A method, system, device and storage medium for generating annual-scale robust scheduling logic for lake water resources and water security collaboration

Abstract

This invention discloses a method, system, device, and storage medium for generating robust scheduling logic on an annual scale for lake water resources and water security collaboration. The system includes a basin hydrology-lake hydrodynamic coupling modeling module for constructing a distributed basin hydrological model and a lake hydrodynamic model; a comprehensive meteorological uncertainty boundary file generation module for obtaining basin meteorological boundary files and lake meteorological boundary files; a basin hydrological simulation module for forming a set of basin inflow-intake boundary files that account for meteorological uncertainty; a lake water level control strategy generation module for generating multiple water level control boundary files; a lake hydrodynamic simulation module; and a water level control strategy determination module for determining the final annual robust water level scheduling strategy. By comprehensively considering the spatiotemporal heterogeneity of rainfall and multi-scenario water level control strategies, the present invention significantly improves the accuracy and robustness of lake water resource management and ensures reliable operation of the system under various meteorological conditions.

Description

Technical Field

[0001] The present invention relates to a lake water level control method, and in particular to a lake water resources-water security coordinated annual-scale robust scheduling logic generation method, system, equipment and storage medium. Background Art

[0002] Existing technologies for regulating lake water levels suffer from the following problems: Most simulations rely on a single meteorological scenario, failing to fully account for the uncertainties surrounding annual and monthly rainfall, rainfall patterns, and spatial variations in rainfall. Basin inflow is typically calculated by multiplying a single or several preset rainfall amounts by a fixed runoff coefficient, accounting for the impact of factors such as prior drought duration and underlying surface differences on runoff production. However, inherent basin uncertainties lead to a nonlinear rainfall-runoff relationship. Consequently, runoff into lakes is subject to significant uncertainty due to the combined influence of the temporal and spatial variability of rainfall and the uncertainty of basin hydrological processes.

[0003] Existing scheduling methods generally rely on manual experience, often screening scenarios based on a limited number of typical hydrological years or preset scenarios such as extreme events. They lack systematic modeling and quantitative responses to multiple sources of uncertainty, such as meteorological and hydrological uncertainties. Because they lack the integration of large-scale scenario simulation and multi-objective optimization mechanisms, the resulting scheduling strategies are prone to failure, lack of robustness, and lack of adaptability when actual operations deviate significantly from the preset scenarios. Summary of the Invention

[0004] Purpose of the invention: The first purpose of the present invention is to provide an annual-scale robust scheduling logic generation method for lake water resources-water security coordination that takes into account meteorological uncertainty and watershed uncertainty and has high control accuracy and strong robustness.

[0005] The second object of the present invention is to provide an annual-scale robust scheduling logic generation system for lake water resources-water security collaboration.

[0006] A third object of the present invention is to provide an electronic device.

[0007] A fourth object of the present invention is to provide a computer-readable storage medium.

[0008] Technical solution: The present invention discloses a method for generating annual-scale robust scheduling logic for lake water resources-water security collaboration, comprising the following steps:

[0009] S1: Collect hydrogeographic data and lake body characteristic data of the target lake's basin, build a distributed basin hydrological model based on the hydrogeographic data, and then use the lake body characteristic data and the output results of the distributed basin hydrological model to construct a lake hydrodynamic model.

[0010] S2: Obtain annual-scale forecast rainfall data and historical meteorological data for the target lake's basin. Use scenario construction techniques to construct multiple boundary scenarios representing meteorological uncertainty based on the annual-scale forecast rainfall data and historical meteorological data. Based on the multiple boundary scenarios, generate a basin meteorological boundary file for use in the distributed basin hydrological model and a lake meteorological boundary file for use in the lake hydrodynamic model.

[0011] S3: Input the generated multiple watershed meteorological boundary files into the distributed watershed hydrological model respectively, extract the time series of water volume entering the lake and the time series of water volume withdrawn from the lake output by the distributed watershed hydrological model after running the corresponding distributed watershed meteorological boundary file, and form a set of watershed inflow-withdrawal boundary files covering meteorological uncertainties;

[0012] S4: Based on the water level monitoring data of the lake body characteristic data, the monthly end-of-month water level distribution range of the target lake is statistically calculated. A grid search method is used to construct multiple annual-scale water level control strategies with differentiated regulation characteristics for the end-of-month water level distribution range. Each annual-scale water level strategy is converted into a water level control boundary file for the lake body hydrodynamic model.

[0013] S5: The lake meteorological boundary file and the basin inflow-water intake boundary file with the same historical meteorological data year are used to form an initial combination. Several initial combinations are arranged and combined with several water level control boundary files to obtain multiple sets of input files for the lake hydrodynamic model. Each set of input files is input into the lake hydrodynamic model for annual-scale simulation, and the lake water level time series and downstream outflow flow time series corresponding to each set of input files after the lake hydrodynamic model simulation are extracted;

[0014] S6: Merge several groups of input files using the same annual-scale water level control strategy to obtain several input groups; conduct a multi-objective comprehensive evaluation of the input groups based on the lake water level time series and downstream outflow flow time series data corresponding to each group of input files, and combine them with lake water resources and water security-related indicators to determine the input group with the best multi-objective comprehensive score, and extract the water level control boundary file used by this input group as the final annual-scale robust water level scheduling logic for the target lake.

[0015] Furthermore, the basin hydrogeographic data described in step S1 include surface elevation data, soil type, land use, hourly meteorological data, measured hydrological data of the basin, spatial distribution information of drainage and water intake facilities, and their operating flow time series data of the basin to which the lake belongs; the lake body characteristic data described in step S1 include underwater three-dimensional terrain data of the lake, lake area meteorological data, water level monitoring data, and continuous monitoring data of inflow and outflow.

[0016] Furthermore, the annual-scale forecast rainfall data described in step S2 refers to the monthly forecast rainfall for the next year predicted by all forecast stations in the basin to which the target lake body belongs; the historical measured meteorological data described in step S2 refers to the many-year historical hourly rainfall data of all rainfall stations in the basin to which the target lake body belongs and the many-year historical hourly multi-index meteorological observation data of all meteorological stations.

[0017] Furthermore, the steps of generating the watershed meteorological boundary file for use in the distributed watershed hydrological model and the lake meteorological boundary file for use in the lake hydrodynamic model in step S2 are as follows:

[0018] S21: Based on the monthly forecast rainfall of each forecast station for the next year, the annual rainfall total of each forecast station is calculated, and all the annual rainfall totals are disturbed and expanded to obtain multiple annual rainfall total samples;

[0019] S22: Based on the historical hourly rainfall data and the historical hourly multi-index meteorological observation data, the weight of the historical monthly rainfall at each rainfall station and meteorological station in the total annual rainfall is calculated to form a historical monthly scale weight sequence. Based on the historical monthly scale weight sequence, several sets of predicted monthly scale weight sequences covering the 12 months of the year are generated;

[0020] S23: The annual rainfall total sample obtained in step S21 and the predicted monthly scale weight sequence obtained in step S22 are arranged and combined to obtain a plurality of annual rainfall total sample-prediction monthly scale weight sequence combinations, and the monthly rainfall of each month in the annual rainfall total sample-prediction monthly scale weight sequence combination constitutes a monthly rainfall time series sample;

[0021] S24: clustering all monthly rainfall time series samples using a clustering method, obtaining k types of typical meteorological scenarios, and using the typical meteorological scenarios as representative samples of the target rainfall time series;

[0022] S25: Combine k representative samples of target monthly rainfall time series with historical meteorological data to predict hourly rainfall time series for all rainfall stations and meteorological stations in the next year;

[0023] S26: Convert the hourly rainfall time series of all rain gauges and meteorological stations within the watershed to which the target lake predicted in step S25 belongs into a watershed meteorological boundary file in the input format of the distributed watershed hydrological model; convert the hourly rainfall time series of the rain gauges and meteorological stations within the specified range of the target lake into a lake meteorological boundary file in the input format of the lake hydrodynamic model.

[0024] Furthermore, in step S22, a method of generating several sets of predicted monthly scale weight sequences including 12 months of a year is as follows:

[0025] All historical monthly weight sequences of each rainfall station and meteorological station in the historical measured meteorological data are classified by month to form 12 historical rainfall weight sets;

[0026] For each month's historical rainfall weight set, a Bernoulli distribution model is constructed based on the binary event results of whether rainfall occurs in that month at each rainfall station and meteorological station recorded in historical meteorological data to simulate the rainfall probability of that month. The Bernoulli distribution expression is as follows: ; where p m is the Bernoulli distribution probability; X m =1 means there is rainfall in month m; X m =0 means no rainfall in month m; m=1,2,...,12;

[0027] In the historical rainfall weight set of each month, a subset of historical monthly rainfall weights corresponding to rainfall events is selected, and a Gamma distribution model is constructed to simulate the probability distribution of the rainfall weight of that month. The expression of the Gamma distribution is: ;in is the monthly rainfall weight of the mth month; is the shape parameter of the Gamma distribution; is the scale parameter of the Gamma distribution;

[0028] Multiple sampling is performed based on the Bernoulli–Gamma joint distribution model to generate multiple sets of predicted monthly scale weight sequences covering 12 months of the year. The steps for a single sampling are as follows:

[0029] Sampling the mth month, first sample from the Bernoulli distribution corresponding to the mth month, if the collected X m =1, then sample from the Gamma distribution of the historical monthly rainfall weight subset corresponding to month m, and collect the monthly rainfall weight ; If the collected X m =0, then let ;

[0030] Repeat the collection 12 times until the monthly rainfall weights for the next 12 months are collected. , and Perform normalization so that the normalized The sum is 1; after normalization is the predicted monthly scale weight series;

[0031] The steps for predicting the hourly rainfall time series for all rainfall stations and meteorological stations in the next year in step S25 are as follows:

[0032] Calculate the ratio b of the monthly rainfall of the k representative samples of the target monthly rainfall time series to the monthly average rainfall of the basin to which the target lake belongs. The calculation formula of b is as follows: ; where q = 1, 2, 3, 4… k， When the qth target month rainfall time series represents the proportion b of the mth month in the nth year of the sample history; is the average monthly rainfall in the mth month of the nth year in the history of the basin to which the target lake belongs; The rainfall time series of the qth target month represents the rainfall of the mth month of the sample;

[0033] Then, based on the ratio b, the historical hourly rainfall of the historical measured meteorological data of all rain gauges and meteorological stations is scaled and mapped to obtain the hourly rainfall time series for the next year. The calculation formula is as follows: ;in Map the qth target month rainfall time series to the rainfall station or meteorological station t to represent the predicted rainfall at the hth hour of the mth month, day of the next year at the sample time; It is the historical rainfall at the hth hour on the dth day of the mth month of the nth year at the rainfall station or meteorological station t.

[0034] Furthermore, in step S4, the steps of constructing multiple annual-scale water level control strategies with differentiated regulation characteristics using a grid search method for the water level distribution range at the end of the month are as follows:

[0035] S41: Based on the monthly end-of-month water level distribution range of the target lake, extract the minimum and maximum values ​​of the monthly end-of-month water level distribution. The extraction expression is as follows: ;in is the historical lowest water level of the target lake at the end of the mth month; is the historical highest water level of the lake at the end of the mth month; N refers to the number of historical years for which lake characteristic data were collected;

[0036] S42: Latin hypercube sampling method is used in each month Multiple sampling is performed within the interval to generate multiple sets of annual water level control strategies covering 12 months of the year; the sampling expression is as follows: ;in Refers to the dispatching water level obtained at the rth sampling in the mth month; LHS refers to Latin hypercube sampling; Refers to the annual water level control strategy obtained from the rth sampling.

[0037] Furthermore, the multi-objective comprehensive evaluation formula in step S6 is: ; x is the number of all lake meteorological boundary files in the current input group whose lake water level time series after the lake hydrodynamic model simulation is within the set standard water level range throughout the year; y is the number of lake meteorological boundary files in the current input group whose downstream outflow flow time series after the lake hydrodynamic model simulation meets the maximum outflow capacity restriction condition of the outflow outlet throughout the whole process; z is the minimum value of the lake water level at the end of the year corresponding to the annual-scale simulation results of all lake hydrodynamic models in the current input group; z min The minimum value of the lake water level at the end of the year corresponding to the annual scale simulation results of the lake hydrodynamic model in all input files; max is the maximum value of the lake water level at the end of the year corresponding to the annual-scale simulation results of the lake hydrodynamic model in all input files; u is the total number of lake meteorological boundary files in step S2.

[0038] Based on the same inventive concept, the present invention also discloses an annual-scale robust scheduling logic generation system for lake water resources-water security collaboration, comprising:

[0039] The basin hydrology-lake hydrodynamic coupling modeling module is used to collect hydrogeographic data of the basin to which the target lake belongs and the lake body characteristic data, build a distributed basin hydrological model based on the hydrogeographic data, and use the lake body characteristic data and the output results of the distributed basin hydrological model to jointly build a lake hydrodynamic model;

[0040] A comprehensive meteorological uncertainty boundary file generation module is used to obtain annual-scale forecast rainfall data and historical measured meteorological data for the target lake basin. Scenario construction technology is used to construct multiple boundary scenarios representing meteorological uncertainty based on the annual-scale forecast rainfall data and historical measured meteorological data. Based on the multiple boundary scenarios, a basin meteorological boundary file for use in the distributed basin hydrological model and a lake meteorological boundary file for use in the lake hydrodynamic model are generated.

[0041] The watershed hydrological simulation module is used to input the generated multiple watershed meteorological boundary files into the distributed watershed hydrological model respectively, extract the time series of water entering the lake and the time series of water withdrawn from the lake output by the distributed watershed hydrological model after running the corresponding distributed watershed meteorological boundary file, and form a set of watershed inflow-withdrawal boundary files that cover meteorological uncertainties;

[0042] The lake water level control strategy generation module can calculate the monthly end-of-month water level distribution range of the target lake based on the water level monitoring data of the lake body's characteristic data. It uses a grid search method to construct multiple annual-scale water level control strategies with differentiated regulation characteristics for the end-of-month water level distribution range, and converts each annual-scale water level strategy into a water level control boundary file for the lake hydrodynamic model.

[0043] The lake hydrodynamic simulation module can form an initial combination of lake meteorological boundary files and basin inflow-water intake boundary files that are consistent with the historical measured meteorological data year. Several initial combinations are arranged and combined with several water level control boundary files to obtain multiple groups of lake hydrodynamic model input files. Each group of input files is input into the lake hydrodynamic model for annual scale simulation, and the lake water level time series and downstream outflow flow time series after the lake hydrodynamic model simulation corresponding to each group of input files are extracted;

[0044] The water level control strategy determination module can merge several groups of input files using the same annual-scale water level control strategy to obtain several input groups; based on the lake water level time series and downstream outflow flow time series data corresponding to each group of input files, and combined with lake water resources and water security-related indicators, the input group is evaluated comprehensively with multiple objectives to determine the input group with the best multi-objective comprehensive score, and the water level control boundary file used by the input group is extracted as the final annual-scale robust water level scheduling logic for the target lake.

[0045] Based on the same inventive concept, the present invention also discloses an electronic device, including one or more processors, one or more memories and one or more programs, wherein the programs are stored in the memories and configured to be executed by the processors. When the programs are loaded into the processors, the steps of the annual-scale robust scheduling logic generation method for lake water resources-water security coordination are implemented.

[0046] Based on the same inventive concept, the present invention also discloses a computer-readable storage medium, which stores a computer program. The computer program includes program instructions, which, when executed by a processor, enable the processor to execute the steps of a method for generating annual-scale robust scheduling logic for lake water resources-water security coordination.

[0047] Beneficial effects: Compared with the existing technology, the present invention has the following significant advantages: based on annual-scale forecast rainfall data and historical measured meteorological data, the present invention takes into account the spatiotemporal heterogeneity of rainfall and generates rainfall uncertainty intervals, which is conducive to improving the adaptability and robustness of the final annual-scale robust water level scheduling logic to climate change from the source; a distributed watershed hydrological model is used to simulate the runoff production and convergence process in the basin, and uncertainty intervals of basin inflow and water intake are generated, which systematically considers basin uncertainty factors such as underlying surface changes, which is conducive to improving the accuracy and stability of the final annual-scale robust water level scheduling logic.

[0048] Based on historical measured meteorological data, the present invention generates uncertainty intervals for monthly controlled water levels and adopts the Latin hypercube method for sampling to obtain a dynamic water level control method suitable for a variety of scenarios. It no longer relies on manual experience and a small number of preset scenarios, which is conducive to improving the reliability of the robust water level scheduling logic at the final annual scale.

[0049] The present invention comprehensively evaluates the water level fluctuations throughout the year, the maximum outflow capacity of the lake outlet and the lake water level at the end of the year, and screens the water level dynamic control scheme, which is conducive to screening the final annual-scale robust water level scheduling logic with strong applicability and high stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 is a flow chart of the method of the present invention;

[0051] Figure 2 This is a Gamma distribution diagram of the monthly sampling set for January in an embodiment of the present invention;

[0052] Figure 3 This is a Gamma distribution diagram of the monthly sampling set for February in an embodiment of the present invention;

[0053] Figure 4 This is a Gamma distribution diagram of the monthly sampling set for March in an embodiment of the present invention;

[0054] Figure 5 This is a Gamma distribution diagram of the monthly sampling set for April in an embodiment of the present invention;

[0055] Figure 6 This is a Gamma distribution diagram of the monthly sampling set for May in an embodiment of the present invention;

[0056] Figure 7 This is a gamma distribution diagram of the monthly sampling set for June in an embodiment of the present invention;

[0057] Figure 8 This is a Gamma distribution diagram of the monthly sampling set for July in an embodiment of the present invention;

[0058] Figure 9 This is a Gamma distribution diagram of the monthly sampling set for August in an embodiment of the present invention;

[0059] Figure 10 This is a gamma distribution diagram of the monthly sampling set for September in an embodiment of the present invention;

[0060] Figure 11 Gamma distribution diagram of the monthly sampling set for October in an embodiment of the present invention;

[0061] Figure 12 This is a gamma distribution diagram of the monthly sampling set for November in an embodiment of the present invention;

[0062] Figure 13 This is a gamma distribution diagram of the monthly sampling set for December in an embodiment of the present invention;

[0063] Figure 14This is a time series sample distribution diagram of target monthly rainfall for the first representative scenario of an embodiment of the present invention;

[0064] Figure 15 This is a time series sample distribution diagram of target monthly rainfall for the second representative scenario of an embodiment of the present invention;

[0065] Figure 16 This is a time series sample distribution diagram of target monthly rainfall for the third representative scenario of an embodiment of the present invention;

[0066] Figure 17 This is a time series sample distribution diagram of target monthly rainfall for the fourth representative scenario of an embodiment of the present invention;

[0067] Figure 18 This is a time series sample distribution diagram of target monthly rainfall for the fifth representative scenario of an embodiment of the present invention;

[0068] Figure 19 This is a time series sample distribution diagram of target monthly rainfall for the sixth representative scenario of an embodiment of the present invention;

[0069] Figure 20 This is a time series sample distribution diagram of target monthly rainfall for the seventh representative scenario of an embodiment of the present invention;

[0070] Figure 21 This is a time series sample distribution diagram of target monthly rainfall for the eighth representative scenario of an embodiment of the present invention;

[0071] Figure 22 This is a time series sample distribution diagram of target monthly rainfall for the ninth representative scenario of an embodiment of the present invention;

[0072] Figure 23 This is a time series sample distribution diagram of target monthly rainfall for the 10th representative scenario of an embodiment of the present invention;

[0073] Figure 24 This is a water level control interval diagram for Lake E in December according to an embodiment of the present invention;

[0074] Figure 25 A multi-objective comprehensive score graph of an input group according to an embodiment of the present invention;

[0075] Figure 26 Schematic diagram of the structure of the system of the present invention. DETAILED DESCRIPTION

[0076] The technical solution of the present invention will be further described below with reference to the accompanying drawings.

[0077] Example 1: The present invention discloses a method for generating annual scale robust scheduling logic for lake water resources and water security collaboration, such as Figure 1As shown, the following steps are included:

[0078] S1: Collect hydrogeographic data from the target lake's basin and lake-body characteristic data. A distributed basin hydrological model is constructed based on the hydrogeographic data. The lake-body characteristic data and the output of the distributed basin hydrological model are then used to construct a lake hydrodynamic model. The distributed basin hydrological model and the lake hydrodynamic model are linked through bidirectional calibration. An iterative feedback mechanism is employed. When significant deviations in the lake hydrodynamic simulation results are detected, the sensitive parameters of the distributed basin hydrological model are reversely corrected, forming a bidirectional coupled calibration loop. This effectively addresses the boundary condition distortion problem in traditional single-model calibration and provides a more reliable model for annual lake operation.

[0079] The basin hydrological and geographical data include the surface elevation data, soil type, land use, hourly meteorological data, measured hydrological data of the basin, spatial distribution information of drainage and water intake facilities, and their operating flow time series data of the basin to which the lake belongs; the lake body characteristic data include the underwater three-dimensional terrain data of the lake, lake area meteorological data, water level monitoring data, and continuous monitoring data of inflow and outflow.

[0080] When constructing a distributed watershed hydrological model within existing watershed hydrological modeling software, the watershed's surface elevation data is used to divide the watershed into sub-basins and sub-river channels. Each sub-basin is divided into different hydrological response units based on soil type, land use, and surface elevation data. Boundary conditions are defined based on hourly meteorological data, measured watershed hydrological data, the spatial distribution of drainage and water intake facilities, and their operational flow time series data to construct the distributed watershed hydrological model. Existing watershed hydrological modeling software includes but is not limited to SWAT, MIKE, and LSPC.

[0081] When constructing a lake hydrodynamic model using existing lake hydrodynamic modeling software, a grid is constructed based on the lake's underwater three-dimensional topographic data. Boundary conditions are defined based on lake area meteorological data, continuous monitoring data on inflow and outflow, and the output of a distributed watershed hydrological model to construct the lake hydrodynamic model. The output of the distributed watershed hydrological model refers to the time series of water entering the lake and the time series of water withdrawn from the lake after the distributed watershed hydrological model is run under the current conditions. The inflow and outflow monitoring data refers to the time series of water flowing into and out of the lake, excluding the time series of water flowing into the lake and the time series of water withdrawn from the lake. Existing lake hydrodynamic modeling software includes, but is not limited to, EFDC, Delft3D, and MIKE; and any zero-dimensional, one-dimensional, two-dimensional, or three-dimensional lake hydrodynamic modeling software is acceptable.

[0082] The distributed basin hydrological model uses the same time series of watershed hydrogeographic data as the lake-body characteristic data used to construct the lake hydrodynamic model. This data spans the same historical year or time interval. The output of the distributed basin hydrological model, as part of the lake hydrodynamic model, establishes a hydraulic coupling relationship between the basin and the lake, effectively addressing the uncertainty of nonlinear rainfall-runoff conversion caused by differences in basin characteristics.

[0083] S2: Obtain the annual-scale forecast rainfall data and historical measured meteorological data of the watershed to which the target lake belongs. Use scenario construction technology to construct multiple boundary scenarios that represent meteorological uncertainty based on the annual-scale forecast rainfall data and historical measured meteorological data. Based on the multiple boundary scenarios, generate the watershed meteorological boundary file for use in the distributed watershed hydrological model and the lake meteorological boundary file for use in the lake hydrodynamic model.

[0084] The annual-scale forecast rainfall data refers to the monthly forecast rainfall for the next year predicted by all forecast stations in the basin to which the target lake belongs; the historical measured meteorological data refers to the many-year historical hourly rainfall data of all rainfall stations in the basin to which the target lake belongs and the many-year historical hourly multi-index meteorological observation data of all meteorological stations.

[0085] The steps to generate a watershed meteorological boundary file for use in a distributed watershed hydrological model and a lake meteorological boundary file for use in a lake hydrodynamic model are as follows:

[0086] S21: Based on the monthly forecast rainfall of each forecast station for the next year, the total annual rainfall of each forecast station is calculated, and all the total annual rainfall is disturbed and expanded to obtain multiple total annual rainfall samples.

[0087] The annual rainfall data for all forecast stations within the target lake's basin are perturbed upward and downward by a specified amplitude. After the perturbation expansion, each forecast station will obtain three annual rainfall samples; the total number of annual rainfall samples obtained after the perturbation expansion is three times the total number of all forecast stations. Preferably, in this embodiment, the specified amplitude is 20%, i.e., a 20% downward perturbation, no perturbation, and a 20% upward perturbation are performed.

[0088] The perturbation formula for any forecast site s is as follows:

[0089] A 20% downward disturbance: ;

[0090] No disturbance: ;

[0091] 20% upward disturbance: ;

[0092] Among them, Ps Refers to the total annual rainfall predicted by forecast station s.

[0093] S22: Based on the historical hourly rainfall data and historical hourly multi-index meteorological observation data, the weight of the historical monthly rainfall at each rainfall station and meteorological station in the total annual rainfall is calculated to form a historical monthly scale weight sequence. Based on the historical monthly scale weight sequence, several groups of predicted monthly scale weight sequences covering the 12 months of the year are generated.

[0094] The historical monthly scale weight series refers to the proportion of the rainfall in December of a certain year at a rain gauge or meteorological station to the total rainfall of that year.

[0095] The calculation formula for the historical monthly rainfall weight of any rainfall station or meteorological station t in a certain month is as follows: ;in is the monthly rainfall weight of rainfall station or meteorological station t in month m of historical year n; is the rainfall at rainfall station or meteorological station t in month m of year n; is the total rainfall at rainfall station or meteorological station t in the historical year n. Constitute the historical monthly scale weight sequence of the historical nth year of rainfall station or meteorological station t.

[0096] The method of generating several sets of forecast monthly weight series containing 12 months of a year is as follows:

[0097] All historical monthly weight sequences of each rainfall station and meteorological station in the historical measured meteorological data are classified by month to form 12 historical rainfall weight sets; for example The historical monthly rainfall weights with the same value in month m constitute the historical rainfall weight set corresponding to month m;

[0098] For each month's historical rainfall weight set, a Bernoulli distribution model is constructed based on the binary event results of whether rainfall occurs in that month at each rainfall station and meteorological station recorded in historical meteorological data to simulate the rainfall probability of that month. The Bernoulli distribution expression is as follows: ; where p m is the Bernoulli distribution probability; X m =1 means there is rainfall in month m; X m =0 means no rainfall in month m; m=1,2,...,12;

[0099] In the historical rainfall weight set of each month, a subset of historical monthly rainfall weights corresponding to rainfall events is selected, and a Gamma distribution model is constructed to simulate the probability distribution of the rainfall weight of that month. The expression of the Gamma distribution is: ;in is the monthly rainfall weight of the mth month; is the shape parameter of the Gamma distribution; is the scale parameter of the Gamma distribution;

[0100] Multiple sampling is performed based on the Bernoulli–Gamma joint distribution model to generate multiple sets of predicted monthly scale weight sequences covering 12 months of the year. The steps for a single sampling are as follows:

[0101] Sampling the mth month, first sample from the Bernoulli distribution corresponding to the mth month, if the collected X m =1, then sample from the Gamma distribution of the historical monthly rainfall weight subset corresponding to month m, and collect the monthly rainfall weight ; If the collected X m =0, then let .

[0102] Repeat the collection 12 times until the monthly rainfall weights for the next 12 months are collected. , and Perform normalization so that the normalized The sum is 1; after normalization is the predicted monthly scale weight series.

[0103] In practical applications, before normalization, first calculate Is the sum of 1? If the sum is 1, then the Constitute a forecast monthly scale weight sequence; if the sum is not 1, then After normalization, the obtained is a forecast monthly scale weight sequence.

[0104] The normalization process is as follows:

[0105]

[0106] After the above treatment , is a forecast monthly scale weight sequence.

[0107] The historical monthly rainfall weights are divided into a set of historical rainfall weights and a subset of historical monthly rainfall weights based on rainfall and non-rainfall scenarios. The set of historical rainfall weights is sampled first, followed by the subset. This two-stage setup aligns with actual physical processes: rainfall occurrence and rainfall amount are two distinct mechanisms. During single sampling, samples are first collected from the set of historical rainfall weights to ensure the possibility of dry months is included. A gamma distribution is used to fit rainfall in the presence of rainfall. The shape and scale parameters of the gamma distribution can be flexibly adjusted to match the mean, variance, and skewness of the actual data. Monte Carlo sampling is the preferred single sampling method. Monte Carlo sampling generates multiple sets of predicted monthly scale weight sequences through a large number of random simulations. This method preserves the statistical regularities of the historical data (such as mean, variance, and probability of extreme values) while also accounting for possible random fluctuations, ultimately providing a physically sound and statistically robust data foundation. The predicted monthly scale weight sequence is obtained by combining the historical rainfall weight set for both cases with and without rainfall, the historical monthly rainfall weight subset for cases with rainfall, the Gamma distribution and the Monte Carlo sampling method. It not only conforms to the actual statistical characteristics of rainfall, but also quantifies the uncertainty of the historical monthly rainfall weights.

[0108] S23: The annual rainfall total sample obtained in step S21 and the predicted monthly scale weight sequence obtained in step S22 are arranged and combined to obtain several annual rainfall total sample-prediction monthly scale weight sequence combinations, and the monthly rainfall of each month in the annual rainfall total sample-prediction monthly scale weight sequence combination constitutes a monthly rainfall time series sample.

[0109] Monthly rainfall time series samples The expression is as follows:

[0110]

[0111] in is the monthly rainfall time series sample composed of the i-th annual rainfall total sample and the j-th predicted monthly scale weight sequence, is the monthly rainfall in the mth month after the i-th annual rainfall total sample is combined with the j-th predicted monthly scale weight sequence; is the total annual rainfall of the i-th annual rainfall sample; is the weight of the mth month in the scale weight sequence of the jth forecast month.

[0112] S24: A clustering method is used to cluster all monthly rainfall time series samples, and k types of typical meteorological scenarios are obtained by clustering. The typical meteorological scenarios are used as representative samples of the target rainfall time series.

[0113] The Kmeans clustering method is used to screen out k typical meteorological scenarios with similar characteristics as the target monthly rainfall time series samples. Kmeans clustering can cluster all monthly rainfall time series samples into k classes, minimizing the intra-class differences and maximizing the inter-class differences.

[0114] S25: Combine k representative samples of target monthly rainfall time series with historical measured meteorological data to predict hourly rainfall time series for all rainfall stations and meteorological stations in the next year.

[0115] The steps to predict hourly rainfall time series for all rainfall stations and meteorological stations for the next year are as follows:

[0116] Calculate the ratio b of the monthly rainfall of the k representative samples of the target monthly rainfall time series to the monthly average rainfall of the basin to which the target lake belongs. The calculation formula of b is as follows: ; where q = 1, 2, 3, 4… k， When the qth target month rainfall time series represents the proportion b of the mth month in the nth year of the sample history; is the average monthly rainfall in the mth month of the nth year in the history of the basin to which the target lake belongs; The rainfall time series of the qth target month represents the rainfall of the mth month of the sample;

[0117] Then, based on the ratio b, the historical hourly rainfall of the historical measured meteorological data of all rain gauges and meteorological stations is scaled and mapped to obtain the hourly rainfall time series for the next year. The calculation formula is as follows: ;in Map the qth target month rainfall time series to the rainfall station or meteorological station t to represent the predicted rainfall at the hth hour of the mth month, day of the next year at the sample time; It is the historical rainfall at the hth hour on the dth day of the mth month of the nth year at the rainfall station or meteorological station t.

[0118] The inherent spatial and temporal differences between rain gauges and meteorological stations are maintained by using historical hourly rainfall data from these stations. The coefficient b is used to capture the overall deviation (e.g., excess / deficient) of the predicted rainfall from historically measured meteorological data across the watershed. This overall deviation is then dynamically distributed to each rain gauge or meteorological station using the coefficient b according to the historical spatial pattern, ensuring that rainfall at these stations changes synchronously with the forecast. The purpose of using the coefficient b to calculate the hourly rainfall time series is to map the target monthly rainfall total at the watershed level of the lake to the hourly rainfall at the rain gauges and meteorological stations, maintaining temporal and spatial consistency and appropriately reflecting forecast changes.

[0119] S26: The hourly rainfall time series from all rainfall and meteorological stations within the target lake's basin, as predicted in step S25, are converted into a basin meteorological boundary file in a format consistent with the input format of the distributed basin hydrological model. The hourly rainfall time series from rainfall and meteorological stations within the designated area of ​​the target lake are converted into a lake meteorological boundary file in a format consistent with the input format of the lake hydrodynamic model. In actual use, the designated area of ​​the lake can be determined based on actual needs, such as an area 20 km from the center of the lake.

[0120] Preferably, before obtaining the basin meteorological boundary file and the lake boundary meteorological file, the hourly rainfall time series of the rainfall station and the meteorological station are compared with the The corresponding historical meteorological data of the nth year, excluding the historical rainfall, is combined and converted into the required basin meteorological boundary file or lake meteorological boundary file. When calculating the historical rainfall at the hth hour of the mth month in 2019 at rain gauge station or meteorological station t, it is necessary to remove the historical hourly meteorological data of the rain gauge station or meteorological station t in 2019 from the historical rainfall.

[0121] When obtaining a watershed meteorological boundary file and a lake meteorological boundary file with comprehensive meteorological uncertainty, the present invention first performs an upper and lower disturbance on the predicted annual rainfall data based on the forecasted annual rainfall total for the next year to generate a predicted annual rainfall total sample covering the possibility of both relatively abundant and relatively dry weather. Secondly, based on the historical monthly rainfall weights, a Bernoulli distribution is used to fit the distribution of rainfall with or without rainfall, and a Gamma distribution is further used to fit the distribution of rainfall weight when rainfall occurs. Finally, a Monte Carlo sampling method is used to obtain multiple groups of predicted monthly scale weight sequences covering the 12 months of the next year. Then, the predicted annual rainfall total sample is combined with the predicted monthly scale weight sequence to obtain several groups of monthly rainfall time series samples. A clustering method is used to cluster all the monthly rainfall time series samples to obtain representative target monthly rainfall time series samples. Finally, based on the monthly rainfall of the representative target monthly rainfall time series samples and the monthly average rainfall of the watershed to which the lake belongs, the target monthly rainfall time series samples are mapped to meteorological stations and rainfall stations in the watershed to obtain multiple groups of hourly rainfall time series for all rainfall stations and meteorological stations in the next year. By combining future forecasted annual rainfall totals with fluctuations in historical observed meteorological data, a probabilistic annual rainfall total ensemble encompassing wet (abundant), dry (drought), and normal years is generated, reflecting the uncertainty of long-term climate forecasts. Annual total samples (e.g., "abundant years") are dynamically matched with monthly weighted samples (e.g., "delayed rainy season") to generate physically consistent monthly rainfall time series (e.g., "abundant years + concentrated rainy season = flood risk scenario"). Clustering is used to extract typical rainfall patterns (e.g., "uniform distribution" and "concentrated heavy rainfall"), reducing computational effort while preserving key extreme scenarios. Based on the spatial relationship between monthly totals and basin-average rainfall, combined with topography and historical station weights, monthly values ​​are decomposed into hourly rainfall at each station (e.g., spatial displacement of the center of a heavy rainfall event). Finally, a ensemble of probabilistic rainfall scenarios is generated for the next year, at hourly scales, and with spatial distribution, accounting for climate forecast errors, intra-year distribution uncertainty, and spatial variability.

[0122] The present invention fully considers meteorological uncertainty factors and is no longer limited to a single meteorological scenario. Instead, it generates rainfall uncertainty intervals through multi-source data of historical measured meteorological data and forecast rainfall data, combined with multiple optimization methods, to obtain hourly-scale rainfall time series for a large number of rainfall stations and meteorological stations in the next year, systematically reflecting the diversity of annual rainfall totals, rainfall types, and spatial distributions, and fundamentally improving the adaptability and robustness of the final annual-scale robust water level scheduling logic to climate change.

[0123] S3: Input the generated multiple watershed meteorological boundary files into the distributed watershed hydrological model respectively, extract the time series of water volume entering the lake and the time series of water volume withdrawn from the lake output by the distributed watershed hydrological model corresponding to each watershed meteorological boundary file, and form a set of watershed inflow-water withdrawal boundary files covering meteorological uncertainties.

[0124] Preferably, in actual application, the time series of water volume entering the lake from each tributary in the basin and the time series of water volume taken from the lake by the basin are extracted from the results output after the distributed watershed hydrological model is run; then these data are integrated and processed, and the water volume of multiple tributaries entering the lake is accumulated to generate a total inflow flow time series, and the water volume of each water intake is summarized to form a total water intake time series; finally, according to the input requirements of the lake hydrodynamic model, the processed inflow and water intake data are converted into a standardized boundary file format to provide the lake hydrodynamic model with the basin inflow and water intake boundary files.

[0125] Preferably, in actual application, after the distributed watershed hydrological model in step S1 is constructed, multiple identical distributed watershed hydrological models can be copied, and the number of distributed watershed hydrological models is consistent with the number of watershed meteorological boundary files. The multiple watershed meteorological boundary files are respectively input into different distributed watershed hydrological models, and multiple distributed watershed hydrological models are driven at the same time, which is conducive to improving the overall operation efficiency.

[0126] The present invention fully reflects the complexity of the watershed system in practical applications. It simulates the runoff generation and confluence process under different meteorological scenarios through a distributed watershed hydrological model, captures the nonlinear relationship between rainfall and runoff, and systematically considers watershed uncertainty factors such as underlying surface changes. Compared with the traditional method of calculating watershed inflow by multiplying a single or several sets of preset rainfall by a fixed runoff coefficient, the present invention effectively makes up for the problem that the traditional method is insufficient in expressing the dynamic response of the watershed.

[0127] S4: Based on the water level monitoring data of the lake body characteristic data, the water level distribution range of the target lake at the end of each month is statistically calculated. The grid search method is used to construct multiple annual-scale water level control strategies with differentiated regulation characteristics for the water level distribution range at the end of each month, and each annual-scale water level strategy is converted into a water level control boundary file of the lake body hydrodynamic model.

[0128] The steps for constructing multiple annual-scale water level control strategies with differentiated regulation characteristics using a grid search method for the water level distribution range at the end of the month are as follows:

[0129] S41: Based on the monthly end-of-month water level distribution range of the target lake, extract the minimum and maximum values ​​of the monthly end-of-month water level distribution. The minimum and maximum values ​​are used to determine the feasible boundary area of ​​the water level scheduling in the corresponding month. The extraction expression is as follows: ;in is the historical lowest water level of the target lake at the end of the mth month; is the historical highest water level of the lake at the end of the mth month; N refers to the number of historical years for which lake characteristic data were collected.

[0130] S42: Latin hypercube sampling method is used in each month Multiple sampling is performed within the interval to generate multiple sets of annual water level control strategies covering 12 months of the year; the sampling expression is as follows: ;in Refers to the dispatching water level obtained at the rth sampling in the mth month; LHS refers to Latin hypercube sampling; Refers to the annual water level control strategy obtained from the rth sampling.

[0131] This method combines historical end-of-month water level distribution data with Latin hypercube sampling to generate year-round lake water level scenarios that take into account historical fluctuations. First, the extreme end-of-month water levels in historical water level monitoring data are statistically analyzed to determine the seasonal range of water level fluctuations and historical extremes. Then, the Latin hypercube sampling (LHS) method is used to efficiently sample within the extreme value interval of each month to generate a large number of annual-scale water level control strategies. This method preserves the inter-month correlation of water level changes while ensuring that the sampling results cover the historical extreme value range, thereby systematically capturing possible water level fluctuation scenarios. The generated annual-scale water level control strategies can be further used to evaluate water resource scheduling strategies, ecological impacts, and flood and drought control measures under different hydrological conditions, providing a probabilistic decision-making basis for comprehensive lake management.

[0132] This invention breaks through the limitations of traditional plans that rely on a small number of scenarios and manual experience. It generates large-scale annual water level control strategies based on methods such as Latin hypercube sampling. It can explore the optimal scheduling strategy that takes into account both water resource utilization and water security in massive scenarios, and can greatly improve the reliability and scientific nature of the annual water level control strategy.

[0133] S5: The lake meteorological boundary file and the basin inflow-water intake boundary file with the same historical measured meteorological data year are used to form an initial combination. Several initial combinations are arranged and combined with several water level control boundary files to obtain multiple groups of input files for the lake hydrodynamic model. Each group of input files is input into the lake hydrodynamic model for annual-scale simulation, and the lake water level time series and downstream outlet flow time series after the lake hydrodynamic model simulation corresponding to each group of input files are extracted.

[0134] Preferably, in actual application, after the lake hydrodynamic model in step S1 is constructed, multiple identical lake hydrodynamic models can be copied, and the number of lake hydrodynamic models is consistent with the number of input files. Multiple input files are input into different lake hydrodynamic models respectively, and multiple lake hydrodynamic models are driven at the same time, which is conducive to improving the overall operating efficiency.

[0135] This invention constructs multiple sets of boundary conditions with temporal and spatial consistency and physical rationality to drive the lake hydrodynamic model to carry out large-scale multi-scenario simulations, comprehensively support the water resource scheduling work of the lake under the background of future climate change, and provide a systematic, probabilistic and operational scientific basis for decision-making.

[0136] Using the output of the distributed watershed hydrological model, the basin inflow and water withdrawal boundary files, as input files, couples basin hydrological processes with lake hydrodynamic processes, enabling a more systematic simulation of the water cycle. Combining the lake meteorological boundary file, the basin inflow-water withdrawal boundary file, and the water level control boundary file as input files allows for coupling meteorological uncertainty, nonlinear basin response models, and large-scale water level control schemes. This optimizes the entire chain from rainfall forecasting to water level control, significantly improving the accuracy and robustness of the resulting year-round water level control scheme.

[0137] S6: Merge several groups of input files using the same annual-scale water level control strategy to obtain several input groups; conduct a multi-objective comprehensive evaluation of the input groups based on the lake water level time series and downstream outflow flow time series data corresponding to each group of input files, and combine them with lake water resources and water security-related indicators to determine the input group with the best multi-objective comprehensive score, and extract the water level control boundary file used by this input group as the final annual-scale robust water level scheduling logic for the target lake.

[0138] The multi-objective scoring formula is: ; x is the number of all lake meteorological boundary files in the current input group whose lake water level time series after the lake hydrodynamic model simulation is within the set standard water level range throughout the year; y is the number of lake meteorological boundary files in the current input group whose downstream outflow flow time series after the lake hydrodynamic model simulation meets the maximum outflow capacity restriction condition of the outflow outlet throughout the whole process; z is the minimum value of the lake water level at the end of the year corresponding to the annual-scale simulation results of all lake hydrodynamic models in the current input group; z min The minimum value of the lake water level at the end of the year corresponding to the annual scale simulation results of the lake hydrodynamic model in all input files; max is the maximum year-end lake water level corresponding to the annual-scale simulation results of the lake hydrodynamic model from all input files; u is the total number of lake meteorological boundary files from step S2. x, y, and z in the multi-objective scoring formula represent indicators related to lake water resources and water security. In practice, the standard water level range and maximum outflow capacity constraints of the outflow outlet can be set based on actual conditions.

[0139] This step selects an annual-scale robust scheduling logic that is stable under different meteorological conditions through multi-objective scoring. The selected annual-scale robust scheduling logic has the following advantages: the setting of x can ensure that the water level fluctuation throughout the year is within the allowable ecological / safe water level range, avoiding the risk of exceeding the limit; the setting of y can strictly monitor the outflow flow, ensure the ecological base flow, and at the same time not exceed the maximum outflow capacity, preventing the dam from overloading or downstream flooding risks; z, z min and z max The system prioritizes options with higher year-end water levels to ensure water supply and ecological needs the following year, while also assessing their adaptability under different hydrological years (flood, normal, and dry). The resulting dynamic water level control strategy provides a quantitative basis for decision-making in actual annual water resource scheduling. The optimal annual regulation strategy maintains water level safety while balancing water storage benefits in most scenarios, demonstrating high adaptability and scalability across a wide range of meteorological conditions.

[0140] The present invention can optimize the entire process from rainfall forecasting to water level control, greatly improving the accuracy, adaptability and robustness of lake water level regulation, and has important engineering application value and promotion potential.

[0141] Example 2: This example discloses the steps of obtaining the water level control strategy of Lake E using the annual-scale robust scheduling logic generation method for lake water resources-water security collaboration in Example 1.

[0142] S1: Collect the hydrogeographic data of the basin to which Lake E belongs and the lake body characteristic data of Lake E, build a distributed basin hydrological model based on the hydrogeographic data, and use the lake body characteristic data and the output results of the distributed basin hydrological model to jointly build a lake hydrodynamic model.

[0143] The basin hydrological and geographical data include the surface elevation data, soil type, land use, hourly meteorological data, measured hydrological data of the basin, spatial distribution information of drainage and water intake facilities, and their operating flow time series data of the basin to which the lake belongs; the lake body characteristic data include the underwater three-dimensional terrain data of the lake, lake area meteorological data, water level monitoring data, and continuous monitoring data of inflow and outflow.

[0144] A distributed basin hydrological model was constructed using the Intelliway-WS model software, a watershed hydrological and water quality model. The sub-basins and sub-rivers were divided based on the surface elevation data of the basin. Each sub-basin was divided into different hydrological response units according to soil type, land use and surface elevation data. Boundary conditions were defined based on hourly meteorological data, measured hydrological data of the basin, spatial distribution information of drainage and water intake facilities, and their operating flow time series data to construct a distributed basin hydrological model.

[0145] The lake hydrodynamic model was constructed using the IWIND-LR hydrodynamic-water-quality-water-ecological modeling software. A grid was constructed based on the lake's underwater three-dimensional topographic data. Boundary conditions were defined based on lake meteorological data, continuous monitoring data on inflows and outflows, and the output of a distributed watershed hydrological model. The distributed watershed hydrological model output refers to the time series of water entering and withdrawing from the lake after the watershed hydrological model is run under the current conditions. The inflow and outflow monitoring data refers to the time series of water flowing into and out of the lake, in addition to the time series of water flowing into and out of the lake.

[0146] The time series of the watershed data for constructing the distributed watershed hydrological model and the lake data for constructing the lake hydrodynamic model are consistent, that is, the watershed data and lake data of the same historical year or the same historical interval are taken.

[0147] S2: Obtain the annual-scale forecast rainfall data and historical measured meteorological data of the watershed to which the target lake belongs. Use scenario construction technology to construct multiple boundary scenarios that represent meteorological uncertainty based on the annual-scale forecast rainfall data and historical measured meteorological data. Based on the multiple boundary scenarios, generate the watershed meteorological boundary file for use in the distributed watershed hydrological model and the lake meteorological boundary file for use in the lake hydrodynamic model.

[0148] The annual-scale forecast rainfall data refers to the monthly forecast rainfall for the next year predicted by three forecast stations in the basin to which Lake E belongs; the historical measured meteorological data refers to the 11-year historical hourly meteorological data from 104 rainfall stations and 7 meteorological stations in the basin to which Lake E belongs.

[0149] The steps to generate a watershed meteorological boundary file for use in a distributed watershed hydrological model and a lake meteorological boundary file for use in a lake hydrodynamic model are as follows:

[0150] S21: Based on the monthly forecast rainfall of each forecast station for the next year, calculate the total annual rainfall of the three forecast stations, and perform disturbance expansion on all the total annual rainfall to obtain multiple total annual rainfall samples.

[0151] The annual rainfall data of three forecast stations in the basin of Lake E were disturbed by 20% up and down. After the disturbance expansion, a total of 9 annual rainfall samples were obtained.

[0152] The perturbation formula for any forecast station s is as follows: 20% downward perturbation: ; Do not disturb: ; Upward 20% disturbance: ; where P s Refers to the total annual rainfall predicted by forecast station s.

[0153] S22: Based on historical hourly rainfall data and historical hourly multi-index meteorological observation data, the weights of historical monthly rainfall at 104 rainfall stations and 7 meteorological stations in the historical 11 years to the total annual rainfall are calculated, forming 1221 sets of historical monthly scale weight sequences covering 12 months of a year. Based on the obtained 1221 sets of historical monthly scale weight sequences, several sets of predicted monthly scale weight sequences covering 12 months of a year are predicted and generated.

[0154] The historical monthly scale weight series refers to the proportion of the rainfall in December of a certain year at a rainfall station or meteorological station to the total rainfall of that year.

[0155] The calculation formula for the historical monthly rainfall weight of any rainfall station or meteorological station t in a certain month is as follows: ;in is the monthly rainfall weight of rainfall station or meteorological station t in month m of historical year n; is the rainfall at rainfall station or meteorological station t in month m of year n; is the total rainfall at rainfall station or meteorological station t in the historical year n. Constitute the historical monthly scale weight sequence of the historical nth year of rainfall station or meteorological station t.

[0156] The method of generating several sets of forecast monthly weight series containing 12 months of a year is as follows:

[0157] All historical monthly weight sequences of each rainfall station and meteorological station in the historical measured meteorological data are classified by month to form 12 historical rainfall weight sets; for example The historical monthly rainfall weights with the same value in month m constitute the historical rainfall weight set corresponding to month m;

[0158] For each month's historical rainfall weight set, a Bernoulli distribution model is constructed based on the binary event results of whether rainfall occurs in that month at each rainfall station and meteorological station recorded in historical meteorological data to simulate the rainfall probability of that month. The Bernoulli distribution expression is as follows: ; where p m is the Bernoulli distribution probability; X m =1 means there is rainfall in month m; X m =0 means no rainfall in month m; m=1,2,...,12;

[0159] In the historical rainfall weight set of each month, a subset of historical monthly rainfall weights corresponding to rainfall events is selected, and a Gamma distribution model is constructed to simulate the probability distribution of the rainfall weight of that month. The expression of the Gamma distribution is: ;in is the monthly rainfall weight of the mth month; is the shape parameter of the Gamma distribution; is the scale parameter of the Gamma distribution. The Gamma distribution of the 12 historical monthly rainfall weight subsets is as follows: Figures 2 to 13 shown.

[0160] Multiple sampling is performed based on the Bernoulli–Gamma joint distribution model to generate multiple sets of predicted monthly scale weight sequences covering 12 months of the year. The steps for a single sampling are as follows:

[0161] Sampling the mth month, first sample from the Bernoulli distribution corresponding to the mth month, if the collected X m =1, then sample from the Gamma distribution of the historical monthly rainfall weight subset corresponding to month m, and collect the monthly rainfall weight ; If the collected X m =0, then let .

[0162] Repeat the collection 12 times until the monthly rainfall weights for the next 12 months are collected. , and Perform normalization so that the normalized The sum is 1; after normalization is the predicted monthly scale weight series.

[0163] In practical applications, before normalization, first calculate Is the sum of 1? If the sum is 1, then the Constitute a forecast monthly scale weight sequence; if the sum is not 1, then After normalization, the obtained is a forecast monthly scale weight sequence.

[0164] S23: Arrange and combine the 9 annual rainfall total samples obtained in S21 and the 1221 predicted monthly scale weight sequences obtained in step S22 to obtain multiple annual rainfall total sample-prediction monthly scale weight sequence combinations, and the monthly rainfall of each month in the annual rainfall total sample-prediction monthly scale weight sequence combination constitutes a monthly rainfall time series sample.

[0165] Monthly rainfall time series samples The expression is as follows:

[0166]

[0167] in is the monthly rainfall time series sample composed of the i-th annual rainfall total sample and the j-th predicted monthly scale weight sequence, is the monthly rainfall in the mth month after the i-th annual rainfall total sample is combined with the j-th predicted monthly scale weight sequence; is the total annual rainfall of the i-th annual rainfall sample; is the weight of the mth month in the scale weight sequence of the jth forecast month.

[0168] S24: Clustering method is used to cluster all monthly rainfall time series samples, and 10 typical meteorological scenarios are obtained by clustering. The typical meteorological scenarios are used as representative samples of the target rainfall time series.

[0169] The time series distribution of monthly rainfall of 10 representative scenarios obtained by Kmeans clustering algorithm is as follows: Figures 14 to 23 As shown, Figures 14 to 23 The median of the monthly rainfall weight in each cluster category and the range of 25%-75% are given in Figures 14 to 23 The "total * samples" means that the cluster contains * months of rainfall time series samples.

[0170] S25: Combine the representative samples of rainfall time series of 10 target months and historical measured meteorological data to predict the hourly rainfall time series of all rainfall stations and meteorological stations in the next year.

[0171] Calculate the ratio b of the monthly rainfall of the k representative samples of the target monthly rainfall time series to the monthly average rainfall of the basin to which the target lake belongs. The calculation formula of b is as follows: ; where q = 1, 2, 3, 4… k， When the qth target month rainfall time series represents the proportion b of the mth month in the nth year of the sample history; is the average monthly rainfall in the mth month of the nth year in the history of the basin to which the target lake belongs; The mth month rainfall of the representative sample of the qth target month rainfall time series is obtained by combining 11 years of historical meteorological data with the 10 target month rainfall time series samples. A total of 11*10=110 groups of hourly rainfall time series samples for rainfall stations and meteorological stations in the E Lake basin are obtained.

[0172] Then, based on the ratio b, the historical hourly rainfall of the historical measured meteorological data of all rain gauges and meteorological stations is scaled and mapped to obtain the hourly rainfall time series for the next year. The calculation formula is as follows: ;in Map the qth target month rainfall time series to the rainfall station or meteorological station t to represent the predicted rainfall at the hth hour of the mth month, day of the next year at the sample time; It is the historical rainfall at the hth hour on the dth day of the mth month of the nth year at the rainfall station or meteorological station t.

[0173] S26: Convert the hourly rainfall time series of all rainfall stations and meteorological stations in the basin to which the target lake predicted in step S25 belongs into a basin meteorological boundary file that conforms to the input format of the distributed basin hydrological model; convert the hourly rainfall time series of the rainfall stations and meteorological stations within 5 km from Lake E into a lake meteorological boundary file that conforms to the input format of the lake hydrodynamic model.

[0174] Before obtaining the basin meteorological boundary file and lake boundary meteorological file, the hourly rainfall time series of the rainfall station and meteorological station are compared with the The corresponding historical measured meteorological data of the nth year excluding the historical rainfall are combined and then converted into the required watershed meteorological boundary file or lake meteorological boundary file.

[0175] S3: Input the generated multiple watershed meteorological boundary files into the distributed watershed hydrological model respectively, extract the time series of water volume entering the lake and the time series of water volume withdrawn from the lake output by the distributed watershed hydrological model corresponding to each watershed meteorological boundary file, and form a set of watershed inflow-water withdrawal boundary files covering meteorological uncertainties.

[0176] The time series of water volume entering the lake from each tributary in the basin and the time series of water volume taken from the lake are extracted from the output of the distributed watershed hydrological model. These data are then integrated and processed, and the water volume of multiple tributaries entering the lake is accumulated to generate a total inflow time series. At the same time, the water volume of each water intake is summarized to form a total water intake time series. Finally, according to the input requirements of the lake hydrodynamic model, the processed inflow and water intake data are converted into a standardized boundary file format to provide the lake hydrodynamic model with the basin inflow and water intake boundary files.

[0177] After the distributed watershed hydrological model in step S1 is constructed, multiple identical distributed watershed hydrological models are copied, and the number of distributed watershed hydrological models is consistent with the number of watershed meteorological boundary files. The multiple watershed meteorological boundary files are respectively input into different distributed watershed hydrological models, and multiple distributed watershed hydrological models are driven at the same time, which is conducive to improving the overall operation efficiency.

[0178] S4: Based on the water level monitoring data of the lake body characteristic data, the water level distribution range of the target lake at the end of each month is statistically calculated. The grid search method is used to construct multiple annual-scale water level control strategies with differentiated regulation characteristics for the water level distribution range at the end of each month, and each annual-scale water level strategy is converted into a water level control boundary file of the lake body hydrodynamic model.

[0179] The steps for constructing multiple annual-scale water level control strategies with differentiated regulation characteristics using a grid search method for the water level distribution range at the end of the month are as follows:

[0180] S41: Based on the monthly end-of-month water level distribution range of the target lake, extract the minimum and maximum values ​​of the monthly end-of-month water level distribution. The extraction expression is as follows: ;in is the historical lowest water level of the target lake at the end of the mth month; is the historical highest water level of the lake at the end of the mth month; N refers to the number of historical years for which lake characteristic data were collected.

[0181] S42: Latin hypercube sampling method is used in each month Multiple sampling is performed within the interval to generate multiple sets of annual water level control strategies covering 12 months of the year; the sampling expression is as follows: ;in Refers to the dispatching water level obtained at the rth sampling in the mth month; LHS refers to Latin hypercube sampling; Refers to the annual water level control strategy obtained from the rth sampling.

[0182] In this embodiment, the water level control intervals of Lake E for each month are as follows: Figure 24 As shown, from January to December, 3, 3, 3, 3, 3, 3, 3, 4, 4, 3, 3, 3, and 2 groups of samples were taken respectively, and a total of 3*3*3*3*3*3*3*3*4*4*3*3*2=629856 groups of annual-scale water level control strategies were obtained.

[0183] S5: The lake meteorological boundary file and the basin inflow-water intake boundary file with the same historical measured meteorological data year are used to form an initial combination. Several initial combinations are arranged and combined with several water level control boundary files to obtain multiple groups of input files for the lake hydrodynamic model. Each group of input files is input into the lake hydrodynamic model for annual-scale simulation, and the lake water level time series and downstream outlet flow time series after the lake hydrodynamic model simulation corresponding to each group of input files are extracted.

[0184] After the lake hydrodynamic model in step S1 is constructed, a total of 110*629856=69284160 lake hydrodynamic models are copied, and multiple input files are input into different lake hydrodynamic models respectively. Multiple lake hydrodynamic models are driven at the same time, which is conducive to improving the overall operation efficiency.

[0185] S6: Merge several groups of input files using the same annual-scale water level control strategy to obtain several input groups; conduct a multi-objective comprehensive evaluation of the input groups based on the lake water level time series and downstream outflow flow time series data corresponding to each group of input files, and combine them with lake water resources and water security-related indicators to determine the input group with the best multi-objective comprehensive score, and extract the water level control boundary file used by this input group as the final annual-scale robust water level scheduling logic for the target lake.

[0186] The multi-objective scoring formula is: ; x is the number of all lake meteorological boundary files in the current input group whose lake water level time series after the lake hydrodynamic model simulation is within the set standard water level range throughout the year; y is the number of lake meteorological boundary files in the current input group whose downstream outflow flow time series after the lake hydrodynamic model simulation meets the maximum outflow capacity restriction condition of the outflow outlet throughout the whole process; z is the minimum value of the lake water level at the end of the year corresponding to the annual-scale simulation results of all lake hydrodynamic models in the current input group; z min The minimum value of the lake water level at the end of the year corresponding to the annual scale simulation results of the lake hydrodynamic model in all input files; max is the maximum value of the lake water level at the end of the year corresponding to the annual-scale simulation results of the lake hydrodynamic model in all input files; u is the total number of lake meteorological boundary files in step S2.

[0187] The multi-objective comprehensive score results of all annual water level control candidate strategies (i.e., input groups) obtained in this example Figure 25 This embodiment ultimately selects the water level control boundary file with the highest multi-objective comprehensive score as the final annual-scale robust water level scheduling logic.

[0188] Example 3: The present invention discloses a lake water resources-water security coordinated annual scale robust scheduling logic generation system, such as Figure 26 As shown, it includes a basin hydrology-lake hydrodynamic coupling modeling module, a comprehensive meteorological uncertainty boundary file generation module, a basin hydrological simulation module, a lake water level control strategy generation module, a lake hydrodynamic simulation module and a water level control strategy determination module.

[0189] The basin hydrology-lake hydrodynamic coupling modeling module is used to collect hydrogeographic data and lake body characteristic data for the basin to which the target lake belongs, construct a distributed basin hydrological model based on the hydrogeographic data, and construct a lake hydrodynamic model using the lake body characteristic data and the output of the distributed basin hydrological model. The basin hydrology-lake hydrodynamic coupling modeling module performs the contents of step S1 in Example 1.

[0190] The integrated meteorological uncertainty boundary file generation module is configured to obtain annual-scale forecast rainfall data and historically measured meteorological data for the target lake's basin. The module then uses scenario construction techniques to construct multiple boundary scenarios representing meteorological uncertainty. Based on these multiple boundary scenarios, the module then generates a basin meteorological boundary file for use in the distributed basin hydrological model and a lake meteorological boundary file for use in the lake hydrodynamic model. The integrated meteorological uncertainty boundary file generation module executes step S2 of step 1.

[0191] The watershed hydrological simulation module is configured to input the generated multiple watershed meteorological boundary files into the distributed watershed hydrological model, extract the time series of water inflow and water withdrawal from the lake output by the distributed watershed hydrological model corresponding to each watershed meteorological boundary file, and form a set of watershed inflow-withdrawal boundary files that account for meteorological uncertainties. The watershed hydrological simulation module executes step S3 in step embodiment 1.

[0192] The lake water level control strategy generation module can calculate the monthly end-of-month water level distribution range of the target lake based on the water level monitoring data of the lake body characteristic data. It uses a grid search method to construct multiple annual-scale water level control strategies with differentiated regulation characteristics for the end-of-month water level distribution range. It then converts each annual-scale water level strategy into a water level control boundary file for the lake hydrodynamic model. The lake water level control strategy generation module executes the contents of step S4 in step embodiment 1.

[0193] The lake hydrodynamic simulation module can construct an initial combination of lake meteorological boundary files and basin inflow-intake boundary files corresponding to the historical measured meteorological data year. These initial combinations are then permuted and combined with multiple water level control boundary files to generate multiple sets of lake hydrodynamic model input files. Each set of input files is then input into the lake hydrodynamic model for annual-scale simulation. The module then extracts the lake water level time series and downstream outflow flow time series corresponding to each set of input files. The lake hydrodynamic simulation module then executes step S5 of step 1.

[0194] The water level control strategy determination module can merge several groups of input files using the same annual-scale water level control strategy to obtain several input groups; based on the lake water level time series and downstream outflow flow time series data corresponding to each group of input files, and combined with lake water resources and water security-related indicators, the input group is evaluated comprehensively with multiple objectives to determine the input group with the best multi-objective comprehensive score, and the water level control boundary file used by the input group is extracted as the final annual-scale robust water level scheduling logic for the target lake.

[0195] Example 4: An electronic device disclosed in the present invention includes one or more processors, one or more memories, and one or more programs. The programs are stored in the memories and are configured to be executed by the processors. When the programs are loaded into the processors, the steps of the annual-scale robust scheduling logic generation method for lake water resources-water security coordination of Example 1 are implemented.

[0196] Example 5: The present invention discloses a computer-readable storage medium, which stores a computer program. The computer program includes program instructions. When the program instructions are executed by a processor, the processor executes the steps of the annual-scale robust scheduling logic generation method for lake water resources-water security coordination in Example 1.

Claims

1. A method for generating annual-scale robust scheduling logic for lake water resources and water security collaboration, characterized by: The following steps are included: S1: Collect hydrogeographic data and lake body characteristic data of the target lake's basin, build a distributed basin hydrological model based on the hydrogeographic data, and then use the lake body characteristic data and the output results of the distributed basin hydrological model to construct a lake hydrodynamic model. S2: Obtain annual-scale forecast rainfall data and historical meteorological data for the target lake's basin. Use scenario construction techniques to construct multiple boundary scenarios representing meteorological uncertainty based on the annual-scale forecast rainfall data and historical meteorological data. Based on the multiple boundary scenarios, generate a basin meteorological boundary file for use in the distributed basin hydrological model and a lake meteorological boundary file for use in the lake hydrodynamic model. S3: Input the generated multiple watershed meteorological boundary files into the distributed watershed hydrological model respectively, extract the time series of water volume entering the lake and the time series of water volume withdrawn from the lake output by the distributed watershed hydrological model after running the corresponding distributed watershed meteorological boundary file, and form a set of watershed inflow-withdrawal boundary files covering meteorological uncertainties; S4: Based on the water level monitoring data of the lake body characteristic data, the monthly end-of-month water level distribution range of the target lake is statistically calculated. A grid search method is used to construct multiple annual-scale water level control strategies with differentiated regulation characteristics for the end-of-month water level distribution range. Each annual-scale water level strategy is converted into a water level control boundary file for the lake body hydrodynamic model. S5: The lake meteorological boundary file and the basin inflow-water intake boundary file with the same historical meteorological data year are used to form an initial combination. Several initial combinations are arranged and combined with several water level control boundary files to obtain multiple sets of input files for the lake hydrodynamic model. Each set of input files is input into the lake hydrodynamic model for annual-scale simulation, and the lake water level time series and downstream outflow flow time series corresponding to each set of input files after the lake hydrodynamic model simulation are extracted; S6: Merge several groups of input files using the same annual-scale water level control strategy to obtain several input groups; conduct a multi-objective comprehensive evaluation of the input groups based on the lake water level time series and downstream outflow flow time series data corresponding to each group of input files, and combine them with lake water resources and water security-related indicators to determine the input group with the best multi-objective comprehensive score, and extract the water level control boundary file used by this input group as the final annual-scale robust water level scheduling logic for the target lake.

2. The method for generating annual-scale robust scheduling logic for lake water resources-water security coordination according to claim 1 is characterized by: The basin hydrogeographic data described in step S1 include the surface elevation data, soil type, land use, hourly meteorological data, measured hydrological data of the basin, spatial distribution information of drainage and water intake facilities, and their operating flow time series data of the basin to which the lake belongs; the lake body characteristic data described in step S1 include the underwater three-dimensional terrain data of the lake, lake area meteorological data, water level monitoring data, and continuous monitoring data of inflow and outflow.

3. The method for generating annual-scale robust scheduling logic for lake water resources-water security coordination according to claim 1 is characterized by: The annual-scale forecast rainfall data described in step S2 refers to the monthly forecast rainfall for the next year predicted by all forecast stations in the basin to which the target lake body belongs; the historical measured meteorological data described in step S2 refers to the many-year historical hourly rainfall data of all rainfall stations in the basin to which the target lake body belongs and the many-year historical hourly multi-index meteorological observation data of all meteorological stations.

4. The method for generating annual-scale robust scheduling logic for lake water resources-water security coordination according to claim 3 is characterized by: The steps for generating the watershed meteorological boundary file for use in the distributed watershed hydrological model and the lake meteorological boundary file for use in the lake hydrodynamic model in step S2 are as follows: S21: Based on the monthly forecast rainfall of each forecast station for the next year, the annual rainfall total of each forecast station is calculated, and all the annual rainfall totals are disturbed and expanded to obtain multiple annual rainfall total samples; S22: Based on the historical hourly rainfall data and the historical hourly multi-index meteorological observation data, the weight of the historical monthly rainfall at each rainfall station and meteorological station in the total annual rainfall is calculated to form a historical monthly scale weight sequence. Based on the historical monthly scale weight sequence, several sets of predicted monthly scale weight sequences covering the 12 months of the year are generated; S23: The annual rainfall total sample obtained in step S21 and the predicted monthly scale weight sequence obtained in step S22 are arranged and combined to obtain a plurality of annual rainfall total sample-prediction monthly scale weight sequence combinations, and the monthly rainfall of each month in the annual rainfall total sample-prediction monthly scale weight sequence combination constitutes a monthly rainfall time series sample; S24: clustering all monthly rainfall time series samples using a clustering method, obtaining k types of typical meteorological scenarios, and using the typical meteorological scenarios as representative samples of the target rainfall time series; S25: Combine k representative samples of target monthly rainfall time series with historical meteorological data to predict hourly rainfall time series for all rainfall stations and meteorological stations in the next year; S26: converting the hourly rainfall time series of all rain gauges and meteorological stations in the watershed to which the target lake predicted in step S25 belongs into a watershed meteorological boundary file in an input format that conforms to the distributed watershed hydrological model; The hourly rainfall time series of rainfall stations and meteorological stations within the specified range of the target lake are converted into lake meteorological boundary files in the input format that conforms to the lake hydrodynamic model.

5. The method for generating annual-scale robust scheduling logic for lake water resources-water security coordination according to claim 4 is characterized by: The method of generating several sets of predicted monthly scale weight sequences containing 12 months of a year in step S22 is as follows: All historical monthly weight sequences of each rainfall station and meteorological station in the historical measured meteorological data are classified by month to form 12 historical rainfall weight sets; For each month's historical rainfall weight set, a Bernoulli distribution model is constructed based on the binary event results of whether rainfall occurs in that month at each rainfall station and meteorological station recorded in historical meteorological data to simulate the rainfall probability of that month. The Bernoulli distribution expression is as follows: ; where p m is the Bernoulli distribution probability; X m =1 means there is rainfall in month m; X m =0 means no rainfall in month m; m=1,2,...,12; In the historical rainfall weight set of each month, a subset of historical monthly rainfall weights corresponding to rainfall events is selected, and a Gamma distribution model is constructed to simulate the probability distribution of the rainfall weight of that month. The expression of the Gamma distribution is: ;in is the monthly rainfall weight of the mth month; is the shape parameter of the Gamma distribution; is the scale parameter of the Gamma distribution; Multiple sampling is performed based on the Bernoulli–Gamma joint distribution model to generate multiple sets of predicted monthly scale weight sequences covering 12 months of the year. The steps for a single sampling are as follows: Sampling the mth month, first sample from the Bernoulli distribution corresponding to the mth month, if the collected X m =1, then sample from the Gamma distribution of the historical monthly rainfall weight subset corresponding to month m, and collect the monthly rainfall weight ; If the collected X m =0, then let ; Repeat the collection 12 times until the monthly rainfall weights for the next 12 months are collected. , and Perform normalization so that the normalized The sum is 1; After normalization is the predicted monthly scale weight series; The steps for predicting the hourly rainfall time series for all rainfall stations and meteorological stations in the next year in step S25 are as follows: Calculate the ratio b of the monthly rainfall of the k representative samples of the target monthly rainfall time series to the monthly average rainfall of the basin to which the target lake belongs. The calculation formula of b is as follows: ; where q = 1, 2, 3, 4… k， When the qth target month rainfall time series represents the proportion b of the mth month in the nth year of the sample history; is the average monthly rainfall in the mth month of the nth year in the history of the basin to which the target lake belongs; The rainfall time series of the qth target month represents the rainfall of the mth month of the sample; Then, based on the ratio b, the historical hourly rainfall of the historical measured meteorological data of all rain gauges and meteorological stations is scaled and mapped to obtain the hourly rainfall time series for the next year. The calculation formula is as follows: ;in Map the qth target month rainfall time series to the rainfall station or meteorological station t to represent the predicted rainfall at the hth hour of the mth month, day of the next year at the sample time; It is the historical rainfall at the hth hour on the dth day of the mth month of the nth year at the rainfall station or meteorological station t.

6. The method for generating annual-scale robust scheduling logic for lake water resources-water security coordination according to claim 1 is characterized by: In step S4, the steps of using the grid search method to construct multiple annual-scale water level control strategies with differentiated regulation characteristics for the water level distribution range at the end of the month are as follows: S41: Based on the monthly end-of-month water level distribution range of the target lake, extract the minimum and maximum values ​​of the monthly end-of-month water level distribution. The extraction expression is as follows: ;in is the historical lowest water level of the target lake at the end of the mth month; is the historical highest water level of the lake at the end of the mth month; N refers to the number of historical years for which lake characteristic data were collected; S42: Latin hypercube sampling method is used in each month Multiple sampling is performed within the interval to generate multiple sets of annual water level control strategies covering 12 months of the year; the sampling expression is as follows: ;in Refers to the dispatching water level obtained at the rth sampling in the mth month; LHS refers to Latin hypercube sampling; Refers to the annual water level control strategy obtained from the rth sampling.

7. The method for generating annual-scale robust scheduling logic for lake water resources-water security coordination according to claim 1 is characterized by: The multi-objective comprehensive evaluation formula in step S6 is: ; x is the number of all lake meteorological boundary files in the current input group whose lake water level time series after the lake hydrodynamic model simulation is within the set standard water level range throughout the year; y is the number of lake meteorological boundary files in the current input group whose downstream outflow flow time series after the lake hydrodynamic model simulation meets the maximum outflow capacity restriction condition of the outflow outlet throughout the whole process; z is the minimum value of the lake water level at the end of the year corresponding to the annual-scale simulation results of all lake hydrodynamic models in the current input group; z min The minimum value of the lake water level at the end of the year corresponding to the annual scale simulation results of the lake hydrodynamic model in all input files; max is the maximum value of the lake water level at the end of the year corresponding to the annual-scale simulation results of the lake hydrodynamic model in all input files; u is the total number of lake meteorological boundary files in step S2.

8. A robust annual-scale scheduling logic generation system for lake water resources and water security collaboration, characterized by: include, The basin hydrology-lake hydrodynamic coupling modeling module is used to collect hydrogeographic data of the basin to which the target lake belongs and the lake body characteristic data, build a distributed basin hydrological model based on the hydrogeographic data, and use the lake body characteristic data and the output results of the distributed basin hydrological model to jointly build a lake hydrodynamic model; A comprehensive meteorological uncertainty boundary file generation module is used to obtain annual-scale forecast rainfall data and historical measured meteorological data for the target lake basin. Scenario construction technology is used to construct multiple boundary scenarios representing meteorological uncertainty based on the annual-scale forecast rainfall data and historical measured meteorological data. Based on the multiple boundary scenarios, a basin meteorological boundary file for use in the distributed basin hydrological model and a lake meteorological boundary file for use in the lake hydrodynamic model are generated. The watershed hydrological simulation module is used to input the generated multiple watershed meteorological boundary files into the distributed watershed hydrological model respectively, extract the time series of water entering the lake and the time series of water withdrawn from the lake output by the distributed watershed hydrological model after running the corresponding distributed watershed meteorological boundary file, and form a set of watershed inflow-withdrawal boundary files that cover meteorological uncertainties; The lake water level control strategy generation module can calculate the monthly end-of-month water level distribution range of the target lake based on the water level monitoring data of the lake body's characteristic data. It uses a grid search method to construct multiple annual-scale water level control strategies with differentiated regulation characteristics for the end-of-month water level distribution range, and converts each annual-scale water level strategy into a water level control boundary file for the lake hydrodynamic model. The lake hydrodynamic simulation module can form an initial combination of lake meteorological boundary files and basin inflow-water intake boundary files that are consistent with the historical measured meteorological data year. Several initial combinations are arranged and combined with several water level control boundary files to obtain multiple groups of lake hydrodynamic model input files. Each group of input files is input into the lake hydrodynamic model for annual scale simulation, and the lake water level time series and downstream outflow flow time series after the lake hydrodynamic model simulation corresponding to each group of input files are extracted; The water level control strategy determination module can merge several groups of input files using the same annual-scale water level control strategy to obtain several input groups; based on the lake water level time series and downstream outflow flow time series data corresponding to each group of input files, and combined with lake water resources and water security-related indicators, the input group is evaluated comprehensively with multiple objectives to determine the input group with the best multi-objective comprehensive score, and the water level control boundary file used by the input group is extracted as the final annual-scale robust water level scheduling logic for the target lake.

9. An electronic device, characterized in that: It includes one or more processors, one or more memories and one or more programs, wherein the programs are stored in the memories and are configured to be executed by the processors, and when the programs are loaded into the processors, the steps of the annual-scale robust scheduling logic generation method for lake water resources-water security coordination are implemented according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, which includes program instructions. When the program instructions are executed by a processor, the processor executes the steps of the annual-scale robust scheduling logic generation method for lake water resources-water security coordination according to any one of claims 1 to 7.

Images