A BIM-based construction simulation and monitoring method for mechanical and electrical equipment of a hydropower station

Abstract

This invention discloses a BIM-based method for simulating and monitoring the construction of electromechanical equipment in hydropower stations, comprising: Step 1: Constructing a three-dimensional perception network of air, ground, and water to achieve real-time, high-precision acquisition of all elements of the construction environment and dynamic digital twin mapping; Step 2: Developing a dynamic inference engine based on BIM and GAN for three factors: progress, resources, and environment, generating a 7-day progress path and automatically identifying critical path conflicts, and dynamically adjusting construction parameters; Step 3: Constructing a three-level early warning system based on a rule engine, deep learning, and knowledge graph to achieve basic early warning, dynamic threshold optimization, and root cause intelligent diagnosis; Step 4: Developing an AR-5G real-time interactive system to overlay BIM models and on-site images with real-time annotation of early warning information, supporting remote consultation and mobile push notifications; Step 5: Introducing blockchain and smart contracts to store the entire early warning process on the blockchain, automatically execute corrections, and iterate algorithm parameters.

Description

Technical Field

[0001] This invention relates to the field of hydropower station electromechanical equipment construction technology, and in particular to a BIM-based method for simulating and monitoring the construction of hydropower station electromechanical equipment. Background Technology

[0002] As a core strategic infrastructure of the clean energy system, hydropower stations are mostly located in canyon and river areas with high mountains and deep valleys and complex geological conditions. The construction of their electromechanical equipment involves multiple dimensions of cross-operations, such as high-precision equipment installation, temperature-controlled pouring of large-volume concrete, excavation and support of underground cavern groups, and deformation control of metal structure welding. This places extremely high demands on construction precision control, multi-professional collaboration efficiency, and safety management throughout the entire life cycle.

[0003] However, traditional construction simulation systems have long suffered from systemic shortcomings. They rely on location data collection from a single type of sensor, have rigid early warning rules that lack dynamic adaptability, and over-reliance on human experience for deviation judgment, leading to frequent problems such as broken data chains, rigid progress projections, and delayed early warning responses. In the traditional construction simulation of a certain hydropower station, the large-volume concrete pouring module was not fully coupled with the meteorological-geological dual-factor dynamic projection engine, resulting in deviations in the calculation of concrete curing time during the rainy season. Therefore, a BIM-based method for construction simulation and monitoring of electromechanical equipment in hydropower stations is proposed. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a BIM-based method for simulating and monitoring the construction of electromechanical equipment in hydropower stations.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: A BIM-based method for construction simulation and monitoring of electromechanical equipment in hydropower stations includes: Step 1: Construct a three-dimensional perception network of "air-ground-water" through quantum inertial navigation sensors, underwater sonar scanning and UAV laser point cloud technology to carry out real-time high-precision acquisition and dynamic digital twin mapping of all elements of the construction environment (spatial positioning, equipment status, environmental parameters), with an accuracy of 0.1mm and resistance to electromagnetic interference, ensuring the synchronization of data across the entire construction environment. Step 2: Based on the BIM model and generative adversarial network, develop a three-factor dynamic simulation engine for "schedule-resource-environment". Bind the construction schedule plan with the attributes of BIM components, generate the schedule path for the next 7 days and automatically identify critical path conflicts. At the same time, combine AI algorithms to calculate resource consumption and meteorological-geological conditions in real time, dynamically adjust construction parameters, and achieve adaptive optimization of schedule simulation. Step 3: Construct a three-level early warning system of "rule engine + deep learning + knowledge graph". Basic early warnings are triggered by fixed threshold rules, and the early warning thresholds are dynamically adjusted by the Transformer attention mechanism to reduce false alarms and missed alarms. Based on the knowledge graph, multi-dimensional data is associated to realize intelligent diagnosis of the root cause of the fault (such as locating whether the excessive vibration of the positioning equipment is due to installation deviation or foundation settlement), thereby improving the accuracy of early warning and diagnosis. Step 4: Develop an AR-5G real-time interactive system. By overlaying the BIM model and the on-site image on AR glasses, the system can provide real-time warnings on equipment installation deviations and progress delays to guide workers' operations. It supports 5G remote consultation, enabling experts to intervene in the annotation of problems in real time via multiple terminals. Warning information is pushed to mobile devices via 5G in seconds, supporting voice broadcasting and image annotation to ensure the high efficiency of on-site and remote collaborative correction. Step 5: Introduce blockchain + smart contract technology to store the entire decision-making process, including early warning triggering, cause analysis, and corrective action plan formulation, on the blockchain to ensure immutability and traceability; smart contracts automatically execute corrective action plans (such as automatic shutdown when exceeding limits) and regularly optimize early warning algorithm parameters based on blockchain data to achieve system self-iteration and continuous improvement, avoiding the early warning-processing cycle problem.

[0006] The above technical solution further includes: Furthermore, the construction of the air-ground-water three-dimensional sensing network for collecting and dynamically mapping all elements of the construction environment includes the following steps: By deploying quantum inertial navigation sensors on construction equipment and key structural nodes, high-precision positioning with electromagnetic interference resistance is achieved through quantum precision measurement technology, and spatial data such as equipment installation deviation and structural displacement are captured in real time to ensure accurate perception of the spatial status of construction elements. For the underwater foundations of hydropower stations and the concealed works of water diversion tunnels, a multi-beam sonar array is used to penetrate turbid water bodies and scan data such as equipment foundation positioning, concrete pouring quality, and underwater structural defects in real time to generate underwater three-dimensional point cloud models, make up for the blind spots of manual underwater environmental detection, and ensure the holographic acquisition of underwater construction elements. The drone equipped with high-precision lidar flies along a preset flight path to dynamically scan the mountain slope and surface construction area, capturing data such as terrain subsidence, landslide risk, and surface flatness of concrete pouring in real time. Combined with AI algorithms, it automatically identifies changes in the construction environment (such as excessive slope displacement) and provides surface and aerial environmental parameters for progress prediction and early warning. By using edge computing nodes to perform real-time cleaning, registration, and fusion of three types of data, a full-domain digital twin foundation covering "air-ground-water" is constructed. This foundation is linked to the BIM model in real time, dynamically mapping all elements of the construction environment (spatial positioning, equipment status, and environmental parameters) into the digital model to form an interactive 3D visualization environment. This enables real-time synchronization and updating of the physical construction scene and the digital model, providing high-precision and high-reliability data support for subsequent progress simulation, multimodal early warning, and AR interaction.

[0007] Furthermore, the development of the three-factor dynamic inference engine based on BIM model and generative adversarial network includes the following steps: The WBS (Work Breakdown Structure) of hydropower station construction is precisely linked to the specific component attributes in the BIM model, such as equipment size, installation location, and concrete grade. For example, the installation task of the turbine generator unit will be bound to the corresponding equipment model in the BIM, so that each construction step has a clear physical correspondence in the digital model, providing a structured data base for subsequent dynamic simulation. Based on historical construction data, GAN generates multiple possible progress paths for the next 7 days through generative adversarial learning. These paths simulate the progress evolution under different construction sequences and resource allocation strategies, automatically identify critical path conflicts, such as the temporal or spatial overlap between equipment installation and concrete pouring, and output the optimal or alternative progress solutions.

[0008] By combining AI algorithms to calculate the consumption requirements of manpower, materials and equipment in real time and comparing them with budget data, a dynamic resource allocation heat map is generated to warn of resource shortages or redundancy risks. At the same time, meteorological and geological data are integrated to automatically adjust construction parameters through environmental models, such as extending concrete curing time during the rainy season and adjusting work periods during high temperatures, so that the progress projection conforms to the actual environmental conditions. The projected schedule path, resource allocation plan, and environmental adaptability adjustment strategy are integrated into a visual result and displayed in real time through a digital twin platform. For example, different colors are used to mark areas of schedule conflict, resource overload nodes, or environmentally sensitive periods to help the construction team intuitively identify risk points and dynamically adjust the construction plan based on the project results, thereby achieving dynamic balance and optimization of the three factors of schedule, resources, and environment.

[0009] Furthermore, the construction of the three-tiered early warning system based on rule engine, deep learning, and knowledge graph includes the following steps: Based on industry norms and safety standards for hydropower station construction, the rule engine has a built-in fixed threshold rule library. When the construction parameters collected in real time trigger the preset threshold, the system automatically generates basic early warning information and pushes it to the mobile terminal of the person in charge, so that the construction behavior complies with the basic norm requirements and forms the first line of defense for safety. Based on the rules engine, the deep learning module analyzes historical construction data (such as false alarm cases of similar projects and the impact of environmental changes on thresholds) through the Transformer attention mechanism, and dynamically adjusts the early warning thresholds. For example, the tolerance for resource allocation deviations can be appropriately relaxed during peak construction periods, while the environmental parameter thresholds are tightened during the rainy season or geologically sensitive periods. This mechanism reduces false alarms and missed alarms caused by fixed thresholds by capturing the dynamic correlation of multi-dimensional data such as construction progress, environmental conditions, and resource allocation, thereby improving the accuracy and adaptability of early warnings and forming a second intelligent optimization defense line. Knowledge graphs construct multi-dimensional data association networks, including equipment parameters, environmental factors, construction logs, and historical failure cases, to intelligently diagnose the root causes of failures. For example, when an equipment vibration exceeding the standard warning is triggered, the knowledge graph can associate information such as equipment installation records, foundation settlement monitoring data, and recent weather changes to automatically infer the main cause of the vibration exceeding the standard and generate a visual diagnostic report containing causal chains, forming a third line of defense for intelligent diagnosis.

[0010] Furthermore, the method of dynamically adjusting the warning threshold using the Transformer attention mechanism includes the following steps: The system collects real-time data on all aspects of construction, including equipment installation accuracy, concrete curing time, ambient temperature and humidity, and resource allocation status. It also integrates multi-dimensional data such as historical construction records, environmental change logs, and false alarm cases from similar projects to form a dynamic data pool as the basis for analysis. Based on the attention mechanism of the Transformer architecture, weights are assigned to multi-dimensional data—focusing on key parameters that are strongly related to the current construction progress and environmental conditions (such as focusing on concrete curing time and humidity data during the rainy season, and focusing on resource allocation deviation analysis during peak construction periods), weakening the interference of non-key parameters, and performing intelligent identification and priority ranking of data correlation. By analyzing the correlation patterns between historical false alarms / missed alarms and construction parameters, the system automatically deduces threshold adjustment logic. For example, during peak construction periods, the tolerance for resource allocation deviations is appropriately relaxed to avoid frequent false alarms; during rainy seasons or geologically sensitive periods, environmental parameter thresholds are tightened to provide early warnings of potential risks. The adjustment logic is based on data-driven adaptive learning rather than fixed rules, enabling the thresholds to dynamically match actual working conditions. The adjusted thresholds are applied in real time to construction parameter monitoring. The system continuously verifies the early warning effect. If the false alarm rate and missed alarm rate decrease after adjustment, the adjustment is confirmed to be effective and included in the threshold library. If the effect does not meet expectations, a secondary optimization mechanism is triggered to further optimize the threshold setting through iterative learning, forming a closed-loop optimization process of "data collection - weight allocation - logical deduction - effect verification" to continuously improve the accuracy and adaptability of early warning.

[0011] Furthermore, the intelligent diagnosis of fault root causes based on knowledge graph-linked multidimensional data includes the following steps: By integrating equipment lifecycle data (such as design parameters, installation records, and maintenance logs), environmental dynamic data (such as temperature and humidity, vibration monitoring values, and geological settlement trends), construction process data (such as schedule execution status and resource allocation status), and a historical fault case library, a knowledge graph network containing entity-attribute-relationship elements is constructed. For example, "excessive equipment vibration" is used as the core entity, and attribute nodes such as "installation deviation records," "foundation settlement monitoring values," and "recent weather changes" are associated to form a multi-dimensional data association network. When the rules engine or deep learning module triggers an alert, the system automatically activates the entity nodes and associated paths related to the alert in the knowledge graph; through the graph query engine, it quickly locates data nodes such as equipment parameters, environmental factors, and construction operations directly related to the alert, forming a preliminary associated dataset; Based on the relationships in the knowledge graph, the system performs multi-step reasoning to trace the root cause of the fault. For example, for the warning of "excessive equipment vibration", the graph can trace backward along the path of "equipment installation deviation → foundation settlement → change in geological conditions", or deduce forward along the path of "fluctuation of environmental temperature and humidity → thermal expansion and contraction of materials → change in structural stress" to identify the potential causal chain that leads to the anomaly. At the same time, the rationality of the root cause hypothesis is further verified by combining the operation records in the construction log. The root cause path derived from reasoning is presented in the form of a visual map, marking key nodes and causal chain weights; at the same time, a decision report containing root cause explanation, impact scope assessment and corrective suggestions is generated to help the construction team quickly locate the root cause of the problem and formulate targeted solutions, realizing closed-loop management from "early warning triggering" to "root cause diagnosis" and then to "corrective execution".

[0012] Furthermore, the method of overlaying the BIM model with the on-site image using AR glasses to provide real-time warning information and guide worker operations includes the following steps: Deploy high-precision AR glasses devices, integrating spatial positioning modules and 5G communication modules; perform spatial positioning calibration of AR glasses by scanning on-site reference points and aligning with BIM model coordinates, so that the spatial position of BIM model and on-site image is accurately matched, providing a spatial reference for subsequent overlay display; The BIM model is lightweighted (e.g., reducing the number of polygons and optimizing texture mapping) to enable smooth rendering in AR glasses; at the same time, the BIM model status is updated in real time based on the construction progress, and the updated model data is transmitted to AR glasses in real time via 5G network to achieve dynamic synchronization between the virtual model and the on-site scene. When the early warning system triggers an alert, the alert information (including alert type, location, and severity) is transmitted to the AR glasses in real time via the 5G network. The AR system overlays the alert information onto the scene in a visual form—for example, marking red warning boxes at equipment deviation locations, displaying progress bar comparisons in areas of lag, and marking warning icons at safety risk points for intuitive prompts. Based on the warning type and construction specifications, the system automatically generates operation guidance information (such as "adjust the equipment installation angle to XX degrees" or "increase manpower input in this area"), and displays it in real time at the corresponding site location in the form of text, arrows, and animations through AR glasses; workers confirm the operation steps through the interactive interface of AR glasses, and the system provides real-time feedback on the operation results, forming a closed-loop operation process of "warning-guidance-execution-feedback". The 5G network supports real-time access from multiple terminals, enabling collaborative interaction between on-site workers and remote experts. Remote experts can view the on-site situation through real-time images on AR glasses, directly marking problem points or providing supplementary operational suggestions on the screen. On-site workers can view the expert's annotations in real time and adjust their operations accordingly, forming a highly efficient guidance model of seamless "on-site-remote" collaboration, thereby improving the accuracy and safety of construction operations.

[0013] Furthermore, the smart contract automatically executes the correction scheme and periodically optimizes the early warning algorithm parameters based on blockchain data, performing self-iteration and continuous improvement, including the following steps: When the early warning system triggers an alert, it automatically packages all decision-making data, including the alert trigger time, trigger conditions, initial diagnostic results, cause analysis process, and corrective action plan formulation records, into encrypted blocks and writes them to the blockchain network through a consensus mechanism. The distributed ledger characteristic of the blockchain ensures that all nodes store the same data, making the decision-making process immutable and traceable, and providing a reliable data foundation for subsequent auditing and optimization. Based on the correction scheme rules stored on the blockchain, smart contracts automatically execute preset operations when trigger conditions are met. For example, when the vibration value of the equipment exceeds the threshold, the smart contract directly sends a stop command to the construction equipment and simultaneously notifies the person in charge. When the progress lags behind the preset threshold, resource allocation is automatically triggered, such as increasing manpower in the area or adjusting the construction sequence, to achieve the immediacy and automation of correction actions and avoid delays caused by human intervention. The system periodically extracts historical early warning data from the blockchain and combines it with dynamic factors such as construction progress and environmental conditions to identify parameter optimization points through statistical analysis or machine learning models. For example, if a certain type of early warning is found to frequently issue false alarms during the rainy season, the system can adjust the early warning threshold or diagnostic logic for that scenario. If the recurrence rate of the problem decreases after the corrective action is implemented, the recommendation weight of that action is strengthened. The optimized parameters are updated to the early warning system through smart contracts, forming a closed-loop iteration of "data accumulation - analysis and optimization - parameter update". The immutability of blockchain supports transparent auditing of the entire decision-making process—regulators or project managers can trace the entire process of early warning triggering, analysis, and correction at any time, verifying the rationality of the decision logic and the effectiveness of its execution. At the same time, the system generates improvement reports regularly based on blockchain data, identifies weak links in the early warning system (such as scenarios with high false alarm rates and inefficient correction schemes), and promotes a new round of parameter optimization or rule adjustment, ensuring that the early warning system continuously improves itself as the construction environment evolves, forming a continuous improvement closed loop of "early warning-execution-optimization".

[0014] The present invention has the following beneficial effects: In this invention, a three-dimensional perception network integrating air, ground, and water is constructed to achieve real-time, high-precision acquisition of all elements of the construction environment and dynamic digital twin mapping. Based on a BIM model and a generative adversarial network, a dynamic simulation engine for the three factors of progress, resources, and environment is developed to generate a 7-day progress path and automatically identify critical path conflicts. Combined with a three-level early warning system, and supplemented by an AR-5G real-time interactive system that overlays BIM models and on-site images to provide real-time annotation of early warning information to guide operations, blockchain and smart contract technologies are introduced to store the entire early warning decision-making process on the blockchain, automatically execute corrective measures using smart contracts, and periodically optimize early warning algorithm parameters based on blockchain data. This forms a closed-loop system of "data acquisition - dynamic simulation - precise early warning - root cause diagnosis - corrective action - self-iteration," effectively solving the problem of insufficient early warning capabilities. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating the steps of a BIM-based construction simulation and monitoring method for electromechanical equipment in a hydropower station, as proposed in this invention. Detailed Implementation

[0016] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Please see Figure 1 As shown, this invention is a BIM-based method for simulating and monitoring the construction of electromechanical equipment in hydropower stations, comprising: A BIM-based method for construction simulation and monitoring of electromechanical equipment in hydropower stations includes: Step 1: Construct a three-dimensional perception network of "air-ground-water" through quantum inertial navigation sensors, underwater sonar scanning and UAV laser point cloud technology to carry out real-time high-precision acquisition and dynamic digital twin mapping of all elements of the construction environment (spatial positioning, equipment status, environmental parameters), with an accuracy of 0.1mm and resistance to electromagnetic interference, ensuring the synchronization of data across the entire construction environment. Step 2: Based on the BIM model and generative adversarial network, develop a three-factor dynamic simulation engine for "schedule-resource-environment". Bind the construction schedule plan with the attributes of BIM components, generate the schedule path for the next 7 days and automatically identify critical path conflicts. At the same time, combine AI algorithms to calculate resource consumption and meteorological-geological conditions in real time, dynamically adjust construction parameters, and achieve adaptive optimization of schedule simulation. Step 3: Construct a three-level early warning system of "rule engine + deep learning + knowledge graph". Basic early warnings are triggered by fixed threshold rules, and the early warning thresholds are dynamically adjusted by the Transformer attention mechanism to reduce false alarms and missed alarms. Based on the knowledge graph, multi-dimensional data is associated to realize intelligent diagnosis of the root cause of the fault (such as locating whether the excessive vibration of the positioning equipment is due to installation deviation or foundation settlement), thereby improving the accuracy of early warning and diagnosis. Step 4: Develop an AR-5G real-time interactive system. By overlaying the BIM model and the on-site image on AR glasses, the system can provide real-time warnings on equipment installation deviations and progress delays to guide workers' operations. It supports 5G remote consultation, enabling experts to intervene in the annotation of problems in real time via multiple terminals. Warning information is pushed to mobile devices via 5G in seconds, supporting voice broadcasting and image annotation to ensure the high efficiency of on-site and remote collaborative correction. Step 5: Introduce blockchain + smart contract technology to store the entire decision-making process, including early warning triggering, cause analysis, and corrective action plan formulation, on the blockchain to ensure immutability and traceability; smart contracts automatically execute corrective action plans (such as automatic shutdown when exceeding limits) and regularly optimize early warning algorithm parameters based on blockchain data to achieve system self-iteration and continuous improvement, avoiding the early warning-processing cycle problem.

[0018] In one embodiment, the construction of an air-ground-water three-dimensional sensing network for collecting and dynamically mapping all elements of the construction environment includes the following steps: By deploying quantum inertial navigation sensors on construction equipment and key structural nodes, high-precision positioning with electromagnetic interference resistance is achieved through quantum precision measurement technology, and spatial data such as equipment installation deviation and structural displacement are captured in real time to ensure accurate perception of the spatial status of construction elements. For the underwater foundations of hydropower stations and the concealed works of water diversion tunnels, a multi-beam sonar array is used to penetrate turbid water bodies and scan data such as equipment foundation positioning, concrete pouring quality, and underwater structural defects in real time to generate underwater three-dimensional point cloud models, make up for the blind spots of manual underwater environmental detection, and ensure the holographic acquisition of underwater construction elements. The drone equipped with high-precision lidar flies along a preset flight path to dynamically scan the mountain slope and surface construction area, capturing data such as terrain subsidence, landslide risk, and surface flatness of concrete pouring in real time. Combined with AI algorithms, it automatically identifies changes in the construction environment (such as excessive slope displacement) and provides surface and aerial environmental parameters for progress prediction and early warning. By using edge computing nodes to perform real-time cleaning, registration, and fusion of three types of data, a full-domain digital twin foundation covering "air-ground-water" is constructed. This foundation is linked to the BIM model in real time, dynamically mapping all elements of the construction environment (spatial positioning, equipment status, and environmental parameters) into the digital model to form an interactive 3D visualization environment. This enables real-time synchronization and updating of the physical construction scene and the digital model, providing high-precision and high-reliability data support for subsequent progress simulation, multimodal early warning, and AR interaction.

[0019] In one embodiment, the BIM model-based generative adversarial network-based dynamic simulation engine for three factors—development progress, resources, and environment—includes the following steps: The WBS (Work Breakdown Structure) of hydropower station construction is precisely linked to the specific component attributes in the BIM model, such as equipment size, installation location, and concrete grade. For example, the installation task of the turbine generator unit will be bound to the corresponding equipment model in the BIM, so that each construction step has a clear physical correspondence in the digital model, providing a structured data base for subsequent dynamic simulation. Based on historical construction data, GAN generates multiple possible progress paths for the next 7 days through generative adversarial learning. These paths simulate the progress evolution under different construction sequences and resource allocation strategies, automatically identify critical path conflicts, such as the temporal or spatial overlap between equipment installation and concrete pouring, and output the optimal or alternative progress solutions.

[0020] By combining AI algorithms to calculate the consumption requirements of manpower, materials and equipment in real time and comparing them with budget data, a dynamic resource allocation heat map is generated to warn of resource shortages or redundancy risks. At the same time, meteorological and geological data are integrated to automatically adjust construction parameters through environmental models, such as extending concrete curing time during the rainy season and adjusting work periods during high temperatures, so that the progress projection conforms to the actual environmental conditions. The projected schedule path, resource allocation plan, and environmental adaptability adjustment strategy are integrated into a visual result and displayed in real time through a digital twin platform. For example, different colors are used to mark areas of schedule conflict, resource overload nodes, or environmentally sensitive periods to help the construction team intuitively identify risk points and dynamically adjust the construction plan based on the project results, thereby achieving dynamic balance and optimization of the three factors of schedule, resources, and environment.

[0021] In one embodiment, the construction of the three-tiered early warning system consisting of a rules engine, deep learning, and a knowledge graph includes the following steps: Based on industry norms and safety standards for hydropower station construction, the rule engine has a built-in fixed threshold rule library. When the construction parameters collected in real time trigger the preset threshold, the system automatically generates basic early warning information and pushes it to the mobile terminal of the person in charge, so that the construction behavior complies with the basic norm requirements and forms the first line of defense for safety. Based on the rules engine, the deep learning module analyzes historical construction data (such as false alarm cases of similar projects and the impact of environmental changes on thresholds) through the Transformer attention mechanism, and dynamically adjusts the early warning thresholds. For example, the tolerance for resource allocation deviations can be appropriately relaxed during peak construction periods, while the environmental parameter thresholds are tightened during the rainy season or geologically sensitive periods. This mechanism reduces false alarms and missed alarms caused by fixed thresholds by capturing the dynamic correlation of multi-dimensional data such as construction progress, environmental conditions, and resource allocation, thereby improving the accuracy and adaptability of early warnings and forming a second intelligent optimization defense line. Knowledge graphs construct multi-dimensional data association networks, including equipment parameters, environmental factors, construction logs, and historical failure cases, to intelligently diagnose the root causes of failures. For example, when an equipment vibration exceeding the standard warning is triggered, the knowledge graph can associate information such as equipment installation records, foundation settlement monitoring data, and recent weather changes to automatically infer the main cause of the vibration exceeding the standard and generate a visual diagnostic report containing causal chains, forming a third line of defense for intelligent diagnosis.

[0022] In one embodiment, dynamically adjusting the warning threshold using the Transformer attention mechanism includes the following steps: The system collects real-time data on all aspects of construction, including equipment installation accuracy, concrete curing time, ambient temperature and humidity, and resource allocation status. It also integrates multi-dimensional data such as historical construction records, environmental change logs, and false alarm cases from similar projects to form a dynamic data pool as the basis for analysis. Based on the attention mechanism of the Transformer architecture, weights are assigned to multi-dimensional data—focusing on key parameters that are strongly related to the current construction progress and environmental conditions (such as focusing on concrete curing time and humidity data during the rainy season, and focusing on resource allocation deviation analysis during peak construction periods), weakening the interference of non-key parameters, and performing intelligent identification and priority ranking of data correlation. By analyzing the correlation patterns between historical false alarms / missed alarms and construction parameters, the system automatically deduces threshold adjustment logic. For example, during peak construction periods, the tolerance for resource allocation deviations is appropriately relaxed to avoid frequent false alarms; during rainy seasons or geologically sensitive periods, environmental parameter thresholds are tightened to provide early warnings of potential risks. The adjustment logic is based on data-driven adaptive learning rather than fixed rules, enabling the thresholds to dynamically match actual working conditions. The adjusted thresholds are applied in real time to construction parameter monitoring. The system continuously verifies the early warning effect. If the false alarm rate and missed alarm rate decrease after adjustment, the adjustment is confirmed to be effective and included in the threshold library. If the effect does not meet expectations, a secondary optimization mechanism is triggered to further optimize the threshold setting through iterative learning, forming a closed-loop optimization process of "data collection - weight allocation - logical deduction - effect verification" to continuously improve the accuracy and adaptability of early warning.

[0023] In one embodiment, the intelligent diagnosis of fault root causes based on knowledge graph-related multidimensional data includes the following steps: By integrating equipment lifecycle data (such as design parameters, installation records, and maintenance logs), environmental dynamic data (such as temperature and humidity, vibration monitoring values, and geological settlement trends), construction process data (such as schedule execution status and resource allocation status), and a historical fault case library, a knowledge graph network containing entity-attribute-relationship elements is constructed. For example, "excessive equipment vibration" is used as the core entity, and attribute nodes such as "installation deviation records," "foundation settlement monitoring values," and "recent weather changes" are associated to form a multi-dimensional data association network. When the rules engine or deep learning module triggers an alert, the system automatically activates the entity nodes and associated paths related to the alert in the knowledge graph; through the graph query engine, it quickly locates data nodes such as equipment parameters, environmental factors, and construction operations directly related to the alert, forming a preliminary associated dataset; Based on the relationships in the knowledge graph, the system performs multi-step reasoning to trace the root cause of the fault. For example, for the warning of "excessive equipment vibration", the graph can trace backward along the path of "equipment installation deviation → foundation settlement → change in geological conditions", or deduce forward along the path of "fluctuation of environmental temperature and humidity → thermal expansion and contraction of materials → change in structural stress" to identify the potential causal chain that leads to the anomaly. At the same time, the rationality of the root cause hypothesis is further verified by combining the operation records in the construction log. The root cause path derived from reasoning is presented in the form of a visual map, marking key nodes and causal chain weights; at the same time, a decision report containing root cause explanation, impact scope assessment and corrective suggestions is generated to help the construction team quickly locate the root cause of the problem and formulate targeted solutions, realizing closed-loop management from "early warning triggering" to "root cause diagnosis" and then to "corrective execution".

[0024] In one embodiment, the step of overlaying the BIM model and the on-site image using AR glasses to provide real-time warning information to guide worker operations includes the following steps: Deploy high-precision AR glasses devices, integrating spatial positioning modules and 5G communication modules; perform spatial positioning calibration of AR glasses by scanning on-site reference points and aligning with BIM model coordinates, so that the spatial position of BIM model and on-site image is accurately matched, providing a spatial reference for subsequent overlay display; The BIM model is lightweighted (e.g., reducing the number of polygons and optimizing texture mapping) to enable smooth rendering in AR glasses; at the same time, the BIM model status is updated in real time based on the construction progress, and the updated model data is transmitted to AR glasses in real time via 5G network to achieve dynamic synchronization between the virtual model and the on-site scene. When the early warning system triggers an alert, the alert information (including alert type, location, and severity) is transmitted to the AR glasses in real time via the 5G network. The AR system overlays the alert information onto the scene in a visual form—for example, marking red warning boxes at equipment deviation locations, displaying progress bar comparisons in areas of lag, and marking warning icons at safety risk points for intuitive prompts. Based on the warning type and construction specifications, the system automatically generates operation guidance information (such as "adjust the equipment installation angle to XX degrees" or "increase manpower input in this area"), and displays it in real time at the corresponding site location in the form of text, arrows, and animations through AR glasses; workers confirm the operation steps through the interactive interface of AR glasses, and the system provides real-time feedback on the operation results, forming a closed-loop operation process of "warning-guidance-execution-feedback". The 5G network supports real-time access from multiple terminals, enabling collaborative interaction between on-site workers and remote experts. Remote experts can view the on-site situation through real-time images on AR glasses, directly marking problem points or providing supplementary operational suggestions on the screen. On-site workers can view the expert's annotations in real time and adjust their operations accordingly, forming a highly efficient guidance model of seamless "on-site-remote" collaboration, thereby improving the accuracy and safety of construction operations.

[0025] In one embodiment, the smart contract automatically executes a correction scheme and periodically optimizes the early warning algorithm parameters based on blockchain data, performing self-iteration and continuous improvement, including the following steps: When the early warning system triggers an alert, it automatically packages all decision-making data, including the alert trigger time, trigger conditions, initial diagnostic results, cause analysis process, and corrective action plan formulation records, into encrypted blocks and writes them to the blockchain network through a consensus mechanism. The distributed ledger characteristic of the blockchain ensures that all nodes store the same data, making the decision-making process immutable and traceable, and providing a reliable data foundation for subsequent auditing and optimization. Based on the correction scheme rules stored on the blockchain, smart contracts automatically execute preset operations when trigger conditions are met. For example, when the vibration value of the equipment exceeds the threshold, the smart contract directly sends a stop command to the construction equipment and simultaneously notifies the person in charge. When the progress lags behind the preset threshold, resource allocation is automatically triggered, such as increasing manpower in the area or adjusting the construction sequence, to achieve the immediacy and automation of correction actions and avoid delays caused by human intervention. The system periodically extracts historical early warning data from the blockchain and combines it with dynamic factors such as construction progress and environmental conditions to identify parameter optimization points through statistical analysis or machine learning models. For example, if a certain type of early warning is found to frequently issue false alarms during the rainy season, the system can adjust the early warning threshold or diagnostic logic for that scenario. If the recurrence rate of the problem decreases after the corrective action is implemented, the recommendation weight of that action is strengthened. The optimized parameters are updated to the early warning system through smart contracts, forming a closed-loop iteration of "data accumulation - analysis and optimization - parameter update". The immutability of blockchain supports transparent auditing of the entire decision-making process—regulators or project managers can trace the entire process of early warning triggering, analysis, and correction at any time, verifying the rationality of the decision logic and the effectiveness of its execution. At the same time, the system generates improvement reports regularly based on blockchain data, identifies weak links in the early warning system (such as scenarios with high false alarm rates and inefficient correction schemes), and promotes a new round of parameter optimization or rule adjustment, ensuring that the early warning system continuously improves itself as the construction environment evolves, forming a continuous improvement closed loop of "early warning-execution-optimization".

[0026] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A BIM-based method for construction simulation and monitoring of electromechanical equipment in hydropower stations, characterized in that, include: Step 1: Construct a three-dimensional perception network of air, ground, and water using quantum inertial navigation sensors, underwater sonar scanning, and UAV laser point cloud technology to collect and dynamically map all elements of the construction environment into a digital twin. Step 2: Based on the BIM model and generative adversarial network, develop a three-factor dynamic simulation engine for progress, resources and environment, bind the construction schedule plan with the attributes of BIM components, generate the progress path for the next 7 days and automatically identify critical path conflicts, and combine AI algorithms to calculate resource consumption and meteorological and geological conditions in real time, and dynamically adjust construction parameters. Step 3: Construct a three-level early warning system consisting of a rule engine, deep learning, and a knowledge graph. Basic early warnings are triggered by fixed threshold rules, and the early warning thresholds are dynamically adjusted using the Transformer attention mechanism. Finally, intelligent diagnosis of the root causes of faults is performed based on multi-dimensional data associated with the knowledge graph. Step 4: Develop an AR-5G real-time interactive system that overlays BIM models and on-site images onto AR glasses, providing real-time annotations and warnings to guide worker operations; supports 5G remote consultations, allowing experts to intervene in annotation issues from multiple terminals in real time. Warning information is pushed to mobile devices, supporting voice broadcast and image annotation; Step 5: Introduce blockchain and smart contract technologies to store the entire decision-making process of early warning triggering, cause analysis, and corrective action plan formulation on the blockchain; the smart contract automatically executes the corrective action plan and regularly optimizes the early warning algorithm parameters based on blockchain data for self-iteration and continuous improvement; The construction of a three-dimensional perception network integrating air, ground, and water, for the collection of comprehensive construction environment data and dynamic digital twin mapping, includes the following steps: Quantum inertial navigation sensors are deployed on construction equipment and key structural nodes. High-precision positioning against electromagnetic interference is achieved through quantum precision measurement technology, and real-time data on equipment installation deviation and structural displacement are captured. For underwater foundations of hydropower stations and hidden projects of water diversion tunnels, multi-beam sonar arrays are used to penetrate turbid water bodies and scan data in real time, including equipment foundation positioning, concrete pouring quality and underwater structural defects, to generate underwater three-dimensional point cloud models. The drone equipped with high-precision lidar flies along a preset flight path to dynamically scan the mountain slope and surface construction area, capturing data in real time including terrain subsidence, landslide risk, and the flatness of concrete pouring surface. Combined with AI algorithms, it automatically identifies changes in the construction environment, providing surface and aerial environmental parameters for progress prediction and early warning. By using edge computing nodes to perform real-time cleaning, registration, and fusion of three types of data, a full-domain digital twin foundation covering "air-ground-water" is constructed. This foundation is linked to the BIM model in real time, dynamically mapping all elements of the construction environment data into the digital model to form an interactive 3D visualization environment. This enables real-time synchronous updates of the physical construction scene and the digital model, providing high-precision and high-reliability data support for subsequent progress simulation, multimodal early warning, and AR interaction.

2. The method for construction simulation and monitoring of electromechanical equipment in a hydropower station based on BIM according to claim 1, characterized in that, The aforementioned BIM model and generative adversarial network-based dynamic simulation engine for three factors—development progress, resources, and environment—includes the following steps: The WBS of hydropower station construction is associated with the specific component attributes in the BIM model, so that each construction step has a clear physical correspondence in the digital model, providing a structured data base for subsequent dynamic simulation. Based on historical construction data, GAN generates multiple possible progress paths for the next 7 days through generative adversarial learning. The paths simulate the progress evolution under different construction sequences and resource allocation strategies, automatically identify critical path conflicts, and output the optimal or alternative progress solutions. By combining AI algorithms to calculate the consumption requirements of manpower, materials, and equipment in real time and comparing them with budget data, a dynamic resource allocation heat map is generated to provide early warning of resource shortages or redundancy risks. At the same time, meteorological and geological data are integrated, and construction parameters are automatically adjusted through environmental models to ensure that the progress simulation conforms to actual environmental conditions. The progress path, resource allocation plan, and environmental adaptability adjustment strategy generated by the simulation are integrated into a visualization result and displayed in real time through a digital twin platform.

3. The method for construction simulation and monitoring of electromechanical equipment in a hydropower station based on BIM according to claim 1, characterized in that, The aforementioned three-tiered early warning system, comprising a rule engine, deep learning, and a knowledge graph, includes the following steps: Based on industry norms and safety standards for hydropower station construction, the rule engine has a built-in fixed threshold rule library. When the construction parameters collected in real time trigger the preset threshold, the system automatically generates basic early warning information and pushes it to the mobile terminal of the person in charge, so that the construction behavior complies with the basic norm requirements and forms the first line of defense for safety. Based on the rules engine, the deep learning module analyzes historical construction data through the Transformer attention mechanism and dynamically adjusts the warning threshold. By capturing the dynamic correlation of multi-dimensional data such as construction progress, environmental conditions, and resource allocation, it reduces false alarms and missed alarms caused by fixed thresholds, improves the accuracy and adaptability of warnings, and forms a second intelligent optimization defense line. The knowledge graph constructs a multi-dimensional data association network of equipment parameters, environmental factors, construction logs, and historical failure cases to conduct intelligent diagnosis of the root causes of failures. When an equipment vibration exceeding the standard warning is triggered, the knowledge graph associates equipment installation records, foundation settlement monitoring data, and recent weather change information to automatically infer the main cause of the vibration exceeding the standard and generate a visual diagnostic report containing causal chains, forming a third line of defense for intelligent diagnosis.

4. The method for construction simulation and monitoring of electromechanical equipment in a hydropower station based on BIM according to claim 1, characterized in that, The method of dynamically adjusting the warning threshold using the Transformer attention mechanism includes the following steps: The system collects real-time data on all aspects of construction, including equipment installation accuracy, concrete curing time, ambient temperature and humidity, and resource allocation status. It also integrates multi-dimensional data from historical construction records, environmental change logs, and false alarm cases from similar projects to form a dynamic data pool as the basis for analysis. Based on the attention mechanism of the Transformer architecture, weights are assigned to multi-dimensional data—focusing on key parameters that are strongly related to the current construction progress and environmental conditions, weakening the interference of non-key parameters, and intelligently identifying and prioritizing data correlations. By analyzing the correlation patterns between historical false alarms, missed alarms and construction parameters, the system automatically deduces the threshold adjustment logic; during peak construction periods, the tolerance for resource allocation deviations is appropriately relaxed to avoid frequent false alarms; during rainy seasons or geologically sensitive periods, environmental parameter thresholds are tightened to provide early warning of potential risks; the adjustment logic is based on data-driven adaptive learning, enabling the thresholds to dynamically match the actual working conditions. The adjusted thresholds are applied in real time to monitor construction parameters. The system continuously verifies the early warning effect. If the false alarm rate and the missed alarm rate decrease after adjustment, the adjustment is confirmed to be effective and included in the threshold library. If the effect does not meet expectations, a secondary optimization mechanism is triggered to further optimize the threshold setting through iterative learning, thereby continuously improving the accuracy and adaptability of the early warning.

5. The method for construction simulation and monitoring of electromechanical equipment in a hydropower station based on BIM according to claim 1, characterized in that, The intelligent diagnosis of root causes of faults based on knowledge graph-linked multidimensional data includes the following steps: By integrating equipment lifecycle data, environmental dynamic data, construction process data, and historical fault case databases, a knowledge graph network containing entity-attribute-relationship data is constructed. When the rule engine or deep learning module triggers an alert, the system automatically activates the entity nodes and associated paths in the knowledge graph related to the alert. Through the graph query engine, the system quickly locates the equipment parameters, environmental factors, and construction operation data nodes directly related to the alert, forming a preliminary associated dataset. Based on the relationships in the knowledge graph, the system performs multi-step reasoning to trace the root cause of the failure. At the same time, it combines the operation records in the construction log to further verify the rationality of the root cause hypothesis. The root cause path derived from the reasoning is presented in the form of a visual graph, with key nodes and causal chain weights marked. Meanwhile, a decision report containing root cause explanations, impact scope assessments, and corrective suggestions is generated to help the construction team quickly locate the root cause of the problem and develop targeted solutions.

6. The method for construction simulation and monitoring of electromechanical equipment in a hydropower station based on BIM according to claim 1, characterized in that, The method of overlaying BIM models and on-site images using AR glasses to provide real-time warning information and guide worker operations includes the following steps: Deploy high-precision AR glasses devices, integrating spatial positioning modules and 5G communication modules; perform spatial positioning calibration of AR glasses by scanning on-site reference points and aligning with BIM model coordinates, so that the spatial position of BIM model and on-site image is accurately matched, providing a spatial reference for subsequent overlay display; The BIM model is lightweighted and rendered smoothly in AR glasses; at the same time, the BIM model status is updated in real time based on the construction progress, and the updated model data is transmitted to AR glasses in real time via 5G network to achieve dynamic synchronization between the virtual model and the on-site scene. When the early warning system triggers an alert, the alert information is transmitted to the AR glasses in real time via the 5G network. The AR system overlays the alert information onto the scene in a visual form, marks the location of equipment deviation with a red warning box, displays a progress bar comparison in areas of lag, and marks warning icons at safety risk points for intuitive prompts. Based on the warning type and construction specifications, the system automatically generates operation guidance information and displays it in real time at the corresponding site location in the form of text, arrows, and animations through AR glasses; workers confirm the operation steps through the interactive interface of AR glasses, and the system provides real-time feedback on the operation results, forming a closed-loop operation process of warning-guidance-execution-feedback. The 5G network supports real-time access from multiple terminals, enabling collaborative interaction between on-site workers and remote experts. Remote experts can view the on-site situation through real-time images on AR glasses, directly marking problem points or providing supplementary operational suggestions on the screen. On-site workers can view the expert's annotations in real time and adjust their operations accordingly, forming an efficient guidance mode for seamless remote collaboration on-site, improving the accuracy and safety of construction operations.

7. The method for construction simulation and monitoring of electromechanical equipment in a hydropower station based on BIM according to claim 1, characterized in that, The smart contract automatically executes the correction scheme and periodically optimizes the early warning algorithm parameters based on blockchain data, performing self-iteration and continuous improvement, including the following steps: When the early warning system triggers an alert, it automatically packages the entire decision-making data, including the alert trigger time, trigger conditions, initial diagnostic results, cause analysis process, and corrective action plan formulation record, into an encrypted block and writes it to the blockchain network through a consensus mechanism. The distributed ledger characteristic of the blockchain ensures that all nodes store the same data, making the decision-making process immutable and traceable, and providing a reliable data foundation for subsequent auditing and optimization. Based on the correction scheme rules stored on the blockchain, the smart contract automatically executes preset operations when the trigger conditions are met; when the equipment vibration value exceeds the threshold, the smart contract directly sends a stop command to the construction equipment and simultaneously notifies the responsible person; when the progress lags behind the preset threshold, it automatically triggers resource allocation adjustment to carry out the correction action in a timely and automated manner. The system periodically extracts historical early warning data from the blockchain and combines it with dynamic factors such as construction progress and environmental conditions to identify parameter optimization points through statistical analysis or machine learning models. If a certain type of early warning is found to frequently issue false alarms during the rainy season, the system adjusts the early warning threshold or diagnostic logic for the scenario. If the recurrence rate of the problem decreases after the corrective action is implemented, the recommendation weight of the corrective action is strengthened. The optimized parameters are updated to the early warning system through smart contracts, forming a closed-loop iteration of data accumulation, analysis and optimization, and parameter updates. The immutability of blockchain supports transparent auditing of the entire decision-making process. Regulators or project managers can trace the entire process of early warning triggering, analysis, and correction at any time to verify the rationality of the decision-making logic and the effectiveness of its execution. At the same time, the system generates improvement reports regularly based on blockchain data to identify weaknesses in the early warning system and promote a new round of parameter optimization or rule adjustment, so that the early warning system can continuously improve itself as the construction environment evolves.

Images