A scene automatic generation method and system based on digital twin
By building high-precision smart models and knowledge maps, combined with Internet of Things technology, the problem of incomplete data acquisition in floodgate management is solved, high-precision dynamic simulation and real-time monitoring of floodgate is achieved, and emergency response capabilities are improved.
Patent Information
- Application Number
- CN202411012481.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-26
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-07-26
AI Technical Summary
The existing technology has incomplete data collection and poor real-time performance in floodgate management, resulting in insufficient accuracy and real-time performance of the digital twin model, and the inability to effectively monitor and optimize the operating status of the floodgate.
By obtaining geospatial data of the floodgate and basic water conservancy data, a high-precision intelligent model and knowledge map are built, and real-time monitoring is carried out in combination with IoT technology, the automatic generation and simulation of the floodgate scene is realized, and the operating status is dynamically simulated, and the accuracy of the model is improved. and practicality.
High-precision dynamic simulation and real-time monitoring of flood diversion gates are realized, the accuracy and practicality of the model are improved, potential problems are discovered in a timely manner, and emergency response efficiency and effectiveness are improved.
Smart Images

Figure CN118551584B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of digital twin technology, and in particular to a method and system for automatically generating scenes based on digital twins. Background Art
[0002] With the rapid development of information technology and intelligent technology, digital twin technology, as an emerging cutting-edge technology, has gradually been widely used in the fields of industry, urban management, water conservancy projects, etc. Digital twin technology comprehensively collects, models and simulates various types of data of physical entities to build a virtual model that is highly consistent with the real world, realizing real-time monitoring, prediction and optimization control of entities.
[0003] In the field of water conservancy projects, diversion gates are key water conservancy facilities, and their operating status and safety are directly related to the effectiveness and safety of flood prevention and disaster reduction. Traditional diversion gate management mainly relies on manual inspections and historical experience, and there are problems such as incomplete data collection, poor real-time performance, and low intelligence level. With the development of sensor technology, Internet of Things technology, and big data technology, diversion gates can be monitored more accurately and comprehensively, providing a basis for the application of digital twin technology in diversion gate management. Existing technologies are mostly single data sources in terms of data collection, and the intelligence level of data fusion is not high, resulting in insufficient accuracy and real-time performance of the model. Summary of the invention
[0004] In order to solve the above-mentioned technical problems, the present invention proposes a method and system for automatic scene generation based on digital twins to solve at least one of the above-mentioned technical problems.
[0005] The present application provides a method for automatically generating a scene based on digital twins, the method comprising:
[0006] S1. Obtaining the geographical spatial data of the flood diversion gate and the basic water conservancy data, and constructing the data base according to the geographical spatial data of the flood diversion gate and the basic water conservancy data to obtain the spatial base data of the flood diversion gate;
[0007] S2. constructing a flood diversion gate intelligent model based on the flood diversion gate spatial base plate data to obtain flood diversion gate intelligent model data;
[0008] S3, obtaining basic data of the flood diversion gate, and constructing a knowledge graph based on the basic data of the flood diversion gate to obtain knowledge graph data of the flood diversion gate;
[0009] S4. Construct a twin engine based on the flood diversion gate intelligent model data and the flood diversion gate knowledge graph data to obtain the flood diversion gate twin engine data for automatic generation of flood diversion gate scenarios.
[0010] The data obtained by the present invention through a variety of high-precision technologies ensures the accuracy and comprehensiveness of geospatial data and basic water conservancy data. Real-time monitoring data and Internet of Things technology are used to ensure the dynamic update and real-time performance of the data base. A high-precision intelligent model of the flood diversion gate is constructed through detailed structural data processing and parameter initialization. The combination of the hydrodynamic model and the intelligent model enables the model to dynamically simulate the operating status of the flood diversion gate, improving the accuracy and practicality of the model. Through the construction of the relational model, the relationship between each knowledge point is clarified, and the organization and utilization efficiency of knowledge is improved. The twin engine can monitor and predict the operating status of the flood diversion gate in real time, and promptly discover and warn potential problems. Different scenarios are automatically generated according to the simulation results, which improves the efficiency and effectiveness of emergency response.
[0011] Optionally, S1 includes:
[0012] Obtain geospatial data of flood diversion gates and basic water conservancy data;
[0013] Perform three-dimensional terrain modeling based on the geospatial data of the flood diversion gate to obtain three-dimensional terrain model data of the flood diversion gate;
[0014] The spatial base plate data of the flood diversion gate is obtained by fusion of geographic data based on the basic water conservancy data and the three-dimensional terrain model data of the flood diversion gate.
[0015] The present invention uses a variety of high-precision technical means to obtain comprehensive data coverage, ensuring the accuracy and integrity of geographic space and water conservancy basic data. The three-dimensional terrain model can accurately show the topographic features, which helps to design and manage flood diversion gates more accurately. The three-dimensional model provides an accurate terrain basis for water flow simulation, flood analysis, etc., which helps to improve the accuracy of analysis. The fused data can improve the accuracy of spatial data and ensure the accuracy of models and simulations.
[0016] Optionally, S2 includes:
[0017] Perform structural data processing based on the spatial bottom plate data of the flood diversion gate to obtain the structural data of the flood diversion gate;
[0018] Initialize key parameters of the model according to the flood diversion gate structure data to obtain key parameter data of the model;
[0019] The hydrodynamic model of the flood diversion gate is constructed according to the key parameter data of the model to obtain the hydrodynamic model data of the flood diversion gate;
[0020] Simulating the state of the flood diversion gate according to the hydrodynamic model data of the flood diversion gate to obtain the simulation data of the state of the flood diversion gate;
[0021] Model integration is performed based on the flood diversion gate state simulation data and the flood diversion gate hydrodynamic model data to obtain the flood diversion gate intelligent model data.
[0022] In the present invention, the structural information of the diversion gate is extracted from the spatial base plate data, including the geometric shape, size, material and position of the gate, dam, spillway, etc., to ensure the accuracy and reliability of the diversion gate structure data. The extraction and classification process ensures the comprehensiveness of key parameters and covers all important information required for model construction. Through detailed parameter classification and initialization, the accuracy and applicability of the model are improved. The hydrodynamic model can dynamically simulate the water flow behavior and operation status of the diversion gate, which improves the realism of the model. Through the analysis of simulation results, potential risks can be identified in advance and the emergency response capability can be improved. The intelligent model provides scientific support for the management and decision-making of the diversion gate, and improves the accuracy and effectiveness of decision-making.
[0023] Optionally, S3 includes:
[0024] Obtain basic data of flood diversion gates;
[0025] Perform knowledge extraction based on the basic data of the flood diversion gate to obtain the knowledge extraction data of the flood diversion gate;
[0026] According to the flood diversion gate knowledge extraction data, a relational model is constructed to obtain the flood diversion gate knowledge relational model;
[0027] The knowledge graph data of the flood diversion gate is obtained by performing knowledge graph conversion according to the knowledge relationship model of the flood diversion gate.
[0028] The present invention collects comprehensive basic data of flood diversion gates to ensure the diversity and integrity of the data, and provides rich source data for subsequent knowledge extraction. Key knowledge points and information are extracted from the basic data, including structural features, operating status, maintenance cycle, fault records, etc. Through knowledge extraction, key knowledge points can be obtained from a large amount of data to provide important information for subsequent analysis. Identify the entities and relationships between entities in the knowledge extraction data, such as the relationship between structural parts and functions, operating steps and maintenance cycles, etc. Through relationship identification and model construction, the relationship between each knowledge point is clarified, and the organization and utilization efficiency of knowledge is improved. The knowledge relationship model is converted into a knowledge graph, and the knowledge points and relationships are represented as nodes and edges using graph database technology. Using semantic technology, the nodes and edges in the knowledge graph are semantically associated to enhance the intelligence and interpretability of the knowledge graph.
[0029] Optionally, S4 includes:
[0030] Mapping the flood diversion gate smart model data according to the flood diversion gate knowledge graph data to obtain the flood diversion gate smart model mapping data;
[0031] Acquire real-time flood diversion gate sensor data, and build a simulation engine based on the real-time flood diversion gate sensor data and flood diversion gate smart model mapping data to obtain primary flood diversion gate twin engine data;
[0032] Multi-scenario simulation calculations are performed based on the primary flood diversion gate twin engine data to obtain the flood diversion gate twin engine data for automatic generation of flood diversion gate scenarios.
[0033] In the present invention, the knowledge graph data of the flood diversion gate is mapped with the intelligent model data of the flood diversion gate, and the entities in the knowledge graph and the corresponding parts of the relationship in the intelligent model are identified and matched. By mapping the knowledge graph data to the intelligent model data, the combination of knowledge and model is realized, and the intelligence level of the model is enhanced. The fusion of real-time data and intelligent model data improves the dynamics and accuracy of the data and ensures the reliability of the simulation results. The simulation engine can accurately simulate the operating status and hydrological changes of the flood diversion gate and provide high-precision simulation data. Different simulation scenarios are set, including normal operation, emergency status, extreme weather conditions, etc., covering various situations that the flood diversion gate may encounter. The primary flood diversion gate twin engine is used to perform multi-scenario simulation calculations to generate operating data and early warning information under different scenarios. Through simulation calculations, potential risks and early warning information can be identified in advance to improve emergency response capabilities.
[0034] Optionally, the relationship model building includes:
[0035] Perform knowledge entity recognition and entity attribute extraction based on the flood diversion gate knowledge extraction data to obtain the flood diversion gate knowledge entity data and flood diversion gate entity attribute data;
[0036] Associating the flood diversion gate knowledge entity data and the flood diversion gate entity attribute data to obtain the flood diversion gate entity association data;
[0037] Entity relationship extraction is performed based on the flood diversion gate entity association data and the flood diversion gate basic data to obtain the flood diversion gate entity relationship data;
[0038] The relational model is constructed based on the entity association data and entity relationship data of the flood diversion gate, and a preliminary knowledge relational model of the flood diversion gate is obtained;
[0039] The knowledge fusion is performed based on the preliminary flood diversion gate knowledge relationship model to obtain the flood diversion gate knowledge relationship model.
[0040] Identifying and extracting knowledge entities and their attributes in the present invention helps to clarify the key components and features of the flood diversion gate. Structuring knowledge entities and attributes is conducive to subsequent data association and relationship extraction, and improves the efficiency of data organization and management. Through entity association, the relationship between each knowledge entity is clarified, and the relevance and comprehensibility of the data are improved. The generation of entity-related data provides rich association information for subsequent relationship extraction and model construction. Through relationship extraction, the various relationships between each knowledge entity can be fully understood and recorded, providing comprehensive data support for model construction. Constructing a preliminary knowledge relationship model, systematically displaying the knowledge entities and their relationships of the flood diversion gate, and improving the efficiency of data organization and management. The information in the preliminary knowledge relationship model is integrated to eliminate redundancy and conflict, and form a unified knowledge relationship model.
[0041] Optionally, the flood diversion gate knowledge relationship model includes a first flood diversion gate knowledge relationship model and a second flood diversion gate knowledge relationship model, and the knowledge fusion includes:
[0042] Performing a first knowledge fusion according to the preliminary flood diversion gate knowledge relationship model to obtain a first flood diversion gate knowledge relationship model;
[0043] Performing second knowledge fusion according to the preliminary flood diversion gate knowledge relationship model to obtain a second flood diversion gate knowledge relationship model, wherein the first knowledge fusion and the second knowledge fusion are different knowledge fusion modes;
[0044] The first knowledge fusion is specifically as follows:
[0045] Mapping is performed according to the basic data of the flood diversion gate and the preset ontology concept list data to obtain ontology concept mapping data;
[0046] The ontology model is constructed according to the ontology concept mapping data to obtain the preliminary flood diversion gate ontology model data;
[0047] Ontology instantiation is performed based on the preliminary flood diversion gate ontology model data and the preliminary flood diversion gate knowledge relationship model to obtain the flood diversion gate ontology data.
[0048] In the present invention, by adopting different knowledge fusion methods, the diversity and comprehensiveness of the knowledge relationship model are ensured, and the applicability and reliability of the model are improved. The first knowledge fusion enhances the standardization and semantic association of the knowledge model through an ontology-driven method, and improves the interpretability of knowledge. The second knowledge fusion improves the accuracy and reliability of the data through a statistically driven method, and effectively solves the problems of data conflict and redundancy.
[0049] Optionally, the second knowledge fusion includes:
[0050] According to the preliminary flood diversion gate knowledge relationship model, a graph is constructed to obtain the flood diversion gate knowledge relationship graph data;
[0051] Perform graph matching according to the flood diversion gate knowledge relationship graph data to obtain matching graph data;
[0052] Perform graph clustering according to the matching graph data to obtain cluster graph data;
[0053] Perform self-attention graph fusion according to cluster graph data to obtain fusion graph data;
[0054] Redundant relationships are eliminated based on the fusion graph data to obtain the second flood diversion gate knowledge relationship model.
[0055] The knowledge relationship model in the present invention is structured in the form of a graph, which improves the efficiency of knowledge organization and management and provides an intuitive visualization method. Through graph matching and clustering, similar knowledge entities and relationships are identified and classified, and the organizational form of knowledge relationships is optimized. The self-attention mechanism can automatically identify and strengthen important nodes and edges, eliminate redundancy and conflict, and improve the criticality and accuracy of the knowledge model. The knowledge relationship model is further optimized to ensure the simplicity and effectiveness of the model and improve the data quality and credibility. The knowledge relationship model provides a comprehensive knowledge basis for the intelligent management and decision support of flood diversion gates, and improves the scientificity and effectiveness of management.
[0056] Optionally, self-attention map fusion includes:
[0057] Perform node feature extraction and edge feature extraction according to the cluster graph data to obtain cluster graph node feature data and cluster graph edge feature data;
[0058] Perform multi-head self-attention calculation based on the cluster graph node feature data and the cluster graph edge feature data to obtain graph node feature self-attention data and graph edge feature self-attention data;
[0059] Perform convolution calculation based on the graph node feature self-attention data and the graph edge feature self-attention data to obtain the graph node feature convolution layer data and the graph edge feature convolution layer data;
[0060] Perform masked self-attention calculation based on cluster graph node feature data and cluster graph edge feature data to obtain graph node feature masked self-attention data and graph edge feature masked self-attention data;
[0061] According to the graph node feature convolution layer data and the graph node feature mask self-attention data, weighted fusion is performed to obtain the cluster graph node weighted feature data, and according to the graph edge feature convolution layer data and the graph edge feature mask self-attention data, weighted fusion is performed to obtain the cluster graph edge weighted feature data;
[0062] According to the weighted feature data of the cluster graph nodes, the key nodes of the cluster graph data are fused to obtain the fused node data;
[0063] According to the weighted feature data of the cluster graph edges, the key edges of the cluster graph data are fused to obtain fused edge data;
[0064] A fusion graph is constructed according to the fusion node data and the fusion edge data to obtain fusion graph data.
[0065] The present invention ensures the comprehensiveness and accuracy of the data through detailed feature extraction and processing, providing a reliable data foundation for subsequent processing. The self-attention mechanism is used to identify and strengthen important nodes and edges, thereby improving the criticality and expressiveness of the data. Advanced features are extracted through convolution calculations, enhancing the quality and richness of feature data. Masked self-attention calculations effectively handle missing values and abnormal data to ensure the integrity and reliability of the data. Feature data from different sources are integrated through weighted fusion to improve the comprehensiveness and accuracy of feature representation. The fusion process of key nodes and edges optimizes the data structure, eliminates redundancy, and improves the simplicity and effectiveness of the knowledge model.
[0066] Optionally, the present application also provides a scene automatic generation system based on digital twins, which is used to execute the scene automatic generation method based on digital twins as described above, and the scene automatic generation system based on digital twins includes:
[0067] A data baseboard construction module is used to obtain the geographical spatial data of the flood diversion gate and the basic water conservancy data, and to construct the data baseboard according to the geographical spatial data of the flood diversion gate and the basic water conservancy data to obtain the spatial baseboard data of the flood diversion gate;
[0068] A flood diversion gate intelligent model construction module is used to construct a flood diversion gate intelligent model according to the flood diversion gate spatial base plate data to obtain the flood diversion gate intelligent model data;
[0069] A flood diversion gate knowledge graph construction module is used to obtain the basic data of the flood diversion gate, and to construct a knowledge graph based on the basic data of the flood diversion gate to obtain the knowledge graph data of the flood diversion gate;
[0070] The flood diversion gate twin engine construction module is used to construct the twin engine based on the flood diversion gate intelligent model data and the flood diversion gate knowledge graph data, and obtain the flood diversion gate twin engine data to perform automatic generation operations of the flood diversion gate scenario.
[0071] The purpose of the present invention is to ensure the comprehensiveness and accuracy of data through multi-source data acquisition and integration, and provide a reliable data basis for model construction and simulation. The intelligent model covers the structure, operating status and hydrodynamic characteristics of the flood diversion gate, and realizes high-precision dynamic simulation and intelligent optimization. The knowledge graph systematically organizes the knowledge of the flood diversion gate, enhances the semantic association and data quality of the knowledge, and improves the interpretability and application value of the knowledge. The twin engine realizes real-time monitoring and multi-scenario simulation of the flood diversion gate, provides comprehensive operating data and early warning information, and improves emergency response capabilities. Through data analysis and optimization, the operating efficiency and safety of the flood diversion gate are improved, providing comprehensive intelligent support for management and decision-making. BRIEF DESCRIPTION OF THE DRAWINGS
[0072] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting implementations made with reference to the following drawings:
[0073] Figure 1 A flowchart of a method for automatically generating a scene based on digital twins according to an embodiment is shown;
[0074] Figure 2 A flowchart showing a method for constructing a data backplane according to an embodiment of the present invention is provided;
[0075] Figure 3 A flowchart showing a method for constructing a flood diversion gate intelligent model according to an embodiment is shown;
[0076] Figure 4 A flowchart showing a method for constructing a knowledge graph of a flood diversion gate according to an embodiment of the present invention is shown;
[0077] Figure 5 A flowchart showing a method for constructing a twin engine of a flood diversion gate according to an embodiment of the present invention is provided;
[0078] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION
[0079] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.
[0080] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.
[0081] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.
[0082] See also Figures 1 to 5 , the present application provides a method for automatically generating a scene based on digital twins, the method comprising:
[0083] S1. Obtaining the geographical spatial data of the flood diversion gate and the basic water conservancy data, and constructing the data base according to the geographical spatial data of the flood diversion gate and the basic water conservancy data to obtain the spatial base data of the flood diversion gate;
[0084] Specifically, use LiDAR technology to perform high-precision terrain scanning to obtain terrain data of the diversion gate and surrounding areas. Use drone aerial photography technology to obtain high-resolution ground image data. Use GIS system to integrate geospatial data such as terrain, landforms, and river channels. Extract design parameters and structural data from diversion gate design documents and engineering drawings. Obtain operation records and maintenance records of diversion gates through water conservancy departments. Use historical hydrological data and monitoring data to supplement basic water conservancy information. Use software such as ArcGIS or AutoCAD Civil 3D to convert geospatial data into three-dimensional terrain models. Calibrate and optimize terrain models to ensure the accuracy and authenticity of the models. Fuse basic water conservancy data with three-dimensional terrain model data, and use GIS technology to integrate data from different sources into a unified geographic information system. Calibrate the fused data to ensure the consistency and accuracy of spatial and attribute data.
[0085] S2. constructing a flood diversion gate intelligent model based on the flood diversion gate spatial base plate data to obtain flood diversion gate intelligent model data;
[0086] Specifically, structural information is extracted from the spatial base plate data of the flood diversion gate, and data cleaning technology is used to remove redundant and noisy data. Data standardization is performed to ensure the consistency of data format and unit. Key parameters such as gate height, width, material properties, etc. are extracted from the processed structural data. Statistical analysis methods are used to classify and initialize key parameters to generate model key parameter data.
[0087] Based on key parameter data, a three-dimensional hydrodynamic model is constructed using hydrodynamic principles and software (such as HEC-RAS) to simulate water flow behavior and flood diversion processes. Model calibration is performed to adjust model parameters using historical data and actual observation data to ensure the accuracy of the model.
[0088] Set up different simulation scenarios, including normal operation, flood emergency, maintenance and overhaul, etc., and use the hydrodynamic model to simulate multiple scenarios. Generate flood diversion gate data under different operating conditions, analyze the results, and identify key conditions and potential risks.
[0089] Integrate state simulation data with hydrodynamic model data, and use data fusion technology to form comprehensive intelligent model data. Introduce machine learning and artificial intelligence technology to intelligently analyze and optimize the integrated data to build a complete intelligent model of flood diversion gates.
[0090] S3, obtaining basic data of the flood diversion gate, and constructing a knowledge graph based on the basic data of the flood diversion gate to obtain knowledge graph data of the flood diversion gate;
[0091] Specifically, the design parameters, operation records, maintenance records, environmental data and historical hydrological data of the flood diversion gate are collected and stored in a unified database to ensure data consistency and integrity.
[0092] Use natural language processing (NLP) technology to extract key knowledge points and information from basic data, identify knowledge entities and their attributes, and annotate and classify the extracted knowledge points to form standardized knowledge extraction data.
[0093] Identify entities and relationships between entities in knowledge extraction data, such as the relationship between structural parts and functions, operation steps and maintenance cycles, etc. Use graph database technology (such as Neo4j) to build a knowledge relationship model to show the relationship between various entities.
[0094] Convert the relational model into a knowledge graph, and use graph database technology to represent knowledge points and relationships as nodes and edges. Optimize the knowledge graph, use semantic technology to semantically associate nodes and edges, and enhance the intelligence and interpretability of the knowledge graph.
[0095] S4. Construct a twin engine based on the flood diversion gate intelligent model data and the flood diversion gate knowledge graph data to obtain the flood diversion gate twin engine data for automatic generation of flood diversion gate scenarios.
[0096] Specifically, the flood diversion gate knowledge graph data is mapped with the flood diversion gate smart model data, and the entities and relationships in the knowledge graph are identified and matched to the corresponding parts in the smart model. The mapped data is integrated to form consistent flood diversion gate smart model mapping data.
[0097] Obtain real-time data from various sensors of the flood diversion gate, including key parameters such as water level, flow, temperature, pressure, etc. Fuse the real-time sensor data with the flood diversion gate intelligent model mapping data to ensure the real-time and accuracy of the data.
[0098] A simulation engine is built based on the fused data, and simulation software (such as Ansys and Matlab) is used to simulate the operating status and hydrological changes of the flood diversion gate to form primary flood diversion gate twin engine data.
[0099] Set different simulation scenarios, including normal operation, emergency status, extreme weather conditions, etc., and perform multi-scenario simulation calculations. Generate operation data and warning information under different scenarios, analyze and optimize simulation results, and identify key risk points and optimization opportunities.
[0100] The data obtained by the present invention through a variety of high-precision technologies ensures the accuracy and comprehensiveness of geospatial data and basic water conservancy data. Real-time monitoring data and Internet of Things technology are used to ensure the dynamic update and real-time performance of the data base. A high-precision intelligent model of the flood diversion gate is constructed through detailed structural data processing and parameter initialization. The combination of the hydrodynamic model and the intelligent model enables the model to dynamically simulate the operating status of the flood diversion gate, improving the accuracy and practicality of the model. Through the construction of the relational model, the relationship between each knowledge point is clarified, and the organization and utilization efficiency of knowledge is improved. The twin engine can monitor and predict the operating status of the flood diversion gate in real time, and promptly discover and warn potential problems. Different scenarios are automatically generated according to the simulation results, which improves the efficiency and effectiveness of emergency response.
[0101] Optionally, S1 includes:
[0102] S11. Obtaining geospatial data of flood diversion gates and basic water conservancy data;
[0103] Specifically, the LiDAR (Light Detection and Ranging) technology is used to scan the flood diversion gate and its surrounding areas to obtain high-precision terrain data. The LiDAR system sends laser pulses and measures the time it takes to reflect back from the target, thereby generating high-resolution point cloud data.
[0104] Use drones equipped with high-resolution cameras to take aerial photos and obtain high-definition images of the ground. Stitch the aerial images into orthophotos and use image recognition technology to extract geographic features.
[0105] Use GIS (Geographic Information System) technology to integrate geographic information from different data sources, including topography, landforms, rivers, etc. Store and manage geospatial data through spatial databases (such as PostGIS).
[0106] Extract key design parameters and structural information from the design documents and engineering drawings of the flood diversion gate, and use CAD software (such as AutoCAD) to digitize the data.
[0107] Collect historical operation and maintenance records of the flood diversion gates and use a database system (such as MySQL) to store and manage this data.
[0108] Obtain historical hydrological data from hydrological monitoring departments, including water level, flow, rainfall, etc., and use a time series database (such as InfluxDB) to store and manage this data.
[0109] S12, performing three-dimensional terrain modeling according to the geospatial data of the flood diversion gate to obtain three-dimensional terrain model data of the flood diversion gate;
[0110] Specifically, the point cloud data obtained from LiDAR is preprocessed to remove noise points and outliers. A filtering algorithm (such as RANSAC) is used to identify and remove outliers. A machine learning algorithm (such as Random Forest) is used to classify the point cloud data into ground points, vegetation points, and building points.
[0111] The ground points are extracted from the classified point cloud data and the digital terrain model (DTM) is generated using the Delaunay triangulation algorithm.
[0112] Map drone aerial images as textures onto 3D terrain models to enhance the visual effects of the models. Use image stitching algorithms (such as SIFT) for image stitching and registration.
[0113] The generated 3D model is optimized and a smoothing algorithm (such as Laplacian Smoothing) is used to smooth the model surface and reduce noise and irregularities.
[0114] S13. Geographic data fusion is performed based on the basic water conservancy data and the three-dimensional terrain model data of the flood diversion gate to obtain the spatial base plate data of the flood diversion gate.
[0115] Specifically, data from different data sources are converted to a unified coordinate system (such as WGS84), and coordinate alignment is performed using a coordinate transformation algorithm (such as Procrustes Analysis).
[0116] Spatial registration of 3D terrain model and water conservancy basic data is performed to ensure spatial consistency of data. ICP (Iterative Closest Point) algorithm is used for precise alignment.
[0117] Associate the water conservancy basic data with the geometric data of the 3D terrain model to construct the spatial base plate data. Use spatial connection algorithms (such as Spatial Join) to achieve the association of attribute data.
[0118] Integrate all data into a unified geographic information system to build a complete spatial base plate data of the flood diversion gate. Use a geographic database (such as PostGIS) to store and manage the merged data.
[0119] The present invention uses a variety of high-precision technical means to obtain comprehensive data coverage, ensuring the accuracy and integrity of geographic space and water conservancy basic data. The three-dimensional terrain model can accurately show the topographic features, which helps to design and manage flood diversion gates more accurately. The three-dimensional model provides an accurate terrain basis for water flow simulation, flood analysis, etc., which helps to improve the accuracy of analysis. The fused data can improve the accuracy of spatial data and ensure the accuracy of models and simulations.
[0120] Optionally, S2 includes:
[0121] S21, performing structural data processing according to the spatial bottom plate data of the flood diversion gate to obtain structural data of the flood diversion gate;
[0122] Specifically, the geometric information of the flood diversion gate, such as the shape and location of the gate, dam, and spillway, is extracted from the spatial base plate data, and the relevant attribute information, such as material, size, design parameters, etc., is extracted from the water conservancy basic data.
[0123] S22, initializing key parameters of the model according to the flood diversion gate structure data to obtain key parameter data of the model;
[0124] Specifically, key geometric parameters are extracted: such as the height, width, thickness of the gate, the length and slope of the dam, etc. Key physical parameters are extracted: such as the density, strength, elastic modulus, etc. of the material.
[0125] Geometric parameter classification: classify the extracted geometric parameters, such as size parameters, shape parameters, etc. Physical parameter classification: classify the extracted physical parameters, such as material property parameters, environmental condition parameters, etc.
[0126] Statistical analysis methods (such as PCA) are used to reduce the dimension and analyze the extracted parameters to ensure the validity and relevance of the parameters. Based on the analysis results, various parameters are initialized to generate key parameter data of the model.
[0127] S23, constructing a flood diversion gate hydrodynamic model according to the key parameter data of the model to obtain flood diversion gate hydrodynamic model data;
[0128] Specifically, the Navier-Stokes equation and the continuity equation are used to perform hydrodynamic analysis to describe the motion law of water flow. The boundary conditions of the model are set, including the water inlet, water outlet, water level, etc.
[0129] The hydrodynamic equations are discretized using the finite element method (FEM) or the finite volume method (FVM) to generate a discrete set of equations, which are then solved using a numerical solution algorithm (such as an explicit solver or implicit solver for CFL conditions) to obtain hydrodynamic model data.
[0130] S24, performing flood diversion gate state simulation according to the flood diversion gate hydrodynamic model data to obtain flood diversion gate state simulation data;
[0131] Specifically, the normal operation status of the flood diversion gate is set, including routine operation and maintenance. Different emergency states are set, including flood, failure, maintenance, etc.
[0132] Use the hydrodynamic model to perform multi-scenario numerical simulations to generate flood diversion gate data under different operating conditions. Analyze the simulation results to identify critical conditions and potential risks.
[0133] Record the status data of each simulation scenario, including parameters such as water level, flow, pressure, etc. Perform risk assessment based on simulation data, and identify and record potential risk points.
[0134] S25. Perform model integration based on the flood diversion gate state simulation data and the flood diversion gate hydrodynamic model data to obtain the flood diversion gate intelligent model data.
[0135] Specifically, the state simulation data is integrated with the hydrodynamic model data, and a data fusion algorithm (such as Kalman filtering) is used to perform data fusion processing. The fused data is checked for consistency to ensure the integrity and accuracy of the data.
[0136] In the present invention, the structural information of the diversion gate is extracted from the spatial base plate data, including the geometric shape, size, material and position of the gate, dam, spillway, etc., to ensure the accuracy and reliability of the diversion gate structure data. The extraction and classification process ensures the comprehensiveness of key parameters and covers all important information required for model construction. Through detailed parameter classification and initialization, the accuracy and applicability of the model are improved. The hydrodynamic model can dynamically simulate the water flow behavior and operation status of the diversion gate, which improves the realism of the model. Through the analysis of simulation results, potential risks can be identified in advance and the emergency response capability can be improved. The intelligent model provides scientific support for the management and decision-making of the diversion gate, and improves the accuracy and effectiveness of decision-making.
[0137] Optionally, S3 includes:
[0138] S31, obtaining basic data of flood diversion gates;
[0139] Specifically, design parameters and structural data were collected from flood diversion gate design documents and engineering drawings. Information from paper documents was extracted and digitized using OCR (optical character recognition) technology. Historical operation and maintenance records of the flood diversion gate were obtained from the water conservancy department. ETL (extraction, transformation, loading) tools such as Talend were used to import data into a unified database. Environmental data and historical hydrological data around the flood diversion gate were obtained. Real-time and historical data were obtained from meteorological and hydrological monitoring systems using API interfaces and stored.
[0140] S32, performing knowledge extraction based on the flood diversion gate basic data to obtain flood diversion gate knowledge extraction data;
[0141] Specifically, NLP technology is used to extract key information from text data. Named entity recognition (NER) algorithms are applied to identify entities and attributes in design parameters, operating status, and maintenance records. Text mining tools (such as NLTK and SpaCy) are used to parse documents and extract key knowledge points and information.
[0142] Label the extracted knowledge entities and use manual labeling tools (such as Labelbox) to verify and label the data to ensure data accuracy. Label the attributes of the knowledge entities and use automated labeling tools (such as Snorkel) to improve labeling efficiency.
[0143] S33, constructing a relational model based on the flood diversion gate knowledge extraction data to obtain a flood diversion gate knowledge relational model;
[0144] Specifically, use graph databases (such as Neo4j) to identify entity relationships in knowledge extraction data. Apply graph matching algorithms (such as Subgraph Isomorphism) to identify associations between entities. Use relationship extraction algorithms (such as OpenIE) to identify relationships between entity attributes and establish attribute associations.
[0145] The identified entities and relationships are constructed into a knowledge graph, and the entities and relationships are represented using the graph data model of the graph database. The constructed relationship model data is stored in the graph database to ensure the structured and queryable nature of the data.
[0146] S34. Perform knowledge graph conversion based on the flood diversion gate knowledge relationship model to obtain flood diversion gate knowledge graph data.
[0147] Specifically, the entities in the knowledge relationship model are used as nodes and the relationships as edges to build a knowledge graph. The nodes and edges are created using the API interface of the graph database. The graph optimization algorithm (such as PageRank) is applied to optimize the knowledge graph to improve the query efficiency and readability of the graph.
[0148] Use ontology-driven semantic enhancement technology to associate nodes and edges in the knowledge graph with domain ontology. Use RDF and OWL to represent semantic information. Apply semantic reasoning engines (such as Apache Jena) to reason about the knowledge graph and generate new knowledge and relationships.
[0149] The present invention collects comprehensive basic data of flood diversion gates to ensure the diversity and integrity of the data, and provides rich source data for subsequent knowledge extraction. Key knowledge points and information are extracted from the basic data, including structural features, operating status, maintenance cycle, fault records, etc. Through knowledge extraction, key knowledge points can be obtained from a large amount of data to provide important information for subsequent analysis. Identify the entities and relationships between entities in the knowledge extraction data, such as the relationship between structural parts and functions, operating steps and maintenance cycles, etc. Through relationship identification and model construction, the relationship between each knowledge point is clarified, and the organization and utilization efficiency of knowledge is improved. The knowledge relationship model is converted into a knowledge graph, and the knowledge points and relationships are represented as nodes and edges using graph database technology. Using semantic technology, the nodes and edges in the knowledge graph are semantically associated to enhance the intelligence and interpretability of the knowledge graph.
[0150] Optionally, S4 includes:
[0151] S41, mapping the flood diversion gate smart model data according to the flood diversion gate knowledge graph data to obtain the flood diversion gate smart model mapping data;
[0152] Specifically, the entities in the knowledge graph are aligned with the entities in the intelligent model, and the entity alignment algorithm (such as Word2Vec) is used to calculate the similarity between entities and identify matching entity pairs. The relationships in the knowledge graph are mapped to the relationships in the intelligent model, and the graph matching algorithm (such as Graph Matching) is used to identify and match the relationships between entities.
[0153] S42, acquiring real-time flood diversion gate sensor data, and constructing a simulation engine based on the real-time flood diversion gate sensor data and flood diversion gate smart model mapping data to obtain primary flood diversion gate twin engine data;
[0154] Specifically, various sensors, including water level sensors, flow sensors, temperature sensors, and pressure sensors, are deployed at key locations of the flood diversion gates, and data is collected using IoT technologies (such as LoRa and NB-IoT). Sensor data is collected in real time through IoT platforms (such as ThingSpeak), and data is transmitted using the MQTT protocol.
[0155] The real-time sensor data is integrated with the intelligent model mapping data, and the data fusion algorithm (such as Kalman filtering) is used for data fusion processing. A simulation engine is built based on the fused data, and simulation software (such as Ansys and MatlabSimulink) is used for modeling to simulate the operation status and hydrological changes of the flood diversion gate, and generate primary flood diversion gate twin engine data.
[0156] S43. Perform multi-scenario simulation calculations based on the primary flood diversion gate twin engine data to obtain flood diversion gate twin engine data for automatic generation of flood diversion gate scenarios.
[0157] Specifically, different simulation scenarios are defined, including normal operation, emergency status, extreme weather conditions, etc., to ensure that various situations that the flood diversion gate may encounter are covered. Corresponding parameters are set for each simulation scenario, including water level, flow, pressure, etc., to ensure the authenticity and effectiveness of the simulation scenario.
[0158] Use the primary flood diversion gate twin engine to perform multi-scenario simulation calculations, use numerical solution algorithms (such as explicit solvers or implicit solvers for CFL conditions) to solve, and generate operation data under different scenarios. Analyze the simulation results, identify key states and potential risks, and use data analysis tools (such as Python Pandas) for data analysis and processing.
[0159] Record the operating data of each simulation scenario, including parameters such as water level, flow, and pressure, and use data storage tools (such as HDF5) for data storage.
[0160] In the present invention, the knowledge graph data of the flood diversion gate is mapped with the intelligent model data of the flood diversion gate, and the entities in the knowledge graph and the corresponding parts of the relationship in the intelligent model are identified and matched. By mapping the knowledge graph data to the intelligent model data, the combination of knowledge and model is realized, and the intelligence level of the model is enhanced. The fusion of real-time data and intelligent model data improves the dynamics and accuracy of the data and ensures the reliability of the simulation results. The simulation engine can accurately simulate the operating status and hydrological changes of the flood diversion gate and provide high-precision simulation data. Different simulation scenarios are set, including normal operation, emergency status, extreme weather conditions, etc., covering various situations that the flood diversion gate may encounter. The primary flood diversion gate twin engine is used to perform multi-scenario simulation calculations to generate operating data and early warning information under different scenarios. Through simulation calculations, potential risks and early warning information can be identified in advance to improve emergency response capabilities.
[0161] Optionally, the relationship model building includes:
[0162] Perform knowledge entity recognition and entity attribute extraction based on the flood diversion gate knowledge extraction data to obtain the flood diversion gate knowledge entity data and flood diversion gate entity attribute data;
[0163] Specifically, natural language processing technology (such as SpaCy or BERT) is used to perform named entity recognition on text data, extract entities related to the flood diversion gate (such as gates, dams, spillways, etc.), and obtain the knowledge entity data of the flood diversion gate. By matching known entity libraries or domain dictionaries, the identified entities are verified to ensure the accuracy of the recognition results.
[0164] Identify the attributes related to the entity (such as size, material, location, status, etc.) from the text, and use the relationship extraction algorithm (such as OpenIE) to identify the attributes. Label and classify the identified attributes, and use the labeling tool (such as Labelbox) to manually check and label the data to obtain the entity attribute data of the flood diversion gate.
[0165] Associating the flood diversion gate knowledge entity data and the flood diversion gate entity attribute data to obtain the flood diversion gate entity association data;
[0166] Specifically, we use relation extraction algorithms (such as RE) to identify the relationships between entities from knowledge entity data and extract entity association information. We verify the identified entity relationships by matching known relation libraries or domain knowledge to ensure the accuracy of the association data.
[0167] Entity relationship extraction is performed based on the flood diversion gate entity association data and the flood diversion gate basic data to obtain the flood diversion gate entity relationship data;
[0168] Specifically, use graph databases (such as Neo4j) to store and manage entity-related data and perform relationship extraction. Apply graph matching algorithms (such as Subgraph Isomorphism) to identify complex relationships between entities and extract entity relationship data.
[0169] The relational model is constructed based on the entity association data and entity relationship data of the flood diversion gate, and a preliminary knowledge relational model of the flood diversion gate is obtained;
[0170] Specifically, we use entities as nodes and relationships as edges to build a preliminary knowledge relationship model, and use the API interface of the graph database to create nodes and edges. We apply graph optimization algorithms (such as PageRank) to optimize the knowledge relationship model to improve the query efficiency and readability of the model. We verify the model by comparing actual data with the relationship model, and use error analysis algorithms (such as RMSE) to evaluate and correct model errors.
[0171] The knowledge fusion is performed based on the preliminary flood diversion gate knowledge relationship model to obtain the flood diversion gate knowledge relationship model.
[0172] Specifically, based on domain ontology, knowledge data from different sources are mapped into a unified ontology structure and integrated through the semantic association of ontology. Statistical methods are used to integrate data, eliminate redundancy and conflict, and ensure data consistency and accuracy.
[0173] Construct a knowledge ontology in the field of flood diversion gates, and use RDF and OWL to represent semantic information. Map the knowledge data to the ontology structure and use semantic associations to fuse the data. Use the semantic associations of the ontology to fuse the data and generate a unified knowledge relationship model.
[0174] Identifying and extracting knowledge entities and their attributes in the present invention helps to clarify the key components and features of the flood diversion gate. Structuring knowledge entities and attributes is conducive to subsequent data association and relationship extraction, and improves the efficiency of data organization and management. Through entity association, the relationship between each knowledge entity is clarified, and the relevance and comprehensibility of the data are improved. The generation of entity-related data provides rich association information for subsequent relationship extraction and model construction. Through relationship extraction, the various relationships between each knowledge entity can be fully understood and recorded, providing comprehensive data support for model construction. Constructing a preliminary knowledge relationship model, systematically displaying the knowledge entities and their relationships of the flood diversion gate, and improving the efficiency of data organization and management. The information in the preliminary knowledge relationship model is integrated to eliminate redundancy and conflict, and form a unified knowledge relationship model.
[0175] Optionally, the flood diversion gate knowledge relationship model includes a first flood diversion gate knowledge relationship model and a second flood diversion gate knowledge relationship model, and the knowledge fusion includes:
[0176] Performing a first knowledge fusion according to the preliminary flood diversion gate knowledge relationship model to obtain a first flood diversion gate knowledge relationship model;
[0177] Specifically, the first flood diversion gate knowledge relationship model is obtained through an ontology-driven knowledge fusion method.
[0178] Performing second knowledge fusion according to the preliminary flood diversion gate knowledge relationship model to obtain a second flood diversion gate knowledge relationship model, wherein the first knowledge fusion and the second knowledge fusion are different knowledge fusion modes;
[0179] Specifically, a second flood diversion gate knowledge relationship model is obtained through a statistically driven knowledge fusion method.
[0180] The first knowledge fusion is specifically as follows:
[0181] Mapping is performed according to the basic data of the flood diversion gate and the preset ontology concept list data to obtain ontology concept mapping data;
[0182] Specifically, key information and concepts are extracted from the basic data of the flood diversion gate, and natural language processing (NLP) technology (such as SpaCy) is used for concept extraction. According to the preset ontology concept list, the extracted concepts are mapped and concept matching is performed using string matching algorithms (such as Jaccard similarity).
[0183] Collect text data such as design documents, operation records, and maintenance records related to the flood diversion gate. Load pre-trained SpaCy models, such as en_core_web_sm. Process the text data to extract named entities, phrases, and key concepts. Merge the extracted named entities and phrases to form a list of key information and concepts. Load an ontology concept list containing preset concepts. Calculate the Jaccard similarity between the extracted concepts and the concepts in the ontology concept list. Determine the matching concepts based on the similarity threshold.
[0184] The ontology model is constructed according to the ontology concept mapping data to obtain the preliminary flood diversion gate ontology model data;
[0185] Specifically, an ontology modeling tool (such as Protégé) is used to build an ontology model of the flood diversion gate, define concepts, attributes and relationships. Based on the concept mapping data, the classes, attributes and relationships of the ontology are defined, and OWL (Web Ontology Language) is used to represent the ontology structure.
[0186] Perform semantic verification on the ontology model to ensure the logical consistency and integrity of the model, and use semantic reasoning engines (such as HermiT) for verification. Apply ontology optimization algorithms (such as ontology reduction algorithms) to optimize the ontology model to improve the simplicity and efficiency of the model.
[0187] Ontology instantiation is performed based on the preliminary flood diversion gate ontology model data and the preliminary flood diversion gate knowledge relationship model to obtain the flood diversion gate ontology data.
[0188] Specifically, the data instances in the preliminary flood diversion gate knowledge relational model are mapped to the ontology model using an instance mapping algorithm (such as Instance Matching). Based on the mapping results, the ontology is instantiated, ontology instances are created, and instance data are represented using RDF and OWL.
[0189] Use semantic reasoning engines (such as Apache Jena) to perform semantic reasoning, generate new knowledge and relationships, and enrich ontology data. Optimize the generated semantic associations, eliminate redundancy and conflicts, and use association optimization algorithms (such as ontology merging algorithms) for optimization processing.
[0190] Make sure Apache Jena is installed on your system and prepare the ontology data files to be processed (such as ontology files in RDF / XML or TTL format). Use Jena API to load ontology data: Create a Jena model and load the ontology data into the model. Use the built-in reasoner (such as RDFS or OWL) to create an inference model. Perform reasoning on the inference model to generate new knowledge and relationships.
[0191] The newly generated knowledge and relations are extracted from the reasoning model using the SPARQL query language.
[0192] In the present invention, by adopting different knowledge fusion methods, the diversity and comprehensiveness of the knowledge relationship model are ensured, and the applicability and reliability of the model are improved. The first knowledge fusion enhances the standardization and semantic association of the knowledge model through an ontology-driven method, and improves the interpretability of knowledge. The second knowledge fusion improves the accuracy and reliability of the data through a statistically driven method, and effectively solves the problems of data conflict and redundancy.
[0193] Optionally, the second knowledge fusion includes:
[0194] According to the preliminary flood diversion gate knowledge relationship model, a graph is constructed to obtain the flood diversion gate knowledge relationship graph data;
[0195] Specifically, the entities in the knowledge relationship model are used as nodes and the relationships as edges to construct a knowledge relationship graph. The nodes and edges are created using the API interface of the graph database (such as Neo4j). The constructed knowledge relationship graph data is stored in the graph database, and the graph database (such as Neo4j) is used to manage the graph data.
[0196] Perform graph matching according to the flood diversion gate knowledge relationship graph data to obtain matching graph data;
[0197] Specifically, a graph matching algorithm (such as Graph Edit Distance or VF2) is used to calculate the similarity of nodes and edges in the graph and identify similar subgraphs. Based on the calculated similarity, similar subgraphs in the knowledge relationship graph are identified and matched.
[0198] Perform graph clustering according to the matching graph data to obtain cluster graph data;
[0199] Specifically, a graph clustering algorithm (such as Louvain or Spectral Clustering) is used to perform cluster analysis on the matching graph data to identify the community structure in the graph. The graph clustering algorithm is run to cluster similar nodes and edges into the same community.
[0200] Specifically, more importantly, a weighted adjacency matrix conversion is performed according to the matching graph data to obtain adjacency matrix conversion data; a degree matrix is calculated according to the adjacency matrix conversion data to obtain degree matrix data; a Laplace matrix is calculated according to the adjacency matrix conversion data and the degree matrix data to obtain Laplace matrix data; eigenvalue decomposition and feature selection are performed according to the Laplace matrix data to obtain eigenvector matrix data; normalization and clustering are performed according to the eigenvector matrix data to obtain clustering graph data;
[0201] The weighted adjacency matrix conversion is specifically as follows: neighboring node similarity calculation and node degree calculation are performed based on the matching graph data to obtain neighboring node similarity data and node degree data; nonlinear activation calculation is performed based on the neighboring node similarity data and node degree data to obtain node attribute data; adjacency matrix calculation is performed based on the matching graph data to obtain adjacency matrix data; adjacency matrix data is weighted according to the node attribute data to obtain adjacency matrix conversion data.
[0202] Perform self-attention graph fusion according to cluster graph data to obtain fusion graph data;
[0203] Specifically, extract node and edge features from cluster graph data, and use feature extraction algorithms (such as Node2Vec or Edge2Vec) to generate feature vectors. Use self-attention mechanisms (such as Multi-Head Attention in Transformer) to perform weighted calculations on node and edge features to identify and strengthen important nodes and edges. Perform convolution operations on the self-attention calculation results, and use graph convolutional networks (GCNs) to aggregate and update features. Fuse the convolved features to generate fused graph data.
[0204] Redundant relationships are eliminated based on the fusion graph data to obtain the second flood diversion gate knowledge relationship model.
[0205] Specifically, a redundancy detection algorithm (such as MinHash or SimHash) is used to identify redundant relations in the fusion graph data, and the identified redundant relations are marked for deletion.
[0206] Delete the marked redundant relationships and use graph editing algorithms (such as Graph Simplification) to simplify and optimize the relationship graph.
[0207] The knowledge relationship model in the present invention is structured in the form of a graph, which improves the efficiency of knowledge organization and management and provides an intuitive visualization method. Through graph matching and clustering, similar knowledge entities and relationships are identified and classified, and the organizational form of knowledge relationships is optimized. The self-attention mechanism can automatically identify and strengthen important nodes and edges, eliminate redundancy and conflict, and improve the criticality and accuracy of the knowledge model. The knowledge relationship model is further optimized to ensure the simplicity and effectiveness of the model and improve the data quality and credibility. The knowledge relationship model provides a comprehensive knowledge basis for the intelligent management and decision support of flood diversion gates, and improves the scientificity and effectiveness of management.
[0208] Optionally, self-attention map fusion includes:
[0209] Perform node feature extraction and edge feature extraction according to the cluster graph data to obtain cluster graph node feature data and cluster graph edge feature data;
[0210] Specifically, select the basic features of the node, such as degree, clustering coefficient, centrality, etc. Use the Node2Vec algorithm to embed the structural features of the node into a low-dimensional vector space to generate a node feature vector.
[0211] Select the basic features of the edge, such as weight, similarity, connectivity strength, etc. Use the Edge2Vec algorithm to embed the structural features of the edge into a low-dimensional vector space to generate an edge feature vector.
[0212] Perform multi-head self-attention calculation based on the cluster graph node feature data and the cluster graph edge feature data to obtain graph node feature self-attention data and graph edge feature self-attention data;
[0213] Specifically, multi-head self-attention calculation is performed on the feature data of each node and edge, and the attention weights of nodes and edges are calculated using the multi-head attention mechanism. The attention matrix of nodes and edges is generated to represent the attention distribution of different heads.
[0214] Perform convolution calculation based on the graph node feature self-attention data and the graph edge feature self-attention data to obtain the graph node feature convolution layer data and the graph edge feature convolution layer data;
[0215] Specifically, a graph convolutional network is used (e.g., the number of convolutional layers, the number of layers, the convolution kernel, the offset, and the activation function are generated according to preset convolutional layer parameters, and the convolution kernels are such as the maximum convolution kernel, the minimum convolution kernel, the average convolution kernel, and the Gaussian convolution kernel) to perform convolution operations on the self-attention features of nodes and edges. The graph node feature convolutional layer data and the graph edge feature convolutional layer data are calculated.
[0216] Perform masked self-attention calculation based on cluster graph node feature data and cluster graph edge feature data to obtain graph node feature masked self-attention data and graph edge feature masked self-attention data;
[0217] Specifically, according to the characteristics of nodes and edges in the clustering graph, a mask matrix is generated to mask out irrelevant or redundant information. The masked self-attention mechanism is used to calculate the attention weights of nodes and edges. , is the graph node feature mask self-attention data / graph edge feature mask self-attention data, softmax() is the softmax function, which is used to convert the input vector into a probability distribution to ensure that the sum of the weights is 1. is the query matrix corresponding to the cluster graph node feature data / cluster graph edge feature data (the matrix after the input feature is linearly transformed), is the key matrix corresponding to the cluster graph node feature data / cluster graph edge feature data (the matrix after the input feature is linearly transformed), is the mask matrix corresponding to the cluster graph node feature data / cluster graph edge feature data, It is the dimension data of the key matrix corresponding to the clustering graph node feature data / clustering graph edge feature data. It is the value matrix corresponding to the cluster graph node feature data / cluster graph edge feature data (the matrix after the input features have been linearly transformed).
[0218] According to the graph node feature convolution layer data and the graph node feature mask self-attention data, weighted fusion is performed to obtain the cluster graph node weighted feature data, and according to the graph edge feature convolution layer data and the graph edge feature mask self-attention data, weighted fusion is performed to obtain the cluster graph edge weighted feature data;
[0219] Specifically, the node feature convolution layer data is weighted fused with the node feature mask self-attention data. The edge feature convolution layer data is weighted fused with the edge feature mask self-attention data, and the same formula is used for calculation. , Weighted feature data for cluster graph nodes / weighted feature data for cluster graph edges, is the weight parameter corresponding to the graph node feature convolution layer data / graph edge feature convolution layer data, is the graph node feature convolution layer data / graph edge feature convolution layer data, is the weight parameter corresponding to the graph node feature mask self-attention data / graph edge feature mask self-attention data, Masked self-attention data for graph node features / masked self-attention data for graph edge features.
[0220] According to the weighted feature data of the cluster graph nodes, the key nodes of the cluster graph data are fused to obtain the fused node data;
[0221] Specifically, the node feature weighted data is used to calculate the node score, and the PageRank algorithm is applied to score the nodes. Key nodes are selected based on the score to ensure the importance and representativeness of the selected nodes in the graph.
[0222] Perform feature aggregation on the selected key nodes and use feature aggregation algorithms (such as Max Pooling or Average Pooling) to fuse node features. Generate fused node data and store it in the database.
[0223] According to the weighted feature data of the cluster graph edges, the key edges of the cluster graph data are fused to obtain fused edge data;
[0224] Specifically, edge feature weighted data is used to calculate edge scores, and edge centrality algorithms (such as EdgeBetweenness Centrality) are applied to score edges. Key edges are selected based on the scores to ensure the importance and representativeness of the selected edges in the graph.
[0225] Perform feature aggregation on the selected key edges and use feature aggregation algorithms (such as Max Pooling or Average Pooling) to fuse edge features. Generate fused edge data and store it in the database.
[0226] A fusion graph is constructed according to the fusion node data and the fusion edge data to obtain fusion graph data.
[0227] Specifically, a fusion graph is constructed based on the fusion node data and fusion edge data, and nodes and edges are created using the graph database API (such as Neo4j). A graph optimization algorithm (such as a graph simplification algorithm) is applied to optimize the fusion graph to improve the query efficiency and readability of the graph.
[0228] The present invention ensures the comprehensiveness and accuracy of the data through detailed feature extraction and processing, providing a reliable data foundation for subsequent processing. The self-attention mechanism is used to identify and strengthen important nodes and edges, thereby improving the criticality and expressiveness of the data. Advanced features are extracted through convolution calculations, enhancing the quality and richness of feature data. Masked self-attention calculations effectively handle missing values and abnormal data to ensure the integrity and reliability of the data. Feature data from different sources are integrated through weighted fusion to improve the comprehensiveness and accuracy of feature representation. The fusion process of key nodes and edges optimizes the data structure, eliminates redundancy, and improves the simplicity and effectiveness of the knowledge model.
[0229] Optionally, the present application also provides a scene automatic generation system based on digital twins, which is used to execute the scene automatic generation method based on digital twins as described above, and the scene automatic generation system based on digital twins includes:
[0230] A data baseboard construction module is used to obtain the geographical spatial data of the flood diversion gate and the basic water conservancy data, and to construct the data baseboard according to the geographical spatial data of the flood diversion gate and the basic water conservancy data to obtain the spatial baseboard data of the flood diversion gate;
[0231] A flood diversion gate intelligent model construction module is used to construct a flood diversion gate intelligent model according to the flood diversion gate spatial base plate data to obtain the flood diversion gate intelligent model data;
[0232] A flood diversion gate knowledge graph construction module is used to obtain the basic data of the flood diversion gate, and to construct a knowledge graph based on the basic data of the flood diversion gate to obtain the knowledge graph data of the flood diversion gate;
[0233] The flood diversion gate twin engine construction module is used to construct the twin engine based on the flood diversion gate intelligent model data and the flood diversion gate knowledge graph data, and obtain the flood diversion gate twin engine data to perform automatic generation operations of the flood diversion gate scenario.
[0234] The purpose of the present invention is to ensure the comprehensiveness and accuracy of data through multi-source data acquisition and integration, and provide a reliable data basis for model construction and simulation. The intelligent model covers the structure, operating status and hydrodynamic characteristics of the flood diversion gate, and realizes high-precision dynamic simulation and intelligent optimization. The knowledge graph systematically organizes the knowledge of the flood diversion gate, enhances the semantic association and data quality of the knowledge, and improves the interpretability and application value of the knowledge. The twin engine realizes real-time monitoring and multi-scenario simulation of the flood diversion gate, provides comprehensive operating data and early warning information, and improves emergency response capabilities. Through data analysis and optimization, the operating efficiency and safety of the flood diversion gate are improved, providing comprehensive intelligent support for management and decision-making.
[0235] Therefore, from any point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the attached application documents rather than the above description, and it is intended that all changes falling within the meaning and scope of equivalent elements of the application documents are included in the present invention.
[0236] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.
Claims
1. A method for automatic scene generation based on digital twins, characterized in that: The method comprises: S1. Obtaining the geographical spatial data of the flood diversion gate and the basic water conservancy data, and constructing the data base according to the geographical spatial data of the flood diversion gate and the basic water conservancy data to obtain the spatial base data of the flood diversion gate; S2. constructing a flood diversion gate intelligent model based on the flood diversion gate spatial base plate data to obtain flood diversion gate intelligent model data; S3 includes: Obtain basic data of flood diversion gates; Perform knowledge extraction based on the basic data of the flood diversion gate to obtain the knowledge extraction data of the flood diversion gate; According to the flood diversion gate knowledge extraction data, a relational model is constructed to obtain the flood diversion gate knowledge relational model; Perform knowledge graph conversion according to the knowledge relationship model of the flood diversion gate to obtain the knowledge graph data of the flood diversion gate; S4. Construct a twin engine based on the flood diversion gate intelligent model data and the flood diversion gate knowledge graph data to obtain the flood diversion gate twin engine data to automatically generate the flood diversion gate scene; The relationship model construction includes: Perform knowledge entity recognition and entity attribute extraction based on the flood diversion gate knowledge extraction data to obtain the flood diversion gate knowledge entity data and flood diversion gate entity attribute data; Associating the flood diversion gate knowledge entity data and the flood diversion gate entity attribute data to obtain the flood diversion gate entity association data; Entity relationship extraction is performed based on the flood diversion gate entity association data and the flood diversion gate basic data to obtain the flood diversion gate entity relationship data; The relational model is constructed based on the entity association data and entity relationship data of the flood diversion gate, and a preliminary knowledge relational model of the flood diversion gate is obtained; According to the preliminary flood diversion gate knowledge relationship model, knowledge fusion is performed to obtain the flood diversion gate knowledge relationship model; The flood diversion gate knowledge relationship model includes the first flood diversion gate knowledge relationship model and the second flood diversion gate knowledge relationship model, and the knowledge fusion includes: Performing a first knowledge fusion according to the preliminary flood diversion gate knowledge relationship model to obtain a first flood diversion gate knowledge relationship model; Performing second knowledge fusion according to the preliminary flood diversion gate knowledge relationship model to obtain a second flood diversion gate knowledge relationship model, wherein the first knowledge fusion and the second knowledge fusion are different knowledge fusion modes; The first knowledge fusion is specifically as follows: Mapping is performed according to the basic data of the flood diversion gate and the preset ontology concept list data to obtain ontology concept mapping data; The ontology model is constructed according to the ontology concept mapping data to obtain the preliminary flood diversion gate ontology model data; According to the preliminary flood diversion gate ontology model data and the preliminary flood diversion gate knowledge relationship model, ontology instantiation is performed to obtain the flood diversion gate ontology data; The second knowledge fusion includes: According to the preliminary flood diversion gate knowledge relationship model, a graph is constructed to obtain the flood diversion gate knowledge relationship graph data; Perform graph matching according to the flood diversion gate knowledge relationship graph data to obtain matching graph data; Perform graph clustering according to the matching graph data to obtain cluster graph data; Perform self-attention graph fusion according to cluster graph data to obtain fusion graph data; Redundant relationships are eliminated based on the fusion graph data to obtain the second flood diversion gate knowledge relationship model.
2. The method according to claim 1, characterized in that S1 includes: Obtain geospatial data of flood diversion gates and basic water conservancy data; Perform three-dimensional terrain modeling based on the geospatial data of the flood diversion gate to obtain three-dimensional terrain model data of the flood diversion gate; The spatial base plate data of the flood diversion gate is obtained by fusion of geographic data based on the basic water conservancy data and the three-dimensional terrain model data of the flood diversion gate.
3. The method according to claim 1, characterized in that S2 include: Perform structural data processing based on the spatial bottom plate data of the flood diversion gate to obtain the structural data of the flood diversion gate; Initialize key parameters of the model according to the flood diversion gate structure data to obtain key parameter data of the model; The hydrodynamic model of the flood diversion gate is constructed according to the key parameter data of the model to obtain the hydrodynamic model data of the flood diversion gate; Simulating the state of the flood diversion gate according to the hydrodynamic model data of the flood diversion gate to obtain the simulation data of the state of the flood diversion gate; Model integration is performed based on the flood diversion gate state simulation data and the flood diversion gate hydrodynamic model data to obtain the flood diversion gate intelligent model data.
4. The method according to claim 1, characterized in that S4 include: Mapping the flood diversion gate smart model data according to the flood diversion gate knowledge graph data to obtain the flood diversion gate smart model mapping data; Acquire real-time flood diversion gate sensor data, and build a simulation engine based on the real-time flood diversion gate sensor data and flood diversion gate smart model mapping data to obtain primary flood diversion gate twin engine data; Multi-scenario simulation calculations are performed based on the primary flood diversion gate twin engine data to obtain the flood diversion gate twin engine data for automatic generation of flood diversion gate scenarios.
5. The method according to claim 1, characterized in that Self-attention graph fusion includes: Perform node feature extraction and edge feature extraction according to the cluster graph data to obtain cluster graph node feature data and cluster graph edge feature data; Perform multi-head self-attention calculation based on the cluster graph node feature data and the cluster graph edge feature data to obtain graph node feature self-attention data and graph edge feature self-attention data; Perform convolution calculation based on the graph node feature self-attention data and the graph edge feature self-attention data to obtain the graph node feature convolution layer data and the graph edge feature convolution layer data; Perform masked self-attention calculation based on cluster graph node feature data and cluster graph edge feature data to obtain graph node feature masked self-attention data and graph edge feature masked self-attention data; According to the graph node feature convolution layer data and the graph node feature mask self-attention data, weighted fusion is performed to obtain the cluster graph node weighted feature data, and according to the graph edge feature convolution layer data and the graph edge feature mask self-attention data, weighted fusion is performed to obtain the cluster graph edge weighted feature data; According to the weighted feature data of the cluster graph nodes, the key nodes of the cluster graph data are fused to obtain the fused node data; Performing key edge fusion on cluster graph data according to cluster graph edge weighted feature data to obtain fused edge data; A fusion graph is constructed according to the fusion node data and the fusion edge data to obtain fusion graph data.
6. A scene automatic generation system based on digital twins, characterized in that: Used to execute the scene automatic generation method based on digital twins as claimed in claim 1, the scene automatic generation system based on digital twins includes: A data baseboard construction module is used to obtain the geographical spatial data of the flood diversion gate and the basic water conservancy data, and to construct the data baseboard according to the geographical spatial data of the flood diversion gate and the basic water conservancy data to obtain the spatial baseboard data of the flood diversion gate; A flood diversion gate intelligent model construction module is used to construct a flood diversion gate intelligent model according to the flood diversion gate spatial base plate data to obtain the flood diversion gate intelligent model data; The flood diversion gate knowledge graph construction module is used to obtain the basic data of the flood diversion gate; perform knowledge extraction based on the basic data of the flood diversion gate to obtain the knowledge extraction data of the flood diversion gate; perform relationship model construction based on the knowledge extraction data of the flood diversion gate to obtain the knowledge relationship model of the flood diversion gate; perform knowledge graph conversion based on the knowledge relationship model of the flood diversion gate to obtain the knowledge graph data of the flood diversion gate; wherein the relationship model construction includes: performing knowledge entity recognition and entity attribute extraction based on the knowledge extraction data of the flood diversion gate to obtain the knowledge entity data of the flood diversion gate and the entity attribute data of the flood diversion gate; performing association based on the knowledge entity data of the flood diversion gate and the entity attribute data of the flood diversion gate to obtain the entity association data of the flood diversion gate; performing entity relationship extraction based on the entity association data of the flood diversion gate and the basic data of the flood diversion gate to obtain the entity relationship data of the flood diversion gate; performing relationship model construction based on the entity association data of the flood diversion gate and the entity relationship data of the flood diversion gate to obtain the preliminary knowledge relationship model of the flood diversion gate; performing knowledge fusion based on the preliminary knowledge relationship model of the flood diversion gate to obtain the knowledge relationship model of the flood diversion gate; wherein the knowledge relationship model of the flood diversion gate includes the first knowledge relationship model of the flood diversion gate and the second knowledge relationship model of the flood diversion gate, The fusion includes: performing a first knowledge fusion according to a preliminary flood diversion gate knowledge relationship model to obtain a first flood diversion gate knowledge relationship model; performing a second knowledge fusion according to the preliminary flood diversion gate knowledge relationship model to obtain a second flood diversion gate knowledge relationship model, wherein the first knowledge fusion and the second knowledge fusion are different knowledge fusion methods; wherein the first knowledge fusion is specifically: mapping according to the flood diversion gate basic data and the preset ontology concept list data to obtain ontology concept mapping data; constructing an ontology model according to the ontology concept mapping data to obtain preliminary flood diversion gate ontology model data; instantiating an ontology according to the preliminary flood diversion gate ontology model data and the preliminary flood diversion gate knowledge relationship model to obtain flood diversion gate ontology data; wherein the second knowledge fusion includes: performing a graph construction according to the preliminary flood diversion gate knowledge relationship model to obtain flood diversion gate knowledge relationship graph data; performing graph matching according to the flood diversion gate knowledge relationship graph data to obtain matching graph data; performing graph clustering according to the matching graph data to obtain clustering graph data; performing self-attention graph fusion according to the clustering graph data to obtain fused graph data; performing redundant relationship elimination according to the fused graph data to obtain a second flood diversion gate knowledge relationship model; The flood diversion gate twin engine construction module is used to construct the twin engine based on the flood diversion gate intelligent model data and the flood diversion gate knowledge graph data, and obtain the flood diversion gate twin engine data to perform automatic generation operations of the flood diversion gate scenario.
