A digital twin basin early warning data efficient processing method, system and device

Abstract

The application discloses a kind of digital twin river basin early warning data efficient processing method, system and equipment, belong to river basin early warning technical field.The existing river basin early warning data processing scheme cannot meet the real-time early warning demand.A kind of digital twin river basin early warning data efficient processing method of the application, by creating data splitting model, template making model, data relocation model, risk focusing model, obtain optimized high-risk early warning data, so that it can greatly improve processing efficiency, reduce the redundancy of data, meet the real-time early warning demand, and can effectively solve the pain point of repeated calculation, avoid memory overflow, so that the efficient processing of digital twin river basin early warning data can be realized.Further, by reusing static geometric grid, the dynamic hydrological parameters are stored, to avoid the repeated calculation of "each time step needs to be independently positioned geometric grid" in the traditional scheme, so that the business demand of "minute level response" of river basin early warning can be met.

Description

Technical Field

[0001] This invention relates to a method, system, and device for efficient processing of digital twin watershed early warning data, belonging to the field of watershed early warning technology. Background Technology

[0002] Existing watershed early warning simulation schemes rely on the calculation results of grid-discrete hydrological and hydrodynamic models (such as TELEMAC), which require extracting valuable information (such as "areas with water depth > 0.5 meters" and "the proportion of grids with risk level ≥ 3") from massive amounts of raw grid data.

[0003] However, the number of grid points and cells in the original grid data can reach tens of millions or even hundreds of millions. When calculating the flooded area, it is necessary to traverse all grids to determine the threshold. Furthermore, for each time step, it is necessary to independently locate the relevant geometric grid points of the village to calculate the risk and flooded area. This results in a large number of redundant calculations, making it possible for the early warning data to take several hours to process at one time, which cannot meet the requirements of real-time early warning (minute-level response). It may even lead to memory overflow, causing data loss, business interruption, and affecting normal simulation operation, further reducing the efficiency of early warning.

[0004] The information disclosed in this background section is only for understanding the background of the inventive concept, and therefore may include information that does not constitute prior art. Summary of the Invention

[0005] In view of the above problems or one of the above problems, the purpose of this invention is to provide a method, system and device for efficient processing of digital twin watershed early warning data, which can significantly improve processing efficiency, reduce data redundancy, meet the needs of real-time early warning, and effectively solve the pain point of repetitive calculation and avoid memory overflow, thus realizing efficient processing of digital twin watershed early warning data.

[0006] To address the aforementioned problems or one of them, the second objective of this invention is to provide a method, system, and device for efficient processing of digital twin watershed early warning data. By reusing static geometric grids and storing dynamic hydrological parameters, this invention avoids the repetitive calculations required in traditional solutions where "each time step requires independent positioning of the geometric grid." This reduces the core time consumption for risk calculation and inundation area statistics from "hours" to "seconds," thereby meeting the operational requirement of "minute-level response" for watershed early warning.

[0007] To address the aforementioned problems or one of the aforementioned problems, the third objective of this invention is to provide a method, system, and device for efficient processing of digital twin watershed early warning data. This method employs a parallel mapping mechanism, dividing thread tasks according to time steps, with each thread independently completing the mapping of dynamic hydrological parameters for a single frame. This significantly shortens the processing cycle of massive amounts of data (tens of millions / hundreds of millions of grid points) and avoids the several hours of time consumption caused by serial traversal.

[0008] To achieve one of the above objectives, the first technical solution of the present invention is as follows:

[0009] A method for efficient processing of digital twin watershed early warning data includes the following:

[0010] Using a pre-created data splitting model, the original grid data of a digital twin watershed is obtained, and the original grid data is spatiotemporally separated to obtain static geometric grids and dynamic hydrological parameters;

[0011] The model is created by using a pre-created template. Based on a static geometric mesh, the river area is removed, and a template mesh with several non-river area mesh points is constructed.

[0012] Based on a pre-created data relocation model, a parallel mapping mechanism is used to map time-varying dynamic hydrological parameters to corresponding non-channel grid points to obtain dynamic template information, so as to reuse static geometric grids to store dynamic hydrological parameters.

[0013] Based on a pre-created risk-focusing model, threshold screening and integral algorithms are integrated to calculate hydrological parameters from dynamic template information, thereby obtaining high-risk early warning data and completing the efficient processing of digital twin watershed early warning data.

[0014] This invention separates the original grid data into static geometric grids and dynamic hydrological parameters by creating a data splitting model, a template creation model, a data relocation model, and a risk focusing model. Based on the static geometric grid, river areas are removed to construct a template grid with several non-river area grid points. Then, a parallel mapping mechanism is used to map the time-varying dynamic hydrological parameters to the corresponding non-river area grid points to obtain dynamic template information, which reuses the static geometric grid to store the dynamic hydrological parameters. Finally, threshold filtering and integration algorithms are combined to calculate hydrological parameters from the dynamic template information to obtain optimized high-risk early warning data. This significantly improves processing efficiency, reduces data redundancy, meets real-time early warning requirements, and effectively solves the problem of repetitive calculations and avoids memory overflow. Therefore, it can achieve efficient processing of digital twin watershed early warning data.

[0015] Furthermore, this invention reuses static geometric grids to store dynamic hydrological parameters, avoiding the repetitive calculations required in traditional schemes where "each time step requires independent positioning of the geometric grid." Taking 1441 frames of watershed data as an example, the core time consumption for risk calculation and inundation area statistics is compressed from "hours" to "seconds," thereby meeting the operational requirement of "minute-level response" for watershed early warning.

[0016] Meanwhile, the present invention adopts a parallel mapping mechanism, which divides thread tasks according to time steps. Each thread independently completes the mapping of dynamic hydrological parameters for a single frame, which greatly shortens the processing cycle of massive data (tens of millions / hundreds of millions of grid points) and avoids the time consumption of several hours caused by serial traversal.

[0017] As a preferred technical measure:

[0018] The method for obtaining static geometric grids and dynamic hydrological parameters by using a pre-created data splitting model to acquire raw grid data of a digital twin watershed and performing spatiotemporal separation on the raw grid data is as follows:

[0019] Read the raw grid data generated by the simulation software modeling system for a digital twin watershed;

[0020] Several time steps are obtained from the original grid data, and single frame data is obtained based on the time steps;

[0021] Dynamically identify the region attribute labels for each single frame of data and create a label array;

[0022] A static geometric mesh is constructed based on the read cell coordinates, cell connection relationships, and label array;

[0023] Based on a static geometric grid, multiple dynamic hydrological parameters are read from the original grid data;

[0024] Dynamic hydrological parameters are loaded as point data into each grid point in the static geometric grid.

[0025] As a preferred technical measure:

[0026] The method for creating a template mesh with several non-river channel grid points by using a pre-created template, based on a static geometric mesh, and removing river channel areas is as follows:

[0027] For each grid point in the static geometric mesh, add an original point identifier, which serves as a unique identifier for the grid point;

[0028] Based on the label array, obtain the region attribute label value for each grid point;

[0029] The regional attribute label value of grid points in the river channel area is -1, while the regional attribute label value of grid points in the non-river channel area is greater than -1.

[0030] Filter each grid point in the static geometric grid, filter out grid points with a region attribute label value equal to -1 to remove the river area, and only keep grid points in the non-river area, i.e., non-river grid points.

[0031] For non-river grid points, only the original point identifier is retained, dynamic hydrological parameters are removed, and a lightweight template grid is generated.

[0032] The origin point identifier is used to point to a location point in the static geometry mesh;

[0033] For each cross-section point on a non-channel cross-section, calculate the Euclidean distance between the cross-section point and each non-channel grid point;

[0034] Find the non-channel grid point with the shortest Euclidean distance and its number to obtain the shortest non-channel point;

[0035] Based on the spatial partitioning algorithm, the static geometric grid point set is divided into multiple sub-regions to accelerate nearest neighbor search;

[0036] Based on the village number, establish a hash mapping table between the village and the shortest non-river channel point, which will serve as the cross-section point mapping table;

[0037] The lightweight template mesh and the cross-section point mapping table are spliced ​​together to obtain a template mesh with several non-channel grid points.

[0038] As a preferred technical measure:

[0039] Based on a pre-created data relocation model, and utilizing a parallel mapping mechanism, dynamic hydrological parameters that change over time are mapped to corresponding non-channel grid points to obtain dynamic template information. The method is as follows:

[0040] Acquire multiple dynamic hydrological parameters that change over time, each of which includes hydrological physical quantities at several time steps;

[0041] A multi-threaded parallel processing architecture is adopted, which divides thread tasks according to time steps, with each thread processing a single frame of data independently and establishing a data mapping mechanism.

[0042] Based on the data mapping mechanism, non-river grid points are indexed by the original point identifier according to the village number. The hydrological and physical quantities of several time steps at the village level are attached to the same non-river grid point, so that each non-river grid point includes the hydrological and physical quantities of several time steps, thereby obtaining dynamic template information.

[0043] This invention separates static geometric meshes from dynamic hydrological parameters in time and space, so that the static geometric mesh stores only one set of mesh information (grid point coordinates and cell connection relationships), while the dynamic hydrological parameters reuse the static geometric mesh in the form of point data mounting. This avoids the repeated storage of geometric information in each frame of data and can effectively avoid memory overflow caused by the full loading of the original mesh, significantly reducing memory and storage costs and lowering the simulation hardware requirements.

[0044] As a preferred technical measure:

[0045] Based on a pre-created risk-focusing model, and by integrating threshold screening and integral algorithms, hydrological parameters are calculated from dynamic template information to obtain high-risk early warning data.

[0046] Extract village-level hydrological physical quantities from dynamic template information, including water depth, flow velocity, and risk level values;

[0047] Based on hydrological physical quantities, statistical values ​​of water depth and flow velocity are calculated, including average and maximum values.

[0048] The number of points with different risk levels is counted, and non-river grid points that exceed the risk threshold are filtered out according to the risk level to obtain the risk grid points;

[0049] Based on risk grid points, risk areas are delineated to reduce interference from low-risk points;

[0050] Calculate the sum, maximum, and average risk values ​​for each risk area, with the maximum and average values ​​serving as reference values ​​for risk decision-making in that risk area;

[0051] Set water depth thresholds based on risk decision reference values;

[0052] For risk areas, risk grid points with water depths greater than the water depth threshold are selected to obtain deep water grid points, which are used to reduce interference in shallow water areas;

[0053] Calculate the actual flooded area of ​​the village based on the deep-water grid points;

[0054] By summarizing the coordinates, water depth, flow velocity, risk level, and actual flooded area of ​​villages at deep-water grid points, high-risk early warning data is obtained.

[0055] The risk focus model of this invention integrates threshold screening and integral algorithms. First, it screens high-risk grid points according to risk level values, and then removes shallow water interference according to water depth thresholds. This ensures that the statistical risk data focuses only on key early warning targets, avoids low-risk data from affecting decision-making, and ensures the reliability of early warning decisions by accurately screening high-risk data, while reducing interference from low-risk data.

[0056] As a preferred technical measure:

[0057] Method for calculating the actual flooded area of ​​a village based on deep-water grid points:

[0058] Obtain several deep-water grid points;

[0059] Based on adjacent deep-water grid points, several triangular units are established;

[0060] Calculate the area of ​​each triangular unit and sum all the areas to obtain the actual submerged area of ​​the village.

[0061] As a preferred technical measure:

[0062] It also includes building a data transformation and storage model to optimize the accuracy of high-risk early warning data and for its storage, which includes the following:

[0063] Based on the simulation requirements of high-risk early warning data, determine the number of decimal places to retain;

[0064] Using the number of decimal places as the exponent and 10 as the base, we obtain the scaling factor;

[0065] Multiply the values ​​with decimal parts in the high-risk warning data by a scaling factor to obtain amplified data, thereby converting the decimal parts in the high-risk warning data into integer parts, resulting in rounded amplified data;

[0066] The rounded, magnified data is reduced back to its original scale to obtain high-risk early warning data with optimized accuracy, and this high-risk early warning data is then stored.

[0067] This invention uses a scaling factor to round data to the nearest integer and then restore it, replacing traditional string truncation. This allows for precise control of the decimal precision of parameters such as water depth, flow rate, and risk value, avoiding deviations in early warning indicators caused by computer floating-point storage errors, such as misjudging a risk value of 0.751 as 0.749.

[0068] To achieve one of the above objectives, the second technical solution of the present invention is as follows:

[0069] A method for efficient processing of digital twin watershed early warning data includes the following steps:

[0070] S1: Using the data reading model, parse the hydrodynamic simulation format file to obtain the original mesh data;

[0071] S2: Create a model using a template, generate a template mesh, and record the geometric information of the original points;

[0072] S3: Based on the data relocation model, a multi-threaded parallel mapping mechanism is used to map the original grid data to the template grid to obtain dynamic template information;

[0073] S4: Based on the risk focus model, the threshold screening and integral algorithm are integrated to calculate hydrological parameters from dynamic template information to obtain high-risk early warning data;

[0074] S5: Employs a data conversion and storage model, using the banker's algorithm to optimize the accuracy of high-risk early warning data and store it, thus achieving efficient processing of digital twin watershed early warning data.

[0075] This invention maps the original grid data to a template grid to reuse the original point geometric information, avoiding the repetitive calculations of "each time step requires independent positioning of the village geometric grid" in traditional schemes. This can significantly improve processing efficiency, reduce data redundancy, meet the needs of real-time early warning, and effectively solve the pain point of repetitive calculations and avoid memory overflow. Therefore, it can achieve efficient processing of digital twin watershed early warning data.

[0076] Furthermore, this invention employs a multi-threaded parallel mapping mechanism, dividing thread tasks according to time steps, with each thread independently completing the mapping of dynamic hydrological parameters for a single frame, significantly shortening the processing cycle of massive data (tens of millions / hundreds of millions of grid points) and avoiding hours of time consumption caused by serial traversal.

[0077] Furthermore, in the data conversion process, this invention employs the Banker's Algorithm to replace the string truncation method in order to retain a fixed number of decimal places, thereby achieving rounding to the nearest even number and thus obtaining more accurate data representation.

[0078] To achieve one of the above objectives, the third technical solution of the present invention is as follows:

[0079] A high-efficiency processing system for digital twin watershed early warning data includes a data splitting module, a template creation module, a data relocation module, and a risk focusing module;

[0080] The data splitting module is used to acquire the original grid data of a digital twin watershed and perform spatiotemporal separation on the original grid data to obtain static geometric grids and dynamic hydrological parameters.

[0081] The template creation module is used to construct a template mesh with several non-river channel grid points by removing river areas based on a static geometric mesh.

[0082] The data relocation module is used to map time-varying dynamic hydrological parameters to corresponding non-channel grid points using a parallel mapping mechanism to obtain dynamic template information, so as to reuse the static geometric grid and store dynamic hydrological parameters.

[0083] The risk focus module is used to integrate threshold screening and integral algorithms to calculate hydrological parameters from dynamic template information, obtain high-risk early warning data, and complete the efficient processing of digital twin watershed early warning data.

[0084] To achieve one of the above objectives, the fourth technical solution of the present invention is as follows:

[0085] A high-efficiency processing device for digital twin watershed early warning data, comprising:

[0086] One or more processing units;

[0087] Storage device for storing one or more programs;

[0088] When the one or more programs are executed by the one or more processing units, the one or more processing units implement the above-described method for efficient processing of digital twin watershed early warning data.

[0089] The storage device can be internal memory, external memory, cache memory, or other special memory.

[0090] The processing unit has signal processing capabilities and can be a general-purpose processor, a digital signal processor, an application-specific integrated circuit, an off-the-shelf programmable gate array, or other programmable logic devices.

[0091] Compared with existing technical solutions, the present invention has the following beneficial effects:

[0092] This invention can significantly improve processing efficiency, reduce data redundancy, meet the needs of real-time early warning, and effectively solve the pain point of repetitive calculations and avoid memory overflow, thus enabling efficient processing of digital twin watershed early warning data.

[0093] Furthermore, this invention reuses static geometric grids to store dynamic hydrological parameters, avoiding the repetitive calculations required in traditional schemes where "each time step requires independent positioning of the geometric grid." Taking 1441 frames of watershed data as an example, the core time consumption for risk calculation and inundation area statistics is compressed from "hours" to "seconds," thereby meeting the operational requirement of "minute-level response" for watershed early warning.

[0094] Meanwhile, the present invention adopts a parallel mapping mechanism, which divides thread tasks according to time steps. Each thread independently completes the mapping of dynamic hydrological parameters for a single frame, which greatly shortens the processing cycle of massive data (tens of millions / hundreds of millions of grid points) and avoids the time consumption of several hours caused by serial traversal. Attached Figure Description

[0095] Figure 1 This is a flowchart illustrating an efficient method for processing digital twin watershed early warning data according to the present invention.

[0096] Figure 2 A schematic diagram illustrating the computation time required to process data from a certain watershed using this invention;

[0097] Figure 3 This diagram illustrates the memory usage of a single operation for processing watershed data using this invention. Detailed Implementation

[0098] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this application. This invention covers any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of the invention as defined by the claims.

[0099] like Figure 1 As shown, the first specific embodiment of the efficient processing method for digital twin watershed early warning data of the present invention is as follows:

[0100] A method for efficient processing of digital twin watershed early warning data includes the following:

[0101] Using a pre-created data splitting model, the original grid data of a digital twin watershed is obtained, and the original grid data is spatiotemporally separated to obtain static geometric grids and dynamic hydrological parameters;

[0102] The model is created by using a pre-created template. Based on a static geometric mesh, the river area is removed, and a template mesh with several non-river area mesh points is constructed.

[0103] Based on a pre-created data relocation model, a parallel mapping mechanism is used to map time-varying dynamic hydrological parameters to corresponding non-channel grid points to obtain dynamic template information, so as to reuse static geometric grids to store dynamic hydrological parameters.

[0104] Based on a pre-created risk-focusing model, threshold screening and integral algorithms are integrated to calculate hydrological parameters from dynamic template information, thereby obtaining high-risk early warning data and completing the efficient processing of digital twin watershed early warning data.

[0105] A specific embodiment of the digital twin watershed early warning data high-efficiency processing system of the present invention:

[0106] A high-efficiency processing system for digital twin watershed early warning data includes a data splitting module, a template creation module, a data relocation module, and a risk focusing module;

[0107] The data splitting module is used to acquire the original grid data of a digital twin watershed and perform spatiotemporal separation on the original grid data to obtain static geometric grids and dynamic hydrological parameters.

[0108] The template creation module is used to construct a template mesh with several non-river channel grid points by removing river areas based on a static geometric mesh.

[0109] The data relocation module is used to map time-varying dynamic hydrological parameters to corresponding non-channel grid points using a parallel mapping mechanism to obtain dynamic template information, so as to reuse the static geometric grid and store dynamic hydrological parameters.

[0110] The risk focus module is used to integrate threshold screening and integral algorithms to calculate hydrological parameters from dynamic template information, obtain high-risk early warning data, and complete the efficient processing of digital twin watershed early warning data.

[0111] like Figure 1 As shown, this is the second specific embodiment of the efficient processing method for digital twin watershed early warning data of the present invention:

[0112] A method for efficient processing of digital twin watershed early warning data includes the following:

[0113] S1: Using the data reading model, the hydrodynamic simulation Serafin format file is parsed and adaptively converted into the VTK data structure to obtain the original mesh data.

[0114] S2: Create a model using a template, generate a dynamic template mesh, and record the original point index information.

[0115] S3: Based on the data relocation model, a multi-threaded parallel mapping mechanism is used to map the original grid data to the template grid to obtain template information.

[0116] S4: Based on the risk focus model and by integrating threshold screening and integral algorithms, hydrological parameters are calculated on the template information to obtain high-risk early warning data.

[0117] S5: Employs a data transformation and storage model, and uses the banker's algorithm to optimize the accuracy of high-risk early warning data and store it.

[0118] This invention is an optimization method for data processing of a watershed early warning digital twin platform based on the visualization toolkit VTK. It can significantly improve the computing speed, reduce the memory usage, and reduce hardware and maintenance costs.

[0119] In this embodiment, in S1, the Serafin format file generated by the TELEMAC modeling system of the simulation software is read. Serafin is a standard format for hydrodynamic / hydrological simulation data developed by the French National Laboratory for Water and Environment (IRSTEA). It is mainly used to store and exchange the calculation results of hydrological and hydrodynamic models, including grid geometric information (such as grid point coordinates and cell connection relationships), dynamic hydrological parameters (such as water depth, flow velocity, water level, etc.) and time step sequence data.

[0120] Then, the total number of time steps is obtained from the Serafin format file, and single-frame data is retrieved based on these time steps. The element type (e.g., triangle, tetrahedron, etc.) is dynamically identified, and an element array is created. A static geometric mesh, vtkUnstructuredGrid, is constructed based on the read coordinates and element information, as follows:

[0121] Define point data vtkPoints and cell connection relationships vtkCellArray, then associate them with a grid object and set the cell type to obtain a static geometric grid vtkUnstructuredGrid.

[0122] Then, dynamic hydrological parameters (such as flow velocity and water depth) at specific time steps are read from the file and mounted as point data into the static geometric grid vtkUnstructuredGrid. Serafin format files typically use node indices starting from 1, while VTK tools use indices starting from 0, so a -1 operation is performed in the code.

[0123] Cross-sectional data is stored in a CSV file. The CSV file is read, the numerical data in the file is split, and every three data points form the three-dimensional coordinates of a point, forming a cross-sectional point data, which is then stored in the structured mesh vtkPolyData of the cross-section. At the same time, the string data is parsed and used as the physical quantity of the cross-section point, which is also stored in the structured mesh vtkPolyData of the cross-section.

[0124] In this embodiment, in S2, a template mesh is created: since the geometric mesh is consistent across all frames, a static geometric mesh, vtkUnstructuredGrid, is selected for processing. An integer-type OriginalPointIds attribute is added to the original mesh points as a unique identifier for each point.

[0125] In the unstructured grid, sub-grids are extracted based on the VILLAGE value of the region attribute label, retaining only the OriginalPointIds attribute. The remaining dynamic hydrological parameters are then removed from the filtered grids to generate a lightweight template, VILLAGEGrid. The purpose of retaining the OriginalPointIds is to indicate the location of points on the template within the overall static geometric grid vtkUnstructuredGrid.

[0126] For each point on the cross-section vtkPolyData, calculate the cross-section point. Grid points on the static geometric mesh vtkUnstructuredGrid Euclidean distance The calculation formula is as follows:

[0127]

[0128] in, For grid points The coordinate values; Cross-section point The coordinate values.

[0129] Find the nearest point ID (PointID) of the static geometric grid vtkUnstructuredGrid based on Euclidean distance. Divide the vtkUnstructuredGrid point set into small regions according to a spatial partitioning algorithm to accelerate nearest neighbor search. Referring to village numbers, establish a hash mapping table (pointIDMap) of grid point identifiers (vid-pointIDs) as a cross-section point mapping table to achieve a query with a time complexity of O(1).

[0130] Based on the VILLAGE value of the regional attribute label, the VILLAGE value of points in the river area is -1, while the VILLAGE value of points in the non-river area is greater than -1. Each point on the unstructured grid is filtered out, and the points in the region with the VILLAGE value equal to -1 are filtered out. Only the grid data of the non-river area is retained, and a template grid for the non-river area is generated.

[0131] In this embodiment, S3 employs a multi-threaded parallel processing architecture, dividing thread tasks according to time steps, with each thread independently processing a single frame of data. A data mapping mechanism is established.

[0132] Based on the village number, the original grid points are indexed by the original point identifier OriginalPointIds. Physical quantities (such as water depth, flow velocity, etc.) are copied to the village-specific lightweight template grid VILLAGEGrid, and a village-level scalar dataset CAScalarData is created to store the calculation results for each village.

[0133] Simultaneously, each point in the template grid of the non-channel area is traversed, and the point with the same identifier is located in the point dataset of the original grid through its stored OriginalPointIds value (which records the correspondence between the original grid point and the original grid point). This point is the original mapping point corresponding to the template grid point, and then the water depth value of the point is obtained to form the data grid of the non-channel area.

[0134] In this embodiment, in S4, risk perception statistical calculations are performed for each frame, including the following:

[0135] For village-level statistics, based on the template grid VILLAGEGrid with physical quantities, memory pointers directly access the VTK data array to extract and analyze hydrological data for the village area from the VTK data structure. Statistical values ​​(average and maximum) of water depth and flow velocity, as well as the number of points at different risk levels, are calculated and stored in the village-level scalar dataset CAScalarData. Based on the risk level hr values ​​from simulation results, points exceeding the risk threshold (0.6-0.8) are selected to define risky areas and prevent low-risk points from interfering with the statistics.

[0136] Next, based on whether the number of selected regional points is sufficiently representative, statistical analysis is performed using these points. For example, the sum, maximum, and average risk hr values ​​of each point are calculated. The maximum and average values ​​are used as risk decision reference values ​​for the region and stored in the village-level scalar dataset CAScalarData. Only point sets and grids with water depths greater than or equal to the water depth threshold (0.08m-0.15m) are processed, ignoring interference from shallow water areas. The area of ​​the region is calculated as the actual flooded area of ​​the village and stored in the village-level scalar dataset CAScalarData.

[0137] For data extraction from cross-section points, water level values ​​are obtained through the cross-section point mapping table pointIDMap and stored in the village-level scalar dataset CAScalarData.

[0138] The calculation of the total inundation area requires data from a non-channel region grid. The grid cell integration module vtkIntegrateAttributes is used to calculate the integral, yielding the total inundation area. The area calculation principle is based on a triangular calculation mechanism. The calculation formula is as follows:

[0139]

[0140] in, , , These are the three vertices of the triangle.

[0141] For the entire mesh, the total area / volume is the sum of the areas / volumes of all cells, expressed as follows:

[0142]

[0143] in, The total number of units, For the first The area of ​​each unit.

[0144] In this embodiment, in step S5, a frame of geometric mesh (point coordinates and topology) is selected and saved to a JSON file. Based on the village-level scalar dataset CAScalarData and time step information, a data JSON file is written and stored locally. Simultaneously, the point data (excluding coordinate data) of each frame of static geometric mesh vtkUnstructuredGrid is directly converted into physical quantity JSON and written locally.

[0145] During data conversion, to retain a fixed number of decimal places, the Banker's Algorithm is used instead of the string truncation method to achieve rounding (rounding to even numbers) and more accurate data representation. The specific processing flow is as follows:

[0146] Multiply the original data value by the scaling factor (N decimal places to retain) to obtain the magnified data. This converts the decimal places that need to be retained into integer parts, and the calculation formula is as follows:

[0147]

[0148] If the decimal part is less than 0.5, the decimal part is discarded; if the decimal part is greater than 0.5, the decimal part is rounded up by 1; if the decimal part is equal to 0.5, the decimal part is rounded to the nearest even number.

[0149] The rounded integer is reduced back to its original scale to obtain the result with the specified precision. The calculation formula is as follows:

[0150]

[0151] This invention employs dynamic templated meshing technology, establishing a mapping relationship between the original mesh and the template through the original point identifier OriginalPointIds, thereby achieving lightweight data reuse. Simultaneously, the template dynamically divides the mesh using the VILLAGE physical quantity as a region attribute label, avoiding redundant calculations.

[0152] Based on a heterogeneous data parallel relocation mechanism, multi-threaded data of templates is mapped, overcoming the bottleneck of traditional serial processing. Furthermore, based on a risk-driven statistical calculation model, a high-risk area-focused statistical method using threshold screening improves computational efficiency.

[0153] Finally, the Banker's Algorithm is used to normalize the data, thus solving the problem of floating-point precision loss in scientific computing.

[0154] Based on the above points, the amount of computation is reduced, the calculation and conversion speed is accelerated, the memory usage during the operation process is reduced, and the requirements for hardware and operating environment are lowered, thus achieving cost reduction and efficiency improvement.

[0155] This application is compatible with reading Serafin format files and can directly convert static geometric meshes (vtkUnstructuredGrid) into single-frame geometric mesh JSON (containing the geometric information of the static geometric mesh) and multi-frame physical quantity JSON (dynamic hydrological parameters at different times). This eliminates the need for repeated inclusion of mesh geometric information within a single rendering file, reducing data redundancy. Geometric information includes point coordinates and cell connectivity; dynamic hydrological parameters include water depth and flow velocity at each point.

[0156] Simultaneously, the static geometric mesh is directly converted into a renderable data structure. The geometric information can be used by the engine to construct the spatial form of the 3D mesh, and dynamic hydrological parameters can be directly mapped to rendering attributes such as color and transparency. No additional mesh type conversion or data format parsing is required; scene rendering can be performed directly based on this JSON, thus achieving efficient rendering.

[0157] In terms of time, the use of multi-threaded parallelism and template meshes significantly reduces the time consumption and computational difficulty, improving the time required from minutes or even hours for conventional traversal to seconds. (See [reference needed]). Figure 2 .

[0158] In terms of performance, over a thousand frames of data are directly stored and post-processed using static geometric mesh containers, requiring no additional processing. This avoids the problem of excessive memory consumption during data conversion, has low hardware performance requirements, and reduces investment costs. (See [link to relevant documentation]). Figure 3 .

[0159] Taking 1441 frames of data from a certain watershed as an example: the size of the Serafin format file is 5.8G, and the size of the dumped MED file (a commonly used data format for watershed simulation) is 13.5G.

[0160] A specific embodiment of applying the method of the present invention to process data in a watershed:

[0161] The method for processing data of a certain watershed using the efficient digital twin watershed early warning data processing method of this invention is as follows:

[0162] Step 1: Read the Serafin format file result.slf generated by the TELEMAC modeling system software.

[0163] Then, the Serafin format file result.slf is processed to obtain the total number of time steps and the static geometric mesh vtkUnstructuredGrid data within each time step.

[0164] The static geometric mesh consists of the geometric information `ReadGeometry` and the physical quantity information `ReadData`. A numerical array `vtkDoubleArray` is created to store node coordinates, storing coordinate values ​​read from a file. Simultaneously, the topology stored in the file is parsed, and the element type is determined to be triangular based on the number of nodes in each element. An element array is then created, ultimately forming the mesh. The numerical array `vtkDoubleArray` is then used to store the physical quantity values ​​parsed from the file, adding them to the point data. Finally, a data array `vector` is obtained. <vtkunstructuredgrid>.

[0165] Read the cross-section point data stored in the CSV file. The data includes the physical quantity identifier vid and the three-dimensional coordinates of the points.

[0166] Step 2: The number of points and cells in each frame of the static geometric mesh vtkUnstructuredGrid is consistent, as is the topology; that is, the geometric data of the mesh is consistent. Therefore, the first frame of the static geometric mesh vtkUnstructuredGrid is selected as the data template. An identifier array vtkIdTypeArray is created, named OriginalPointIds, with 1 component, a length equal to the number of points, and a value representing the point's unique position within the complete mesh.

[0167] Based on the known village ID array and region attribute labels VILLAGEIDs, the threshold segmentation function vtkThreshold is used to locate each village in the static geometric grid vtkUnstructuredGrid, extract the corresponding grid data, and form the village basic template VILLAGEGrids. For each grid point in the village basic template VILLAGEGrids, dynamic hydrological parameters are removed, and the point number, i.e., the original point identifier OriginalPointIds, is retained, resulting in the final village-specific template VILLAGEGrids.

[0168] For each point in the cross-section vtkPolyData, the spatial indexing tool vtkPointLocator is used to obtain the pointIDs of the nearest corresponding point on the grid, with the target grid being the background geometry grid vtkUnstructuredGrid. Based on the previously obtained pointIDs, the points are grouped according to the village ID's regional attribute label VILLAGEIDs. That is, when the cross-section point identifier vid equals the regional attribute label VILLAGEID, the point corresponding to that pointID is a cross-section point on that village. This process yields the final village cross-section point retrieval array pointIDMap.

[0169] The VILLAGE attribute label for physical quantities not only marks the village's ID, but also the numbers of the river and other areas. For the calculation of the flooded area in the whole region, the river area needs to be removed. The threshold segmentation function vtkThreshold can be used to filter out the grid data of the river area from the static geometric grid vtkUnstructuredGrid to obtain the non-river data grid riverRemoveGird. Similarly, dynamic hydrological parameters are removed, and the point numbers are retained as the template grid.

[0170] Step 3: Create 16 threads to process subsequent steps of each frame of static geometric mesh vtkUnstructuredGrid data in parallel using multithreading, and store the calculation results for each village using the village-level scalar dataset CAScalarData.

[0171] The village-level scalar dataset CAScalarData includes time, maximum water depth, average water depth, maximum flow velocity, average flow velocity, average risk value, maximum risk value, inundated area, number of points with different hr values, and cross-sectional water level values.

[0172] Based on the village number, copy the village template mesh and create a new array of physical quantities. Names and component counts are created according to the frame data. Point counts are obtained from the static geometric mesh vtkUnstructuredGrid, based on the village mesh template. The physical quantities are copied according to the original point number mapping, resulting in a data region attribute label VILLAGEdata based on the village template and containing the corresponding frame physical quantities.

[0173] Similarly, copy the non-river region template riverRemoveGird, and based on the original point number mapping, copy the water depth physical quantity from vtkUnstructuredGrid to form the non-river region data grid riverRemoveData.

[0174] Step 4: Access VTK data using memory pointers, iterate through the points, sum the values ​​of each physical quantity to obtain the total sum, then divide by the number of points to obtain the average value of the physical quantity. Simultaneously, count the risk values ​​(hr) for different risk levels based on the hr value. Filter points with risk values ​​greater than 0.75 to avoid low-risk points interfering with the results. If the total number of filtered points is greater than or equal to 4, the data used for calculating the maximum and average values ​​of the risk values ​​comes from the filtered grid points. Based on the physical quantity WATERDEPTH, filter areas that meet the water depth requirement of greater than 0.1, and determine the flooded area of ​​the village based on the area of ​​this region.

[0175] Based on village grouping, find the cross-sectional points of the villages, obtain the cross-sectional water level value of the point according to the point number, and store it in the village-level scalar dataset CAScalarData.

[0176] Then calculate the total inundated area of ​​the entire region and obtain the inundation ratio.

[0177] Step 5: Since the grid point positions are consistent across all frames of the static geometric grid vtkUnstructuredGrid, meaning the geometric units are identical, a separate JSON file containing the data for each geometric unit is generated and saved locally to reduce data redundancy. Because the point coordinate values ​​are of type double, the data type casting function `static_cast` and the batch data assignment function `assign` are directly called for memory copying, reducing data conversion time and minimizing data precision loss. Similarly, the village-level scalar dataset `CAScalarData` and the total flooding data are also converted and written to the JSON file.

[0178] The dynamic hydrological parameters of each frame are then converted into frame data separately. During the conversion process, decimal places are retained. The floating-point rounding mode is controlled by C standard library functions. The rounding mode is set to the Banker's algorithm. The "round to even" rule is used to obtain amplified data. Then, the amplified data is reduced back to the original scale to obtain the required early warning data, thus achieving efficient processing of watershed early warning data.

[0179] A server embodiment applying the method of the present invention:

[0180] A server comprising:

[0181] One or more processing units;

[0182] Storage device for storing one or more programs;

[0183] When the one or more programs are executed by the one or more processing units, the one or more processing units implement the above-described method for efficient processing of digital twin watershed early warning data.

[0184] The storage device can be internal memory, external memory, cache memory, or other special memory. The processing unit has signal processing capabilities and can be a general-purpose processor, a digital signal processor, an application-specific integrated circuit, an off-the-shelf programmable gate array, or other programmable logic device.

[0185] An embodiment of a device applying the method of the present invention:

[0186] An electronic device is provided with a computer-readable storage medium on which a computer program is stored. When the program is executed by a processing unit, it implements the above-described method for efficient processing of digital twin watershed early warning data.

[0187] Computer-readable storage media refers to physical carriers capable of storing computer-recognizable data, instructions, or programs. These media must meet the core characteristic of being "readable by a computer" (i.e., the data exists in the form of electrical, magnetic, or optical signals and can be converted into binary information that a computer can process through appropriate devices). The physical carrier can be a magnetic storage medium, optical storage medium, semiconductor storage medium, or other storage media.

[0188] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, optical storage, etc.) containing computer-usable program code.

[0189] The model in this application is an object that uses physical or virtual representation to objectively describe the form and structure. The object is not the same as a physical object, and is not limited to physical or virtual. It can be a data processing function, software program, processing mode, usage method, operation mode, workflow, application process, electronic hardware, circuit module, processing system, system imitation or simulation object.

[0190] Finally, it should be noted that the above-described embodiments are merely specific implementations of the present invention, used to illustrate the technical solutions of the present invention, and not to limit it. The scope of protection of the present invention is not limited thereto. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features within the scope of the technology disclosed in the present invention; and these modifications or substitutions will not cause the substance of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. Any modifications or equivalent substitutions that do not deviate from the spirit and scope of the present invention should be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.< / vtkunstructuredgrid>

Claims

1. A method for efficient processing of digital twin watershed early warning data, characterized in that: Includes the following: Using a pre-created data splitting model, the original grid data of a digital twin watershed is obtained, and the original grid data is spatiotemporally separated to obtain static geometric grids and dynamic hydrological parameters; The model is created by using a pre-created template. Based on a static geometric mesh, the river area is removed, and a template mesh with several non-river area mesh points is constructed. Based on a pre-created data relocation model, a parallel mapping mechanism is used to map time-varying dynamic hydrological parameters to corresponding non-channel grid points to obtain dynamic template information, so as to reuse static geometric grids to store dynamic hydrological parameters. Based on a pre-created risk-focusing model, threshold screening and integral algorithms are combined to calculate hydrological parameters from dynamic template information, thereby obtaining high-risk early warning data. The method for obtaining high-risk early warning data is as follows: Extract village-level hydrological physical quantities from dynamic template information, including water depth, flow velocity, and risk level values; Based on hydrological physical quantities, statistical values ​​of water depth and flow velocity are calculated, including average and maximum values. The number of points with different risk levels is counted, and non-river grid points that exceed the risk threshold are filtered out according to the risk level to obtain the risk grid points; Based on risk grid points, risk areas are delineated to reduce interference from low-risk points; Calculate the sum, maximum, and average risk values ​​for each risk area, with the maximum and average values ​​serving as reference values ​​for risk decision-making in that risk area; Set water depth thresholds based on risk decision reference values; For risk areas, risk grid points with water depths greater than the water depth threshold are selected to obtain deep water grid points, which are used to reduce interference in shallow water areas; Calculate the actual flooded area of ​​the village based on the deep-water grid points; By summarizing the coordinates, water depth, flow velocity, risk level, and actual flooded area of ​​villages at deep-water grid points, high-risk early warning data is obtained, thus completing the efficient processing of digital twin watershed early warning data.

2. The efficient processing method for digital twin watershed early warning data as described in claim 1, characterized in that: The method for obtaining static geometric grids and dynamic hydrological parameters by using a pre-created data splitting model to acquire raw grid data of a digital twin watershed and performing spatiotemporal separation on the raw grid data is as follows: Read the raw grid data generated by the simulation software modeling system for a digital twin watershed; Several time steps are obtained from the original grid data, and single frame data is obtained based on the time steps; Dynamically identify the region attribute labels for each single frame of data and create a label array; A static geometric mesh is constructed based on the read cell coordinates, cell connection relationships, and label array; Based on a static geometric grid, multiple dynamic hydrological parameters are read from the original grid data; Dynamic hydrological parameters are loaded as point data into each grid point in the static geometric grid.

3. The efficient processing method for digital twin watershed early warning data as described in claim 1, characterized in that: The method for creating a template mesh with several non-river channel grid points by using a pre-created template, based on a static geometric mesh, and removing river channel areas is as follows: For each grid point in the static geometric mesh, add an original point identifier, which serves as a unique identifier for the grid point; Based on the label array, obtain the region attribute label value for each grid point; The regional attribute label value of grid points in the river channel area is -1, while the regional attribute label value of grid points in the non-river channel area is greater than -1. Filter each grid point in the static geometric grid, filter out grid points with a region attribute label value equal to -1 to remove the river area, and only keep grid points in the non-river area, i.e., non-river grid points. For non-river grid points, only the original point identifier is retained, dynamic hydrological parameters are removed, and a lightweight template grid is generated. The origin point identifier is used to point to a location point in the static geometry mesh; For each cross-section point on a non-channel cross-section, calculate the Euclidean distance between the cross-section point and each non-channel grid point; Find the non-channel grid point with the shortest Euclidean distance and its number to obtain the shortest non-channel point; Based on the spatial partitioning algorithm, the static geometric grid point set is divided into multiple sub-regions to accelerate nearest neighbor search; Based on the village number, establish a hash mapping table between the village and the shortest non-river channel point, which will serve as the cross-section point mapping table; The lightweight template mesh and the cross-section point mapping table are spliced ​​together to obtain a template mesh with several non-channel grid points.

4. The efficient processing method for digital twin watershed early warning data as described in claim 1, characterized in that: Based on a pre-created data relocation model, and utilizing a parallel mapping mechanism, dynamic hydrological parameters that change over time are mapped to corresponding non-channel grid points to obtain dynamic template information. The method is as follows: Acquire multiple dynamic hydrological parameters that change over time, each of which includes hydrological physical quantities at several time steps; A multi-threaded parallel processing architecture is adopted, which divides thread tasks according to time steps, with each thread processing a single frame of data independently and establishing a data mapping mechanism. Based on the data mapping mechanism, non-river grid points are indexed by the original point identifier according to the village number. The hydrological and physical quantities of several time steps at the village level are attached to the same non-river grid point, so that each non-river grid point includes the hydrological and physical quantities of several time steps, thereby obtaining dynamic template information.

5. The efficient processing method for digital twin watershed early warning data as described in claim 1, characterized in that: Method for calculating the actual flooded area of ​​a village based on deep-water grid points: Obtain several deep-water grid points; Based on adjacent deep-water grid points, several triangular units are established; Calculate the area of ​​each triangular unit and sum all the areas to obtain the actual submerged area of ​​the village.

6. The efficient processing method for digital twin watershed early warning data as described in claim 1, characterized in that: It also includes building a data transformation and storage model to optimize the accuracy of high-risk early warning data and for its storage, which includes the following: Based on the simulation requirements of high-risk early warning data, determine the number of decimal places to retain; Using the number of decimal places as the exponent and 10 as the base, we obtain the scaling factor; Multiply the values ​​with decimal parts in the high-risk warning data by a scaling factor to obtain amplified data, thereby converting the decimal parts in the high-risk warning data into integer parts, resulting in rounded amplified data; The rounded, magnified data is reduced back to its original scale to obtain high-risk early warning data with optimized accuracy, and this high-risk early warning data is then stored.

7. A method for efficient processing of digital twin watershed early warning data, characterized in that: Includes the following steps: S1: Using the data reading model, parse the hydrodynamic simulation format file to obtain the original mesh data; S2: Create a model using a template, generate a template mesh, and record the geometric information of the original points; It includes the following: Spatiotemporal separation is performed on the original grid data to obtain static geometric grids and dynamic hydrological parameters; Add an integer-type original point identifier attribute to the original grid points as an identifier to ensure the uniqueness of the points; In unstructured grids, subgrids are extracted based on regional attribute label values, retaining only the original point identifier attribute. The remaining dynamic hydrological parameters are removed from the filtered grids to generate a lightweight template. Filter each point on the unstructured grid, filter out the points in the region with the region attribute label equal to -1, and keep only the grid data of the non-river area to generate a template grid exclusive to the non-river area. S3: Based on the data relocation model, a multi-threaded parallel mapping mechanism is used to map the original grid data to the template grid to obtain dynamic template information; S4: Based on the risk-focused model, and integrating threshold screening and integral algorithms, hydrological parameters are calculated from dynamic template information to obtain high-risk early warning data; this includes the following: Extract village-level hydrological physical quantities from dynamic template information, including water depth, flow velocity, and risk level values; Based on hydrological physical quantities, statistical values ​​of water depth and flow velocity are calculated, including average and maximum values. The number of points with different risk levels is counted, and non-river grid points that exceed the risk threshold are filtered out according to the risk level to obtain the risk grid points; Based on risk grid points, risk areas are delineated to reduce interference from low-risk points; Calculate the sum, maximum, and average risk values ​​for each risk area, with the maximum and average values ​​serving as reference values ​​for risk decision-making in that risk area; Set water depth thresholds based on risk decision reference values; For risk areas, risk grid points with water depths greater than the water depth threshold are selected to obtain deep water grid points, which are used to reduce interference in shallow water areas; Calculate the actual flooded area of ​​the village based on the deep-water grid points; By summarizing the coordinates, water depth, flow velocity, risk level, and actual flooded area of ​​villages at deep-water grid points, high-risk early warning data is obtained. S5: Employs a data conversion and storage model, using the banker's algorithm to optimize the accuracy of high-risk early warning data and store it, thus achieving efficient processing of digital twin watershed early warning data.

8. A high-efficiency processing system for digital twin watershed early warning data, characterized in that: It includes a data splitting module, a template creation module, a data relocation module, and a risk focusing module; The data splitting module is used to acquire the original grid data of a digital twin watershed and perform spatiotemporal separation on the original grid data to obtain static geometric grids and dynamic hydrological parameters. The template creation module is used to construct a template mesh with several non-river channel grid points by removing river areas based on a static geometric mesh. The data relocation module is used to map time-varying dynamic hydrological parameters to corresponding non-channel grid points using a parallel mapping mechanism to obtain dynamic template information, so as to reuse the static geometric grid and store dynamic hydrological parameters. The risk focus module is used to integrate threshold screening and integral algorithms to calculate hydrological parameters from dynamic template information and obtain high-risk early warning data. It includes the following: Extract village-level hydrological physical quantities from dynamic template information, including water depth, flow velocity, and risk level values; Based on hydrological physical quantities, statistical values ​​of water depth and flow velocity are calculated, including average and maximum values. The number of points with different risk levels is counted, and non-river grid points that exceed the risk threshold are filtered out according to the risk level to obtain the risk grid points; Based on risk grid points, risk areas are delineated to reduce interference from low-risk points; Calculate the sum, maximum, and average risk values ​​for each risk area, with the maximum and average values ​​serving as reference values ​​for risk decision-making in that risk area; Set water depth thresholds based on risk decision reference values; For risk areas, risk grid points with water depths greater than the water depth threshold are selected to obtain deep water grid points, which are used to reduce interference in shallow water areas; Calculate the actual flooded area of ​​the village based on the deep-water grid points; By summarizing the coordinates, water depth, flow velocity, risk level, and actual flooded area of ​​villages at deep-water grid points, high-risk early warning data is obtained, thus completing the efficient processing of digital twin watershed early warning data.

9. A high-efficiency processing device for digital twin watershed early warning data, characterized in that: It includes: One or more processing units; Storage device for storing one or more programs; When the one or more programs are executed by the one or more processing units, the one or more processing units implement a method for efficient processing of digital twin watershed early warning data as described in any one of claims 1-7.

Images