A hydrological and hydrodynamic coupling driven reservoir irrigation area algal bloom prediction and prevention method and system

Abstract

This invention provides a method and system for predicting and controlling algal blooms in reservoir irrigation areas driven by hydrological and hydrodynamic coupling, belonging to the field of water ecological protection technology in water conservancy projects. This invention collects and standardizes basic data of all elements to construct a two-way coupled model of a non-point source pollution confluence model in the irrigation area and a two-dimensional hydrodynamic-water quality model of the reservoir. This model simulates the nitrogen and phosphorus loads of irrigation runoff, the flow field of the reservoir area, and the distribution of nutrients under different scheduling and irrigation conditions. It integrates the coupled simulation results with water quality and meteorological data, and constructs a prediction model based on LSTM to achieve 7-15 day algal bloom risk prediction. Based on the prediction results, a risk index is constructed for graded early warning, and the weights are dynamically adjusted through multi-objective optimization to automatically generate a scheduling-control scheme adapted to water conservancy engineering rules. It utilizes incremental learning and iterative optimization based on real-time monitoring data. This invention forms a closed loop of "prediction-early warning-control-scheduling," taking into account algal bloom control and water supply, flood control, and irrigation safety.

Description

Technical Field

[0001] This invention relates to the field of water ecological protection technology in water conservancy projects, specifically to a method and system for predicting and controlling algal blooms in reservoir irrigation areas driven by hydrological and hydrodynamic coupling. Background Technology

[0002] Reservoirs, as core engineering facilities for water resource regulation in my country, undertake multiple functions including urban and rural water supply, agricultural irrigation, flood control and disaster reduction, and ecological water replenishment. Irrigation districts, as core water-using units downstream of reservoirs, form a strongly coupled system with reservoirs in terms of hydrological connectivity, hydraulic linkages, and water quality impacts. In recent years, affected by multiple factors such as climate change, non-point source pollution input from farmland runoff in irrigation districts, and reservoir water conservancy scheduling methods, many reservoirs in my country have experienced prominent eutrophication problems and frequent algal blooms. These not only seriously threaten the safety of drinking water sources in reservoirs but also spread with downstream water flow to irrigation district water conveyance systems and downstream river channels, damaging the health of the watershed's aquatic ecosystem and becoming a core bottleneck restricting the ecological operation of water conservancy projects.

[0003] Existing algal bloom prediction and control technologies have the following significant shortcomings: 1. The prediction model is out of touch with the actual water conservancy scenario: it only takes a single reservoir area as the object and does not consider key factors such as irrigation area drainage and reservoir scheduling. The prediction lead time is only 3-5 days, and the accuracy is low. 2. Severe separation between forecasting and prevention: Prevention measures are not linked to the operation rules of water conservancy projects, which may easily conflict with water supply, irrigation and flood control safety; 3. The governance model is too emergency-oriented: it relies on chemical disinfection and mechanical retrieval, which are costly, prone to secondary pollution, and lack source control and long-term mechanisms.

[0004] Therefore, this invention provides a method and system for predicting and controlling algal blooms in reservoir irrigation areas driven by hydrological-hydraulic coupling, realizing an integrated closed-loop operation of advanced and accurate early warning and scheduling control. Summary of the Invention

[0005] The purpose of this invention is to overcome at least one technical problem existing in the prior art and to provide a method and system for predicting and controlling algal blooms in reservoir irrigation areas driven by hydrological and hydrodynamic coupling.

[0006] On one hand, embodiments of the present invention provide a method for predicting and controlling algal blooms in reservoir irrigation areas driven by hydrological-hydraulic coupling. The method includes: Step S1, collecting long-sequence hydrological data, irrigation and farmland runoff data, water quality data, meteorological data, and historical algal bloom data of the target reservoir-irrigation area system, completing outlier removal, missing value completion, and dimensionless standardization processing to construct a standardized dataset; Step S2, based on the standardized dataset, constructing a non-point source pollution confluence model for the irrigation area and a two-dimensional hydrodynamic-water quality model for the reservoir area, respectively, achieving bidirectional coupling through spatiotemporal scale matching and boundary condition interaction, generating hydrological-hydraulic coupling simulation results under different water conservancy scheduling conditions and irrigation cycles, the simulation results including the nitrogen and phosphorus load input process of irrigation area runoff, reservoir water flow field, and spatiotemporal distribution data of nutrients; Step S3, using the hydrological-hydraulic coupling simulation results, water quality data, and meteorological data from the standardized dataset as input feature sets, and using historical algal density and algal bloom events as output labels, constructing a time-series prediction model based on a long short-term memory network and performing training, verification, and calibration. Determine the critical threshold for algal bloom under different operating conditions and output the algal bloom risk prediction result for the next N days; Step S4: Based on the predicted values ​​of chlorophyll a concentration, algal bloom probability, and total nitrogen and total phosphorus concentrations in the risk prediction result, construct an algal bloom risk comprehensive index, classify the risk into three levels (low, medium, and high) and corresponding to different warning levels, and simultaneously push the warning information and risk spatial distribution to the reservoir and irrigation area management units, while transmitting the risk level as a trigger signal to step S5; Step S5: Based on the risk level transmitted in step S4... With algal bloom control, water supply security, irrigation guarantee, and flood control safety as the core objectives, a multi-objective optimized water conservancy scheduling model is constructed. The weights of each objective are dynamically adjusted, and a hierarchical linkage prevention and control scheme adapted to the operation rules of reservoir-irrigation area water conservancy projects is automatically generated and the scheme is issued for execution. Step S6: Real-time collection of reservoir hydrological, water quality, and algal density data after the implementation of the hierarchical linkage prevention and control scheme described in step S5. Based on the incremental learning method, the parameters of the time series prediction model and the coupled model are self-learned and iteratively optimized to continuously improve the accuracy of risk prediction and the adaptability of the prevention and control scheme.

[0007] Further, step S2 includes: Step S21, constructing a non-point source pollution runoff model for the irrigation area, including: based on the SWAT model, using hydrological response units as the basic calculation units, inputting the standardized dataset, calculating the surface irrigation runoff using the SCS-CN model, calculating the total nitrogen load and total phosphorus load using dissolved nitrogen load formula and adsorbed phosphorus load formula respectively, and outputting the daily irrigation area runoff flow, daily irrigation area runoff total nitrogen load, and daily irrigation area runoff total phosphorus load after calibration; Step S22, constructing a two-dimensional hydrodynamic-water quality model for the reservoir area, including: based on the MIKE21 model, discretizing the reservoir area into an unstructured grid, using the two-dimensional shallow water equation as the hydrodynamic control equation and coupling it with a convection-diffusion water quality module, inputting the standardized dataset, and outputting the model after calibration. The daily water depth, flow velocity, water level, and concentrations of total nitrogen, total phosphorus, and chlorophyll a in each grid are output, along with the daily outflow rate. Step S23, bidirectional coupling, includes: using a preset time step, the daily outflow rate, total nitrogen load, and total phosphorus load of the irrigation area outflow from the irrigation area non-point source pollution runoff model are used as the inflow boundary conditions of the reservoir area two-dimensional hydrodynamic-water quality model, and converted into the inflow rate and pollutant concentration at the reservoir inlet according to the coupling interface formula; simultaneously, the daily outflow rate output by the reservoir model is used as the upper limit of available water for irrigation water intake in the irrigation area model for the next day, updating the irrigation satisfaction rate of the irrigation area and recalculating the outflow process, thereby achieving bidirectional coupling to simulate the hydrological-hydrodynamic coupling simulation results under different water conservancy scheduling conditions and irrigation cycles.

[0008] Furthermore, it is characterized by, The formula for dissolved nitrogen loading is: ; The formula for the adsorbed phosphorus loading is: ; In the formula, Dissolved nitrogen load from surface runoff, in kg; Surface irrigation runoff volume, unit: ; This represents the concentration of dissolved nitrogen in the runoff, expressed in mg / L. The nitrogen transport loss coefficient in the ditch; Phosphorus loading in sediment adsorbed state, in kg; This refers to the sediment transport volume of the irrigation area's runoff, expressed in tons (t). The phosphorus content in the sediment is expressed in mg / kg. This represents the phosphorus sediment transport ratio.

[0009] Furthermore, in the two-dimensional hydrodynamic-water quality model of the reservoir area, the continuity equation of the two-dimensional shallow water equation is: ; The momentum equation in the x-direction is: ; The momentum equation in the y-direction is: ; In the formula, h is the water depth, in meters (m). The water level is represented by meters (m); u and v are the velocity components in the x and y directions, respectively, in meters per second (m / s); t is time, in seconds (s); and g is the acceleration due to gravity, in cubic meters per second (m / s). ; This is the reference density for water, in units of... ; This represents the actual density of the water, in units of... ; , The wind stress component on the water surface is expressed in units of 1000 ppm. ; , The shear stress component of the bed surface is expressed in units of 1000 kJ / m². ; , , and is the viscous stress tensor component; f is the Coriolis force coefficient, in units of . S represents source and sink terms, including irrigation district runoff inflow, reservoir outflow, and intake flow rate, in units of... ; , Let x be the momentum source and sink terms in the x and y directions.

[0010] Furthermore, the coupling interface formula in the bidirectional coupling is: ; ; In the formula, Let be the inflow rate at the k-th inlet of the reservoir at time t, in units of . ; Let be the flow rate of the irrigation area's runoff into the reservoir at time t, in units of . ; Let be the nitrogen / phosphorus pollutant concentration at the k-th inlet at time t, in units of . ; This represents the nitrogen / phosphorus load output from the irrigation district at time t, in units of... .

[0011] Furthermore, in step S3, the core computational unit formula of the Long Short-Term Memory network includes: Forgotten Gate: ; Input Gate: ; Candidate cell status: ; Cell status update: ; Output gate: ; Hidden layer output: ; In the formula, , , These are the output vectors of the forget gate, input gate, and output gate, respectively. Let t be the cell state vector at time t; Let t be the hidden layer output vector; Let be the input feature vector at time t; , , , These are the weight matrices for each network layer; , , , For the corresponding bias term; sigmoid is the activation function; tanh is the hyperbolic tangent activation function; For Hadama accumulation.

[0012] Furthermore, in step S3, for different water conservancy scheduling conditions and / or irrigation cycles, the corresponding critical threshold for algal blooms is determined using ROC curves and the Youden index maximization method; the Youden index is calculated according to the following formula: J = Sensitivity + Specificity - 1; In the formula, sensitivity is the proportion of algal blooms correctly predicted by the time series prediction model, and specificity is the proportion of algal blooms not predicted by the time series prediction model. The chlorophyll a concentration, algal bloom probability, total nitrogen concentration, and total phosphorus concentration corresponding to the maximum value of the Youden index are taken as the critical threshold for algal blooms under this operating condition or cycle. The calibration performance of the time series forecasting model is evaluated using the Nash efficiency coefficient (NSE), calculated as follows: ; In the formula, This is the measured value of algal density. These are simulated values ​​from a time series prediction model. Let n be the arithmetic mean of the measured value sequence, and n be the sample size.

[0013] Furthermore, in step S4, the formula for calculating the Algal Bloom Risk Composite Index (ARI) is as follows: ; In the formula, This is the predicted value for chlorophyll a concentration. The critical threshold for chlorophyll a burst; This is a predicted probability value for algal blooms. This represents the critical threshold for the probability of an outbreak. , These are predicted values ​​for total phosphorus and total nitrogen concentrations. , These are the critical thresholds for total phosphorus and total nitrogen; , , , The weights of each indicator are determined using the Analytic Hierarchy Process (AHP), and satisfy the following conditions: ; The risk level classification criteria are as follows: Low risk: , ; Medium risk: , ; High risk: , ; in, , .

[0014] Furthermore, in step S5, the comprehensive objective function of the multi-objective optimization water conservancy scheduling model is: ; In the formula, F is the comprehensive objective function; To achieve the goal of algal bloom control, the core indicators are maximizing the average flow velocity and minimizing the hydraulic residence time in the reservoir area. To ensure water supply security, the core indicator is maximizing the utilization rate of reservoir capacity. To ensure irrigation, the core indicator is to maximize the rate at which irrigation water demand is met in the irrigation district. To achieve the goal of flood control safety, the core indicator is to maximize the compliance rate of reservoir water level and outflow. , , , The weights of each target are dynamically adjusted according to the warning level, and the following conditions are met: ; The model constraints include reservoir water balance constraints, reservoir water level constraints, outflow constraints, and irrigation district constraints. The formula for the reservoir water balance constraint is: ; In the formula, , The reservoir capacity is given by the units t and t+1. ; The average inbound flow rate during the time period, in units of ; The average discharge flow rate over the period is expressed in units of... ; Evaporation loss flow rate in the reservoir area during the specified time period, in units of ; The unit for calculating the duration is seconds (s). Furthermore, the weights of each target are dynamically adjusted according to the warning level, and a hierarchical linkage prevention and control plan adapted to the operation rules of reservoir-irrigation area water conservancy projects is automatically generated.

[0015] Secondly, embodiments of the present invention provide a reservoir irrigation area algal bloom prediction and control system driven by hydrological-hydraulic coupling. The system is implemented using the aforementioned hydrological-hydraulic coupling-driven method for predicting and controlling algal blooms in reservoir irrigation areas. The system includes: a basic data acquisition and standardization module, suitable for collecting long-sequence hydrological data, irrigation and farmland drainage data, water quality data, meteorological data, and historical algal bloom outbreak data of the target reservoir-irrigation area system, completing outlier removal, missing value completion, and dimensionless standardization processing to construct a standardized dataset; and a reservoir-irrigation area hydrological-hydraulic coupling model construction module, suitable for building models based on the standardized dataset. A non-point source pollution runoff model for irrigation districts and a two-dimensional hydrodynamic-water quality model for reservoir areas are bidirectionally coupled through spatiotemporal scale matching and boundary condition interaction. This generates hydrological-hydrodynamic coupling simulation results under different water conservancy scheduling conditions and irrigation cycles. The simulation results include nitrogen and phosphorus load input processes from irrigation district runoff, water flow field data in the reservoir area, and spatiotemporal distribution data of nutrients. A module for constructing an algal bloom risk prediction model is also included. This module is suitable for using the hydrological-hydrodynamic coupling simulation results along with water quality and meteorological data from a standardized dataset as input feature sets, and historical algal density and algal bloom events as output labels. A time-series prediction model is constructed based on a long short-term memory network and then trained, validated, and... The calibration module determines the critical threshold for algal blooms under different operating conditions and outputs the algal bloom risk prediction results for the next N days. The risk classification, early warning, and threshold triggering module is suitable for constructing a comprehensive algal bloom risk index based on the predicted chlorophyll a concentration, algal bloom probability, and total nitrogen and total phosphorus concentrations from the risk prediction results. This index classifies risks into low, medium, and high levels with corresponding early warning levels, and synchronously pushes the early warning information and risk spatial distribution to reservoir and irrigation district management units. Simultaneously, the risk level is used as a trigger signal to transmit the water conservancy scheduling-ecological prevention and control linkage scheme generation module. The water conservancy scheduling-ecological prevention and control linkage scheme generation module is suitable for generating algal bloom risk based on the risk classification and early warning results. The risk level transmitted by the warning and threshold triggering module, with algal bloom control, water supply safety, irrigation guarantee, and flood control safety as core objectives, constructs a multi-objective optimized water conservancy scheduling model, dynamically adjusts the weights of each objective, automatically generates a graded linkage prevention and control scheme adapted to the operation rules of reservoir-irrigation area water conservancy projects, and issues the scheme for execution; the scheme iteration and effect feedback module is suitable for real-time collection of reservoir area hydrological, water quality, and algal density data after the implementation of the graded linkage prevention and control scheme generated by the water conservancy scheduling-ecological prevention and control linkage scheme generation module, and completes parameter self-learning and iterative optimization of the time series prediction model and the coupled model based on the incremental learning method, continuously improving the accuracy of risk prediction and the adaptability of prevention and control schemes.

[0016] Thirdly, embodiments of the present invention also provide an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program, when executed by the processor, implements the above-described method for predicting and controlling algal blooms in reservoir irrigation areas driven by hydrological and hydrodynamic coupling.

[0017] Fourthly, embodiments of the present invention also provide a readable storage medium, which, when the instructions in the storage medium are executed by the processor of an electronic device, enables the electronic device to execute the above-described hydrological and hydrodynamic coupled-driven method for predicting and controlling algal blooms in reservoir irrigation areas.

[0018] The advantages of this invention compared to the prior art are: 1. Advanced Prediction: For the first time, all elements of reservoir-irrigation area water management are incorporated into the algal bloom prediction model. A two-way coupling model of non-point source pollution runoff in the irrigation area and hydrodynamics-water quality in the reservoir area is constructed, accurately analyzing the driving and regulatory mechanisms of water management and irrigation runoff on algal bloom outbreaks in the reservoir area. The lead time for algal bloom risk prediction is increased from 3-5 days to 7-15 days. Compared with traditional prediction models that only use meteorological and water quality data, the prediction accuracy is improved by more than 40%, and the location error of the outbreak area is ≤5%.

[0019] 2. Closed-loop linkage: A closed-loop linkage system covering the entire process of "prediction-early warning-prevention-water conservancy scheduling" has been constructed. A graded prevention and control plan that is fully adapted to the operation rules of water conservancy projects is generated for different risk levels. Through a multi-objective optimization scheduling model, the system achieves synergy between algal bloom prevention and control and the core functions of water supply, flood control and irrigation. It is fully adapted to the actual management needs of water conservancy management units under the river and lake chief system and solves the industry problem of the disconnect between existing technology prediction and prevention and control.

[0020] 3. Ecological long-term effectiveness: With optimized water conservancy scheduling as the core, combined with ecological prevention and control measures, the nutrient base and hydrodynamic conditions for algal blooms are reduced from the source, replacing the traditional emergency chemical disinfection and mechanical dredging treatment model. This significantly reduces the cost of algal bloom control, while avoiding the risk of secondary pollution, and achieving the dual goals of long-term protection of the water ecology of the reservoir-irrigation area system and sustainable operation of water conservancy projects. Attached Figure Description

[0021] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0022] Figure 1 This is a flowchart of a method for predicting and controlling algal blooms in reservoir irrigation areas driven by hydrological and hydrodynamic coupling, provided in Embodiment 1 of the present invention.

[0023] Figure 2 This is a logic diagram of a method for predicting and controlling algal blooms in reservoir irrigation areas driven by hydrological and hydrodynamic coupling, provided in Embodiment 1 of the present invention.

[0024] Figure 3 This is a schematic diagram of a hydrological and hydrodynamic coupled-driven reservoir irrigation area algal bloom prediction and control system provided in Embodiment 2 of the present invention.

[0025] Figure 4 This is a partial block diagram of the electronic device provided in Embodiment 3 of the present invention. Detailed Implementation

[0026] Before discussing the exemplary embodiments in more detail, it should be noted that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the operations as sequential processes, many of these operations can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but may also have additional steps not included in the figures. The process can correspond to a method, function, procedure, subroutine, subroutine, etc.

[0027] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0028] The present invention will now be described in detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0029] For ease of understanding, the following technical terms are explained here: 1. SWAT Model: SWAT (Soil and Water Assessment Tool) is a watershed-scale distributed hydrological model based on physical mechanisms, developed by the USDA Agricultural Research Service. This model can simulate surface runoff, soil erosion, nutrient cycling, and pesticide transport processes under different land uses, soil types, and agricultural management practices. In this invention, the SWAT model is used to construct a non-point source pollution runoff model for irrigation districts, calculating the runoff volume of irrigation runoff and the total nitrogen and total phosphorus loads carried by the runoff.

[0030] 2. Hydrological Response Unit (HRU): An HRU is the smallest computational unit in the SWAT model, referring to an area with the same combination of land use type, soil type, and slope. Hydrological processes and pollutant transport processes within each HRU are considered homogeneous. By dividing the irrigation district into several HRUs, the model can reflect the impact of spatial heterogeneity within the irrigation district on runoff and pollution generation.

[0031] 3. SCS-CN Model: (Soil Conservation Service Curve Number) is an empirical method proposed by the U.S. Department of Agriculture's Soil Conservation Service for calculating surface runoff. This model determines a dimensionless curve number (CN value) by comprehensively considering factors such as soil type, land use, and previous soil moisture, and then calculates the depth of surface runoff generated by rainfall (or irrigation). In this invention, the SCS-CN model is used to calculate surface irrigation runoff.

[0032] 4. MIKE21 Model: A professional two-dimensional hydrodynamic-water quality simulation software developed by DHI, Denmark. This software is widely used for hydrodynamic, wave, sediment transport, and water quality simulation of rivers, lakes, reservoirs, estuaries, and coastal zones. In this invention, the MIKE21 model is used to construct a two-dimensional hydrodynamic-water quality model of a reservoir area, simulating the velocity field, water level changes, and spatiotemporal distribution of water quality indicators such as total nitrogen, total phosphorus, and chlorophyll a.

[0033] 5. Two-Dimensional Shallow Water Equations: These are a set of partial differential equations describing the motion of shallow water bodies (where the depth is less than the horizontal scale). They consist of a continuity equation and momentum equations in the x and y directions. Derived based on the hydrostatic pressure assumption, these equations can simulate changes in water level and flow velocity across the water surface. In this invention, the two-dimensional shallow water equations are the governing equations of the hydrodynamic module in the MIKE21 model.

[0034] 6. Convection-Diffusion Water Quality Module: This is a numerical calculation module in the hydrodynamic-water quality model used to simulate the transport and diffusion of dissolved substances in water bodies. Convection refers to the overall movement of substances with the water flow, while diffusion refers to the dispersion of substances from high-concentration areas to low-concentration areas. This module typically also includes biochemical reaction terms, such as degradation, algal growth, and nutrient cycling. In this invention, the convection-diffusion water quality module is used to simulate the transport and transformation processes of total nitrogen, total phosphorus, and chlorophyll a in the reservoir, and includes algal growth kinetics.

[0035] 7. Long Short-Term Memory (LSTM): LSTM is a special type of recurrent neural network (RNN) specifically designed for processing time-series data. By introducing forget gates, input gates, output gates, and cell states, LSTM effectively solves the gradient vanishing or exploding problems of traditional RNNs, thereby learning long-term dependencies. In this invention, LSTM is used to construct an algal bloom risk prediction model. Using historical hydrological, water quality, meteorological, and hydrodynamic coupled simulation results as input, it predicts the chlorophyll a concentration, algal bloom probability, and total nitrogen and total phosphorus concentrations for the next N days.

[0036] 8. Youden Index: This is a comprehensive index for evaluating the accuracy of binary diagnostic tests, calculated as J = Sensitivity + Specificity − 1. Sensitivity (true positive rate) refers to the proportion of algal blooms correctly predicted by the model, while specificity (true negative rate) refers to the proportion of algal blooms correctly predicted by the model that did not occur. The Youden Index ranges from [0,1], with a higher value indicating a stronger overall discriminative ability of the model at that threshold. In this invention, the optimal critical threshold for algal blooms is determined using the Youden Index maximization method.

[0037] 9. Nash Efficiency Coefficient: The Nash-Sutcliffe Efficiency Coefficient (NSE) is a dimensionless index commonly used to evaluate the simulation performance of hydrological models. When NSE = 1, it indicates that the simulated value is completely consistent with the measured value; NSE ≥ 0.75 is generally considered to indicate good model simulation performance. In this invention, NSE is used to evaluate the calibration and validation performance of irrigation district models, reservoir models, and algal bloom prediction models.

[0038] 10. Analytic Hierarchy Process (AHP): A multi-objective decision analysis method combining qualitative and quantitative approaches, proposed by American operations researcher Satie. This method decomposes complex problems into several levels and factors, determining the relative importance weights of each factor based on pairwise comparisons. In this invention, AHP is used to determine the weight coefficients of each indicator (chlorophyll a concentration, algal bloom probability, total phosphorus concentration, and total nitrogen concentration) in the Algal Bloom Risk Index (ARI).

[0039] 11. Multi-objective optimization: Multi-objective optimization refers to a mathematical programming problem that seeks a set of decision variables to maximize the effectiveness of all objectives under multiple potentially conflicting objective functions. Since there are often trade-offs between the objectives, the solution to a multi-objective optimization problem is typically a Pareto optimal solution set. In this invention, the multi-objective optimization water management model simultaneously pursues four objectives: algal bloom control, water supply security, irrigation guarantee, and flood control security, and transforms it into a single-objective problem for solution using a weighted method.

[0040] 12. Incremental Learning: This is a machine learning paradigm where a model continuously learns from and updates its parameters based on newly arriving data, without needing to retrain using all historical data. Incremental learning can significantly reduce computational costs and enable the model to adapt to changes in data distribution over time. In this invention, the incremental learning method is used to continuously update the parameters of the hydrological-hydrodynamic coupling model and the algal bloom prediction model using real-time monitoring data collected after the implementation of the scheme, achieving self-iterative optimization of the model.

[0041] Example 1 To facilitate understanding, before elaborating on the specific solutions of this embodiment, the overall inventive concept of the present invention is described here: the reservoir and irrigation area are regarded as a two-way coupled whole. The impact of water conservancy scheduling on hydrodynamics and nutrients is simulated through a physical model. The simulation results are then input into a machine learning model to achieve advance prediction. The scheduling scheme is dynamically optimized according to the risk level. Finally, through feedback loop, continuous self-improvement is achieved, thereby transforming water conservancy scheduling from a passive factor to an active prevention and control measure, and realizing long-term intelligent prevention and control of algal blooms.

[0042] The specific implementation method is as follows: like Figure 1-2 The diagram shown is a flowchart of a method for predicting and controlling algal blooms in reservoir irrigation areas driven by hydrological and hydrodynamic coupling, provided by the present invention.

[0043] As an example, the method includes: Step S1, collecting long-sequence hydrological data, irrigation and farmland runoff data, water quality data, meteorological data, and historical algal bloom data of the target reservoir-irrigation area system; completing outlier removal, missing value completion, and dimensionless standardization to construct a standardized dataset; Step S2, based on the standardized dataset, constructing a non-point source pollution confluence model for the irrigation area and a two-dimensional hydrodynamic-water quality model for the reservoir area; achieving bidirectional coupling through spatiotemporal scale matching and boundary condition interaction to generate hydrological-hydrodynamic coupling simulation results under different water conservancy scheduling conditions and irrigation cycles; the simulation results include the nitrogen and phosphorus load input process of irrigation area runoff, and the spatiotemporal distribution data of reservoir water flow field and nutrients; Step S3, using the hydrological-hydrodynamic coupling simulation results, water quality data, and meteorological data from the standardized dataset as the input feature set, and historical algal density and algal bloom events as output labels, constructing a time-series prediction model based on a long short-term memory network and training, validating, and calibrating it to determine the critical threshold for algal blooms under different operating conditions. Output the risk prediction results of algal blooms for the next N days; Step S4: Based on the predicted values ​​of chlorophyll a concentration, algal bloom probability, and total nitrogen and total phosphorus concentrations in the risk prediction results, construct an algal bloom risk comprehensive index, classify the risk into three levels (low, medium, and high) and corresponding to different warning levels, and push the warning information and risk spatial distribution to the reservoir and irrigation area management units simultaneously, while transmitting the risk level as a trigger signal to Step S5; Step S5: Based on the risk level transmitted in Step S4, construct a multi-objective optimized water conservancy scheduling model with algal bloom control, water supply safety, irrigation guarantee, and flood control safety as the core objectives, dynamically adjust the weights of each objective, automatically generate a graded linkage prevention and control scheme adapted to the operation rules of reservoir-irrigation area water conservancy projects, and issue the scheme for execution; Step S6: Collect real-time hydrological, water quality, and algal density data of the reservoir area after the implementation of the graded linkage prevention and control scheme described in Step S5, complete the parameter self-learning and iterative optimization of the time series prediction model and the coupled model based on the incremental learning method, and continuously improve the accuracy of risk prediction and the adaptability of the prevention and control scheme. Where N = 7 to 15.

[0044] In some feasible implementations, in step S1, data standardization adopts the min-max normalization method, and the calculation formula is: ; In the formula, For the standardized i-th sample data, For the original sample data, These represent the maximum and minimum values ​​of the corresponding indicator data series, respectively. In step S1, outlier removal is performed using... The criteria and determination formula are as follows: ; In the formula, It is the arithmetic mean of the data sequence. Let be the standard deviation of the data sequence, satisfying The data in the criterion judgment formula is identified as outlier, and linear interpolation is used to correct it.

[0045] Specifically, long-term basic data of the target reservoir-irrigation area system will be collected, including: hydrological data: daily water level, reservoir capacity, inflow, and outflow of the reservoir; irrigation and farmland drainage data: irrigation cycle, daily irrigation water consumption, farmland drainage patterns, crop planting structure, and soil type data; water quality data: daily total nitrogen (TN), total phosphorus (TP), chlorophyll a, permanganate index, and dissolved oxygen (DO) of the reservoir and irrigation area water conveyance system; meteorological data: daily temperature, rainfall, wind speed, wind direction, sunshine duration, and solar radiation intensity; and historical algal bloom data: historical algal bloom time, spatial distribution range, algal density, and bloom level.

[0046] Preprocessing of the collected raw data: Outlier removal: using... Criteria, that is, when Values ​​identified as outliers are filled using linear interpolation; missing value filling is also done using linear interpolation; dimensionless standardization is achieved using the min-max normalization method, with the following formula: All indicators are mapped to the [0,1] interval to construct a standardized dataset.

[0047] In some feasible implementations, step S2 includes: Step S21, constructing a non-point source pollution runoff model for the irrigation area, including: based on the SWAT model, using hydrological response units as the basic calculation units, inputting the standardized dataset, calculating the surface irrigation runoff using the SCS-CN model, calculating the total nitrogen load and total phosphorus load using dissolved nitrogen load formula and adsorbed phosphorus load formula respectively, and outputting the daily irrigation area runoff, daily irrigation area runoff total nitrogen load, and daily irrigation area runoff total phosphorus load after calibration; Step S22, constructing a two-dimensional hydrodynamic-water quality model for the reservoir area, including: based on the MIKE21 model, discretizing the reservoir area into an unstructured grid, using the two-dimensional shallow water equation as the hydrodynamic control equation and coupling it with a convection-diffusion water quality module, inputting the standardized dataset, and outputting the daily runoff... The data includes the water depth, flow velocity, water level, and concentrations of total nitrogen, total phosphorus, and chlorophyll a in each grid, as well as the daily outflow. Step S23, bidirectional coupling, includes: using a preset time step (with one day as the external coupling step), the daily outflow, total nitrogen load, and total phosphorus load of the irrigation area outflow output by the irrigation area non-point source pollution runoff model are used as the inflow boundary conditions of the reservoir area two-dimensional hydrodynamic-water quality model, and converted into the inflow and pollutant concentration at the reservoir inlet according to the coupling interface formula; at the same time, the daily outflow output by the reservoir model is used as the upper limit of the available water volume for irrigation water intake of the irrigation area model on the next day, updating the irrigation satisfaction rate of the irrigation area and recalculating the outflow process, realizing bidirectional coupling to simulate the hydrological-hydrodynamic coupling simulation results under different water conservancy scheduling conditions and irrigation cycles.

[0048] Preferably, the dissolved nitrogen loading formula is: ; The formula for the adsorbed phosphorus loading is: ; In the formula, Dissolved nitrogen load from surface runoff, in kg; Surface irrigation runoff volume, unit: ; This represents the concentration of dissolved nitrogen in the runoff, expressed in mg / L. The nitrogen transport loss coefficient in the ditch; Phosphorus loading in sediment adsorbed state, in kg; This refers to the sediment transport volume of the irrigation area's runoff, expressed in tons (t). The phosphorus content in the sediment is expressed in mg / kg. This represents the phosphorus sediment transport ratio.

[0049] Preferably, in the two-dimensional hydrodynamic-water quality model of the reservoir area, the continuity equation of the two-dimensional shallow water equation is: ; The momentum equation in the x-direction is: ; The momentum equation in the y-direction is: ; In the formula, h is the water depth, in meters (m). The water level is represented by meters (m); u and v are the velocity components in the x and y directions, respectively, in meters per second (m / s); t is time, in seconds (s); and g is the acceleration due to gravity, in cubic meters per second (m / s). ; This is the reference density for water, in units of... ; This represents the actual density of the water, in units of... ; , The wind stress component on the water surface is expressed in units of 1000 ppm. ; , The shear stress component of the bed surface is expressed in units of 1000 kJ / m². ; , , and is the viscous stress tensor component; f is the Coriolis force coefficient, in units of . S represents source and sink terms, including irrigation district runoff inflow, reservoir outflow, and intake flow rate, in units of... ; , Let x be the momentum source and sink terms in the x and y directions.

[0050] Preferably, the coupling interface formula in the bidirectional coupling is: ; ; In the formula, Let be the inflow rate at the k-th inlet of the reservoir at time t, in units of . ; Let be the flow rate of the irrigation area's runoff into the reservoir at time t, in units of . ; Let be the nitrogen / phosphorus pollutant concentration at the k-th inlet at time t, in units of . ; This represents the nitrogen / phosphorus load output from the irrigation district at time t, in units of... .

[0051] Specifically, step S2 is the core of this embodiment. By bidirectionally coupling the irrigation district non-point source pollution runoff model (SWAT) and the reservoir two-dimensional hydrodynamic-water quality model (MIKE21), the flow field and nutrient distribution in the reservoir under different water conservancy scheduling and irrigation conditions are simulated. Based on the SWAT model, the irrigation district non-point source pollution runoff model is built, dividing the irrigation district into 28 hydrological response units. The model parameters are calibrated, and the daily discharge flow and TN / TP pollution load inflow process of the irrigation district under different irrigation cycles are simulated and calculated. Based on the MIKE21 model, the reservoir two-dimensional hydrodynamic-water quality model is built. The reservoir is discretized using an unstructured grid, with a total of 12,862 grids. The hydrodynamic and water quality parameters are calibrated, and the spatiotemporal distribution of water flow field, velocity, and nutrients in the reservoir is simulated. By matching spatiotemporal scales, the daily discharge flow and nitrogen and phosphorus load of the irrigation district model are used as the inflow boundary conditions of the reservoir model, and the outflow process of the reservoir model is used as the irrigation water intake boundary of the irrigation district model, achieving bidirectional coupling between the two models. The NSE of the model during the calibration period is greater than 0.78, and the NSE during the validation period is greater than 0.75, meeting the simulation accuracy requirements. That is, forward coupling: the daily discharge flow, total nitrogen load, and total phosphorus load of the irrigation district output from the irrigation district non-point source pollution runoff model are used as the inflow boundary conditions of the reservoir model, and converted into the flow rate and pollutant concentration at the reservoir inlet according to the coupling interface formula; reverse coupling: the daily outflow output from the two-dimensional hydrodynamic-water quality model of the reservoir area is used as the upper limit of the available water volume for irrigation water intake of the irrigation district non-point source pollution runoff model on the next day, updating the irrigation satisfaction rate of the irrigation district and recalculating the discharge process. Through the aforementioned two-way coupling, S2 can simulate the impact of irrigation district runoff nitrogen and phosphorus input, reservoir water level and flow changes on the reservoir water flow field and spatiotemporal distribution of nutrients under different water conservancy scheduling conditions (such as changes in outflow) and irrigation cycles. These simulation results provide key "hydrodynamic-nutrient" features for the algal bloom prediction model in step S3, enabling the machine learning model to learn the causal relationship of "scheduling → flow field → algal response," thereby achieving 7-15 day advance prediction. The output of step S2 includes: the irrigation district runoff nitrogen and phosphorus load input process (i.e., the daily irrigation district runoff flow, daily irrigation district runoff total nitrogen load, and daily irrigation district runoff total phosphorus load), the reservoir water flow field (including flow velocity, water level, etc.), and nutrient spatiotemporal distribution data (including the spatial distribution of total nitrogen, total phosphorus, and chlorophyll a concentrations).

[0052] In some feasible implementations, in step S3, the core computational unit formula of the Long Short-Term Memory network includes: Forgotten Gate: ; Input Gate: ; Candidate cell status: ; Cell status update: ; Output gate: ; Hidden layer output: ; In the formula, , , These are the output vectors of the forget gate, input gate, and output gate, respectively. Let t be the cell state vector at time t; Let t be the hidden layer output vector; Let be the input feature vector at time t; , , , These are the weight matrices for each network layer; , , , For the corresponding bias term; sigmoid is the activation function; tanh is the hyperbolic tangent activation function; For Hadama accumulation.

[0053] Preferably, although the coupled model in S2 can simulate hydrodynamics and nutrients, directly simulating algal growth (especially cyanobacterial blooms) suffers from problems such as numerous parameters, high computational cost, and high uncertainty. Therefore, in step S3, the data output from S2, along with historical water quality and meteorological data from the standardized dataset, are used as input features; historically observed algal densities and algal bloom events are used as standard answers (output labels); and then a time-series prediction model is trained using a Long Short-Term Memory (LSTM) network. The trained model can predict chlorophyll a concentration, algal bloom probability, etc., for the next 7-15 days based on the current and past few days' input features, thereby achieving advanced risk prediction.

[0054] Specifically, this embodiment adopts a data-driven + physical prior strategy: using the output of the S2 coupled model and measured water quality and meteorological data as input features, and historical algal density and outbreak events as labels, an LSTM time-series prediction model is trained. The reservoir water level, reservoir flow velocity field, hydraulic residence time, and irrigation runoff nitrogen and phosphorus load from the hydro-hydrodynamic coupled simulation output, combined with water quality and meteorological data from the standardized dataset, are used as input features. Historical measured algal density and algal bloom events are used as output labels. A time-series prediction model is constructed based on a Long Short-Term Memory (LSTM) network. The standardized dataset is divided into training, validation, and test sets in a 7:2:1 ratio to complete model training and validation. The Nash efficiency coefficient (NSE) is used to evaluate the model's fit; an NSE ≥ 0.75 is considered a successful model calibration. The critical threshold for algal blooms under different water conservancy scheduling conditions and irrigation cycles is determined using ROC curves and the Youden exponent maximization method. Ultimately, the model achieves risk prediction and spatial distribution simulation of algal blooms 7-15 days in the future, with an algal bloom area location error ≤ 5%.

[0055] In some feasible implementations, in step S3, for different water conservancy scheduling conditions and / or irrigation cycles, the corresponding critical threshold for algal blooms is determined by using ROC curves and the Youden index maximization method; the Youden index is calculated according to the following formula: J = Sensitivity + Specificity - 1; In the formula, sensitivity is the proportion of algal blooms correctly predicted by the time series prediction model, and specificity is the proportion of algal blooms not predicted by the time series prediction model. The chlorophyll a concentration, algal bloom probability, total nitrogen concentration, and total phosphorus concentration corresponding to the maximum value of the Youden index are taken as the critical threshold for algal blooms under this operating condition or cycle. The calibration performance of the time series forecasting model is evaluated using the Nash efficiency coefficient (NSE), calculated as follows: ; In the formula, This is the measured value of algal density. These are simulated values ​​from a time series prediction model. Let n be the arithmetic mean of the measured value sequence, and n be the sample size.

[0056] Specifically, the critical thresholds are determined as follows: using the validation set, the thresholds for chlorophyll a concentration, algal bloom probability, total nitrogen concentration, and total phosphorus concentration are determined using the Youden exponent maximization method. The Youden exponent J = sensitivity + specificity - 1; a larger J indicates stronger diagnostic capability. Taking chlorophyll a as an example, possible thresholds (e.g., from 10 to 100) are iterated, and the sensitivity and specificity at each threshold are calculated. The threshold with the largest J is taken as the optimal critical threshold.

[0057] In some feasible implementations, the formula for calculating the Algal Bloom Risk Composite Index (ARI) in step S4 is as follows: ; In the formula, This is the predicted value for chlorophyll a concentration. The critical threshold for chlorophyll a burst; This is a predicted probability value for algal blooms. This represents the critical threshold for the probability of an outbreak. , These are predicted values ​​for total phosphorus and total nitrogen concentrations. , These are the critical thresholds for total phosphorus and total nitrogen; , , , The weights of each indicator are determined using the Analytic Hierarchy Process (AHP), and satisfy the following conditions: ; The risk level classification criteria are as follows: Low risk: , ; Medium risk: , ; High risk: , ; in, , .in, The value is 0.6. The value is 1.0. Take 30% The value is 70%.

[0058] Preferably, since algal blooms are influenced by multiple factors, a single indicator (such as chlorophyll a) may produce false alarms. This invention constructs a comprehensive index that integrates four dimensions: chlorophyll a, bloom probability, total phosphorus, and total nitrogen; the risk level classification is shown in Table 1. Table 1:

[0059] In some feasible implementations, in step S5, the comprehensive objective function of the multi-objective optimization water conservancy scheduling model is: ; In the formula, F is the comprehensive objective function; To achieve the goal of algal bloom control, the core indicators are maximizing the average flow velocity and minimizing the hydraulic residence time in the reservoir area. To ensure water supply security, the core indicator is maximizing the utilization rate of reservoir capacity. To ensure irrigation, the core indicator is to maximize the rate at which irrigation water demand is met in the irrigation district. To achieve the goal of flood control safety, the core indicator is to maximize the compliance rate of reservoir water level and outflow. , , , The weights of each target are dynamically adjusted according to the warning level, and the following conditions are met: .

[0060] Specifically, (Algal bloom control objectives): Maximize the average flow velocity and minimize the hydraulic residence time in the reservoir area. Increasing flow velocity and shortening residence time can inhibit algal aggregation and growth. (Water supply security target): Maximize the utilization rate of reservoir capacity, that is, the ratio of actual water storage to the designed utilization capacity. (Irrigation guarantee target): Maximize the irrigation water demand satisfaction rate of the irrigation district, that is, the ratio of actual water supply to water demand. (Flood control safety objectives): Maximize the compliance rate of reservoir water level not exceeding the flood limit level and downstream flow not exceeding the safe flow of the river channel.

[0061] Dynamic weighting strategies include: dynamically adjusting based on the warning level. The adjustment strategy is shown in Table 2: Table 2:

[0062] In low-risk situations, the focus is on ecological control, while in high-risk situations, priority is given to ensuring flood control and water supply safety, while also taking into account algal bloom control (through emergency measures).

[0063] In some feasible implementations, the model constraints include reservoir water balance constraints, reservoir water level constraints, outflow constraints, and irrigation district constraints, wherein the formula for the reservoir water balance constraint is: ; In the formula, , The reservoir capacity is given by the units t and t+1. ; The average inbound flow rate during the time period, in units of ; The average discharge flow rate over the period is expressed in units of... ; Evaporation loss flow rate in the reservoir area during the specified time period, in units of ; The unit for calculating the duration is seconds (s). Furthermore, the weights of each target are dynamically adjusted according to the warning level, and a hierarchical linkage prevention and control plan adapted to the operation rules of reservoir-irrigation area water conservancy projects is automatically generated.

[0064] Preferred, tiered, coordinated prevention and control measures include: Low risk (blue alert): Focus on optimizing water resource management. For example: increase reservoir discharge to raise the average flow velocity in the reservoir area from 0.02 m / s to over 0.05 m / s; adjust irrigation schedules to avoid periods of concentrated rainfall and reduce nitrogen and phosphorus from farmland runoff entering the reservoir; implement stratified water intake, changing from surface water intake to mid-layer water intake (due to high algae concentrations in the surface layer). Chemical or biological measures will not be used.

[0065] Medium risk (yellow alert): Based on low-risk management, add ecological control measures. For example: deploy straw allelopathic ecological control devices at drainage outlets and shallow water areas of reservoir bays—straw (rice, wheat, etc.) is soaked and fermented to make slow-release bags, which are then floated or suspended in the water to release allelopathic substances (such as phenolic acids) to inhibit algae growth while simultaneously adsorbing nitrogen and phosphorus. This measure has no secondary pollution and is low-cost.

[0066] High Risk (Red Alert): Prioritize water supply and flood control safety, while initiating emergency management measures. Specifically, this includes: adjusting the stratified water intake plan to ensure drinking water intakes avoid algae-rich water layers; developing an emergency discharge plan: if water levels permit, increase discharge to flush the reservoir area and quickly remove algae-rich water; and activating an air-ground-ship collaborative emergency response system: drones will patrol and monitor algae bloom dynamics, shore-based UV / ultrasonic disinfection equipment will treat near-shore algae-rich areas, and shipborne dredging devices will collect floating algae.

[0067] In some feasible implementations, in step S6, the weight update formula for the incremental learning iteration of the model is: ; In the formula, This is the updated model weight matrix; This is the original weight matrix before iteration; L is the learning rate; L is the model loss function, using the mean squared error (MSE), calculated as follows: ; In the formula, These are the measured values ​​of algae density and water quality indicators after the implementation of the plan; is the predicted value corresponding to the model; n is the number of newly added measured samples.

[0068] Specifically, for the LSTM prediction model, the weights are updated using the incremental gradient descent method; for the coupled physical model of S2 (SWAT-MIKE21), the key parameters of the model (such as Manning coefficient and pollutant degradation coefficient) are updated periodically (e.g., quarterly) using newly added measured hydrological and water quality data through ensemble Kalman filtering or parameter recalibration methods, in order to maintain the model's adaptability to dynamic changes in the system.

[0069] To intuitively demonstrate the beneficial effects of this embodiment, a large reservoir-irrigation system in southern my country is used as the implementation object. The reservoir has a total capacity of 520 million m³, making it a large (2) type reservoir. It undertakes the irrigation of 320,000 mu of farmland downstream and provides drinking water sources for two surrounding prefecture-level cities. In recent years, due to the impact of farmland runoff in the irrigation area, the eutrophication problem of the reservoir has become prominent, with algal blooms of varying degrees occurring every spring and summer, threatening the safety of drinking water sources. The method described in this invention is used to carry out algal bloom prediction and prevention linkage. The specific implementation steps are as follows: 1. Basic Data Collection and Standardization: Long-series basic data for the reservoir-irrigation area system from 2018 to 2024 were collected, including: daily water level, capacity, inflow, and outflow of the reservoir; irrigation system, daily irrigation water consumption, crop planting structure, and farmland drainage monitoring data; daily TN, TP, chlorophyll a, and dissolved oxygen data from 12 water quality monitoring sections in the reservoir area; daily temperature, rainfall, wind speed, and sunshine data from the basin meteorological stations; and temporal and spatial distribution and algal density data of historical algal bloom events from 2018 to 2024. [Data was adopted...] The criteria are to remove outliers from the data, use linear interpolation to fill in missing data, and use the min-max normalization method to complete the dimensionless standardization of all data to construct a standardized dataset.

[0070] 2. Construction of a reservoir-irrigation district hydrological-hydraulic coupled model: A non-point source pollution runoff model for the irrigation district was built based on the SWAT model, dividing the irrigation district into 28 hydrological response units. Model parameters were calibrated, and the daily discharge flow and TN / TP pollution load inflow process of the irrigation district under different irrigation cycles were simulated. A two-dimensional hydrodynamic-water quality model of the reservoir was built based on the MIKE21 model, using an unstructured mesh to discretize the reservoir area, with a total of 12,862 meshes. Hydrodynamic and water quality parameters were calibrated, and the spatiotemporal distribution of water flow field, velocity, and nutrients in the reservoir area was simulated. Through spatiotemporal scale matching, the daily discharge flow and nitrogen and phosphorus load of the irrigation district model were used as the inflow boundary conditions of the reservoir model, and the outflow process of the reservoir model was used as the irrigation water intake boundary of the irrigation district model, realizing bidirectional coupling between the two models. The NSE during the model calibration period was greater than 0.78, and the NSE during the validation period was greater than 0.75, meeting the simulation accuracy requirements.

[0071] 3. Training and Calibration of Algal Bloom Risk Prediction Model: Using 18 indicators—reservoir water level, average reservoir flow velocity, hydraulic retention time, and irrigation runoff TN / TP load—combined with water quality and meteorological data from the coupled model as input features, and measured chlorophyll a concentration and algal bloom events in the reservoir area as output labels, an LSTM time-series prediction model was constructed. The standardized dataset was divided into training, validation, and test sets in a 7:2:1 ratio. The model time step was set to 7 days, the number of hidden layer neurons to be 64, and the number of iterations to be 200. Model training and validation were completed. After calibration, the model's prediction NSEs for chlorophyll a concentrations in the next 7, 10, and 15 days were 0.82, 0.79, and 0.76, respectively, meeting the accuracy requirements. Using the Youden exponent maximization method, the critical threshold for algal blooms in this reservoir was determined to be a chlorophyll a concentration of 20... The critical value for algal bloom probability is 50%.

[0072] 4. Risk Classification, Early Warning, and Threshold Triggering: Using the analytic hierarchy process (AHP), the weights of chlorophyll a concentration, algal bloom probability, TP concentration, and TN concentration were determined to be 0.4, 0.3, 0.2, and 0.1, respectively. An algal bloom risk index (ARI) was constructed, classifying risk levels as: low risk (ARI < 0.6, blue alert), medium risk (0.6 ≤ ARI < 1.0, yellow alert), and high risk (ARI ≥ 1.0, red alert). Based on model predictions, an early warning is automatically triggered when the risk index reaches the corresponding threshold, and the warning information is simultaneously pushed to the reservoir management office, irrigation district management bureau, and local river and lake chief system office.

[0073] 5. Generation of Water Conservancy Scheduling-Ecological Prevention and Control Linkage Scheme: Construct a multi-objective optimized water conservancy scheduling model, and dynamically adjust the objective weights according to different early warning levels: for low risk, the weights are set to... , , , =0.1, optimizing the reservoir discharge flow from the daily average Upgraded to Adjust the irrigation schedule in the irrigation area to avoid periods of concentrated rainfall, reduce the amount of nitrogen and phosphorus entering the reservoir from farmland runoff, and suppress algae growth through hydraulic disturbance; for medium-risk situations, the weighting is set to... , , , Based on optimized scheduling, straw allelopathic ecological control devices are deployed at the reservoir inlet and shallow water areas to reduce nitrogen and phosphorus loads in the water and inhibit algae growth; during high-risk periods, the weighting is set to... , , , Priority was given to ensuring the safety of drinking water intakes, adjusting the stratified water intake plan, formulating an emergency discharge scheduling plan, and simultaneously launching a coordinated emergency response involving air, ground, and shipborne operations, including drone patrols, shore-based disinfection, and shipborne salvage.

[0074] 6. Scheme Iteration and Effect Feedback: Hydrological, water quality, and algae density data were collected in real time after the scheme was implemented through in-situ water quality monitoring stations in the reservoir area and UAV remote sensing monitoring. An incremental learning dataset was constructed, and model parameters were iterated and optimized monthly. Implementation verification showed that the method of this invention achieved an accuracy rate of 89% in predicting algal blooms in the reservoir, a 42% improvement over the original model, and provided risk warnings 10 days in advance. Through the implementation of the joint prevention and control scheme, no large-scale algal blooms occurred in the reservoir in 2025, the drinking water source quality compliance rate reached 100%, and the irrigation water demand of the irrigation area and the flood control safety of downstream rivers were fully guaranteed, achieving significant application results.

[0075] Example 2 Please see Figure 3 This embodiment provides a schematic diagram of a reservoir irrigation area algal bloom prediction and control system driven by hydrological and hydrodynamic coupling.

[0076] As an example, the system is implemented using the hydrological-hydraulic coupled-driven method for predicting and controlling algal blooms in reservoir irrigation areas as described in Example 1. The system includes: The basic data acquisition and standardization processing module 210 is suitable for collecting long-sequence hydrological data, irrigation and farmland drainage data, water quality data, meteorological data, and historical algal bloom data of the target reservoir-irrigation area system. It completes outlier removal, missing value completion, and dimensionless standardization processing to construct a standardized dataset. The reservoir-irrigation district hydrological-hydraulic coupling model construction module 220 is suitable for building a non-point source pollution confluence model of the irrigation district and a two-dimensional hydrodynamic-water quality model of the reservoir district based on the standardized dataset. It achieves bidirectional coupling through spatiotemporal scale matching and boundary condition interaction, and generates hydrological-hydraulic coupling simulation results under different water conservancy scheduling conditions and irrigation cycles. The simulation results include the nitrogen and phosphorus load input process of irrigation district runoff, water flow field and spatiotemporal distribution data of nutrients in the reservoir district. The algal bloom risk prediction model construction module 230 is suitable for taking the hydro-hydrodynamic coupling simulation results and water quality data and meteorological data in the standardized dataset as input feature set, taking historical algal density and algal bloom events as output labels, constructing a time series prediction model based on long short-term memory network and training, validating and calibrating it, determining the critical threshold of algal bloom under different working conditions, and outputting the algal bloom risk prediction result for the next N days. The risk classification, early warning, and threshold triggering module 240 is applicable to constructing a comprehensive algal bloom risk index based on the predicted values ​​of chlorophyll a concentration, algal bloom probability, and total nitrogen and total phosphorus concentrations in the risk prediction results. It divides the risk into three levels: low, medium, and high, and corresponds to different early warning levels. It pushes the early warning information and risk spatial distribution to the reservoir and irrigation area management units simultaneously. At the same time, the risk level is transmitted as a trigger signal to the water conservancy scheduling-ecological prevention and control linkage scheme generation module. The Water Conservancy Dispatch-Ecological Prevention and Control Linkage Scheme Generation Module 250 is applicable to constructing a multi-objective optimized water conservancy dispatch model based on the risk level transmitted by the risk classification early warning and threshold triggering module, with algal bloom control, water supply security, irrigation guarantee, and flood control security as the core objectives. It dynamically adjusts the weight of each objective, automatically generates a graded linkage prevention and control scheme adapted to the operation rules of reservoir-irrigation area water conservancy projects, and issues the scheme for execution. The scheme iteration and effect feedback module 260 is suitable for real-time collection of reservoir hydrological, water quality, and algae density data after the implementation of the hierarchical linkage prevention and control scheme described in the water conservancy scheduling-ecological prevention and control linkage scheme generation module. Based on incremental learning methods, it completes parameter self-learning and iterative optimization of the time-series prediction model and the coupled model, continuously improving the accuracy of risk prediction and the adaptability of the prevention and control scheme. It is not difficult to see that this implementation method is a system embodiment corresponding to the first implementation method, and this implementation method can be implemented in conjunction with the first implementation method. The relevant technical details mentioned in the first implementation method are still valid in this implementation method, and will not be repeated here to reduce repetition. Correspondingly, the relevant technical details mentioned in this implementation method can also be applied to the first implementation method.

[0077] It is worth mentioning that all modules involved in this embodiment are logical units. In practical applications, a logical unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. Furthermore, to highlight the innovative aspects of this invention, this embodiment does not introduce units that are not closely related to solving the technical problem proposed by this invention; however, this does not mean that other units are absent from this embodiment.

[0078] Example 3 Please see Figure 4 The present invention also provides an electronic device, including: a memory and a processor; the memory stores at least one program instruction; the processor loads and executes the at least one program instruction to implement the hydrological and hydrodynamic coupled-driven method for predicting and controlling algal blooms in reservoir irrigation areas provided in Embodiment 1.

[0079] The memory 702 and processor 701 are connected via a bus, which may include any number of interconnecting buses and bridges, connecting various circuits of one or more processors 701 and memory 702 together. The bus may also connect various other circuits, such as peripheral devices, voltage regulators, and power management circuits, which are well known in the art and therefore will not be described further herein. A bus interface provides an interface between the bus and the transceiver. The transceiver may be a single element or multiple elements, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by processor 701 is transmitted over a wireless medium via an antenna, which further receives data and transmits it to processor 701.

[0080] Processor 701 is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory 702 can be used to store data used by processor 701 during operation.

[0081] Example 4 This invention also proposes a storage medium storing a hydrologically and hydrodynamically coupled method for predicting and controlling algal blooms in reservoir irrigation areas. When executed by a processor, the hydrologically and hydrodynamically coupled algal bloom prediction and control program implements the steps of the hydrologically and hydrodynamically coupled algal bloom prediction and control method described above. Since this storage medium employs all the technical solutions of the above embodiments, it possesses at least all the beneficial effects brought about by the technical solutions of the above embodiments, which will not be elaborated upon further here.

[0082] The above descriptions are merely embodiments of the present invention. Commonly known structures and characteristics are not described in detail here. Those skilled in the art are aware of all common technical knowledge in the field prior to the application date or priority date, are aware of all existing technologies in that field, and have the ability to apply conventional experimental methods prior to that date. Those skilled in the art can, based on the guidance provided in this application, improve and implement this solution in combination with their own capabilities. Some typical known structures or methods should not be obstacles for those skilled in the art to implement this application. It should be noted that those skilled in the art can make several modifications and improvements without departing from the structure of the present invention. These should also be considered within the scope of protection of the present invention, and will not affect the effectiveness of the implementation of the present invention or the practicality of the patent. The scope of protection claimed in this application should be determined by the content of its claims, and the specific embodiments described in the specification can be used to interpret the content of the claims.

Claims

1. A method for predicting and controlling algal blooms in reservoir irrigation areas driven by hydrological and hydrodynamic coupling, characterized in that, The method includes: Step S1: Collect long-sequence hydrological data, irrigation and farmland drainage data, water quality data, meteorological data, and historical algal bloom data of the target reservoir-irrigation area system; complete outlier removal, missing value completion, and dimensionless standardization processing to construct a standardized dataset. Step S2: Based on the standardized dataset, a non-point source pollution confluence model for the irrigation area and a two-dimensional hydrodynamic-water quality model for the reservoir area are built respectively. Two-way coupling is achieved through spatiotemporal scale matching and boundary condition interaction to generate hydrological-hydrodynamic coupling simulation results under different water conservancy scheduling conditions and irrigation cycles. The simulation results include the nitrogen and phosphorus load input process of irrigation area runoff, water flow field and spatiotemporal distribution data of nutrients in the reservoir area. Step S3: Take the hydro-hydrodynamic coupling simulation results and the water quality data and meteorological data in the standardized dataset as the input feature set, take the historical algal density and algal bloom events as the output labels, construct a time series prediction model based on the long short-term memory network and train, verify and calibrate it, determine the critical threshold of algal bloom under different working conditions, and output the algal bloom risk prediction result for the next N days. Step S4: Based on the predicted values ​​of chlorophyll a concentration, algal bloom probability, and total nitrogen and total phosphorus concentration in the risk prediction results, construct an algal bloom risk comprehensive index, divide the risk into three levels: low, medium, and high, and correspond to different warning levels. Simultaneously push the warning information and risk spatial distribution to the reservoir and irrigation area management units, and transmit the risk level as a trigger signal to step S5. Step S5: Based on the risk level transmitted in Step S4, with algal bloom control, water supply security, irrigation guarantee, and flood control security as the core objectives, construct a multi-objective optimized water conservancy scheduling model, dynamically adjust the weight of each objective, automatically generate a hierarchical linkage prevention and control plan that is compatible with the operation rules of reservoir-irrigation area water conservancy projects, and issue the plan for implementation. Step S6: Collect real-time data on reservoir hydrology, water quality, and algae density after the implementation of the hierarchical linkage prevention and control scheme described in Step S5. Based on the incremental learning method, complete the parameter self-learning and iterative optimization of the time series prediction model and the coupled model to continuously improve the accuracy of risk prediction and the adaptability of the prevention and control scheme.

2. The method for predicting and controlling algal blooms in reservoir irrigation areas driven by hydrodynamic coupling according to claim 1, characterized in that, Step S2 includes: Step S21: Construct a non-point source pollution runoff model for the irrigation area, including: based on the SWAT model, using the hydrological response unit as the basic calculation unit, inputting the standardized dataset, using the SCS-CN model to calculate the surface irrigation runoff, using the dissolved nitrogen load formula and the adsorbed phosphorus load formula to calculate the total nitrogen load and total phosphorus load respectively, and outputting the daily irrigation area runoff, daily irrigation area runoff total nitrogen load and daily irrigation area runoff total phosphorus load after calibration; Step S22: Construct a two-dimensional hydrodynamic-water quality model for the reservoir area, including: based on the MIKE21 model, discretizing the reservoir area into an unstructured grid, using the two-dimensional shallow water equation as the hydrodynamic control equation and coupling it with the convection-diffusion water quality module, inputting the standardized dataset, and outputting the daily water depth, flow velocity, water level, and total nitrogen, total phosphorus, and chlorophyll a concentrations, as well as the daily outflow for each grid after calibration. Step S23, bidirectional coupling, includes: using a preset time step, taking the daily irrigation area runoff flow, daily irrigation area runoff total nitrogen load, and daily irrigation area runoff total phosphorus load output by the irrigation area non-point source pollution runoff model as the inflow boundary conditions of the reservoir area two-dimensional hydrodynamic-water quality model, and converting them into the flow rate and pollutant concentration at the reservoir inlet according to the coupling interface formula; at the same time, taking the daily outflow output by the reservoir model as the upper limit of available water for irrigation water intake of the irrigation area model for the next day, updating the irrigation area irrigation satisfaction rate and recalculating the runoff process, to achieve bidirectional coupling, so as to simulate the hydrological-hydrodynamic coupling simulation results under different water conservancy scheduling conditions and irrigation cycles.

3. The method for predicting and controlling algal blooms in reservoir irrigation areas driven by hydrodynamic coupling according to claim 2, characterized in that, The formula for dissolved nitrogen loading is: ； The formula for the adsorbed phosphorus loading is: ； In the formula, Dissolved nitrogen load from surface runoff, in kg; Surface irrigation runoff volume, unit: ; This represents the concentration of dissolved nitrogen in the runoff, expressed in mg / L. The nitrogen transport loss coefficient in the ditch; Phosphorus loading in sediment adsorbed state, in kg; This refers to the sediment transport volume of the irrigation area's runoff, expressed in tons (t). The phosphorus content in the sediment is expressed in mg / kg. This represents the phosphorus sediment transport ratio.

4. The method for predicting and controlling algal blooms in reservoir irrigation areas driven by hydrodynamic coupling according to claim 2, characterized in that, In the two-dimensional hydrodynamic-water quality model of the reservoir area, the continuity equation of the two-dimensional shallow water equation is: ； The momentum equation in the x-direction is: ； The momentum equation in the y-direction is: ； In the formula, h is the water depth, in meters (m). This refers to the water level, measured in meters (m). u and v represent the velocity components in the x and y directions, respectively, in m / s; t represents time, in seconds; g represents gravitational acceleration, in units of... ; This is the reference density for water, in units of... ; This represents the actual density of the water, in units of... ; , The wind stress component on the water surface is expressed in units of 1000 ppm. ; , The shear stress component of the bed surface is expressed in units of 1000 kJ / m². ; , , and These are components of the viscous stress tensor; f is the Coriolis force coefficient, with units of . S represents source and sink terms, including irrigation district runoff inflow, reservoir outflow, and intake flow rate, in units of... ; , Let x be the momentum source and sink terms in the x and y directions.

5. The method for predicting and controlling algal blooms in reservoir irrigation areas driven by hydrodynamic coupling according to claim 2, characterized in that, The coupling interface formula in the bidirectional coupling is: ； ； In the formula, Let be the inflow rate at the k-th inlet of the reservoir at time t, in units of . ; Let be the flow rate of the irrigation area's runoff into the reservoir at time t, in units of . ; Let be the nitrogen / phosphorus pollutant concentration at the k-th inlet at time t, in units of . ; This represents the nitrogen / phosphorus load output from the irrigation district at time t, in units of... .

6. The method for predicting and controlling algal blooms in reservoir irrigation areas driven by hydrodynamic coupling according to claim 1, characterized in that, In step S3, the core computational unit formula of the Long Short-Term Memory network includes: Forgotten Gate: ; Input Gate: ; Candidate cell status: ; Cell status update: ; Output gate: ; Hidden layer output: ; In the formula, , , These are the output vectors of the forget gate, input gate, and output gate, respectively. Let t be the cell state vector at time t; Let t be the hidden layer output vector; Let be the input feature vector at time t; , , , These are the weight matrices for each network layer; , , , For the corresponding bias term; sigmoid is the activation function; tanh is the hyperbolic tangent activation function; For Hadama accumulation.

7. The method for predicting and controlling algal blooms in reservoir irrigation areas driven by hydrodynamic coupling according to claim 1, characterized in that, In step S3, for different water conservancy scheduling conditions and / or irrigation cycles, the corresponding critical threshold for algal blooms is determined using ROC curves and the Yoden index maximization method; the Yoden index is calculated according to the following formula: J = Sensitivity + Specificity - 1; In the formula, sensitivity is the proportion of algal blooms correctly predicted by the time series prediction model, and specificity is the proportion of algal blooms not predicted by the time series prediction model. The chlorophyll a concentration, algal bloom probability, total nitrogen concentration, and total phosphorus concentration corresponding to the maximum value of the Youden index are taken as the critical threshold for algal blooms under this operating condition or cycle. The calibration performance of the time series forecasting model is evaluated using the Nash efficiency coefficient (NSE), calculated as follows: ； In the formula, This is the measured value of algal density. These are simulated values ​​from a time series prediction model. Let n be the arithmetic mean of the measured value sequence, and n be the sample size.

8. The method for predicting and controlling algal blooms in reservoir irrigation areas driven by hydrodynamic coupling according to claim 7, characterized in that, In step S4, the formula for calculating the Algal Bloom Risk Index (ARI) is as follows: ； In the formula, This is the predicted value for chlorophyll a concentration. The critical threshold for chlorophyll a burst; This is the predicted probability value for algal blooms. This represents the critical threshold for the probability of an outbreak. , These are predicted values ​​for total phosphorus and total nitrogen concentrations. , These are the critical thresholds for total phosphorus and total nitrogen; , , , The weights of each indicator are determined using the Analytic Hierarchy Process (AHP), and satisfy the following conditions: ; The risk level classification criteria are as follows: Low risk: , ; Medium risk: , ; High risk: , ; in, , .

9. The method for predicting and controlling algal blooms in reservoir irrigation areas driven by hydrodynamic coupling according to claim 8, characterized in that, In step S5, the comprehensive objective function of the multi-objective optimization water conservancy scheduling model is: ； In the formula, F is the comprehensive objective function; To achieve the goal of algal bloom control, the core indicators are maximizing the average flow velocity and minimizing the hydraulic residence time in the reservoir area. To ensure water supply security, the core indicator is maximizing the utilization rate of reservoir capacity. To ensure irrigation, the core indicator is to maximize the rate at which irrigation water demand is met in the irrigation district. To achieve the goal of flood control safety, the core indicator is to maximize the compliance rate of reservoir water level and outflow. , , , The weights of each target are dynamically adjusted according to the warning level, and the following conditions are met: ; The model constraints include reservoir water balance constraints, reservoir water level constraints, outflow constraints, and irrigation district constraints. The formula for the reservoir water balance constraint is: ； In the formula, , The reservoir capacity is given by the units t and t+1. ; The average inbound flow rate during the time period, in units of ; The average discharge flow rate over the period is expressed in units of... ; Evaporation loss flow rate in the reservoir area during the specified time period, in units of ; The unit for calculating the duration is seconds (s). Furthermore, the weights of each target are dynamically adjusted according to the warning level, and a hierarchical linkage prevention and control plan adapted to the operation rules of reservoir-irrigation area water conservancy projects is automatically generated.

10. A hydrologically and hydrodynamically coupled system for predicting and controlling algal blooms in reservoir irrigation areas, wherein the system is implemented using the hydrologically and hydrodynamically coupled method for predicting and controlling algal blooms in reservoir irrigation areas as described in any one of claims 1-9, characterized in that... The system includes: The basic data acquisition and standardization module is suitable for collecting long-sequence hydrological data, irrigation and farmland drainage data, water quality data, meteorological data, and historical algal bloom data of the target reservoir-irrigation area system. It completes outlier removal, missing value completion, and dimensionless standardization processing to construct a standardized dataset. The reservoir-irrigation district hydrological-hydraulic coupling model construction module is suitable for building a non-point source pollution confluence model of the irrigation district and a two-dimensional hydrodynamic-water quality model of the reservoir district based on the standardized dataset. It achieves bidirectional coupling through spatiotemporal scale matching and boundary condition interaction, and generates hydrological-hydraulic coupling simulation results under different water conservancy scheduling conditions and irrigation cycles. The simulation results include the nitrogen and phosphorus load input process of irrigation district runoff, water flow field and spatiotemporal distribution data of nutrients in the reservoir district. The module for constructing a risk prediction model for algal blooms is suitable for using the hydro-hydrodynamic coupling simulation results and water quality and meteorological data in the standardized dataset as input feature sets, using historical algal density and algal bloom events as output labels, constructing a time-series prediction model based on a long short-term memory network and training, validating and calibrating it, determining the critical threshold for algal blooms under different operating conditions, and outputting the prediction results of algal bloom risk for the next N days. The risk classification, early warning, and threshold triggering module is applicable to constructing a comprehensive algal bloom risk index based on the predicted values ​​of chlorophyll a concentration, algal bloom probability, and total nitrogen and total phosphorus concentrations in the risk prediction results. It divides the risk into three levels: low, medium, and high, and corresponds to different early warning levels. The early warning information and risk spatial distribution are pushed to reservoir and irrigation area management units simultaneously. At the same time, the risk level is used as a trigger signal to transmit to the water conservancy scheduling-ecological prevention and control linkage scheme generation module. The water conservancy scheduling-ecological prevention and control linkage scheme generation module is suitable for constructing a multi-objective optimized water conservancy scheduling model based on the risk level transmitted by the risk classification early warning and threshold triggering module, with algal bloom control, water supply security, irrigation guarantee and flood control safety as the core objectives. It dynamically adjusts the weight of each objective, automatically generates a graded linkage prevention and control scheme that is compatible with the operation rules of reservoir-irrigation area water conservancy projects, and issues the scheme for execution. The scheme iteration and effect feedback module is suitable for real-time collection of reservoir hydrological, water quality and algae density data after the implementation of the hierarchical linkage prevention and control scheme generated by the water conservancy scheduling-ecological prevention and control linkage scheme generation module. Based on the incremental learning method, it completes the parameter self-learning and iterative optimization of the time series prediction model and the coupled model, and continuously improves the accuracy of risk prediction and the adaptability of prevention and control scheme.

Images