Hydropower station safety integrated management system and method based on digital twinborn scene

Abstract

The invention relates to a hydropower station safety comprehensive management system based on a digital twinborn scene, which designs a hydropower station data resource coding system, establishes a hydropower engineering digital twinborn base, integrates all related geospatial data, basic data, monitoring data and business management data of a hydropower station, and constructs the hydropower station digital twinborn scene. Predicting displacement and deformation conditions of a building structure of the hydropower station and a change trend of important facility and equipment state indexes of the hydropower station, and predicting and pre-judging safety risks therein; the reservoir flow of the hydropower station, the upstream and downstream river section water level, the flow change and the power generation output and the power generation capacity of the hydropower station are forecasted or predicted, corresponding plans are provided for flood early warning or meteorological disaster early warning or water level over-limit early warning, and key links of the plans are simulated in combination with a simulation engine; and performing simulation calculation on different power generation and reservoir scheduling strategies to obtain an optimal scheduling scheme. According to the invention, the intelligent management of hydropower station engineering safety, scheduling safety and power generation safety is realized.

Description

Technical Field

[0001] The present invention belongs to the field of smart hydropower stations, and specifically relates to a comprehensive safety management system for hydropower stations based on digital twin scenarios. Background Art

[0002] Digital Twin, also known as digital mapping and digital mirroring, is a simulation process that makes full use of data such as physical models, sensor updates, and operation history, integrates multi-disciplinary, multi-physical quantities, multi-scales, and multi-probabilities, and completes mapping in virtual space. Digital Twin of Hydropower Station is an intelligent solution using digital technology. By simulating the facilities, equipment, and hydrological data of the hydropower station, it can monitor, optimize, and predict the hydropower station, which is conducive to improving the construction, operation, and safety management level of the hydropower station. However, the existing hydropower station safety management system or method using digital twin technology has a low degree of intelligence and wisdom. For example, the invention application "A Reservoir Safety Lifecycle Management System Based on Digital Twin" with publication number CN116362689A discloses a reservoir safety full lifecycle management system, including data collection, data support, data processing, application support, application management, and visualization modules, which collect, analyze, and process various reservoir data to realize the monitoring and management of all elements of the reservoir. The system disclosed in this invention application focuses on data collection, analysis, processing and monitoring and management of reservoir operations, and mainly provides query and display services. Its application in engineering safety prediction, early warning, rehearsal, and emergency plans is not sufficient, and it fails to combine digital twin scenarios to achieve simulation, simulation, and deduction of safety risks and operating conditions, making it difficult to assist operation and maintenance units in comprehensive safety management and scheduling decisions. Summary of the invention

[0003] The purpose of the present invention is to provide a comprehensive safety management system for hydropower stations based on digital twin scenarios in order to solve the above problems, design a data resource coding system for hydropower stations, establish a digital twin base for hydropower projects, integrate all relevant geographic spatial data, basic data, monitoring data, and business management data of hydropower stations, and construct a digital twin scenario for hydropower stations to provide a three-dimensional visualization carrier for the safety management of hydropower stations; predict the displacement and deformation of the building structure of the hydropower station and the changing trend of the status indicators of the important facilities and equipment of the hydropower station through the AI ​​model, and predict and prejudge the safety risks therein; respectively use the hydrological forecast model, the hydrodynamic model, and the scheduling calculation model Predict or predict the inflow of hydropower stations, water levels of upstream and downstream river sections, flow changes, and power output and power generation of hydropower stations. Provide corresponding plans for flood warnings, meteorological disaster warnings, or water level over-limit warnings. Combined with the simulation engine, simulate the key links of the plan; simulate different power generation and reservoir scheduling strategies to obtain the optimal scheduling plan; combine the digital twin scene of the hydropower station to simulate the dynamic effects of flood evolution, gate discharge, unit power generation, water level rise and fall, and submergence; through the hydropower station safety integrated management system based on the digital twin scene, realize the intelligent management of hydropower station engineering safety, scheduling safety, and power generation safety.

[0004] In order to achieve the above object, the technical solution provided by the present invention is: The comprehensive safety management system for hydropower stations based on digital twin scenarios includes an engineering digital twin base as well as engineering safety subsystems, dispatching safety subsystems and power generation safety subsystems.

[0005] The engineering safety subsystem uses an AI model to predict the deformation of key areas of hydropower station buildings at deformation monitoring points of hydropower station buildings. It uses a finite element analysis model to calculate the displacement and stress distribution of the hydropower station building structure under design conditions, typical operating conditions, extreme conditions, custom conditions or different load combinations, and displays the displacement field and stress field of the hydropower station building structure in the form of a rendered cloud map. Based on the digital twin scenario, abnormal measuring points and risk areas with safety risks are located and marked.

[0006] The dispatching safety subsystem uses a hydrological forecasting model to predict the water inflow process of the hydropower station, calculates the outflow flow, reservoir water level and unit output under different dispatching schemes, compares the calculation results of different dispatching strategies, selects the optimal dispatching scheme, and simulates the dynamic effects of gate discharge, unit power generation and flood storage through the digital twin engine.

[0007] The power generation safety subsystem uses AI models to predict the changing trends of the status indicators of the unit equipment. When safety risks occur in the unit equipment, it locates abnormal measurement points and risky equipment in the digital twin scene, recommends relevant plans, and simulates and rehearses the plan handling process.

[0008] The digital twin base of the project integrates all relevant geospatial data, basic data, monitoring data, business management data and external shared data of the hydropower station; through multi-scale data mapping of time and space, a digital twin scene of the hydropower station is constructed to provide a three-dimensional visualization carrier for various business applications.

[0009] Furthermore, the engineering digital twin base includes a data acquisition module, a data management module, a data model module and a data sharing module.

[0010] The data acquisition module integrates all relevant geospatial data, basic data, monitoring data, business management data and external shared data of the hydropower station according to the hydropower station data resource coding system.

[0011] The data management module implements resource directory management, data quality management, and data security management based on the data standard system.

[0012] The data model module constructs a digital twin scene of a hydropower station that unifies basic data, aggregates monitoring data, and integrates two and three dimensions through multi-scale data mapping in time and space, providing a three-dimensional visualization carrier for various business applications.

[0013] The data sharing module provides an information channel between the hydropower station management department and its superior and subordinate management units and external related units, realizing data sharing across networks, industries and levels.

[0014] Preferably, the hydropower station data resource coding system adopts a three-level coding AB1B2C1C2D1D2.E1E2F1F2G1G2.H1H2H3, the first-level coding AB1B2C1C2D1D2 is the type code of the data resource, where A represents the major category of data resources, and the major categories of data resources include basic data, monitoring data, business management data, cross-industry shared data and geospatial data; B1B2C1C2D1D2 represents the subcategory of data resources, where B1B2 represents the type of monitoring data, and the monitoring data type It includes three categories: engineering safety monitoring, environmental quantity monitoring, and equipment operation monitoring; C1C2 represents the monitoring data subdivision type, which includes deformation, seepage, hydrology, meteorology, water machine equipment operation, electrical equipment operation, and metal structure equipment operation types; D1D2 represents the subtype of the monitoring data subdivision type; the second-level code E1E2F1F2G1G2 is the hierarchical code of the data resource, which is set according to the business level and the subordinate level; the third-level code H1H2H3 is the sequential code of the data resource, which is encoded according to the equipment sequence and the measuring point sequence.

[0015] Furthermore, the engineering safety subsystem includes an engineering safety prediction module, an engineering safety early warning module, an engineering safety rehearsal module and an engineering safety emergency plan module.

[0016] The engineering safety prediction module constructs a gradient boosting regression tree model for deformation prediction of key areas of hydropower station buildings based on the deformation monitoring indicators of hydropower station buildings; renders the prediction results obtained by the gradient boosting regression tree model into the digital twin scene to form a heat map that can reflect the future deformation distribution of key sections of the building, and publishes the rendering results to the WEB for display.

[0017] The engineering safety early warning module issues early warning information when the measured value or predicted value of the monitoring indicator exceeds the threshold, displays the early warning time, related measuring points, causes and data changes, distinguishes the early warning levels with different colors, and locates abnormal measuring points and risk areas in the three-dimensional digital twin model of the engineering structure.

[0018] The engineering safety rehearsal module uses the finite element analysis model of the hydropower station building structure to simulate and calculate the overall displacement and stress-strain of the structure under preset conditions and custom conditions; the user selects the design condition, typical operating condition, extreme condition, custom condition or a custom combination of live load, temperature load and wind load to calculate the distribution of the displacement and stress of the hydropower station building structure, and displays the displacement field and stress field of the hydropower station building structure in the form of a rendered cloud map; the finite element results are rendered on the server side through a visualization engine to generate a corresponding thermal map, and the rendering results are published to the WEB side for display.

[0019] The engineering safety plan module, when the engineering safety monitoring, prediction, and rehearsal results trigger an alarm, searches the built-in plan library to find the plan information applicable to the alarm scenario, and displays the processing methods and steps of the plan; combined with the simulation engine, it simulates the key links of the plan, including material preparation, personnel evacuation, equipment inspection, and dynamic operation process.

[0020] Furthermore, in the engineering safety prediction module, the method of rendering the deformation prediction result of the gradient boosting regression tree model to the digital twin scene includes: 1) Estimate the distribution density of profile measurement points through the coordinate positions of the measurement points; 2) Set the color gradient, map the displacement prediction data to different color bands, and process the color value of each measuring point according to the numerical range; 3) The displacement prediction results are rendered on the server side through a visualization engine to form a heat map that can reflect the future deformation distribution of key sections of the building; 4) Publish the rendering results to the WEB for display, integrate the heat map with the building twin scene, and drag the timeline to view the building deformation trend at different times in the future.

[0021] Furthermore, in the engineering safety rehearsal module, the method of fusing the stress-strain data obtained by the finite element analysis model with the digital twin scene includes: 1. Data preparation; The calculation data of the finite element analysis model includes node information, unit information and physical properties. The node information includes the node number and x, y, z coordinates. The unit information includes the unit number and the node numbers of the constituent units. Each node or unit contains the corresponding calculated physical properties. Before rendering the finite element analysis results, the finite element analysis model calculation result file is converted into a JSON data file, which contains node coordinates, element indexes and result values; 2. Data standardization processing; In order to map the finite element calculation results to different color intervals, the displacement and stress data are standardized to be normalized between (0, 1); 3. Read JSON data; Read the vertex coordinates, triangle index and calculation result data in the JSON file through the visualization engine; 4. Create triangular patches; Reconstruct the node and unit information of the finite element model, and split the triangles and quadrilaterals contained in the mesh into triangular facets; 5. Define color mapping; Map displacement and stress data to color values, set color gradients, specify color bands of physical properties using RGB codes, and process the color value of each node according to the numerical range; 6. Draw triangles and color them; Load the JSON data and triangular patches of the finite element object in the visualization engine, use the previously defined color mapping to render the displacement and stress calculation data on the server side, and generate the corresponding heat map; 7. Achievements display; The rendering results are published to the WEB for display, and the displacement and stress distribution cloud maps in the XYZ directions on the surface of the hydropower station building structure are viewed. The static cloud map is further rendered into a deformation animation for display through the interpolation algorithm.

[0022] Furthermore, the dispatching safety subsystem includes a dispatching safety prediction module, a dispatching safety early warning module, a dispatching safety rehearsal module and a dispatching safety plan module.

[0023] The dispatching safety prediction module takes rainfall forecast as input and predicts the inflow of the hydropower station through the hydrological forecast model; takes the predicted inflow as input and predicts the water level and flow changes of the upstream and downstream river sections through the hydrodynamic model; takes the gate unit control mode as input and predicts the power generation output and power generation of the hydropower station through the dispatching calculation model.

[0024] The dispatching safety warning module compares the real-time water level with the characteristic water level. The characteristic water level includes the flood limit water level, normal water storage level, design flood level, and verification flood level. An alarm is triggered when the characteristic water level limit is exceeded. When warning information of natural disasters such as floods, heavy rains, strong winds, and high temperatures is received, the warning information is displayed in a flashing and highlighted manner in the digital twin scene. The warning information includes the warning level, estimated time, and warning value, and a warning text message is sent to the relevant responsible person.

[0025] The dispatching safety rehearsal module inputs the inflow process, and the user selects real-time, historical, forecast, or preset inflow for calculation, or customizes the inflow for calculation by uploading a template; according to the dispatching regulations or dispatching instructions, the user selects the control method of water level, output, and outflow, and inputs the starting water level and control value; through the dispatching calculation model, the change process of dispatching factors such as outflow, reservoir water level, reservoir capacity, and output is calculated; through the hydrodynamic model, the water level and flow change process of the upstream and downstream river sections of the hydropower station is calculated; the calculation results of the dispatching calculation model are integrated into the digital twin scene to simulate the dynamic effects of flood evolution, gate discharge, unit power generation, water level rise and fall, and submergence; the simulation results of different dispatching strategies are compared, the optimal dispatching plan is selected, and the optimal operation mode of the gate hole and the unit is given.

[0026] When a hydropower station encounters an excessive flood or a related accident, the dispatching safety plan module searches the built-in plan library to find plan information applicable to the scenario and displays the methods and steps for handling dangerous situations of different levels and types. It combines with the simulation engine to simulate the key links of the plan, which includes the dynamic process of dispatching measures, material preparation, flood prevention and rescue, personnel evacuation, equipment inspection, and plugging and repair.

[0027] Furthermore, the power generation safety subsystem includes a power generation safety prediction module, a power generation safety early warning module and a power generation safety plan module.

[0028] The power generation safety prediction module uses a gradient boosting regression tree or an AI model to predict the changing trend of the equipment status indicators of important facilities and equipment of the hydropower station, taking the equipment operating conditions as input.

[0029] The power generation safety early warning module issues early warning information when the status indicators of important equipment in the hydropower station exceed the threshold, reminds users to pay attention in a flashing and highlighted manner, displays the early warning time, related measuring points, data changes, and handling status, distinguishes the early warning levels with different colors, and locates abnormal measuring points in the three-dimensional digital twin model of the facilities and equipment.

[0030] When a hydropower station encounters an alarm event, the power generation safety plan module searches the built-in plan library to find the plan information applicable to the alarm scenario and displays the plan handling methods and steps; combined with the simulation engine, it simulates the key links of the plan, and the plan simulation includes the dynamic process of fault repair, personnel evacuation, and equipment operation.

[0031] Furthermore, in the integrated safety management system of the hydropower station, remote sensing images, oblique photography and BIM modeling data are used to create a digital twin scene, in which the geographical information of the hydropower station and its affected area, civil engineering projects, electrical equipment and facilities, turbines, and metal structures are presented in the form of a digital twin model.

[0032] The engineering safety control method of the hydropower station integrated safety management system based on the digital twin scenario includes the following steps: Step 1: Predict the deformation of the safety monitoring points, conduct a preview of the displacement, stress and strain of the hydropower station building structure, analyze the anomalies, and predict safety risks; Step 1.1: Based on the deformation monitoring indicators of the hydropower station buildings, deformation prediction is performed on the safety monitoring points of the hydropower station, the future deformation trend is predicted, potential risks are predicted, and abnormal measurement points are identified; Step 1.2: Select the working condition or load combination mode, conduct finite element analysis on the hydropower station building structure, and preview the displacement, stress and strain of the building structure; analyze the safety risk of the hydropower station building structure based on the displacement, stress and strain of the building structure obtained by preview; Step 2: If the measured value or predicted value of the monitoring indicator of the safety monitoring point exceeds the threshold or the method in step 1 identifies an abnormality or safety risk, the warning information is issued through the comprehensive safety management system, displaying the warning time, related measuring points, cause analysis and data changes, distinguishing the warning levels with different colors, and locating abnormal measuring points and risk areas in the digital twin model; Step 3: Search the plan library to find the plan information applicable to the early warning alarm scenario in step 2, display the processing method and steps of the plan, and provide the user with a recommended plan for the early warning alarm scenario in step 2; Step 4: According to the plan obtained in step 3, take corresponding safety measures to handle the abnormality in step 1, control safety risks, and ensure the safe operation of the hydropower station building structure.

[0033] The dispatching safety control method of the hydropower station integrated safety management system based on the digital twin scenario includes the following steps: Step 1: Forecast the inflow of the hydropower station, the power generation output, and the water level and flow of the upstream and downstream river sections; Step 2: Issue warnings for real-time water level exceeding limits or predicted water level exceeding limits or exceeding standard flood or meteorological disaster forecasts; Step 2.1: If the real-time water level or the predicted water level obtained in step 1 exceeds the water level limit, an alarm is triggered and a water level limit warning is issued; Step 2.2: If the inflow of the hydropower station is predicted to exceed the standard flood in step 1, a flood warning is issued; Step 2.3: Issue meteorological disaster warnings based on meteorological disaster forecasts; Step 3: Search the plan library to find the plan applicable to the warning scenario in step 2; Step 4: Rehearse the plan obtained in step 3; Step 4.1: Select the forecasted inflow or preset inflow; set the start water level and control value; Step 4.2: Calculate the change process of outflow, reservoir water level, reservoir capacity and output through the dispatching calculation model; calculate the change process of water level and flow in the upstream and downstream river sections of the hydropower station through the hydrodynamic model; Step 4.3: Compare the simulation results of different scheduling strategies and select the optimal scheduling solution based on the plan; Step 5: According to the optimal dispatching plan obtained in step 4, control the safe and stable operation of the hydropower station.

[0034] Compared with the prior art, the beneficial effects of the present invention include: 1) The present invention provides a standardized method for constructing a digital twin base for a hydropower station and a hierarchical data resource coding system, which realizes the integration of all relevant geographic spatial data, basic data, monitoring data, and business management data of the hydropower station, and constructs a digital twin scene for the hydropower station; establishes a comprehensive safety management system for a hydropower station based on the digital twin scene, and realizes the intelligent management of all-round safety prediction, early warning, plans, and rehearsals for the engineering safety, power generation safety, and dispatching safety of the hydropower station, so as to facilitate the operation and management personnel of the hydropower station to fully grasp the safety situation, changing trends, and safety risks of the hydropower station project, improve the accuracy of early warning and the scientific nature of decision-making, and provide technical guarantees in reducing safety hazards, controlling safety risks, and assisting management decisions. It can enhance the comprehensive safety management capabilities of hydropower stations, improve the advancement, timeliness, and intelligence of safety management, and provide guidance and reference for the efficient, safe, and reliable operation of hydropower stations.

[0035] 2) The present invention provides a hydropower station data resource coding system that includes data resource type coding, monitoring data type coding, and data resource level coding. Through the digital twin model and digital twin scene of the hydropower station entity, multi-dimensional and multi-type data are collected and integrated, realizing the sharing and collaborative management of various types of data, and improving the integration and utilization efficiency of hydropower station data resources.

[0036] 3) The present invention establishes a cascade-boosting regression tree model for the deformation monitoring points of hydropower station buildings, and realizes the deformation prediction of key areas of hydropower station buildings; constructs a finite element analysis model for the structure of hydropower station buildings, calculates the distribution of displacement and stress of the building structure under different working conditions and different load combinations, and displays it in the form of a rendered cloud map; when safety risks occur, the digital twin scene of the hydropower station is used to locate abnormal measuring points and risk areas, recommend relevant plans, and realize the simulation of the plan handling process, which greatly improves the intelligence and wisdom level of the safety of hydropower station projects.

[0037] 4) The present invention predicts the water inflow process of the hydropower station through a hydrological forecasting model; takes the predicted inflow as input, and calculates the outflow, reservoir water level, and unit output process under different scheduling schemes through a scheduling calculation model; compares the calculation results of different scheduling strategies and selects the optimal scheduling scheme; simulates the dynamic effects of gate discharge, unit power generation, and flood storage through a digital twin engine; after an excessive flood or related accident occurs, the early warning information is displayed in a flashing, highlighted, etc. manner in the digital twin scene of the hydropower station, and relevant plans are recommended, realizing the simulation of the plan handling process, which greatly improves the intelligence and wisdom level of the hydropower station scheduling safety.

[0038] 5) The present invention predicts the changing trends of equipment status indicators such as temperature and vibration of hydropower station units through intelligent algorithms such as gradient boosting decision trees and long- and short-term neural networks. When there are risks in equipment operation, the abnormal measurement points and risky equipment are located in the digital twin scenario, and relevant emergency plans are recommended, realizing the simulation of the emergency plan handling process, which greatly improves the intelligence and wisdom level of hydropower station power generation safety.

[0039] 6) The engineering safety control method provided by the present invention can discover potential safety risks and abnormal measurement points in advance by predicting the deformation of safety monitoring points and rehearsing the displacement, stress and strain of the hydropower station building structure, so as to achieve early prevention and early treatment, and effectively avoid or reduce the occurrence of safety accidents; when the measured value or predicted value of the monitoring indicator exceeds the threshold, or anomalies and safety risks are identified, early warning information can be issued, and the warning levels can be distinguished by different colors, and abnormal measurement points and risk areas can be located in the digital twin model; this visual management method makes the early warning information more intuitive and easy to understand, which helps to quickly respond to and deal with safety issues; it can quickly find and display applicable plan information according to the early warning alarm scenario, including processing methods and steps; it provides users with targeted solutions, which helps to take measures quickly in emergency situations, control safety risks, and ensure the safe operation of the hydropower station building structure.

[0040] 7) The dispatching safety control method provided by the present invention can grasp the key parameters and future trends of the operation of the hydropower station in advance through comprehensive prediction and forecasting of the inflow flow, power generation output, and water level and flow of the upstream and downstream river sections, which is helpful to timely discover potential safety risks; for different situations such as real-time water level exceeding the limit, predicted water level exceeding the limit, super-standard floods and meteorological disasters, targeted warnings are issued; this precise warning mechanism can ensure that relevant personnel respond quickly, take necessary dispatching measures, effectively control safety risks, and ensure the safe and stable operation of the hydropower station. The system has a built-in plan library, which can quickly search and provide applicable plans according to the warning scenario, providing users with scientific decision-making basis and reference, and helping to quickly formulate and implement effective dispatching plans in emergency situations. Through the dispatching calculation model and the hydrodynamic model, the outflow flow, reservoir water level, storage capacity, output, and water level and flow change process of the upstream and downstream river sections under different dispatching strategies can be simulated, providing users with intuitive simulation results, which is helpful to compare the effects of different dispatching strategies and select the optimal dispatching plan. According to the optimal scheduling plan, the safe and stable operation of the hydropower station can be controlled, which helps to ensure that the parameters such as the hydropower station's power output, water level and flow rate fluctuate within a safe range, avoid the occurrence of safety accidents, and ensure the long-term stable operation of the hydropower station. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0042] Figure 1 It is a flow chart of a hydropower station engineering safety control method according to an embodiment of the present invention.

[0043] Figure 2 It is a flow chart of a hydropower station dispatching safety control method according to an embodiment of the present invention.

[0044] Figure 3 It is a flow chart of a method for controlling power generation safety in a hydropower station according to an embodiment of the present invention.

[0045] Figure 4 It is a flow chart of a method for constructing a digital twin model of a hydropower station in an embodiment of the present invention.

[0046] Figure 5 This is a deformation simulation interface diagram of a ship lift lock building of a hydropower station according to an embodiment of the present invention.

[0047] Figure 6 This is a diagram of a reservoir dispatching calculation interface according to an embodiment of the present invention.

[0048] Figure 7 This is a diagram of the power generation prediction interface of a hydropower station unit according to an embodiment of the present invention. DETAILED DESCRIPTION

[0049] The comprehensive safety management system for hydropower stations based on digital twin scenarios includes an engineering digital twin base as well as engineering safety subsystems, dispatching safety subsystems and power generation safety subsystems.

[0050] The engineering safety subsystem uses a gradient boosting regression tree model to predict the deformation of key areas of hydropower station buildings at deformation monitoring points of hydropower station buildings. It uses a finite element analysis model to calculate the displacement and stress distribution of the hydropower station building structure under design conditions, typical operating conditions, extreme conditions, custom conditions or different load combinations, and displays the displacement field and stress field of the hydropower station building structure in the form of a rendered cloud map. Based on the digital twin scenario, it locates and marks abnormal measuring points and risk areas with safety risks.

[0051] The engineering safety subsystem includes engineering safety prediction module, engineering safety early warning module, engineering safety rehearsal module and engineering safety plan module.

[0052] The engineering safety prediction module builds a gradient boosting regression tree model for deformation prediction of key areas of hydropower station buildings based on the deformation monitoring indicators of hydropower station buildings; the prediction results obtained by the gradient boosting regression tree model are rendered into the digital twin scene to form a heat map that can reflect the future deformation distribution of key sections of the building, and the rendering results are published to the WEB for display.

[0053] The engineering safety early warning module issues early warning information when the measured or predicted value of the monitoring indicator exceeds the threshold, displays the warning time, related measuring points, causes and data changes, distinguishes the warning levels with different colors, and locates abnormal measuring points and risk areas in the three-dimensional digital twin model of the engineering structure.

[0054] The engineering safety rehearsal module uses the finite element analysis model of the hydropower station building structure to simulate and calculate the overall displacement and stress-strain of the structure under preset and custom conditions. The user selects the design condition, typical operating condition, extreme condition, custom condition or a custom combination of live load, temperature load and wind load to calculate the displacement and stress distribution of the hydropower station building structure, and displays the displacement field and stress field of the hydropower station building structure in the form of a rendered cloud map. The finite element results are rendered on the server side through a visualization engine to generate the corresponding thermal map, and the rendering results are published to the WEB for display.

[0055] The engineering safety plan module searches the built-in plan library to find the plan information applicable to the alarm scenario when the engineering safety monitoring, prediction, and rehearsal results trigger an alarm, and displays the processing methods and steps of the plan; combined with the simulation engine, it simulates the key links of the plan, including material preparation, personnel evacuation, equipment inspection, and dynamic operation process.

[0056] The dispatching safety subsystem uses a hydrological forecasting model to predict the water inflow process of the hydropower station, calculates the outflow flow, reservoir water level and unit output under different dispatching schemes, compares the calculation results of different dispatching strategies, selects the optimal dispatching scheme, and simulates the dynamic effects of gate discharge, unit power generation and flood storage through the digital twin engine.

[0057] The dispatching safety prediction module uses rainfall forecast as input and the hydrological forecast model to predict the inflow of the hydropower station. It uses the predicted inflow as input and the hydrodynamic model to predict the water level and flow changes of the upstream and downstream river sections. It uses the gate unit control mode as input and the dispatching calculation model to predict the power generation output and power generation of the hydropower station. The dispatching safety warning module compares the real-time water level and the characteristic water level. The characteristic water levels include the flood limit water level, normal water storage level, design flood level, and verification flood level. An alarm is triggered when the characteristic water level limit is exceeded. When warning information of natural disasters such as floods, heavy rains, strong winds, and high temperatures is received, the warning information is displayed in a flashing and highlighted manner in the digital twin scene. The warning information includes the warning level, estimated time, and warning value, and a warning text message is sent to the relevant responsible persons.

[0058] In the dispatching safety rehearsal module, the inflow process is input, and the user selects real-time, historical, forecast, or preset inflow for calculation, or customizes the inflow calculation by uploading a template; according to the dispatching regulations or dispatching instructions, the user selects the control method of water level, output, and outflow, and inputs the starting water level and control value; the dispatching calculation model is used to calculate the change process of dispatching factors such as outflow, reservoir water level, reservoir capacity, and output; the water level and flow change process of the upstream and downstream river sections of the hydropower station is calculated through the hydrodynamic model; the calculation results of the dispatching calculation model are integrated into the digital twin scene to simulate the dynamic effects of flood evolution, gate discharge, unit power generation, water level rise and fall, and submergence; the simulation results of different dispatching strategies are compared, the optimal dispatching plan is selected, and the best operation mode of the gate hole and unit is given.

[0059] The dispatching safety plan module, when the hydropower station encounters an excessive flood or related accident, searches the built-in plan library to find the plan information applicable to the scenario, and displays the handling methods and steps for dangerous situations of different levels and types; combined with the simulation engine, it simulates the key links of the plan. The simulation of the plan includes the dynamic process of dispatching measures, material preparation, flood prevention and rescue, personnel evacuation, equipment inspection, and plugging and repair.

[0060] The power generation safety subsystem uses the LSTM network to predict the changing trends of the status indicators of the unit equipment. When safety risks occur in the unit equipment, it locates abnormal measurement points and risky equipment in the digital twin scenario, recommends relevant plans, and simulates and rehearses the plan handling process.

[0061] The power generation safety subsystem includes a power generation safety prediction module, a power generation safety early warning module and a power generation safety plan module.

[0062] The power generation safety prediction module uses a gradient boosting regression tree or AI model to predict the changing trend of the equipment status indicators of important facilities and equipment in the hydropower station, with the equipment operating conditions as input.

[0063] The power generation safety early warning module issues early warning information when the status indicators of important equipment in the hydropower station exceed the threshold, reminding users to pay attention in a flashing and highlighted manner, displaying the early warning time, related measuring points, data changes, and handling status, distinguishing the early warning levels with different colors, and locating abnormal measuring points in the three-dimensional digital twin model of the facilities and equipment.

[0064] The power generation safety plan module, when the hydropower station encounters an alarm event, searches the built-in plan library to find the plan information applicable to the alarm scenario and displays the plan handling methods and steps; combined with the simulation engine, it simulates the key links of the plan. The plan simulation includes the dynamic process of fault repair, personnel evacuation, and equipment operation.

[0065] The digital twin base of the project integrates all relevant geospatial data, basic data, monitoring data, business management data and external shared data of the hydropower station; through multi-scale data mapping of time and space, it constructs a digital twin scene of the hydropower station and provides a three-dimensional visualization carrier for various business applications.

[0066] The engineering digital twin base includes data acquisition module, data management module, data model module and data sharing module.

[0067] The data acquisition module integrates all relevant geospatial data, basic data, monitoring data, business management data and external shared data of the hydropower station according to the hydropower station data resource coding system.

[0068] In the embodiment, the hydropower station data resource coding system is divided into three levels, recorded as AB1B2C1C2D1D2.E1E2F1F2G1G2.H1H2H3. The first 7 bits are the type code of the data resource, the middle 6 bits are the level code of the data resource, and the last 3 bits are the sequence code of the data resource, which is distinguished by ".".

[0069] Code A refers to the major category code of data resources, including five categories: basic data, monitoring data, business management data, cross-industry shared data, and geospatial data. The codes are set from 1 to 5.

[0070] Code B1B2C1C2D1D2 refers to the sub-class code of data resources, which is subdivided for each type of resource. Taking monitoring data as an example, code B1B2 divides monitoring data into three categories: engineering safety monitoring, environmental monitoring, and equipment operation monitoring; code C1C2 is subdivided into deformation, seepage, hydrology, meteorology, water machine equipment operation, electrical equipment operation, and metal structure equipment operation under the previous level classification; code D1D2 is subdivided into displacement, settlement, uplift pressure, foundation reaction force, water level, flow, rainfall, vibration, working temperature, power, opening, and other monitoring indicators under the previous level classification. All levels of codes are set according to 01~09.

[0071] The code E1E2F1F2G1G2 refers to the hierarchical code of the data resource, which can be set according to the business level and subordinate level. Taking monitoring data as an example, the code E1E2 refers to the project or equipment object to which the measuring point belongs, such as the dam body, dam top, dam foundation, gate head, turbine unit, gate, transformer, etc.; the code F1F2 refers to the decomposition structure of the project or equipment, such as dam section #1, dam section #2, dam section #3, etc.; the code G1G2 refers to the next level of decomposition of F1F2, such as unit #1 can be divided into water guide bearing, frame, top cover, stator, volute, impeller, etc. All levels of codes are set according to 01~09.

[0072] The code H1H2H3 refers to the sequence code of the data resource, which is coded according to the equipment sequence and the measurement point sequence, and is set from 001 to 009. Taking the monitoring data as an example, the coding system of the hydropower station data resource is analyzed, as shown in Table 1.

[0073] Table 1

[0074] The data management module implements resource directory management, data quality management, and data security management based on the data standard system.

[0075] Data resource directory: Based on database construction and resource coding system, a data resource directory is established to provide directory and index services for user objects and various business systems. Data virtualization and metadata technology are used to achieve unified data maintenance, providing data catalog management, data directory publishing, data directory services, data directory maintenance and other functions.

[0076] Data quality management: Analyze various data quality issues throughout the entire life cycle of data collection, storage, and application, including key data identification, measurement scheme rule design, data quality measurement, and data quality measurement report release, to provide effective support for improving data quality in each system.

[0077] Data standard system: Based on the demands of unifying data caliber, indicating data location, analyzing data relationships, and managing model changes, basic common standards, data classification standards, coding systems, key technology standards, tool / platform standards, evaluation standards, and security standards are formed to support the efficient management of hydropower station data assets.

[0078] Data security management: Carry out data classification and identification based on the data resource directory, label the data classification, formulate data classification management regulations, and manage data storage, access, backup, and disposal.

[0079] The data model module constructs a digital twin scene of a hydropower station that unifies basic data, aggregates monitoring data, and integrates two and three dimensions through multi-scale data mapping in time and space, providing a three-dimensional visualization carrier for various business applications.

[0080] In the embodiment, remote sensing images, oblique photography and BIM modeling data are used to create a digital twin scene, in which the geographical information of the hydropower station and its affected area, civil engineering, electrical equipment and facilities, turbines, and metal structures are presented in the form of a digital twin model.

[0081] like Figure 4 As shown in the figure, the construction process of the hydropower station digital twin scene includes: Model processing: Integrate the BIM model and 3D model of the hydropower station engineering structure and ancillary electromechanical equipment, the oblique photography model of the surrounding impact area, and other related remote sensing and DOM / DEM models, and convert various models into a unified coordinate system such as the CGCS2000 coordinate system, the 1985 national elevation datum and a unified 3D general standard data format.

[0082] Model editing and beautification: Use vector lines to crop conflicting terrain images to achieve seamless splicing and fusion of digital twins; refine and map textures of buildings, equipment, roads, terrain, vegetation, water systems, etc. that do not conform to reality to beautify the model presentation effect.

[0083] Model data fusion: Establish the relationship between business, monitoring and other data and digital twins to form a digital twin scene of the hydropower station, so that the effects of the outside world on the physical entity can be reflected in real time on the model. The basic data of the engineering structure and equipment and the monitoring data of the measuring points are attached to the three-dimensional model, the gate and unit status are converted into the gate and unit opening and closing actions, the scheduling execution effect is converted into simulated scenes such as flood discharge, power generation, and water level rise and fall, and the alarm warning situation is converted into a highlight flashing point in the scene.

[0084] The data sharing module provides an information channel between the hydropower station management department and its superior and subordinate management units and external related units, realizing data sharing across networks, industries and levels.

[0085] Follow data sharing standards and specifications, form a data exchange link, realize the reporting, distribution and synchronization of relevant data information between the hydropower station management department and the upper and lower management units and external related units, and realize cross-network, cross-industry and cross-level data sharing. By clarifying the main responsibility of data assets, authorization timing and update mechanisms, unified access, exchange status monitoring, data query, data subscription, visual reporting and other data sharing services are realized to meet the needs of real-time data services and batch data sharing.

[0086] In the engineering safety prediction module, the input variables of the gradient boosting regression tree model for deformation prediction of key areas of hydropower station buildings include upstream water level, downstream water level, upstream and downstream water level difference, rainfall and ambient temperature, which are recorded as x ; The output variable is the deformation monitoring index, denoted as y .

[0087] The calculation process of the gradient boosting regression tree model is: a) The training set is denoted as ,in represents the input variable of the i-th sample data, Indicates i Output variable of sample data; The expression of the gradient boosting regression tree model is: ; in represents the number of regression trees, Indicates The parameters of the regression tree, Indicates A regression tree classifier, The expression of a regression tree is: ; The gradient boosted regression tree model uses a squared loss function: ; b) Initialize the first regression tree : ; Where argmin represents the parameter to obtain the minimum value, and c is the category of the root node; c) For data samples i , i =1,2,3... N , calculate the negative gradient of the loss function as an estimate of the residual, ; Calculate residual value , , and as the new true value of the sample, As the next set of training data, fit a new regression tree and get The leaf node area of ​​the regression tree ,in is the sequence number of the leaf node; d) For =1,2,3... J , J is the number of leaf nodes, calculate the value of the leaf node area, minimize the loss function, and calculate the optimal fit value: ; In the formula Represents the fitted value of the jth leaf node of the tth regression tree; e) Update the ladder regression tree model: ; In the formula is the indicator function, for the input variable Determine whether it belongs to the leaf node area ; f) Get the final regression tree, that is, add the leaf node values ​​of each tree: ; g) After the regression tree model is built, the regression tree model is trained using the training set to obtain the optimal trapezoidal regression tree model; h) Use the validation set to test and evaluate the optimal trapezoidal regression tree model. Calculate the mean absolute error (MAE), mean square error (MSE), root mean square error (RMSE), and determination coefficient (R) of the regression tree model. 2 The closer MAE, MSE, and RMSE are to 0, the higher the R 2 The closer it is to 1, the better the model fits the data. The final regression tree model is determined based on the verification results.

[0088] By analyzing the safety monitoring of hydropower station buildings, we selected key cross-sections and longitudinal sections that affect building safety and have dense measurement points, and built corresponding gradient regression tree models for each displacement measurement point on the longitudinal and cross-sections. The gradient regression tree model uses the predicted upstream and downstream water levels, rainfall, and temperature as inputs to calculate the displacement change process of each measurement point in the section during the corresponding forecast period.

[0089] After the calculation is completed, the prediction results need to be rendered into a three-dimensional scene. The specific steps are as follows: estimate the distribution density of the profile measurement points through the coordinate positions of the measurement points; set the color gradient, map the displacement prediction data to different color bands, and process the color value of each measurement point according to the numerical range; render the displacement prediction results on the server side through the visualization engine to form a heat map that can reflect the future deformation distribution of the key sections of the building.

[0090] Publish the rendering results to the WEB for display, merge the heat map with the building twin scene, and drag the timeline to view the deformation trend of the building at different times in the future.

[0091] In the engineering safety rehearsal module, the methods for integrating the stress-strain data obtained from the finite element analysis model with the digital twin scenario include: a) Data preparation; The calculation data of the finite element model consists of node information, unit information and physical properties. The node information consists of the node number and x, y, and z coordinates. The unit information consists of the unit number and the node numbers of the constituent units. Each node or unit will correspond to the calculated physical properties, such as displacement, stress, etc.

[0092] Before rendering the finite element results, the finite element model result file needs to be converted into a JSON data file. The JSON data file contains node coordinates, element indexes, and result values.

[0093] Node coordinates is a two-dimensional array where each element is a subarray containing x, y, and z coordinates.

[0094] The cell index is a two-dimensional array where each element is the index of the three or four nodes that form a triangle or quadrilateral.

[0095] The result is an array of values ​​corresponding to the triangle or quadrilateral, including the displacement and stress values ​​in the x-, y-, and z-directions.

[0096] b) Data standardization; In order to map the finite element calculation results to different color intervals, it is necessary to normalize the displacement, stress and other result data to (0, 1). Specific steps: Calculate the minimum and maximum values ​​of the result values, subtract the minimum value from each result value, and divide it by the difference between the maximum and minimum values ​​to get the standardized result value.

[0097] c) Read JSON data; The vertex coordinates, triangle index and calculation result data in the JSON file are read through the visualization engine for subsequent numerical calculation and processing.

[0098] d) Create triangular patches; The solid elements of the finite element model cannot be directly rendered to the WEB side. The solid elements need to be converted into renderable 3D graphics for storage through the visualization engine. The node and unit information of the finite element model is reconstructed, and the triangles and quadrilaterals contained in the mesh are split into triangular facets. For example, a hexahedral element can be split into 12 triangular facets, and a hexahedral node can be split into 36 triangular facet nodes.

[0099] e) define color mapping; Map result data such as displacement and stress to color values, set color gradients, specify color bands of physical properties with RGB codes, and process the color value of each node according to the numerical range.

[0100] f) Draw triangles and color them; The visualization engine loads the JSON data and triangular patches processed by the finite element object, uses the previously defined color mapping, renders the displacement and stress calculation data on the server side, and generates the corresponding heat map.

[0101] g) Achievements display; The rendering results are published to the WEB for display, and the displacement and stress distribution cloud maps in the XYZ directions on the surface of the hydropower station building structure can be viewed. The static cloud map can be further rendered into a deformation animation for display through the interpolation algorithm.

[0102] At the same time, the system supports the model sectioning function. The section formed after sectioning is drawn into a cloud map using the calculation node data inside the structure for display, which can view and analyze the mechanical properties of the entire model.

[0103] like Figure 1 As shown in the figure, the engineering safety control method of the hydropower station safety integrated management system based on the digital twin scenario includes the following steps: Step 1: Predict the deformation of the safety monitoring points, conduct a preview of the displacement, stress and strain of the hydropower station building structure, analyze the anomalies, and predict safety risks; Step 1.1: Based on the deformation monitoring indicators of the hydropower station buildings, deformation prediction is performed on the safety monitoring points of the hydropower station, the future deformation trend is predicted, potential risks are predicted, and abnormal measurement points are identified; Step 1.2: Select the working condition or load combination mode, conduct finite element analysis on the hydropower station building structure, and preview the displacement, stress and strain of the building structure; analyze the safety risk of the hydropower station building structure based on the displacement, stress and strain of the building structure obtained by preview; Step 2: If the measured value or predicted value of the monitoring indicator of the safety monitoring point exceeds the threshold or the method in step 1 identifies an abnormality or safety risk, the comprehensive safety management system of the present invention issues warning information, displays the warning time, related measuring points, cause analysis and data changes, distinguishes the warning levels with different colors, and locates abnormal measuring points and risk areas in the digital twin model; Step 3: Search the plan library to find the plan information applicable to the early warning alarm scenario in step 2, display the processing method and steps of the plan, and provide the user with a recommended plan for the early warning alarm scenario in step 2; Step 4: According to the plan obtained in step 3, take corresponding safety measures to handle the abnormality in step 1, control safety risks, and ensure the safe operation of the hydropower station building structure.

[0104] In the embodiment, the deformation of the ship lift building structure of the hydropower station under the custom working conditions is predicted, such as Figure 5 shown.

[0105] like Figure 2 As shown, the dispatching safety control method of the hydropower station safety integrated management system based on the digital twin scenario includes the following steps: Step 1: Forecast the inflow of the hydropower station, the power generation output, and the water level and flow of the upstream and downstream river sections; Step 2: Issue warnings for real-time water level exceeding limits or predicted water level exceeding limits or exceeding standard flood or meteorological disaster forecasts; Step 2.1: If the real-time water level or the predicted water level obtained in step 1 exceeds the water level limit, an alarm is triggered and a water level limit warning is issued; Step 2.2: If the inflow of the hydropower station is predicted to exceed the standard flood in step 1, a flood warning is issued; Step 2.3: Issue meteorological disaster warnings based on meteorological disaster forecasts; Step 3: Search the plan library to find the plan applicable to the warning scenario in step 2; Step 4: Rehearse the plan obtained in step 3; Step 4.1: Select the forecasted inflow or preset inflow; set the start water level and control value; Step 4.2: Calculate the change process of outflow, reservoir water level, reservoir capacity and output through the dispatching calculation model; calculate the change process of water level and flow in the upstream and downstream river sections of the hydropower station through the hydrodynamic model; Step 4.3: Compare the simulation results of different scheduling strategies and select the optimal scheduling solution based on the plan; Step 5: According to the optimal dispatching plan obtained in step 4, control the safe and stable operation of the hydropower station.

[0106] In the embodiment, the hydropower station reservoir dispatching preview under flood warning is as follows: Figure 6 shown.

[0107] The power generation safety control method of the embodiment is as follows Figure 3 As shown in the figure, the prediction results of thrust bearing temperature, upper guide bearing temperature and water guide bearing temperature of the hydropower station are as follows Figure 7 shown.

Claims

1. The integrated safety management system for hydropower stations based on digital twin scenarios is characterized by: Including the engineering digital twin base as well as the engineering safety subsystem, dispatching safety subsystem and power generation safety subsystem; The engineering safety subsystem uses an AI model to predict deformation of key areas of hydropower station buildings based on deformation monitoring points of hydropower station buildings; The finite element analysis model is used to calculate the displacement and stress distribution of the hydropower station building structure under design conditions, typical operating conditions, extreme conditions, custom conditions or different load combinations, and the displacement field and stress field of the hydropower station building structure are displayed in the form of rendered cloud maps; Based on the digital twin scenario, abnormal measurement points and risk areas with safety risks are located and marked; The dispatching safety subsystem predicts the water inflow process of the hydropower station, calculates the outflow flow, reservoir water level and unit output under different dispatching schemes, compares the calculation results of different dispatching strategies, selects the optimal dispatching scheme, and simulates the dynamic effects of gate discharge, unit power generation and flood interception and storage through the digital twin engine; The power generation safety subsystem uses AI models to predict the changing trends of the status indicators of the unit equipment. When safety risks occur in the unit equipment, it locates abnormal measurement points and risky equipment in the digital twin scene, recommends relevant plans, and simulates and rehearses the plan handling process. The digital twin base of the project integrates all relevant geospatial data, basic data, monitoring data, business management data and external shared data of the hydropower station; through multi-scale data mapping of time and space, a digital twin scene of the hydropower station is constructed to provide a three-dimensional visualization carrier for various business applications.

2. The hydropower station safety integrated management system based on digital twin scenario according to claim 1 is characterized in that: The engineering digital twin base includes a data acquisition module, a data management module, a data model module and a data sharing module; The data acquisition module integrates all relevant geospatial data, basic data, monitoring data, business management data and external shared data of the hydropower station according to the hydropower station data resource coding system; The data management module implements resource catalog management, data quality management, and data security management based on the data standard system; The data model module builds a digital twin scene of a hydropower station that unifies basic data, aggregates monitoring data, and integrates two- and three-dimensional data through spatiotemporal multi-scale data mapping, providing a three-dimensional visualization carrier for various business applications; The data sharing module provides an information channel between the hydropower station management department and its superior and subordinate management units and external related units, realizing data sharing across networks, industries and levels.

3. The hydropower station safety integrated management system based on digital twin scenario according to claim 2 is characterized in that: The hydropower station data resource coding system adopts a three-level coding AB1B2C1C2D1D2.E1E2F1F2G1G2.H1H2H3, the first-level coding AB1B2C1C2D1D2 is the type coding of data resources, wherein A represents the major category of data resources, which includes basic data, monitoring data, business management data, cross-industry shared data and geographic spatial data; B1B2C1C2D1D2 represents the sub-category of data resources, wherein B1B2 represents the monitoring data type, which includes three categories: engineering safety monitoring, environmental quantity monitoring and equipment operation monitoring; C1C2 represents the monitoring data subdivision type, which includes deformation, seepage, hydrology, meteorology, water machine equipment operation, electrical equipment operation and metal structure equipment operation; D1D2 represents the sub-type of the monitoring data subdivision type; The second level code E1E2F1F2G1G2 is the hierarchical code of the data resource, which is set according to the business level and subordinate level; The third level coding H1H2H3 is the sequential coding of data resources, which is encoded according to the sequence of equipment and measuring points.

4. The hydropower station safety integrated management system based on digital twin scenario according to claim 1 is characterized in that: The engineering safety subsystem includes an engineering safety prediction module, an engineering safety early warning module, an engineering safety rehearsal module and an engineering safety emergency plan module; The engineering safety prediction module constructs a gradient boosting regression tree model for deformation prediction of key areas of hydropower station buildings based on the deformation monitoring indicators of hydropower station buildings; renders the prediction results obtained by the gradient boosting regression tree model into the digital twin scene to form a heat map that can reflect the future deformation distribution of key sections of the building, and publishes the rendering results to the WEB end for display; The engineering safety early warning module issues early warning information when the measured value or predicted value of the monitoring indicator exceeds the threshold, displays the early warning time, related measuring points, causes and data changes, distinguishes the early warning levels with different colors, and locates abnormal measuring points and risk areas in the three-dimensional digital twin model of the engineering structure; The engineering safety rehearsal module uses the finite element analysis model of the hydropower station building structure to simulate and calculate the overall displacement and stress-strain of the structure under preset working conditions and custom working conditions; The user selects the design condition, typical operating condition, extreme condition, custom condition or custom live load, temperature load, wind load combination, calculates the displacement and stress distribution of the hydropower station building structure, and displays the displacement field and stress field of the hydropower station building structure in the form of rendered cloud map; On the server side, the finite element results are rendered through a visualization engine to generate corresponding heat maps, and the rendering results are published to the WEB side for display; The engineering safety plan module, when the engineering safety monitoring, prediction, and rehearsal results trigger an alarm, searches the built-in plan library to find the plan information applicable to the alarm scenario, and displays the processing methods and steps of the plan; combined with the simulation engine, it simulates the key links of the plan, including material preparation, personnel evacuation, equipment inspection, and dynamic operation process.

5. The integrated safety management system for hydropower stations based on digital twin scenarios according to claim 4 is characterized in that: In the engineering safety prediction module, the input variables of the gradient boosting regression tree model for deformation prediction of key areas of hydropower station buildings include upstream water level, downstream water level, upstream and downstream water level difference, rainfall and ambient temperature, denoted as x ; The output variable is the deformation monitoring index, denoted as y ; The calculation process of the gradient boosting regression tree model is: a) The training set is denoted as ,in represents the input variable of the i-th sample data, Indicates i Output variable of sample data; The expression of the gradient boosting regression tree model is: ； in represents the number of regression trees, Indicates The parameters of the regression tree, Indicates A regression tree classifier, The expression of a regression tree is: ； The gradient boosted regression tree model uses a squared loss function: ； b) Initialize the first regression tree : ； Where argmin represents the parameter to obtain the minimum value, and c is the category of the root node; c) For data samples i , i =1,2,3... N , calculate the negative gradient of the loss function as an estimate of the residual, ； Calculate residual value , , and as the new true value of the sample, As the next set of training data, fit a new regression tree and get The leaf node area of ​​the regression tree ,in is the sequence number of the leaf node; d) For =1,2,3... J , J is the number of leaf nodes, calculate the value of the leaf node area, minimize the loss function, and calculate the optimal fit value: ； In the formula Represents the fitted value of the jth leaf node of the tth regression tree; e) Update the ladder regression tree model: ； In the formula is the indicator function, for the input variable Determine whether it belongs to the leaf node area ; f) Get the final regression tree, ； g) After the regression tree model is built, the regression tree model is trained using the training set to obtain the optimal trapezoidal regression tree model; h) Use the validation set to test and evaluate the optimal ladder regression tree model.

6. The hydropower station safety integrated management system based on digital twin scenario according to claim 1 is characterized in that: In the engineering safety rehearsal module, the method of fusing the stress-strain data obtained from the finite element analysis model with the digital twin scene includes:

1. Data preparation; The calculation data of the finite element analysis model includes node information, unit information and physical properties. The node information includes the node number and x, y, z coordinates. The unit information includes the unit number and the node numbers of the constituent units. Each node or unit contains the corresponding calculated physical properties. Before rendering the finite element analysis results, the finite element analysis model calculation result file is converted into a JSON data file, which contains node coordinates, element indexes and result values; 2. Data standardization processing; Standardize the displacement and stress data to the range of (0, 1); 3. Read JSON data; Read the vertex coordinates, triangle index and calculation result data in the JSON file through the visualization engine; 4. Create triangular patches; Reconstruct the node and unit information of the finite element model, and split the triangles and quadrilaterals contained in the mesh into triangular facets; 5. Define color mapping; Map displacement and stress data to color values, set color gradients, specify color bands of physical properties using RGB codes, and process the color value of each node according to the numerical range; 6. Draw triangles and color them; Load the JSON data and triangular patches of the finite element object in the visualization engine, use the previously defined color mapping to render the displacement and stress calculation data on the server side, and generate the corresponding heat map; 7. Achievements display; The rendering results are published to the WEB for display, and the displacement and stress distribution cloud maps in the XYZ directions on the surface of the hydropower station building structure are viewed. The static cloud map is further rendered into a deformation animation for display through the interpolation algorithm.

7. The hydropower station safety integrated management system based on digital twin scenario according to claim 1 is characterized in that: The dispatch safety subsystem includes a dispatch safety prediction module, a dispatch safety early warning module, a dispatch safety rehearsal module and a dispatch safety plan module; The dispatching safety prediction module takes the rainfall forecast as input and predicts the inflow of the hydropower station through the hydrological forecast model; takes the predicted inflow as input and predicts the water level and flow changes of the upstream and downstream river sections through the hydrodynamic model; takes the gate unit control mode as input and predicts the power generation output and power generation of the hydropower station through the dispatching calculation model; The dispatching safety warning module compares the real-time water level with the characteristic water level, which includes the flood limit water level, normal water storage level, design flood level, and verification flood level. When the characteristic water level limit is exceeded, an alarm is triggered; when natural disaster warning information is received, the warning information is displayed in a flashing and highlighted manner in the digital twin scene, and the warning information includes the warning level, estimated time, and warning value, and a warning SMS is sent to the relevant responsible person; The dispatching safety rehearsal module inputs the inflow process, and the user selects real-time, historical, forecast, or preset inflow for calculation, or customizes the inflow by uploading a template; according to the dispatching regulations or dispatching instructions, the user selects the control method of water level, output, and outflow, and inputs the starting water level and control value; the dispatching element change process is calculated through the dispatching calculation model; the water level and flow change process of the upstream and downstream river sections of the hydropower station is calculated through the hydrodynamic model; the calculation results of the dispatching calculation model are integrated into the digital twin scene to simulate the dynamic effects of flood evolution, gate discharge, unit power generation, water level rise and fall, and submergence; the simulation results of different dispatching strategies are compared, the optimal dispatching plan is selected, and the best operation mode of the gate hole and the unit is given; When a hydropower station encounters an excessive flood or a related accident, the dispatching safety plan module searches the built-in plan library to find plan information applicable to the scenario and displays the methods and steps for handling dangerous situations of different levels and types. It combines with the simulation engine to simulate the key links of the plan, which includes the dynamic process of dispatching measures, material preparation, flood prevention and rescue, personnel evacuation, equipment inspection, and plugging and repair.

8. The hydropower station safety integrated management system based on digital twin scenario according to claim 1 is characterized in that: The power generation safety subsystem includes a power generation safety prediction module, a power generation safety early warning module and a power generation safety plan module; The power generation safety prediction module predicts the change trend of the status indicators of important facilities and equipment of the hydropower station through a gradient boosting regression tree or an AI model, taking the equipment operating conditions as input; The power generation safety early warning module, when the status indicators of important equipment in the hydropower station exceed the threshold, issues early warning information, reminds users to pay attention in a flashing and highlighted manner, displays the early warning time, related measuring points, data changes, and disposal status, distinguishes the early warning levels with different colors, and locates abnormal measuring points in the three-dimensional digital twin model of the facilities and equipment; When a hydropower station encounters an alarm event, the power generation safety plan module searches the built-in plan library to find the plan information applicable to the alarm scenario and displays the plan handling methods and steps; combined with the simulation engine, it simulates the key links of the plan, and the plan simulation includes the dynamic process of fault repair, personnel evacuation, and equipment operation.

9. The hydropower station safety integrated management system based on digital twin scenario according to claim 1 is characterized in that: In the hydropower station safety integrated management system, remote sensing images, oblique photography and BIM modeling data are used to create a digital twin scene, in which the hydropower station and its affected area geographic information, civil engineering, electrical equipment and facilities, turbines, and metal structures are presented in the form of a digital twin model; The construction process of the hydropower station digital twin scene includes: (1) Model data preparation: Integrate the BIM model and 3D model data of the hydropower station engineering structure and auxiliary electromechanical equipment, the oblique photography model data of the surrounding impact area, and the remote sensing and DOM / DEM model data, and convert various model data into a unified coordinate system and a unified 3D universal standard data format; (2) Model editing and beautification: Use vector lines to crop conflicting terrain images to achieve seamless splicing and fusion of digital twin models; refine and map textures of buildings, equipment, roads, terrain, vegetation, and water systems that do not conform to reality to beautify the presentation of digital twin models; (3) Model data fusion: Establish the relationship between business and monitoring data and the corresponding digital twin model to form a digital twin scene of the hydropower station, so that the effects of the outside world on the physical entity can be reflected in real time on the digital twin model; attach the basic data of the engineering structure, equipment and facilities, and the monitoring data of the measuring points to the digital twin model, convert the gate and unit status into the gate and unit opening and closing actions, convert the scheduling execution effect into a simulation scene of flood discharge, power generation, and water level rise and fall, and convert the alarm warning situation into a highlight flashing point in the scene.

10. The engineering safety control method of the hydropower station safety integrated management system based on the digital twin scenario according to any one of claims 1 to 9, characterized in that: The following steps are involved: Step 1: Predict the deformation of the safety monitoring points, conduct a preview of the displacement, stress and strain of the hydropower station building structure, analyze the anomalies, and predict safety risks; Step 1.1: Based on the deformation monitoring indicators of the hydropower station buildings, deformation prediction is performed on the safety monitoring points of the hydropower station, the future deformation trend is predicted, potential risks are predicted, and abnormal measurement points are identified; Step 1.2: Select the working condition or load combination mode, conduct finite element analysis on the hydropower station building structure, and preview the displacement, stress and strain of the building structure; analyze the safety risk of the hydropower station building structure based on the displacement, stress and strain of the building structure obtained by preview; Step 2: If the measured value or predicted value of the monitoring indicator of the safety monitoring point exceeds the threshold or the method in step 1 identifies an abnormality or safety risk, the warning information is issued through the comprehensive safety management system, displaying the warning time, related measuring points, cause analysis and data changes, distinguishing the warning levels with different colors, and locating abnormal measuring points and risk areas in the digital twin model; Step 3: Search the plan library to find the plan information applicable to the early warning alarm scenario in step 2, display the processing method and steps of the plan, and provide the user with a recommended plan for the early warning alarm scenario in step 2; Step 4: According to the plan obtained in step 3, take corresponding safety measures to handle the abnormality in step 1, control safety risks, and ensure the safe operation of the hydropower station building structure.

Images