Water conservancy project digital management method and system based on BIM

Abstract

The embodiment of the invention provides a BIM-based hydraulic engineering digital management method and system. The method comprises the following steps: performing multi-dimensional monitoring system deployment on a to-be-monitored area to obtain multi-dimensional hydrological data; constructing a first BIM based on the topographic data of the to-be-monitored area, the real-time work area image, the hydraulic engineering construction information of the design stage and the equipment deployment information; fusing the multi-dimensional hydrological data, the water area change data, the construction progress data and the operation and maintenance monitoring data into the first BIM, and constructing a second BIM including a full life cycle; and through the equipment characteristic curve and the water conservancy project physical model, in combination with the multi-dimensional hydrological data and the water area change data, predicting a water conservancy project structure change trend and potential risk factors in the second BIM, fusing the water conservancy project structure change trend and the potential risk factors into the second BIM, and displaying an implementation effect and improvement suggestions in real time in the second BIM. A user is assisted to realize hydraulic engineering digital management, and the engineering management efficiency is improved.

Description

Technical Field

[0001] The present application relates to the field of data processing, and in particular to a BIM-based digital management method and system for water conservancy projects. Background Art

[0002] In the current water conservancy monitoring system, some monitoring stations rely on manual data reporting. For example, small reservoir flood control communication systems are not yet widespread, resulting in delayed updates of key information such as water levels and rainfall, making it impossible to meet real-time scheduling requirements. Water conservancy information infrastructure is weak, network coverage is insufficient, and grassroots data transmission relies on outdated communication technologies. For example, some hydrological stations still use a combination of manual inspections and terrestrial communication networks (such as CDMA and GPRS) for monitoring. In remote areas, weak signals lead to data transmission failures, necessitating reliance on satellite communications, resulting in delayed updates of hydrological information.

[0003] Therefore, it is urgent to design a technical solution to overcome at least one technical problem in the related art. Summary of the Invention

[0004] The main purpose of the embodiments of the present application is to provide a BIM-based digital management method and system for water conservancy projects, aiming to solve at least one technical problem in related technologies: lack of unified supervision of water conservancy data, insufficient degree of data digitization, and delayed updating of hydrological information.

[0005] In a first aspect, an embodiment of the present application provides a BIM-based digital management method for water conservancy projects, comprising:

[0006] For the monitored area where the water conservancy project is located, a multi-dimensional monitoring system is deployed in the monitored area based on the topographic data of the monitored area and the water conservancy project information to obtain multi-dimensional hydrological data of the monitored area; wherein the topographic data of the monitored area includes at least: terrain elevation, slope, geological structure, and surface vegetation cover;

[0007] Acquire real-time images of the work area to be monitored through drones; construct the first building information model (BIM) based on the terrain data of the monitored area, real-time images of the work area, and water conservancy project construction information and equipment deployment information during the design phase;

[0008] Acquire water area change data in the monitored area through satellite remote sensing; integrate multi-dimensional hydrological data, water area change data, construction progress data during the construction phase, and operation and maintenance monitoring data during the operation and maintenance phase into the first BIM to construct a second BIM covering the entire life cycle;

[0009] By using equipment characteristic curves and the physical model of water conservancy projects, combined with multi-dimensional hydrological data and water area change data of the monitored area, the structural change trends and potential risk factors of water conservancy projects in the second BIM are predicted, and these structural change trends and potential risk factors are integrated into the second BIM and displayed to users in real time.

[0010] In response to the user's management operations on the second BIM, the implementation effect of the water conservancy project after the corresponding management strategy is implemented is predicted, and the implementation effect and management improvement suggestions are displayed in the second BIM to assist users in realizing digital management of water conservancy projects.

[0011] In a second aspect, an embodiment of the present application provides a BIM-based digital management system for water conservancy projects, including:

[0012] The acquisition unit is used to deploy a multi-dimensional monitoring system in the monitored area where the water conservancy project is located, based on the terrain data of the monitored area and the water conservancy project information, to obtain multi-dimensional hydrological data of the monitored area; wherein the terrain data of the monitored area includes at least: terrain elevation, slope, geological structure, and surface vegetation cover; obtain real-time work area images of the monitored area through drones; and obtain water area change data of the monitored area through satellite remote sensing;

[0013] The construction unit is used to construct a first building information model (BIM) based on the terrain data of the area to be monitored, real-time work area images, water conservancy project construction information and equipment deployment information in the design phase; multi-dimensional hydrological data, water area change data, construction progress data in the construction phase, and operation and maintenance monitoring data in the operation and maintenance phase are integrated into the first BIM to construct a second BIM covering the entire life cycle;

[0014] The prediction unit is used to predict the structural change trend and potential risk factors of the water conservancy project in the second BIM by combining the equipment characteristic curve and the physical model of the water conservancy project with the multi-dimensional hydrological data and water area change data of the monitored area, and integrate the structural change trend and potential risk factors of the water conservancy project into the second BIM for real-time display to users;

[0015] The interactive unit is used to respond to the user's management operations on the second BIM, predict the implementation effect of the water conservancy project after the corresponding management strategy is implemented, and display the implementation effect and management improvement suggestions in the second BIM to assist the user in realizing digital management of the water conservancy project.

[0016] In a third aspect, an embodiment of the present application further provides a terminal device, which includes a processor and a memory for storing computer programs; the processor is used to execute the computer program and implement the BIM-based digital management method for water conservancy projects described in the first aspect or any embodiment of the present application when executing the computer program.

[0017] The present invention provides a BIM-based digital management method and system for water conservancy projects. First, a multidimensional monitoring system is deployed based on terrain data (such as terrain elevation, slope, geological structure, and surface vegetation cover) and water conservancy project information in the monitored area. Monitoring equipment layout is adaptively planned to obtain multidimensional hydrological data for the monitored area, providing basic data support for subsequent management. Next, a drone is used to capture real-time images of the work area. This data is combined with terrain data, construction information from the design phase, and equipment deployment information to construct a first BIM model, achieving a digital representation of the water conservancy project design phase. Next, water area change data is acquired using satellite remote sensing. Multidimensional hydrological data, construction progress data, and operation and maintenance monitoring data are integrated into the first BIM model to form a second BIM model covering the entire lifecycle. Based on this, the system uses equipment characteristic curves and a physical model of the water conservancy project, combined with multidimensional hydrological and water area change data, to predict project structural change trends and potential risks, and present them in real time in the second BIM model. Finally, when users perform management operations on the second BIM, simulations predict the effectiveness of corresponding management strategies and provide management improvement suggestions to further assist users in decision-making. The embodiment of the present application realizes efficient integration and visualization of data, breaks down data silos, unifies multi-source heterogeneous data on the BIM platform, and can intuitively view information at all stages of the project, improve decision-making efficiency, significantly enhance risk prevention and control capabilities, optimize the entire life cycle management process, and assist in making scientific decisions on water conservancy projects through visual simulation of the effects of different management strategies, thereby effectively improving the management level of water conservancy projects. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 A flow chart of a BIM-based digital management method for water conservancy projects provided in an embodiment of the present application;

[0019] Figure 2 A schematic diagram of the module structure of a BIM-based digital management system for water conservancy projects provided in an embodiment of the present application;

[0020] Figure 3 A schematic block diagram of the structure of a terminal device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0021] In response to the technical problems existing in the relevant technologies, the embodiments of the present application propose a BIM-based digital management method and system for water conservancy projects. Specifically, the BIM-based digital management method for water conservancy projects solves the existing technical problems in water conservancy project management from multiple levels such as data integration, process optimization, and real-time monitoring. First, in the embodiments of the present application, a unified data architecture is built with the BIM model as the core, and a cross-departmental data sharing platform is established. By clarifying the data docking specifications of multiple departments such as water conservancy, meteorology, environmental protection, and land, the data resources scattered in various departments are integrated to eliminate data duplication and conflicts. At the same time, a unified data standard covering data format, coding rules, quality assessment, etc. is formulated to enable data between different systems such as flood control scheduling systems and water resources management systems to communicate with each other, avoiding the tedious process of manually integrating multi-source data and greatly improving data processing efficiency. For databases that lack maintenance, a regular data update and maintenance mechanism is integrated, and digital technologies such as image recognition and text extraction are used to repair and store damaged historical data to ensure data integrity, so that these historical data can also be integrated into the BIM model to provide more comprehensive data support for water conservancy project management.

[0022] Second, in the data collection process, a unified collection specification is formulated based on the embodiments of this application, and clear provisions are made for monitoring indicators, collection frequency, collection methods, etc. In response to the problem of manual data reporting, combined with the deployment of a multi-dimensional monitoring system, investment in automated monitoring equipment is increased, flood control communication systems are popularized in small reservoirs, and smart sensors are installed to collect key information such as water level and rainfall in real time, and the collected multi-dimensional hydrological data is directly incorporated into the BIM model. In cases where the degree of digitization of historical data is insufficient, a special team is formed to carry out digital processing, integrate the processed data into the BIM model, and establish a complete backup mechanism to prevent data loss and ensure the accuracy and integrity of BIM model data.

[0023] Third, in response to the problems of weak water conservancy information infrastructure and delayed data transmission, in the embodiments of the present application, in terms of information infrastructure construction, the coverage of the water conservancy information network is strengthened, and old communication technologies are upgraded and renovated. In remote areas, a combination of satellite communication and 5G communication is adopted to ensure the stable transmission of hydrological site data. At the same time, a data transmission redundancy mechanism is established. When one communication method fails, it automatically switches to the backup communication line to avoid data transmission failure. Whether it is water area change data obtained by satellite remote sensing, or construction progress data, operation and maintenance monitoring data, they can be integrated into the BIM model in a timely and accurate manner to meet the real-time scheduling needs of water conservancy projects. Based on real-time and accurate data, through equipment characteristic curves and water conservancy project physical models, the project structure change trends and potential risks are predicted in the BIM model and displayed to users in real time. When users manage and operate the BIM model, the system can also predict the implementation effect of the management strategy, assist users in making scientific decisions, and improve project management efficiency.

[0024] Embodiments of the present application provide a BIM-based digital management method and system for water conservancy projects. The BIM-based digital management method for water conservancy projects can be applied to terminal devices, such as mobile phones, virtual reality devices, tablet computers, laptop computers, desktop computers, wearable devices, and other electronic devices. The terminal device can be a server connected to a tower crane or a server cluster. The aforementioned connection method can be implemented via hardware circuits or a communication module.

[0025] The following is a detailed description of some embodiments of the present application in conjunction with the accompanying drawings. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other. Figure 1 , Figure 1 A flow chart of a BIM-based digital management method for water conservancy projects provided in an embodiment of the present application.

[0026] like Figure 1 As shown, the BIM-based digital management method for water conservancy projects includes the following steps S101 to S107.

[0027] Step S101: For the area to be monitored where the water conservancy project is located, a multi-dimensional monitoring system is deployed in the area to be monitored based on the terrain data of the area to be monitored and the water conservancy project information to obtain multi-dimensional hydrological data of the area to be monitored.

[0028] Step S102: Acquire a real-time work area image of the area to be monitored by using a drone.

[0029] Step S103: constructing a first building information model (BIM) based on the terrain data of the area to be monitored, the real-time work area image, the water conservancy project construction information and the equipment deployment information in the design phase.

[0030] Step S104: Acquire water area change data of the area to be monitored through satellite remote sensing.

[0031] Step S105: Integrate the multi-dimensional hydrological data, water area change data, construction progress data in the construction phase, and operation and maintenance monitoring data in the operation and maintenance phase into the first BIM to construct a second BIM covering the entire life cycle.

[0032] Step S106: Using the equipment characteristic curve and the physical model of the water conservancy project, combined with the multi-dimensional hydrological data and water area change data of the monitored area, predict the structural change trend and potential risk factors of the water conservancy project in the second BIM, and integrate the structural change trend and potential risk factors of the water conservancy project into the second BIM for real-time display to the user.

[0033] Step S107: In response to the user's management operation on the second BIM, predict the implementation effect of the water conservancy project after the corresponding management strategy is implemented, and display the implementation effect and management improvement suggestions in the second BIM to assist the user in realizing digital management of the water conservancy project.

[0034] In the embodiment of the present application, Building Information Modeling (BIM) is a building life cycle management method based on digital technology. It integrates multi-dimensional data such as geometric information, functional parameters, material properties, schedule costs, etc. of the construction project to construct a visual, parameterized, and collaborative virtual model, and uses this as the core to realize information sharing and collaborative work throughout the entire project process.

[0035] The essence of BIM is to simulate a real building using a digital model. Its core principles are embodied in three aspects: It integrates a building's geometric form (such as the dimensions and location of walls, beams, and columns) with non-geometric information (such as material properties, equipment models, and cost data) into a unified model, creating a detailed digital twin. For example, a BIM model of a water conservancy project not only displays its exterior and internal structure but also correlates equipment performance and energy consumption parameters for each device. During the design phase, engineers from different disciplines (architecture, structure, and mechanical and electrical engineering) can work concurrently on the same model, reviewing each other's designs in real time to avoid pipeline collisions and structural conflicts. When a design component is modified, the model automatically updates its associated content to ensure information consistency. Models are used throughout the entire building process, from planning and design to construction and operation and maintenance. For example, during the construction phase, models can be used for schedule simulation (4D BIM) and cost control (5D BIM). During the operation and maintenance phase, models are used to manage equipment maintenance, energy consumption, and other aspects, extending the building's lifecycle. During the architectural design phase, architects use BIM models to visually visualize the building's exterior and interior layout. Structural engineers simultaneously verify the proper load-bearing of beams and columns. Mechanical and electrical engineers simulate the routing of ventilation ducts, proactively identifying collision points between pipes and structural columns and adjusting the design, thus reducing the "errors, omissions, collisions, and gaps" common in traditional two-dimensional drawings. During the construction management phase, the construction team uses BIM models to simulate construction processes, pre-planning the concrete pouring sequence and machinery operation paths to avoid on-site chaos. Furthermore, by linking schedule and cost data, they monitor project progress and budget overruns in real time. For example, if a section of pile foundation construction ran three days overdue, the system automatically alerts and prompts adjustments to subsequent plans. During the operations and maintenance phase, the maintenance team uses BIM models to quickly locate faulty equipment, access information such as equipment purchase contracts and maintenance records, and develop maintenance plans. The model also integrates energy consumption data, analyzing electricity and water usage in each area and assisting in optimizing energy management strategies.

[0036] This reduces human error and shortens the design cycle through visual models and collaborative design. Rework is also reduced during the construction phase, lowering costs. During the planning phase, models are used to simulate the sunlight and ventilation effects of different design options to assist in selecting the optimal solution. During the operations and maintenance phase, historical data is used to predict equipment lifespans, allowing replacements to be scheduled in advance and avoiding unexpected failures. BIM supports green building analysis, such as simulating building energy consumption and carbon emissions, optimizing insulation material selection or photovoltaic panel layout, and helping achieve low-carbon goals. This provides a complete data foundation for subsequent renovations and expansions, avoiding the issues of lost drawings or information gaps found in traditional models.

[0037] In the embodiments of this application, topographic data of the monitored area is an important foundation for the digital management of water conservancy projects. Data such as terrain elevation, slope, geological structure, and surface vegetation cover provide a basis for project construction and monitoring from different dimensions. For example, spatial geographic information, such as digital elevation models, digital orthophotos, and vector topographic maps of the project area, is used to accurately integrate the BIM model with the terrain, enhancing the model's authenticity and practicality.

[0038] Specifically, the terrain data of the area to be monitored includes at least: terrain elevation, slope, geological structure, and surface vegetation coverage.

[0039] Terrain elevation data plays a critical role in the construction of reservoirs and dams. For example, during the construction of the Three Gorges Dam on the Yangtze River, precise terrain elevation measurements enabled accurate information on the riverbed height at the dam site and the height of the mountains along its banks. Engineers used this data to determine the dam's foundation depth and height, ensuring that the dam can intercept sufficient water for power generation and flood control while also maintaining its stability under varying water levels. Furthermore, during flood monitoring, terrain elevation data can help determine the extent of flooding, providing an accurate basis for evacuating people and deploying flood control supplies in downstream areas.

[0040] Slope data is crucial for the site selection and construction planning of water conservancy projects. For example, when constructing irrigation canals in mountainous areas, if the slope is steep, the canal design must consider the velocity of the water flow and its ability to resist scour. This may necessitate reinforcement measures or changes in the canal's direction to prevent damage from excessive water flow. Around small reservoirs, slope data can be used to assess the risk of landslides. If the slopes of the mountains upstream of the reservoir are steep, landslides could block the river or impact the reservoir dam during heavy rain. Slope data can help identify high-risk areas in advance, allowing preventive measures like planting trees to stabilize the soil and building retaining walls.

[0041] Geological structure directly impacts the safety and durability of water conservancy projects. For example, when constructing water conservancy projects in the karst regions of the southwest, due to the region's numerous caves, underground rivers, and other unique geological structures, geological structure data is required during the project site selection phase to conduct detailed surveys of the underground rock structure and avoid unfavorable geological areas such as caves and faults. In the planning of some routes for the South-to-North Water Diversion Project, geological structure data helped engineers determine the most suitable water transmission routes, avoiding ruptures or leakage in water pipelines due to geological instability and ensuring the long-term stable operation of the project.

[0042] Surface vegetation cover has a significant impact on the hydrological cycle and soil and water conservation. Analysis of surface vegetation cover data in soil and water conservation projects in the middle and upper reaches of the Yellow River revealed severe soil erosion in areas with sparse vegetation. Based on this, project personnel implemented ecological restoration measures such as afforestation and grass planting in these areas to increase vegetation coverage. The restoration of vegetation reduced direct soil erosion by rainwater, lowered river sediment concentrations, and improved water quality. Furthermore, the water conservation function of vegetation increased soil moisture, providing a more stable water source for irrigation of surrounding farmland.

[0043] In the embodiment of the present application, water conservancy project information runs through the entire life cycle of the project and is the core support for the BIM-based digital management method. From design and construction to operation and maintenance, and then to environmental and social impacts, information at each stage plays an important role.

[0044] Information during the design phase primarily serves as the blueprint for project construction. Basic engineering design data covers the overall project layout, including the reservoir dam type (gravity dam, arch dam, etc.), axis position and control coordinates, channel orientation and pipe diameter, and hydropower plant floor plan. These data define the overall structure and scale of the project. Structural design parameters, including dam material strength, impermeable layer thickness, foundation treatment method, gate opening and closing force, and water-stop device parameters, determine the safety and functionality of the project structure. Equipment selection data determines the model, rated flow / power, and installation location and technical specifications of equipment such as pumps and turbines to ensure proper operation of the project equipment. Regarding geological and environmental data, the engineering geological survey report provides information such as borehole histograms, geotechnical parameters, and groundwater levels and flow directions to assist in assessing engineering geological conditions. Environmental impact assessment data identifies the distribution of ecologically sensitive areas within the construction area, noise and dust control standards, and soil and water conservation plans to ensure that the project construction meets environmental requirements.

[0045] Information during the construction phase is crucial for the smooth progress of a project. Progress and resource data includes the construction schedule and critical paths for each sub-project. Resource input data records the use of labor, materials, and machinery and equipment, facilitating the rational allocation of resources and controlling the construction progress. Quality and safety data are crucial. Concealed engineering acceptance records reflect the construction quality of key areas like foundation treatment and anti-seepage walls. Construction change records describe the reasons for design changes, their content, and their impact on the construction period and costs. Safety accident and hazard records cover high-slope work protection and construction machinery troubleshooting, ensuring that construction safety and quality standards are met.

[0046] Information from the operation and maintenance phase is a core element for the long-term, stable operation of the project. Equipment operation data provides real-time monitoring of equipment parameters such as gate opening and closing status, pump head flow, and transmission line current and voltage. This data, combined with equipment records, documents equipment procurement, maintenance, and service life, enabling full lifecycle management of the equipment. Operation and maintenance management data includes inspection records documenting issues such as dam cracks and their treatments; dispatch instruction data covers flood control scheduling and water resource allocation plans; and emergency plan data includes emergency response procedures, material reserves, and deployment routes, ensuring project operational safety and the rational allocation of water resources.

[0047] Environmental and social data are also involved. This information reflects the project's impact on the surrounding environment and society. Hydrological and meteorological data collects real-time data on water levels, flow, rainfall, and other indicators, accumulating historical statistical data to provide a basis for flood control, drought relief, and water resources management. Ecological and social impact data monitors ecological indicators such as reservoir water quality, downstream river ecological flow, and surrounding vegetation coverage. It also collects social feedback such as complaints from surrounding residents regarding project operations and satisfaction with resettlement, helping the project achieve ecological friendliness and social harmony. Regulations and standards, including national and local water conservancy project construction specifications and industry standards, are used to verify the compliance of BIM models and ensure that project construction and management meet regulatory requirements.

[0048] Through multi-source collection, classification integration, and dynamic updating, this information is integrated into the BIM model to build a full-chain digital archive, providing support for multi-dimensional monitoring, risk prediction, and management decision-making of water conservancy projects, achieving responsibility traceability, and improving the level of intelligent management.

[0049] As an optional embodiment, in step S101, a three-dimensional terrain model of the area to be monitored is established based on the terrain elevation, slope, geological structure, and surface vegetation coverage in the terrain data; the spatial layout and functional zoning of the water conservancy project are determined based on the structural design data, equipment deployment status, and construction log in the water conservancy project information; based on the spatial layout and functional zoning of the water conservancy project, the three-dimensional building framework of the water conservancy project is superimposed on the three-dimensional terrain model, and the collision between the three-dimensional building framework and the surrounding scenery is detected to obtain a water conservancy project model that includes the surrounding environment and terrain characteristics; through the hydrological monitoring inference chain, the key monitoring points in the water conservancy project model are identified, and corresponding monitoring equipment and monitoring ranges are configured for the key monitoring points to obtain a multi-dimensional monitoring system deployment plan; the coverage rate of the monitoring equipment and the monitoring range is tested; if the coverage rate meets the qualified conditions, the corresponding multi-dimensional monitoring system deployment plan is executed, and the multi-dimensional hydrological data of the area to be monitored is collected through the deployed monitoring equipment.

[0050] In this way, by integrating terrain and project data to construct a precise model, and then intelligently planning the monitoring layout based on the model, efficient data collection is achieved. In principle, the monitoring system is constructed and optimized in a hierarchical manner, based on terrain data and water conservancy project information. First, spatial modeling techniques are used to construct a three-dimensional terrain model using terrain data such as elevation, slope, geological structure, and surface vegetation cover, faithfully reproducing the regional geomorphological characteristics. Second, based on the structural design data, equipment deployment, and construction logs of the water conservancy project, the spatial layout and functional zoning of the project, such as the dam structure and the equipment distribution of the hydropower station, are determined. Next, the three-dimensional architectural framework of the water conservancy project is superimposed on the three-dimensional terrain model. Collision detection algorithms are used to avoid spatial conflicts between the project and the surrounding environment (such as mountains and vegetation), resulting in a realistic water conservancy project model. Finally, using the hydrological monitoring inference chain, key elements in the model, such as flow paths and structural weaknesses, are analyzed to identify key monitoring points, match them with appropriate monitoring equipment and coverage, and formulate a deployment plan. Furthermore, coverage testing ensures that the monitoring network is comprehensive, and the plan is implemented only when the target is met, ensuring comprehensive data collection.

[0051] Taking the construction of a reservoir in a mountainous area as an example, during the early planning phase, high-precision mapping was used to obtain terrain elevation data for the area. Combined with slope analysis, steep slopes and flat areas were identified. Geological structural data revealed faults in some areas, requiring focused monitoring. Based on vegetation cover, dense forests were avoided to minimize monitoring interference, resulting in a three-dimensional terrain model. Combined with the reservoir's dam design drawings, turbine installation locations, and other hydraulic engineering information, functional zones such as the dam, water pipeline, and power plant were identified. The three-dimensional building framework was then integrated into the 3D terrain model. Collision detection was used to adjust the layout to avoid conflicts between construction and the surrounding environment. Subsequently, a hydrological monitoring inference chain was used to identify key monitoring points, such as the dam foundation and water pipeline interfaces. Displacement sensors and pressure sensors were deployed at these locations, and monitoring ranges were set to create a deployment plan. Once coverage passed the test, the equipment was deployed according to the plan to collect multi-dimensional hydrological data, including water level, water pressure, and dam displacement, in real time.

[0052] In terms of technical effectiveness, it has significantly improved the scientific nature and accuracy of water conservancy project monitoring. Through three-dimensional modeling and collision detection, the rationality of the water conservancy project layout is ensured, avoiding construction risks and subsequent operation and maintenance hazards caused by spatial conflicts. Hydrological monitoring reasoning chains can also be used to determine key monitoring points, changing the blindness of traditional monitoring point layout, enabling monitoring equipment to accurately cover core areas, reducing resource waste, and improving the pertinence and effectiveness of data collection. The coverage detection mechanism ensures the integrity of the monitoring network, enabling managers to fully understand the operating status of water conservancy projects and changes in the surrounding hydrological environment, providing reliable data support for decisions such as flood control scheduling and project maintenance.

[0053] It's worth noting that the hydrological monitoring inference chain is a dynamic deduction mechanism based on hydrological process logic, data correlation analysis, and domain knowledge. Its core principle is to establish a logical chain of "data collection - feature extraction - causal analysis - trend prediction - decision-making recommendations" by integrating multi-source monitoring data (such as water level, flow, water quality, and soil moisture), combining hydrological models (such as runoff models and hydrodynamic models), and expert experience rules. Specifically, it identifies the spatiotemporal correlations between hydrological elements (such as the lagged response relationship between rainfall and river flow), physical mechanisms (such as the impact of terrain slope on runoff velocity), and historical patterns (such as the evolution of seasonal floods). It conducts a multi-level analysis of the hydrological state of the monitored area, thereby achieving real-time monitoring of key hydrological processes, anomaly identification, and future deduction.

[0054] Taking a monitoring scenario involving a reservoir complex in a particular river basin as an example, real-time data from rainfall stations, water level stations, weather radar, and other equipment within the basin was first collected. For example, at 8:00 a.m. on a particular day, rainfall in the upstream mountainous area reached 50 mm / h (exceeding the 90th percentile for the same period in history), and inflows to the three upstream reservoirs increased by 30% over the previous two hours. Using the "Rainfall-Runoff Generation" module of the inference chain, combined with the regional soil type (sandy loam with a high permeability coefficient) and vegetation cover (65%), the runoff generation coefficient for this rainfall was calculated to be approximately 0.4, predicting a flood peak in the downstream river within 12 hours. Further invocation of the reservoir scheduling rule library revealed that the current water levels in all reservoirs were close to the flood control limit, and that the downstream river embankments were designed for a 50-year flood response. Combined with hydrodynamic model simulations, it was determined that without pre-discharge measures, the flood peak could exceed the downstream embankment's safe water level by 1.2 meters. The floodgates at upstream Reservoir A were immediately opened, increasing the discharge from 80 m³ / s to 150 m³ / s. Downstream townships were notified to prepare for early warnings. Subsequent monitoring data showed that the peak water level after the adjustment was 0.8 meters lower than predicted, remaining within the levee's safety threshold.

[0055] The hydrological monitoring inference chain overcomes the limitations of single-parameter monitoring. By correlating multiple data elements, such as precipitation, runoff, sediment, and water quality, it reveals the complex coupling relationships within hydrological processes (such as the carryover effect of stormwater runoff on non-point source pollution), thereby enhancing the comprehensiveness of monitoring. Threshold rules trained on historical data and machine learning models (such as LSTM time series prediction) can automatically identify abnormal events such as sudden water level rises and falls and sudden changes in water quality. For example, if the ammonia nitrogen concentration in a lake exceeds the Class III water standard for three consecutive hours, the inference chain automatically triggers the "pollution source tracing - emergency response" process, shortening the response time to within one hour. By embedding physical models (such as the HEC-RAS river hydrodynamic model) and statistical models, disasters such as floods, droughts, and saltwater tides can be simulated in advance. In a coastal estuary application, the inference chain's prediction error for saltwater tide upstreaming was less than 5%, providing 48 hours of preparation time for water diversion operations in the irrigation district. Expert experience is transformed into an executable rule engine (such as reservoir operation diagrams and gate opening and closing strategies), and operational plans are dynamically generated based on real-time data, reducing the subjectivity of manual decision-making. After application in a large irrigation area, irrigation water efficiency increased by 18%, and the time required for water scheduling decisions was shortened from 4 hours to 30 minutes.

[0056] Alternatively, graph neural networks (GNNs) can be introduced to model the spatial topological relationships of watersheds and combined with traditional hydrological models (such as SWMM) to improve the simulation accuracy of runoff distribution in complex terrain. For example, in mountainous and hilly areas, the fusion model can reduce the prediction error of flood peak arrival times in small watersheds. Lightweight inference models (such as TensorFlowLite) can be deployed on monitoring equipment to enable real-time anomaly detection (such as sensor fault identification and data jump filtering) at the data acquisition end, reducing invalid data transmission and alleviating cloud computing pressure. This extends beyond single hydrological element monitoring to multi-objective analysis encompassing "hydrology, ecology, and society." For example, in reservoir operation, ecological flow assurance, irrigation water demand satisfaction, and power generation efficiency can be simultaneously optimized, generating Pareto optimal solutions using a multi-objective genetic algorithm (NSGA-II). Data output from the inference chain, such as flood evolution and water quality diffusion range, can be connected to a VR platform for intuitive 3D dynamic display, helping decision makers understand complex hydrological scenarios. During a flood prevention drill in a particular river basin, VR visualization improved the efficiency of flood risk assessment by command personnel.

[0057] Further optionally, in 101, before identifying the key monitoring points in the water conservancy project model and configuring the corresponding monitoring equipment and monitoring range for the key monitoring points to obtain the multi-dimensional monitoring system deployment plan, the natural geographical entities and construction entities in the area to be monitored can also be obtained; the natural geographical entities include at least: topography, geological structure, and vegetation coverage; the construction entities include at least: water conservancy project buildings, hydrological monitoring equipment, and management units. Establish a spatial correlation relationship, a temporal causal chain, and a functional correlation relationship between the natural geographical entities and the construction entities in the area to be monitored, and obtain a dynamic correlation relationship between the natural geographical entities and the construction entities. The monitoring point site selection rules, equipment configuration rules, and coverage range setting rules are added to the dynamic correlation relationship through a dynamic monitoring logic reasoning chain to construct a hydrological monitoring reasoning chain with a closed data flow loop; the hydrological monitoring reasoning chain is used to dynamically construct a monitoring point addressing logic and a device deployment logic adapted to the area to be monitored.

[0058] Based on exploring the inherent connections between natural geographic entities and construction entities, a dynamic relational network is constructed. Natural geographic entities (topography, geological structure, and vegetation cover) and construction entities (hydraulic engineering structures, monitoring equipment, and management units) do not exist independently but rather influence each other spatially, temporally, and functionally. For example, topography determines the site selection and flow paths of hydraulic engineering projects, which in turn influences the layout of hydrological monitoring equipment. Changes in geological structure during construction can also create new requirements for project stability and monitoring needs. A dynamic relational map is formed by establishing spatial relationships (such as the relative position of engineering structures and terrain), temporal causal chains (such as the phased impact of construction progress on monitoring needs), and functional relationships (such as the connection between vegetation cover and soil erosion monitoring). Furthermore, professional knowledge, such as monitoring point site selection rules, equipment configuration rules, and coverage setting rules, is integrated into the dynamic relations. Leveraging a dynamic monitoring logic chain, intelligent deduction from entity relationships to monitoring strategies is achieved, constructing a closed-loop hydrological monitoring reasoning chain. This reasoning chain can dynamically generate monitoring point addressing logic (determine the optimal location of monitoring points) and equipment deployment logic (select appropriate equipment and monitoring range) according to the characteristics of different monitored areas.

[0059] Taking a large-scale reservoir project in a mountainous area as an example, during the initial planning phase, the natural geographical features of the area to be monitored were first identified, including steep mountainous terrain, complex folded geological structures, and uneven vegetation cover. Construction entities included hydraulic engineering structures such as the dam and spillway, as well as the planned deployment of water level and flow monitoring equipment and management units. Analysis revealed that the dam foundation is located near a fault, necessitating spatial monitoring of geological structural changes. Areas with poor vegetation cover are prone to soil erosion during the rainy season, impacting reservoir water quality, necessitating enhanced monitoring during the rainy season. The spillway's function is linked to the downstream river's flood-carrying capacity, necessitating functionally linked monitoring of upstream and downstream water level changes. Based on these dynamic correlations, combined with the rules that monitoring points should be selected in geologically weak areas, equipment configuration should take accuracy and environmental adaptability into consideration, and the coverage should ensure that there are no blind spots, the hydrological monitoring reasoning chain was used to conclude that: displacement sensors should be arranged at the dam foundation to monitor geological deformation; soil moisture and water quality monitoring equipment should be set up in the upstream area with sparse vegetation; water level gauges should be deployed upstream and downstream of the spillway, and appropriate monitoring ranges should be set, thus forming a multi-dimensional monitoring system deployment plan.

[0060] Traditional monitoring point placement often relies on experience, which can lead to problems such as irrational layout, blind spots, or redundant equipment. However, solutions based on the hydrological monitoring inference chain precisely analyze the dynamic relationships between entities, making monitoring point selection more tailored to actual needs and equipment configuration more targeted, effectively avoiding waste of resources. Furthermore, the closed-loop data flow design ensures that monitoring data is promptly fed back into the inference chain, allowing monitoring strategies to be dynamically adjusted based on actual conditions. This improves the adaptive capabilities of the monitoring system and provides more reliable data support for the safe operation and scientific management of water conservancy projects.

[0061] Alternatively, extensive historical monitoring data and engineering case studies can be used to continuously optimize the rules and logic within the inference chain, enabling it to more accurately adapt to diverse engineering scenarios. Furthermore, integration with IoT technology can be strengthened, with intelligent sensors collecting real-time data on entity state changes and promptly updating dynamic relationships. This will enhance the real-time and dynamic response capabilities of the hydrological monitoring inference chain. Furthermore, this inference chain can be combined with virtual reality and digital twin technologies to simulate the deployment of monitoring systems in a virtual environment, assessing the feasibility of solutions in advance and providing more intuitive and efficient decision-making support for the planning of water conservancy project monitoring systems.

[0062] In step S102, a real-time work area image of the area to be monitored is obtained by using a drone.

[0063] For example, drones played a key role in acquiring real-time images of the work area during a hazard removal and reinforcement project at a medium-sized reservoir in a mountainous area. The reservoir's clay core dam had developed hazards such as cracks on the dam slope and aging of the impermeable structure due to long-term operation. During the hazard removal and reinforcement construction phase, drones regularly (twice weekly) captured aerial images of the dam area, spillway, and surrounding mountainous terrain, acquiring real-time images of the work area with a resolution of up to 2 cm.

[0064] Drone footage clearly captures the progress of construction on the upstream and downstream slopes of the dam. Details such as the layered thickness of the clay core fill, the continuity of the filter layer, and the quality of the concrete slope protection are clearly visible. For example, software analysis of aerial images revealed three honeycomb defects, approximately 5-10 cm in diameter, on the concrete slope protection on the downstream slope of the dam's left bank. Construction personnel were promptly notified to repair these defects, preventing potential water seepage. Drones can also monitor the slope stability of the spillway expansion project. Comparing images from different periods revealed signs of localized landslides on the right bank slope (displacement of surface vegetation and increased exposed soil). Construction personnel immediately adjusted blasting parameters and added anchor support to prevent further landslides.

[0065] During environmental monitoring around the reservoir area, drone imagery can identify potential risks such as mountain cracks and changes in vegetation cover. For example, by comparing multiple images, a new crack approximately 20 meters long and 3 centimeters wide was discovered on the right bank of the reservoir tail. Combined with topographic data, the site was identified as a landslide-prone area, prompting the timely establishment of additional surface displacement monitoring points and the activation of an early warning mechanism. Furthermore, real-time drone imagery is used to measure construction progress. Image recognition technology automatically calculates the volume of excavated earthwork and concrete poured, comparing it to the bill of quantities in the BIM model to monitor progress deviations in real time. In one case, when rain delayed earthwork excavation by three days in a certain section, the system automatically issued an early warning and adjusted machinery configuration to ensure that the construction period of key lines was not affected.

[0066] In the embodiment of the present application, the drone, with its high maneuverability, high-precision imaging and real-time data return capabilities, has become a key component of water conservancy project construction sites, effectively making up for the limitations of manual inspections in high-risk areas and complex terrains, and providing intuitive and timely visualization basis for project quality control, safety monitoring and progress management.

[0067] In step S103, a first building information model (BIM) is constructed based on the terrain data of the area to be monitored, the real-time work area image, the water conservancy project construction information in the design phase, and the equipment deployment information.

[0068] As an optional embodiment, in step S103, a static terrain model is established based on the terrain elevation, slope, geological structure, and surface vegetation coverage in the terrain data, and a dynamic terrain model is established based on the real-time work area image. The terrain background model of the water conservancy project is obtained through multi-dimensional point cloud matching and fusion; a three-dimensional building model of the water conservancy project is constructed based on the water conservancy project construction information, and the three-dimensional building model is geometrically embedded and matched with the terrain background model, and superimposed and fused to obtain a first BIM; the building parameters, equipment parameters, and construction progress data of the water conservancy project in the design stage are associated with each model component of the first BIM; a building surface structure model of the water conservancy project is constructed based on the real-time work area image, and the building surface structure model is obtained in the first BIM through multi-dimensional point cloud matching and fusion, and surface defects of the engineering structure in the building surface structure model are identified and marked in the defect detection layer of the first BIM.

[0069] In principle, the first BIM model is constructed through layered modeling and data fusion. First, a static terrain model is created using traditional surveying and GIS technologies, leveraging terrain data including elevation, slope, geological structure, and surface vegetation cover. This model reflects the terrain's basic characteristics and long-term stability. Simultaneously, a dynamic terrain model is constructed using photogrammetry and computer vision techniques, using real-time drone-generated worksite images to capture the terrain's real-time state due to construction and natural changes. Multi-dimensional point cloud matching and fusion of the static and dynamic models are combined to create a terrain background model that reflects the current status and fundamental characteristics of the terrain. Second, based on the water conservancy project construction information, 3D modeling software is used to construct a 3D building model of the project, including structures such as dams, channels, and powerhouses. Using geometric embedding matching technology, the 3D building model is precisely overlaid on the terrain background model, forming the foundational architecture of the first BIM. Then, architectural parameters (such as material strength and dimensions), equipment parameters (such as model and performance indicators), and construction progress data from the design phase are linked to each component of the BIM model, enabling dynamic model updates and data integration. Finally, the real-time work area images are used again to construct a surface structural model of the building, which is then embedded into the first BIM through multi-dimensional point cloud matching and fusion. The image recognition algorithm is then used to detect defects such as cracks and wear on the surface of the engineering structure, which are marked in the defect detection layer to achieve a detailed display of the engineering status.

[0070] Taking a large-scale inter-basin water diversion project as an example, during the initial model construction, a static terrain model was created based on regional topographic data, showcasing topographic features such as mountain orientation and river distribution. Regular drone aerial photography was used to obtain real-time images of the work area, and a dynamic terrain model was constructed to capture topographic changes during channel excavation and dam filling. The static and dynamic models were then fused through point cloud matching to produce an accurate terrain background model. Next, based on the design drawings and parameters of the water diversion project, a 3D building model, including the water diversion channel, pumping station, and sluice gates, was constructed. This model was then overlaid and integrated with the terrain background model to form the foundational framework for the first BIM. The design parameters of each building structure, equipment specifications, and construction schedule were linked to the corresponding model components, allowing the model to intuitively demonstrate the project design and construction progress. Furthermore, drone imagery was used to construct a surface structural model of the building. Image recognition technology detected cracks in a certain channel slope section and annotated the defect in the BIM model's defect detection layer, providing a basis for project quality control and maintenance decisions.

[0071] Because traditional BIM model construction often relies on design drawings, it is difficult to reflect terrain changes and construction status in real time. However, this application integrates static and dynamic terrain models to ensure that the model is highly consistent with the actual terrain. The integration of multi-source data enables the model to contain rich engineering information, facilitating a comprehensive understanding of the project status by design, construction, and management personnel. The combination of building surface structural models and defect detection enables visual monitoring of project quality. Compared with manual inspections, structural defects can be detected more efficiently and accurately, providing strong support for project quality control and safe operation.

[0072] Alternatively, integration with IoT devices could be strengthened, integrating real-time sensor data (such as structural stress and vibration) into the BIM model to enable dynamic model updates and intelligent early warnings. Furthermore, virtual reality and augmented reality technologies could be used to transform the primary BIM model into an immersive experience, facilitating virtual inspections and solution verification for managers.

[0073] In addition, we can explore deep integration with geographic information systems (GIS) to enable BIM models to have more powerful spatial analysis and decision-making support capabilities, further improving the level of digital management of water conservancy projects.

[0074] Further, optionally, in the above steps, after geometrically embedding and matching the three-dimensional building model with the terrain background model and superimposing and fusing them to obtain the first BIM, the Douglas-Peucker algorithm can be used to optimize and compress the dynamic trajectories associated with the construction progress and equipment deployment progress in the first BIM. Furthermore, the first BIM can be geometrically simplified using LOD grading technology to achieve model compression of the first BIM. In the process of constructing a BIM model for a water conservancy project, dynamic trajectory optimization and geometric simplification of the model are key steps to improve model performance. The application of the Douglas-Peucker algorithm and LOD grading technology provides an effective means for this.

[0075] It is understandable that the core principle of the Douglas-Peucker algorithm is to compress and simplify the curve or trajectory by iteratively screening key points. Based on the error threshold, the algorithm selects the first and last points in a broken line as the initial nodes, calculates the perpendicular distances from the middle points to the line segment, retains the point with the largest distance, and divides the broken line into two segments with this point as the boundary. Repeat the above operation for these two segments until the distances from all points to the corresponding line segments are less than the threshold. After algorithm processing, redundant points in the broken line are eliminated, and the retained points can not only reflect the overall shape of the broken line, but also greatly reduce the amount of data. For example, the core formula of the Douglas-Peucker algorithm is used to calculate the perpendicular distance from a point to a line segment to determine whether to retain the point. Let the two end points of the line segment be A( , ) and B( , ), a point outside the line segment is C( , ), then the perpendicular distance from point C to line segment AB is The formula is: In the above formula, the numerator is calculated by vector cross product to calculate twice the area of ​​the triangle (the absolute value represents the area size), and the denominator is the length of the line segment AB. The two are divided to get the vertical distance from the point to the line segment. This algorithm is achieved by setting a distance threshold (such as ), when the distance of a point is greater than the threshold, the point is retained, otherwise it is eliminated, thereby achieving simplified compression of the trajectory.

[0076] In BIM modeling of dynamic trajectories for water conservancy project construction (such as machinery movement paths and equipment installation process nodes), the original trajectories are composed of densely sampled points, resulting in high data redundancy. The Douglas-Peucker algorithm can filter key nodes according to a preset accuracy (e.g., centimeter-level error tolerance) and eliminate minor points. This reduces data volume while maintaining trajectory morphological characteristics (such as turning angles and path direction), improving model loading speed and visualization efficiency.

[0077] In the application of BIM models for water conservancy projects, the Douglas-Peucker algorithm can be used to optimize dynamic trajectories associated with construction progress and equipment deployment. For example, during the construction of a large hydropower station, the BIM model recorded dynamic data such as the movement trajectories of concrete mixer trucks on the construction site and the paths of crane hoisting equipment. This trajectories are voluminous, and directly storing and displaying them would consume significant system resources. The Douglas-Peucker algorithm optimizes and compresses the trajectories, removing unnecessary path points. While preserving the basic trajectory shape, the data volume is reduced by approximately 60%. This makes the model smoother in displaying dynamic changes in construction progress and reduces the burden on system operations.

[0078] For example, during the construction of a reservoir dam, the daily trajectory of a bulldozer was recorded in the BIM model as a continuous polyline containing tens of thousands of points. When applying the Douglas-Peucker algorithm, with a threshold of 0.5 meters (the maximum distance the trajectory was allowed to deviate from the actual path was 0.5 meters), it automatically retained points reflecting key locations such as the bulldozer's round-trip unloading, climbing, and turning, while eliminating redundant points in straight-line sections. The resulting trajectory had fewer points, while still clearly depicting the construction machinery's operating patterns and efficiency bottlenecks (such as the frequent turning points at the dam-slope junction).

[0079] The Level of Detail (LOD) grading technology performs different degrees of geometric simplification on the three-dimensional model according to the viewing distance or display requirements. When the user views the model from a farther perspective, the system automatically loads the model with a low level of detail, retaining only the main outline and structure of the model. As the perspective zooms in, it gradually switches to a model with a high level of detail to show more details. LOD technology is applied in the first BIM model of the water conservancy project. For example, for a large reservoir dam model, only the overall outline and key structures of the dam are displayed when viewed from a distance. When viewed from a close distance, the surface texture of the dam, gate details, etc. are presented. In this way, lightweight display of the model in different scenarios is achieved, effectively reducing the resource consumption required for model rendering and transmission.

[0080] These steps significantly reduce model data volume, optimize model storage and transmission efficiency, and enable smooth model loading and operation across various terminal devices. This improves model display performance, enabling rapid response for both macroscopic inspections of the overall project layout and microscopic analysis of local structural details. This provides an efficient and convenient model application experience for personnel at all stages of the design, construction, and management.

[0081] In step S104, water area change data of the area to be monitored is acquired through satellite remote sensing.

[0082] For example, in a plateau lake ecological restoration project, satellite remote sensing plays a key role in monitoring watershed changes. Due to agricultural irrigation and climate change within the basin, the lake's water area has shrunk by 15% over the past decade, and shoreline erosion has intensified. Satellites equipped with multispectral sensors (such as Landsat 8 and Sentinel-2) regularly acquire remote sensing images of the lake (every 16 days) to extract key data such as water boundaries, water depth distribution, and algae cover. Processing these images with water indices (such as the Normalized Difference Water Index (NDWI)) clearly distinguishes the lake from the surrounding land. For example, remote sensing images from May 2024 show that the lake's northwestern water area has decreased by 2.3 square kilometers compared to the same period last year. Combined with topographic data, this area is an alluvial fan fed by the main river inflow. Excessive groundwater extraction has led to a drop in groundwater levels and increased seepage from the lake. Further analysis of the subtle deformation of the lakeshore line using synthetic aperture radar (SAR) data from radar satellites (such as Sentinel-1) revealed an erosion zone about 500 meters long on the southwestern lakeshore and abnormal surface radar echo signals, indicating that the soil on the lakeshore is loose and prone to collapse.

[0083] In evaluating the effectiveness of ecological remediation, satellite remote sensing continuously monitors trends in watershed changes. Following the project's implementation, through measures such as restoring the ecological flow of rivers flowing into the lake and constructing artificial wetlands, remote sensing imagery from March 2025 shows that the lake's water area has increased by 1.8 square kilometers compared to pre-remediation levels. The northwestern seepage zone has expanded landward due to rising groundwater levels. Furthermore, chlorophyll-a concentration inversion results indicate a decrease in algal cover, improving water quality from Class V to Class IV. Satellite remote sensing can also capture remote areas difficult to reach with traditional manual monitoring, such as the marsh wetlands southeast of the lake. Multi-temporal image comparisons reveal increased vegetation cover, demonstrating significant success of ecological restoration measures. In these examples, satellite remote sensing, with its wide coverage, stable observation period, and high data consistency, enables macro-dynamic monitoring of changes in plateau lake waters. This provides full-cycle data support for the development of ecological water replenishment plans, site selection for lakeshore protection projects, and evaluation of remediation effectiveness, complementing the limitations of ground-based monitoring in terms of spatial scale and timeliness.

[0084] In step S105, multi-dimensional hydrological data, water area change data, construction progress data in the construction phase, and operation and maintenance monitoring data in the operation and maintenance phase are integrated into the first BIM to construct a second BIM covering the entire life cycle.

[0085] As an optional embodiment, in step S105, multi-dimensional hydrological data, water area change data, construction progress data during the construction phase, and operation and maintenance monitoring data during the operation and maintenance phase are integrated into the first BIM to construct a second BIM covering the entire life cycle, including:

[0086] Multidimensional hydrological data, water area change data, and construction progress data are spatiotemporally aligned and spatiotemporally correlated to obtain a first spatiotemporal cube model for displaying the correlation between construction progress and actual environmental conditions; multidimensional hydrological data, water area change data, and operation and maintenance monitoring data are spatiotemporally aligned and spatiotemporally correlated to obtain a second spatiotemporal cube model for displaying the correlation between operation and maintenance monitoring operations and actual environmental conditions; the first spatiotemporal cube model and the second spatiotemporal cube model are respectively used as three-dimensional layers, and the three-dimensional layers are integrated into the first BIM through a geometric semantic dual-driven matching method to construct a second BIM; based on real-time work area images and construction progress data, an image rollback axis is constructed in the second BIM to trace the actual construction situation, and the actual construction scenes at different times are dynamically displayed through the image rollback axis.

[0087] The above steps achieve deep integration of information throughout the entire life cycle of water conservancy projects through spatiotemporal data integration and dynamic display technology. The following explains the principles, examples, technical effects, and optimization directions.

[0088] In principle, the system relies on structured processing of spatiotemporal data and fusion of multi-source information. First, multidimensional hydrological data (such as water level, flow, and water quality), watershed change data (area and shoreline evolution), and construction progress data (process completion and resource input) are spatiotemporally aligned, unifying them into the same temporal and spatial coordinate system. Relationships between these data sets are then explored, such as analyzing the correlation between rainfall intensity and excavation progress. This results in a first spatiotemporal cube model, visually depicting the interaction between the construction process and environmental changes. Similarly, for the operation and maintenance phase, hydrological, watershed, and operation and maintenance monitoring data (equipment operating parameters and inspection records) are integrated into a second spatiotemporal cube model, demonstrating the inherent connections between equipment maintenance, scheduling, and environmental conditions. Subsequently, using dual-driven geometric and semantic matching, the spatiotemporal cube model is integrated into the first BIM as a 3D layer, based on both spatial geometry (such as equipment installation coordinates) and semantic attributes (such as equipment type and monitoring indicators), creating a second BIM covering the entire lifecycle of design, construction, and operation and maintenance. Finally, an image rollback axis is constructed using real-time work area images and construction progress data. Dynamic backtracking of the construction scene is achieved by sliding the timeline, making it convenient for users to intuitively compare the planned and actual construction conditions.

[0089] For example, during the construction phase of a large-scale river regulation project, a first space-time cube model was constructed by temporally and spatially aligning daily water level and flow data with the progress of river dredging and revetment construction. For example, during a flood, the model clearly demonstrated a correlation between a three-day suspension of dredging operations due to rising water levels and a 5% delay in revetment concrete pouring. During the operation and maintenance phase, a second space-time cube model was integrated with water quality monitoring data, gate operation and closing records, and changes in water area. This model revealed that after a sewage discharge event, the downstream water area decreased by 2%, and the frequency of gate openings increased to accelerate water replacement. Through dual-driven geometric semantic matching, the two space-time cube models were integrated into the first BIM model to form a complete second BIM. Furthermore, an image scroll axis was constructed based on real-time drone-captured work site images and construction logs. Users can scroll through the timeline to view the entire process of a particular river revetment section, from foundation excavation to completion, and compare it with the design and progress data in the BIM model to promptly identify construction deviations.

[0090] At the data management level, the space-time cube model enables structured integration of multi-source heterogeneous data, breaking down information silos and improving data utilization. In terms of visualization and decision support, the integration of image scrolling and full-lifecycle BIM provides managers with intuitive tools for historical tracing and current situation analysis, improving decision-making efficiency. Furthermore, by demonstrating the connections between construction, operations, and the environment, it helps to predict the impact of environmental changes on projects in advance, such as the risk of construction delays due to flooding or the potential threat to equipment operation due to deteriorating water quality, allowing preventive measures to be taken.

[0091] Furthermore, integrating virtual reality (VR) or augmented reality (AR) technology with image scrolling provides users with an immersive construction review and operations management experience. Furthermore, the integration with the City Information Model (CIM) can be explored, integrating water conservancy project BIM models into a larger urban space management system to achieve multi-project collaborative management and optimized resource allocation.

[0092] For example, in a large-scale coastal cross-sea bridge water conservancy project, the spatiotemporal alignment and correlation of multidimensional hydrological data, watershed change data, and construction progress data provided critical support for the project. The project encompassed pier foundation construction and breakwater construction, with the construction area significantly affected by tides and waves. During construction, multiple coastal hydrological monitoring stations collected hourly multidimensional hydrological data, including tide level, flow velocity, and wave height. Satellite remote sensing was used to acquire weekly watershed imagery, analyzing shoreline changes, erosion, and siltation in the surrounding waters to generate watershed change data. Construction progress data, such as the pier pile driving depth and breakwater concrete pouring volume, was recorded daily. For spatiotemporal alignment, all data was based on Coordinated Universal Time (UTC) and the unified geographic coordinate system for the project area. For example, at 8:00 AM on July 15, 2024 (time alignment), the high tide level of 4.2 meters recorded by the tide monitoring station, the increase of 100 square meters of erosion in the waters near the bridge piers as indicated by satellite remote sensing, and the third pile foundation of Pier 2, completed that day and driven to a depth of 15 meters (spatially aligned to the pier coordinates), were integrated into the same spatiotemporal framework. Spatiotemporal correlation analysis revealed that whenever the tide exceeded 3.5 meters and the current velocity exceeded 1.5 meters per second (multi-dimensional hydrological data combination conditions), breakwater concrete pouring operations needed to be suspended (construction progress data) to prevent wave and current impacts that could affect construction quality. Furthermore, five consecutive days of high waves (multi-dimensional hydrology) led to increased sediment scour in the waters around the bridge piers (water area change data), prompting adjustments to the construction plan and earlier implementation of scour prevention and protection measures (construction progress data changes). By establishing this spatiotemporal correlation between the data, the engineering team was able to predict the impact of the hydrological environment on the construction progress and adjust the construction plan in a timely manner. For example, by concentrating pile foundation work during low tide periods, this effectively reduced construction delays caused by hydrological conditions and ensured the smooth progress of the project.

[0093] Further optionally, in the above steps, based on the real-time work area image and construction progress data, an image rollback axis for retracing the actual construction situation is constructed in the second BIM, and the actual construction scenes at different times are dynamically displayed through the image rollback axis, including: using a dynamic differential correction algorithm to correct the positioning error in the real-time work area image, so that the real-time building position in the real-time work area image, the building position in the construction progress data, and the standard building position in the water conservancy project construction information are aligned with each other; identifying the construction machinery position, personnel position, construction progress, and equipment deployment progress in the real-time work area image, marking the identification results to the model components in the second BIM, and performing spatiotemporal association with the first space-time cube model to obtain a dynamic construction image in the second BIM; in response to the user's rollback instruction for the image rollback axis, displaying the dynamic construction image at the corresponding time to the user to assist the user in checking the potential risks in the actual construction scene.

[0094] Thus, through error correction, information recognition, and dynamic display technologies, accurate construction scene backtracking and risk prediction are achieved, effectively addressing the shortcomings of traditional BIM models in construction process visualization and risk management. Furthermore, a dynamic differential correction algorithm uses the standard building positions in water conservancy project construction information as a benchmark. By comparing the real-time building positions in the real-time work area image with the building positions in the construction progress data, the spatial deviation between the three is analyzed. This algorithm then corrects the positioning errors in the real-time work area image, ensuring spatial consistency between the image, the BIM model, and the construction progress data. Furthermore, computer vision and image recognition technologies are used to extract key information such as the location of construction machinery, personnel, construction progress, and equipment deployment progress from the corrected real-time work area image and annotate it to the corresponding model components in the second BIM. This information is spatiotemporally correlated with the first space-time cube model, integrating dynamic elements of the construction process into the full lifecycle model, creating a dynamic construction image that incorporates actual construction details. When a user issues a rollback command, the dynamic construction image at the corresponding moment is retrieved based on the timeline, visually presenting the construction scene and helping users quickly identify potential risks.

[0095] Taking a large reservoir dam pouring project as an example, drones regularly collected real-time images of the work area daily during construction, recording operations such as concrete pouring and formwork erection. The construction team also simultaneously updated construction progress data. Using a dynamic differential correction algorithm, the dam pouring height in the image was compared with the construction progress data and the standard height in the design drawings, correcting for image positioning errors caused by changes in the drone's flight altitude. Using image recognition technology, the positions of construction machinery such as concrete pump trucks and cranes, as well as the workers' work areas, were identified from the corrected images and annotated into the secondary BIM model. For example, during a rollback review, the user discovered that the crane's operating area at a certain time conflicted with the design plan and was close to an unreinforced slope, posing a safety hazard. Prompt measures were taken to adjust the crane's position and strengthen slope protection, preventing an accident.

[0096] This approach significantly improves the positioning accuracy of real-time work area images, ensuring accurate representation of construction scenes and avoiding misjudgments due to image errors. Furthermore, image recognition and spatiotemporal correlation technologies integrate dynamic information from the construction process into the BIM model, creating a visual, dynamic image of the construction process that more intuitively reflects the actual construction situation than traditional static models. Using the image scroll bar, users can quickly review construction history and promptly identify potential risks such as unreasonable construction procedures and inadequate safety measures. This improves risk identification efficiency and effectively reduces construction safety hazards and quality issues.

[0097] It is worth noting that the core principle of the dynamic differential correction algorithm is to dynamically calibrate the positioning error of the real-time work area image by establishing a spatiotemporal mapping relationship between multi-source data, thereby ensuring spatial consistency with the BIM model and construction progress data.

[0098] Using the standard building locations in water conservancy project construction information (such as coordinate points in design drawings) as an absolute benchmark, feature points in real-time work area images (such as building corners and equipment outlines) are aligned with location information in construction progress data (such as the coordinates of completed pile foundations) in a spatial coordinate system. For example, using Global Navigation Satellite System (GNSS) control points or ground landmarks, the coordinates of drone-captured images are converted to an engineering coordinate system consistent with the BIM model, eliminating systematic deviations caused by different shooting angles and projection methods.

[0099] Computer vision technology is used to identify targets with stable geometric features (such as bridge piers and gates) in real-time work area images and match them with corresponding components in the BIM model. During construction, the error between the image and the model changes dynamically due to factors such as equipment movement and terrain changes. The algorithm continuously tracks the relative position changes of these feature points, calculating the translation, rotation, and scaling parameters for each frame, thereby extracting dynamic error characteristics. For example, if the position of a bridge pier in the image is found to be offset by 20 cm compared to the BIM model and changes linearly over time, it can be determined that the image positioning system has accumulated errors.

[0100] Based on the extracted dynamic error features, a differential fusion strategy is used to perform real-time corrections to image positioning. For image sequences with adjacent timestamps, an error correction model is established by comparing the positional changes (differential information) of the same feature point and combining it with the actual displacement in the construction progress data (such as the increase in concrete pour height). For example, if an image shows a 5-centimeter increase in dam height over a certain period of time, while the construction progress record shows an 8-centimeter increase, the height displayed in the image is corrected through scaling to align the two. Prediction algorithms such as Kalman filtering are also introduced to smooth the error correction process, reducing correction fluctuations caused by image noise or feature point mismatches.

[0101] During the construction of a certain cross-river bridge, the images of the piers collected daily by drones exhibited a positioning error of approximately 0.5 meters, which increased with increasing shooting distance. A dynamic differential correction algorithm, based on the design coordinates of the piers in the BIM model, identified characteristic points of the pier reinforcement skeleton in the image and calculated their deviation from the designed position. By combining the daily pile foundation sinking depth recorded in the construction progress, the image coordinates were dynamically adjusted to keep the positional error between the corrected images and the corresponding piers in the BIM model within 5 centimeters. This significantly improved the spatial accuracy of construction process backtracking and provided a reliable basis for subsequent quality inspections and safety assessments.

[0102] Furthermore, in the above steps, after identifying the construction machinery position, personnel position, construction progress, and equipment deployment progress in the real-time work area image, the construction machinery position and personnel position can also be compared with the collision rule library to identify the pipeline collision risk area and generate a collision avoidance plan; the collision risk prompt corresponding to the collision avoidance plan and the avoidance scheduling path are displayed in the second BIM.

[0103] First, computer vision algorithms (such as YOLO object detection) extract the 2D positions of construction machinery (such as cranes and excavators) and personnel from real-time worksite images. Combined with the image positioning correction results, these positions are converted into 3D coordinates within the BIM model. A collision rule library pre-generates and stores the spatial locations of pipelines, structures, high-risk areas, and other objects in water conservancy projects, along with safety distance rules (e.g., a horizontal distance of ≥10 meters between machinery and high-pressure pipelines, and a distance of ≥5 meters between personnel and the edge of a deep foundation pit). Once the real-time location data is received, the system uses spatial geometry algorithms (such as bounding box collision detection) to calculate the spatial distance between the machinery / personnel coordinates and the risk objects in the rule library. If the distance is less than a safety threshold, a risk warning is triggered. Subsequently, a path planning algorithm (such as the A* algorithm) is invoked to generate a dispatch path within the traversable area of ​​the BIM model that avoids risk areas. This path must meet constraints such as the machinery's operating radius and the road's load-bearing capacity. Finally, risk warnings (e.g., red highlighted collision areas) and avoidance paths are visually overlaid on the secondary BIM to guide on-site dispatching.

[0104] For example, during a reservoir spillway renovation and expansion project, a crawler crane (positioning coordinates X=123.5, Y=45.8, Z=8.2) was lifting gate components. Its real-time position was identified by drone imagery and imported into a secondary BIM. When the system compared the collision rule library, it discovered that the crane was directly above a water pipeline buried 2 meters deep. The safety distance threshold is ≥3 meters in the vertical direction, but the current distance is only 1.8 meters, triggering a pipeline collision risk warning. The crane retreated 5 meters along the construction road to the right of the spillway, turned upstream and detoured to the gate installation location. The dispatching path was marked with a yellow arrow in the BIM model, and the pipeline risk area was highlighted with a red translucent block. On-site operators received instructions through the BIM mobile terminal and adjusted the machine's position according to the planned path, avoiding the risk of pipeline damage caused by blind operation.

[0105] Through real-time collision detection, the time required to identify dynamic risks (such as potential conflicts between machinery movement and underground pipelines) that are difficult to detect during traditional manual inspections has been reduced from minutes to seconds, improving the accuracy of risk warnings. The automatic generation of collision avoidance plans improves machine scheduling efficiency and reduces energy consumption caused by equipment idling due to improper route planning. Furthermore, visual risk alerts and route guidance reduce decision-making pressure on operators. After implementation in one construction site, the collision accident rate between construction machinery and structures has been significantly reduced, improving construction safety.

[0106] Further optionally, in the above steps, the first space-time cube model and the second space-time cube model are respectively used as three-dimensional layers, and the three-dimensional layers are fused into the first BIM through a geometric semantic dual-driven matching method to construct a second BIM, including: learning the correlation between environmental factors and engineering risks in the area to be monitored through a graph convolutional network, and adjusting the fusion weights of different environmental areas based on the learned correlation; wherein the environmental factors include at least: terrain slope, water flow direction, and vegetation density; using the adjusted fusion weights, based on the point cloud feature extraction method of PointNet++, the first space-time cube model is fused into the first BIM through a geometric semantic dual-driven matching method to construct a second BIM. The empty cube model and the second space-time cube model are respectively matched with the geometric body of the first BIM; and, using a graph embedding algorithm, based on the dynamic relationship map between the water conservancy project building entities, the environmental attributes in the first space-time cube model and the second space-time cube model are respectively mapped to the model components of the first BIM to obtain the second BIM; during the construction stage and / or operation and maintenance stage, based on the multi-dimensional hydrological data, water area change data, construction progress data, and operation and maintenance monitoring data, the water conservancy project building morphology change table is updated, and the second BIM is partially reconstructed based on the water conservancy project building morphology change table.

[0107] The second BIM construction method achieves intelligent association of data throughout the entire life cycle and dynamic evolution of models through the deep integration of graph neural networks and point cloud processing technology. Specifically, a graph convolutional network (GCN) is first used to build an association model between environmental factors and engineering risks. Environmental factors such as terrain slope, water flow direction, and vegetation density are converted into node features in the graph structure. Engineering risks (such as slope landslides and pipeline leakage) are used as edge weights. The dependencies between nodes are learned through graph convolution operations. For example, it is found that in areas with terrain slopes greater than 30° and vegetation density less than 20%, the risk of slope instability caused by water erosion increases by 40%. Based on this, a higher fusion weight is assigned to this area, and data mapping for such high-risk areas is strengthened during model fusion.

[0108] Secondly, the PointNet++ algorithm was used to perform feature matching between the space-time cube model and the geometry of the first BIM. The point cloud data of the first space-time cube model (including construction progress and environmental data) and the second space-time cube model (including operation and maintenance data and environmental data) were grouped and extracted through multiple levels of hierarchy. Geometric features at different scales (such as dam surface texture and channel slope profile) were captured and then point-to-point matched with the design geometry in the first BIM, ensuring precise alignment of the space-time data with the model's spatial location.

[0109] Then, using graph embedding algorithms (such as Node2Vec), environmental attributes are mapped to BIM model components. A dynamic relationship graph of water conservancy project entities (such as gates and pumping stations) is constructed, with nodes representing project components and edges representing functional relationships between components (such as the causal relationship between gate opening and closing and downstream water level changes). Through graph embedding, environmental attributes in the space-time cube model (such as flow velocity and water quality indicators over a certain period of time) are converted into low-dimensional vectors and mapped to the attribute space of the corresponding components, achieving a semantic association between "environmental change, component status, and project risk."

[0110] Specifically, the graph embedding algorithm achieves precise association of complex data in the mapping between BIM models and environmental attributes by constructing a knowledge graph, vector representation, weight calculation, multi-dimensional mapping, and dynamic updating. First, physical entities such as dams and gates in the BIM model are abstracted as graph structure nodes, with relationships such as water flow direction and structural support as edges. At the same time, environmental attributes such as terrain slope and vegetation coverage are connected as dynamic attributes or independent nodes to form a multimodal knowledge network. Next, algorithms such as Node2Vec and GraphSAGE are used to convert spatial attributes, time series attributes, and semantic attributes into low-dimensional vectors. Through random walk sampling and neural network training, similar environmental features are clustered in the vector space. Then, the graph attention network (GAT) is introduced to dynamically calculate the impact weight of each BIM component based on the strength of the association between environmental attributes and project risks, such as assigning a higher weight to water flow velocity for riverbank gates.

[0111] During the multidimensional mapping phase, spatial alignment of environmental nodes and BIM components is achieved based on geographic coordinates. Environmental attributes are associated with component functions according to engineering knowledge rules, and environmental status is converted into risk indicators using a predefined risk matrix. Finally, using incremental graph embedding technology, newly incoming monitoring data, such as water level changes and equipment status, is processed in real time, rapidly updating node vectors and triggering adjustments to local BIM model parameters. Attention weights are then optimized through reinforcement learning.

[0112] Taking a cross-basin water diversion project as an example, a knowledge graph was constructed that included engineering nodes such as the water transfer tunnel and bend revetment, as well as environmental nodes such as flow velocity and sediment concentration. The flow velocity and sediment concentration at the bend revetment were mapped into a 128-dimensional environmental vector. GAT calculated the impact of this vector on revetment scour and wear as 0.85, and then incorporated the scour rate into the revetment component attributes. When the scour rate exceeded a threshold, an early warning for revetment reinforcement was automatically triggered, enabling the engineering team to identify abnormal scour risks in advance and avoid embankment breaches.

[0113] Finally, during the construction or operation and maintenance phase, the "Water Conservancy Project Architectural Morphology Change Table" is updated based on real-time monitoring data to record changes in component dimensions (such as height changes caused by dam settlement) and material property changes (such as concrete strength attenuation), triggering the model's local morphology reconstruction algorithm to automatically adjust the geometric parameters and topological relationships of related components in BIM to maintain consistency between the model and the actual project.

[0114] For example, in a reservoir reinforcement project in a mountainous area, graph convolutional network analysis found that the terrain slope on the left bank of the reservoir reached 35°, the vegetation coverage rate was only 15%, and it was located in the water bend erosion area. It was determined that the risk level of landslides on the slopes in this area during heavy rain was high. Therefore, when fusing the space-time cube model, this area was assigned a fusion weight 1.5 times that of other areas to strengthen the integration of construction monitoring data in this area (such as slope displacement meter data).

[0115] The PointNet++ algorithm matched the slope construction point cloud data captured by the drone with the designed slope model in the first BIM. It identified that the actual excavation contour was offset by 0.8 meters from the design. The system automatically marked this area, triggered local morphological reconstruction, and updated the slope geometry parameters in the BIM. Simultaneously, a graph embedding algorithm mapped the slope area's "slope-vegetation-water flow" relationship to the slope component attributes in the BIM. When subsequent monitoring detected that the soil moisture content in this area exceeded a threshold, the model automatically issued a landslide risk warning and, combined with historical construction data, generated a reinforcement plan (such as increasing anchor density).

[0116] This improves the environmental adaptability of the second BIM model, increases data fusion accuracy in risk areas to the centimeter level, and reduces the latency of dynamic model updates. In a river regulation project, a graph convolutional network (GCN) pre-identified five high-risk scour points that were often overlooked by traditional methods. Combined with real-time matching using PointNet++, this prevented embankment collapses caused by design and construction discrepancies. The local morphology reconstruction function improves the model's response to project deformation, increases the accuracy of equipment failure prediction during the operation and maintenance phase, and reduces unplanned downtime.

[0117] Furthermore, a Graph Attention Network (GAT) can be introduced to enhance the ability to weight complex environmental factors, for example, distinguishing the differential impact of different rainfall intensities on slope risk. This can be combined with a Generative Adversarial Network (GAN) to simulate changes in engineering morphology under extreme environments, preemptively verifying the robustness of model reconstruction.

[0118] Furthermore, we are exploring deep integration with digital twin technology, enabling a fully closed-loop management system from monitoring to analysis to control through real-time, two-way interaction between physical entities and virtual models. For example, environmental factors can be expanded to include ecological indicators (such as the biodiversity index), building a multi-dimensional intelligent BIM system encompassing engineering, environmental, and ecological aspects.

[0119] As an optional embodiment, in step S106, the structural change trend and potential risk factors of the water conservancy project in the second BIM are predicted by combining the equipment characteristic curve and the water conservancy project physical model with the multi-dimensional hydrological data and water area change data of the monitored area, and the structural change trend and potential risk factors of the water conservancy project are integrated into the second BIM for real-time display to the user, including: identifying key building components and component types in the second BIM; using the structural stiffness matrix of the component type as a physical constraint, training a neural network to predict the stress field distribution of key building components through the equipment characteristic curve and the water conservancy project physical model to obtain a physical information neural hybrid model; using physical information neural hybrid model to predict the stress field distribution of key building components ... The information neural hybrid model, combined with multi-dimensional hydrological data and water area change data, predicts the stress field distribution probability of key building components in the future period; the stress field distribution probability prediction value is converted into the water conservancy equipment failure knowledge graph using the knowledge graph embedding algorithm to obtain the risk transmission map of key building components; among them, the water conservancy equipment failure knowledge graph is constructed based on the type of water conservancy building failure, the number of equipment opening and closing times, the vibration spectrum, and the seepage pressure fluctuation data; through the multimodal visualization engine, the stress field distribution probability and the risk transmission map are mapped to the semantic expression space of the second BIM to obtain a model legend for indicating the changes in building structure strength and potential risk factors.

[0120] Specifically, in step S106, key building components (such as dam heels, sluice piers, and water pipeline elbows) are first identified based on the second BIM model. Based on component type (gravity dam, steel structure, concrete structure, etc.), their structural stiffness matrices (such as elastic modulus, Poisson's ratio, and section moment of inertia) are extracted and embedded as physical constraints in the neural network training process. Equipment characteristic curves (such as pump head-power curves and gate opening and closing force-displacement curves) and hydraulic engineering physics models (such as elastic equilibrium equations and seepage continuity equations) provide prior knowledge, constraining the stress field distribution predicted by the neural network to conform to basic physical laws (such as stress equilibrium and deformation coordination).

[0121] For example, the elastic mechanics equilibrium equation is embedded in the neural network training: the equilibrium coefficient is , the data fitting loss is , physical constraint loss: is the predicted stress field. To predict the stress field and observed stress fields exist The mean square error at each sampling point. To predict the divergence of the stress field and body force exist The mean square error at each sampling point. is the number of observed stress data points, The larger it is, the higher the statistical reliability of the data fit, but the computational cost also increases. For location The observed stress value at . Meaning is the number of physical constraint sampling points, The larger it is, the more comprehensive the spatial coverage of physical constraints is, but the computational efficiency is reduced. The meaning is the position predicted by the neural network The stress tensor at can be directly predicted by the neural network output layer or derived from the predicted displacement field. The meaning of is the divergence of the stress field (vector), which represents the resultant force per unit volume. The meaning of location The volume force vector at . It can be viewed as a regularization parameter that balances the model's sensitivity to data noise and the strictness of the physical laws.

[0122] The trained physical-information neural hybrid model combines real-time multidimensional hydrological data (water level, flow velocity, water temperature, and sediment content) with water area change data (riverbed scour depth, bank displacement, and ice slab movement trajectories) to output the stress field distribution probability of key components in future time periods (e.g., the next 24 hours, rainy season cycle), quantifying the likelihood of stress exceedance (exceeding material design strength) at different locations. Subsequently, using a knowledge graph embedding algorithm, the predicted stress probability values ​​are semantically linked to the hydraulic equipment failure knowledge graph (integrating historical failure types, equipment operating parameters, and environmental response data). This generates a risk transmission map from "stress anomaly to material fatigue to structural damage to failure occurrence," clearly illustrating the risk diffusion path from local components to the overall structure. Finally, a multimodal visualization engine maps information such as stress distribution heat maps, risk transmission arrows, and failure probability annotations into the three-dimensional space of the BIM model, creating an intuitive and easy-to-understand dynamic legend that provides real-time insights into structural strength changes and potential risks.

[0123] For example, in a sluice gate project on a plain river network, three high-risk pier components (the middle pier, which is subject to long-term erosion) were first identified. Their structural stiffness matrices were constructed based on concrete material parameters (elastic modulus 30 GPa, compressive strength 30 MPa) and pier geometry. A physical-informed neural hybrid model was trained using historical monitoring data (pier strain gauge data, upstream and downstream water level differences, and gate opening and closing frequency) to predict the pier stress distribution under three days of continuous heavy rain. The model output showed a 75% probability of exceeding the tensile stress limit at the interface between the bottom of the middle pier and the foundation, exceeding the design tensile strength of C30 concrete (1.43 MPa).

[0124] Through a knowledge graph embedding algorithm, this stress anomaly was linked to the "concrete pier crack" fault node, and the risk transmission path was deduced: tensile stress exceeds the limit → concrete microcrack expansion → accelerated steel corrosion → crack penetration and leakage. The multimodal visualization engine marked high-stress areas in the BIM model with red translucent blocks and dynamically displayed the direction of possible crack expansion with orange arrows. A risk warning ("Stress exceeds the limit at the bottom of the middle pier. It is recommended to immediately lower the gate opening height and intensify seepage monitoring") was also displayed. The engineering management team adjusted the scheduling plan accordingly, reducing the gate opening from 80% to 50%. After 24 hours, the stress monitoring value returned to a safe range, preventing the pier cracking accident caused by stress concentration.

[0125] In another example, full-factor data is first extracted from the second BIM model, including the geometric parameters of all building components (dimensions, coordinates, material type), the structural stiffness matrix from the design drawings (such as elastic modulus, Poisson's ratio, and shear strength), and equipment characteristic curves (such as the load-stroke relationship of the gate hoist and the efficiency-flow curve of the pump). Using the structural stiffness matrices of key components as physical constraints, the equipment characteristic curves and the physical model of the hydraulic engineering project (such as the equilibrium equations of elasticity and Darcy's law of seepage) are converted into constraints in the neural network's loss function. For example, when training a neural network to predict dam foundation stress, in addition to minimizing the error between the predicted values ​​and historical monitored values, an additional physical constraint requiring the predicted stress field to satisfy the equilibrium equations is added to ensure that the model output conforms to basic mechanical principles. The network parameters are optimized using a gradient descent algorithm, ensuring that the model fits the historical data while maintaining physical consistency. Training data includes monitoring data under historical operating conditions (such as dam body strain distribution at different water levels) and simulated data under design conditions (such as finite element analysis results under extreme floods). This creates a hybrid training set that combines measured and simulated data, improving the model's generalization capabilities under complex operating conditions. Furthermore, real-time multidimensional hydrological data (e.g., current upstream water level of 120 m, flow velocity of 2.5 m / s) and water area change data (e.g., recent riverbed scour resulting in a localized 0.5 m reduction in the dam foundation depth) are fed into a trained physical-informed neural hybrid model. Combined with the current geometric state of the components (e.g., changes in dam height due to settlement), the model outputs the probability of stress field distribution for key components in the future (e.g., the next 24 hours, the next flood cycle). This probability reflects the likelihood that stress at each component location will exceed the material design strength under specific operating conditions (e.g., an 85% probability of exceeding the tensile stress limit at the dam heel), rather than a single, fixed value. This provides a quantitative basis for risk assessment. The model supports parallel predictions for multiple operating conditions, such as simultaneously calculating stress evolution trends under different scenarios, including normal operation, heavy rain warnings, and equipment failures.

[0126] Using knowledge graph embedding algorithms (such as TransE and ComplEx), the predicted stress field distribution probability values ​​are mapped to a pre-built knowledge graph for hydraulic equipment failures. This graph, based on historical failure data, includes nodes such as failure type (cracks, leakage, deformation), inducing factors (load overrun, material aging, environmental erosion), equipment operating parameters (number of starts and stops, load amplitude), and monitoring indicators (vibration frequency, seepage pressure fluctuation). It also includes directed edges such as "stress overrun → material fatigue" and "crack propagation → leakage occurrence." For example, when the model predicts a 70% probability of tensile stress overrun at the bottom of a pier, the algorithm automatically links it to the "concrete pier crack" node in the knowledge graph and generates a risk transmission map for that component, following the transmission path from "stress overrun → microcrack initiation → crack propagation → structural leakage." This map annotates the risk probability and impact range of each link (for example, the probability of abnormalities at seepage monitoring points within 50 meters downstream increases by 40%).

[0127] In particular, the multimodal visualization engine maps stress field distribution probabilities (e.g., heat maps, with red indicating high-risk areas), risk transmission pathways (arrows indicating the direction of risk diffusion), and fault-related information (historical similar failure cases and treatment measures) into the three-dimensional space of the secondary BIM. Component areas with a probability of stress exceeding 50% or higher are highlighted in different colors within the BIM model; users can click to view the specific stress value, excess multiple, and material strength margin. Risk transmission pathways (e.g., seepage diffusion trajectory from the bottom of the pier to the upstream blanket) are displayed with animated arrows, with the thickness of the arrows reflecting the intensity of risk transmission. When the mouse hovers over a high-risk area, a comprehensive information box pops up containing information such as "current stress distribution, evolution trend over the next three days, recommended monitoring frequency, and historical similar failure treatment plans." The risk legend supports 3D model browsing on computers, lightweight viewing on mobile devices, and centralized display on large-screen visualization systems, meeting management needs in different scenarios.

[0128] Hybrid modeling, combining physical constraints and data-driven analysis, avoids the physical inexplicability of pure data models, reduces stress prediction errors compared to simplified traditional finite element analysis methods, and improves the accuracy of over-limit risk identification under complex working conditions. Risk transmission maps can predict potential structural damage (such as cracks in dam foundations and cracks in pipeline welds) in advance, extending the time window for early risk detection several times compared to manual inspections or regular monitoring. Multimodal graphics transform abstract stress data into intuitive three-dimensional risk maps, allowing managers to quickly locate high-risk components and view associated failure cases and recommended solutions, improving decision-making efficiency. The model automatically records component stress evolution histories, fault correlations, and solution outcomes, creating a digital twin archive that provides data support for equipment operation and maintenance strategies (such as maintenance cycle adjustments and material durability assessments).

[0129] As an optional embodiment, in step S107, based on the intelligent agent cluster of the PPO algorithm, each intelligent agent corresponds to a water conservancy project management target; management strategy collaboration is achieved through centralized training and decentralized execution of CTDE; in response to the user's management operation on the second BIM, the corresponding target intelligent agent is called to parallelly deduce the operation status of the water conservancy project under different scheduling schemes in the target management strategy in a high-fidelity physical simulation environment, and the implementation effect of the water conservancy project under the target management strategy is obtained; the SHAP algorithm is used to quantify the contribution of each scheduling scheme and the actual status of the project to the implementation effect, and generate visual management improvement suggestions.

[0130] Specifically, a cluster of intelligent agents is constructed based on the Proximal Policy Optimization (PPO) algorithm, with each agent corresponding to a specific water conservancy project management objective (e.g., flood control safety, optimal water resource allocation, and minimized equipment energy consumption). Through the centralized training decentralized execution (CTDE) framework, the agents share global environmental states (e.g., basin rainfall distribution, reservoir water levels, and equipment operating status) during the training phase to collaboratively optimize management strategies. During the execution phase, each agent makes independent decisions based on local observations, achieving multi-objective collaborative control. For example, during training, the flood control agent and the irrigation agent jointly optimize reservoir scheduling rules to ensure that flood control storage capacity requirements are met while preserving as much irrigation water as possible when a flood occurs. When a user initiates a management operation in the secondary BIM (e.g., adjusting the scheduling rules for a reservoir gate), the system calls the corresponding target agent to concurrently evaluate different scheduling scenarios within a high-fidelity physical simulation environment. The simulation environment, built on physical models of hydraulic engineering (such as the Saint-Venant equations and hydrodynamic models), accurately simulates dynamic processes such as flow evolution, silt accumulation, and equipment stresses. It outputs project operational indicators (such as downstream water level rise, irrigated area, and equipment wear rate) for each scenario. The SHAP (SHapley Additive exPlanations) algorithm is used to analyze simulation results, quantifying the contribution of each scheduling scenario parameter (such as gate opening time and flow threshold) and the actual project status (such as current reservoir storage and river flow capacity) to implementation effectiveness. For example, it was found that in the "8-meter gate opening height" scenario, 60% of the downstream water level control effect came from the current river siltation level, and 30% came from the gate opening and closing speed. This in turn generates visual management improvement recommendations (such as prioritizing silt removal to improve flow capacity before adjusting gate scheduling).

[0131] In the embodiments of the present application, efficient data integration and visualization are achieved, and multi-source heterogeneous data are unified in the BIM system, so that information at each stage of the project can be viewed intuitively, improving decision-making efficiency. By predicting structural change trends and potential risks in advance, the warning time is greatly extended and the probability of engineering accidents is reduced. The entire life cycle management process is optimized, reducing changes in the design phase, controlling progress in the construction phase, and reducing costs and equipment failures in the operation and maintenance phase. It provides powerful decision-making support, and through visual simulation of the effects of different management strategies, it assists in making scientific decisions on water conservancy projects and effectively improves the management level of water conservancy projects.

[0132] See also Figure 2 , Figure 2 A BIM-based digital management system for water conservancy projects is provided in an embodiment of the present application. The BIM-based digital management system for water conservancy projects includes the following modules: an acquisition unit for deploying a multi-dimensional monitoring system for the monitored area where the water conservancy project is located, based on the terrain data of the monitored area and the water conservancy project information, so as to obtain multi-dimensional hydrological data of the monitored area; wherein the terrain data of the monitored area includes at least: terrain elevation, slope, geological structure, and surface vegetation coverage; real-time work area images of the monitored area are obtained by drones; water area change data of the monitored area are obtained by satellite remote sensing; a construction unit is used to construct a first building information model based on the terrain data of the monitored area, real-time work area images, water conservancy project construction information in the design stage, and equipment deployment information. The invention relates to a BIM-based water conservancy project management system, wherein the first BIM is used to integrate multidimensional hydrological data, water area change data, construction progress data during the construction phase, and operation and maintenance monitoring data during the operation and maintenance phase into the first BIM to construct a second BIM covering the entire life cycle; a prediction unit is used to predict the structural change trend and potential risk factors of the water conservancy project in the second BIM by combining the multidimensional hydrological data and water area change data of the monitored area with the equipment characteristic curve and the physical model of the water conservancy project, and integrate the structural change trend and potential risk factors of the water conservancy project into the second BIM for real-time display to the user; an interaction unit is used to predict the implementation effect of the corresponding management strategy after the water conservancy project is implemented in response to the user's management operation on the second BIM, and display the implementation effect and management improvement suggestions in the second BIM to assist the user in realizing digital management of the water conservancy project. In some embodiments, the BIM-based water conservancy project digital management system can be applied to terminal devices. It should be noted that for the convenience and simplicity of description, the specific working process of the BIM-based water conservancy project digital management system described above can refer to the corresponding process in the aforementioned BIM-based water conservancy project digital management method embodiment, and will not be repeated here.

[0133] See also Figure 3 , Figure 3This is a schematic block diagram of the structure of a terminal device provided in an embodiment of the present application. Figure 3 As shown, the terminal device 300 includes a processor 301 and a memory 302, and the processor 301 and the memory 302 are connected via a bus 303, such as I 2 C bus. Specifically, processor 301 is used to provide computing and control capabilities, supporting the operation of the entire terminal device. Processor 301 can be a central processing unit, or it can also be other general-purpose processors, digital signal processors, application-specific integrated circuits, etc. A general-purpose processor can be a microprocessor or any conventional processor. Specifically, memory 302 can be a Flash chip, a read-only memory disk, an optical disk, a USB flash drive, or a removable hard disk.

[0134] Those skilled in the art will understand that Figure 3 The structure shown in is only a block diagram of a part of the structure related to the embodiment of the present application, and does not constitute a limitation on the terminal device to which the embodiment of the present application is applied. The specific server may include more or fewer components than shown in the figure, or combine certain components, or have a different arrangement of components. Among them, the processor is used to run the computer program stored in the memory, and implement any one of the BIM-based digital management methods for water conservancy projects provided in the embodiment of the present application when executing the computer program. It should be noted that those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working process of the terminal device described above can refer to the aforementioned embodiment of the BIM-based digital management method for water conservancy projects, and will not be repeated here.

Claims

1. A BIM-based digital management method for water conservancy projects, characterized by: include: For the monitored area where the water conservancy project is located, a multi-dimensional monitoring system is deployed in the monitored area based on the topographic data of the monitored area and the water conservancy project information to obtain multi-dimensional hydrological data of the monitored area; wherein the topographic data of the monitored area includes at least: terrain elevation, slope, geological structure, and surface vegetation cover; Acquire real-time images of the work area to be monitored through drones; construct the first building information model (BIM) based on the terrain data of the monitored area, real-time images of the work area, and water conservancy project construction information and equipment deployment information during the design phase; Acquire water area change data in the monitored area through satellite remote sensing; integrate multi-dimensional hydrological data, water area change data, construction progress data during the construction phase, and operation and maintenance monitoring data during the operation and maintenance phase into the first BIM to construct a second BIM covering the entire life cycle; By using equipment characteristic curves and the physical model of water conservancy projects, combined with multi-dimensional hydrological data and water area change data of the monitored area, the structural change trends and potential risk factors of water conservancy projects in the second BIM are predicted, and these structural change trends and potential risk factors are integrated into the second BIM and displayed to users in real time. In response to the user's management operations on the second BIM, the implementation effect of the water conservancy project after the corresponding management strategy is implemented is predicted, and the implementation effect and management improvement suggestions are displayed in the second BIM to assist users in realizing digital management of water conservancy projects.

2. The method according to claim 1, characterized in that The multi-dimensional monitoring system is deployed in the monitored area based on the terrain data and water conservancy project information of the monitored area to obtain multi-dimensional hydrological data of the monitored area, including: Based on the terrain elevation, slope, geological structure, and surface vegetation coverage in the terrain data, a three-dimensional terrain model of the area to be monitored is established; Determine the spatial layout and functional zoning of water conservancy projects based on structural design data, equipment deployment, and construction logs contained in water conservancy project information; Based on the spatial layout and functional zoning of the water conservancy project, the three-dimensional building framework of the water conservancy project is superimposed on the three-dimensional terrain model, and the collision between the three-dimensional building framework and the surrounding scenery is detected to obtain a water conservancy project model that includes the surrounding environment and terrain characteristics; Through the hydrological monitoring reasoning chain, key monitoring points in the water conservancy project model are identified, and corresponding monitoring equipment and monitoring ranges are configured for the key monitoring points to obtain a multi-dimensional monitoring system deployment plan; The coverage rate of the monitoring equipment and the monitoring range is tested; if the coverage rate meets the qualified conditions, the corresponding multi-dimensional monitoring system deployment plan is executed, and the multi-dimensional hydrological data of the monitored area is collected through the deployed monitoring equipment.

3. The method according to claim 2, characterized in that Before developing a multi-dimensional monitoring system deployment plan, the following steps are also included: Obtain the natural geographical entities and construction entities in the area to be monitored; the natural geographical entities include at least: topography, geological structure, and vegetation cover; the construction entities include at least: water conservancy project buildings, hydrological monitoring equipment, and management units; Establish spatial correlation, temporal causal chain, and functional correlation between natural geographical entities and construction entities in the monitored area, and obtain dynamic correlation between natural geographical entities and construction entities; The monitoring point site selection rules, equipment configuration rules, and coverage range setting rules are added to the dynamic association relationship through a dynamic monitoring logic reasoning chain to construct a hydrological monitoring reasoning chain with a closed data flow loop; the hydrological monitoring reasoning chain is used to dynamically construct monitoring point addressing logic and equipment deployment logic that are adapted to the area to be monitored.

4. The method according to claim 1, wherein The first building information model (BIM) is constructed based on the terrain data of the area to be monitored, the real-time work area image, the water conservancy project construction information and the equipment deployment information in the design phase, including: Based on the terrain elevation, slope, geological structure, and surface vegetation coverage in the terrain data, a static terrain model is established. Based on the real-time work area images, a dynamic terrain model is established. The terrain background model of the water conservancy project is obtained through multi-dimensional point cloud matching and fusion. Constructing a three-dimensional building model of the water conservancy project based on the water conservancy project construction information, geometrically embedding and matching the three-dimensional building model with the terrain background model, and superimposing and fusing them to obtain a first BIM; Associating the water conservancy project building parameters, equipment parameters, and construction progress data in the design phase with the various model components of the first BIM; A building surface structure model of the water conservancy project is constructed based on the real-time work area image, and the building surface structure model is fused into the first BIM through multi-dimensional point cloud matching. Surface defects of the engineering structure in the building surface structure model are identified and marked in the defect detection layer of the first BIM.

5. The method according to claim 4, characterized in that After geometrically embedding and matching the three-dimensional building model and the terrain background model, and superimposing and fusing them to obtain the first BIM, the method further includes: The Douglas-Peucker algorithm is used to optimize and compress the dynamic trajectories associated with the construction progress and equipment deployment progress in the first BIM; The LOD grading technology is used to geometrically simplify the first BIM to achieve model compression of the first BIM.

6. The method according to claim 1, characterized in that The multi-dimensional hydrological data, water area change data, construction progress data during the construction phase, and operation and maintenance monitoring data during the operation and maintenance phase are integrated into the first BIM to construct a second BIM covering the entire life cycle, including: Performing spatiotemporal alignment and spatiotemporal correlation on multidimensional hydrological data, water area change data, and construction progress data to obtain a first spatiotemporal cube model for demonstrating the correlation between construction progress and actual environmental conditions; The multi-dimensional hydrological data, water area change data, and operation and maintenance monitoring data are temporally aligned and temporally correlated to obtain a second space-time cube model for demonstrating the correlation between operation and maintenance monitoring operations and actual environmental conditions; The first space-time cube model and the second space-time cube model are respectively used as three-dimensional layers, and the three-dimensional layers are integrated into the first BIM through a geometric semantic dual-driven matching method to construct a second BIM. Based on real-time work area images and construction progress data, an image rollback axis is constructed in the second BIM to trace the actual construction situation, and the actual construction scenes at different times are dynamically displayed through the image rollback axis.

7. The method according to claim 6, characterized in that Based on the real-time work area image and construction progress data, an image rollback axis for reviewing the actual construction situation is constructed in the second BIM, and the actual construction scenes at different times are dynamically displayed through the image rollback axis, including: A dynamic differential correction algorithm is used to correct positioning errors in real-time work area images, so that the real-time building positions in the real-time work area images, the building positions in the construction progress data, and the standard building positions in the water conservancy project construction information are aligned with each other; Identify the locations of construction machinery, personnel, construction progress, and equipment deployment progress in the real-time work area image, annotate the identification results to the model components in the second BIM, and perform spatiotemporal association with the first space-time cube model to obtain a dynamic construction image in the second BIM; In response to the user's rollback instruction on the image rollback axis, the dynamic construction image at the corresponding moment is displayed to the user to assist the user in checking the potential risks in the actual construction scene.

8. The method according to claim 7, characterized in that The first space-time cube model and the second space-time cube model are respectively used as three-dimensional layers, and the three-dimensional layers are integrated into the first BIM through a geometric semantic dual-driven matching method to construct a second BIM, including: Through the graph convolutional network, the correlation between environmental factors and engineering risks in the monitored area is learned, and the fusion weights of different environmental areas are adjusted based on the learned correlation. The environmental factors include at least: terrain slope, water flow direction, and vegetation density. Using the adjusted fusion weights and the point cloud feature extraction method based on PointNet++, the first space-time cube model and the second space-time cube model are respectively matched with the geometry of the first BIM; and Using a graph embedding algorithm, based on the dynamic relationship graph between water conservancy project building entities, the environmental attributes in the first space-time cube model and the second space-time cube model are respectively mapped to the model components of the first BIM to obtain the second BIM; During the construction phase and / or operation and maintenance phase, the water conservancy project building morphology change table is updated based on multi-dimensional hydrological data, water area change data, construction progress data, and operation and maintenance monitoring data, and the second BIM is partially reconstructed based on the water conservancy project building morphology change table.

9. The method according to claim 1, characterized in that The method uses the equipment characteristic curve and the water conservancy project physical model, combined with the multi-dimensional hydrological data and water area change data of the monitored area, to predict the water conservancy project structure change trend and potential risk factors in the second BIM, and integrates the water conservancy project structure change trend and potential risk factors into the second BIM for real-time display to users, including: Identify key building components and component types in the second BIM; Using the structural stiffness matrix of component types as physical constraints, the neural network is trained to predict the stress field distribution of key building components through equipment characteristic curves and water conservancy engineering physical models to obtain a physical information neural hybrid model; A physical information neural hybrid model is used, combined with multi-dimensional hydrological data and water area change data, to predict the stress field distribution probability of key building components in the future period; The predicted values ​​of stress field distribution probability are converted into the knowledge graph of hydraulic equipment failure using a knowledge graph embedding algorithm to obtain a risk transmission graph of key building components. The knowledge graph of hydraulic equipment failure is constructed based on hydraulic building failure types, equipment opening and closing times, vibration spectrum, and seepage pressure fluctuation data. Through the multimodal visualization engine, the stress field distribution probability and risk conduction map are mapped into the semantic expression space of the second BIM to obtain a model legend for indicating the changes in building structure strength and potential risk factors.

10. A BIM-based digital management system for water conservancy projects, characterized by: The system includes the following units, The acquisition unit is used to deploy a multi-dimensional monitoring system in the monitored area where the water conservancy project is located, based on the terrain data of the monitored area and the water conservancy project information, to obtain multi-dimensional hydrological data of the monitored area; wherein the terrain data of the monitored area includes at least: terrain elevation, slope, geological structure, and surface vegetation cover; obtain real-time work area images of the monitored area through drones; and obtain water area change data of the monitored area through satellite remote sensing; The construction unit is used to construct a first building information model (BIM) based on the terrain data of the area to be monitored, real-time work area images, water conservancy project construction information and equipment deployment information in the design phase; multi-dimensional hydrological data, water area change data, construction progress data in the construction phase, and operation and maintenance monitoring data in the operation and maintenance phase are integrated into the first BIM to construct a second BIM covering the entire life cycle; The prediction unit is used to predict the structural change trend and potential risk factors of the water conservancy project in the second BIM by combining the equipment characteristic curve and the physical model of the water conservancy project with the multi-dimensional hydrological data and water area change data of the monitored area, and integrate the structural change trend and potential risk factors of the water conservancy project into the second BIM for real-time display to users; The interactive unit is used to respond to the user's management operations on the second BIM, predict the implementation effect of the water conservancy project after the corresponding management strategy is implemented, and display the implementation effect and management improvement suggestions in the second BIM to assist the user in realizing digital management of the water conservancy project.

Images