CN119990755B

CN119990755B - Multi-working-system flood control and disaster reduction method based on four pre-coupling models

Info

Publication number: CN119990755B
Application number: CN202510067432.2A
Authority: CN
Inventors: 田福昌; 张安邦; 李孝彬; 刘明辉; 杨长青; 刘建伟; 刘金蕊; 于海滨; 毛新宇; 石文涛; 王海涛
Original assignee: Chengde Water Conservancy Workstation; Tianjin University
Current assignee: Chengde Water Conservancy Workstation; Tianjin University
Priority date: 2025-01-16
Filing date: 2025-01-16
Publication date: 2025-08-08
Anticipated expiration: 2045-01-16
Also published as: CN119990755A

Abstract

The invention discloses a multi-working procedure flood control and disaster reduction method based on four pre-coupling models, which comprises the steps of constructing a flood control and disaster reduction forecast model of a reservoir-river-embankment multi-engineering system, constructing a flood control and disaster reduction early warning model of the reservoir-river-embankment multi-engineering system, constructing a flood control and disaster reduction forecast model of the reservoir-river-embankment multi-engineering system, planning and designing a monitoring network according to a certain rule, and carrying out four pre-dynamic couplings including forecast, early warning, forecast and forecast. Compared with the prior art, the invention comprehensively considers the synergistic effect and influence restriction relation of the flood control process of the reservoir, the river channel and the embankment, couples the flood control four-pre-step dynamic coupling model through the multi-step system, realizes the chained dynamic connection of the flood control and disaster reduction four-pre-step links, and realizes the accurate flood forecast, scientific flood control and disaster reduction decision and refined flood risk management.

Description

Multi-working-system flood control and disaster reduction method based on four pre-coupling models

Technical Field

The invention relates to the field of emergency disaster prevention and flood control and disaster reduction intelligent schemes, in particular to a four-pre-coupling model for flood control and disaster reduction of a reservoir-river channel-embankment multi-engineering system.

Background

The reservoir-river-embankment multi-engineering system flood control and disaster reduction four-pre-coupling model takes a reservoir-river-embankment as an integrated unit, realizes a model of coupling functions of flood forecasting, early warning, forecasting and planning, and is one of important non-engineering measures for comprehensively forecasting, evaluating and managing flood disaster risks and reducing flood disaster losses.

The reservoir, the river course and the embankment are taken as important infrastructure for flood control and disaster reduction, and the exertion of the flood control capability plays an important role in protecting the safety of downstream people and reducing disaster area. At present, the flood control and disaster reduction field has constructed a diversified and technical intensive model system, the core of which mainly relies on hydrologic models, hydrodynamic models, remote sensing and geographic information system technologies and numerical simulation technologies, and artificial intelligence and data mining technologies are introduced, so that the flood control and disaster reduction model is relatively abundant in technical research and application. However, in the current flood control and disaster reduction model research and practice, the reservoir-river channel-embankment three are not completely researched as an integrated unit, but are researched singly or combined in pairs, so that systematic thinking and planning from the whole angle are easily lacking, natural synergistic serial effect and mutual influence constraint relation among the three are difficult to comprehensively consider, the system toughness and the whole effect of flood control and disaster reduction are reduced, and scientific scheduling and management of flood risks cannot be realized. More importantly, the current research is lack of consideration of the inherent dynamic coupling system of the four pre-prediction function, which causes difficulty in implementing the forward-looking pre-modeling of the whole physical basin elements and the whole water conservancy management activity process when facing complex and changeable flood situations, and ensures the early detection of risks, early release of early warning, early formulation of schemes and early implementation of measures. Meanwhile, the coupling system between the existing four pre-models is not perfect, and the lack of deep integration and cooperative optimization leads to obvious defects of information sharing and functional complementation between the models, so that the chain dynamic connection of flood control and disaster reduction links is difficult to realize, the full play of the overall flood control and disaster reduction efficiency is affected to a certain extent, and a high-efficiency and unified flood control and disaster reduction system is not formed. Therefore, the reservoir-river channel-embankment multi-engineering system is taken as an integral unit, and the four pre-coupling modes of flood control and disaster reduction of the complex engineering system are established and perfected, so that the method has important significance for improving the scientificity and the accuracy of flood control risk scheduling of the river basin.

In summary, although the flood control and disaster reduction field already has diversified water conservancy professional models with rich functions, the defects of lack of coupling and linkage among the functional models of 'reservoirs', 'river channels' or 'embankments', 'four-prediction' (forecasting, early warning, previewing and planning) all form key factors for restricting the further development of the flood control and disaster reduction field. In view of the above, the invention combines the integrated unit of reservoir-river-embankment with the four pre-coupling model system by scientific methods such as theoretical analysis, numerical simulation and the like, and from the integral angle of reservoir-river-embankment, the four pre-coupling model method for flood control and disaster reduction of the reservoir-river-embankment multi-engineering system is provided by taking prediction as a basis, early warning as a whistle, and prediction as a key and the purpose of the prediction as a forward-looking thought, realizing four pre-functions of prediction, early warning, prediction and prediction as well as flood control and disaster reduction of the prediction, strengthening dynamic coupling among models and defining interaction and influence mechanisms among the models.

Disclosure of Invention

Aiming at the defect that the coupling system among the reservoir, river and embankment in the current flood control and disaster reduction field has defects, the invention aims to focus the important requirements and development trend of reservoir flood control emergency management, and provides a multi-process system flood control and disaster reduction method based on the four pre-coupling model, which takes the reservoir, river and embankment as an integrated research unit and realizes a model system of forecasting, early warning, forecasting and disaster reduction for a reservoir, river and embankment multi-project system through coupling integration and collaborative optimization among the four pre-coupling model models.

The invention is realized by the following technical scheme:

A multi-working procedure flood control and disaster reduction method based on a four-pre-coupling model specifically comprises the following steps:

The method comprises the steps of constructing a flood control and disaster reduction forecast model of a reservoir-river channel-embankment multi-engineering system, wherein the forecast model at least comprises a weather rainfall forecast model, a flood forecast model of an upstream area of the reservoir, a reservoir dispatching and downstream river channel embankment risk forecast model and a flood submerged disaster forecast model which are deeply coupled in a multi-dimensional mode; the weather rainfall forecast model takes meshed rainfall live data as input and meshed weather rainfall forecast data as output; the reservoir dispatching and downstream river course embankment risk forecasting model takes the output of the reservoir upstream area flood forecasting model as the input, the model comprises a reservoir engineering flood dispatching sub-model and a reservoir downstream river course-embankment two-dimensional hydrodynamic coupling sub-model, the reservoir engineering flood dispatching sub-model combines multisource data comprising topography elevation, river water system and hydraulic engineering facility dispatching rules to obtain the input condition that the output 1 is the reservoir dispatching lower discharge flow process and is taken as the reservoir downstream river course-embankment flood break two-dimensional hydrodynamic coupling sub-model, the reservoir downstream flood evolution and two-bank flooding risk simulation is realized by the reservoir downstream river course-embankment two-dimensional hydrodynamic coupling sub-model, the output 2 is obtained by dynamically simulating reservoir dispatching and downstream river course flood evolution, embankment two-bank flooding risk, the flooding range caused by typical frequency flood is calculated, and the reservoir flooding safety of different levels is evaluated for the reservoir flooding is forecasted, the flood submerged disaster prediction model takes the output of a flood prediction model of an upstream area of the reservoir, the output 1 and the output 2 of the flood prediction model of the reservoir schedule and the downstream river bank, and takes the flood submerged disaster prediction result of a critical-focus extreme disaster event of the flood submerged dam break of the reservoir and the river bank break of the river bank for the submerged disaster risk of the downstream area as the output;

The method comprises the steps of constructing a flood control and disaster reduction early warning model of a reservoir-river channel-embankment multi-engineering system, wherein the early warning model is a model system formed by multi-dimensional deep coupling of a meteorological risk early warning model, a reservoir risk early warning model, an under-dam river channel embankment risk early warning model and a reservoir downstream resident inundation risk early warning model; the weather risk early warning model takes regional rainfall as an input, forecast rainfall data for 24 hours per hour is formed, and early warning results of weather risk grades are taken as an output, the reservoir risk early warning model further comprises a rainfall station monitoring early warning sub model, a storage flood early warning sub model, a reservoir storage area flood inundation risk early warning sub model and a dam engineering safety early warning sub model, the rainfall station monitoring early warning sub model takes actual measurement rainfall data as an input, whether rainfall triggers rainfall early warning response judgment results as an output, the reservoir storage area flood inundation risk early warning sub model takes reservoir area water level all-weather monitoring data as an input, and flood inundation risk forecast results as an output, the dam engineering safety early warning sub model comprises flood inundation risk early warning processing and dam body instability early warning processing, specifically, the flood inundation risk early warning processing takes storage flood and reservoir level monitoring data, combines a water level storage capacity relation and reservoir drainage capacity input 1, the reservoir water level development trend is predicted as an output 1 of the sub model, the dam instability early warning processing takes water level crack, the water content, the dam elevation degree, the bank height and the deformation coefficient and the safety coefficient 2 are taken as an input, taking the structural safety condition of a real-time monitoring dam (embankment) as an output 2 of the model; the under-dam river channel embankment risk early warning model further comprises a rainfall station monitoring early warning sub-model, a river channel flood risk simulation early warning sub-model and an embankment engineering safety early warning sub-model, and specifically, the rainfall station monitoring early warning sub-model takes rainfall monitoring data as input and triggers rainfall early warning response of corresponding grades as output; the river course flood risk simulation early warning sub-model takes a model calculation boundary condition output by the reservoir engineering flood control scheduling sub-model as input and takes a downstream river course flood risk early warning response result as output of the model, the embankment engineering safety early warning sub-model takes basic data at least comprising embankment soil moisture content, crack development degree, infiltration line height, soil deformation and bank slope balance safety coefficient as input and takes an embankment disaster accident early warning response result as output of the model, the reservoir downstream resident flooding risk early warning model further comprises a monitoring early warning sub-model and a simulation early warning sub-model, the monitoring early warning sub-model takes a dynamic monitoring image of a possible flooding area as input of the sub-model and takes disaster evaluation and rescue work as output, and the simulation early warning sub-model is based on reservoir downstream river course flood evolution and two-bank flooding risk simulation treatment;

A flood control and disaster reduction previewing model of a reservoir-river channel-embankment multi-engineering system is built, the topography elevation of a downstream area of the reservoir, a river water system and remote sensing image data are taken as inputs, and dynamic visualization display of the whole process of rainfall runoff, reservoir scheduling, flood evolution and flooding is taken as outputs;

constructing a flood control and disaster reduction plan model of a reservoir-river-embankment multi-engineering system, and combining the flood control and disaster reduction forecast model of the reservoir-river-embankment multi-engineering system, the flood control and disaster reduction early warning model of the reservoir-river-embankment multi-engineering system and the flood control and disaster reduction forecast model of the reservoir-river-embankment multi-engineering system, wherein the output of the forecast model, the early warning model and the forecast model is used as input, and the risk regulation and control scheme of the reservoir-embankment-flood zone multi-engineering system under the standard flood condition is used as output;

The monitoring network is planned and designed, and at least comprises a rainfall station, a hydrological station, a water level station and a video station, and four pre-dynamic couplings including forecasting, early warning, previewing and planning are carried out.

In some embodiments, the reservoir upstream region flood forecast model further comprises a physical mechanism flood forecast sub-model and a data driven flood forecast sub-model.

In some embodiments, the physical mechanism flood forecast sub-model is a combined model consisting of an SCS current generation sub-model, an instantaneous unit line sink sub-model, and Ma Sijing flood sink sub-models coupled in series.

In some embodiments, the data-driven flood forecast sub-model adopts an LSTM model, wherein the input of the flood forecast LSTM model is a rainfall process of the first n hours, a rainfall process of the runoff, a rainfall process of the next m hours, and a runoff process of the next m hours, the output is a runoff process of the next m hours, the relevant information which is not important with the flood forecast is abandoned by using a forgetting gate, the retention degree of the information is controlled, the neural network model is updated to be new important information by using an updating gate, a candidate value is generated, the important information which needs to be output is determined by using an output gate, and a hidden state is generated for predicting or serving as the input of the next time step.

Compared with the prior art, the invention has the following positive technical effects:

The method is characterized in that the cooperative effect and the influence restriction relation of the flood control process of a reservoir, a river channel and a embankment can be comprehensively considered from the overall view angle, the hydrokinetic coupling flood control four-pre dynamic coupling model of a multi-process system on the upstream side, the downstream side, the left side and the right side of the reservoir is considered by constructing the reservoir as a core, the chain dynamic connection of the flood control and disaster reduction four-pre links is realized, the flood forecast accuracy, the flood control and disaster reduction decision science and the flood risk management refinement are realized by taking the river basin as the overall view angle, flood control, flood discharge and flood peak reduction are timely carried out, flood control is effectively carried out, the emergency management intelligence level of the reservoir flood control is improved, the flood control decision efficiency and accuracy are remarkably reduced, and important technical support is provided for water safety guarantee.

Drawings

FIG. 1 is a flow chart of a flood control and disaster reduction method of a multi-working system based on a four pre-coupling model;

FIG. 2 is a block diagram of an algorithm of an LSTM model of the used artificial neural network;

FIG. 3 is a diagram of geographic locations of a study area;

FIG. 4 is a diagram of a process line for forecasting flood in an area upstream of a mountain reservoir;

FIG. 5 is a process line of the discharge flow under a certain mountain reservoir schedule;

FIG. 6 is a pre-modeling result-flood maximum submerged depth map I (accent submerged area);

FIG. 7 is a pre-modeling result-flood maximum submerged depth map II (accent submerged area);

Fig. 8 is a diagram of a four pre-coupling model planning monitoring station network for a mountain reservoir.

Detailed Description

The invention will be described in further detail with reference to the accompanying drawings and examples.

As shown in FIG. 1, the overall flow of the multi-working-procedure flood control and disaster reduction method based on the four pre-coupling model specifically comprises the following steps:

firstly, constructing a flood control and disaster reduction forecast model of a reservoir-river channel-embankment multi-engineering system;

The forecasting model is a model system formed by multi-dimensional deep coupling of a weather rainfall forecasting model, a flood forecasting model of an upstream area of a reservoir, a reservoir dispatching and downstream river dike risk forecasting model and a flood inundation disaster forecasting model. The forecasting model meets the real-time rolling forecasting capability, sets water levels, flow rates, water amounts and the like of different points or sections as forecasting elements, quantifies the water levels, the water amounts and the like, performs all-round and multi-layer flood disaster forecasting of different forecasting periods on the reservoir and the upstream and downstream river dike areas, and analyzes the dynamic coupling mechanism among various forecasting methods. The specific description is as follows:

1.1, a weather rainfall forecast model is used for observing and forecasting rainfall in different space scale ranges in different areas by deploying phased array type water conservancy rain measuring radar in a storm flood prone area and a mountain flood disaster prone area and weather radar around cities, so that the forward movement from 'falling rain' to 'rain in cloud' in rain condition monitoring is realized. The method comprises the steps of using a water conservancy rain measuring radar to observe liquid water in the atmosphere of a near-ground layer in a height range from the ground to 2km, generating high-precision 30m multiplied by 30m grids in real time, using continuous high-space-time resolution grid rainfall live data with time resolution up to 40s as input of the model, using grid weather rainfall forecast data with high space-time resolution extrapolated for 1-3h as output of the model on the basis of 3-part rain measuring radar networking application, using the weather radar to observe all weather elements in the atmosphere in a height range from the ground to the top of a troposphere and 20-30km, carrying out large-range sampling observation on the atmosphere, solving the problem of vertical detection of the atmosphere, and meeting the requirement of flood control forecast on rainfall forecast.

The model takes grid weather rainfall forecast data output by the weather rainfall forecast model in the step 1.1 as input, particularly a physical and data dual-driven flood high-precision forecast model of the upstream area of the reservoir, wherein the physical mechanism flood forecast sub-model mainly comprises an SCS flow-producing sub-model, an instantaneous unit line flow-converging sub-model and Ma Sijing flood evolution sub-models, the three models are in series coupling relation, the output of the last model is used as the input of the next model, and the SCS flow-producing sub-model is a runoff process model generated in a rainfall loss stage. The instantaneous unit line sink sub-model realizes the simulation of the whole runoff process. Ma Sijing flood evolution sub-models consider the water flow speed, the flow rate and the water level height to realize the flood process simulation. The data-driven flood forecasting sub-model mainly comprises an LSTM model and other artificial neural network models, wherein the inputs of the flood forecasting LSTM model are the first n hours of rainfall process and runoff process, the next m hours of rainfall process, the outputs are the next m hours of runoff process, n and m are generally integer multiples of 6 in the invention), n and m are generally integer multiples of 6 in the invention, the relevant information which is not important for flood forecasting is abandoned by a forgetting gate, the retention degree of the information is controlled, the new important information is updated in the neural network model by an updating gate to generate candidate values, the important information which needs to be output is determined by an output gate to generate a hidden state, and the hidden state is used for forecasting or serving as the input of the next time step, as shown in fig. 2, so that the real-time dynamic coupling of the water conservancy rain measuring radar monitoring forecasting data and the upstream area flood forecasting model is realized, meanwhile, the model calibration rate is carried out by combining with a historical experience database, the key indexes (the key indexes of the flood peak, the flood peak arrival time, the total quantity and the rainfall exploration time) of the upstream area flood forecasting model are acquired after the parameters of the flood forecasting model are adjusted, and then the reservoir continuous reservoir is influenced on the reservoir process.

1.3, Reservoir dispatching and downstream river course embankment risk prediction model, take the output result of the reservoir upstream region flood prediction model in 1.2 as the input of the model, the model comprises a reservoir engineering flood dispatching sub-model and a reservoir downstream river course-embankment flood break two-dimensional hydrodynamic coupling sub-model, the reservoir engineering flood dispatching sub-model combines multi-source data such as high-precision terrain elevation, river water system, hydraulic engineering facility dispatching rules and the like to obtain output 1 which is the input condition of the reservoir dispatching lower discharge flow process and is used as the reservoir downstream river course-embankment flood break two-dimensional hydrodynamic coupling sub-model, the reservoir downstream river course flood evolution and two-dimensional hydrodynamic coupling sub-model realizes reservoir downstream river course flood evolution and two-bank flood risk simulation, output 2 is dynamic simulation dispatching and downstream flood evolution and embankment two-bank flood risk, the flood flooding range caused by typical frequency is calculated, and the possible influence of different levels of flood on reservoir safety, embankment safety and bank protection area flood risk is predicted and evaluated. And taking the output 1 and the output 2 together as the output of the reservoir dispatching and downstream river dike risk prediction model.

1.4, A flood inundation disaster prediction model, wherein the output 1 and the output 2 of the reservoir upstream area flood prediction model in 1.2 and the reservoir dispatching and downstream river dike risk prediction model in 1.3 are used as inputs of the model, flood inundation disaster prediction results of critical extremely disaster events such as reservoir overdam break, river dike overdam break and the like on the flood inundation disaster risks in the downstream area are used as outputs of the model, and basic statistical indexes such as population, cultivated land, total production value and the like are combined, and the model output is evaluated based on GIS construction flood disaster influence analysis and loss evaluation model so as to comprehensively predict and evaluate the disaster influence loss.

Step two, constructing a flood control and disaster reduction early warning model of the reservoir-river channel-embankment multi-engineering system;

The early warning model is a model system formed by multi-dimensional deep coupling of a meteorological risk early warning model, a reservoir risk early warning model, an under-dam river dike risk early warning model and a reservoir downstream resident inundation risk early warning model, and provides a planning monitoring station network layout scheme according to flood control and disaster reduction 'early warning' requirements, wherein the reservoir risk early warning model further comprises a rainfall station monitoring early warning sub-model, a warehouse-in flood early warning sub-model and a reservoir area flood inundation risk early warning sub-model, and the under-dam river dike risk early warning model further comprises a rainfall station monitoring early warning sub-model, a river flood risk early warning sub-model and a dike engineering safety early warning sub-model. The method comprises the steps of comprehensively considering the change factors including at least the characteristics of storm flood in a small river basin and early rainfall or soil water content as the input of the model, analyzing the natural characteristics of flood, the safety of reservoir-river channel-embankment engineering and the social and economic effects of flood disasters, combining meteorological monitoring and flood forecasting results, determining a flood prevention and reduction early warning index system consisting of indexes such as critical rainfall, critical water level and critical flow, and the like, taking reservoir-river channel-embankment design data, hydrologic and historical flood disaster data as the input of the model, determining a flood risk early warning index threshold value to realize the real-time dynamic early warning of the occurrence time and occurrence scale of the flood disasters as the output of the model, and issuing early warning information for specific areas and specific crowds to support emergency evacuation and rescue arrangement of disaster area personnel.

2.1, A weather risk early warning model, wherein regional rainfall is observed and 24h hour-by-hour forecast rainfall data are formed as input of the model, and the regional rainfall is observed and 24h hour-by-hour forecast rainfall data are formed through a phased array type water conservancy rain measuring radar networking and an meteorological weather radar which are easy to deploy in storm floods and mountain torrents in areas on the upstream and downstream of a reservoir. Grid rainfall (changing along with the state of the water content of the soil) of each early warning period (3 h, 6h, 12h and 24 h) is used as a flood disaster weather risk early warning index, early warning index thresholds corresponding to 4 weather risk levels of low (possible occurrence, blue early warning), medium (possible high, yellow early warning), high (possible high, orange early warning) and extremely high (possible high, red early warning) are calculated respectively, and early warning results of the weather risk levels are output.

2.2, Reservoir risk early warning model, specifically including following submodel:

(1) The rainfall station monitoring and early warning sub-model is used for monitoring rainfall data of each early warning period in real time by depending on a rainfall station network densely distributed on the upstream and downstream of a reservoir, and observing and recording various rainfall indexes such as the maximum rainfall, 24h rainfall, regional average rainfall, accumulated rainfall and the like. The measured rainfall data is used as the input of the submodel, the analysis and calculation method of disaster water level and design of the rainstorm flood reverse-push critical rainfall is adopted, the rainfall in each early warning period is selected as a real-time dynamic early warning index, and whether the rainfall triggers the rainfall early warning response judgment result is used as the output of the submodel.

(2) And in addition, key indexes such as the flow of the entering flood are obtained in real time by depending on a reservoir tail flow monitoring system, and the reservoir area flood safety is analyzed by combining the water storage capacity and the incoming flow condition, so that the real-time monitoring and early warning of the rain and flood are realized.

(3) The reservoir area flood risk early warning submodel relies on a reservoir area water level all-weather monitoring system to analyze water level dynamic change in real time, the reservoir area water level all-weather monitoring data is used as input of the submodel, a flood risk corresponding water level early warning index threshold value is set, whether an out-of-standard water level condition occurs or is about to occur is judged in real time, and if the out-of-standard water level condition occurs, flood risk monitoring early warning is triggered. In addition, based on the output of the flood inundation disaster prediction model in 1.4 and the reservoir engineering flood control scheduling sub-model in 1.3, the water level change process during reservoir operation scheduling is dynamically calculated by combining the upstream water supply, the water level storage capacity relation, the reservoir scheduling operation mode and the rainfall prediction result, the flood inundation risk prediction result is taken as the output of the mold model, the flood trend and the potential inundation risk of the reservoir are comprehensively evaluated, whether the standard water level is exceeded or not is judged, and if the standard water level exceeds the standard water level, the flood inundation risk prediction and early warning are triggered.

(4) The dam engineering safety early warning sub-model comprises ① dam-spreading risk early warning processing, namely, input 1 of the sub-model is formed by combining water level storage capacity relation and reservoir drainage capacity according to actual measured water level monitoring data of the storage flood and the reservoir, output 1 of the sub-model is formed by predicting reservoir water level development trend according to the water level storage capacity relation and the reservoir drainage capacity according to actual measured water level monitoring data of the storage flood and the reservoir, and dam-spreading disaster accident early warning response is triggered if the reservoir water level exceeds a dam top elevation. ② Dam body instability early warning treatment, namely, dam body dangerous situation data such as dam body water content, crack development degree, infiltration line height, soil deformation, bank slope balance safety coefficient and the like are used as input 2 of the model, dam body dangerous situation data such as dam body water content, crack development degree, infiltration line height, soil deformation, bank slope balance safety coefficient and the like are monitored by using a high-precision sensor, structural safety conditions of a real-time monitoring dam (embankment) are used as output 2 of the model by using modes such as video remote monitoring, regular inspection by staff and the like, a preset safety threshold value and a dam design standard are compared, stability and risk level of the dam are evaluated, and dam instability damage disaster accident early warning response is triggered.

2.3, River dike risk early warning model under dam, specifically including the following submodel:

(1) And the rainfall station monitoring and early warning sub-model takes rainfall monitoring data as input of the sub-model, relies on a network of a rainfall station under a dam, monitors the rainfall data in real time, adopts an analysis method of disaster water level and design of a heavy rainfall flood back-push critical rainfall, selects the rainfall of each early warning period of a river channel two-bank protection object as a real-time dynamic early warning index, triggers the rainfall early warning response of a corresponding level as output of the sub-model, judges the condition of the exceeding standard rainfall, and triggers the rainfall early warning response of the corresponding level. The early warning indexes are divided into two stages, and the early warning indexes respectively correspond to the preparation transfer and the immediate transfer of the disaster prevention object.

(2) The river channel flood risk simulation early warning sub-model takes a model calculation boundary condition output by the reservoir engineering flood control scheduling sub-model in 1.3 as input, based on the output of the reservoir engineering flood control scheduling sub-model and combined with information such as river channel topography, embankment elevation and remote sensing images, a reservoir downstream river channel flood evolution and two-bank flooding risk simulation coupling model is built, a reservoir downstream area flood evolution process is dynamically simulated, a predicted flood or typical frequency flood flooding range is calculated, key parameters such as river channel water level and flow are monitored in real time, the possibility that river channel flood exceeds the embankment warning water level and guarantees the water level and is higher than the elevation of a two-bank danger area or a protection area ground table is comprehensively estimated, downstream river channel flood risk early warning response is triggered, and a downstream river channel flood risk early warning response result is taken as the output of the model.

(3) The embankment engineering safety early warning sub-model takes basic data such as the water content of embankment soil, the development degree of cracks, the height of a infiltration line, the deformation of soil, the balance safety coefficient of a bank slope and the like as input of the sub-model, monitors the basic data such as the water content of the embankment soil, the development degree of cracks, the height of the infiltration line, the deformation of soil, the balance safety coefficient of the bank slope and the like according to a high-precision sensor, monitors the structural safety condition of the embankment in real time by means of video remote monitoring, regular inspection by staff and the like, compares a preset safety threshold value with a embankment flood design standard, evaluates the stability and risk level of the embankment, triggers the embankment disaster accident early warning response, and takes the embankment disaster accident early warning response result as output of the model.

2.4, Reservoir downstream resident inundation risk early warning model, specifically comprising the following sub-models:

(1) The monitoring and early warning submodel takes a dynamic monitoring image of a possible submerged area as the input of the submodel, utilizes the real-time three-dimensional monitoring means of space, ground and the like such as high-definition video monitoring, satellite remote sensing image identification, unmanned aerial vehicle aerial photography and the like to realize the dynamic monitoring of the possible submerged area, captures the real-time image of the flood submerged process in real time, accurately grasps the flood spreading range, clearly identifies the target object of the submerged area, evaluates the index data such as the water depth, the water flow velocity and the like of the flood submerged area, and accordingly triggers the flood risk monitoring and early warning response and provides data support and decision basis for the follow-up emergency response, the prevention of secondary disasters and the disaster assessment and rescue work as the output of the submodel.

(2) Based on reservoir downstream river flood evolution and two-bank submerged risk simulation, the simulated early warning sub-model focuses on extreme disaster events such as reservoir flood break, river embankment flood break and the like, combines social economic data (statistical indexes such as population, cultivated land and GDP) and simulates and predicts submerged ranges, submerged depths and disaster losses of downstream resident areas of different disaster event scenes according to a disaster situation statistics and loss assessment method, and triggers submerged disaster early warning response of reservoir downstream resident according to risk grades.

According to the flood control and disaster reduction forecast and early warning requirements, a monitoring network is planned and designed according to a certain rule (for example, the main principle of being capable of covering and monitoring rain conditions, water conditions, work conditions and dangerous conditions of reservoirs, riverways, embankments and two-bank inundation dangerous areas), for example, the rainfall capacity of a rainfall station monitoring area is generally arranged near the centroid of a river basin, a hydrological station and a water level station are used for monitoring the water surface elevation of the reservoirs and the riverways, the hydrological station can be additionally used for monitoring the flow process of the riverways and is generally arranged at a reservoir or an embankment water position, the video station is used for monitoring whether the water level reaches the pre-warning water surface elevation, monitoring whether the dangerous areas are submerged by flood or not, and the like, and is arranged at a position capable of being monitored by remote images, including the hydrological station, the rainfall station, the rain measuring radar, the video station and the like, so as to improve the sensing capacity including forecasting, early warning, and pre-playing and scheme, namely 'four pre-warning'.

Step three, constructing a flood control and disaster reduction previewing model of the reservoir-river channel-embankment multi-engineering system;

Taking data information such as a reservoir downstream area terrain elevation, a river water system, a remote sensing image and the like as input of the model, taking multiple influences of a constraint function of a dike and interval inflow into consideration, determining model boundary conditions, river course and rough region roughness and the like, dynamically linking the reservoir upstream area flood forecasting model in 1.2, the reservoir engineering flood control scheduling sub-model in 1.3 and a reservoir downstream river course-dike flood break two-dimensional hydrodynamic coupling sub-model at the same time, and simulating the flood submerging process and distribution characteristics under the conditions of super-standard flood and flood break. And constructing a two-dimensional plane and three-dimensional integrated digital twin scene by adopting technical means such as GIS, three-dimensional modeling and rendering, hydrokinetic model and the like, supporting rainfall runoff, reservoir dispatching, flood evolution and submerged whole-process dynamic visual display, and outputting the rainfall runoff, reservoir dispatching, flood evolution and submerged whole-process dynamic visual display as the module.

The related art is described as follows:

GIS technology is mainly used for processing geospatial data including spatial distribution of river, road, terrain and other elements.

2. The three-dimensional modeling and rendering technology mainly utilizes three-dimensional modules in three-dimensional modeling software or GIS software to construct a three-dimensional scene according to geographic space data, wherein the three-dimensional scene comprises river flow direction, road layout, topography fluctuation and the like, so that reasonable illumination and shadow effects are set for the three-dimensional model, materials and textures are added, and the sense of reality and the sense of three-dimensional sense of the scene are enhanced, so that the three-dimensional scene is more lifelike.

3. The hydrologic hydrodynamics model is mainly characterized in that a flood overall process calculation model is built according to parameters such as river basin topographic features and meteorological conditions, initial conditions, boundary conditions and calculation parameters are set, risk information such as a flood submerging range, water depth and flow velocity is obtained after the model is operated, simulation result data are integrated through a GIS or a data interface of a three-dimensional platform, and after processes such as color coding, transparency adjustment and animation production, a flood evolution submerging process is intuitively simulated in a three-dimensional digital scene.

Step four, constructing a flood control and disaster reduction plan model of the reservoir-river channel-embankment multi-engineering system;

The flood control and disaster reduction 'pre-scheme' model mainly focuses on disaster prevention and control and risk reduction, combines the reservoir-river-dike multi-engineering system flood control and disaster reduction prediction model in the first step, the reservoir-river-dike multi-engineering system flood control and disaster reduction early warning model in the second step and the reservoir-river-dike multi-engineering system flood control and disaster reduction pre-scheme in the third step, takes the output of the prediction model, the early warning model and the pre-scheme model as the input of the pre-scheme model, simulates the reservoir scheduling operation process under different storm situations, takes the downstream flood risk minimization principle, prefers different working condition flood scheduling defense schemes, and accordingly establishes a reservoir-dike-flood area multi-engineering system risk regulation scheme under the super-standard flood condition as the output of the pre-scheme. The specific description is as follows:

1. The disaster prevention and control plan is based on a rainwater condition monitoring and early warning system (comprising satellite remote sensing, radar rain measurement, hydrologic automatic monitoring stations and the like) to collect information in real time, analyze hydrologic weather conditions in the reservoir and the river basin range, and predict future flood development situation in the reservoir-river channel-embankment multi-engineering system range by combining a forecasting and early warning model, so as to establish a zonal grading early warning release scheme and suggested prevention and control measures. Wherein:

(1) The early warning release scheme is mainly used for defining the release flow, standard and business specification of early warning information, establishing a county, village (town), village, group and household five-level flood disaster defense responsibility system, pushing the early warning information to different regional responsibility people and masses, realizing targeted early warning and ensuring timely transmission and effective reception of the early warning information.

(2) The proposal of the prevention and control measures mainly comprises ① of perfecting non-engineering measures. Configuring video monitoring, data monitoring equipment, communication equipment, light lighting equipment, emergency materials (material quantity, storage place, use rule and responsible person) and the like, and ② establishing a reservoir day and night duty system. Implement the station responsibility system, strengthen the patrol, monitor and protection work, ③ observe the engineering change condition. The method comprises the steps of carrying out reservoir engineering safety detection condition analysis, monitoring engineering horizontal, vertical displacement, infiltration lines and the like by using instrument facilities, and observing whether a dam bank is broken by cracks, landslide, collapse leakage, piping and the like at a heavy point, reporting found problems in time, further taking effective emergency protection measures, and analyzing the harmfulness of flood events by ④. And analyzing risks of different grades corresponding to different water levels, defining downstream influence areas and disposal measures of different grades, carrying out reservoir dam break inundation risk analysis, and manufacturing a reservoir superstandard flood and dam break flood risk map. ⑤ An emergency guarantee system is established and mainly comprises organization guarantee, team guarantee and the like, and ⑥ propaganda flood disaster defending knowledge is carried out. Carrying out flood disaster professional knowledge training and technical training on responsible persons, technicians, masses and the like at all levels, and organizing and developing flood disaster avoidance exercises and the like.

2. The risk reduction plan mainly comprises flood risk regulation and control, reservoir and embankment engineering accident rescue, personnel flood prevention transfer, emergency rescue and the like, and relates to multi-department collaborative disaster prevention, disaster reduction and relief. Wherein:

(1) Flood risk regulation. The method mainly prepares flood control schemes such as reservoir flood diversion and discharge and the like by comprehensively evaluating factors such as storm flood characteristics, reservoir and embankment engineering safety conditions, possible flood disaster conditions and the like, thereby obtaining the optimal flow discharge process of the reservoir and maximizing the control of flood risks.

(2) And (5) rescue from engineering accidents. The method comprises the steps of carrying out emergency repair on engineering facilities such as reservoirs, embankments and the like damaged in a flood period, reinforcing rescue dams, and planning coping strategies such as rescue team organization, personnel configuration, task allocation, material preparation, typical dangerous case disposal schemes, information communication coordination modes and the like in advance.

(3) Personnel flood-avoiding transfer and emergency rescue. The method is characterized by comprising the steps of making a detailed transfer placement plan aiming at flood inundation areas in low-lying zones and pre-modeling, wherein the main content comprises the aspects of pool bottoms of dangerous areas, transfer time, transfer routes and placement points, transfer vehicles, medical rescue points, health rescue teams, life-saving equipment, living goods and materials and the like.

Step five, carrying out dynamic coupling of four pre-models of 'forecasting-early warning-previewing-planning', wherein:

The flood control and disaster reduction are based on 'flood' as a main line, the space-time dynamic characteristics of four water flows are mainly reflected, the prediction-early warning-previewing-plan covers the space areas of the upstream and downstream sides and the left and right sides with the reservoir as a core, and from the perspective of flood calculation, rainfall runoff, reservoir scheduling, river channel flood evolution, flood flooding existence time and space continuity are dynamically changed in time and space.

The method comprises the steps of (1) calculating a model, namely, coupling input and output conditions among four pre-models, wherein a flood forecast result can be used as input conditions of flood pre-warning, flood pre-modeling and flood planning, and sequentially determining whether pre-warning, submerged disaster risk and flood defense are generated, judging specific disaster risk grades by using the flood pre-warning as the flood pre-modeling input conditions, determining whether to start and how to start a flood prevention plan, comprehensively obtaining a flood defense plan according to flood forecast, pre-warning and pre-modeling results, and visually analyzing the flood risk regulation and disaster reduction effects of the plan through the flood pre-modeling after the flood defense plan is adopted, so that scientific decisions of the flood prevention plan are supported. (2) The model functions are that from the aspect of four pre-services of flood control and disaster reduction, the prediction is a flood control and disaster reduction basis, the early warning is a flood control and disaster reduction whistle, the previewing is a flood control and disaster reduction key, the pre-planning is a flood control and disaster reduction purpose, and the four pre-services are organically combined, coupled and mutually fed, so that the scientificalness level of flood control decisions is improved together.

The four-pre-flood-control and disaster-reduction model has a modularized link relation, four types of models are connected in a loop-to-loop manner, progressive layer by layer, circularly reciprocating and iteratively coupled, from forecasting to early warning to forecasting and planning, and through data sharing and information transmission, a full-chain, inter-correlation and full-coupling four-pre-flood-control and disaster-reduction management system is built, so that the organic integration of the four-pre-flood-reduction model of the reservoir-river-embankment multi-engineering system is realized.

1. Forecast is the basis of flood control and disaster reduction. The method utilizes real-time monitoring information such as rainfall, water level, reservoir capacity, flow and the like, and meteorological rainfall forecast and the like to predict the water level, flow and flood inundation influence in different foreseeable periods (short term, medium term and long term) through calculation of a flood forecast model, so as to carry out real-time rolling forecast on a possible flood process and a possible flood disaster, and provide important risk information for flood control early warning.

2. The early warning is a flood control disaster relief whistle. The early warning is closely dependent on real-time rolling forecast information, potential disaster risks are identified, and the early warning level is lifted or eliminated, so that early warning information is timely and accurately released, work such as engineering inspection, engineering scheduling and personnel transfer is arranged, early warning timeliness and accuracy are improved, guidance is provided for starting the pre-modeling work, and early warning information is provided for the pre-planning.

3. Previewing is a key to flood control and disaster reduction. Integrating forecast, early warning and corresponding plan information, reasonably determining reservoir flood control scheduling targets, forecast nodes, boundary conditions and the like, carrying out analog simulation on reservoir application and flood evolution conditions triggering early warning, simultaneously realizing forward and reverse two-dimensional forecast of typical historical disaster situations in a digital twin flow field play-back, obtaining flood risk situations and influences by forward forecast, obtaining reservoir safe operation limiting conditions by reverse forecast, timely finding flood control safety problems, updating forecast early warning information in real time, ensuring timeliness and accuracy of information, realizing three-dimensional visual display of disaster situations, improving intuitiveness of disaster situations, providing accurate control areas for planning of plans, and scientifically making and optimizing reservoir scheduling schemes.

4. The plan is the aim of flood control and disaster reduction. According to flood disaster prediction results under different working conditions, reservoir safety conditions, population and socioeconomic distribution conditions and the like are considered, reservoir scheduling operation, non-engineering measures, organization implementation modes, disaster handling measures, resource allocation schemes, post-disaster recovery plans and the like are determined, flood control disaster reduction plan libraries adapting to different situations are formed, prediction updating situations are fed back through the plan measures, iteration is repeated if the plans are unreasonable, visualization display of plan effects is achieved, and optimal emergency defense plans are selected, so that rationality and feasibility of the plans are ensured.

The more detailed related art explanation related to the above-described flow of the present invention is as follows:

1. With respect to the model of the SCS streamlet,

The SCS runoff model is used for small-river-basin hydrologic forecasting, namely, calculating runoff depth under a given rainfall, and is a mathematical model for estimating surface runoff. The method is characterized in that the size of the rainwater runoff of the ground surface during rainfall can be obtained by assuming two basic assumption conditions through the relationship between a water balance equation and a proportional equality assumption and a primary loss (maximum potential hold-up) relationship. The basic principle is that when rainfall, if the rainfall does not reach the initial absorption value I _a of the soil, no runoff is generated, and when the rainfall reaches I _a, the surface direct runoff Q is equal to the remainder of the total rainfall minus the actual infiltration F and the initial absorption value I _a of the soil.

(1) According to the assumption that the ratio of the actual infiltration quantity F to the possible maximum retention S at that time is equal to the ratio of the actual surface direct runoff quantity Q to the possible maximum runoff quantity P-I _a, the expression is:

Wherein P is total rainfall, Q is surface direct runoff, I _a is initial loss (initial absorption value of soil), mm comprises interception, surface water storage and the like, F is accumulated infiltration amount (actual infiltration amount) excluding I _a, and S is possible maximum retention (maximum potential retention) at that time.

(2) According to a river basin water quantity balance equation, namely that the total rainfall P is equal to the sum of the initial loss quantity (initial absorption value of soil) I _a, the actual infiltration quantity F and the surface direct runoff quantity Q of a water collecting area in a river basin, the expression is as follows:

P=I_a+F+Q (1.2)

(3) Based on the assumption that the relation between the initial loss and the maximum retention possible at that time, i.e. the initial loss I _a is proportional to the maximum retention possible at that time F, the expression is:

I_a=λS (1.3)

where λ is the initial loss coefficient, and the empirical value λ=0.2 is usually taken.

(4) The formulas (1.1) - (1.3) are combined to obtain a calculation formula of the direct surface runoff quantity Q in the SCS model:

(5) The model introduces the parameter CN for the maximum possible hold-up S for the basin at the time. From equation (1.4), Q is determined by P and S, and S is related to factors such as land utilization in the flow field, soil type, and soil wettability before precipitation. S is different in different watershed, so that the value of S is difficult to take, and a comprehensive parameter CN reflecting the characteristics of the watershed is introduced into the SCS model to obtain the value of S. The relation is as follows:

Wherein CN reflects the runoff producing capability of the underlying surface unit of the area, the influencing factors comprise the underlying surface factors such as early soil wetting degree, soil type, land utilization type, gradient, vegetation and the like, the CN value ranges from 0 to 100 according to the influencing factor information, and the smaller CN indicates the more infiltration.

2. Model for transient unit line current collection

The Nash instantaneous unit line refers to a surface runoff process line formed on the section of the outlet of the river basin through the regulation action of n linear reservoirs with the storage constant K connected in series when the total water quantity is input to be 1 and the unit surface is uniformly distributed on the river basin for purifying rain in the range of the given river basin at the moment of infinitely small duration, and can be used for representing the regulation and storage capacity of the river basin for purifying the ground rain and suitable for the confluence calculation of the surface runoffs of the small and medium river basins which lack data. The mathematical expression is as follows:

wherein u (0, t) is the ordinate of an instantaneous unit line at the moment t, wherein 0 represents that the net rain duration tends to be infinitesimal, t represents that at any moment, K is the storage constant of a linear reservoir, is equivalent to the parameter of the confluence time of a river basin, Γ (n) is a gamma function about n, n is the number of the linear reservoirs, is equivalent to the adjustment times, reflects the energy-storage capacity of the river basin, and e is the natural logarithmic base.

3. Model for Ma Sijing flood evolutions

The river flood calculation method for predicting the downstream flow of the river by the upstream flow of the river has simple calculation and low data requirement. The Ma Sijing models adopt a water balance equation and a tank storage equation to replace a complex hydrodynamic equation, and the simplified equation is as follows:

Wherein W is the tank accumulation, t is time, I is the flow of the inflow section, Q is the flow of the outflow section, K is the tank accumulation coefficient, and x is the flow specific gravity factor.

4. LSTM model for flood forecast

The LSTM model is a variant of the recurrent neural network model, each module of which is composed of a forgetting gate f _t, an updating gate u _t, an output gate o _t, 3 gating units and 1 cell state unitComposition is prepared. The LSTM model has stronger memory capacity, and is suitable for processing long-time sequence data and solving the long-term dependence problem. Each time step has 3 inputs, namely an external input x _t at the time t, an LSTM unit output h _t-1 at the time t-1 and a cell state C _t -1 at the time t-1. The input data is processed by the LSTM gating unit to obtain an output h _t and a cell state C _t.

The output calculation process and formula of the gate control unit are as follows:

Wherein f _t is a forget gate, u _t is an update gate, For the cell state, o _t is the output gate, σ is the activation equation, sigmoid function, tanh is hyperbolic tangent function, h _t-1 is the t-1 time LSTM unit output, h _t is the t time LSTM unit output, x _t is the t time external input, C _t-1 is the t-1 time cell state, C _t is the t time cell state, W _f、W_u、W_c、W_o and b _f、b_u、b_c、b_o represent the forgetting gate, the updating gate, the cell state and the weight and bias matrix vector of the output gate respectively, and the initial values are randomly set for the weight and bias matrix and are obtained through data training. The internal structure of the unit is shown in fig. 2. As can be seen from fig. 2, the LSTM cell output h _t-1 at the previous time and the current state variable x _t are input as input quantities to the forget gate, the input gate and the output gate in parallel to obtain candidate values. The forgetting gate is mainly responsible for discarding noise information of C _t-1, retaining key information, selectively storing information of critical state C _t by the input gate so as to update the memory unit C _t, jointly obtaining output quantity h _t at the current moment by the output gate and the updated memory unit C _t, and obtaining a better LSTM model by reducing errors through multiple iterations.

5. Reservoir flood control scheduling model

The flood control dispatching model mainly comprises a highest water level change process and a lower drainage flow change process in front of the reservoir dam. The model mainly considers flood control and drainage, and needs to be operated by combining a specific application mode of reservoir flood control scheduling, a water level storage capacity relation and building parameters.

(1) Objective function:

And the maximum peak clipping principle is adopted, the safety of the dam and the flood prevention disaster in the downstream area of the reservoir are comprehensively considered, and a minimum model of the maximum downward leakage flow is established by taking the minimum of the flood peak flow of the downstream flood control section as a main target.

Wherein: For the maximum discharging flow, q _t is the outlet flow of the reservoir in the period T, I _t is the inflow flow of the reservoir in the period downstream of the flood control section in the period T, T is the length of the scheduling period, and T is the number of the scheduling period.

(2) Constraint conditions:

1) The expression of the water balance constraint is as follows:

Wherein V _t-1 and V _t are the initial and final water storage amounts of the t period respectively, Q _t-1 and Q _t are the initial and final reservoir flow amounts of the t period respectively, Q _t-1 and Q _t are the initial and final reservoir outlet flow amounts of the reservoir of the t period respectively, and Δt is the calculated period length.

2) The expression of the reservoir level constraint is as follows:

Z_min≤Z_t≤Z_max (5.3)

Wherein Z _t、Z_min、Z_max is the reservoir water level, the minimum and maximum allowable water level of the period t.

3) The expression of the reservoir discharge capacity constraint is as follows:

Q_min≤Q_t≤Q_max(Zt) (5.4)

Wherein Q _t is the discharge flow rate of the period t, Q _max(Zt) is the maximum discharge flow rate corresponding to Z _t, and Q _min is the minimum discharge flow rate of the reservoir.

4) The expression of the relation between the water storage level and the reservoir capacity is as follows:

wherein: the reservoir water level corresponding to the reservoir capacity of the reservoir in the period t.

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

|Q_out(t)-Q_out(t-1)|≤Δq (5.6)

wherein deltaq is the maximum allowable amplitude of reservoir delivery flow in adjacent time periods.

6) Boundary condition constraint for reservoir capacity

V₀＝V_b;V_T＝V_e (5.7)

V_min≤V_t≤V_max (5.8)

Wherein V _b and V _e are respectively the reservoir capacity corresponding to the water level of the reservoir at the beginning of the dispatching period and the reservoir capacity corresponding to the water level to be fallen back at the end of the dispatching period, and V _max and V _min respectively represent the upper limit and the lower limit of the water storage capacity of the reservoir.

7) The expression of the flood discharge capacity constraint of the flood discharge facility is as follows:

0≤q_i,t≤q_i(Z_t,B_t)≤q_i,max (5.9)

Wherein q _i,t is the discharge flow of the flood discharge facility i, m ³/s,g_i,max is the maximum allowable discharge flow of the flood discharge facility i, m ³/s,q_i(Z_t,B_t) is the maximum discharging capacity of the flood discharge facility i when the reservoir water storage level is Z _t and the facility opening is B _t, Z _t is the reservoir water storage level at the moment t, B _t is the operation mode of the water discharge building, and the corresponding gate opening is selected according to the corresponding relation between the facility opening and the discharging capacity.

8) Ma Sijing flow calculation constraints

Wherein Q _k-1 (t) and Q _k (t) are respectively the outflow of the kth Ma Sijing calculated river reach at the beginning and the end of the nth period, Q _k-1 (t-1) and Q _k-1 (t) are respectively the outflow of the kth-1 Ma Sijing calculated river reach at the beginning and the end of the nth period,Calculating the interval inflow of the river reach for the kth Ma Sijing,The calculation parameters of the Ma Sijing th calculation river reach are respectively calculated.

9) Non-negative constraints, all variables in the calculation process have no negative value.

6. Reservoir downstream river course flood evolution and two-bank inundation risk simulation model

The reservoir downstream river course flood evolution and two-bank submerged risk simulation model adopts a river course one-dimensional hydrodynamic model, a plane two-dimensional hydrodynamic model and a two-dimensional coupling model.

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

the basic equations of the one-dimensional model are the san-valance equation set, including the continuous equation (law of conservation of mass) and the momentum equation (newton's second law), which equations are as follows:

wherein Q is flow, Q is lateral inflow, A is water passing area, h is water level, R is hydraulic radius, C is thank coefficient, and alpha is momentum correction coefficient.

(2) Two-dimensional planar free surface flow model principle:

the two-dimensional unsteady flow calculation basis equations include a continuity equation and a momentum equation as follows:

continuity equation:

Momentum equation:

wherein: for flow velocity based on water depth average, t is time, x, y and z are Cartesian coordinates, eta is river bottom elevation, d is still water depth, h=eta+d is total head, u, v are velocity components in x, y directions, g is gravity acceleration, ρ is water density, S _xx、s_xy、s_yx、s_yy is radiation stress component, p _a is atmospheric pressure, ρ ₀ is relative density of water, S is point source flow size, u _s、v_s is flow velocity of source sink item water flow.

7. Monitoring station network planning design principle for flood control and disaster reduction forecast and early warning requirements

Regarding the monitoring station network planning of a reservoir-river channel-embankment system, sky ground monitoring requirements should be comprehensively considered, and a multi-mode, multi-layer and integrated monitoring station network system consisting of a meteorological satellite, a rain measuring radar, a rainfall station, a hydrologic station, a water level station and the like is constructed, and the invention provides the following principles:

(1) The hydrologic station network is that a reservoir is provided with a main dry tributary warehouse-in and warehouse-out hydrologic station, a mountain torrent disaster easy-to-develop area is provided with a river of flood control targets such as medium and small towns, villages, industrial and mining enterprises, important infrastructures and the like at the downstream, and the hydrologic station is arranged at the mountain outlet or at the upstream of the flood control targets.

(2) The water level station network is that a representative water level station is arranged in front of a reservoir dam, the water level station is arranged near the river mouth of a branch flow of a warehouse (lake) and along the narrow or widened position of the water level of a water course, the water level station is arranged at the tail and middle section of a reservoir area affected by changed backwater, and the water level station is arranged on a river with flood disaster threat to protection targets such as urban residential areas, industrial and mining enterprises, important infrastructure and the like.

(3) The rainfall station network is uniformly distributed in a planning range, the average single station area is not more than 100km ², rainfall stations are distributed near the gravity center of a river basin, rainfall stations are distributed in a region with concentrated source areas of storm flood and important functions for flood prevention forecast, early warning and the like, the rainfall stations are distributed in a reservoir control river basin range, and at least 1 rainfall station is distributed in a small reservoir in an encryption mode.

(4) And the rain measuring radar station is arranged in a storm flood concentrated source area and a mountain torrent disaster incident area and is deployed in a networking way.

(5) The video station is a reservoir distribution video monitoring station, wherein the distribution of communication conditions is not less than 2 video monitoring stations, and the video monitoring stations are preferably distributed along the river channel area at the downstream of the reservoir according to the flooding danger degree.

In summary, the invention adopts technical means such as theoretical analysis, numerical simulation and the like, and respectively constructs 'forecast', 'early warning', 'previewing', 'planning' model and 'planning' model of the reservoir-river channel-embankment multi-engineering system through multi-disciplinary intersection such as hydrology, hydraulics, calamity, artificial intelligence and the like, and defines a dynamic association mechanism among four pre-models to realize high-precision forecasting and early warning of flood, dynamic previewing of flood risk and scientific preparation of flood control plans.

Modifications and variations will occur to those skilled in the art in light of the foregoing disclosure. Any modification, improvement and variation made on the basis of the present invention shall fall within the scope of the present invention.

Claims

1. A multi-working-procedure flood control and disaster reduction method based on a four-pre-coupling model is characterized by comprising the following steps:

The method comprises the steps of constructing a flood control and disaster reduction early warning model of a reservoir-river channel-embankment multi-engineering system, wherein the early warning model is a model system formed by multi-dimensional deep coupling of a meteorological risk early warning model, a reservoir risk early warning model, an under-dam river channel embankment risk early warning model and a reservoir downstream resident inundation risk early warning model; the weather risk early warning model takes regional rainfall observation and 24-hour forecast rainfall data as input and takes early warning results of weather risk grades as output; the reservoir risk early-warning model further comprises a rainfall station monitoring early-warning sub-model, a storage flood early-warning sub-model, a reservoir area flood risk early-warning sub-model and a dam engineering safety early-warning sub-model, wherein the rainfall station monitoring early-warning sub-model takes actual measurement rainfall data as input and takes a rainfall triggering rainfall early-warning response judgment result as output, the reservoir area flood risk early-warning sub-model takes all-weather reservoir area water level monitoring data as input and takes a flood risk prediction result as output, the dam engineering safety early-warning sub-model comprises flood dam risk early-warning processing and dam body instability early-warning processing, specifically, the flood risk early-warning processing takes actual measurement water level monitoring data of the storage flood and the reservoir, the reservoir level development trend is predicted to be output 1 of the sub-model by combining a water level storage capacity relation and reservoir drainage capacity input 1, the dam body instability processing takes dam body water content, crack development degree, infiltration line height, soil mass deformation and bank balance safety coefficient dam body safety coefficient safety condition data as input 2 and takes a real-time dam structure safety risk monitoring result as output 2 of the dam, and the reservoir risk early-warning model further comprises a river channel risk early-warning sub-warning model, and the reservoir risk early-warning model comprises the rain warning sub-model, the system comprises a reservoir downstream resident inundation risk early warning model, a reservoir flood control scheduling sub-model, a river flood risk simulation early warning sub-model, a embankment engineering safety early warning sub-model, a simulation early warning sub-model, a reservoir downstream resident inundation risk resident early warning model, a reservoir downstream resident inundation risk early warning model, a disaster assessment and rescue operation providing data support and decision basis, wherein the reservoir flood risk simulation early warning sub-model takes rainfall monitoring data as input and takes rainfall early warning response of triggering corresponding grades as output, takes model calculation boundary conditions output by the reservoir engineering flood control scheduling sub-model as input, and takes downstream river flood risk early warning response results as output, and takes basic data at least comprising the soil moisture content of the embankment, crack development degree, the infiltration line height, the soil deformation and the bank slope balance safety coefficient as input, takes the output of the model, and the reservoir downstream resident inundation risk early warning model further comprises a monitoring early warning sub-model and a simulation early warning sub-model, takes the dynamic monitoring image of a possible inundation area as input of the sub-model, takes disaster assessment and rescue operation as output;

2. The multi-process flood control and mitigation method based on the four pre-coupling model of claim 1, wherein the reservoir upstream area flood forecast model further comprises a physical mechanism flood forecast sub-model and a data driven flood forecast sub-model.

3. The multi-process flood control and mitigation method based on a four pre-coupling model according to claim 2, wherein the physical mechanism flood forecast sub-model is a combined model consisting of an SCS production flow sub-model, an instant unit line sink sub-model and Ma Sijing flood evolution sub-models coupled in series.

4. The multi-process flood control and disaster reduction method based on the four pre-coupling model according to claim 2, wherein the data-driven flood forecast sub-model adopts an LSTM model, wherein the input of the flood forecast LSTM model is a rainfall process of the first n hours and a runoff process, the output of the flood forecast LSTM model is a rainfall process of the next m hours, the correlation information which is not important with the flood forecast is abandoned by a forgetting gate, the retention degree of the information is controlled, the neural network model is updated to be new important information by an updating gate to generate candidate values, the important information which needs to be output is determined by an output gate to generate a hidden state for predicting or serving as the input of the next time step.