Water conservancy sensing system based on edge calculation

Abstract

The invention relates to the technical field of energy efficiency management, in particular to a water conservancy sensing system based on edge computing. The system comprises the following modules: a riverway edge sensing module, a water conservancy characteristic analysis module, a sediment migration simulation module and a deposition scouring early warning module. Constructing a river channel water conservancy sensing network; performing real-time water conservancy monitoring data acquisition according to the river channel water conservancy sensing network to obtain river channel water conservancy sensing data; performing three-dimensional flow field model processing on the river channel water conservancy sensing data to obtain a three-dimensional sediment flow field network model; carrying out particle flow field simulation on the three-dimensional sediment flow field network model to generate sediment migration space-time distribution data; and according to the sediment migration space-time distribution data, carrying out siltation and scour area identification, and uploading the siltation and scour area data to a cloud platform through an edge computing node to carry out abnormal event alarm so as to obtain river section monitoring data. According to the invention, real-time, accurate and intelligent monitoring and early warning of the sediment deposition and scouring process of the river channel are realized.

Description

Technical Field

[0001] The present invention relates to the field of water conservancy measurement technology, and in particular to a water conservancy sensing system based on edge computing. Background Art

[0002] Rivers are an important part of water resources, and they undertake multiple functions such as water supply, navigation, and ecology. Sedimentation and scouring are the basic processes of river evolution, which have a profound impact on river morphology, navigation capacity, water ecological environment, and the safety of water conservancy projects. Therefore, accurately and timely grasping the dynamics of river sedimentation and scouring is of great significance for ensuring the safety of river flood discharge, maintaining the smooth flow of waterways, optimizing water resources allocation, and guiding the design and management of water conservancy projects. It is crucial to monitor and effectively manage the process of river sedimentation and scouring in real time. However, traditional water conservancy sensing systems mainly rely on manual measurement, regular sampling and analysis, or the use of fixed hydrological stations. Manual measurement and sampling analysis are time-consuming and labor-intensive, and it is difficult to achieve high-frequency real-time monitoring. Especially in river bends, narrow river sections, or areas with complex riverbed structures, the degree of sedimentation and scouring often varies greatly, and traditional monitoring points are difficult to capture these local changes; in addition, traditional hydrological measurement data usually need to go through manual processing, analysis, and reporting, which takes a long time and cannot meet the needs of real-time monitoring and early warning. This results in a delayed response to changes in river sediment, making it difficult to achieve refined management and risk prevention of water conservancy projects. Summary of the invention

[0003] Based on this, the present invention provides a water conservancy sensing system based on edge computing to solve at least one of the above technical problems.

[0004] To achieve the above objectives, a water conservancy perception system based on edge computing includes the following modules: The river channel edge perception module includes a high-definition camera, an acoustic Doppler current profiler, and a laser radar scanner, which is used to connect the high-definition camera, the acoustic Doppler current profiler, and the laser radar scanner to edge computing nodes to build a river channel water conservancy perception network; real-time water conservancy monitoring data is collected based on the river channel water conservancy perception network to obtain river channel water conservancy perception data; The water conservancy characteristic analysis module is used to synchronize the time and space of the river water conservancy sensing data to obtain the spatiotemporal river channel sensing data; perform water conservancy monitoring characteristic analysis based on the spatiotemporal river channel sensing data to obtain real-time cross-section monitoring characteristic data; perform three-dimensional flow field model processing based on the real-time cross-section monitoring characteristic data to obtain a three-dimensional sediment flow field network model; The sediment migration simulation module is used to simulate the particle flow field of the three-dimensional sediment flow field network model and generate particle trajectory data; the sediment migration analysis is performed based on the particle trajectory data to generate the sediment migration time and space distribution data; The siltation and scour warning module is used to identify the siltation and scour areas based on the spatiotemporal distribution data of sediment migration and obtain the siltation and scour area data; upload the siltation and scour area data to the cloud platform through the edge computing node to alarm for abnormal events and obtain the river section monitoring data.

[0005] Preferably, the river edge perception module includes the following functions: Obtaining the elevation terrain data of the target river; Mark the key sections of the target river elevation terrain data and generate key section location data; According to the key section location data, high-definition cameras, acoustic Doppler flow profilers and lidar scanners are installed at the key sections of the river to obtain the layout data of water conservancy sensing equipment; Based on the deployment data of water conservancy sensing equipment, edge computing nodes of high-definition cameras, acoustic Doppler current profilers and lidar scanners are connected to build a river water conservancy sensing network; Real-time water conservancy monitoring data is collected based on the river water conservancy sensing network to obtain river water conservancy sensing data.

[0006] Preferably, the marking of key sections of the target river elevation terrain data includes: Extract the centerline of the target river channel from the elevation terrain data and identify the left and right riverbanks to obtain the geometric constraint data of the river channel; Based on the river channel geometric constraint data, the preset equal spacing threshold is used to divide the section perpendicular to the river channel centerline to obtain the river channel section data; The river section data is spatially superimposed through the target river elevation terrain data to generate river section elevation data; Perform cross-sectional morphology analysis based on the river section elevation data to obtain cross-sectional geometric parameters; Calculate the cross-sectional shape coefficient according to the cross-sectional geometric shape parameters to generate the cross-sectional shape coefficient; Based on the preset key section identification threshold, the river section data is screened for key sections through the section morphology coefficient to generate key section location data.

[0007] Preferably, the water conservancy feature analysis module includes the following functions: Perform time synchronization and spatial registration on river water conservancy sensing data to obtain spatiotemporal river sensing data; The spatiotemporal river channel sensing data is transmitted to the edge computing node, and the data is preprocessed to generate water conservancy monitoring data; wherein the water conservancy monitoring data includes cross-sectional water body image data, cross-sectional flow velocity profile data, and cross-sectional riverbed topography data; Conduct water conservancy monitoring feature analysis based on water conservancy monitoring data to obtain real-time cross-section monitoring feature data; Construct a three-dimensional spatial network based on the target river elevation terrain data to generate a three-dimensional spatial model of the river; The three-dimensional flow field model of the river channel is processed by real-time cross-section monitoring feature data to obtain a three-dimensional sediment flow field network model.

[0008] Preferably, the water conservancy monitoring feature analysis includes: Perform sediment particle texture recognition on the cross-sectional water body image data to obtain sediment particle texture data; Perform particle size statistics based on sediment particle texture data and calculate the mean particle size to generate real-time sediment particle size data; Fitting the near-bottom velocity distribution curve to the cross-sectional velocity profile data to obtain the near-bottom velocity distribution curve data; Based on the cross-sectional riverbed topography data and using the near-bottom velocity distribution curve data, the riverbed bottom shear stress is calculated to generate bottom shear stress data; According to the cross-section velocity profile data and bottom shear stress data, the cross-section real-time concentration is estimated to generate the cross-section sediment concentration distribution data; The real-time sediment particle size data, cross-section sediment concentration distribution data and bottom shear stress data are integrated into water conservancy monitoring characteristics to generate real-time cross-section monitoring characteristic data.

[0009] Preferably, the cross-section real-time concentration estimation includes: According to the cross-sectional velocity profile data, the echo intensity is corrected and the suspended sediment reference concentration is calculated to generate the suspended sediment reference concentration data; The sediment settling velocity data is queried by using the preset sediment particle size-sedimentation velocity relationship database to obtain the sediment settling velocity data; Calculate the probability of sediment initiation based on sediment settling velocity data and bottom shear stress data to generate sediment initiation probability data; The sediment suspended concentration is estimated based on the suspended sediment reference concentration data and the sediment initiation probability data, and the cross-sectional concentration distribution is processed to generate the cross-sectional sediment concentration distribution data.

[0010] Preferably, the three-dimensional flow field model processing of the three-dimensional spatial model of the river channel includes: Perform cross-section three-dimensional coordinate transformation based on cross-section sediment concentration distribution data to generate cross-section concentration position data; The three-dimensional spatial model of the river channel is interpolated through the cross-section concentration position data to obtain a three-dimensional sediment concentration network model; The bottom boundary conditions of the riverbed are set according to the cross-section riverbed topography data and the bottom shear stress data, and the bottom boundary condition data are generated; Based on the bottom boundary condition data and the three-dimensional sediment concentration network model, a three-dimensional flow field numerical simulation model is constructed through the Navier-Stokes equation to generate a three-dimensional sediment flow field network model.

[0011] Preferably, the sediment migration simulation module includes the following functions: Calculate the grid unit volume based on the three-dimensional sediment flow field network model to generate grid unit volume data; Mark the cross-section monitoring points of the three-dimensional sediment flow field network model, set the virtual particle release position, and obtain the virtual particle release position data; The particle number distribution of the three-dimensional sediment flow field network model is carried out through the cross-section sediment concentration distribution data to obtain the virtual particle number data; Based on the virtual particle release position data and virtual particle quantity data, the particle flow field simulation is performed on the three-dimensional sediment flow field network model, and Lagrangian particle tracking is performed to generate particle trajectory data; Sediment migration analysis is performed based on particle trajectory data to generate spatiotemporal distribution data of sediment migration.

[0012] Preferably, the performing sediment migration analysis based on particle trajectory data comprises: Perform grid particle average displacement vector statistics based on particle trajectory data to generate grid average displacement data; Calculate the sediment migration flux based on the grid average displacement data to generate sediment migration flux data; The theoretical particle sediment carrying capacity is converted according to the sediment migration flux data to obtain the theoretical particle sediment carrying capacity; The sediment migration rate is calculated for the three-dimensional sediment flow field network model through the sediment migration flux data to generate sediment migration rate data; The theoretical particle sediment carrying capacity and sediment migration rate data are integrated into the grid spatiotemporal distribution to generate the spatiotemporal distribution data of sediment migration.

[0013] Preferably, the siltation and scouring early warning module includes the following functions: Calculate the net sediment transport of grid cells based on the temporal and spatial distribution data of sediment migration to generate net sediment transport data; Based on the net sediment transport data, the grid riverbed elevation change is calculated through a preset time step to obtain the riverbed elevation change data; The target river elevation terrain data and the riverbed elevation change data are superimposed, and compared with the preset riverbed terrain scour threshold to obtain the siltation scour excess data; According to the over-limit data of siltation and scour, the siltation and scour area is identified to obtain the siltation and scour area data; Calculate the sediment scouring and silting intensity based on the real-time cross-section monitoring characteristic data to generate grid unit scouring and silting intensity data; Early warning events are judged based on the excessive siltation and scouring data and the grid unit scouring and silting intensity data, and the siltation and scouring area data are uploaded to the cloud platform through the edge computing node for abnormal event alarm to obtain river section monitoring data.

[0014] The present invention deploys a perception module that integrates high-definition cameras, acoustic Doppler current profilers (ADCPs) and laser radar scanners at the edge of the river channel, and uses edge computing nodes for connection to build a highly integrated, multi-source heterogeneous river water conservancy perception network. This network can achieve all-round and three-dimensional perception of multiple factors such as river flow, sediment, and riverbed topography. High-definition cameras provide real-time video images of the river surface, intuitively reflecting the floating objects on the water surface, water color changes, etc.; ADCP can accurately measure key parameters such as water flow velocity, flow rate, and sediment concentration; and laser radar scanners can quickly obtain high-precision three-dimensional riverbed topographic data. This multi-sensor fusion method makes up for the limitations of a single sensor and realizes comprehensive and accurate perception of the river channel state. Through the deployment of edge computing nodes, real-time collection and preprocessing of raw perception data are realized. Unlike the traditional water conservancy perception system that transmits all data to a remote data center for processing, this system completes the time synchronization, spatial registration, and water conservancy feature extraction processing steps of the data on the edge side close to the data source. Time synchronization ensures the temporal consistency of data collected by different sensors, and the spatial matching criterion unifies these data into the same geographic coordinate system, thereby obtaining river channel perception data that is consistent in time and space. Based on these data, the system further extracts key characteristic parameters that reflect the state of water flow and sediment movement in the river channel, providing high-quality input data for building a three-dimensional flow field model. The introduction of edge computing has greatly reduced the latency and bandwidth requirements of data transmission, improved the efficiency of data processing, and enabled the system to respond quickly to changes in river channel sediment. Especially when dealing with sudden sediment flow events, this low-latency data processing capability can buy valuable time for timely warning and disposal. Using the extracted real-time cross-section monitoring feature data and combining it with advanced computational fluid dynamics (CFD) technology, a sophisticated three-dimensional sediment flow field network model is constructed. This model can not only reflect the overall water flow movement state of the river channel, but also depict the complex processes of sediment particle transport, suspension, and sedimentation in the water flow. Compared with traditional empirical formulas or simplified models, the three-dimensional flow field model can more accurately simulate the actual situation of river channel sediment movement, especially in river sections with complex terrain and variable water flow, where its advantages are more obvious. The system simulates the movement trajectories of a large number of sediment particles under the action of water flow, and then generates the spatiotemporal distribution data of sediment migration. This simulation can understand the sedimentation and scouring process of the river channel, and can also predict the evolution trend of the river channel in the future. By analyzing the movement trajectory and deposition location of sediment particles, areas prone to sedimentation or scouring can be accurately identified, providing important reference information for river management and waterway maintenance. Based on the spatiotemporal distribution data of sediment migration, the system can automatically identify high-risk areas for sedimentation and scouring, and upload this information to the cloud platform in real time through edge computing nodes, triggering abnormal event alarms.This active early warning mechanism has changed the passive response mode of traditional water conservancy monitoring, and realized real-time monitoring and risk warning of river sediment changes. The application of the cloud platform enables managers to remotely grasp the real-time status of the river and take timely response measures, such as adjusting the operation mode of water conservancy projects, dredging waterways, and reinforcing embankments, so as to effectively ensure the flood safety and navigation capacity of the river. Therefore, a water conservancy perception system based on edge computing of the present invention deploys high-definition cameras, acoustic Doppler current profilers (ADCPs) and lidar scanners at key sections of the river to build a multi-source heterogeneous river water conservancy perception network, thereby realizing all-round perception of water flow, sediment and riverbed topography. Edge computing nodes are used to perform real-time time synchronization and spatial registration of collected data, and to extract water conservancy monitoring features. Further combined with computational fluid dynamics (CFD) technology, a three-dimensional sediment flow field network model is constructed to simulate the sediment migration process, automatically identify siltation and scour areas, and use edge computing nodes to upload early warning information to the cloud platform in real time, thereby achieving real-time monitoring and early warning of river sediment changes, effectively improving the level of intelligent river management. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 It is a schematic diagram of the module flow of the water conservancy perception system based on edge computing of the present invention; Figure 2 for Figure 1 Schematic diagram of the detailed implementation process of the sediment transport simulation module; Figure 3 for Figure 1 Detailed implementation flow diagram of the sedimentation and scour warning module; 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

[0016] 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.

[0017] 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.

[0018] 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.

[0019] To achieve this, please refer to Figures 1 to 3 The present invention provides a water conservancy perception system based on edge computing, comprising the following modules: The river channel edge perception module includes a high-definition camera, an acoustic Doppler current profiler, and a laser radar scanner, which is used to connect the high-definition camera, the acoustic Doppler current profiler, and the laser radar scanner to edge computing nodes to build a river channel water conservancy perception network; real-time water conservancy monitoring data is collected based on the river channel water conservancy perception network to obtain river channel water conservancy perception data; The water conservancy characteristic analysis module is used to synchronize the time and space of the river water conservancy sensing data to obtain the spatiotemporal river channel sensing data; perform water conservancy monitoring characteristic analysis based on the spatiotemporal river channel sensing data to obtain real-time cross-section monitoring characteristic data; perform three-dimensional flow field model processing based on the real-time cross-section monitoring characteristic data to obtain a three-dimensional sediment flow field network model; The sediment migration simulation module is used to simulate the particle flow field of the three-dimensional sediment flow field network model and generate particle trajectory data; the sediment migration analysis is performed based on the particle trajectory data to generate the sediment migration time and space distribution data; The siltation and scour warning module is used to identify the siltation and scour areas based on the spatiotemporal distribution data of sediment migration and obtain the siltation and scour area data; upload the siltation and scour area data to the cloud platform through the edge computing node to alarm for abnormal events and obtain the river section monitoring data.

[0020] In an embodiment of the present invention, the water conservancy perception system based on edge computing includes the following modules: S1: River channel edge perception module, including high-definition camera, acoustic Doppler current profiler and lidar scanner, used to connect the high-definition camera, acoustic Doppler current profiler and lidar scanner to edge computing nodes, build a river channel water conservancy perception network; collect real-time water conservancy monitoring data according to the river channel water conservancy perception network to obtain river channel water conservancy perception data; In an embodiment of the present invention, sensing equipment is installed at the key section according to the river channel DEM data and key section marking results obtained in the early stage. The high-definition camera is fixed on a customized waterproof bracket, and the bracket is fixed to the riverbed or embedded parts by expansion bolts to ensure that the camera is stable and the lens is facing the center of the river. The acoustic Doppler current profiler (ADCP) is installed in a buoy style. The buoy is fixed to a predetermined position on the riverbed by an anchor chain, and the ADCP probe is facing downward to ensure that the complete flow profile is measured. The lidar scanner (such as RIEGL VZ-400i or FARO Focus S70) is installed at a commanding height on the river bank (such as a bridge pier or a special tower). By adjusting the pitch angle and azimuth angle of the scanner, ensure that the laser beam covers the entire key section and its surrounding area. After the installation is completed, the edge computing node is connected. An edge computing node (for example, an Advantech MIC-7700 industrial computer or an industrial computer with similar configuration) is set up near each key section. The node is equipped with an Intel Core i7 processor, 16GB memory, a 512GB solid-state drive, and an Ubuntu 20.04 operating system. The camera is connected to the PoE (Power over Ethernet) interface of the edge computing node via a Gigabit Ethernet cable to transmit real-time video streams. The ADCP is connected to the serial-to-Ethernet module of the edge computing node via an RS-485 serial cable to transmit data such as flow rate and water depth. The lidar scanner is connected to the edge computing node via a Gigabit Ethernet cable to transmit point cloud data. The edge computing nodes are connected by optical fiber to form a local area network. At the same time, each edge computing node is connected to the Internet through a 4G / 5G wireless router (for example, Huawei AR502H or an industrial router with similar functions) to achieve data interaction with the cloud platform. After the network is built, data collection is started. The high-definition camera shoots 1080P video encoded in H.264 at a rate of 25 frames per second. ADCP measures velocity profiles at a frequency of 1 Hz, and each profile contains velocity data for 20 cells. LiDAR scanners scan the river section once an hour, generating point cloud data containing millions of points.

[0021] S2: Water conservancy feature analysis module, used to perform time synchronization and spatial registration on river water conservancy sensing data to obtain spatiotemporal river channel sensing data; perform water conservancy monitoring feature analysis based on spatiotemporal river channel sensing data to obtain real-time cross-section monitoring feature data; perform three-dimensional flow field model processing based on real-time cross-section monitoring feature data to obtain a three-dimensional sediment flow field network model; In an embodiment of the present invention, the received data is time synchronized and spatially aligned. All data are time synchronized based on GPS time. For data with different sampling frequencies, a linear interpolation method is used to unify the data to one data point per minute. In terms of spatial alignment, the camera image is geometrically corrected by the camera's internal and external parameters (obtained by calibration), and the pixel coordinates are converted to WGS84 geographic coordinates. ADCP and lidar data directly use their own GPS positioning information. Data preprocessing is performed on the edge computing node. The camera video stream uses H.264 hardware decoding, and uses the OpenCV library for image enhancement (such as histogram equalization) and denoising (such as median filtering). ADCP data is subjected to bad value removal (according to the signal-to-noise ratio threshold) and coordinate conversion (converting the instrument coordinate system to the geographic coordinate system). The lidar point cloud data is filtered (such as statistical filtering) to remove noise points. The preprocessed data is classified according to the data type to form cross-sectional water body image data (JPEG format), cross-sectional flow velocity profile data (CSV format) and cross-sectional riverbed topography data (LAS format). Image processing algorithms (such as Canny edge detection) are used to extract the position of the water surface line. ADCP data is used to calculate the average flow velocity, maximum flow velocity, flow rate and other parameters of the section. LiDAR point cloud data is used to calculate parameters such as cross-sectional area, wetted perimeter, hydraulic radius, etc. These characteristic parameters are stored in JSON format to form real-time cross-sectional monitoring characteristic data. The three-dimensional model of the river channel is constructed using the pre-processed LiDAR point cloud data. CloudCompare software is used for point cloud splicing, registration and triangulation to generate a three-dimensional model in STL format. The real-time cross-sectional monitoring characteristic data (such as flow velocity and water level) are combined with the three-dimensional model, and CFD simulation is performed using OpenFOAM software to obtain a three-dimensional sediment flow field network model.

[0022] S3: Sediment migration simulation module, used to simulate the particle flow field of the three-dimensional sediment flow field network model and generate particle trajectory data; perform sediment migration analysis based on the particle trajectory data and generate sediment migration spatiotemporal distribution data; In the embodiment of the present invention, ParaView or other software supporting VTK format is used to read the three-dimensional sediment flow field network model, which is usually an unstructured grid (Unstructured Grid), including node coordinates, unit connection relationships, and flow velocity, sediment concentration and other data at each node. The volume of each grid unit (such as a tetrahedron or hexahedron) is calculated. The volume calculation formula depends on the type of grid unit. For tetrahedral units, the formula V=(1 / 6) is used |(ad)·((bd)×(cd))|, where a, b, c, and d are vertex coordinates. Mark the corresponding grid nodes or cells in the model according to the actual monitoring section location. For example, if there are ADCPs and water level gauges near the center point and both banks of a section, find the grid nodes closest to these locations in the model and mark them. Set the virtual particle release position near the marked point. For example, 10 meters upstream of the marked point, a release point can be set every 1 meter along the section direction, and a release point can be set every 0.5 meters in the vertical direction of each release point. Record the three-dimensional coordinates of each release point. Determine the number of particles at each release position based on the cross-section sediment concentration distribution data obtained by the hydraulic characteristic analysis module. Assuming that a total of 10,000 virtual particles are released, the sediment concentration of the grid cell where each release position is located is normalized, and then the total number of particles is multiplied by the normalized concentration to obtain the number of particles at the release position. Use OpenFOAM or other CFD software that supports particle tracking to perform Lagrangian particle tracking. Each particle moves according to the flow velocity at its location, taking into account the sediment settling velocity (calculated from the particle size-sedimentation velocity relationship) and random diffusion (simulated by adding random displacements). The trajectory of each particle is recorded, including the position and velocity at each time step.

[0023] S4: Siltation and scour warning module, which is used to identify the siltation and scour area according to the spatiotemporal distribution data of sediment migration, and obtain the siltation and scour area data; upload the siltation and scour area data to the cloud platform through the edge computing node for abnormal event alarm, and obtain the river section monitoring data.

[0024] In an embodiment of the present invention, the net sediment transport of each grid unit is calculated based on the particle trajectory data. The number of particles entering and leaving the grid unit in each time step is counted, and the net transport is calculated based on the theoretical sediment carrying capacity of each particle (the initial value is determined based on the total amount of sediment and the total number of particles, and is subsequently adjusted based on the flux). The riverbed elevation change of each grid unit is calculated based on the net sediment transport and the preset time step (e.g., 1 hour). The net transport is divided by the packing density and the bottom area of ​​the grid unit to obtain the elevation change. The initial river elevation data (from the lidar scan) is superimposed with the elevation change to obtain a new riverbed elevation. The new riverbed elevation is compared with the preset scouring threshold (e.g., 0.5 meters) and silting threshold (e.g., 0.3 meters), and the grid units that exceed the limit are marked. Using OpenCV or other image processing libraries, the connected region analysis is performed on the grid units that exceed the limit, continuous silting or scouring areas are identified, and the area of ​​each area is calculated. Based on the real-time cross-section monitoring characteristic data (such as flow velocity and bottom shear stress measured by ADCP), the scouring and silting intensity of each grid unit is calculated. For example, the bottom shear stress can be directly used as an indicator of scouring and silting intensity. Finally, set the warning rules, such as: (1) there is an area of ​​over-limit scouring or silting area greater than 100 square meters; (2) the average scouring and silting intensity in the over-limit area is greater than 0.1Pa; (3) the duration exceeds 6 hours. If any two of the above conditions are met, it is determined that a warning event has occurred. The edge computing node sends the warning information (including regional location, area, scouring and silting intensity, etc.) to the cloud platform via the 4G / 5G network. The cloud platform triggers the alarm mechanism (such as SMS notification) and stores the data in the database.

[0025] Preferably, the river edge perception module includes the following functions: Obtaining the elevation terrain data of the target river; Mark the key sections of the target river elevation terrain data and generate key section location data; According to the key section location data, high-definition cameras, acoustic Doppler flow profilers and lidar scanners are installed at the key sections of the river to obtain the layout data of water conservancy sensing equipment; Based on the deployment data of water conservancy sensing equipment, edge computing nodes of high-definition cameras, acoustic Doppler current profilers and lidar scanners are connected to build a river water conservancy sensing network; Real-time water conservancy monitoring data is collected based on the river water conservancy sensing network to obtain river water conservancy sensing data.

[0026] In an embodiment of the present invention, the drone flies according to a pre-planned route, and the laser radar scanner emits laser pulses and receives reflected echoes. By calculating the laser round-trip time difference, the three-dimensional coordinates (X, Y, Z) of each sampling point on the surface are obtained. During the aerial survey process, the drone needs to be equipped with real-time differential positioning (RTK) or network RTK (NRTK) equipment to achieve centimeter-level positioning accuracy by receiving differential correction signals from ground base stations. The acquired original laser point cloud data is filtered to remove non-ground points such as vegetation and buildings, and retain ground points reflecting the river terrain. Then, the triangulated ground points after filtering are triangulated using the triangulated grid method (TIN) to construct an irregular triangulated network. Through an interpolation algorithm, such as the Kriging interpolation method, the irregular triangulated network is converted into a regular grid to generate elevation raster data, that is, the elevation terrain data of the target river. The water flow direction of the river is analyzed to determine the mainstream line of the river. Along the mainstream line, section lines are set at fixed distance intervals (for example, every 50 meters) or in areas where the river morphology changes significantly (for example, where the river widens, narrows, and bends). In order to ensure that the section line is perpendicular to the direction of the water flow, it is necessary to use hydrological analysis software to calculate the average flow angle of each section and adjust the direction of the section line to make it perpendicular to the average flow angle. Section lines must also be set at important hydrological control points (such as bridges, dams, and confluences). Each section line is numbered, and the geographic coordinates (latitude and longitude) at both ends of the section line and all elevation point data on the section line are recorded. The selection criteria for key sections include: the section shape is representative (such as U-shaped, V-shaped), which can reflect the main hydraulic characteristics of the river; the section location is near the river control node (such as reservoirs, bridges, dangerous sections); the section location is convenient for the installation and maintenance of water conservancy sensing equipment. At each key section, the equipment installation plan is determined according to the section shape and water depth conditions. The high-definition camera is installed in a customized waterproof housing and fixed to the river bottom or bank slope through a bracket, with the lens facing the center of the river to ensure that a clear underwater image is captured. Acoustic Doppler current profiler (ADCP) is usually installed on a ship or buoy. The ship-borne ADCP is fixed to the bottom of the measuring vessel and measures the cross-sectional flow velocity distribution by sailing; the buoy-type ADCP is fixed to the buoy and fixed to the river bottom by an anchor chain. The lidar scanner is installed high on the river bank, such as on the top of a bridge or a special tower, and the scanning range covers the entire key section and its surrounding area. During the installation process, the model, serial number, installation location (latitude and longitude, elevation), installation time and other information of each device are recorded, and this information is associated with the key section number to form the layout data of the water conservancy sensing equipment. An edge computing node is set near each key section. The edge computing node uses an industrial-grade embedded computer with powerful computing power, rich interfaces and good environmental adaptability. Connect the high-definition camera, acoustic Doppler flow profiler and lidar scanner to the corresponding edge computing node through a wired or wireless network (such as 4G / 5G, optical fiber).The high-definition camera is connected to the edge computing node through the Ethernet interface or HDMI interface to transmit real-time video stream. The acoustic Doppler current profiler is connected to the edge computing node through the RS-232 or RS-485 serial port to transmit data such as flow velocity and water depth. The lidar scanner is connected to the edge computing node through the Ethernet interface to transmit point cloud data. After the river water conservancy perception network is built, each sensor collects data according to the preset frequency and working mode. The high-definition camera continuously shoots underwater video to record information such as water flow status, riverbed morphology, and suspended matter. The acoustic Doppler current profiler continuously measures water flow velocity and water depth to obtain cross-sectional flow velocity distribution data. The lidar scanner regularly scans the river section and surrounding terrain to obtain high-precision point cloud data for monitoring water level changes, bank slope stability, etc. The edge computing node receives and processes the data collected by the sensor in real time, and extracts key information such as average flow velocity, maximum flow velocity, water level, and water surface line.

[0027] Preferably, the marking of key sections of the target river elevation terrain data includes: Extract the centerline of the target river channel from the elevation terrain data and identify the left and right riverbanks to obtain the geometric constraint data of the river channel; Based on the river channel geometric constraint data, the preset equal spacing threshold is used to divide the section perpendicular to the river channel centerline to obtain the river channel section data; The river section data is spatially superimposed through the target river elevation terrain data to generate river section elevation data; Perform cross-sectional morphology analysis based on the river section elevation data to obtain cross-sectional geometric parameters; Calculate the cross-sectional shape coefficient according to the cross-sectional geometric shape parameters to generate the cross-sectional shape coefficient; Based on the preset key section identification threshold, the river section data is screened for key sections through the section morphology coefficient to generate key section location data.

[0028] In the embodiment of the present invention, based on the acquired digital elevation model (DEM), the DEM is filled with depressions to eliminate the influence of the depressions on the water flow simulation. Then, the flow direction of each grid cell is calculated, that is, to which unit of the eight adjacent cells the water flows from the cell. The flow direction calculation adopts the D8 algorithm, that is, according to the elevation difference between each cell and its eight adjacent cells, the maximum slope direction is determined as the flow direction. Next, the flow accumulation is calculated, that is, the area of ​​the catchment area upstream of each grid cell. The grid cells with high flow accumulation values ​​represent the locations where the water flows converge. Finally, a threshold is set according to the flow accumulation value, for example, an area with a flow accumulation value greater than 1000 grid cells is defined as a river channel. By connecting these grid cells, a continuous vector line is formed, that is, the center line of the river channel. For the identification of the river bank line, the generated river channel center line is extended to both sides with a fixed distance buffer zone, and the DEM data is used to set the elevation gradient threshold by comparing the change rate or gradient of the DEM elevation value in the buffer area, and the position above the threshold is identified as a point on the river bank line. These points are connected in sequence to form two riverbanks on the left and right. The river centerline and the left and right riverbanks together constitute the river geometric constraint data. Based on the extracted river centerline, section lines are generated along the centerline according to the preset equal spacing threshold. For example, if the preset equal spacing threshold is 50 meters, a section line perpendicular to the centerline is generated every 50 meters. The vertical relationship is realized by calculating the tangent direction of each segment point on the centerline, and then generating a straight line segment perpendicular to the tangent direction at this point. The length of the straight line segment needs to be greater than the width of the river to ensure that it intersects with the left and right riverbanks. In practical applications, since the river channel is not completely regular, simple equal spacing division will result in missing sections at certain key locations. Therefore, it is allowed to manually add or adjust section lines on the basis of equal spacing division. For example, section lines are added at sharp bends in the river channel or at locations with obvious hydraulic structures. Each generated section line records the coordinates of its starting and ending points, as well as the coordinates of the intersection with the centerline. This information constitutes the river section data. Get the intersection of each section line and DEM, and extract the elevation values ​​of these intersections. For each section line, organize the intersection points and their corresponding elevation values ​​into an ordered sequence from the left bank to the right bank (or vice versa). This sequence describes the elevation change on the section. Each point in the sequence contains X, Y coordinates (plane position) and Z coordinates (elevation value). Using the trapezoidal rule or Simpson's rule, integrate the section elevation data to obtain the cross-sectional water area. Then, calculate the wetted perimeter, that is, the perimeter of the contact between the water flow and the riverbed. The wetted perimeter length is obtained by accumulating the distance between two adjacent points in the section elevation data. Next, calculate the hydraulic radius, that is, the ratio of the cross-sectional area to the wetted perimeter. The hydraulic radius is an important parameter reflecting the hydraulic efficiency of the river channel. In addition, parameters such as the maximum water depth, average water depth, and section width can also be calculated. The section shape coefficient is calculated using the following formula: Shape coefficient = wetted perimeter / (2 sqrt(cross-sectional area)). The larger the shape coefficient, the more irregular the cross-sectional shape. Calculate the cross-sectional width-to-depth ratio, which reflects the width of the cross-sectional area. The cross-sectional width-to-depth ratio can be calculated using the following formula: width-to-depth ratio = cross-sectional width / average water depth of the cross-sectional area. Among them, the cross-sectional width refers to the width of the river channel at the cross-sectional area, and the average water depth of the cross-sectional area refers to the cross-sectional area divided by the cross-sectional width. Calculate the cross-sectional hydraulic efficiency, which reflects the water flow transport capacity of the cross-sectional area. The larger the hydraulic radius and the smaller the roughness, the higher the hydraulic efficiency. The cross-sectional hydraulic efficiency can be calculated using the following formula: hydraulic efficiency = hydraulic radius / river roughness. Associate the calculated cross-sectional shape coefficient, cross-sectional width-to-depth ratio, cross-sectional hydraulic efficiency and other morphological coefficients with the corresponding cross-sectional numbers to form a cross-sectional shape coefficient. For the range of cross-sectional shape coefficient, cross-sectional width-to-depth ratio, and cross-sectional hydraulic efficiency morphological coefficient. For example, the shape coefficient threshold is set to be greater than 1.5, the width-to-depth ratio threshold is set to be greater than 10, and the hydraulic efficiency threshold is set to be greater than 0.5. For each section, determine whether its section shape coefficient meets the set threshold condition. If all the shape coefficients of a section meet the set threshold condition, the section is determined as a critical section. The geometric shape and geographical location information of the critical section are stored as vector data and associated with the critical section number to form the critical section location data.

[0029] Preferably, the water conservancy feature analysis module includes the following functions: Perform time synchronization and spatial registration on river water conservancy sensing data to obtain spatiotemporal river sensing data; The spatiotemporal river channel sensing data is transmitted to the edge computing node, and the data is preprocessed to generate water conservancy monitoring data; wherein the water conservancy monitoring data includes cross-sectional water body image data, cross-sectional flow velocity profile data, and cross-sectional riverbed topography data; Conduct water conservancy monitoring feature analysis based on water conservancy monitoring data to obtain real-time cross-section monitoring feature data; Construct a three-dimensional spatial network based on the target river elevation terrain data to generate a three-dimensional spatial model of the river; The three-dimensional flow field model of the river channel is processed by real-time cross-section monitoring feature data to obtain a three-dimensional sediment flow field network model.

[0030] In an embodiment of the present invention, time synchronization is performed on data from different sensors. Since different sensors use different clocks or there is clock drift, it is necessary to calibrate the timestamps of all data to a unified time reference. The clock of the central server is selected as the time reference. Each edge computing node periodically synchronizes time with the central server to obtain the current standard time. When uploading data, the edge computing node corrects the timestamp of the local data with the standard time to ensure that the timestamps of all data are based on the same time reference. Then, it is necessary to spatially align the data from different locations. The device geographic coordinates recorded in the water conservancy sensing device layout data are used to associate the data collected by each sensor with the corresponding geographic location. Using the spatial coordinate conversion function of the GIS software, all data are converted to a unified geographic coordinate system (e.g., WGS84 coordinate system). The time-synchronized data and the spatially aligned data are integrated to form spatiotemporal river channel perception data. The spatiotemporal river channel perception data is transmitted to the corresponding edge computing node via a wired or wireless network (such as 4G / 5G, optical fiber). After receiving the data, the edge computing node first performs data format conversion to convert the data of different sensors into a unified format, such as JSON or CSV. Then, data cleaning is performed to remove outliers and erroneous data. For example, for flow rate data, a reasonable range is set, and data outside the range is considered an outlier. For image data, noise and interference in the image are detected and removed. Next, data compression is performed to reduce the amount of data and improve transmission and storage efficiency. For example, for video data, H.264 or H.265 encoding is used for compression. For flow velocity profile data and riverbed topography data, data structure optimization is performed, such as converting a two-dimensional array into a one-dimensional array. The pre-processed data is classified according to the data type to form cross-sectional water body image data (processed video frames), cross-sectional flow velocity profile data (two-dimensional flow velocity matrix), and cross-sectional riverbed topography data (point cloud or raster data), collectively referred to as water conservancy monitoring data. For cross-sectional water body image data, image processing algorithms are used to analyze the color, texture and other characteristics of the water body, and evaluate water quality indicators such as turbidity and transparency of the water body. Floating objects in water body images are detected, the types and quantities of floating objects are identified, and the degree of river pollution is evaluated. For cross-sectional velocity profile data, analyze the distribution of velocity and calculate hydrological parameters such as average velocity, maximum velocity, velocity gradient, etc. Analyze the change trend of flow and evaluate the change of flow in the river. For cross-sectional riverbed terrain data, analyze the ups and downs of the riverbed and calculate the average elevation, standard deviation and other terrain parameters of the riverbed. Analyze the scouring and siltation of the riverbed and evaluate the stability of the river. Integrate the analyzed cross-sectional water quality indicators, hydrological parameters, terrain parameters and other information to form real-time cross-sectional monitoring feature data. Grid the elevation terrain data and divide the river area into a series of three-dimensional grid cells. The size of the grid cell depends on the accuracy of the data and the computing resources.Then, each grid cell is triangulated and divided into a series of triangular patches. Triangular patches can better approximate the complex surface of the river. The elevation data is triangulated using the triangular mesh method (TIN) to construct an irregular triangulated network. The irregular triangulated network is optimized, such as Delaunay triangulation, to ensure the quality of the triangular patches. All triangular patches are connected to form a three-dimensional mesh model. In order to improve the visualization of the model, the model can be textured to map the elevation value to the surface color of the model. The generated three-dimensional mesh model is used as the three-dimensional spatial model of the river. The cross-section monitoring data is interpolated onto the three-dimensional spatial grid. For example, the cross-section flow velocity profile data is interpolated onto the corresponding cross-section in the three-dimensional model to obtain the flow velocity value at each grid node. The interpolation method can use Kriging interpolation, inverse distance weighted interpolation or other spatial interpolation methods. Then, the three-dimensional Navier-Stokes equations are solved using computational fluid dynamics (CFD) methods, such as the finite volume method or the finite element method, to simulate the water flow movement in the river. CFD simulation requires setting boundary conditions, such as inlet velocity, outlet water level, riverbed roughness, etc. These boundary conditions can be obtained from real-time monitoring data. For sediment movement, sediment transport equations, such as the Meyer-Peter & Müller formula or the Engelund-Hansen formula, are used to calculate the sediment transport rate. The sediment transport rate is coupled with the water flow field to simulate the scouring and deposition process of sediment. The resulting three-dimensional sediment flow field network model.

[0031] Preferably, the water conservancy monitoring feature analysis includes: Perform sediment particle texture recognition on the cross-sectional water body image data to obtain sediment particle texture data; Perform particle size statistics based on sediment particle texture data and calculate the mean particle size to generate real-time sediment particle size data; Fitting the near-bottom velocity distribution curve to the cross-sectional velocity profile data to obtain the near-bottom velocity distribution curve data; Based on the cross-sectional riverbed topography data and using the near-bottom velocity distribution curve data, the riverbed bottom shear stress is calculated to generate bottom shear stress data; According to the cross-section velocity profile data and bottom shear stress data, the cross-section real-time concentration is estimated to generate the cross-section sediment concentration distribution data; The real-time sediment particle size data, cross-section sediment concentration distribution data and bottom shear stress data are integrated into water conservancy monitoring characteristics to generate real-time cross-section monitoring characteristic data.

[0032] In an embodiment of the present invention, the cross-sectional water body image data is preprocessed, including image enhancement, denoising and other operations to improve the image quality. An image segmentation technique, such as a threshold-based segmentation method, is used to separate the silt particles in the image from the background water body. A threshold is set, and the pixels in the image with grayscale values ​​higher than the threshold are determined as silt particles, and the pixels lower than the threshold are determined as background water bodies. In order to improve the segmentation accuracy, morphological operations, such as opening and closing operations, are used to optimize the segmentation results and remove small isolated areas and connection fracture areas. For the segmented silt particles, their texture features are extracted. The texture features can reflect the surface roughness, directionality and other information of the silt particles. The gray level co-occurrence matrix (GLCM) method can be used to extract the texture features. Based on the silt particle texture data, the number of silt particles of different particle sizes is counted, and the average particle size of the silt particles is calculated. The edge detection of the silt particles is performed using an image processing algorithm, such as an edge detection algorithm, to extract the boundaries of the silt particles. According to the boundaries of the silt particles, the area and perimeter of the silt particles are calculated. Using the area and perimeter, estimate the particle size of the sediment particles. The following formula can be used to estimate the particle size of the sediment particles: particle size = sqrt(4 Area / pi). Classify the sediment particles according to their particle size, such as fine sand, medium sand, coarse sand and other different particle size groups. Count the number of sediment particles in each particle size group. Calculate the average particle size of the sediment particles. The average particle size can be calculated using the following formula: Average particle size = (∑(particle size Number)) / ∑number. Among them, particle size refers to the particle size of each particle size group, and number refers to the number of sediment particles in each particle size group. The calculated average particle size is used as the real-time sediment particle size data. The flow velocity data within a certain height range from the bottom of the riverbed is selected as the near-bottom flow velocity data. The flow velocity data within a range of 0.1 meters from the bottom of the riverbed can be selected. The logarithmic flow velocity formula is used to fit the near-bottom flow velocity data. The logarithmic flow velocity formula describes the change of flow velocity with height in the turbulent boundary layer: u=(u / k) ln(z / z0). Among them, u refers to the flow velocity at a height of z from the bottom of the riverbed. refers to the friction velocity, k refers to the Karman constant (usually 0.4), and z0 refers to the roughness height. The least squares method is used to fit the logarithmic flow velocity formula to obtain the friction velocity and the value of the roughness height z0. Based on the friction velocity and roughness height obtained by fitting, the near-bottom velocity distribution curve is generated. The shear stress at the bottom of the riverbed is calculated using the following formula: τ=ρ ^2. Where τ refers to the shear stress at the bottom of the riverbed, and ρ refers to the density of water. Refers to friction velocity. The friction velocity can be obtained from the fitting results of the near-bottom velocity distribution curve. Riverbed topography data can be used to calculate the average slope of the riverbed. The slope can affect the magnitude of the shear stress at the bottom of the riverbed. The calculated riverbed bottom shear stress is associated with its corresponding cross-sectional position information to form the bottom shear stress data. The Rouse formula can be used to estimate the sediment concentration distribution. The Rouse formula describes the change of suspended sediment concentration with height: c / ca=((Hz) / z za / (H-za))^Z. Among them, c refers to the sediment concentration at a height of z from the bottom of the riverbed, ca refers to the sediment concentration at a reference height za, H refers to the water depth, and Z refers to the Rouse number. The Rouse number can reflect the suspension capacity of sediment, and its calculation formula is: Z=ws / (k ). Among them, ws refers to the sediment settling velocity, k refers to the Karman constant (usually 0.4), Refers to friction speed. Real-time sediment particle size data, cross-section sediment concentration distribution data, bottom shear stress data and their corresponding cross-section position information and time information are associated to form a comprehensive real-time cross-section monitoring feature data.

[0033] Preferably, the cross-section real-time concentration estimation includes: According to the cross-sectional velocity profile data, the echo intensity is corrected and the suspended sediment reference concentration is calculated to generate the suspended sediment reference concentration data; The sediment settling velocity data is queried by using the preset sediment particle size-sedimentation velocity relationship database to obtain the sediment settling velocity data; Calculate the probability of sediment initiation based on sediment settling velocity data and bottom shear stress data to generate sediment initiation probability data; The sediment suspended concentration is estimated based on the suspended sediment reference concentration data and the sediment initiation probability data, and the cross-sectional concentration distribution is processed to generate the cross-sectional sediment concentration distribution data.

[0034] In the embodiment of the present invention, in the cross-sectional velocity profile data, a height close to the bottom of the riverbed is selected as the reference height. A height of 0.05 meters from the bottom of the riverbed can be selected as the reference height. The suspended sediment concentration at the reference height is calculated using the relationship between echo intensity and suspended sediment concentration. The relationship between echo intensity and suspended sediment concentration is usually a linear relationship or an exponential relationship, which can be obtained through experimental calibration. The suspended sediment concentration can be estimated using the following formula: Cref=a Echo+b. Among them, Cref refers to the suspended sediment concentration at the reference height, Echo refers to the echo intensity at the reference height, and a and b are calibration coefficients. The calculated suspended sediment concentration at the reference height is used as the suspended sediment reference concentration data. The relationship between sediment particle size and settling velocity is used to determine the settling velocity of sediments of different particle sizes. Establish a sediment particle size-sedimentation velocity relationship database. The database can contain empirical formulas, such as the Rubey formula, the Stokes formula, etc., or contain experimental measurement data. The Rubey formula is suitable for calculating the settling velocity of larger particle size sediments, and the Stokes formula is suitable for calculating the settling velocity of smaller particle size sediments. For each particle size value in the real-time sediment particle size data, the corresponding settling velocity is searched in the sediment particle size-sedimentation velocity relationship database. If there is no completely matching particle size value, the interpolation method can be used to estimate the settling velocity. The queried sediment settling velocity is used as the sediment settling velocity data. Calculate the Shields number based on the bottom shear stress data: Shields number = / (( -p) g d), where is the bottom shear stress, is the density of sediment, p is the density of water, g is the acceleration of gravity, and d is the particle size of sediment. Then, the particle Reynolds number is calculated based on the sediment settling velocity and the friction flow velocity. The calculated Shields number and particle Reynolds number are compared with the Shields curve. If the calculation point is above the Shields curve, the sediment is considered to be moving; if it is below the curve, the sediment is considered to be not moving. In order to more accurately describe the probability of sediment starting, a probability model can be used, such as P=1-exp(-k (Shields number - Shields number_c)), where P is the probability of sediment initiation, k is an empirical coefficient, and Shields number_c is the critical Shields number. The calculated sediment initiation probability is a value between 0 and 1, indicating the probability of sediment particles initiating under the current conditions. This value is the sediment initiation probability data. Using the Rouse formula or its improved form: C(z)=C_a ((hz) / z a / (ha))^Z, where C(z) is the sediment concentration at a height z from the riverbed, C_a is the sediment concentration at a reference height a (i.e., the reference concentration of suspended sediment), h is the water depth, and Z is the Rouse number. Z=ω / (β k ), ω is the sediment settling velocity, β is a coefficient, and k is the Karman constant (usually 0.4). The Rouse formula describes the distribution of suspended sediment content with water depth. In the calculation process, sediment settling velocity data (calculation Z), bottom shear stress data, and suspended sediment reference concentration data (as C_a) are needed. For the near-bottom area, the calculated sediment concentration is arranged in the vertical and horizontal directions (along the cross section) to form a two-dimensional array. This array represents the sediment concentration distribution of the entire cross section.

[0035] Preferably, the three-dimensional flow field model processing of the three-dimensional spatial model of the river channel includes: Perform cross-section three-dimensional coordinate transformation based on cross-section sediment concentration distribution data to generate cross-section concentration position data; The three-dimensional spatial model of the river channel is interpolated through the cross-section concentration position data to obtain a three-dimensional sediment concentration network model; The bottom boundary conditions of the riverbed are set according to the cross-section riverbed topography data and the bottom shear stress data, and the bottom boundary condition data are generated; Based on the bottom boundary condition data and the three-dimensional sediment concentration network model, a three-dimensional flow field numerical simulation model is constructed through the Navier-Stokes equation to generate a three-dimensional sediment flow field network model.

[0036] In an embodiment of the present invention, the cross-sectional sediment concentration distribution data is based on a two-dimensional cross-sectional coordinate system (usually an xz coordinate system, with the x-axis along the cross-sectional direction and the z-axis perpendicular to the riverbed). In order to apply it to a three-dimensional river channel model, coordinate conversion is required. First, it is necessary to know the position and direction of each two-dimensional cross-section in the three-dimensional model. This can be determined by previously generated key cross-sectional position data and river channel geometric constraint data. For each cross-section, the coordinates (latitude and longitude) and direction (perpendicular to the centerline of the river channel) of its center point are known. Then, each point (x, z) in the two-dimensional cross-sectional coordinate system is converted to a point (X, Y, Z) in the three-dimensional coordinate system. The conversion formula depends on the specific coordinate system definition and cross-sectional position. Generally, the conversion can be performed by the following steps: 1. The origin of the two-dimensional coordinate system is translated to the position of the cross-sectional center point in the three-dimensional model. 2. The x-axis of the two-dimensional coordinate system is rotated to be consistent with the cross-sectional direction. 3. The z coordinate in the two-dimensional coordinate system is used as the Z coordinate in the three-dimensional coordinate system. After conversion, each sediment concentration data point corresponds to a three-dimensional coordinate (X, Y, Z), which indicates the position of the concentration value in three-dimensional space. The existing cross-sectional concentration position data (discrete point data) is interpolated into the entire three-dimensional river channel model to obtain a continuous three-dimensional concentration field. The geometry of the riverbed is determined based on the cross-sectional riverbed topography data. This can be achieved by extending the cross-sectional topography data to the entire three-dimensional model. Then, the roughness of the riverbed is set based on the bottom shear stress data. Roughness can be expressed by parameters such as the Manning coefficient, the Scheher coefficient, or the roughness height. There is a certain relationship between the bottom shear stress and the roughness, which can be converted based on empirical formulas or measured data. For example, the roughness height z0 can be estimated based on the friction velocity and logarithmic velocity distribution formula. The riverbed geometry and roughness parameters are combined to form the bottom boundary condition data. Using the computational fluid dynamics (CFD) method, a three-dimensional flow field numerical simulation model is constructed based on the Navier-Stokes equation (NS equation) and the sediment transport equation. The Navier-Stokes equations describe the conservation of momentum of the fluid, and the sediment transport equations describe the convection and diffusion of sediment. These equations are partial differential equations and require numerical discretization to be solved. Commonly used numerical methods include the finite volume method (FVM), the finite difference method (FDM), and the finite element method (FEM). The three-dimensional sediment concentration network model is used as the initial condition, and the bottom boundary condition data is used as the boundary condition, and is input into the numerical simulation software, such as OpenFOAM, ANSYS Fluent, or COMSOL Multiphysics. Simulation parameters such as time step, convergence criterion, etc. are set in the software. Then, a numerical solution is performed. The solution process usually requires a lot of computing resources and time. After the solution is completed, a three-dimensional flow field is obtained, including information such as velocity, pressure, sediment concentration, etc. at each grid node.

[0037] Preferably, the sediment migration simulation module includes the following functions: Calculate the grid unit volume based on the three-dimensional sediment flow field network model to generate grid unit volume data; Mark the cross-section monitoring points of the three-dimensional sediment flow field network model, set the virtual particle release position, and obtain the virtual particle release position data; The particle number distribution of the three-dimensional sediment flow field network model is carried out through the cross-section sediment concentration distribution data to obtain the virtual particle number data; Based on the virtual particle release position data and virtual particle quantity data, the particle flow field simulation is performed on the three-dimensional sediment flow field network model, and Lagrangian particle tracking is performed to generate particle trajectory data; Sediment migration analysis is performed based on particle trajectory data to generate spatiotemporal distribution data of sediment migration.

[0038] As an example of the present invention, refer to Figure 2 As shown, Figure 1 Detailed implementation flow diagram of the sediment migration simulation module, which includes the following functions: S31: Calculate the volume of grid cells according to the three-dimensional sediment flow field network model to generate grid cell volume data; In the embodiment of the present invention, the three-dimensional sediment flow field network model is usually composed of a large number of three-dimensional grid units (such as tetrahedrons, hexahedrons, etc.). In order to perform subsequent sediment migration simulation, it is necessary to calculate the volume of each grid unit. The method for calculating the volume of the grid unit depends on the type of grid unit. For a tetrahedral unit, the volume V can be calculated by the coordinates of its four vertices: V=(1 / 6) |(ad)·((bd) (cd))|, where a, b, c, d are the coordinates of the four vertices of the tetrahedron, "·" represents the vector dot product, "×" represents the vector cross product, and "||" represents the vector modulus. For a hexahedral unit, it can be decomposed into multiple tetrahedral units, and then the volume is calculated separately and summed. Alternatively, the volume can be calculated directly using the vertex coordinates of the hexahedral unit and the Gaussian integral formula.

[0039] S32: marking cross-section monitoring points of the three-dimensional sediment flow field network model, and setting virtual particle release positions to obtain virtual particle release position data; In an embodiment of the present invention, a Lagrangian particle tracking method is used. It is necessary to release virtual particles in the model and track their motion trajectories. First, according to the actual section monitoring position, the corresponding grid nodes or grid cells are marked in the three-dimensional sediment flow field network model. These marked points represent the actual observation positions for comparison with the simulation results. Then, the release position of the virtual particles is determined. The release position can be set according to the research purpose and simulation requirements. For example, particles can be released evenly at the entrance section of the river channel, or they can be released in a concentrated manner in a specific area (such as the bottom of the riverbed or the water surface). The release position can be a point, a line, a surface, or a body. For each release position, its three-dimensional coordinates (X, Y, Z) are recorded. These coordinate data constitute the virtual particle release position data. The data is stored in the form of a list or array, and each element represents a release position.

[0040] S33: performing particle number distribution on the three-dimensional sediment flow field network model according to the cross-section sediment concentration distribution data to obtain virtual particle number data; In an embodiment of the present invention, the number of virtual particles released at each release position is determined based on the cross-section sediment concentration distribution data. The principle of particle number distribution is: the higher the sediment concentration, the more particles are released. This can more realistically reflect the distribution and migration of sediment. The specific distribution method can adopt the following steps: 1. Calculate the sediment concentration of the grid unit where each release position is located. If the release position is a point, take the concentration value of the grid unit where the point is located; if the release position is an area, take the average concentration value of all grid units in the area. 2. Normalize the concentration values ​​of all release positions so that their sum is 1. 3. Based on the preset total number of particles and the normalized concentration value, calculate the number of particles to be allocated to each release position: Ni=N_total Ci_norm, where Ni is the number of particles at the ith release position, N_total is the total number of particles, and Ci_norm is the normalized concentration value at the ith release position. The calculated number of particles at each release position is stored in integer form, forming a one-dimensional array with the same number of release positions, i.e., virtual particle number data.

[0041] S34: performing particle flow field simulation on the three-dimensional sediment flow field network model based on the virtual particle release position data and the virtual particle quantity data, and performing Lagrangian particle tracking to generate particle trajectory data; In an embodiment of the present invention, a particle tracking simulation is performed in a three-dimensional sediment flow field network model. The Lagrangian method is used to treat each virtual particle as an independent particle and track its motion trajectory in the flow field. The equation of motion for each particle is: dx / dt=u(x,t), where x is the position vector of the particle, t is time, and u(x,t) is the velocity vector at the location of the particle. The velocity vector u(x,t) is obtained from the three-dimensional sediment flow field network model and obtained by spatial interpolation. The equation of motion of the particle is solved by numerical integration methods, such as the Euler method, the Runge-Kutta method, etc. The Euler method is relatively simple, but has low accuracy; the Runge-Kutta method has high accuracy, but a large amount of calculation. In each time step, the position of the next time step is calculated based on the current position and velocity of the particle. At the same time, the sedimentation effect of the sediment is considered. In the vertical direction, a downward sedimentation velocity ω is applied to the particle, and the value of ω is determined according to the relationship between the sediment particle size and the sedimentation velocity. In addition, the random diffusion effect of the sediment can also be considered. In each time step, a random displacement is applied to the particle to simulate the turbulent diffusion of sediment. The tracking process continues until the particle leaves the simulation area or the preset simulation time is reached. The motion trajectory of each particle is recorded, including the position coordinates (X, Y, Z) and time of each time step. These data constitute the particle trajectory data. The data is stored in the form of a list or array, and each element represents the trajectory of a particle.

[0042] S35: Perform sediment migration analysis based on the particle trajectory data to generate sediment migration spatiotemporal distribution data.

[0043] In an embodiment of the present invention, the number of particles in different areas in different time periods is counted to reflect the change in sediment concentration. For example, the change in the number of particles on each section over time can be counted to reflect the sediment transport along the way. Then, the average migration velocity, migration distance and other parameters of the particles are calculated to reflect the sediment transport rate and range. The average migration velocity can be calculated by the ratio of the displacement of the particles over a period of time to the time. The migration distance can be calculated by accumulating the displacement of the particles in each time step. Next, the spatial distribution characteristics of the particles are analyzed, such as whether there are aggregation areas, retention areas, etc. This can be achieved by drawing a particle distribution density map or a contour map.

[0044] Preferably, the performing sediment migration analysis based on particle trajectory data comprises: Perform grid particle average displacement vector statistics based on particle trajectory data to generate grid average displacement data; Calculate the sediment migration flux based on the grid average displacement data to generate sediment migration flux data; The theoretical particle sediment carrying capacity is converted according to the sediment migration flux data to obtain the theoretical particle sediment carrying capacity; The sediment migration rate is calculated for the three-dimensional sediment flow field network model through the sediment migration flux data to generate sediment migration rate data; The theoretical particle sediment carrying capacity and sediment migration rate data are integrated into the grid spatiotemporal distribution to generate the spatiotemporal distribution data of sediment migration.

[0045] In an embodiment of the present invention, the particle trajectory data is grouped according to the grid unit. For each grid unit, all particles passing through the unit are found. Then, for each particle, its displacement vector in the grid unit is calculated. The displacement vector is defined as the position of the particle when it leaves the grid unit minus the position when it enters the grid unit. If a particle passes through multiple grid units in one time step, the displacement vector needs to be decomposed into each grid unit. Then, the displacement vectors of all particles in the same grid unit are averaged to obtain the average displacement vector of the grid unit. The averaging method can be a simple arithmetic average or a weighted average, and the weight can be determined according to the residence time or path length of the particle in the grid unit. The calculated average displacement vector of each grid unit contains three components (X, Y, and Z directions), which represent the average migration direction and distance of the sediment in the grid unit. For a tetrahedral grid, the interface is a triangle; for a hexahedral grid, the interface is a quadrilateral. Then, the area and normal vector of each interface are calculated. The normal vector points to the outside of the interface. Then, for each interface, the dot product of the average displacement vectors of the grid units on both sides is calculated, and then multiplied by the area of ​​the interface. This dot product reflects the velocity component of the sediment passing through the interface. Finally, this velocity component is multiplied by the sediment concentration (which can be obtained from the three-dimensional sediment concentration network model) to obtain the sediment mass flux through the interface. It should be noted that the sediment concentration should be taken as the average value of the grid cells on both sides of the interface or the upstream value. The calculated sediment migration flux of each interface is a scalar, which represents the mass of sediment passing through the interface per unit time. Since virtual particles are used in the simulation to represent actual sediment particles, it is necessary to establish a connection between virtual particles and actual sediment. One method is to regard each virtual particle as representing a certain mass of sediment. This mass can be determined by the following steps: 1. Calculate the total mass of sediment in the entire simulation area. This can be obtained by integrating the three-dimensional sediment concentration network model. 2. Count the total number of virtual particles released in the simulation. 3. Divide the total mass of sediment by the total number of particles to obtain the average mass of sediment represented by each virtual particle. This average mass is the initial value of the theoretical particle sediment carrying capacity. During the simulation, the mass of sediment represented by each particle will change due to the scouring and siltation of sediment. Therefore, it is necessary to dynamically adjust the sediment carrying capacity of each particle based on the sediment transport flux data. For example, if a particle passes through an interface of sediment deposition, its sediment carrying capacity should increase; if it passes through an interface of sediment scouring, its sediment carrying capacity should decrease. The magnitude of the adjustment is proportional to the sediment transport flux. For each grid cell, calculate the sum of the sediment transport fluxes of all its interfaces. It should be noted that the positive and negative signs of the flux indicate whether the sediment flows into or out of the grid cell. Inflow is positive and outflow is negative. Then, divide the sum of the fluxes by the volume of the grid cell to obtain the rate of change of sediment mass per unit volume. This rate of change is the sediment transport rate.The calculated sediment migration rate is a scalar that represents the change in sediment mass per unit time and per unit volume. A positive value indicates sediment deposition, while a negative value indicates sediment scouring. The theoretical particle sediment carrying capacity and sediment migration rate data are associated with the three-dimensional sediment flow field network model to determine the theoretical particle sediment carrying capacity and sediment migration rate of each grid unit. The theoretical particle sediment carrying capacity and sediment migration rate data can be used as attributes of the grid unit and stored in the data structure of the grid unit.

[0046] Preferably, the siltation and scouring early warning module includes the following functions: Calculate the net sediment transport of grid cells based on the temporal and spatial distribution data of sediment migration to generate net sediment transport data; Based on the net sediment transport data, the grid riverbed elevation change is calculated through a preset time step to obtain the riverbed elevation change data; The target river elevation terrain data and the riverbed elevation change data are superimposed, and compared with the preset riverbed terrain scour threshold to obtain the siltation scour excess data; According to the over-limit data of siltation and scour, the siltation and scour area is identified to obtain the siltation and scour area data; Calculate the sediment scouring and silting intensity based on the real-time cross-section monitoring characteristic data to generate grid unit scouring and silting intensity data; Early warning events are judged based on the excessive siltation and scouring data and the grid unit scouring and silting intensity data, and the siltation and scouring area data are uploaded to the cloud platform through the edge computing node for abnormal event alarm to obtain river section monitoring data.

[0047] As an example of the present invention, refer to Figure 3 As shown, Figure 1 The detailed implementation flow diagram of the siltation and scouring warning module in this example includes the following functions: S41: Calculate the net sediment transport of grid cells according to the temporal and spatial distribution data of sediment migration to generate net sediment transport data; In an embodiment of the present invention, a time period is determined, such as a simulation step or several simulation steps. Then, for each grid unit, the sediment migration rate of all time steps in the time period is multiplied by the time step to obtain the change in sediment mass at each time step. Next, the change in sediment mass at all time steps is added up to obtain the net sediment transport in the time period. If the net transport is positive, it indicates sediment deposition in the grid unit; if it is negative, it indicates sediment scouring. The calculated net sediment transport for each grid unit is a numerical value, which indicates the net change in sediment mass in the grid unit during the time period. These numerical values ​​constitute the net sediment transport data.

[0048] S42: Calculating the riverbed elevation change based on the net sediment transport data at a preset time step to obtain riverbed elevation change data; In the embodiment of the present invention, the packing density ρb of the sediment is determined. The packing density can be determined by measured data or empirical values. Then, the bottom area of ​​each grid unit is calculated. For a three-dimensional grid, the bottom area refers to the area of ​​the surface where the grid unit contacts the riverbed. For a two-dimensional grid, the bottom area is the area of ​​the grid unit. Next, the net sediment transport of each grid unit is divided by the packing density and the bottom area to obtain the riverbed elevation change: Δh=ΔM / (ρb A), where Δh is the change in riverbed elevation, ΔM is the net sediment transport, ρb is the bulk density, and A is the bottom area. The calculated change in riverbed elevation is a numerical value that represents the change in riverbed elevation of the grid cell during the time period.

[0049] S43: superimposing the target river course elevation terrain data and the riverbed elevation change data, and comparing them with a preset riverbed terrain scour threshold to obtain siltation scour excess data; In an embodiment of the present invention, the initial elevation value of each grid cell is added to the corresponding elevation change. Then, the new riverbed elevation data is compared with the preset riverbed terrain scour threshold. The scour threshold is a preset value that indicates the allowable riverbed scour depth. The setting of the threshold needs to be determined based on the characteristics of the specific river channel and flood control requirements. For each grid cell, if its new riverbed elevation value is lower than the initial elevation value minus the scour threshold, it is considered that the grid cell has undergone excessive scour; if its new riverbed elevation value is higher than the initial elevation value plus the sedimentation threshold (another preset value), it is considered that the grid cell has undergone excessive sedimentation. These excessive grid cells are marked, and their location and excessive type (scour or sedimentation) are recorded.

[0050] S44: identifying the siltation and scour area according to the siltation and scour exceeding limit data, and obtaining the siltation and scour area data; In an embodiment of the present invention, continuous siltation or scour areas are identified based on the obtained siltation and scour excess data. This can be achieved through the connected region analysis method in image processing. The grid cells marked as exceeding the limit are regarded as pixels in the image, with a value of 1 (or different values ​​are assigned according to the type of exceeding the limit), and other grid cells are regarded as background pixels, with a value of 0. Then, the adjacent exceeding-limit grid cells are connected using a four-neighborhood or eight-neighborhood connectivity analysis method to form connected regions. Each connected region represents a siltation area or scour area. The location, area, shape and other information of each area are recorded. For example, a rectangular frame or a polygon can be used to represent the scope of each area, which constitutes the siltation and scour area data.

[0051] S45: Calculate the sediment scouring and silting intensity according to the real-time cross-section monitoring characteristic data to generate grid unit scouring and silting intensity data; In an embodiment of the present invention, parameters related to scouring and silting intensity are extracted from the real-time cross-section monitoring characteristic data, such as water flow velocity, water depth, sediment concentration, bottom shear stress, etc. These parameters can reflect the scouring and silting ability of water flow on the riverbed. Then, the scouring and silting intensity of each grid unit is calculated using empirical formulas or hydrodynamic models. Scouring and silting intensity can be expressed as the amount of sediment scouring or silting per unit area per unit time. Commonly used scouring and silting intensity calculation formulas include Einstein formula, Meyer-Peter Müller formula, etc.

[0052] S46: According to the excessive siltation and scouring data and the grid unit scouring and silting intensity data, the early warning event is judged, and the siltation and scouring area data is uploaded to the cloud platform through the edge computing node for abnormal event alarm, and the river section monitoring data is obtained.

[0053] In an embodiment of the present invention, the over-limit data of siltation and scouring and the scouring and silting intensity data of the grid unit are considered to determine whether an early warning event occurs. The judgment criteria of the early warning event can be set according to specific needs. For example, the following criteria can be set: 1. There is an over-limit scouring or silting area, and the area of ​​the area is greater than a certain threshold. 2. The average scouring and silting intensity in the over-limit area is greater than a certain threshold. 3. The duration of the over-limit area is greater than a certain threshold. If one or more of the above criteria are met, it is determined that an early warning event has occurred. Once it is determined that an early warning event has occurred, the edge computing node immediately uploads the siltation and scouring area data (including location, range, over-limit type, scouring and silting intensity, etc.) to the cloud platform through the network (such as 4G / 5G, optical fiber). After the cloud platform receives the data, it triggers an alarm mechanism, such as sending SMS, email or APP notification to relevant personnel. At the same time, the cloud platform can store the data in the database and display it visually, such as marking the siltation and scouring area on the map. These alarm information and visualization data constitute the river section monitoring data, which is used to monitor the river conditions in real time and promptly discover and handle abnormal situations.

[0054] The present application is to build a highly integrated, multi-source heterogeneous river water conservancy perception network by deploying a perception module integrating a high-definition camera, an acoustic Doppler current profiler (ADCP) and a lidar scanner at the edge of the river channel, and connecting them using edge computing nodes. This network can achieve all-round and three-dimensional perception of multiple factors such as river flow, sediment, riverbed topography, etc. The system completes the time synchronization, spatial registration and water conservancy feature extraction of data on the edge side close to the data source, ensuring the temporal consistency of data collected by different sensors. The spatial registration criterion unifies these data into the same geographic coordinate system, thereby obtaining river channel perception data that is consistent in time and space. Based on these data, the system further extracts key characteristic parameters reflecting the state of river flow and sediment movement, providing high-quality input data for constructing a three-dimensional flow field model. The introduction of edge computing greatly reduces the latency and bandwidth requirements of data transmission, improves the efficiency of data processing, and enables the system to respond quickly to changes in river sediment. Especially when dealing with sudden sediment flow events, this low-latency data processing capability can buy valuable time for timely warning and disposal. Based on the spatiotemporal distribution data of sediment migration, the system can automatically identify high-risk areas for siltation and scouring, and upload this information to the cloud platform in real time through edge computing nodes, triggering abnormal event alarms. This active early warning mechanism has changed the passive response mode of traditional water conservancy monitoring, and realized real-time monitoring and risk warning of river sediment changes. The application of the cloud platform enables managers to remotely grasp the real-time status of the river and take timely response measures, such as adjusting the operation mode of water conservancy projects, dredging waterways, and strengthening embankments, thereby effectively ensuring the flood safety and navigation capacity of the river.

[0055] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0056] 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 water conservancy perception system based on edge computing, characterized in that: Includes the following modules: The river channel edge perception module includes a high-definition camera, an acoustic Doppler current profiler, and a laser radar scanner, which is used to connect the high-definition camera, the acoustic Doppler current profiler, and the laser radar scanner to edge computing nodes to build a river channel water conservancy perception network; real-time water conservancy monitoring data is collected based on the river channel water conservancy perception network to obtain river channel water conservancy perception data; The water conservancy feature analysis module is used to synchronize the time and space of the river water conservancy perception data to obtain the spatiotemporal river perception data; the water conservancy monitoring feature analysis is performed based on the spatiotemporal river perception data to obtain the real-time section monitoring feature data; The three-dimensional flow field model is processed according to the real-time cross-section monitoring characteristic data to obtain a three-dimensional sediment flow field network model; Sediment migration simulation module, used to simulate the particle flow field of the three-dimensional sediment flow field network model and generate particle trajectory data; Perform sediment migration analysis based on particle trajectory data to generate spatial and temporal distribution data of sediment migration; The sedimentation and scouring early warning module is used to identify the sedimentation and scouring areas based on the temporal and spatial distribution data of sediment migration and obtain the sedimentation and scouring area data; The data of siltation and scouring areas are uploaded to the cloud platform through edge computing nodes for abnormal event alarms, and river section monitoring data is obtained.

2. The water conservancy sensing system based on edge computing according to claim 1 is characterized in that: The river edge perception module includes the following functions: Obtaining the elevation terrain data of the target river; Mark the key sections of the target river elevation terrain data and generate key section location data; According to the key section location data, high-definition cameras, acoustic Doppler flow profilers and lidar scanners are installed at the key sections of the river to obtain the layout data of water conservancy sensing equipment; Based on the deployment data of water conservancy sensing equipment, edge computing nodes of high-definition cameras, acoustic Doppler current profilers and lidar scanners are connected to build a river water conservancy sensing network; Real-time water conservancy monitoring data is collected based on the river water conservancy sensing network to obtain river water conservancy sensing data.

3. The water conservancy sensing system based on edge computing according to claim 2 is characterized in that: The marking of key sections of the target river elevation terrain data includes: Extract the centerline of the target river channel from the elevation terrain data and identify the left and right riverbanks to obtain the geometric constraint data of the river channel; Based on the river channel geometric constraint data, the preset equal spacing threshold is used to divide the section perpendicular to the river channel centerline to obtain the river channel section data; The river section data is spatially superimposed through the target river elevation terrain data to generate river section elevation data; Perform cross-sectional morphology analysis based on the river section elevation data to obtain cross-sectional geometric parameters; Calculate the cross-sectional shape coefficient according to the cross-sectional geometric shape parameters to generate the cross-sectional shape coefficient; Based on the preset key section identification threshold, the river section data is screened for key sections through the section morphology coefficient to generate key section location data.

4. The water conservancy sensing system based on edge computing according to claim 2 is characterized in that: The water conservancy feature analysis module includes the following functions: Perform time synchronization and spatial registration on river water conservancy sensing data to obtain spatiotemporal river sensing data; The spatiotemporal river channel sensing data is transmitted to the edge computing node, and the data is preprocessed to generate water conservancy monitoring data; wherein the water conservancy monitoring data includes cross-sectional water body image data, cross-sectional flow velocity profile data, and cross-sectional riverbed topography data; Conduct water conservancy monitoring feature analysis based on water conservancy monitoring data to obtain real-time cross-section monitoring feature data; Construct a three-dimensional spatial network based on the target river elevation terrain data to generate a three-dimensional spatial model of the river; The three-dimensional flow field model of the river channel is processed by real-time cross-section monitoring feature data to obtain a three-dimensional sediment flow field network model.

5. The water conservancy sensing system based on edge computing according to claim 4 is characterized in that: The water conservancy monitoring feature analysis includes: Perform sediment particle texture recognition on the cross-sectional water body image data to obtain sediment particle texture data; According to the sediment particle texture data, the particle size statistics are carried out, and the particle size mean is calculated to generate real-time sediment particle size data; Fitting the near-bottom velocity distribution curve to the cross-sectional velocity profile data to obtain the near-bottom velocity distribution curve data; Based on the cross-sectional riverbed topography data and using the near-bottom velocity distribution curve data, the riverbed bottom shear stress is calculated to generate bottom shear stress data; According to the cross-section velocity profile data and bottom shear stress data, the cross-section real-time concentration is estimated to generate the cross-section sediment concentration distribution data; The real-time sediment particle size data, cross-section sediment concentration distribution data and bottom shear stress data are integrated into water conservancy monitoring characteristics to generate real-time cross-section monitoring characteristic data.

6. The water conservancy sensing system based on edge computing according to claim 5 is characterized in that: The cross-section real-time concentration estimation includes: According to the cross-sectional velocity profile data, the echo intensity is corrected and the suspended sediment reference concentration is calculated to generate the suspended sediment reference concentration data; The sediment settling velocity data is queried by using the preset sediment particle size-sedimentation velocity relationship database to obtain the sediment settling velocity data; Calculate the probability of sediment initiation based on sediment settling velocity data and bottom shear stress data to generate sediment initiation probability data; The sediment suspended concentration is estimated based on the suspended sediment reference concentration data and the sediment initiation probability data, and the cross-sectional concentration distribution is processed to generate the cross-sectional sediment concentration distribution data.

7. The water conservancy sensing system based on edge computing according to claim 5 is characterized in that: The three-dimensional flow field model processing of the three-dimensional spatial model of the river channel includes: Perform cross-section three-dimensional coordinate transformation based on cross-section sediment concentration distribution data to generate cross-section concentration position data; The three-dimensional spatial model of the river channel is interpolated through the cross-section concentration position data to obtain a three-dimensional sediment concentration network model; The bottom boundary conditions of the riverbed are set according to the cross-section riverbed topography data and the bottom shear stress data, and the bottom boundary condition data are generated; Based on the bottom boundary condition data and the three-dimensional sediment concentration network model, a three-dimensional flow field numerical simulation model is constructed through the Navier-Stokes equation to generate a three-dimensional sediment flow field network model.

8. The water conservancy sensing system based on edge computing according to claim 5 is characterized in that: The sediment migration simulation module includes the following functions: Calculate the grid unit volume based on the three-dimensional sediment flow field network model to generate grid unit volume data; Mark the cross-section monitoring points of the three-dimensional sediment flow field network model, set the virtual particle release position, and obtain the virtual particle release position data; The particle number distribution of the three-dimensional sediment flow field network model is carried out through the cross-section sediment concentration distribution data to obtain the virtual particle number data; Based on the virtual particle release position data and virtual particle quantity data, the particle flow field simulation is performed on the three-dimensional sediment flow field network model, and Lagrangian particle tracking is performed to generate particle trajectory data; Sediment migration analysis is performed based on particle trajectory data to generate spatiotemporal distribution data of sediment migration.

9. The water conservancy sensing system based on edge computing according to claim 8 is characterized in that: The sediment migration analysis based on particle trajectory data includes: Perform grid particle average displacement vector statistics based on particle trajectory data to generate grid average displacement data; Calculate the sediment migration flux based on the grid average displacement data to generate sediment migration flux data; The theoretical particle sediment carrying capacity is converted according to the sediment migration flux data to obtain the theoretical particle sediment carrying capacity; The sediment migration rate is calculated for the three-dimensional sediment flow field network model through the sediment migration flux data to generate sediment migration rate data; The theoretical particle sediment carrying capacity and sediment migration rate data are integrated into the grid spatiotemporal distribution to generate the spatiotemporal distribution data of sediment migration.

10. The water conservancy sensing system based on edge computing according to claim 2 is characterized in that: The siltation and scouring early warning module includes the following functions: Calculate the net sediment transport of grid cells based on the temporal and spatial distribution data of sediment migration to generate net sediment transport data; Based on the net sediment transport data, the grid riverbed elevation change is calculated through a preset time step to obtain the riverbed elevation change data; The target river elevation terrain data and the riverbed elevation change data are superimposed, and compared with the preset riverbed terrain scour threshold to obtain the siltation scour excess data; According to the over-limit data of siltation and scour, the siltation and scour area is identified to obtain the siltation and scour area data; Calculate the sediment scouring and silting intensity based on the real-time cross-section monitoring characteristic data to generate grid unit scouring and silting intensity data; Early warning events are judged based on the excessive siltation and scouring data and the grid unit scouring and silting intensity data, and the siltation and scouring area data are uploaded to the cloud platform through the edge computing node for abnormal event alarm to obtain river section monitoring data.

Images