A reservoir scheduling method and system based on hydrological simulation

Abstract

This invention discloses a reservoir scheduling method and system based on hydrological simulation. The method includes: acquiring historical hydrological data, real-time monitoring data, and numerical weather prediction data of the watershed; constructing and screening a distributed hydrological model based on historical data; updating the model state using real-time monitoring data for data assimilation, and simulating and predicting future inflow processes using weather forecast data as input; generating at least one set of reservoir scheduling plans based on the predicted runoff, combined with reservoir engineering constraints and scheduling objectives; outputting visualized results and generating operation instructions. The system includes a data acquisition module, a hydrological simulation module, a scheduling decision module, and an output and display module. This invention improves the accuracy of inflow runoff prediction and the scientific nature of scheduling decisions through adaptive hydrological models, data assimilation, and multi-objective optimization algorithms, achieving multi-objective synergy of flood control, power generation, and ecology, and significantly improving the intelligence level and execution efficiency of reservoir scheduling.

Description

Technical Field

[0001] This invention relates to the field of reservoir scheduling and hydrological simulation technology, and in particular to a reservoir scheduling method and system based on hydrological simulation. Background Technology

[0002] Reservoir scheduling is a core component of optimizing water resource allocation, flood control and disaster reduction, and ensuring ecological and domestic water use. Its rationality directly affects the efficiency of water resource utilization, flood control safety, and ecological sustainability in the basin.

[0003] With the development of hydrological monitoring and numerical simulation technologies, some reservoir scheduling has begun to incorporate hydrological models to assist decision-making. However, these technologies have significant shortcomings. On the one hand, most hydrological models cannot adapt to the spatiotemporal variability of watershed hydrological processes, resulting in low accuracy in inflow runoff prediction and consequently affecting the reliability of scheduling plans. On the other hand, scheduling systems struggle to balance multiple objectives such as flood control, power generation, ecology, and water supply, lack effective scenario analysis and visualization capabilities, and suffer from low efficiency in converting plans into actual operational instructions, failing to achieve seamless integration between scheduling decisions and on-site execution.

[0004] Furthermore, the systems in these technologies generally lack feedback correction mechanisms, which causes the model prediction accuracy to gradually decline over time, making it difficult to meet long-term scheduling requirements. Summary of the Invention

[0005] The present invention aims to provide a reservoir scheduling method and system based on hydrological simulation, which realizes accurate prediction of hydrological processes, intelligent scheduling of multiple objectives and command execution, thereby improving the scientificity and efficiency of reservoir scheduling.

[0006] This specification provides one or more embodiments of a reservoir scheduling method based on hydrological simulation, including the following steps:

[0007] Step S1: Obtain historical hydrological data, real-time monitoring data, and numerical weather prediction data for the watershed where the target reservoir is located;

[0008] Step S2: Based on the historical hydrological data, construct and select a distributed hydrological model suitable for the watershed;

[0009] Step S3: Use the real-time monitoring data to assimilate and update the state variables of the distributed hydrological model, and use the numerical weather forecast data as the model input to simulate and predict the inflow process in the future period.

[0010] Step S4: Based on the inflow process obtained from the simulation prediction, and in combination with the engineering constraints and scheduling objectives of the target reservoir, generate at least one reservoir scheduling plan;

[0011] Step S5: Output the visualization results of the reservoir scheduling plan, and generate operation instructions for controlling the reservoir water release facilities according to the reservoir scheduling plan.

[0012] In some embodiments, the distributed hydrological model constructed in step S2 is either a SWAT model based on physical mechanisms or a Xin'anjiang model based on conceptual mechanisms; the screening process employs the SCE-UA global optimization algorithm or a Bayesian parameter estimation method. By selectively choosing a hydrological model suitable for the watershed characteristics and using a scientific parameter screening algorithm, the model's simulation accuracy of watershed hydrological processes can be improved, providing a reliable foundation for subsequent runoff prediction and scheduling decisions. The SWAT model is suitable for long-term hydrological simulation at the watershed scale and can accurately characterize the hydrological cycle process under different underlying surface conditions; the Xin'anjiang model has a simple structure and clear physical meaning of its parameters, making it suitable for hydrological simulation in humid and semi-humid regions. The SCE-UA global optimization algorithm has the advantages of strong global search capability and fast convergence speed, and can efficiently determine the optimal parameters of the model; the Bayesian parameter estimation method can combine prior information and observational data to reduce parameter uncertainty and improve the reliability of model parameters.

[0013] In some embodiments, the data assimilation update in step S3 employs an ensemble Kalman filter or a particle filter algorithm. Data assimilation technology organically integrates real-time monitoring data with model simulation results, dynamically corrects model state variables, effectively reduces model prediction errors, and improves the prediction accuracy of inflow runoff processes. Among these, the ensemble Kalman filter has high computational efficiency and is suitable for real-time updates of large-scale distributed hydrological models; the particle filter algorithm is highly adaptable to nonlinear and non-Gaussian systems, can cope with the uncertainties of complex hydrological processes, and further improves the accuracy of model updates.

[0014] In some embodiments, the scheduling objectives include at least one of the following: flood control safety objective, power generation efficiency maximization objective, ecological flow guarantee objective, and water supply satisfaction objective; the engineering constraints include at least one of the following: maximum operating water level of the reservoir, minimum operating water level of the reservoir, upper limit of outflow, and water level fluctuation limit. The scheduling objectives can be flexibly adjusted according to the actual needs of the basin to achieve multi-objective synergistic optimization; the engineering constraints are determined based on the reservoir engineering design standards and safe operation requirements to ensure the feasibility and safety of the scheduling plan and avoid safety accidents caused by scheduling operations exceeding the engineering carrying capacity.

[0015] In some embodiments, generating at least one set of reservoir scheduling plans in step S4 specifically includes: generating a Pareto front solution set using a multi-objective optimization algorithm for scheduling personnel to choose from. The multi-objective optimization algorithm can balance the conflicting relationships between various scheduling objectives while satisfying engineering constraints, generating multiple non-dominated scheduling schemes (Pareto front solution sets). Scheduling personnel can flexibly select the optimal scheduling plan based on actual hydrological conditions, watershed demand preferences, and other factors, improving the flexibility and relevance of scheduling decisions.

[0016] One or more embodiments of this specification also provide a reservoir scheduling system based on hydrological simulation, used to implement the above-described reservoir scheduling method based on hydrological simulation. The system includes: a data acquisition module, a hydrological simulation module, a scheduling decision module, and an output and display module.

[0017] The data acquisition module is used to acquire historical hydrological data, real-time monitoring data, and numerical weather prediction data for the watershed where the target reservoir is located. Historical hydrological data includes long-term observational data such as historical precipitation, runoff, evaporation, and temperature within the watershed, used for model building and parameter selection. Real-time monitoring data includes reservoir water level, outflow, and real-time precipitation data for the watershed, used for model updates and error correction. Numerical weather prediction data includes predicted precipitation and temperature for future periods, used for predicting inflow runoff. The data acquisition module can automatically collect and integrate multi-source data by connecting to hydrological monitoring stations and meteorological databases, ensuring the real-time nature and completeness of the data.

[0018] The hydrological simulation module is used to construct and screen distributed hydrological models based on the historical hydrological data. It updates the state variables of the distributed hydrological model using real-time monitoring data and uses numerical weather prediction data as model input to simulate and predict inflow processes in future periods. The hydrological simulation module is the core computing unit of the system. By constructing a distributed hydrological model adapted to the watershed characteristics and combining it with data assimilation technology, it achieves accurate prediction of inflow processes, providing data support for scheduling decisions.

[0019] The scheduling decision module is used to generate at least one reservoir scheduling plan based on the inflow runoff process prediction results output by the hydrological simulation module, combined with the engineering constraints and scheduling objectives of the target reservoir. The scheduling decision module uses a multi-objective optimization algorithm to balance various scheduling objectives and constraints, generating feasible scheduling plans and providing a basis for decision-making by scheduling personnel.

[0020] Output and Display Module: This module outputs the visualized results of the reservoir scheduling plan and generates operational instructions for controlling the reservoir's water release facilities. Visual display provides an intuitive view of the implementation effects of the scheduling plan, facilitating quick access to relevant information for dispatchers. The automatic generation of operational instructions ensures seamless integration between scheduling decisions and on-site execution, improving scheduling efficiency.

[0021] In some embodiments, the hydrological simulation module integrates a data assimilation unit. This data assimilation unit employs an ensemble Kalman filter or particle filter algorithm to periodically assimilate the real-time monitoring data into the state variables of the distributed hydrological model. The data assimilation unit enables dynamic updates to the model state, periodically corrects model prediction biases, ensures the accuracy of inflow runoff prediction, and provides a guarantee for the scientific nature of scheduling decisions.

[0022] In some embodiments, the scheduling decision module includes: a scenario generation unit, a multi-objective optimization unit, and a scheme selection unit. The scenario generation unit generates multiple sets of inflow scenarios based on sources of uncertainty, covering inflow scenarios under different hydrological conditions, thereby improving the resilience of the scheduling plan. The multi-objective optimization unit uses the NSGA-II algorithm or the MOEA / D algorithm to solve for the Pareto optimal scheduling scheme set under the engineering constraints. The NSGA-II algorithm has advantages such as fast non-dominated sorting and elite retention strategies, while the MOEA / D algorithm improves optimization efficiency by decomposing the multi-objective problem into multiple single-objective problems. Both can efficiently achieve multi-objective scheduling optimization. The scheme selection unit provides an interactive interface, receives user-input preference weights, and selects a final scheduling plan from the Pareto optimal scheduling scheme set, realizing human-machine collaborative decision-making and balancing scheduling scientificity with practical needs.

[0023] In some embodiments, the output and display module includes a GIS display unit and an instruction generation unit. The GIS display unit is used to dynamically display the predicted runoff process, reservoir water level change process, and downstream inundation risk range on a geographic information system map, intuitively presenting the impact of the scheduling plan on the watershed, and facilitating dispatchers to quickly judge the rationality of the plan. The instruction generation unit is used to automatically convert the final scheduling plan into gate opening control instructions or discharge flow setpoints, and send them directly to the reservoir discharge facility control system, realizing the rapid execution of scheduling instructions, reducing manual intervention, and improving scheduling efficiency.

[0024] In some embodiments, the system further includes a feedback correction module. The feedback correction module compares the actual hydrological process with the inflow process predicted by the hydrological simulation module, calculates the prediction error, and uses this error to correct the parameters of the distributed hydrological model online. Through the feedback correction mechanism, dynamic optimization of model parameters can be achieved, gradually improving the model's prediction accuracy, ensuring the long-term reliability and adaptability of the system, and addressing the spatiotemporal variability of watershed hydrological processes.

[0025] The beneficial effects that this application may bring include, but are not limited to:

[0026] (1) High accuracy of hydrological simulation: By selecting a distributed hydrological model that is suitable for the characteristics of the watershed, and combining scientific parameter screening algorithms and data assimilation technology, the model state is dynamically corrected, which effectively improves the prediction accuracy of the inflow runoff process and provides reliable data support for scheduling decisions.

[0027] (2) Intelligent scheduling decision-making: The Pareto front solution set is generated by using a multi-objective optimization algorithm and combined with the human-machine collaborative decision-making mode to achieve balanced optimization of multiple objectives such as flood control, power generation, ecology and water supply, thereby improving the scientific nature and flexibility of scheduling decision-making;

[0028] (3) Efficient scheduling execution: The effect of the plan is presented intuitively through visualization, and operation instructions such as gate control are automatically generated to achieve seamless connection between scheduling decisions and on-site execution, reduce manual intervention and improve scheduling efficiency;

[0029] (4) Strong system adaptability: The model parameters are corrected online through the feedback correction module, which adapts to the spatiotemporal variability of the watershed hydrological process and ensures the reliability of the system in long-term operation;

[0030] (5) Strong risk resistance: Multiple water inflow scenarios are constructed through scenario generation units, and the generated scheduling plans can cope with different hydrological situations, enhance the risk resistance of reservoir scheduling, and ensure the safety of flood control and the rational use of water resources in the basin.

[0031] It should be noted that different embodiments may produce different beneficial effects. In different embodiments, the beneficial effects may be any one or a combination of the above, or any other possible beneficial effects.

[0032] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0033] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0034] Figure 1 This is a schematic diagram illustrating an application scenario of a reservoir scheduling system based on hydrological simulation, according to some embodiments of this specification.

[0035] Figure 2 This is an exemplary flowchart of a reservoir scheduling method based on hydrological simulation, as shown in some embodiments of this specification.

[0036] Figure 3 This is an exemplary flowchart of the parameter screening for a distributed hydrological model according to some embodiments of this specification;

[0037] Figure 4 This is an exemplary flowchart illustrating the generation of scheduling plans through multi-objective optimization according to some embodiments of this specification;

[0038] Figure 5 This is a schematic diagram of the functional modules of a reservoir scheduling system based on hydrological simulation, as shown in some embodiments of this specification.

[0039] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. Detailed Implementation

[0040] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0041] Reference Appendix Figures 1-5 This specification provides one or more embodiments of a reservoir scheduling method based on hydrological simulation, comprising the following steps:

[0042] Step S1: Obtain historical hydrological data, real-time monitoring data, and numerical weather prediction data for the watershed where the target reservoir is located;

[0043] Step S2: Based on the historical hydrological data, construct and select a distributed hydrological model suitable for the watershed;

[0044] Step S3: Use the real-time monitoring data to assimilate and update the state variables of the distributed hydrological model, and use the numerical weather forecast data as the model input to simulate and predict the inflow process in the future period.

[0045] Step S4: Based on the inflow process obtained from the simulation prediction, and in combination with the engineering constraints and scheduling objectives of the target reservoir, generate at least one reservoir scheduling plan;

[0046] Step S5: Output the visualization results of the reservoir scheduling plan, and generate operation instructions for controlling the reservoir water release facilities according to the reservoir scheduling plan.

[0047] In some embodiments, the distributed hydrological model constructed in step S2 is either a SWAT model based on physical mechanisms or a Xin'anjiang model based on conceptual mechanisms; the screening process employs the SCE-UA global optimization algorithm or a Bayesian parameter estimation method. By selectively choosing a hydrological model suitable for the watershed characteristics and using a scientific parameter screening algorithm, the model's simulation accuracy of watershed hydrological processes can be improved, providing a reliable foundation for subsequent runoff prediction and scheduling decisions. The SWAT model is suitable for long-term hydrological simulation at the watershed scale and can accurately characterize the hydrological cycle process under different underlying surface conditions; the Xin'anjiang model has a simple structure and clear physical meaning of its parameters, making it suitable for hydrological simulation in humid and semi-humid regions. The SCE-UA global optimization algorithm has the advantages of strong global search capability and fast convergence speed, and can efficiently determine the optimal parameters of the model; the Bayesian parameter estimation method can combine prior information and observational data to reduce parameter uncertainty and improve the reliability of model parameters.

[0048] In some embodiments, the data assimilation update in step S3 employs an ensemble Kalman filter or a particle filter algorithm. Data assimilation technology organically integrates real-time monitoring data with model simulation results, dynamically corrects model state variables, effectively reduces model prediction errors, and improves the prediction accuracy of inflow runoff processes. Among these, the ensemble Kalman filter has high computational efficiency and is suitable for real-time updates of large-scale distributed hydrological models; the particle filter algorithm is highly adaptable to nonlinear and non-Gaussian systems, can cope with the uncertainties of complex hydrological processes, and further improves the accuracy of model updates.

[0049] In some embodiments, the scheduling objectives include at least one of the following: flood control safety objective, power generation efficiency maximization objective, ecological flow guarantee objective, and water supply satisfaction objective; the engineering constraints include at least one of the following: maximum operating water level of the reservoir, minimum operating water level of the reservoir, upper limit of outflow, and water level fluctuation limit. The scheduling objectives can be flexibly adjusted according to the actual needs of the basin to achieve multi-objective synergistic optimization; the engineering constraints are determined based on the reservoir engineering design standards and safe operation requirements to ensure the feasibility and safety of the scheduling plan and avoid safety accidents caused by scheduling operations exceeding the engineering carrying capacity.

[0050] In some embodiments, generating at least one set of reservoir scheduling plans in step S4 specifically includes: generating a Pareto front solution set using a multi-objective optimization algorithm for scheduling personnel to choose from. The multi-objective optimization algorithm can balance the conflicting relationships between various scheduling objectives while satisfying engineering constraints, generating multiple non-dominated scheduling schemes (Pareto front solution sets). Scheduling personnel can flexibly select the optimal scheduling plan based on actual hydrological conditions, watershed demand preferences, and other factors, improving the flexibility and relevance of scheduling decisions.

[0051] One or more embodiments of this specification also provide a reservoir scheduling system based on hydrological simulation, used to implement the above-described reservoir scheduling method based on hydrological simulation. The system includes: a data acquisition module, a hydrological simulation module, a scheduling decision module, and an output and display module.

[0052] The data acquisition module is used to acquire historical hydrological data, real-time monitoring data, and numerical weather prediction data for the watershed where the target reservoir is located. Historical hydrological data includes long-term observational data such as historical precipitation, runoff, evaporation, and temperature within the watershed, used for model building and parameter selection. Real-time monitoring data includes reservoir water level, outflow, and real-time precipitation data for the watershed, used for model updates and error correction. Numerical weather prediction data includes predicted precipitation and temperature for future periods, used for predicting inflow runoff. The data acquisition module can automatically collect and integrate multi-source data by connecting to hydrological monitoring stations and meteorological databases, ensuring the real-time nature and completeness of the data.

[0053] The hydrological simulation module is used to construct and screen distributed hydrological models based on the historical hydrological data. It updates the state variables of the distributed hydrological model using real-time monitoring data and uses numerical weather prediction data as model input to simulate and predict inflow processes in future periods. The hydrological simulation module is the core computing unit of the system. By constructing a distributed hydrological model adapted to the watershed characteristics and combining it with data assimilation technology, it achieves accurate prediction of inflow processes, providing data support for scheduling decisions.

[0054] The scheduling decision module is used to generate at least one reservoir scheduling plan based on the inflow runoff process prediction results output by the hydrological simulation module, combined with the engineering constraints and scheduling objectives of the target reservoir. The scheduling decision module uses a multi-objective optimization algorithm to balance various scheduling objectives and constraints, generating feasible scheduling plans and providing a basis for decision-making by scheduling personnel.

[0055] Output and Display Module: This module outputs the visualized results of the reservoir scheduling plan and generates operational instructions for controlling the reservoir's water release facilities. Visual display provides an intuitive view of the implementation effects of the scheduling plan, facilitating quick access to relevant information for dispatchers. The automatic generation of operational instructions ensures seamless integration between scheduling decisions and on-site execution, improving scheduling efficiency.

[0056] In some embodiments, the hydrological simulation module integrates a data assimilation unit. This data assimilation unit employs an ensemble Kalman filter or particle filter algorithm to periodically assimilate the real-time monitoring data into the state variables of the distributed hydrological model. The data assimilation unit enables dynamic updates to the model state, periodically corrects model prediction biases, ensures the accuracy of inflow runoff prediction, and provides a guarantee for the scientific nature of scheduling decisions.

[0057] In some embodiments, the scheduling decision module includes: a scenario generation unit, a multi-objective optimization unit, and a scheme selection unit. The scenario generation unit generates multiple sets of inflow scenarios based on sources of uncertainty, covering inflow scenarios under different hydrological conditions, thereby improving the resilience of the scheduling plan. The multi-objective optimization unit uses the NSGA-II algorithm or the MOEA / D algorithm to solve for the Pareto optimal scheduling scheme set under the engineering constraints. The NSGA-II algorithm has advantages such as fast non-dominated sorting and elite retention strategies, while the MOEA / D algorithm improves optimization efficiency by decomposing the multi-objective problem into multiple single-objective problems. Both can efficiently achieve multi-objective scheduling optimization. The scheme selection unit provides an interactive interface, receives user-input preference weights, and selects a final scheduling plan from the Pareto optimal scheduling scheme set, realizing human-machine collaborative decision-making and balancing scheduling scientificity with practical needs.

[0058] In some embodiments, the output and display module includes a GIS display unit and an instruction generation unit. The GIS display unit is used to dynamically display the predicted runoff process, reservoir water level change process, and downstream inundation risk range on a geographic information system map, intuitively presenting the impact of the scheduling plan on the watershed, and facilitating dispatchers to quickly judge the rationality of the plan. The instruction generation unit is used to automatically convert the final scheduling plan into gate opening control instructions or discharge flow setpoints, and send them directly to the reservoir discharge facility control system, realizing the rapid execution of scheduling instructions, reducing manual intervention, and improving scheduling efficiency.

[0059] In some embodiments, the system further includes a feedback correction module. The feedback correction module compares the actual hydrological process with the inflow process predicted by the hydrological simulation module, calculates the prediction error, and uses this error to correct the parameters of the distributed hydrological model online. Through the feedback correction mechanism, dynamic optimization of model parameters can be achieved, gradually improving the model's prediction accuracy, ensuring the long-term reliability and adaptability of the system, and addressing the spatiotemporal variability of watershed hydrological processes.

[0060] The beneficial effects that this application may bring include, but are not limited to:

[0061] (1) High accuracy of hydrological simulation: By selecting a distributed hydrological model that is suitable for the characteristics of the watershed, and combining scientific parameter screening algorithms and data assimilation technology, the model state is dynamically corrected, which effectively improves the prediction accuracy of the inflow runoff process and provides reliable data support for scheduling decisions.

[0062] (2) Intelligent scheduling decision-making: The Pareto front solution set is generated by using a multi-objective optimization algorithm and combined with the human-machine collaborative decision-making mode to achieve balanced optimization of multiple objectives such as flood control, power generation, ecology and water supply, thereby improving the scientific nature and flexibility of scheduling decision-making;

[0063] (3) Efficient scheduling execution: The effect of the plan is presented intuitively through visualization, and operation instructions such as gate control are automatically generated to achieve seamless connection between scheduling decisions and on-site execution, reduce manual intervention and improve scheduling efficiency;

[0064] (4) Strong system adaptability: The model parameters are corrected online through the feedback correction module, which adapts to the spatiotemporal variability of the watershed hydrological process and ensures the reliability of the system in long-term operation;

[0065] (5) Strong risk resistance: Multiple water inflow scenarios are constructed through scenario generation units, and the generated scheduling plans can cope with different hydrological situations, enhance the risk resistance of reservoir scheduling, and ensure the safety of flood control and the rational use of water resources in the basin.

[0066] In practical applications of reservoir operation, one of the core challenges faced by dispatchers is how to quickly formulate scientific and reasonable dispatch plans under complex and ever-changing hydrological conditions, taking into account multiple objectives such as flood control, power generation, ecology, and water supply, while ensuring the efficient execution of dispatch instructions. Traditional dispatch methods rely on experience-based judgment, making it difficult to cope with extreme hydrological events, and suffer from low dispatch accuracy and poor execution efficiency. Existing hydrological simulation-assisted dispatch systems suffer from poor model adaptability, insufficient prediction accuracy, difficulty in balancing multiple objectives, and a lack of dynamic correction capabilities, failing to meet the needs of modern reservoir operation.

[0067] Example 1: A Case Study of Reservoir Scheduling for Coordinated Optimization of Flood Control and Power Generation

[0068] This embodiment uses the "Biyuan Reservoir," a large-scale water conservancy project in southern my country, as an example to illustrate the implementation process of the present invention in detail. This reservoir is a comprehensive reservoir primarily for flood control and power generation, while also considering water supply and ecological protection. The reservoir controls a drainage area of ​​approximately 3,200 square kilometers, located in a typical humid and semi-humid region with abundant annual rainfall, high and large flood peaks during the flood season, and significant runoff variations during the dry season.

[0069] Step S1: Acquisition of multi-source data

[0070] First, the following data acquisition work is performed through the "data acquisition module" of this invention:

[0071] Historical hydrological data: Daily precipitation, runoff, evaporation, and temperature data for the Biyuan Reservoir basin over the past 30 years (1994-2023), as well as hourly interval data for 10 typical flood events that occurred during this period, were obtained. The data were sourced from 23 rain gauge stations, 3 evaporation stations, and 1 inflow hydrological station within the basin.

[0072] Real-time monitoring data: The latest data (with a time resolution of 1 hour) from all rain gauge stations and water level stations in the basin are accessed in real time through the hydrological telemetry system, including: the current water level of the reservoir (e.g., the flood limit water level of 185.0 meters), the real-time inflow, and the cumulative rainfall of each rain gauge station in the past 6 hours.

[0073] Numerical weather forecast data: Connecting with the National Meteorological Center's high-resolution regional ensemble forecast system (CMA-GEPS), we obtain gridded precipitation and temperature forecast data for the next 72 hours, with a temporal resolution of 1 hour and a spatial resolution of 3km×3km.

[0074] Step S2: Hydrological Model Construction and Screening

[0075] Given the complex underlying surface conditions and significant wetting characteristics of the Biyuan Reservoir basin, this embodiment selects the Xin'anjiang model based on a conceptual mechanism as the distributed hydrological model. The model construction process is as follows:

[0076] The watershed was divided into 45 natural sub-watersheds based on the digital elevation model (DEM) and river network distribution.

[0077] Within each sub-basin, based on land use and soil type data, it is further divided into several hydrological response units.

[0078] The SCE-UA global optimization algorithm was used to screen key model parameters (such as evapotranspiration conversion factor K, free water storage capacity SM, and lag parameter L). Calibration was performed using daily runoff data from 1990 to 2010, and validation was performed using data from 2011 to 2020. Calibration results show that the Nash-Sutcliffe efficiency coefficient (NSE) reached 0.92 and 0.89 during the calibration and validation periods, respectively, indicating that the model can accurately simulate the hydrological processes in the Biyuan Reservoir watershed.

[0079] Step S3: Data Assimilation and Inflow Runoff Prediction

[0080] Entering the actual scheduling phase (assuming the current time is 8:00 AM on July 15, 2024), perform the following operations:

[0081] Data Assimilation Update: The "Data Assimilation Unit" in the hydrological simulation module is activated, employing an ensemble Kalman filter algorithm. The algorithm sets up 50 ensemble members to assimilate real-time monitoring data from the past 6 hours (basin rainfall, inflow calculated from reservoir water levels) into the state variables of the Xin'anjiang model (such as soil moisture content, free water storage, etc.), thereby dynamically correcting the model's initial state and eliminating early forecast errors.

[0082] Runoff Forecast: Using the 72-hour numerical weather forecast obtained in step S1 as model input, the assimilated and updated Xin'anjiang model is driven to simulate and predict the inflow runoff process over the next 72 hours (from 8:00 on July 15th to 8:00 on July 18th). The forecast results show that a flood peak will occur in the next 24 hours, with a peak flow of approximately 3200 cubic meters per second.

[0083] Step S4: Generation of Dispatch Plan

[0084] The scheduling decision module will perform multi-objective optimization based on the above-mentioned predicted runoff process.

[0085] Input constraints:

[0086] The highest operating water level of the reservoir is 195.0 meters (design flood level).

[0087] Minimum operating water level of the reservoir: 170.0 meters (dead water level)

[0088] Maximum discharge flow: The safe discharge flow rate in the downstream river channel is 5000 cubic meters per second.

[0089] Water level fluctuation limit: The water level shall not rise or fall by more than 3 meters within 24 hours.

[0090] Setting the scheduling objective: The objective of this scheduling is to "maximize power generation efficiency while prioritizing flood control safety." That is, firstly, ensure that the highest water level of the reservoir does not exceed the design flood level and the outflow does not exceed the safe discharge capacity. Under this premise, maximize the power generation head and water volume as much as possible.

[0091] Algorithm Execution: The "Multi-Objective Optimization Unit" built into the scheduling decision module employs the NSGA-II algorithm (Non-Dominated Sorting Genetic Algorithm), with a population size of 200 and an evolutionary generation of 500. Under the premise of satisfying all constraints, the algorithm generates a Pareto front solution set containing 48 non-dominated scheduling schemes. Each scheme corresponds to a set of hourly reservoir outflow processes for the next 72 hours.

[0092] Through the interactive interface of the "Scheme Optimization Unit," dispatchers, based on the current forecast consultation conclusion that "significant rainfall is still expected in the later stages," selected a relatively conservative plan: pre-emptive discharge to lower the reservoir level, followed by flood control and storage of the remaining water. The key operations corresponding to this plan are: within the next 12 hours, gradually increase the discharge to 1800 cubic meters per second, pre-discharging the reservoir level from 185.0 meters to 183.5 meters; when the flood peak arrives, control the maximum discharge to 2500 cubic meters per second, achieving a peak reduction rate of 22%; after the flood peak, reduce the discharge to store the remaining water, raising the water level to 192.0 meters for subsequent power generation.

[0093] Step S5: Output and Instruction Generation

[0094] Finally, the output and display module executes:

[0095] Visualization: The map within the GIS display unit dynamically shows the next 72 hours:

[0096] A comparison chart of the flow rates for predicted inbound and planned outbound shipments.

[0097] The reservoir water level change curve clearly marks the highest water level (192.8 meters, below the safe upper limit of 195.0 meters).

[0098] The flood inundation risk range of key downstream sections (the results show that all sections are below the warning level).

[0099] Operation instruction generation: The instruction generation unit automatically converts the selected scheduling plan into a specific sequence of operation instructions, including:

[0100] Hours 1-12: The gate opening is gradually increased from 2.0 meters to 3.5 meters, corresponding to a discharge flow of 1200-1800 cubic meters per second.

[0101] Hours 13-28: Maintain the gate opening at 4.0 meters and control the discharge to not exceed 2500 cubic meters per second.

[0102] Hours 29-72: Adjustments are made dynamically based on actual water inflow, gradually closing the gates. These instructions are sent directly to the reservoir's automatic gate control system via industrial Ethernet, enabling precise execution with minimal or no human intervention.

[0103] System feedback correction

[0104] After this round of scheduling concluded (July 18, 8:00 AM), the system's "feedback correction module" automatically activated. It compared the actual rainfall and inflow over the past 72 hours with the model's previous predictions, time-by-time. Calculations showed that the predicted peak flow error was +5%, and the peak occurrence time error was -1 hour. Using this error information, the feedback correction module fine-tuned the Xin'anjiang model's channel confluence parameters (such as Muskinggan coefficients) online using a Bayesian update algorithm to improve the accuracy of the next prediction. This iterative process ensures the system maintains high adaptability and reliability during the reservoir's long-term operation.

[0105] Example 2: Supplementary Case of Dry Season Scheduling for Ecological Protection

[0106] Unlike Example 1, during the dry season (e.g., January 2025), the scheduling objective of Biyuan Reservoir shifts to "maximizing ecological flow guarantee and water supply satisfaction." At this time, numerical weather forecasts indicate no effective rainfall for the next 7 days. The implementation process of this invention is adjusted as follows:

[0107] In step S3, the data assimilation unit employs a particle filtering algorithm to better adapt to the low-flow, nonlinear hydrological processes during the dry season.

[0108] In step S4, two constraints are added to the scheduling objectives: "downstream ecological base flow not less than 50 cubic meters per second" and "urban water supply guarantee rate not less than 95%". The multi-objective optimization algorithm (MOEA / D algorithm used in this example) generates multiple scheduling plans characterized by different discharge patterns under a low inflow scenario. The final selected plan is: "uniform discharge to protect the ecology," meaning that the average daily discharge flow will remain stable at 55 cubic meters per second for the next 7 days, the reservoir water level will slowly decrease, and the expected final water level will remain above 175.0 meters, higher than the minimum operating water level, thus simultaneously meeting the requirements of ecology, water supply, and the reservoir's own safety.

[0109] Through the detailed description of the two specific embodiments above, it can be clearly seen that the reservoir scheduling method and system based on hydrological simulation provided by the present invention can flexibly, accurately and efficiently generate and execute scientific scheduling plans according to different watershed characteristics and scheduling objectives, significantly improving the comprehensive utilization benefits of water resources and the level of flood control safety.

[0110] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention. The actual content is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar embodiments and examples without departing from the spirit of the present invention, they should all fall within the protection scope of the present invention.

Claims

1. A reservoir scheduling method based on hydrological simulation, characterized in that, include: Step S1: Obtain historical hydrological data, real-time monitoring data, and numerical weather prediction data for the watershed where the target reservoir is located; Step S2: Based on the historical hydrological data, construct and screen a distributed hydrological model suitable for the watershed; Step S3: Use the real-time monitoring data to update the state variables of the distributed hydrological model, and use the numerical weather forecast data as the model input to simulate and predict the inflow process in the future period; Step S4: Based on the simulated and predicted inflow process, and combined with the engineering constraints and scheduling objectives of the target reservoir, generate at least one reservoir scheduling plan; Step S5: Output the visualization results of the reservoir scheduling plan, and generate operation instructions for controlling the reservoir water release facilities according to the reservoir scheduling plan.

2. The reservoir scheduling method based on hydrological simulation according to claim 1, characterized in that, The distributed hydrological model constructed in step S2 is either the SWAT model based on physical mechanisms or the Xin'anjiang model based on conceptual mechanisms; the screening adopts the SCE-UA global optimization algorithm or the Bayesian parameter estimation method.

3. The reservoir scheduling method based on hydrological simulation according to claim 1, characterized in that, The data assimilation update in step S3 uses an ensemble Kalman filter or particle filter algorithm.

4. The reservoir scheduling method based on hydrological simulation according to claim 1, characterized in that, The scheduling objectives include at least one of the following: flood control safety objective, power generation efficiency maximization objective, ecological flow guarantee objective, and water supply satisfaction objective; the engineering constraints include at least one of the following: maximum operating water level of the reservoir, minimum operating water level of the reservoir, upper limit of outflow, and water level fluctuation limit.

5. The reservoir scheduling method based on hydrological simulation according to claim 1, characterized in that, The process of generating at least one reservoir scheduling plan in step S4 specifically includes: generating a Pareto front solution set using a multi-objective optimization algorithm for scheduling personnel to select from.

6. A reservoir scheduling system based on hydrological simulation, characterized in that, include: Data acquisition module: used to acquire historical hydrological data, real-time monitoring data and numerical weather forecast data of the watershed where the target reservoir is located; Hydrological simulation module: used to construct and screen a distributed hydrological model based on the historical hydrological data, use the real-time monitoring data to update the state variables of the distributed hydrological model, and use the numerical weather forecast data as the model input to simulate and predict the inflow process of the reservoir in future periods. Scheduling decision module: Based on the inflow runoff process prediction results output by the hydrological simulation module, and combined with the engineering constraints and scheduling objectives of the target reservoir, it generates at least one set of reservoir scheduling plans. Output and Display Module: Used to output the visualization results of the reservoir scheduling plan and generate operation instructions for controlling the reservoir's water release facilities.

7. The reservoir scheduling system based on hydrological simulation according to claim 6, characterized in that, The hydrological simulation module integrates a data assimilation unit, which uses an ensemble Kalman filter or particle filter algorithm to periodically assimilate the real-time monitoring data into the state variables of the distributed hydrological model.

8. The reservoir scheduling system based on hydrological simulation according to claim 6, characterized in that, The scheduling decision module includes: a scenario generation unit, used to generate multiple water inflow scenarios based on sources of uncertainty; a multi-objective optimization unit, used to solve the Pareto optimal scheduling scheme set under the engineering constraints using the NSGA-II algorithm or the MOEA / D algorithm; and a scheme selection unit, used to provide an interactive interface, receive user-input preference weights, and select a final scheduling plan from the Pareto optimal scheduling scheme set.

9. The reservoir scheduling system based on hydrological simulation according to claim 6, characterized in that, The output and display module includes: a GIS display unit, used to dynamically display the predicted runoff process, reservoir water level change process, and downstream inundation risk range on a geographic information system map; and an instruction generation unit, used to automatically convert the final scheduling plan into gate opening control instructions or discharge flow setting values.

10. The reservoir scheduling system based on hydrological simulation according to claim 6, characterized in that, Also includes: Feedback correction module: It is used to compare the actual hydrological process with the inflow process predicted by the hydrological simulation module, calculate the prediction error, and use the error to correct the parameters of the distributed hydrological model online.

Images