Multi-engineering system flood control and disaster reduction method based on four-pre-coupling model

Abstract

The invention discloses a multi-project flood control and disaster reduction method based on a four-pre-coupling model, and the method comprises the steps: constructing a reservoir-river channel-dike multi-project system flood control and disaster reduction forecasting model, constructing a reservoir-river channel-dike multi-project system flood control and disaster reduction early warning model, and constructing a reservoir-river channel-dike multi-project system flood control and disaster reduction rehearsal model. A flood control and disaster reduction plan model of a reservoir-river channel-dike multi-project system is constructed, a monitoring network is planned and designed according to a certain rule, and four-pre dynamic coupling including forecasting, early warning, rehearsal and pre-warning is carried out. Compared with the prior art, the method comprehensively considers the synergistic effect and influence restriction relation of flood control processes of a reservoir, a river channel and a dike, and realizes chain type dynamic connection of four pre-links of flood control and disaster reduction through a multi-engineering system coupling flood control four-pre dynamic coupling model; and flood forecasting precision, flood control and disaster reduction decision scientization and flood risk management refinement are realized.

Description

Technical Field

[0001] The present invention relates to the field of emergency disaster prevention and intelligent flood control and disaster reduction solutions, and in particular to a reservoir-river-dike 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 takes "reservoir-river-levee" as an integrated unit to realize the coupling functions of flood "forecast-warning-rehearsal-plan". It is one of the important non-engineering measures to comprehensively predict, evaluate and manage flood disaster risks and reduce flood disaster losses.

[0003] Reservoirs, rivers and levees are important infrastructure for flood control and disaster reduction. Their flood control capacity plays an important role in protecting the safety of people downstream and reducing the affected area. At present, a diversified and technology-intensive model system has been built in the field of flood control and disaster reduction. Its core mainly relies on hydrological models, hydrodynamic models, remote sensing and geographic information system technologies, numerical simulation technologies, and the introduction of artificial intelligence and data mining technologies, making flood control and disaster reduction models relatively rich in technical research and application. However, in the current research and practice of flood control and disaster reduction models, the three "reservoirs-rivers-levees" have not yet been fully studied as an integrated unit. Instead, the three are studied separately or in combination. This easily lacks systematic thinking and planning from an overall perspective, and it is difficult to fully consider the natural synergistic series effect and mutual influence and constraint relationship between the three, thereby reducing the system resilience and overall effect of flood control and disaster reduction, and failing to achieve scientific scheduling and management of flood risks. More importantly, current research lacks an inherent dynamic coupling mechanism that considers the "four-prediction" functions. This makes it difficult to achieve a forward-looking rehearsal of all elements of the physical basin and the entire process of water conservancy management activities in the face of complex and changeable flood situations, ensuring early discovery of risks, early release of warnings, early formulation of plans, and early implementation of measures. At the same time, the coupling mechanism between the existing "four-prediction" models is still imperfect, lacking in deep integration and collaborative optimization, resulting in obvious deficiencies in information sharing and functional complementarity between models, making it difficult to achieve chain-like dynamic connection of flood control and disaster reduction links, which to a certain extent affects the full play of the overall flood control and disaster reduction effectiveness, and has not yet formed an efficient and unified flood control and disaster reduction system. Therefore, taking the reservoir-river-levee multi-engineering system as the overall unit, establishing and improving the four-prediction coupling model of flood control and disaster reduction in complex engineering systems is of great significance to improving the scientificity and accuracy of basin flood control risk scheduling.

[0004] In summary, although the field of flood control and disaster reduction already has diversified water conservancy professional models with relatively rich functions, the separate research of "reservoirs", "rivers" or "levees", and the lack of coupling mechanisms between the "four predictions" (forecast, warning, rehearsal, and plan) functional models, etc., are the 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", the forward-looking idea is that forecast is the basis, warning is the outpost, rehearsal is the key, and plan is the purpose, to achieve the four prediction functions of forecast, warning, rehearsal, and plan for flood control and disaster reduction, and at the same time strengthen the dynamic coupling between models, clarify the interaction and influence mechanism between each model, and thus propose a reservoir-river-levee multi-engineering system flood control and disaster reduction four-preparation coupling model method. Summary of the invention

[0005] In view of 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, focusing on the major needs and development trends of reservoir flood control emergency management, the present invention aims to propose a multi-engineering system flood control and disaster reduction method based on the four-pre-coupling model, taking "reservoir-river-levee" as an integrated research unit, and realizing the "forecast-warning-rehearsal-plan" model system for flood control and disaster reduction for the reservoir-river-levee multi-engineering system through the coupling integration and coordinated optimization of the four-pre-coupling models.

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

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

[0008] Construct a flood control and disaster reduction forecasting model for a reservoir-river-levee multi-engineering system, wherein 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 dispatching 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 dispatching and downstream river levee risk forecasting model takes the output of the reservoir upstream area flood forecasting model as input, and the model includes a reservoir engineering flood control dispatching submodel and a reservoir downstream river channel-levee overflow and breach one- and two-dimensional hydrodynamic coupling submodel, wherein the reservoir engineering flood control dispatching submodel combines multi-source data including terrain elevation, river system, and water conservancy engineering facility dispatching rules to obtain an output 1 as the reservoir dispatching discharge flow The process is used as the input condition of the one-dimensional hydrodynamic coupling submodel of the downstream river channel and dike overtopping of the reservoir. The one-dimensional hydrodynamic coupling submodel of the downstream river channel and dike overtopping of the reservoir realizes the simulation of the flood evolution and flooding risk of the downstream river channel of the reservoir, and the output 2 is obtained, which is a dynamic simulation of the reservoir operation and the downstream river channel flood evolution, the flood inundation risk of the dike on both sides, calculates the flood inundation range caused by typical frequency floods, and predicts and evaluates the impact of different magnitude floods on the flood control safety of the reservoir, the dike safety and the flooding risk that may be caused to the coastal area; the output 1 and the output 2 are used together as the output of the reservoir operation and downstream river channel dike risk prediction model; the flood inundation disaster prediction model takes the output of the flood prediction model of the upstream area of ​​the reservoir, the output 1 and the output 2 of the reservoir operation and downstream river channel dike risk prediction model as input, and takes the flood inundation disaster prediction result of the extreme catastrophic events of reservoir overtopping and river channel dike overtopping on the flood disaster risk of 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 river channel and levee risk early warning model below the dam, and a reservoir downstream residential flooding risk early warning model; wherein the meteorological risk early warning model takes the observation of regional rainfall and the formation of 24h hourly forecast rainfall data as input, and takes the early warning result of the meteorological risk level as output; the reservoir risk early warning model further includes a rainfall station monitoring and 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 station monitoring and 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 danger 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 risk warning model for the river embankment below the dam further includes a rain gauge monitoring and warning sub-model, a river 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 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 the model; the levee engineering safety early warning sub-model takes as input the basic data including at least the levee soil moisture content, crack development degree, infiltration line height, soil deformation, and slope balance safety factor, and takes the levee disaster accident early warning response result 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 the dynamic monitoring image of the possible flooded area as the input of the sub-model, and provides data support and decision-making basis for disaster assessment and rescue work as the output; the simulation early warning sub-model is based on the evolution of river floods downstream of the reservoir and the simulation of flooding risks on both sides of the river;

[0010] Construct a flood prevention and disaster reduction rehearsal model for a multi-engineering system of reservoirs, rivers and levees, with terrain elevation, river systems and remote sensing image data in the downstream area of ​​the reservoir as input, and dynamic visualization of the entire process of rainfall runoff, reservoir scheduling, 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 reservoir-river-levee multi-engineering system flood control and disaster reduction forecast model, the reservoir-river-levee multi-engineering system flood control and disaster reduction early warning model and the reservoir-river-levee multi-engineering system flood control and disaster reduction rehearsal model, taking the outputs of the forecast model, early warning model and rehearsal model as input, and taking the reservoir-levee-flooding area multi-engineering system risk control plan under standard flood conditions as output;

[0012] Plan and design a monitoring network, including at least rainfall stations, hydrological stations, water level stations and video stations, and carry out 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 forecasting sub-model adopts an LSTM model, wherein the input of the flood forecasting 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 related to flood forecasting that is not important to 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 a hidden state is generated for prediction or as input for the next time step.

[0016] Compared with the prior art, the positive technical effects achieved by the present invention are as follows:

[0017] Starting from the perspective of the whole and the whole, it is possible to comprehensively consider the synergistic effect and influencing and restricting relationship of the three elements of "reservoir, river channel and embankment" in the flood control process. By constructing a dynamic coupling model of flood control "four pre-emptions" with a reservoir as the core and considering the upstream and downstream and left and right banks of a multi-engineering system, the model can realize the chain dynamic connection of the "four pre-emptions" links of flood control and disaster reduction. This will help to take the river basin as the overall perspective, realize the precision of flood forecasting, the scientific decision-making of flood control and disaster reduction, and the refinement of flood risk management, timely intercept and discharge flood water, 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 It is an overall flow chart of the flood prevention and disaster reduction method of 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 below;

[0020] Figure 3 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 process line for a mountain 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 flooded areas);

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

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

[0027] like Figure 1 As shown, the overall process of a multi-engineering flood prevention and disaster reduction method based on a 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 model system that is deeply coupled in multiple dimensions, including a meteorological rainfall forecast model, a flood forecast model for the upstream area of ​​the reservoir, a reservoir dispatch and downstream river embankment risk forecast model, and a flood inundation disaster forecast model. The forecast model meets the real-time rolling forecast capability, sets the water level, flow, and water volume 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 forecast methods. The specific description is as follows:

[0030] 1.1. Meteorological rainfall forecast model: By deploying phased array hydro-rainfall radars in areas prone to rainstorms and flash floods, and deploying meteorological weather radars around cities, rainfall in different spatial scales in different regions can be observed and forecasted, realizing the shift of rainfall monitoring from "falling rain" to "rain in the cloud". Among them, the hydro-rainfall radar can observe liquid water in the near-surface atmosphere from above the ground to 2km in height, and generate high-precision 30m×30m grids and continuous high spatiotemporal resolution gridded rainfall data with a time resolution of 40s in real time as the input of the model, and the gridded meteorological rainfall forecast data with high spatiotemporal resolution of 1-3h extrapolated based on the application of the network of three rainfall radars as the output of the model; the meteorological weather radar can observe all meteorological elements in the atmosphere from above the ground to the top of the troposphere, within the range of 20-30km in height, and conduct large-scale sampling observation of the atmosphere, solve the problem of vertical detection of the atmosphere, and meet the needs of flood control forecasting for rainfall forecasting.

[0031] 1.2. Flood forecasting model for the upstream area of ​​the reservoir. This model 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 forecasting model for the upstream area of ​​the reservoir driven by both physics and data. The physical mechanism flood forecasting submodel mainly includes the SCS runoff generation submodel, the instantaneous unit line confluence submodel and the Muskingum flood evolution submodel. The three models of runoff generation, confluence and flood evolution are coupled in series, and the output of the previous model is used as the input of the next model. The SCS runoff generation submodel is a model of the runoff process generated by rainfall through the loss stage. The instantaneous unit line confluence submodel realizes the simulation of the entire runoff process. The Muskingum flood evolution submodel takes into account the water flow velocity, flow rate and water level to simulate the flood process. The data-driven flood forecast sub-model mainly includes artificial neural network models such as LSTM model. The input of 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 for flood forecasting 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 empirical rainfall type database, and the key indicators of the flood process line (flood peak value, flood peak arrival time, total flood volume, 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 upstream of the reservoir.

[0032] 1.3. Reservoir dispatching and downstream river embankment risk prediction model, with the output result of the reservoir upstream area flood prediction model described in 1.2 as the input of the model, which includes a reservoir engineering flood control dispatching submodel and a reservoir downstream river channel-embankment overtopping and dyke breaching one-dimensional hydrodynamic coupling submodel. The reservoir engineering flood control dispatching submodel combines multi-source data such as high-precision terrain elevation, river system, and water conservancy project facility dispatching rules to obtain output 1, which is the reservoir dispatching discharge process and uses it as the input condition of the reservoir downstream river channel-embankment overtopping and dyke breaching one-dimensional hydrodynamic coupling submodel. The reservoir downstream river channel-embankment overtopping and dyke breaching one-dimensional hydrodynamic coupling submodel realizes the downstream river channel flood evolution and the flooding risk simulation of both banks, and obtains output 2, which is a dynamic simulation of reservoir dispatching and downstream river channel flood evolution, flooding risk on both banks of the embankment, calculates the flooding range caused by typical frequency floods, and predicts and evaluates the impact of different magnitudes of floods on reservoir flood control safety, embankment safety and possible flooding risks of coastal areas. Output 1 and output 2 are used together as the output of the reservoir regulation and downstream river embankment risk prediction model.

[0033] 1.4. Flood inundation disaster forecasting model, with the output 1 and output 2 of the reservoir upstream area flood forecasting model mentioned in 1.2 and the reservoir dispatch and downstream river embankment risk forecasting model mentioned in 1.3 as the input of the model, and the flood inundation disaster forecasting results of the downstream area inundation disaster risk caused by extreme catastrophic events such as reservoir dam breach and river embankment breach as the output of the model, combined with socio-economic data (basic statistical indicators such as population, cultivated land, and gross domestic product), and further including the evaluation of the output of the model by constructing a flood disaster impact analysis and loss assessment model based on GIS, so as to comprehensively forecast and evaluate the impact and loss of flood disasters.

[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-down river embankment risk warning model, and a reservoir downstream resident 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 station monitoring and early warning sub-model, an inflow flood warning sub-model, and a reservoir area flooding risk warning sub-model; the dam-down river embankment risk warning model further includes a rain gauge station monitoring and early warning sub-model, a river flood risk 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 outputs real-time dynamic early warnings of the time and scale of flood disasters. Early warning information is issued for specific areas and specific groups of people to support emergency evacuation and rescue and resettlement of people in disaster areas.

[0036] 2.1. Meteorological risk warning model, with regional rainfall observed and 24h hourly forecast rainfall data as input, by deploying phased array water conservancy rainfall radar network and meteorological weather radar in the upstream and downstream areas of the reservoir, regional rainfall is observed and 24h hourly forecast rainfall data is formed. The grid rainfall (which changes with the soil moisture state) in each warning period (3h, 6h, 12h and 24h) is used as the meteorological risk warning index of flood disasters, and the warning index thresholds corresponding to the four meteorological risk levels of low (possible occurrence, blue warning), medium (high possibility, yellow warning), high (high possibility, orange warning), and extremely high (high possibility, red warning) are calculated respectively, and the warning results of the meteorological risk level are used as output.

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

[0038] (1) Rain gauge monitoring and early warning sub-model: Relying on the dense network of rain gauge stations upstream and downstream of the reservoir, real-time monitoring of rainfall data in each early warning period, observation and recording of various rainfall indicators such as hourly maximum rainfall, 24-hour rainfall, regional average rainfall, and cumulative rainfall. The measured rainfall data is used as the input of this sub-model, and the analysis and calculation method of inverse calculation of critical rainfall from disaster water level and design rainstorm flood is adopted. The rainfall in each early warning period is selected as the real-time dynamic early warning indicator, and the judgment result of whether the rainfall triggers the rainfall early warning response is used as the output of this sub-model.

[0039] (2) 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 flood process calculation and real-time warning of the flood entering the reservoir are realized. In addition, the reservoir tail flow monitoring system is used to obtain key indicators such as the flood flow entering the reservoir 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 reservoir area water level all-weather monitoring data is used as the input of this sub-model, and the threshold of the water level warning indicator corresponding to the flooding risk is set to determine 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 outputs of the flood disaster prediction model in 1.4 and the reservoir engineering 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, and the flooding risk prediction result is used as the output of the model to comprehensively evaluate the reservoir's flooding trend and potential flooding risk, determine 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 to predict the development trend of the reservoir water level 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 dam body danger data such as dam body moisture content, crack development degree, infiltration line height, soil deformation, slope balance safety factor, etc. are used as the input of the model2. High-precision sensors are used to monitor the dam body danger data such as dam body moisture content, crack development degree, infiltration line height, soil deformation, slope balance safety factor, etc. The structural safety status of the dam (embankment) is monitored in real time through video remote monitoring and regular inspections by staff, etc., and the output of the model2 is to compare the preset safety threshold and dam design standards, evaluate the stability and risk level of the dam, and trigger the early warning response of dam instability and damage disaster accidents.

[0042] 2.3. Risk early warning model for river embankment below the dam, including the following sub-models:

[0043] (1) Rain gauge monitoring and early warning sub-model: The sub-model uses rainfall monitoring data as input, relies on the network of rainfall gauge stations below the dam, monitors rainfall data in real time, uses the analysis method of inverting critical rainfall from disaster-causing water levels and designed rainstorms and floods, selects rainfall during each warning period for protected objects on both sides of the river as real-time dynamic early warning indicators, and uses the rainfall early warning response of the corresponding level as the output of the sub-model to determine the excessive rainfall and trigger the rainfall early warning response of the corresponding level. The early warning indicators are divided into two levels, corresponding to the preparation and immediate transfer of disaster prevention objects.

[0044] (2) River flood risk simulation and early warning submodel: Taking the model calculation boundary conditions output by the reservoir engineering flood control and dispatching submodel in 1.3 as input, based on the output of the reservoir engineering flood control and dispatching submodel, combined with information such as river topography, levee elevation, and remote sensing images, a coupled model for simulating flood evolution and flood risk on both banks of the reservoir downstream is established. The flood evolution process in the downstream area of ​​the reservoir is dynamically simulated, and the flood or flood inundation range of typical frequency is calculated and predicted. At the same time, key parameters such as river water level and flow are monitored in real time. The possibility of river flood exceeding the levee warning water level and guaranteed water level, and being higher than the surface elevation of the dangerous areas or protection areas on both banks is comprehensively assessed, triggering a flood risk early warning response in the downstream river. The flood risk early warning response result in the downstream river 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. It 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. It uses video remote monitoring and regular staff inspections to monitor the structural safety status of the levee in real time, compares the preset safety threshold and levee flood control design standards, evaluates the stability and risk level of the levee, triggers the levee disaster accident early warning response, and uses the levee disaster accident early warning response results 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 surveillance, satellite remote sensing image recognition, drone aerial photography and other "air-ground-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, clearly identify the target objects in the flooded area, and evaluate the water depth and water flow velocity in the flooded area, thereby triggering flood risk monitoring and early warning responses 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 banks, the model focuses on extreme disaster events such as reservoir dam breaches and river embankment breaches. Combined with socio-economic data (statistical indicators such as population, cultivated land, and GDP), the model simulates and predicts the inundation range, inundation depth, and disaster losses of downstream residential areas under different disaster event scenarios based on disaster statistics and loss assessment methods, and triggers an early warning response for inundation disasters in the downstream areas of the reservoir according to the risk level.

[0049] The monitoring network is planned and designed according to the needs of flood control and disaster reduction forecasting and early warning, and in accordance with 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, the rain gauge monitors regional rainfall and is generally arranged near the centroid of the basin; the hydrological station and water level station monitor the water surface elevation of reservoirs and rivers. The hydrological station can also monitor the flow process of the river and is generally arranged at the reservoir bank or embankment near the water; the video station monitors whether the water level reaches the water surface elevation for early warning, whether the monitoring danger zone is flooded, etc., and is arranged in a location 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 predictions" including forecasting, early warning, rehearsal and emergency plans.

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

[0051] The model takes the terrain elevation, river system, remote sensing images and other data information of the downstream area of ​​the reservoir as input, considers the constraints of the levee and the multiple effects of the interval inflow, determines the model boundary conditions, river channel and flood area roughness, etc., and dynamically links the flood forecast model of the upstream area of ​​the reservoir in 1.2, the reservoir engineering flood control dispatch submodel in 1.3 and the one-dimensional hydrodynamic coupling submodel of the downstream river channel-levee overburden in the reservoir to simulate the flood inundation process and its distribution characteristics under the super-standard flood and overburden scenario. Using GIS, three-dimensional modeling and rendering, hydrological and hydrodynamic models and other technical means, a two-dimensional plane and three-dimensional stereoscopic integrated digital twin scene is constructed to support the dynamic visualization display of the whole process of rainfall runoff, reservoir dispatch, flood evolution and inundation. The dynamic visualization display of the whole process of rainfall runoff, reservoir dispatch, flood evolution and inundation is the output of this module.

[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. Three-dimensional modeling and rendering technology mainly uses the three-dimensional module in three-dimensional modeling software or GIS software to construct three-dimensional scenes according to geographic spatial data, including river flow direction, road layout, terrain undulations, etc., and then set reasonable lighting and shadow effects for the three-dimensional model and add materials and textures to enhance the realism and three-dimensional sense of the scene and make 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 terrain 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, flow rate, etc. 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 in step one, the reservoir-river-levee multi-engineering system flood control and disaster reduction early warning model in step two, and the reservoir-river-levee multi-engineering system flood control and disaster reduction rehearsal model in step three. The output of the 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 the risk of downstream flooding, the flood scheduling defense plans under different working conditions are optimized, so as to formulate the risk control plan of the reservoir-levee-flood-region multi-engineering system under super-standard flood conditions, and the risk control plan of the reservoir-levee-flood-region multi-engineering system 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 monitoring and early warning system (including satellite remote sensing, radar rainfall measurement, hydrological automatic 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, and establish a regional and graded early warning release plan and recommended prevention and control measures. Among them:

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

[0060] (2) The 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 displacements and infiltration lines of the project, and focus on observing whether the dam embankment has cracks, landslides, collapses, leakages, pipe bursts and other damages. Report 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 flood 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 Publicize knowledge on flood disaster prevention. 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 levee engineering accident rescue, personnel evacuation and emergency rescue, etc., involving multi-departmental coordinated disaster prevention, mitigation and relief. Among them:

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

[0063] (2) Engineering accident rescue. Emergency repair of reservoirs, levees and other engineering facilities damaged during floods, reinforcement of emergency levees, and pre-planning of emergency response strategies such as rescue team organization, personnel allocation, task allocation, material preparation, typical emergency response plans, and information communication and coordination methods.

[0064] (3) Flood evacuation and emergency rescue. A detailed relocation and resettlement plan is formulated for low-lying areas and flood-inundated areas in the rehearsal model. The main contents include surveying the people in the danger zone, evacuation time, evacuation routes and resettlement points, evacuation transportation, medical assistance points and health rescue teams, life-saving equipment, daily necessities, etc.

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

[0066] About dynamics: Flood prevention and disaster reduction takes "flood" as the main line, mainly reflecting the spatiotemporal dynamic characteristics of water flow in the four forecast periods. Forecast-warning-rehearsal-plan all cover the upstream and downstream and 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 continuity in time and space, and flood movement is dynamically changing in time and space.

[0067] Regarding coupling: (1) Model calculation: The coupling of input and output conditions between the four prediction models. Usually, flood forecast results can be used as input conditions for flood warning, flood rehearsal, and flood plan, and determine whether a warning is issued, whether there is a risk of flooding disaster, and whether flood defense is needed; flood warning as a flood rehearsal input condition can determine the specific disaster risk level and determine whether to start and how to start 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 regulation and disaster reduction, thereby supporting the scientific decision-making of the defense plan. (2) Model function: From the perspective of the four predictions of flood prevention and disaster reduction, forecast is the basis 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 feedback of the four predictions will 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 rehearsal 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 basis of flood prevention and disaster reduction. Using real-time monitoring information such as rainfall, water level, reservoir capacity, flow, and meteorological rainfall forecasts, the flood forecast model is used to predict water levels, flow, and flood inundation impacts in different forecast periods (short-term, medium-term, and long-term), and to make real-time rolling forecasts of possible flood processes and flood disasters, providing important risk information for flood prevention and early warning.

[0070] 2. Early warning is the outpost of flood prevention and disaster reduction. Early warning is closely dependent on real-time rolling forecast information, identifying potential disaster risks, and realizing the increase or decrease or elimination of early warning levels, so as to timely and accurately release early warning information, arrange and deploy engineering inspections, engineering scheduling, personnel transfer and other work, improve the timeliness and accuracy of early warning, provide guidance for launching rehearsal work, and provide early warning information for emergency plans.

[0071] 3. Rehearsal is the key to flood prevention and disaster reduction. Integrate forecast, warning and corresponding plan information, reasonably determine reservoir flood control dispatching targets, rehearsal nodes, boundary conditions, etc., simulate the reservoir operation and flood evolution that triggers the warning, and replay typical historical disaster scenarios in the digital twin basin to achieve "forward" and "reverse" two-dimensional rehearsals. Forward rehearsals obtain flood risk situations and impacts, and reverse rehearsals obtain reservoir safety operation restrictions, timely discover flood control safety issues, and update forecast and warning information in real time to ensure the timeliness and accuracy of information. At the same time, realize three-dimensional visualization of disaster scenarios, improve the intuitiveness of disaster scenarios, and provide precise prevention and control areas for plan formulation, so as to scientifically formulate and optimize reservoir dispatching plans.

[0072] 4. Plans are the purpose of flood prevention and disaster reduction. Based on the results of flood disaster rehearsals under different working conditions, and taking into account the safety of reservoirs, population and socio-economic distribution, etc., determine the reservoir scheduling and use, non-engineering measures, organizational implementation methods, disaster response measures, resource allocation plans and post-disaster recovery plans, etc., to form a flood prevention and disaster reduction plan library that adapts to different scenarios. The scenarios are updated through feedback from plan measures and rehearsals. If the plan is unreasonable, it is repeatedly iterated to achieve a visual display of the plan effect and select the optimal emergency defense plan, thereby ensuring the rationality and feasibility of the plan.

[0073] The more detailed related technical explanations involved in the above process of the present invention are as follows:

[0074] 1. About SCS flow model,

[0075] The SCS runoff model is used for hydrological forecasting of small watersheds, that is, to calculate the runoff depth under a given rainfall. It is a mathematical model for estimating surface runoff. It can derive the amount of surface runoff during rainfall through two basic assumptions: the water balance equation and the assumption of equal proportions, and the assumption of the relationship between initial loss (maximum potential retention). The basic principle is: 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 possible maximum retention volume S at that time is equal to the actual surface direct runoff volume Q and the possible maximum 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 assumption that the initial loss is related to 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 surface direct runoff Q in the SCS model:

[0085]

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

[0087]

[0088] Where: CN reflects the runoff capacity of the regional underlying surface unit, and its influencing factors include the previous soil moisture level, soil type, land use type, slope, vegetation and other underlying surface factors. According to the influencing factor information, the CN value range is 0-100. The smaller the CN, the more infiltration.

[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 the basin within a given basin, at the moment of infinitesimal duration, when the total water input is 1 and the unit surface net rainfall is evenly distributed in the basin, through the regulation and storage of n series-connected linear reservoirs with a storage constant of K. It can be used to represent the basin's regulation and storage capacity for surface net rainfall, and is suitable for the calculation of surface runoff in small and medium-sized basins with a lack of data. 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 be 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 about n, n is the number of linear reservoirs, which is equivalent to the number of adjustments, reflecting the basin regulation and storage capacity, and e is the base of the natural logarithm.

[0093] 3. About Muskingum flood evolution model

[0094] The river flood routing method that predicts the downstream flow of the river by the upstream flow of the river is simple to calculate and has low data requirements. The Muskingum model uses the water balance equation and the tank storage equation to replace the complex hydrodynamic equation. The simplified equation is as follows:

[0095]

[0096] In the formula: 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 outlet section, K is the tank storage coefficient, and x is the flow weight factor.

[0097] 4. About the flood forecasting LSTM model

[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 is 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 cell C t Together we 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 dispatch model

[0103] The flood control dispatch model mainly includes the change process of the highest water level in front of the reservoir dam and the change process of the downstream discharge flow. The model mainly considers flood control discharge, and needs to be combined with the specific application mode of reservoir flood control dispatch, water level and storage capacity relationship, and building parameter operation.

[0104] (1) Objective function:

[0105] The maximum peak reduction principle is adopted, and the safety of the dam and the protection of the downstream areas of the reservoir from flood disasters are comprehensively considered. The maximum downstream discharge flow minimum model is established with the minimum flood peak flow in the downstream flood control section as the main goal.

[0106]

[0107] Where: is the maximum discharge 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 dispatching period, and t is the dispatching 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 in period t, respectively, and 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, Q min is the minimum outflow rate 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 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 respectively represent the upper and lower limits of the reservoir water storage capacity.

[0128] 7) The expression of 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 routing 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 of flood evolution and flooding risk on both banks of the river downstream of the reservoir

[0136] The model for simulating flood evolution and inundation risk on both banks of the river downstream of the reservoir adopts a one-dimensional hydrodynamic model of the river, a two-dimensional planar hydrodynamic model and a one- and two-dimensional coupling 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), and the equations 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 calculation include the continuity equation and momentum equation, as follows:

[0143] Continuity equation:

[0144]

[0145] Momentum equation:

[0146]

[0147] Where: is the average velocity 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 water head, u and v are the velocity components in the x and y directions, g is the gravitational acceleration, ρ is the water density, 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 flow rate of the point source, u s 、v s is the flow rate of the source-sink water flow.

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

[0149] Regarding the planning of the monitoring station network for the reservoir-river-levee system, the needs of air and ground monitoring should be comprehensively considered to build a multi-mode, multi-level, integrated monitoring station network system consisting of meteorological satellites and rainfall radars, rainfall stations, hydrological stations, and water level stations. 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 flash floods and rivers with small and medium-sized towns, villages, industrial and mining enterprises, important infrastructure and other flood control targets downstream, hydrological stations are set up at the exits of the mountains 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; water level stations are set up near the mouths of tributaries entering the reservoir (lake) and at locations where the water surface narrows or widens along the way; water level stations are set up at the end of the reservoir and in the middle of the reservoir area affected by variable 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 an 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 basin controlled by the reservoir, and at least one rain gauge is set up in a small reservoir.

[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, and are networked.

[0154] (5) Video surveillance stations: Video surveillance stations shall be set up in reservoirs. If communication conditions are available, no less than two video surveillance stations shall be set up. Video surveillance stations shall be set up in the river area downstream of the reservoir according to the degree of flooding danger.

[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, artificial intelligence, etc., respectively constructs the "forecast", "warning", "rehearsal" and "plan" models of the reservoir-river-embankment multi-engineering system, and clarifies the dynamic correlation mechanism among the four prediction models, so as to realize high-precision flood forecasting and warning, dynamic rehearsal of flood risks and scientific preparation of flood control plans.

[0156] For those skilled in the art, improvements and changes can be made according to the above invention content. Any modification, improvement and change made on the basis of the present invention should fall within the protection scope of the present invention.

Claims

1. A multi-engineering flood prevention and disaster reduction method based on a four-precoupling model, characterized in that: Specifically include: Construct a flood control and disaster reduction forecasting model for a reservoir-river-levee multi-engineering system, wherein 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 dispatching 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 dispatching and downstream river levee risk forecasting model takes the output of the reservoir upstream area flood forecasting model as input, and the model includes a reservoir engineering flood control dispatching submodel and a reservoir downstream river channel-levee overflow and breach one- and two-dimensional hydrodynamic coupling submodel, wherein the reservoir engineering flood control dispatching submodel combines multi-source data including terrain elevation, river system, and water conservancy engineering facility dispatching rules to obtain an output 1 as the reservoir dispatching discharge flow The process is used as the input condition of the one-dimensional hydrodynamic coupling submodel of the downstream river channel and dike overtopping of the reservoir. The one-dimensional hydrodynamic coupling submodel of the downstream river channel and dike overtopping of the reservoir realizes the simulation of the flood evolution and flooding risk of the downstream river channel of the reservoir, and the output 2 is obtained, which is a dynamic simulation of the reservoir operation and the downstream river channel flood evolution, the flood inundation risk of the dike on both sides, calculates the flood inundation range caused by typical frequency floods, and predicts and evaluates the impact of different magnitude floods on the flood control safety of the reservoir, the dike safety and the flooding risk that may be caused to the coastal area; the output 1 and the output 2 are used together as the output of the reservoir operation and downstream river channel dike risk prediction model; the flood inundation disaster prediction model takes the output of the flood prediction model of the upstream area of ​​the reservoir, the output 1 and the output 2 of the reservoir operation and downstream river channel dike risk prediction model as input, and takes the flood inundation disaster prediction result of the extreme catastrophic events of reservoir overtopping and river channel dike overtopping on the flood disaster risk of the downstream area as the output; 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 river channel and levee risk early warning model below the dam, and a reservoir downstream residential flooding risk early warning model; wherein the meteorological risk early warning model takes the observation of regional rainfall and the formation of 24h hourly forecast rainfall data as input, and takes the early warning result of the meteorological risk level as output; the reservoir risk early warning model further includes a rainfall station monitoring and 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 station monitoring and 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 danger 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 risk warning model for the river embankment below the dam further includes a rain gauge monitoring and warning sub-model, a river 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 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 the model; the levee engineering safety early warning sub-model takes as input the basic data including at least the levee soil moisture content, crack development degree, infiltration line height, soil deformation, and slope balance safety factor, and takes the levee disaster accident early warning response result 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 the dynamic monitoring image of the possible flooded area as the input of the sub-model, and provides data support and decision-making basis for disaster assessment and rescue work as the output; the simulation early warning sub-model is based on the evolution of river floods downstream of the reservoir and the simulation of flooding risks on both sides of the river; Construct a flood prevention and disaster reduction rehearsal model for a multi-engineering system of reservoirs, rivers and levees, with terrain elevation, river systems and remote sensing image data in the downstream area of ​​the reservoir as input, and dynamic visualization of the entire process of rainfall runoff, reservoir scheduling, flood evolution and inundation as output; Constructing a reservoir-river-levee multi-engineering system flood control and disaster reduction plan model by combining the reservoir-river-levee multi-engineering system flood control and disaster reduction forecast model, the reservoir-river-levee multi-engineering system flood control and disaster reduction early warning model and the reservoir-river-levee multi-engineering system flood control and disaster reduction rehearsal model, taking the outputs of the forecast model, early warning model and rehearsal model as input, and taking the reservoir-levee-flooding area multi-engineering system risk control plan under standard flood conditions as output; Plan and design a monitoring network, including at least rainfall stations, hydrological stations, water level stations and video stations, and carry out four-pronged dynamic coupling including forecast, warning, rehearsal and plan.

2. A multi-engineering 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. According to claim 2, a multi-engineering flood prevention and disaster reduction method based on a four-pre-coupling model, wherein 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.

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 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. The forget gate is used to discard the unimportant related information of the flood forecast 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