Flood control and disaster reduction method for multi-engineering systems based on four pre-coupling models

Abstract

The present invention discloses a multi-project flood prevention and disaster reduction method based on a four-pre-coupling model. The method constructs a reservoir-river-levee multi-project system flood prevention and disaster reduction forecast model, a reservoir-river-levee multi-project system flood prevention and disaster reduction early warning model, a reservoir-river-levee multi-project system flood prevention and disaster reduction rehearsal model, and a reservoir-river-levee multi-project system flood prevention and disaster reduction contingency plan model. The method plans and designs a monitoring network according to certain rules, and performs a dynamic coupling of the four pre-couplings, including forecasting, early warning, rehearsal, and pre-coupling. Compared with existing technologies, the present invention comprehensively considers the synergistic effects and influencing constraints of the flood prevention processes of reservoirs, rivers, and levees. By coupling the multi-project system flood prevention "four pre-couplings" dynamic coupling model, the method achieves a chain-like dynamic connection of the four pre-couplings for flood prevention and disaster reduction, achieving precise flood forecasting, scientific flood prevention and disaster reduction decision-making, and refined flood risk management.

Description

Technical Field

[0001] The present invention relates to the field of emergency disaster prevention and flood control and disaster reduction intelligent solutions, and in particular to a reservoir-river-levee multi-engineering system flood control and disaster reduction four-pre-coupling model. Background Art

[0002] The reservoir-river-levee multi-engineering system flood control and disaster reduction four-pre-coupling model is a model that uses "reservoir-river-levee" as an integrated unit to realize the flood "forecast-warning-rehearsal-plan" coupling functions. It is one of the important non-engineering measures to comprehensively predict, assess and manage flood disaster risks and reduce flood disaster losses.

[0003] Reservoirs, rivers, and levees are crucial infrastructure for flood control and disaster reduction. Their flood control capabilities are crucial for protecting people downstream and minimizing the area affected by disasters. Currently, a diverse, technology-intensive model system has been established in the field of flood control and disaster reduction. This system relies primarily on hydrological and hydrodynamic models, remote sensing and geographic information system technologies, and numerical simulation techniques, with the incorporation of artificial intelligence and data mining technologies. This has led to a relatively rich development in the research and application of flood control and disaster reduction models. However, current research and practice on flood control and disaster reduction models has not yet fully integrated the "reservoir-river-levee" system as an integrated unit. Instead, these three components are studied separately or in pairs. This often lacks systematic thinking and planning from a holistic perspective, making it difficult to fully consider the natural synergistic effects and mutual influences among the three. This, in turn, reduces the resilience and overall effectiveness of the flood control and disaster reduction system, hindering the scientific scheduling and management of flood risks. More critically, current research lacks an inherent dynamic coupling mechanism that considers the "four pre-emptive" functions. This makes it difficult to proactively rehearse the entire physical basin and the entire process of water conservancy management activities in the face of complex and changing flood situations, ensuring early risk detection, early warning issuance, early plan development, and early implementation of measures. Furthermore, the coupling mechanism between existing "four pre-emptive" models is incomplete, lacking deep integration and collaborative optimization. This leads to significant deficiencies in information sharing and functional complementarity between models, making it difficult to achieve a chain-like dynamic connection between flood control and disaster reduction links. This, to a certain extent, hinders the full realization of overall flood control and disaster reduction effectiveness, and the lack of an efficient and unified flood control and disaster reduction system. Therefore, establishing and improving a four-preemptive coupled model for flood control and disaster reduction in complex engineering systems, taking the reservoir-river-levee multi-engineering system as the overall unit, is of great significance for improving the scientific nature and precision of basin-wide flood risk management.

[0004] In summary, although the field of flood control and disaster reduction already has diversified water conservancy professional models with relatively rich functions, the independent research of "reservoirs", "rivers" or "levees" and the lack of coupling mechanisms between the "four predictions" (forecast, early warning, rehearsal, and contingency plan) functional models are key factors that restrict the further development of the field of flood control and disaster reduction. In view of this, the present invention integrates the "reservoir-river-levee" integrated unit and the "four predictions" model coupling mechanism through scientific methods such as theoretical analysis and numerical simulation. Starting from the overall perspective of "reservoir-river-levee", with forecast as the basis, early warning as the outpost, rehearsal as the key, and contingency plan as the purpose as the forward-looking idea, the four prediction functions of forecast, early warning, rehearsal, and contingency plan for flood control and disaster reduction are realized. At the same time, the dynamic coupling between models is strengthened, and the interaction and influence mechanism between each model are clarified, thereby proposing a reservoir-river-levee multi-engineering system flood control and disaster reduction four-prediction coupling model method. Summary of the Invention

[0005] Aiming at the shortcomings of the current separate research on "reservoirs, rivers and levees" in the field of flood control and disaster reduction, and the defects in the coupling mechanism between the "four pre-" functional models, and focusing on the major needs and development trends of reservoir flood control emergency management, this invention aims to propose a multi-engineering system flood control and disaster reduction method based on the four pre-coupling models. "Reservoir-river-levee" is taken as an integrated research unit. Through the coupling integration and coordinated optimization of the four pre-coupling models, a "forecast-warning-rehearsal-plan" model system for flood control and disaster reduction for the reservoir-river-levee multi-engineering system is realized.

[0006] The present invention is achieved by utilizing the following technical solutions:

[0007] A multi-project flood control and disaster reduction method based on a four-precoupling model specifically includes:

[0008] A flood control and disaster reduction forecasting model for a reservoir-river-levee multi-engineering system is constructed. The forecasting model is at least a model system formed by multi-dimensional deep coupling including a meteorological rainfall forecasting model, a reservoir upstream area flood forecasting model, a reservoir scheduling and downstream river levee risk forecasting model and a flood inundation disaster forecasting model; wherein the meteorological rainfall forecasting model takes gridded actual rainfall data as input and gridded meteorological rainfall forecast data as output; the reservoir upstream area flood forecasting model takes gridded meteorological rainfall forecast data output by the meteorological rainfall forecasting model as input; the reservoir scheduling and downstream river levee risk forecasting model takes the output of the reservoir upstream area flood forecasting model as input. The model includes a reservoir engineering flood control scheduling sub-model and a reservoir downstream river channel-levee overflow and breach one- and two-dimensional hydrodynamic coupling sub-model. The reservoir engineering flood control scheduling sub-model combines multi-source data including terrain elevation, river system, and water conservancy project facility scheduling rules to obtain output 1 as the reservoir scheduling discharge flow. The process is used as the input condition of the one-dimensional hydrodynamic coupling sub-model of the river channel and levee overflow and breach downstream of the reservoir. The one-dimensional hydrodynamic coupling sub-model of the river channel and levee overflow and breach downstream of the reservoir realizes the simulation of the flood evolution and the flooding risk of the river channel downstream of the reservoir, and obtains output 2, which is a dynamic simulation of reservoir operation and downstream river channel flood evolution, flooding risk of both sides of the levee, calculates the flooding range caused by typical frequency floods, and predicts and evaluates the impact of different magnitude floods on reservoir flood control safety, levee safety and possible flooding risk of coastal areas; output 1 and output 2 are used together as the output of the reservoir operation and downstream river channel levee risk prediction model; the flood disaster prediction model takes the output of the flood prediction model of the upstream area of the reservoir and output 1 and output 2 of the reservoir operation and downstream river channel levee risk prediction model as input, and takes the flood disaster prediction result of the extreme disaster events of reservoir overburden and river channel levee overburden on the downstream area as the output;

[0009] Construct a flood prevention and disaster reduction early warning model for a multi-engineering system of reservoirs, rivers and levees. The early warning model is at least a model system formed by multi-dimensional deep coupling including a meteorological risk early warning model, a reservoir risk early warning model, a dam-down river levee risk early warning model, and a reservoir downstream residents flooding risk early warning model; wherein, the meteorological risk early warning model takes observation of regional rainfall and the formation of 24h hourly rainfall forecast data as input, and takes meteorological risk level early warning results as output; the reservoir risk early warning model further includes a rain gauge monitoring early warning sub-model, an inflow flood early warning sub-model, a reservoir area flooding risk early warning sub-model and a dam engineering safety early warning sub-model, and the rain gauge monitoring early warning sub-model, a reservoir flooding risk early warning sub-model and a dam engineering safety early warning sub-model. The monitoring and early warning sub-model of the measuring station takes the measured rainfall data as input and takes the judgment result of whether the rainfall triggers the rainfall early warning response as output; the reservoir area flooding risk early warning sub-model takes the reservoir area water level all-weather monitoring data as input and takes the flooding risk forecast result as output; the dam engineering safety early warning sub-model includes dam overflow risk early warning processing and dam body instability early warning processing. Specifically, the dam overflow risk early warning processing takes the inflow flood and the measured reservoir water level monitoring data, combined with the water level storage capacity relationship and the reservoir discharge capacity input 1, to predict the reservoir water level development trend as the output 1 of the sub-model; the dam body instability early warning processing takes the dam body moisture content, crack development degree, The dam body hazard data of the infiltration line height, soil deformation and slope balance safety factor are input 2, and the real-time monitoring of the structural safety status of the dam (embankment) is the output 2 of the model; the dam downstream river embankment risk warning model further includes a rain station monitoring and warning sub-model, a river flood risk simulation and warning sub-model and an embankment engineering safety warning sub-model. Specifically, the rain station monitoring and warning sub-model takes rainfall monitoring data as input and triggers a rainfall warning response of the corresponding level as output; the river flood risk simulation and warning sub-model takes the model calculation boundary conditions output by the reservoir engineering flood control scheduling sub-model as input, and the downstream river flood risk warning response result is The output of this model; the embankment engineering safety early warning sub-model takes as input basic data including at least the embankment soil moisture content, crack development degree, infiltration line height, soil deformation, and slope balance safety factor, and takes the embankment disaster accident early warning response results as the output of this model; the reservoir downstream residents flooding risk early warning model further includes a monitoring early warning sub-model and a simulation early warning sub-model. The monitoring early warning sub-model takes as input dynamic monitoring images of possible flooded areas, and provides data support and decision-making basis for disaster assessment and rescue work as output; the simulation early warning sub-model is based on the simulation of flood evolution in the river downstream of the reservoir and flooding risks on both sides of the river;

[0010] Construct a flood control and disaster reduction simulation model for a multi-engineering system of reservoirs, river channels, and levees. This model takes terrain elevation, river systems, and remote sensing imagery data of the downstream reservoir area as input, and provides a dynamic visualization of the entire process of rainfall runoff, reservoir operation, flood evolution, and inundation as output.

[0011] Constructing a reservoir-river-levee multi-engineering system flood control and disaster reduction plan model by combining the aforementioned reservoir-river-levee multi-engineering system flood control and disaster reduction forecast model, the aforementioned reservoir-river-levee multi-engineering system flood control and disaster reduction early warning model, and the aforementioned reservoir-river-levee multi-engineering system flood control and disaster reduction rehearsal model, using the outputs of the aforementioned forecast model, early warning model, and rehearsal model as inputs, and using the reservoir-levee-flood zone multi-engineering system risk control plan under standard flood conditions as output;

[0012] Plan and design a monitoring network, including at least rain gauges, hydrological stations, water level stations and video stations, and conduct four-pronged dynamic coupling including forecast, warning, rehearsal and plan.

[0013] In some embodiments, the reservoir upstream area flood forecasting model further includes a physical mechanism flood forecasting sub-model and a data-driven flood forecasting sub-model.

[0014] In some embodiments, the physical mechanism flood forecasting submodel is a combined model consisting of a series-coupled SCS flow generation submodel, an instantaneous unit line flow submodel, and a Muskingum flood evolution submodel.

[0015] In some embodiments, the data-driven flood forecast sub-model adopts an LSTM model, wherein the input of the flood forecast LSTM model is the rainfall process and runoff process in the previous n hours and the rainfall process in the next m hours, and the output is the runoff process in the next m hours; a forget gate is used to discard information that is not important to the flood forecast and control the degree of information retention; an update gate is used to update the neural network model with new important information to generate candidate values; an output gate is used to determine the important information that needs to be output and generate a hidden state for prediction or as input for the next time step.

[0016] Compared with the prior art, the present invention has the following positive technical effects:

[0017] From the perspective of overall view and global view, we can comprehensively consider the synergistic effect and influence constraint relationship of the three flood control processes of "reservoirs, rivers and embankments". By constructing a dynamic coupling model of flood control "four pre-emptives" with reservoirs as the core and considering the upstream and downstream and left and right banks of multiple engineering systems, we can realize the chain dynamic connection of the "four pre-emptives" links of flood control and disaster reduction. This will help to realize the precision of flood forecasting, scientific flood control and disaster reduction decision-making and refined flood risk management from the perspective of the river basin as a whole, timely intercept and discharge floods, reduce flood peaks, effectively control floods, improve the intelligence level of reservoir flood control emergency management, improve the efficiency and accuracy of flood control decision-making, significantly reduce the impact of flood disasters in the upstream and downstream areas of the reservoir, and provide important technical support for water security. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is an overall flow chart of the flood prevention and disaster reduction method for a multi-engineering system based on a four-precoupling model of the present invention;

[0019] Figure 2 The algorithm block diagram of the artificial neural network LSTM model used is shown in the figure.

[0020] Figure 3 This is the geographical location map of the study area;

[0021] Figure 4 This is the flood forecast process line for the upstream area of a certain mountain pass reservoir;

[0022] Figure 5 Dispatching the discharge flow process line for a certain mountain pass reservoir;

[0023] Figure 6 This is the preview model result - maximum flood depth map I (key flooding area);

[0024] Figure 7 This is the preview model result - maximum flood depth map II (key flooding areas);

[0025] Figure 8 Planning a monitoring station network diagram for a four-pre-coupling model of a mountain pass reservoir. DETAILED DESCRIPTION

[0026] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0027] like Figure 1 As shown, the overall process of the multi-project flood prevention and disaster reduction method based on the four-precoupling model of the present invention specifically includes the following steps:

[0028] Step 1: Construct a flood control and disaster reduction forecast model for the reservoir-river-levee multi-engineering system;

[0029] The forecast model is at least a multi-dimensional, deeply coupled model system consisting of a meteorological rainfall forecast model, a flood forecast model for the upstream area of the reservoir, a reservoir operation and downstream river embankment risk forecast model, and a flood inundation disaster forecast model. This forecast model meets the requirements of real-time rolling forecasting, sets water levels, flow rates, and water volumes at different points or sections as forecast elements and quantifies them, conducts all-round, multi-level, and different forecast period flood disaster forecasts for the reservoir and its upstream and downstream river embankment areas, and analyzes the dynamic coupling mechanism between the various forecast methods. The specific description is as follows:

[0030] 1.1. The meteorological rainfall forecast model deploys phased array hydro-rainfall radars in areas prone to rainstorms and flash floods, as well as meteorological weather radars around cities, to observe and forecast rainfall at different spatial scales in different regions, shifting rainfall monitoring from "falling rain" to "rain in the clouds." The hydro-rainfall radars can observe liquid water in the near-surface atmosphere from above ground level to an altitude of 2 km, generating high-precision, gridded rainfall data with a 30m×30m grid and a temporal resolution of 40 seconds in real time. The model outputs high-resolution, gridded meteorological rainfall forecast data with an extrapolation time of 1-3 hours, based on a network of three rainfall radars. The meteorological weather radars can observe all atmospheric meteorological elements from above ground level to the tropopause, within an altitude of 20-30 km, providing a wide-area sampling observation of the atmosphere, addressing vertical atmospheric detection challenges and meeting the rainfall forecast requirements for flood control.

[0031] 1.2. The flood forecast model for the upstream reservoir area uses the gridded meteorological rainfall forecast data output by the meteorological rainfall forecast model described in 1.1 as input. Specifically, it is a high-precision flood forecast model for the upstream reservoir area, driven by both physics and data. The physical mechanism flood forecast submodel primarily includes the SCS runoff generation submodel, the instantaneous unit line confluence submodel, and the Muskingum flood evolution submodel. The runoff generation, confluence, and flood evolution models are coupled in series, with the output of each model serving as the input to the next. The SCS runoff generation submodel models the process of runoff generation from rainfall through the loss phase. The instantaneous unit line confluence submodel simulates the entire runoff process. The Muskingum flood evolution submodel considers flow velocity, discharge, and water level to simulate the flood process. The data-driven flood forecast sub-model mainly includes artificial neural network models such as LSTM models. The input of the flood forecast LSTM model is the rainfall process and runoff process in the previous n hours and the rainfall process in the next m hours. The output is the runoff process in the next m hours. In the present invention, n and m are generally integer multiples of 6. In the present invention, n and m are generally integer multiples of 6. The forget gate is used to discard the relevant information that is not important to the flood forecast and control the degree of information retention. The update gate is used to update the new important information in the neural network model and generate candidate values. The output gate is used to determine the important information that needs to be output and generate a hidden state for prediction or as the input of the next time step, such as Figure 2 As shown in the figure, the real-time dynamic coupling of the water conservancy rainfall radar monitoring and forecasting data and the flood forecasting model in the upstream area of the reservoir is realized; at the same time, the model is calibrated and determined in combination with the historical experience rainfall type database, and the key indicators of the flood process line (flood peak value, flood peak arrival time, total flood volume, and flood duration) are obtained after adjusting the parameters of the flood forecasting model in the upstream area of the reservoir, thereby exploring the impact mechanism of regional rainfall on the flood process in the upstream area of the reservoir.

[0032] 1.3. A reservoir operation and downstream river levee risk prediction model uses the output of the reservoir upstream flood prediction model described in 1.2 as input. This model includes a reservoir engineering flood control operation submodel and a one- and two-dimensional hydrodynamic coupling submodel for the reservoir downstream river channel and levee overtopping and breach. The reservoir engineering flood control operation submodel combines multi-source data such as high-precision terrain elevation, river systems, and water conservancy project operation rules. Output 1 is the reservoir operation discharge process, which serves as the input condition for the one- and two-dimensional hydrodynamic coupling submodel for the reservoir downstream river channel and levee overtopping and breach. The one- and two-dimensional hydrodynamic coupling submodel for the reservoir downstream river channel and levee overtopping and breach simulates the downstream river channel flood evolution and the inundation risk of the banks. Output 2 is a dynamic simulation of reservoir operation, downstream river channel flood evolution, and levee overtopping and breach risk. It calculates the flood inundation range caused by typical frequency floods and forecasts and assesses the impact of floods of different magnitudes on reservoir flood control safety, levee safety, and the potential inundation risk of coastal areas. Output 1 and output 2 are used together as the output of the reservoir operation and downstream river embankment risk prediction model.

[0033] 1.4. Flood inundation disaster forecasting model. The model uses outputs 1 and 2 of the reservoir upstream flood forecasting model described in 1.2 and the reservoir scheduling and downstream river embankment risk forecasting model described in 1.3 as inputs. The model outputs flood inundation disaster forecast results for the downstream area inundation disaster risk caused by extreme catastrophic events such as reservoir dam breaches and river embankment breaches, which are of particular concern. Combined with socioeconomic data (basic statistical indicators such as population, cultivated land, and gross domestic product), the model output is evaluated by constructing a GIS-based flood disaster impact analysis and loss assessment model to comprehensively forecast and assess flood impact losses.

[0034] Step 2: Construct a flood prevention and disaster reduction early warning model for the reservoir-river-levee multi-engineering system;

[0035] The early warning model is at least a model system formed by multi-dimensional deep coupling including a meteorological risk warning model, a reservoir risk warning model, a dam downstream river embankment risk warning model, and a reservoir downstream residential flooding risk warning model, and the model proposes a planning monitoring station network layout plan based on the "early warning" needs of flood control and disaster reduction; wherein, the reservoir risk warning model further includes a rain gauge monitoring and early warning sub-model, an inflow flood early warning sub-model and a reservoir area rising and flooding risk early warning sub-model; the dam downstream river embankment risk early warning model further includes a rain gauge monitoring and early warning sub-model, a river flood risk early warning sub-model and an embankment engineering safety early warning sub-model. The model takes into account at least the changing factors including the characteristics of rainstorms and floods in small watersheds, previous rainfall or soil moisture content as input, analyzes the natural characteristics of floods, the safety of reservoir-river-levee projects and the socio-economic impact of flood disasters, and combines meteorological monitoring and flood forecast results to determine a flood control and disaster reduction early warning indicator system consisting of indicators such as critical rainfall, critical water level and critical flow. The model uses reservoir-river-levee design data, hydrological and historical flood disaster data as input, determines the threshold of flood risk early warning indicators, and realizes real-time dynamic early warning of the time and scale of flood disasters as output. Warning information is issued for specific areas and specific groups of people to support the emergency evacuation, rescue and resettlement of people in the disaster area.

[0036] 2.1. The meteorological risk warning model uses regional rainfall observations and 24-hour rainfall forecasts as input. A network of phased array hydrological rainfall radars and meteorological weather radars are deployed in areas upstream and downstream of reservoirs prone to rainstorms, floods, and flash floods. Gridded rainfall (which varies with soil moisture) for each warning period (3, 6, 12, and 24 hours) serves as a meteorological flood risk warning indicator. Warning indicator thresholds are calculated for four meteorological risk levels: low (possible, blue warning), medium (high probability, yellow warning), high (high probability, orange warning), and extremely high (very high probability, red warning). Warning results for each meteorological risk level are output.

[0037] 2.2 Reservoir risk early warning model, specifically including the following sub-models:

[0038] (1) Rain gauge monitoring and early warning sub-model: Relying on a dense network of rain gauge stations upstream and downstream of the reservoir, real-time monitoring of rainfall data during each early warning period is carried out, and various rainfall indicators such as hourly maximum rainfall, 24-hour rainfall, regional average rainfall, and cumulative rainfall are observed and recorded. The sub-model uses the measured rainfall data as input and adopts the analytical calculation method of inversely estimating the critical rainfall from the disaster water level and the design rainstorm flood. The rainfall during each early warning period is selected as the real-time dynamic early warning indicator, and the output of the sub-model is the judgment result of whether the rainfall triggers a rainfall early warning response.

[0039] (2) Inflow flood warning sub-model: The rainfall observation data from the surface rain gauge station upstream of the reservoir is used as the input of this sub-model. Combined with the output of the flood inundation disaster forecasting model, the inflow flood process calculation and real-time warning are realized. In addition, the reservoir tail flow monitoring system is used to obtain key indicators such as the inflow flood flow in real time. Combined with the reservoir water storage capacity and inflow conditions, the flood safety of the reservoir area is analyzed to realize real-time monitoring and warning of rainfall and flood.

[0040] (3) Reservoir flooding risk warning sub-model: Relying on the reservoir area water level all-weather monitoring system, the dynamic changes of water level are analyzed in real time. The all-weather monitoring data of the reservoir area water level is used as the input of this sub-model. The threshold of the water level warning indicator corresponding to the flooding risk is set to judge in real time whether the water level exceeds the standard or is about to exceed the standard. If so, the flooding risk monitoring and warning is triggered. In addition, based on the output of the flood disaster prediction model in 1.4 and the reservoir project flood control scheduling sub-model in 1.3, combined with the upstream water, water level and storage capacity relationship, reservoir scheduling and operation mode and rainfall forecast results, the water level change process during the reservoir operation is dynamically calculated. The flooding risk prediction result is used as the output of this model to comprehensively evaluate the reservoir's rising trend and potential flooding risk, judge whether the standard water level exceeds the standard, and trigger the flooding risk prediction and warning if so.

[0041] (4) Dam engineering safety early warning sub-model: It includes two aspects of processing: ① Dam overflow risk early warning processing: The input of this sub-model is the flood inflow and the actual water level monitoring data of the reservoir, combined with the water level storage capacity relationship and the reservoir discharge capacity. The output of this sub-model is the predicted reservoir water level development trend based on the flood inflow and the actual water level monitoring data of the reservoir, combined with the water level storage capacity relationship and the reservoir discharge capacity. If the reservoir water level exceeds the dam top elevation, the dam overflow disaster accident early warning response is triggered. ② Dam instability early warning processing: The model uses dam hazard data such as dam moisture content, crack development degree, seepage line height, soil deformation, and slope balance safety factor as input2. High-precision sensors are used to monitor dam hazard data such as dam moisture content, crack development degree, seepage line height, soil deformation, and slope balance safety factor. The model outputs real-time monitoring of the dam (embankment) structural safety status through video remote monitoring and regular staff inspections2. By comparing the model with preset safety thresholds and dam design standards, the model assesses the dam's stability and risk level, triggering an early warning response to dam instability and damage disasters.

[0042] 2.3. The dam downstream river embankment risk early warning model includes the following sub-models:

[0043] (1) Rain gauge monitoring and early warning sub-model: This sub-model uses rainfall monitoring data as input and relies on the network of rainfall gauges below the dam to monitor rainfall data in real time. The critical rainfall is inferred from the disaster water level and the design rainstorm flood. The rainfall during each warning period for the protected areas on both sides of the river is selected as the real-time dynamic early warning indicator. The output of this sub-model is the rainfall warning response that triggers the corresponding level. The sub-model determines if the rainfall exceeds the standard and triggers the corresponding level of rainfall warning response. The early warning indicators are divided into two levels, corresponding to the preparation and immediate relocation of the disaster prevention objects.

[0044] (2) River flood risk simulation and early warning sub-model: Taking the model calculation boundary conditions output by the reservoir engineering flood control and dispatching sub-model in 1.3 as input, based on the output of the reservoir engineering flood control and dispatching sub-model, and combined with information such as river topography, embankment elevation, and remote sensing images, a coupled model for simulating the evolution of river floods in the downstream area of the reservoir and the inundation risk of both banks is established. The model dynamically simulates the evolution of floods in the downstream area of the reservoir, calculates and predicts the inundation range of floods or floods with typical frequency, and monitors key parameters such as river water level and flow in real time. The model comprehensively evaluates the possibility that river floods exceed the embankment warning water level and guaranteed water level, and exceed the surface elevation of the dangerous areas or protection areas on both banks, triggering a downstream river flood risk early warning response. The downstream river flood risk early warning response result is the output of the model.

[0045] (3) Levee engineering safety early warning sub-model: The sub-model takes the basic data such as levee soil moisture content, crack development degree, infiltration line height, soil deformation, and slope balance safety factor as input. The sub-model monitors the basic data such as levee soil moisture content, crack development degree, infiltration line height, soil deformation, and slope balance safety factor through high-precision sensors. The sub-model uses video remote monitoring and regular staff inspections to monitor the structural safety of the levee in real time. The sub-model compares the preset safety threshold and the levee flood control design standard, evaluates the stability and risk level of the levee, triggers the levee disaster early warning response, and uses the levee disaster early warning response result as the output of the model.

[0046] 2.4. Early warning model for flooding risk of residents downstream of reservoirs, including the following sub-models:

[0047] (1) Monitoring and early warning sub-model: The dynamic monitoring images of the possible flooded areas are used as the input of this sub-model. High-definition video monitoring, satellite remote sensing image recognition, drone aerial photography and other "air-ground" real-time three-dimensional monitoring methods are used to achieve dynamic monitoring of the possible flooded areas, instantly capture the real-life images of the flood inundation process, accurately grasp the flood spread range, clearly identify the target objects in the flooded area, and evaluate the water depth and water flow velocity of the flooded area, thereby triggering the flood risk monitoring and early warning response and providing data support and decision-making basis for subsequent emergency response, prevention of secondary disasters, disaster assessment and rescue work as the output of this sub-model.

[0048] (2) Simulation and early warning sub-model: Based on the evolution of river floods downstream of the reservoir and the simulation of inundation risks on both sides of the river, it focuses on extreme disaster events such as reservoir dam breaches and river embankment breaches. Combined with socioeconomic data (statistical indicators such as population, cultivated land, and GDP), according to disaster statistics and loss assessment methods, it simulates and predicts the inundation range, inundation depth, and disaster losses of downstream residential areas under different disaster event scenarios, and triggers inundation disaster early warning responses for residents downstream of the reservoir according to the risk level.

[0049] Based on the needs of flood control and disaster reduction forecasting and early warning, the monitoring network is planned and designed according to certain rules (for example, the main principle is to cover the rainfall, water conditions, engineering conditions and dangerous conditions of monitoring reservoirs, rivers, embankments and flooding risk areas on both sides). For example, rain gauges monitor regional rainfall and are generally arranged near the centroid of the basin; hydrological stations and water level stations monitor the water surface elevation of reservoirs and rivers. Hydrological stations can also increase the monitoring of river flow processes and are generally arranged at the reservoir bank or embankment near the water; video stations monitor whether the water level reaches the water surface elevation for early warning, whether the monitoring risk area is flooded, etc., and are arranged in locations that can be monitored by remote images, including hydrological stations, rain gauges, rainfall radars, video stations, etc., to enhance the perception capabilities of the "four warnings" including forecasting, early warning, rehearsal and planning.

[0050] Step 3: Construct a flood control and disaster reduction pre-test model for the reservoir-river-levee multi-engineering system;

[0051] The model uses data such as terrain elevation, river systems, and remote sensing imagery downstream of the reservoir as input. It considers the constraints of levees and the multiple influences of interval inflow to determine model boundary conditions, river channel and floodplain roughness, and other factors. The model dynamically links the upstream flood forecast model described in 1.2, the reservoir engineering flood control and scheduling submodel described in 1.3, and the 1-D hydrodynamic coupling submodel of the downstream river channel and levee overburden to simulate the flood inundation process and distribution characteristics under super-standard flood and overburden scenarios. Using GIS, 3D modeling and rendering, and hydrological and hydrodynamic modeling, an integrated 2D and 3D digital twin scene is constructed, supporting dynamic visualization of the entire process of rainfall runoff, reservoir scheduling, flood evolution, and inundation. This dynamic visualization of the entire process of rainfall runoff, reservoir scheduling, flood evolution, and inundation serves as the module's output.

[0052] The relevant technical description is as follows:

[0053] 1. GIS technology is mainly used to process geographic spatial data, including the spatial distribution of elements such as rivers, roads, and terrain.

[0054] 2. 3D modeling and rendering technology mainly uses 3D modeling software or 3D modules in GIS software to construct 3D scenes based on geographic spatial data, including river flow direction, road layout, terrain undulations, etc., and then set reasonable lighting and shadow effects for the 3D model and add materials and textures to enhance the realism and three-dimensionality of the scene, making it more realistic.

[0055] 3. The hydrological and hydrodynamic model mainly constructs a calculation model of the entire flood process based on parameters such as the basin's topographic characteristics and meteorological conditions, sets initial conditions, boundary conditions and calculation parameters, and then runs the model to obtain risk information such as flood inundation range, water depth, and flow rate. The simulation result data is then integrated through the data interface of GIS or a three-dimensional platform. After processes such as color coding, transparency adjustment, and animation production, the flood evolution and inundation process is intuitively simulated in a three-dimensional digital scene.

[0056] Step 4: Construct a flood control and disaster reduction plan model for the reservoir-river-levee multi-engineering system;

[0057] The flood control and disaster reduction "plan" model mainly focuses on disaster prevention and control and risk reduction. It combines the reservoir-river-levee multi-engineering system flood control and disaster reduction forecast model of step one, the reservoir-river-levee multi-engineering system flood control and disaster reduction early warning model of step two, and the reservoir-river-levee multi-engineering system flood control and disaster reduction rehearsal model of step three. The output of the aforementioned forecast model, early warning model, and rehearsal model are used as the input of the plan model to simulate the reservoir scheduling and operation process under different rainstorm scenarios. Based on the principle of minimizing downstream inundation risk, the flood scheduling defense schemes under different working conditions are optimized, thereby formulating the reservoir-levee-pan-region multi-engineering system risk control scheme under super-standard flood conditions, and the reservoir-levee-pan-region multi-engineering system risk control scheme under standard flood conditions is used as the output of the plan model. The specific description is as follows:

[0058] 1. Disaster prevention and control plan, relying on the rain and water monitoring and early warning system (including satellite remote sensing, radar rainfall measurement, automatic hydrological monitoring stations, etc.) to collect information in real time, analyze the hydrological and meteorological conditions within the reservoir and its basin, and combine the forecast and early warning model to predict the future flood development situation within the reservoir-river-levee multi-engineering system, establish a regional and graded early warning release plan and recommend prevention and control measures. Among them:

[0059] (1) The early warning release plan mainly clarifies the release process, standards and business specifications of early warning information, establishes a five-level flood disaster prevention responsibility system at the county, township (town), village, group and household levels, pushes early warning information to responsible persons and the public in different districts, realizes targeted early warning, and ensures timely communication and effective reception of early warning information.

[0060] (2) Recommended prevention and control measures mainly include: ① Improve non-engineering measures. Configure video surveillance, data monitoring equipment, communication equipment, lighting equipment, emergency materials (material quantity, storage location, usage rules and responsible persons), etc.; ② Establish a day and night duty system for reservoirs. Implement the post responsibility system and strengthen inspection, monitoring and protection work; ③ Observe changes in the project. Carry out safety inspection analysis of reservoir projects, use instruments and facilities to monitor the horizontal and vertical displacement and infiltration line of the project, and focus on observing whether the dam embankment has cracks, landslides, collapses, leakage, pipe bursts and other damages. Report any problems in a timely manner and take effective emergency measures; ④ Analyze the hazards of flood events. Analyze the different levels of risks corresponding to different water levels, clarify the downstream impact areas and disposal measures of different levels, carry out reservoir dam break and flooding risk analysis, and prepare reservoir super-standard flood and dam break flood risk maps. ⑤ Establish an emergency support system, mainly including organizational support and team support; ⑥ Promote flood disaster prevention knowledge. Provide professional knowledge and technical training on flood disasters to responsible persons at all levels, technical personnel, the general public, etc., and organize flood disaster prevention drills.

[0061] 2. Risk reduction plan: The risk reduction plan mainly includes flood risk control, reservoir and embankment accident rescue, personnel evacuation and emergency rescue, etc., involving multi-departmental coordinated disaster prevention, reduction and relief. Among them:

[0062] (1) Flood risk control. This is mainly achieved by comprehensively evaluating factors such as the characteristics of rainstorms and floods, the safety status of reservoirs and embankment projects, and possible flood disasters. Flood control plans such as reservoir diversion and flood discharge are formulated to obtain the optimal reservoir discharge process and maximize flood risk control.

[0063] (2) Engineering accident rescue. Carry out emergency repairs on reservoirs, levees and other engineering facilities damaged during floods, reinforce emergency levees, and pre-plan response strategies such as rescue team organization, personnel allocation, task allocation, material preparation, typical emergency situation handling plans, and information communication and coordination methods.

[0064] (3) Flood evacuation and emergency rescue. A detailed resettlement plan should be developed for low-lying areas and flood-inundated areas in the rehearsal model. The main contents include a survey of the people in the danger zone, the time of evacuation, the evacuation route and resettlement point, transportation, medical assistance points and medical rescue teams, life-saving equipment, and daily necessities.

[0065] Step 5: Dynamically couple the four pre-plan models of “forecast-warning-rehearsal-preparatory plan”;

[0066] About dynamics: Flood prevention and disaster reduction takes "flood" as the main line, which mainly reflects the spatiotemporal dynamic characteristics of water flow in the four forecast periods. Forecast-warning-rehearsal-plan all cover the upstream and downstream, left and right bank spatial areas with the reservoir as the core. From the perspective of flood calculation, rainfall runoff, reservoir scheduling, river flood evolution, and flood inundation have temporal and spatial continuity, and flood movement is dynamic in time and space.

[0067] Regarding coupling: (1) Model calculation: The coupling of input and output conditions between the four prediction models. Generally, flood forecast results can be used as input conditions for flood warning, flood rehearsal, and flood plan, which can determine whether a warning is issued, whether there is a risk of flooding disaster, and whether flood defense is needed. Flood warning, as an input condition for flood rehearsal, can determine the specific disaster risk level and determine whether and how to activate the flood prevention plan. Based on the results of flood forecast, warning, and rehearsal, a flood defense plan can be comprehensively derived. After adopting the flood defense plan, the flood rehearsal model can be used to visualize and analyze the effect of the plan on flood risk control and disaster reduction, thereby supporting scientific decision-making on the defense plan. (2) Model function: From the perspective of the four predictions for flood prevention and disaster reduction, forecast is the foundation of flood prevention and disaster reduction, warning is the outpost of flood prevention and disaster reduction, rehearsal is the key to flood prevention and disaster reduction, and plan is the purpose of flood prevention and disaster reduction. The organic combination, coupling, and mutual feedback of the four predictions can jointly improve the scientific level of flood prevention decision-making.

[0068] The "four-prediction" models for flood control and disaster reduction have a modular link relationship. The four types of models are linked together, progressive, cyclical, and iteratively coupled. From forecasting to early warning, and then to rehearsals and plans, through data sharing and information transmission, a full-chain, interconnected, and fully coupled "four-prediction" flood control and disaster reduction management system is built to achieve the organic integration of the "four-prediction" models of the reservoir-river-embankment multi-engineering system.

[0069] 1. Forecasting is the foundation of flood prevention and disaster reduction. Using real-time monitoring information on rainfall, water levels, reservoir capacity, and flow, as well as meteorological rainfall forecasts, flood forecasting models are used to predict water levels, flow, and flood inundation impacts for different forecast periods (short-term, medium-term, and long-term). Real-time rolling forecasts of potential flood processes and flood disasters are provided, providing critical risk information for flood prevention and early warning.

[0070] 2. Early warning is the vanguard of flood prevention and disaster reduction. Early warning relies closely on real-time rolling forecast information to identify potential disaster risks and adjust warning levels, raising or lowering them. This allows for timely and accurate release of early warning information, arranging and deploying project inspections, project scheduling, and personnel transfers. This improves the timeliness and accuracy of early warnings, provides guidance for initiating rehearsals, and provides early warning information for emergency plans.

[0071] 3. Rehearsals are key to flood prevention and disaster reduction. By integrating forecast, warning, and corresponding emergency plan information, we can rationally determine reservoir flood control scheduling targets, rehearsal nodes, and boundary conditions. We can simulate reservoir operation and flood evolution that triggers warnings. Simultaneously, we can replay typical historical disaster scenarios in the digital twin watershed, achieving both "forward" and "reverse" rehearsals. Forward rehearsals reveal flood risk situations and impacts, while reverse rehearsals identify reservoir safety operating constraints. This allows for timely identification of flood control safety issues and real-time updates of forecast and warning information to ensure immediacy and accuracy. Furthermore, we can achieve three-dimensional visualization of disaster scenarios, making them more intuitive and providing precise control areas for emergency plan formulation, thereby scientifically formulating and optimizing reservoir scheduling plans.

[0072] 4. Plans are the purpose of flood prevention and disaster reduction. Based on the results of flood disaster rehearsals under different operating conditions, while also considering reservoir safety, population, and socioeconomic distribution, we determine reservoir operation, non-engineering measures, organizational implementation, disaster response measures, resource allocation plans, and post-disaster recovery plans. This creates a flood prevention and disaster reduction plan library that adapts to different scenarios. Rehearsals update scenarios based on feedback from plan measures. If a plan is not reasonable, it is repeatedly iterated to visualize its effectiveness and select the optimal emergency response plan, thereby ensuring its rationality and feasibility.

[0073] The following are more detailed explanations of the related technologies involved in the above process of the present invention:

[0074] 1. About SCS runoff model,

[0075] The SCS runoff model is used for hydrological forecasting in small watersheds. It calculates runoff depth under a given rainfall and is a mathematical model for estimating surface runoff. It uses two basic assumptions: the water balance equation and the proportional equality assumption, and the initial loss (maximum potential retention) relationship assumption to determine the magnitude of surface runoff during rainfall. The basic principle is that during rainfall, if the precipitation does not reach the initial absorption value of the soil, a , no runoff is generated; when the precipitation reaches I a After that, the surface runoff Q is equal to the total rainfall minus the actual infiltration F and the initial absorption value of the soil I a The remainder after .

[0076] (1) According to the assumption of equal proportions, the ratio of the actual infiltration volume F to the maximum possible retention volume S at that time is equal to the actual surface direct runoff volume Q and the maximum possible runoff volume PI a The ratio of is expressed as:

[0077]

[0078] Where: P is the total rainfall, Q is the direct surface runoff, I ais the initial loss (initial absorption value of soil), mm, including interception, surface water storage, etc., F is excluding I a The cumulative infiltration volume (actual infiltration volume) is , and S is the maximum possible retention volume (maximum potential retention volume) at that time.

[0079] (2) According to the water balance equation of the basin, the total rainfall P is equal to the initial loss of the catchment area in the basin (the initial absorption value of the soil) I a , the sum of the actual infiltration volume F and the surface direct runoff volume Q, and its expression is:

[0080] P=I a +F+Q (1.2)

[0081] (3) Based on the relationship between the initial loss and the maximum possible retention at that time, the initial loss I a It is proportional to the maximum possible retention volume F at that time, and its expression is:

[0082] I a =λS (1.3)

[0083] Where: λ is the initial loss coefficient, and the empirical value λ is usually taken as 0.2.

[0084] (4) Combining equations (1.1)-(1.3), we can obtain the calculation formula for the direct surface runoff Q in the SCS model:

[0085]

[0086] (5) To determine the maximum possible retention capacity S of the basin at that time, the model introduces the parameter CN. As can be seen from Equation (1.4), Q is determined by P and S, and S is related to factors such as land use, soil type, and soil moisture before precipitation within the basin. S varies in different basins, making it difficult to determine its value. Therefore, the comprehensive parameter CN that reflects the basin characteristics is introduced into the SCS model to determine the value of S. The relationship is:

[0087]

[0088] Where: CN reflects the runoff capacity of the regional underlying surface unit, which is affected by factors such as the previous soil moisture level, soil type, land use type, slope, vegetation, and other underlying surface factors. Based on the influencing factor information, the CN value ranges from 0 to 100, and the smaller the CN, the greater the infiltration amount.

[0089] 2. About the instantaneous unit line confluence model

[0090] The Nash instantaneous unit line refers to the surface runoff process line formed at the outlet section of a given watershed, at the instant of infinitesimal duration, when the total water input is 1 and the unit surface net rainfall is evenly distributed over the watershed, after being regulated by n series-connected linear reservoirs with a storage constant of K. It can be used to represent the watershed's storage capacity for surface net rainfall and is suitable for calculating surface runoff confluence in small and medium-sized watersheds where data is scarce. Its mathematical expression is:

[0091]

[0092] Where u(0,t) is the ordinate of the instantaneous unit line at time t, where 0 indicates that the net rainfall duration tends to infinitesimal, t represents any time, K is the storage constant of the linear reservoir, which is equivalent to the parameter of the basin confluence time, Γ(n) is the gamma function with respect to n, n is the number of linear reservoirs, which is equivalent to the number of regulation times and reflects the basin storage capacity, and e is the base of the natural logarithm.

[0093] 3. About the Muskingum Flood Evolution Model

[0094] The river flood routing method, which uses upstream flow to predict downstream flow, is computationally simple and requires minimal data. The Muskingum model replaces the complex hydrodynamic equations with water balance equations and channel storage equations. The simplified equations are as follows:

[0095]

[0096] Where: W is the tank storage capacity, t is time, I is the flow rate of the inlet section, Q is the flow rate of the outflow section, K is the tank storage coefficient, and x is the flow weighting factor.

[0097] 4. About the LSTM model for flood forecasting

[0098] The LSTM model is a variant of the recurrent neural network model. Each module of the model consists of a forget gate f t , update gate u t , output gate o t , 3 gating units and 1 cell state unit The LSTM model has strong memory capacity and is suitable for processing long-term series data and solving long-term dependency problems. Each time step has three inputs: the external input x at time t t , LSTM unit output h at time t-1 t-1 , cell state C at time t-1 t -1. The above input data is processed by the LSTM gating unit to obtain the output h t and cell state C t .

[0099] The gate control unit output calculation process and formula are as follows:

[0100]

[0101] Where: f t For the forget gate, u t To update the gate, is the cell state, o t is the output gate, σ is the activation equation, sigmoid function, tanh is the hyperbolic tangent function, h t-1 is the LSTM unit output at time t-1, h t is the LSTM unit output at time t, x t is the external input at time t, C t-1 is the cell state at time t-1, C t is the cell state at time t, W f 、W u 、W c 、W o and b f 、b u 、b c 、b o Represent the weight and bias matrix vectors of the forget gate, update gate, cell state and output gate respectively. The above weight and bias matrices are randomly set to initial values and obtained through data training. The internal structure of the unit is as follows Figure 2 As shown. Figure 2 It can be seen that the LSTM unit output h at the previous moment t-1 and the current state variable x t As the input quantity, it is input to the forget gate, input gate and output gate in parallel to obtain the candidate value. The forget gate is mainly responsible for discarding C t-1 Noise information, retain key information, the input gate selectively stores the critical state C t The information is then updated to update the memory unit C t , output gate and updated memory unit C t Together we can get the current output h t , a better LSTM model can be obtained by reducing the error through multiple iterations.

[0102] 5. Reservoir flood control operation model

[0103] The flood control operation model primarily includes the changes in the maximum water level in front of the reservoir dam and the changes in the downstream flow. The model mainly considers flood control and discharge, and needs to be combined with the specific application of reservoir flood control operation, the relationship between water level and storage capacity, and the operation parameters of the structure.

[0104] (1) Objective function:

[0105] Adopting the maximum peak reduction principle, comprehensively considering the dam safety and the protection of the downstream areas of the reservoir from flood disasters, with the minimization of the peak flow in the downstream flood control section as the main goal, a maximum downstream flow minimization model is established.

[0106]

[0107] Where: is the maximum downstream flow, q t is the outflow of the reservoir during period t, I t is the interval inflow flow of the flood control section downstream of the reservoir in period t, T is the length of the scheduling period, and t is the scheduling period number.

[0108] (2) Constraints:

[0109] 1) The expression of water balance constraint is as follows:

[0110]

[0111] Where: V t-1 and V t are the initial and final water storage in period t, Q t-1 and Q t are the initial and final inflows of the reservoir during period t, q t-1 and q t are the initial and final outflows of the reservoir during period t, and Δt is the length of the calculation period.

[0112] 2) The expression of reservoir water level constraint is as follows:

[0113] Z min ≤Z t ≤Z max (5.3)

[0114] Where: Z t 、Z min 、Z max is the reservoir water level, minimum and maximum allowable water levels during period t.

[0115] 3) The expression of reservoir discharge capacity constraint is as follows:

[0116] Q min ≤Q t ≤Q max(Zt) (5.4)

[0117] Where: Q t is the discharge flow in period t, Q max(Zt) Z t The corresponding maximum downstream flow rate, Q min is the minimum outflow of the reservoir.

[0118] 4) The expression of the relationship between water level and reservoir capacity is as follows:

[0119]

[0120] Where: is the reservoir water level corresponding to the reservoir capacity in period t.

[0121] 5) The expression of the leakage amplitude constraint is as follows:

[0122] |Q out (t)-Q out (t-1)|≤Δq (5.6)

[0123] Where: Δq is the maximum allowable fluctuation of reservoir outflow between adjacent time periods.

[0124] 6) Reservoir capacity boundary condition constraints

[0125] V0=V b ; V T =V e (5.7)

[0126] V min ≤V t ≤V max (5.8)

[0127] Where: V b and V e are the storage capacity corresponding to the water level at the beginning of the reservoir operation period and the storage capacity corresponding to the water level that should be dropped to at the end of the operation period, V max and V min They represent the upper and lower limits of the reservoir water storage capacity respectively.

[0128] 7) The expression for the discharge capacity constraint of the flood discharge facility is as follows:

[0129] 0≤q i,t ≤q i (Z t , B t )≤q i,max (5.9)

[0130] Where: q i，t is the discharge flow of flood discharge facility i, m 3 / s,g i，max is the maximum allowable discharge flow of flood discharge facility i, m 3 / s,q i (Z t , B t ) is the flood discharge facility i when the reservoir water level is Z t , the facility opening is B t Maximum discharge capacity at Z tis the reservoir water level at time t, B t For the operation mode of the discharge structure, the corresponding gate opening is selected according to the corresponding relationship between the facility opening and the discharge capacity.

[0131] 8) Muskingum flow routing constraints

[0132]

[0133] Where: Q k-1 (t) and Q k (t) are the outflow at the beginning and end of the tth period of the kth Muskingum Routing reach, respectively, Q k-1 (t-1) and Q k-1 (t) are the outflow at the beginning and end of the tth period of the k-1th Muskingum Routing reach, is the interval inflow of the kth Muskingum Routed reach, are the routing parameters of the kth Muskingum routing reach respectively.

[0134] 9) Non-negative constraints: all variables do not have negative values during the calculation process.

[0135] 6. Simulation model for flood evolution and inundation risk in the downstream river channel of the reservoir

[0136] The simulation model for flood evolution and inundation risk of the river channel downstream of the reservoir adopts a one-dimensional river channel hydrodynamic model, a two-dimensional planar hydrodynamic model and a one- and two-dimensional coupled model.

[0137] (1) The expression of the one-dimensional hydrodynamic model of the river channel is as follows:

[0138] The basic equations of the one-dimensional model are the Saint-Venant equations, which include the continuity equation (law of conservation of mass) and the momentum equation (Newton's second law), which are as follows:

[0139]

[0140] Where: Q is the flow rate, q is the lateral inflow, A is the water flow area, h is the water level, R is the hydraulic radius, C is the Xie Cai coefficient, and α is the momentum correction coefficient.

[0141] (2) Principle of two-dimensional plane free surface flow model:

[0142] The basic equations for two-dimensional unsteady flow calculations include the continuity equation and the momentum equation, as follows:

[0143] Continuity equation:

[0144]

[0145] Momentum equation:

[0146]

[0147] Where: is the velocity averaged based on water depth, t is time, x, y and z are Cartesian coordinates, η is the riverbed elevation, d is the still water depth, h = η + d is the total head, u and v are the velocity components in the x and y directions, g is the acceleration due to gravity, ρ is the density of water, and s xx 、s xy 、s yx 、s yy is the component of radiation stress, p a is the atmospheric pressure, ρ0 is the relative density of water, S is the point source flow rate, u s 、v s is the flow velocity of the source-sink water flow.

[0148] 7. Planning and design principles for monitoring station networks to meet flood control and disaster reduction forecasting and early warning needs

[0149] Regarding the planning of a monitoring station network for the reservoir-river-levee system, the needs of ground-based and air-based monitoring should be comprehensively considered. A multi-mode, multi-layered, integrated monitoring station network system consisting of meteorological satellites, rainfall radars, rain gauges, hydrological stations, and water level stations should be constructed. The present invention proposes the following principles:

[0150] (1) Hydrological station network: Hydrological stations are set up on the main tributaries of the reservoir; in areas prone to mountain torrents and rivers with flood control targets such as small and medium-sized towns, villages, industrial and mining enterprises, and important infrastructure, hydrological stations are set up at the mountain outlets or upstream of the flood control targets.

[0151] (2) Water level station network: Representative water level stations are set up in front of the reservoir dam, near the mouth of the tributary into the reservoir (lake), and at the location where the water surface narrows or widens along the way, and at the end of the reservoir and the middle section of the reservoir area affected by the fluctuating backwater; water level stations are set up in rivers where there is a threat of flood disasters to urban residential areas, industrial and mining enterprises, and important infrastructure.

[0152] (3) Rainfall measuring station network: evenly distributed within the planning area, with the average single station area not exceeding 100 km 2 ; Rain gauges are set up near the center of the basin; precipitation stations are set up in areas where rainstorms and floods are concentrated and in areas that play an important role in flood control forecasting and early warning; rain gauges are set up more densely within the reservoir control basin, and at least one rain gauge is set up in small reservoirs.

[0153] (4) Rainfall radar stations: Rainfall radar stations are deployed in areas where heavy rain and floods are concentrated and areas prone to flash floods, forming a network.

[0154] (5) Video surveillance stations: Video surveillance stations should be set up in reservoirs. If communication conditions are available, at least two video surveillance stations should be set up. Video surveillance stations should be set up along the river downstream of the reservoir according to the degree of flooding risk.

[0155] In summary, the present invention adopts theoretical analysis, numerical simulation and other technical means, and through the intersection of multiple disciplines such as hydrology, hydraulics, disaster science, and artificial intelligence, respectively constructs the "forecast", "early warning", "rehearsal", and "plan" models of the reservoir-river-embankment multi-engineering system, and clarifies the dynamic correlation mechanism between the four prediction models, to achieve high-precision flood forecasting and warning, dynamic rehearsal of flood risks and scientific compilation of flood control plans.

[0156] Those skilled in the art can make improvements and changes based on the above invention. Any modifications, improvements and changes made on the basis of the present invention should fall within the scope of protection of the present invention.

Claims

1. A multi-project flood prevention and disaster reduction method based on a four-precoupling model, characterized in that: Specifically include: A flood control and disaster reduction forecasting model for a reservoir-river-levee multi-engineering system is constructed. The forecasting model is at least a model system formed by multi-dimensional deep coupling including a meteorological rainfall forecasting model, a reservoir upstream area flood forecasting model, a reservoir scheduling and downstream river levee risk forecasting model and a flood inundation disaster forecasting model; wherein the meteorological rainfall forecasting model takes gridded actual rainfall data as input and gridded meteorological rainfall forecast data as output; the reservoir upstream area flood forecasting model takes gridded meteorological rainfall forecast data output by the meteorological rainfall forecasting model as input; the reservoir scheduling and downstream river levee risk forecasting model takes the output of the reservoir upstream area flood forecasting model as input. The model includes a reservoir engineering flood control scheduling sub-model and a reservoir downstream river channel-levee overflow and breach one- and two-dimensional hydrodynamic coupling sub-model. The reservoir engineering flood control scheduling sub-model combines multi-source data including terrain elevation, river system, and water conservancy project facility scheduling rules to obtain output 1 as the reservoir scheduling discharge flow. The process is used as the input condition of the one-dimensional hydrodynamic coupling sub-model of the river channel and levee overflow and breach downstream of the reservoir. The one-dimensional hydrodynamic coupling sub-model of the river channel and levee overflow and breach downstream of the reservoir realizes the simulation of the flood evolution and the flooding risk of the river channel downstream of the reservoir, and obtains output 2, which is a dynamic simulation of reservoir operation and downstream river channel flood evolution, flooding risk of both sides of the levee, calculates the flooding range caused by typical frequency floods, and predicts and evaluates the impact of different magnitude floods on reservoir flood control safety, levee safety and possible flooding risk of coastal areas; output 1 and output 2 are used together as the output of the reservoir operation and downstream river channel levee risk prediction model; the flood disaster prediction model takes the output of the flood prediction model of the upstream area of the reservoir and output 1 and output 2 of the reservoir operation and downstream river channel levee risk prediction model as input, and takes the flood disaster prediction result of the extreme disaster events of reservoir overburden and river channel levee overburden on the downstream area as the output; Construct a flood prevention and disaster reduction early warning model for a reservoir-river-levee multi-engineering system. The early warning model is at least a model system formed by multi-dimensional deep coupling including a meteorological risk early warning model, a reservoir risk early warning model, a dam-down river levee risk early warning model, and a reservoir downstream resident flooding risk early warning model; wherein, the meteorological risk early warning model takes observation of regional rainfall and the formation of 24h hourly rainfall forecast data as input, and takes meteorological risk level early warning results as output; the reservoir risk early warning model further includes a rain gauge monitoring early warning sub-model, an inflow flood early warning sub-model, a reservoir area flooding risk early warning sub-model and a dam engineering safety early warning sub-model, the The rain gauge monitoring and early warning sub-model takes the measured rainfall data as input and the judgment result of whether the rainfall triggers the rainfall early warning response as output; the reservoir area flooding risk early warning sub-model takes the reservoir area water level all-weather monitoring data as input and the flooding risk forecast result as output; the dam engineering safety early warning sub-model includes dam overflow risk early warning processing and dam body instability early warning processing. Specifically, the dam overflow risk early warning processing takes the inflow flood and the measured reservoir water level monitoring data, combined with the water level storage capacity relationship and the reservoir discharge capacity as input 1, to predict the reservoir water level development trend as the output 1 of the sub-model; the dam body instability early warning processing takes the dam body water content, crack development process and other factors as input. The dam body hazard data of depth, infiltration line height, soil deformation and slope balance safety factor are taken as input 2, and the structural safety status of the dam is monitored in real time as the output 2 of the model; the risk warning model for the river channel embankment below the dam further includes a rain gauge monitoring and warning sub-model, a river channel flood risk simulation and warning sub-model and a embankment engineering safety warning sub-model. Specifically, the rain gauge monitoring and warning sub-model takes rainfall monitoring data as input and triggers a rainfall warning response of a corresponding level as output; the river channel flood risk simulation and warning sub-model takes the model calculation boundary conditions output by the reservoir engineering flood control scheduling sub-model as input, and the downstream river channel flood risk warning response result is the output of the model. The output of the model; the embankment engineering safety early warning sub-model takes as input basic data including at least the embankment soil moisture content, crack development degree, infiltration line height, soil deformation, and slope balance safety factor, and takes the embankment disaster accident early warning response results as the output of the model; the reservoir downstream residents flooding risk early warning model further includes a monitoring early warning sub-model and a simulation early warning sub-model. The monitoring early warning sub-model takes as input the dynamic monitoring images of the possible flooded area, and provides data support and decision-making basis for disaster assessment and rescue work as output; the simulation early warning sub-model is based on the simulation of flood evolution in the river downstream of the reservoir and flooding risk on both sides of the river; Construct a flood control and disaster reduction simulation model for a multi-engineering system of reservoirs, river channels, and levees. This model takes terrain elevation, river systems, and remote sensing imagery data of the downstream reservoir area as input, and provides a dynamic visualization of the entire process of rainfall runoff, reservoir operation, flood evolution, and inundation as output. Constructing a reservoir-river-levee multi-engineering system flood control and disaster reduction plan model by combining the aforementioned reservoir-river-levee multi-engineering system flood control and disaster reduction forecast model, the aforementioned reservoir-river-levee multi-engineering system flood control and disaster reduction early warning model, and the aforementioned reservoir-river-levee multi-engineering system flood control and disaster reduction rehearsal model, using the outputs of the aforementioned forecast model, early warning model, and rehearsal model as inputs, and using the reservoir-levee-flood zone multi-engineering system risk control plan under standard flood conditions as output; Plan and design a monitoring network, including at least rain gauges, hydrological stations, water level stations and video stations, and conduct four-pronged dynamic coupling including forecast, warning, rehearsal and emergency plan.

2. A multi-project flood prevention and disaster reduction method based on a four-precoupling model according to claim 1, wherein: The reservoir upstream area flood forecasting model further includes a physical mechanism flood forecasting sub-model and a data driven flood forecasting sub-model.

3. A multi-project flood prevention and disaster reduction method based on a four-precoupling model according to claim 2, wherein the physical mechanism flood forecast submodel is a combined model consisting of a series-coupled SCS flow generation submodel, an instantaneous unit line flow confluence submodel, and a Muskingum flood evolution submodel.

4. A multi-project flood prevention and disaster reduction method based on a four-precoupling model according to claim 2, wherein the data-driven flood forecasting sub-model adopts an LSTM model, wherein: The input of the flood forecast LSTM model is the rainfall and runoff processes in the previous n hours and the rainfall processes in the next m hours, and the output is the runoff processes in the next m hours. The forget gate is used to discard information that is not important to flood forecasting and control the degree of information retention. The update gate is used to update the neural network model with new important information to generate candidate values. The output gate is used to determine the important information that needs to be output and generate hidden states for prediction or as input for the next time step.

Images