructions. The optimization instructions include structured instruction sets and structured solution sets, ultimately improving resource utilization efficiency and significantly reducing construction costs; convert optimization instructions and structured decision suggestions through the data visualization module, and display the conversion results in a visualization interface. The conversion results include a construction progress overview, a construction progress Gantt chart, a resource distribution heat map, and a resource allocation plan. The complex construction data is presented to decision makers in an intuitive and easy-to-understand manner through a visualization interface, and multiple views such as global overview and dynamic scheduling are supported to significantly improve the decision-making efficiency and quality of decision makers, providing a strong guarantee for the successful implementation of the project.

[0008] Optionally, the data processing module includes a data fusion unit, a feature extraction unit and a graph neural network (GNN) reasoning unit, wherein: the data fusion unit is used to perform spatiotemporal alignment processing on the sensor data, the video stream data and the external data with the BIM data respectively through a spatiotemporal interpolation algorithm to obtain spatiotemporal aligned multimodal data, store the associated data in the multimodal data based on a graph database, construct a construction knowledge graph, and fuse the multimodal data and the construction knowledge graph to obtain the fused data set; the feature extraction unit is used to refine the structured features of the fused data set to obtain the standardized graph structure data, and the standardized graph structure data includes a node feature matrix, an edge feature matrix and graph structure metadata; the GNN reasoning unit is used to model the standardized graph structure data according to GNN, complete the prediction and reasoning task, and obtain the structured decision recommendation.

[0009] Optionally, the construction knowledge graph includes BIM entities, BIM relationships and BIM attributes, wherein: the BIM entities include personnel entities, equipment entities, material entities, component entities, process entities and safety specification entities; the BIM relationships are the relationships between the BIM entities; and the BIM attributes are dynamic attributes added between the BIM entities and the BIM relationships.

[0010] Optionally, the extracting of the structured features of the fused data set to obtain the standardized graph structure data includes: extracting the feature vectors of the BIM entities to form the node feature matrix; converting the BIM relationships into edge feature vectors to form the edge feature matrix; injecting the external data into the construction knowledge graph as a global context feature vector, and extracting the temporal feature vectors in the construction knowledge graph through a long short-term memory network (LSTM) to form the graph structure metadata.

[0011] Optionally, the GNN reasoning unit includes a message passing subunit, a feature update subunit, a graph attention subunit and a decision generation subunit, wherein: the message passing subunit is used to aggregate feature information of adjacent nodes in the standardized graph structure data; the feature update subunit is used to dynamically adjust the node priority in the standardized graph structure data in combination with the external data through a gated recurrent unit (GRU); the graph attention subunit is used to assign weights to key process nodes in the standardized graph structure data through a self-attention mechanism; and the decision generation subunit is used to generate the structured decision recommendation based on the updated standardized graph structure data.

[0012] Optionally, the data application module includes a dynamic scheduling unit and a resource optimization unit, wherein: the dynamic scheduling unit is used to complete the scheduling operation based on the fused data set, the construction progress prediction results and the decision suggestions, combined with the genetic algorithm, to obtain the structured instruction set, and the structured instruction set includes process adjustment instructions and resource allocation instructions; the resource optimization unit is used to complete the resource optimization operation based on the fused data set, the construction progress prediction results and the resource optimization suggestions, combined with the linear programming algorithm, and output the Pareto optimal solution to obtain the structured solution set, and the structured solution set includes a process adjustment solution and a resource allocation solution.

[0013] Optionally, the visualization interface supports the superimposed display of three-dimensional BIM and real-time three-dimensional BIM data, and the conversion of the optimization instructions and the structured decision suggestions includes: aggregating actual construction progress data and resource status data to generate key performance indicators, wherein the key performance indicators include process delay rate and equipment utilization rate; generating a construction progress Gantt chart based on the comparison results of the actual construction progress data and the construction progress forecast results; generating a resource distribution heat map based on the resource allocation plan; and updating the multi-source heterogeneous data, the construction progress overview, the key performance indicators, the construction progress Gantt chart and the resource distribution heat map in real time through a web page long connection protocol.

[0014] Optionally, the data visualization module includes a global overview visualization unit and a dynamic scheduling visualization unit, wherein; the global overview visualization unit is used to display the construction progress overview and the resource distribution heat map; the dynamic scheduling visualization unit is used to display the construction progress Gantt chart and the resource allocation plan.

[0015] In the second aspect, an embodiment of the present application provides a method for prefabricated building construction management based on BIM and GNN models, which is applied to a prefabricated building construction management system based on BIM and GNN models. The system includes a data acquisition module, a data processing module, a data application module and a data visualization module. The method includes: collecting multi-source heterogeneous data of the entire construction cycle, the multi-source heterogeneous data including BIM data, sensor data, video stream data and external data; processing the multi-source heterogeneous data to obtain data processing results, the data processing results including fused data sets, standardized graph structure data and structured decision suggestions, the structured decision suggestions including construction progress prediction results, decision suggestions and resource optimization suggestions; adjusting resource allocation according to the data processing results, generating optimization instructions, the optimization instructions including a structured instruction set and a structured scheme set; converting the optimization instructions and the structured decision suggestions, and displaying the conversion results in a visualization interface, the conversion results including a construction progress overview, a construction progress Gantt chart, a resource distribution heat map and a resource allocation scheme.

[0016] The beneficial effects of the technical solutions provided in the embodiments of the present application include at least: This application provides an assembly-type building construction management system and method based on BIM and GNN models. The data acquisition module collects multi-source heterogeneous data of the entire construction cycle. The multi-source heterogeneous data includes BIM data, sensor data, video stream data and external data, so as to realize real-time and comprehensive monitoring of detailed information of the entire construction process, thereby completing the accurate simulation of the construction plan and automatically triggering the emergency plan when a dangerous situation is detected, thus avoiding potential safety accidents and greatly improving the safety management level in construction. The data processing module processes the multi-source heterogeneous data to obtain the data processing results, which include fused data sets, standardized graph structure data and structured decision suggestions. The structured decision suggestions include construction progress prediction results, decision suggestions and resource optimization suggestions, thereby realizing the construction progress. Accurately predict the degree and risk of construction delays, and dynamically adjust the construction plan based on the prediction to reduce delays and ensure the project is completed on time; in the data application module, adjust resource allocation according to the data processing results and generate optimization instructions. The optimization instructions include structured instruction sets and structured solution sets, ultimately improving resource utilization efficiency and significantly reducing construction costs; through the data visualization module, optimize instructions and structured decision suggestions are converted and displayed in the visualization interface. The conversion results include construction progress overview, construction progress Gantt chart, resource distribution heat map and resource allocation plan, presenting complex construction data to decision makers in an intuitive and easy-to-understand way through the visualization interface, and supporting multiple views such as global overview and dynamic scheduling, so as to greatly improve the decision-making efficiency and quality of decision makers, providing a strong guarantee for the successful implementation of the project. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present application. Those skilled in the art can also derive other drawings based on these drawings without inventive work, among which: Figure 1 A schematic diagram of an assembled building construction management system based on BIM and GNN models provided in an embodiment of the present application; Figure 2 A schematic diagram of the refinement process of an assembled building construction management system based on BIM and GNN models provided in an embodiment of the present application; Figure 3 A flowchart of a prefabricated building construction management method based on BIM and GNN models is provided in an embodiment of the present application. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all of the embodiments. The following examples are used to illustrate the present application, but are not intended to limit the scope of the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0019] In the following description, reference is made to “some embodiments”, which describes a subset of all possible embodiments, but it will be understood that “some embodiments” may be the same subset or different subsets of all possible embodiments and may be combined with each other without conflict.

[0020] It should be pointed out that the terms "first\second\third" involved in the embodiments of the present application are merely used to distinguish similar objects and do not represent a specific ordering of the objects. It can be understood that "first\second\third" can be interchanged with a specific order or sequence where permitted, so that the embodiments of the present application described here can be implemented in an order other than that illustrated or described here.

[0021] Those skilled in the art will understand that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as generally understood by those skilled in the art in the art to which the embodiments of the present application belong. It should also be understood that terms such as those defined in common dictionaries should be understood to have meanings consistent with their meanings in the context of the prior art, and will not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0022] The embodiments of the present application will be further described below with reference to the accompanying drawings.

[0023] In view of the current problems in the management of prefabricated building construction in the field of building informatization and intelligent technology, the embodiment of the present application provides a prefabricated building construction management system and method based on BIM and GNN models.

[0024] The technical solution of the present application is introduced below, and first the system embodiment of the present application is introduced.

[0025] Please refer to Figure 1 , which shows a schematic diagram of an assembled building construction management system based on BIM and GNN models provided by an embodiment of the present application, such as Figure 1As shown, the system includes a data acquisition module 01, a data processing module 02, a data application module 03 and a data visualization module 04. The data acquisition module 01, the data processing module 02, the data application module 03 and the data visualization module 04 are connected in sequence.

[0026] The data acquisition module 01 is used to collect multi-source heterogeneous data throughout the entire construction cycle, and the multi-source heterogeneous data includes BIM data, sensor data, video stream data and external data; the data processing module 02 is used to process the multi-source heterogeneous data to obtain data processing results, and the data processing results include fused data sets, standardized graph structure data and structured decision suggestions, and the structured decision suggestions include construction progress prediction results, decision suggestions and resource optimization suggestions; the data application module 03 is used to adjust resource allocation according to the data processing results and generate optimization instructions, and the optimization instructions include structured instruction sets and structured solution sets; the data visualization module 04 is used to convert the optimization instructions and the structured decision suggestions, and display the conversion results in a visualization interface, and the conversion results include a construction progress overview, a construction progress Gantt chart, a resource distribution heat map and a resource allocation plan.

[0027] In the embodiment of the present application, the data acquisition module 01 is used to collect multi-source heterogeneous data throughout the entire construction cycle. The multi-source heterogeneous data includes BIM data, sensor data, video stream data, and external data. Specifically, BIM data is a core component of multi-source heterogeneous data. BIM data includes data on the entire life cycle of prefabricated buildings from initial design to operation and maintenance after commissioning. These data not only cover static data such as component geometry (such as size, shape, spatial position information, etc.), material properties (such as strength grade, manufacturer, etc.), and process requirements (such as connection method, installation sequence, etc.), but also dynamic data such as construction progress, quality acceptance results, and equipment maintenance records. In addition, BIM data supports collaborative work among multiple participants, such as design institutes, general contractors, subcontractors, and suppliers working together on the same platform. Throughout the entire construction cycle, BIM data supports real-time dynamic updates. Any design changes, construction schedule adjustments, or quality acceptance results will be recorded and updated in real time in the BIM library. In addition, BIM data also includes different data such as Revit, Tekla, and Archicad. The Industry Foundation Classes (IFC) standard can be used to achieve interaction between different data, thereby ensuring data consistency in all aspects of design, construction, operation, and maintenance.

[0028] In an embodiment of the present application, sensor data is real-time perception data of multi-source heterogeneous data. Various sensors deployed at the construction site (such as temperature and humidity sensors, stress sensors, displacement sensors, and personnel positioning sensors, etc.) collect in real time environmental parameters (such as temperature and humidity, wind speed, and noise) that have an important impact on construction links such as concrete curing and worker safety, structural health monitoring data (such as stress, displacement, and vibration), etc., which are used to monitor the health status of components and structures and thus prevent the occurrence of safety accidents, equipment status information data (such as tower crane load, transport vehicle location, welding robot working status), etc., which support real-time monitoring and scheduling of equipment, thereby improving equipment utilization, and data such as personnel location information data. Personnel location information data can be tracked in real time through ultra-wideband technology, and combined with personnel location information data and electronic fences to conduct safety management and control of construction personnel to ensure the safety of construction personnel. Specifically, ultra-wideband technology uses high-precision positioning capabilities to track workers' locations in real time, and combined with geofencing for intelligent safety management and control. Ultra-wideband technology utilizes nanosecond-level narrow pulse signals to achieve positioning accuracy of 10 to 30 centimeters. This allows for rapid deployment of positioning base stations without the need for site mapping to adapt to the dynamic construction environment. UWB tags are embedded in workers' hard hats or badges, enabling real-time upload of location data to the BIM platform. The geofence feature uses BIM to pre-set virtual safety zones, supporting arbitrary polygonal areas (minimum 3 square meters, maximum 150×150 square meters). It also integrates whitelist and blacklist management, as well as timeout alarms. When UWB positioning data triggers geofence boundary conditions, such as intrusion into a hazardous area or unauthorized entry into a confidential area, the system immediately triggers an audible and visual alarm and notifies management via mobile devices. The system also integrates with the BIM platform's visual interface to locate personnel locations, facilitating rapid decision-making. For example, geofencing can be set up in a steel structure hoisting area. If a worker strays into the area, the system automatically freezes operations and triggers an emergency response plan. The technical solution provided in the embodiment of the present application collects various types of data at the construction site in real time through various sensors, which can realize real-time and comprehensive monitoring of the entire construction process, track personnel location information data in real time through ultra-wideband technology, and conduct safety management and control of construction personnel through electronic fences. When a dangerous situation is detected, the system automatically triggers the emergency plan, thereby ensuring the safety of construction personnel to the greatest extent possible.

[0029] In the embodiments of the present application, video stream data is visual memory data from multi-source heterogeneous data. High-definition video data of the construction site is collected by fixed cameras or drones. Fixed cameras are deployed in key areas such as tower cranes, entrances and exits, and material yards to monitor the construction panorama 24 hours a day. For example, a project deployed 4K high-definition cameras in 10 key areas to achieve 360-degree coverage without blind spots. The images in key areas are refreshed at 25 frames per second to ensure action continuity. For another example, a super-high-rise project uses drones to generate panoramic images once a week with sub-centimeter accuracy, which can clearly identify construction deviations of 0.1 meters.

[0030] In the embodiments of the present application, external data is a key supplementary information source to support construction decisions. By acquiring parameters such as temperature, humidity, wind speed, and rainfall probability in real time as external data, a basis for dynamic adjustment of the construction plan can be provided. Specifically, the external data integrates GIS data, which can be used to analyze the impact of topography and geological conditions on construction; the external data is linked to the market conditions of building materials and can be used to dynamically adjust procurement strategies; the external data also includes real-time tracking information of prefabricated components (such as the location of transport vehicles and estimated arrival time). Combining real-time tracking information with traffic congestion data can optimize the route of prefabricated components entering the site; the external data also includes obtained policy documents to ensure the legality and compliance of the entire construction process.

[0031] In the embodiment of the present application, the data processing module 02 includes a data fusion unit, a feature extraction unit and a GNN reasoning unit, wherein the data fusion unit is the core component of the data processing module 02, which is used to complete the key tasks of integrating multi-source heterogeneous data and building a construction knowledge graph. The four-dimensional data including BIM data, sensor data, video stream data and external data are input into the data fusion unit. The data fusion unit uses a spatiotemporal interpolation algorithm to perform spatiotemporal alignment processing on the sensor data, video stream data and external data with the BIM data to obtain spatiotemporal aligned multimodal data. Specifically, first, the BIM data is converted into a Geographic Information System (GIS) data according to the data conversion platform (Feature Manipulation Engine, FME). Furthermore, the sensor data, video stream data, and external data are aligned to the BIM time axis and spatial coordinate system through a spatiotemporal interpolation algorithm (or inverse distance weighted method), thereby completing the spatiotemporal alignment with the BIM data and obtaining spatiotemporally aligned multimodal data. In addition, since different data may have semantic differences in describing the same entity or relationship when fusing multi-source heterogeneous data, in addition to spatiotemporal alignment, semantic alignment can also be performed on the multi-source heterogeneous data to further achieve semantic alignment. For example, methods such as ontology matching and semantic similarity calculation can be used to ensure data consistency at the semantic level. Furthermore, semantic tags can be added to the video stream data (for example, "Camera: Wall panel lifting, time 14:00") to facilitate subsequent retrieval and analysis. Finally, a graph database is used to store the associated data within the multimodal data and construct a construction knowledge graph. For example, the wall panels, tower cranes, and construction worker Zhang Gong can be associated, and their associations recorded in the construction knowledge graph. Furthermore, we integrate spatiotemporally aligned multimodal data (e.g., “BIM attributes of a component at a certain moment, sensor readings, video footage, and external weather”) with construction knowledge graphs (e.g., “Wall panel A depends on floor panel B, and the current installation progress lags behind by 2 days”) to obtain a fused dataset.

[0032] In an optional embodiment, the construction knowledge graph includes BIM entities, BIM relationships and BIM attributes, wherein BIM entities include six major categories of entities: personnel entities, equipment entities, material entities, component entities, process entities and safety specification entities. For example, personnel entities include attributes such as name, job type, qualification certificate, etc., and equipment entities record information such as model, location coordinates, maintenance records, etc.; BIM relationships are relationships between BIM entities, including more than ten types of relationships such as operation relationships, consumption relationships, compliance relationships and spatial relationships. For example, the "personnel-operation-equipment" relationship records the association between operators and equipment, and the "component-installation-location" relationship reflects the specific coordinates of components at the construction site. The timing relationship and causal relationship (such as equipment failure causing process delays) characterize the dynamic changes in the construction process; BIM attributes are dynamic attributes added between BIM entities and BIM relationships, such as the real-time status of equipment (running / stopped) and the probability of process delay. These attributes are updated in real time through sensor data and video stream data to ensure that the construction knowledge graph reflects the latest status of the construction site in real time. The technical solution provided in the embodiments of this application constructs a construction knowledge graph using three elements: BIM entities, BIM relationships, and BIM attributes. This allows for refined modeling of important elements and their relationships throughout the construction process. The construction knowledge graph provides structured input for the GNN model, supporting deep learning tasks such as construction progress prediction and resource optimization. Ultimately, the construction knowledge graph is presented in three dimensions within a visual interface, helping managers achieve dynamic scheduling and risk management.

[0033] Furthermore, the feature extraction unit is used to refine the structured features of the fused dataset to obtain standardized graph structure data. The standardized graph structure data includes a node feature matrix, an edge feature matrix, and graph structure metadata. Specifically, the main steps of the structured feature extraction process are as follows: First, the feature vector of each BIM entity (e.g., component, equipment) is extracted to form a node feature matrix. The node feature matrix includes static attribute features (e.g., wall panel size, tower crane model) and dynamic state features (e.g., installation progress, load utilization), with each row of the node feature matrix corresponding to the feature vector of a BIM entity. Second, BIM relationships (e.g., support, adjacency, data association) are converted into edge feature vectors to form an edge feature matrix. The edge feature matrix includes relationship type, strength, and timestamp, with each row corresponding to the relationship feature of an edge. Finally, external data (e.g., weather, policy) is injected into the construction knowledge graph as a global context feature vector. The time series feature vectors in the construction knowledge graph (e.g., the 3-day moving average of the installation progress of a component) are extracted through LSTM to form graph structure metadata. The graph structure metadata includes the global attributes of the construction knowledge graph and the node type mapping table. Standardized graph structure data is constructed based on the node feature matrix, edge feature matrix, and graph structure metadata. In the technical solution provided in the embodiment of this application, a feature extraction unit is used to construct a feature system that can perceive construction management scenarios in real time and dynamically evolve. This can provide the GNN reasoning unit with structured input that combines domain knowledge and data-driven input, thereby supporting accurate construction decision-making.

[0034] The following uses a specific example to illustrate the specific process of extracting time series features through LSTM. Taking the three-day moving average of a component installation progress as an example, the time series feature extraction process is as follows.

[0035] First, the raw time series data is preprocessed, arranging the component installation progress data in chronological order to form a time series. For example, the completion percentage of each component installation is recorded each day, forming a sequence such as [Day 1: 10%, Day 2: 25%, Day 3: 40%, ...]. Furthermore, the data is normalized to eliminate dimensionality effects, for example, by mapping the completion percentage to the interval [0, 1]. Next, an LSTM model is constructed. The LSTM, through its unique gating mechanism, can capture long-term dependencies in time series. The network input is the preprocessed time series data, with each time step corresponding to an installation progress value. Internally, the network uses structures such as forget gates, input gates, and output gates to selectively retain or forget historical information, gradually extracting time series features. As the LSTM network processes data, it generates a series of hidden states. These hidden states embody the network's understanding of the time series data and can be considered an encoding of the time series features. To extract the three-day moving average feature, the LSTM output is combined with the raw data. Specifically, a sliding window technique is used to calculate the average of the data within each window on the raw time series. At the same time, the hidden state of the LSTM network at each time step is associated with the corresponding moving average, thereby combining the time series features extracted by LSTM with the moving average features; finally, by training LSTM, it can accurately capture the time series laws of component installation progress and output relevant features including the 3-day moving average. These features can be used for subsequent tasks such as construction progress prediction, providing strong support for prefabricated building construction management.

[0036] During feature extraction, time-series features are dynamically updated to reflect real-time changes in construction. This process includes real-time data stream access, sliding window maintenance, LSTM incremental learning, feature fusion and push, and an exception feedback loop. The real-time data stream access step involves continuously receiving time-series data such as component installation progress and sensor status through a message queue and preprocessing (cleaning and standardizing) the time-series data. The sliding window maintenance step involves using a three-day window. Whenever new data arrives, the oldest data point is removed, the latest data is added, and the moving average is recalculated (e.g., using a double-ended queue for efficient implementation). The LSTM incremental learning step utilizes an online LSTM model, updating only the hidden and cell states rather than global retraining, to rapidly extract updated time-series patterns (e.g., progress trends). The feature fusion and push step combines dynamic time-series features (e.g., the three-day moving average installation rate) with static features (e.g., component attributes) and pushes them to the GNN module in real time via a publish-subscribe model. The exception feedback loop step monitors fluctuations in time-series features. If an anomaly is detected (e.g., progress lag exceeding a threshold), the scheduling module is triggered to adjust the plan (e.g., dispatching additional resources), forming a real-time response closed loop. The dynamic update process of time series features can ensure that the time series features evolve synchronously with the construction site, thereby supporting the GNN model to make timely decisions.

[0037] The process of selecting key features and assigning weights is as follows: first, complete the feature classification oriented by the construction goal. According to the construction management goals (progress, quality, safety, cost), give priority to the key features that directly affect the goals to complete the node feature screening, such as progress management (process timing dependency, component installation sequence, equipment availability), safety management (personnel qualification matching, equipment status abnormality index, high-altitude operation compliance), focus on the features that reflect the construction logic association to complete the edge feature screening, such as the resource competition relationship between processes, the spatial adjacency of component transportation paths, and the compliance link between safety regulations and operations; secondly, through expert experience injection (i.e., through domain knowledge graph prediction), the key features that directly affect the construction goal are selected. Static weight initialization is completed by setting basic weights, such as setting the impact of "equipment failure rate" on progress to 0.7) and historical data calibration (using historical project data to quantify feature contributions through Lasso regression, such as increasing the weight of "weather factors" by 30% during rainy season construction). Finally, dynamic weight adjustment is completed through context perception (dynamically adjusting weights according to the construction stage, such as increasing the weight of "component connection quality" from 0.5 to 0.8 during the main structure construction period) and real-time feedback optimization (through a reinforcement learning framework, using indicators such as scheduling efficiency and resource utilization as feedback, iteratively optimizing weight distribution. For example, after a sudden failure of equipment A, the weight of its "maintenance record" feature is immediately adjusted upward).

[0038] Furthermore, the GNN inference unit is used to model the standardized graph structure data based on GNN, complete the predictive reasoning task, and obtain structured decision recommendations. Specifically, the GNN inference unit includes a message passing subunit, a feature update subunit, a graph attention subunit, and a decision generation subunit. The message passing subunit is used to aggregate the feature information of adjacent nodes in the standardized graph structure data; the feature update subunit is used to dynamically adjust the node priority in the standardized graph structure data by combining external data through GRU; the graph attention subunit is used to assign weights to key process nodes in the standardized graph structure data through the self-attention mechanism; and the decision generation subunit is used to generate structured decision recommendations based on the updated standardized graph structure data.

[0039] In a specific embodiment, the GNN model is first trained in the GNN inference unit, and the training process needs to be deeply integrated with the graph reasoning mechanism. The specific steps are as follows: first, input standardized graph structure data (such as converting construction processes into nodes and encoding dependencies as edges); second, define the GNN layer structure, such as selecting GraphSAGE or GAT, to aggregate neighbor information to update the node representation to simulate the transmission of construction information between processes; then, design a loss function, such as cross entropy loss for classification tasks (such as process type identification), or mean square error for regression tasks (such as progress prediction), and at the same time, select the optimizer to configure hyperparameters such as learning rate. In the training cycle, the forward propagation stage generates node embeddings through the GNN layer and calculates the loss; the backpropagation stage updates the model parameters according to the loss and adjusts the weights according to the optimizer to minimize the loss. In addition, a batch training strategy can be adopted, such as sub-graph sampling of large-scale graphs, and only loading part of the graph data in each iteration to improve training efficiency; finally, the model performance is monitored through the validation set, such as accuracy, loss value and other indicators to prevent overfitting. When the model performs stably on the validation set, the final performance can be evaluated on the test set, and the trained GNN model can be integrated into the GNN reasoning module to complete predictive reasoning tasks, such as dynamically adjusting the construction plan. In the GNN inference module, the performance of the trained GNN model needs to be verified through multi-dimensional evaluation indicators, and targeted optimization methods are used to improve the effect, as follows: the evaluation indicators include three dimensions. The first dimension is for classification tasks, and the accuracy of the model prediction is evaluated using indicators such as accuracy, precision, recall, and F1 value, reflecting the performance of the model in various categories of prediction tasks in construction scenarios, such as component type classification and process stage identification; the second dimension is for regression tasks, and the mean square error (MSE) or mean absolute error (MAE) is used to measure the deviation between the model prediction value and the true value. In construction management, it can be used to evaluate the model's prediction accuracy for continuous values ​​such as schedule delay time and resource consumption; the third dimension is specific indicators for construction scenarios, including schedule deviation rate (the proportion of the difference between the predicted construction period and the actual construction period), resource utilization fluctuation coefficient, and safety accident prediction accuracy, which are directly related to the core goals of construction management and can fully reflect the value of the model in practical applications. The optimization method is hyperparameter tuning. By adjusting key parameters such as the learning rate, number of graph attention heads, and dropout ratio through grid search or Bayesian optimization, the model's prediction accuracy and generalization ability can be improved. Early stopping is used to prevent overfitting and help stop training in time when the model performance reaches saturation.

[0040] Furthermore, the feature information of adjacent nodes in the standardized graph structure data is aggregated through the message passing sub-unit, such as the installation progress of adjacent floor slab nodes aggregated by the wall panel node; in the feature update sub-unit, the node priority in the standardized graph structure data is dynamically adjusted by combining GRU with external data, such as adjusting the component installation priority according to weather data; in the graph attention sub-unit, weights are assigned to key process nodes in the standardized graph structure data through the self-attention mechanism, such as assigning higher weights to process nodes that are about to expire; in the decision generation sub-unit, structured decision suggestions are generated based on the updated standardized graph structure data, and finally, structured decision suggestions including construction progress prediction results, decision suggestions and resource optimization suggestions are generated. In the technical solution provided by the embodiment of the present application, the GNN reasoning module uses GNN to model the standardized graph structure data of the construction process, completes the prediction and reasoning task of the construction progress, outputs more detailed decision suggestions for the dynamic scheduling unit, and prepares for the resource optimization unit to output more detailed resource optimization suggestions.

[0041] In an embodiment of the present application, the data application module 03 includes a dynamic scheduling unit and a resource optimization unit, wherein the dynamic scheduling unit is used to complete the scheduling operation based on the fused data set, the construction progress prediction results and the decision suggestions, combined with the genetic algorithm, to obtain a structured instruction set, and the structured instruction set includes process adjustment instructions and resource allocation instructions; the resource optimization unit is used to complete the resource optimization operation based on the fused data set, the construction progress prediction results and the resource optimization suggestions, combined with the linear programming algorithm, and output the Pareto optimal solution to obtain a structured solution set, and the structured solution set includes a process adjustment solution and a resource allocation solution.

[0042] In a specific embodiment, the dynamic scheduling unit's input is primarily the construction progress predictions and decision recommendations output by the GNN inference module, integrated with real-time construction status data (e.g., progress tracking and resource status) from the data fusion module. The scheduling process begins by detecting scheduling conflicts based on process dependencies within the BIM model and visualizing them using a Gantt chart. If resource conflicts are detected, pre-set rules are applied to generate an initial scheduling solution. An optimized solution is then generated using a genetic algorithm, combined with the GNN recommendations. Finally, the optimized solution is broken down into specific instructions, which are prioritized and output as a structured instruction set for construction teams and equipment. These instructions include process adjustment instructions (e.g., adjusted process start / end times) and resource allocation instructions (e.g., equipment scheduling solutions and personnel deployment plans).

[0043] In a specific embodiment, the resource optimization unit's input primarily consists of the construction progress predictions and resource optimization recommendations output by the GNN inference module, integrated with real-time construction status data (progress tracking, resource status) from the data fusion module. The resource optimization module transforms this input data into an executable optimization plan through the following three layers of logic: First, resource distribution is visualized using Gantt charts and heat maps (e.g., "crane T1 high load" areas are marked in red) to identify inefficient resources. Second, future resource demand is predicted based on the GNN-generated construction progress predictions. Finally, an initial optimization plan is generated by applying pre-set rules (e.g., automatic redeployment of equipment when idle for more than one hour, triggering procurement of material inventory below a threshold). Linear programming, combined with the GNN recommendations, produces a Pareto-optimal solution. The final output is a structured set of solutions for construction teams and equipment, including equipment scheduling plans, equipment maintenance plans, purchase order recommendations, and personnel deployment plans. The technical solution provided in the embodiment of the present application uses a resource optimization unit to make trade-offs and decisions among multiple objectives to generate a resource allocation plan that is both in line with project requirements and feasible. This not only helps to improve the overall benefits and efficiency of the construction project, but also provides strong support for decision makers, helping them make more wise and reasonable decisions.

[0044] In an embodiment of the present application, the data visualization module 04 converts optimization instructions and structured decision suggestions to generate conversion results, which include a construction progress overview, a construction progress Gantt chart, a resource distribution heat map, and a resource allocation plan. The conversion process is as follows: actual construction progress data and resource status data are aggregated by time (day, week, month) and space (region, floor) to generate key performance indicators (KPIs), including process delay rate and equipment utilization rate; a construction progress Gantt chart is generated based on the comparison results of actual construction progress data and construction progress forecast results; a resource distribution heat map is generated based on the resource allocation plan; and multi-source heterogeneous data, the construction progress overview, KPIs, the construction progress Gantt chart, and the resource distribution heat map are updated in real time via a web page persistent connection protocol. The conversion results are displayed in a visualization interface. Clicking on a chart element can view details (e.g., clicking on a "crane" icon to display its full-day load curve). Data can also be filtered by time, region, or resource type (e.g., only displaying construction activities in Area A on May 8). In addition, the visualization interface supports the overlay display of 3D BIM and real-time 3D BIM data.

[0045] In an embodiment of the present application, the data visualization module 04 includes a global overview visualization unit and a dynamic scheduling visualization unit, wherein the global overview visualization unit displays the construction progress overview and the resource distribution heat map; the dynamic scheduling visualization unit displays the construction progress Gantt chart and the resource allocation plan.

[0046] Please refer to Figure 2, which shows a detailed schematic diagram of an assembled building construction management system based on BIM and GNN models provided by an embodiment of the present application. The assembled building construction management system includes a data acquisition module, a data processing module, a data application module and a data visualization module. The data acquisition module includes the collected building information model data, sensor data, video stream data and external data; the data processing module includes a data fusion unit, a feature extraction unit and a graph neural network reasoning unit. In the data fusion unit, the sensor data, video stream data and external data are respectively aligned with the building information model data in time and space through a spatiotemporal interpolation algorithm to finally obtain a fused data set. In the feature extraction unit, the structured features of the fused data set are extracted to obtain standardized graph structure data. In the graph neural network reasoning unit, the standardized graph structure data is modeled according to the graph neural network to complete the prediction and reasoning task, and obtain structured decision recommendations including construction progress prediction results, decision recommendations and resource optimization recommendations; the data application module includes a dynamic scheduling unit and a resource optimization unit. In the dynamic scheduling unit, according to the fused data set, construction progress prediction results and decision recommendations, combined The genetic algorithm completes the scheduling operation and obtains a structured instruction set, which includes process adjustment instructions and resource allocation instructions. In the resource optimization unit, the resource optimization operation is completed based on the fusion data set, construction progress prediction results and resource optimization suggestions, combined with the linear programming algorithm, to obtain a structured solution set, which includes process adjustment solutions and resource allocation solutions. In the data visualization module, the optimization instructions (structured instruction set and structured solution set) and structured decision suggestions are converted and the conversion results are displayed. The conversion results include a construction progress overview, a construction progress Gantt chart, a resource distribution heat map and a resource allocation solution. The data visualization module includes a global overview visualization unit and a dynamic scheduling visualization unit. In the global overview visualization unit, the construction progress overview and the resource distribution heat map are displayed. In the dynamic scheduling visualization unit, the construction progress Gantt chart and the resource allocation solution are displayed.

[0047] In summary, the embodiment of the present application provides an assembled building construction management system based on BIM and GNN model, which collects multi-source heterogeneous data of the entire construction cycle through the data acquisition module. The multi-source heterogeneous data includes BIM data, sensor data, video stream data and external data, and realizes real-time and comprehensive monitoring of detailed information of the entire construction process, thereby completing the accurate simulation of the construction plan and automatically triggering the emergency plan when a dangerous situation is detected, avoiding potential safety accidents and greatly improving the safety management level in construction; the multi-source heterogeneous data is processed by the data processing module to obtain data processing results, which include fused data sets, standardized graph structure data and structured decision suggestions. The structured decision suggestions include construction progress prediction results, decision suggestions and resource optimization suggestions, thereby realizing Accurately predict construction progress and risks, and dynamically adjust construction plans based on the predictions to reduce delays and ensure timely completion of projects; adjust resource allocation based on data processing results in the data application module, generate optimization instructions, which include structured instruction sets and structured solution sets, ultimately improving resource utilization efficiency and significantly reducing construction costs; convert optimization instructions and structured decision suggestions through the data visualization module, and display the conversion results in a visualization interface. The conversion results include a construction progress overview, a construction progress Gantt chart, a resource distribution heat map, and a resource allocation plan. The complex construction data is presented to decision makers in an intuitive and easy-to-understand manner through a visualization interface, and multiple views such as global overview and dynamic scheduling are supported to significantly improve the decision-making efficiency and quality of decision makers, providing a strong guarantee for the successful implementation of the project.

[0048] The above is an introduction to the system embodiment of the present application. Based on the aforementioned embodiment, the method embodiment of the present application is introduced below.

[0049] Please refer to Figure 3 , which shows a flowchart of a method for prefabricated building construction management based on BIM and GNN model provided by an embodiment of the present application, which is applied to Figure 1 The system includes a data acquisition module, a data processing module, a data application module and a data visualization module, which are connected in sequence. Figure 3 As shown, the startup method includes the following steps S310 to S340.

[0050] Step S310: collecting multi-source heterogeneous data of the entire construction cycle, wherein the multi-source heterogeneous data includes building information model data, sensor data, video stream data and external data.

[0051] Step S320, processing the multi-source heterogeneous data to obtain data processing results, the data processing results including fused data sets, standardized graph structure data and structured decision suggestions, the structured decision suggestions including construction progress prediction results, decision suggestions and resource optimization suggestions.

[0052] In an embodiment of the present application, the sensor data, the video stream data and the external data are respectively subjected to spatiotemporal alignment processing with the BIM data through a spatiotemporal interpolation algorithm to obtain spatiotemporal aligned multimodal data, the associated data in the multimodal data is stored based on a graph database, a construction knowledge graph is constructed, and the multimodal data and the construction knowledge graph are fused to obtain the fused data set; the standardized graph structure data is obtained by refining the structured features of the fused data set, and the standardized graph structure data includes a node feature matrix, an edge feature matrix and graph structure metadata; the standardized graph structure data is modeled according to GNN to complete the prediction and reasoning task and obtain the structured decision recommendation.

[0053] In an embodiment of the present application, the construction knowledge graph includes BIM entities, BIM relationships and BIM attributes, wherein: the BIM entities include personnel entities, equipment entities, material entities, component entities, process entities and safety specification entities; the BIM relationships are the relationships between the BIM entities; the BIM attributes are dynamic attributes added between the BIM entities and the BIM relationships.

[0054] In an embodiment of the present application, the feature vectors of the BIM entities are extracted to form the node feature matrix; the BIM relationships are converted into edge feature vectors to form the edge feature matrix; the external data is injected into the construction knowledge graph as a global context feature vector, and the temporal feature vectors in the construction knowledge graph are extracted through LSTM to form the graph structure metadata.

[0055] In an embodiment of the present application, feature information of adjacent nodes in the standardized graph structure data is aggregated; node priorities in the standardized graph structure data are dynamically adjusted by GRU in combination with the external data; weights are assigned to key process nodes in the standardized graph structure data by a self-attention mechanism; and the structured decision recommendations are generated based on the updated standardized graph structure data.

[0056] Step S330 : adjusting resource allocation according to the data processing result and generating optimization instructions, wherein the optimization instructions include a structured instruction set and a structured solution set.

[0057] In an embodiment of the present application, based on the fused data set, the construction progress prediction results and the decision suggestions, the scheduling operation is completed in combination with a genetic algorithm to obtain the structured instruction set, which includes process adjustment instructions and resource allocation instructions; based on the fused data set, the construction progress prediction results and the resource optimization suggestions, the resource optimization operation is completed in combination with a linear programming algorithm, and the Pareto optimal solution is output to obtain the structured solution set, which includes a process adjustment solution and a resource allocation solution.

[0058] Step S340: convert the optimization instructions and the structured decision suggestions, and display the conversion results in a visual interface. The conversion results include a construction progress overview, a construction progress Gantt chart, a resource distribution heat map, and a resource allocation plan.

[0059] In an embodiment of the present application, the visualization interface supports the superimposed display of three-dimensional BIM and real-time three-dimensional BIM data, and the conversion of the optimization instructions and the structured decision suggestions includes: aggregating actual construction progress data and resource status data to generate key performance indicators, and the key performance indicators include process delay rate and equipment utilization rate; generating a construction progress Gantt chart based on the comparison results of the actual construction progress data and the construction progress forecast results; generating a resource distribution heat map based on the resource allocation plan; and updating the multi-source heterogeneous data, the construction progress overview, the key performance indicators, the construction progress Gantt chart and the resource distribution heat map in real time through a web page long connection protocol.

[0060] In an embodiment of the present application, the construction progress overview and the resource distribution heat map are displayed in a global overview visualization unit; and the construction progress Gantt chart and the resource allocation plan are displayed in a dynamic scheduling visualization unit.

[0061] In summary, the embodiment of the present application provides a method for prefabricated building construction management based on BIM and GNN model, which collects multi-source heterogeneous data of the entire construction cycle. The multi-source heterogeneous data includes BIM data, sensor data, video stream data and external data, and realizes real-time and comprehensive monitoring of detailed information of the entire construction process, thereby completing accurate simulation of the construction plan and automatically triggering emergency plans when dangerous situations are detected, avoiding potential safety accidents and greatly improving the safety management level in construction; processing multi-source heterogeneous data to obtain data processing results, which include fused data sets, standardized graph structure data and structured decision suggestions, and structured decision suggestions include construction progress prediction results, decision suggestions and resource optimization suggestions, thereby realizing Accurately predict construction progress and risks, and dynamically adjust construction plans based on predictions to reduce delays and ensure timely project completion; adjust resource allocation based on data processing results and generate optimization instructions. The optimization instructions include structured instruction sets and structured solution sets, ultimately improving resource utilization efficiency and significantly reducing construction costs; convert optimization instructions and structured decision suggestions, and display the conversion results in a visual interface. The conversion results include a construction progress overview, a construction progress Gantt chart, a resource distribution heat map, and a resource allocation plan. The complex construction data is presented to decision makers in an intuitive and easy-to-understand manner through a visual interface, and multiple views such as global overview and dynamic scheduling are supported to significantly improve the decision-making efficiency and quality of decision makers, providing a strong guarantee for the successful implementation of the project.

[0062] It should be understood that "one embodiment" or "an embodiment" mentioned throughout the specification means that the specific features, structures or characteristics related to the embodiment are included in at least one embodiment of the present application. Therefore, "in one embodiment" or "in an embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. In addition, these specific features, structures or characteristics can be combined in one or more embodiments in any suitable manner. It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application. The above-mentioned serial numbers of the embodiments of the present application are for description only and do not represent the advantages and disadvantages of the embodiments.

[0063] It should be noted that, in this document, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of other identical elements in the process, method, article, or apparatus comprising the element.

[0064] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0065] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed across multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the embodiment of the present application.

[0066] In addition, all functional units in the embodiments of the present application can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the above-mentioned integrated units can be implemented in the form of hardware or in the form of hardware plus software functional units.

[0067] Alternatively, if the above-mentioned integrated unit of the present application is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present application, or the part that contributes to the relevant technology, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling the automatic test line of the device to execute all or part of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROMs, magnetic disks or optical disks.

[0068] The methods disclosed in the several method embodiments provided in this application can be arbitrarily combined without conflict to obtain new method embodiments.

[0069] The features disclosed in the several method or device embodiments provided in this application can be arbitrarily combined without conflict to obtain new method embodiments or device embodiments.

[0070] The above is merely an embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

Claims

1. A prefabricated building construction management system based on BIM and GNN model, characterized by: The system comprises: A data acquisition module is used to collect multi-source heterogeneous data throughout the construction cycle, including building information model data, sensor data, video stream data, and external data; A data processing module, configured to process the multi-source heterogeneous data to obtain data processing results, wherein the data processing results include a fused data set, standardized graph structure data, and structured decision suggestions, wherein the structured decision suggestions include construction progress prediction results, decision suggestions, and resource optimization suggestions; A data application module, configured to adjust resource allocation according to the data processing results and generate optimization instructions, wherein the optimization instructions include a structured instruction set and a structured solution set; A data visualization module is used to convert the optimization instructions and the structured decision suggestions, and display the conversion results in a visualization interface. The conversion results include a construction progress overview, a construction progress Gantt chart, a resource distribution heat map, and a resource allocation plan.

2. The system according to claim 1, wherein: The data processing module includes a data fusion unit, a feature extraction unit and a graph neural network inference unit, wherein: The data fusion unit is configured to perform spatiotemporal alignment processing on the sensor data, the video stream data, and the external data with the building information model data respectively through a spatiotemporal interpolation algorithm to obtain spatiotemporal aligned multimodal data, store associated data within the multimodal data based on a graph database, construct a construction knowledge graph, and fuse the multimodal data with the construction knowledge graph to obtain the fused data set; The feature extraction unit is used to refine the structural features of the fused data set to obtain the standardized graph structure data, wherein the standardized graph structure data includes a node feature matrix, an edge feature matrix and graph structure metadata; The graph neural network reasoning unit is used to model the standardized graph structure data according to the graph neural network, complete the prediction and reasoning tasks, and obtain the structured decision recommendations.

3. The system according to claim 2, characterized in that The construction knowledge graph includes building information model entities, building information model relationships and building information model attributes, where: The building information model entities include personnel entities, equipment entities, material entities, component entities, process entities and safety specification entities; The building information model relationship is the relationship between the building information model entities; The building information model attribute is a dynamic attribute added between the building information model entity and the building information model relationship.

4. The system according to claim 2, wherein: The extracting the structural features of the fused data set to obtain the standardized graph structure data includes: Extracting the characteristic vector of the building information model entity to form the node characteristic matrix; Converting the building information model relationship into an edge feature vector to form the edge feature matrix; The external data is injected into the construction knowledge graph as a global context feature vector, and the temporal feature vectors in the construction knowledge graph are extracted through a long short-term memory network to form the graph structure metadata.

5. The system according to claim 2, wherein: The graph neural network reasoning unit includes a message passing subunit, a feature updating subunit, a graph attention subunit and a decision generation subunit, wherein: The message passing subunit is used to aggregate feature information of adjacent nodes in the standardized graph structure data; The feature updating subunit is used to dynamically adjust the node priority in the standardized graph structure data in combination with the external data through a gated recurrent unit; The graph attention subunit is used to assign weights to key process nodes in the standardized graph structure data through a self-attention mechanism; The decision generating subunit is used to generate the structured decision suggestion according to the updated standardized graph structure data.

6. The system according to claim 1, wherein: The data application module includes a dynamic scheduling unit and a resource optimization unit, wherein: The dynamic scheduling unit is used to complete the scheduling operation based on the fused data set, the construction progress prediction result and the decision suggestion in combination with the genetic algorithm to obtain the structured instruction set, which includes process adjustment instructions and resource allocation instructions; The resource optimization unit is used to complete the resource optimization operation based on the fused data set, the construction progress prediction results and the resource optimization suggestions, combined with the linear programming algorithm, and output the Pareto optimal solution to obtain the structured solution set, which includes a process adjustment solution and a resource allocation solution.

7. The system according to claim 1, wherein: The visualization interface supports the overlay display of the 3D building information model and the real-time 3D building information model data, and the conversion of the optimization instructions and the structured decision suggestions includes: Aggregating actual construction progress data and resource status data to generate key performance indicators, including process delay rate and equipment utilization rate; Generating a construction progress Gantt chart based on a comparison result of the actual construction progress data and the construction progress prediction result; generating a resource distribution heat map according to the resource allocation plan; The multi-source heterogeneous data, the construction progress overview, the key performance indicators, the construction progress Gantt chart and the resource distribution heat map are updated in real time through a web page long connection protocol.

8. The system according to claim 1, wherein: The data visualization module includes a global overview visualization unit and a dynamic scheduling visualization unit, wherein; The global overview visualization unit is used to display the construction progress overview and the resource distribution heat map; The dynamic scheduling visualization unit is used to display the construction progress Gantt chart and the resource allocation plan.

9. A method for managing prefabricated building construction based on BIM and GNN model, characterized in that: Applied to an assembly-type building construction management system based on BIM and GNN models, the system includes a data acquisition module, a data processing module, a data application module, and a data visualization module. The method includes: Collecting multi-source heterogeneous data throughout the construction cycle, including building information model data, sensor data, video stream data, and external data; Processing the multi-source heterogeneous data to obtain data processing results, the data processing results including a fused data set, standardized graph structure data, and structured decision suggestions, the structured decision suggestions including construction progress prediction results, decision suggestions, and resource optimization suggestions; Adjust resource allocation according to the data processing result and generate optimization instructions, wherein the optimization instructions include a structured instruction set and a structured solution set; The optimization instructions and the structured decision suggestions are converted and the conversion results are displayed in a visual interface. The conversion results include a construction progress overview, a construction progress Gantt chart, a resource distribution heat map, and a resource allocation plan.

Images