Reservoir flood control dispatching system and method

Abstract

The invention discloses a reservoir flood control dispatching system and method, and belongs to the technical field of reservoir flood control, and the dispatching system comprises an incoming water forecasting module which employs a reservoir incoming water forecasting model to forecast flood or runoff incoming water; the flood control dispatching model is used for carrying out flood control dispatching under the control of water balance and power balance; the reservoir real-time scheduling module comprises an incoming water forecast result correction unit which is used for accessing flood or runoff incoming water forecast results in real time and correcting the forecast results in real time; the scheduling model unit is used for selecting a scheduling scheme or model according to the scheduling model configuration and the incoming water forecast result, and correcting the scheduling model at a fixed time period; the scheduling result unit is used for calculating a scheduling result in real time according to the real-time incoming water forecast result and a real-time scheduling model or scheme; and the scheduling instruction unit is used for generating a scheduling instruction according to the scheduling result in a fixed format. The system improves the digitization level of reservoir dispatching and enhances the flood control capacity of the reservoir.

Description

Technical Field

[0001] The invention belongs to the technical field of flood control, and in particular relates to a reservoir flood control dispatching system and method. Background Art

[0002] Reservoir flood control dispatching models have been widely used in actual business. However, with the increasing uncertainty brought about by climate change, there is an urgent need to develop dispatching models that can reflect uncertainty, accurately predict the size of risks, and improve the safety of dispatching decisions based on existing models and theories. As flood control engineering systems continue to expand and become more complex, the needs and requirements for flood control dispatching are also increasing day by day.

[0003] Reservoir flood control scheduling problems are characterized by high complexity, multi-objective attribute decision-making, multi-stage attribute decision-making, strong uncertainty and certain risks. Conventional scheduling mainly completes flood control decisions based on scheduling diagrams or scheduling rules. Many domestic scholars have conducted extensive research and applied their research results to engineering practice. Flood control optimization scheduling is expressed with the help of mathematical models. Generally, the flood control optimization scheduling model is established with the lowest maximum flood level of the reservoir, the minimum maximum utilization of flood control storage capacity, the maximum peak reduction, the minimum upstream flooding loss, the shortest downstream disaster duration, or the minimum downstream flood control area flood volume as the optimization goals. Mathematical programming methods and intelligent algorithms are used to solve the optimal solution, so as to provide accurate and powerful flood information decision support.

[0004] However, the integration of reservoir flood control optimization and advanced digital twin technology is not high at present, and the advancement of information technology has not been fully utilized. The prediction accuracy of model input data is not high, resulting in low calculation efficiency of reservoir scheduling model and slow emergency flood control response.

[0005] So far, traditional reservoir dispatching methods have the following disadvantages:

[0006] 1) The conventional scheduling calculation method is simple and easy to operate, but it is too dependent on historical information, has a poor ability to accept real-time information and forecast information, and is limited to historical flood thresholds. If a flood beyond historical records occurs, it will exceed the dispatchable range and cause unnecessary losses.

[0007] 2) The integration of reservoir flood control optimization and advanced digital twin technology is not high, and the advanced nature of information technology has not been fully utilized;

[0008] 3) The low accuracy of reservoir inflow forecasting leads to low computational efficiency of reservoir operation models and slow emergency flood control response;

[0009] 4) The integrated intelligent dispatching system of “simulation-analysis-dispatching” for reservoir flood control has not yet been integrated and developed, and the semi-graphic method of traditional hydrological calculation is time-consuming and labor-intensive. Summary of the invention

[0010] In order to improve the accuracy and efficiency of reservoir inflow forecast, enhance the support for reservoir flood control dispatching decision-making, and strengthen digital twin water conservancy construction, the present invention provides a reservoir flood control dispatching system and method, which improves the digital level of reservoir dispatching and enhances the flood control capacity of the reservoir.

[0011] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0012] In a first aspect, the present invention provides a reservoir flood control dispatching system, comprising:

[0013] The water inflow forecast module uses the reservoir water inflow forecast model to forecast flood or runoff water inflow;

[0014] Flood control dispatch model, which conducts flood control dispatch based on water balance and dynamic balance;

[0015] Reservoir real-time dispatching module, which includes:

[0016] The water inflow forecast result correction unit is used to access the flood or runoff water inflow forecast results in real time and correct the forecast results in real time;

[0017] The dispatch model unit is used to select a dispatch scheme or model according to the dispatch model configuration and the water inflow forecast results, and to calibrate the dispatch model at regular intervals;

[0018] A dispatch result unit, used to calculate the dispatch result in real time according to the real-time water inflow forecast results and the real-time dispatch model or plan;

[0019] The scheduling instruction unit is used to generate scheduling instructions according to a fixed format based on the scheduling results.

[0020] As a further improvement of the present invention, it also includes: a reservoir group joint dispatching module, which includes:

[0021] The reservoir group optimization dispatching unit is used for flood control-oriented reservoir groups, which adopt compensation mode dispatching, with the upstream reservoir of the cascade reservoir or the parallel reservoir far from the flood protection zone being compensated first, so that the reservoir with a larger proportion of flood control, higher flood regulation capacity and closer to the downstream flood protection zone will control the discharge volume last;

[0022] The real-time dispatching unit of the reservoir group is used to realize joint flood control dispatching for the reservoir group and interval combined basin with dispatching capacity, taking into account the comprehensive peak-shaving capacity of the reservoir group, and ensuring the maximization of the comprehensive water storage capacity of the reservoir group under the premise of ensuring the safety of the reservoir group, interval and upstream and downstream.

[0023] As a further improvement of the present invention, the reservoir group real-time dispatching unit comprises:

[0024] Establish a reservoir group joint dispatch model sub-unit, which is used to formulate a reservoir group joint dispatch mode and establish a reservoir group joint dispatch model based on the single reservoir dispatch, according to the geographical location of each reservoir, the topological relationship of the water system, the mutual influence of upstream and downstream, the proportion of flood control, the size of flood control storage capacity, the comprehensive utilization task, the interval combination and encounter of floods;

[0025] The joint dispatching scheme formulation subunit is used to carry out joint dispatching and optimization decision-making based on the water inflow forecast of each upstream and downstream reservoir and interval, and the joint dispatching model of the reservoir group, using the measures of early pre-discharge, compensation regulation, and peak shaving, to determine the dispatching indicators of the water storage index, flood discharge sequence, flood discharge time, and gate opening volume of each upstream and downstream reservoir, and formulate a joint dispatching scheme;

[0026] The sub-unit for issuing joint dispatching instructions is used to analyze whether the upstream and downstream reservoirs are safe, and whether the various sections and upstream and downstream rivers exceed the standards based on the joint dispatching plan; if unsafe or exceeding conditions occur, disaster warning information and disposal measures are generated in combination with the flood control plans of each reservoir and town, and joint dispatching instructions are issued. At the same time, changes in the water conditions of each reservoir gate and river channel are collected for comparison, and plans and warning decisions are revised in real time.

[0027] As a further improvement of the present invention, it also includes:

[0028] Reservoir routine dispatch module, used to guide reservoir operation based on historical data;

[0029] The reservoir optimization scheduling module is used to calculate the outflow process of the reservoir or the change of the reservoir water storage according to the reservoir scheduling rules;

[0030] As a further improvement of the present invention, it also includes: a scheduling scheme management module, which includes:

[0031] A scheduling plan query unit, used to query the scheduling plan details by the plan name or the time when the scheduling plan was generated, and export the scheduling plan;

[0032] A scheduling scheme management unit, used to start implementing a scheduling scheme or to view and delete a scheduling scheme by scheme name or time when the scheduling scheme was generated;

[0033] The scheduling scheme comparison unit is used to compare and analyze different scheduling schemes for a group of reservoirs or a single reservoir in the basin, and compare and analyze their highest water levels and their occurrence time, maximum inflows and their occurrence time, maximum outflows and their occurrence time, and inflow process lines.

[0034] As a further improvement of the present invention, the reservoir water inflow forecasting model adopts the Sanshuiyuan Xin'anjiang model and the API model to forecast the reservoir inflow flood, and adopts the VMD-APSO-LSTM model to forecast the runoff;

[0035] The Sanshuiyuan Xin'anjiang model adopts the gradient descent method to determine the model parameter calibration;

[0036] The VMD-APSO-LSTM model is used to predict runoff, specifically including: determining the penalty factor α of runoff data at each scale by a parameter adjustment method, determining different decomposition levels K by a center frequency method, inputting the determined parameters into VMD to obtain multiple sub-modes; each sub-mode is input into the APSO-LSTM model as an independent sequence, setting the data before decomposition as input and the corresponding data of the next year as output for training, and using the model for forecasting after the training is completed; and accumulating the sub-sequences obtained by each forecast accordingly to obtain the final forecast result.

[0037] As a further improvement of the present invention, the flood control dispatch model selects a typical flood process from measured data, and then amplifies it according to the designed flood peak flow and the designed flood volume to obtain the designed flood process line;

[0038] The model has built-in multiple flood processes. According to the relationship between the water level and storage capacity of the reservoir, the water level-storage capacity relationship curve is fitted, and each calculation point in the relationship line is obtained by linear interpolation. The calculation structure is constructed together with the reservoir water balance formula and the Saint-Venant equation group. The dynamic equation used is fitted and calculated using the storage-discharge relationship curve of the reservoir area, and the cyclic trial algorithm is used to calculate the discharge flow.

[0039] As a further improvement of the present invention, the real-time flood control dispatching process of the flood control dispatching model includes:

[0040] Extract and sort the peak flow of the design flood process line under each return period, and use it as the basis for classifying the flood level entering the reservoir; classify the flood into levels, read the peak flow data in the flood forecast data entering the reservoir, classify the predicted flood according to the data, and obtain the flood situation estimation level;

[0041] According to the forecast information of the forecast period, the operation strategy of the forecast period is obtained, and then the decision of the facing period is selected from the operation strategy for implementation;

[0042] Before implementing the optimization strategy for the next forecast period of the water diversion project, the water use forecast information for the next period is corrected using correlation analysis and other techniques based on the actual water use information of the previous periods, and the actual state of the project at the end of the previous period is used to replace the reservoir water storage calculated in the previous time, as the initial value of the actual state of the project for the next period. Then, flood dispatching for the next forecast period is continued, and the corresponding operation strategy is obtained, and the decision of the first period is selected for implementation. In this way, the forecast information and project status of the forecast period are continuously updated and corrected in sequence and time series, and real-time operation decisions are made for each period.

[0043] As a further improvement of the present invention, the correction process of the flood control dispatch model includes:

[0044] According to the conclusions drawn from the flood situation forecast, floods are divided into two categories: minor floods and super-standard floods. For minor floods, water storage is given priority while ensuring that the reservoir water level does not exceed the flood limit water level. The discharge flow is calculated based on the real-time forecast inflow data.

[0045] When the situation is predicted to be an extraordinary flood, the dispatching plan corresponding to the matching design flood process line is corrected in real time in combination with the real-time inflow forecast data, and the corresponding downstream flow data is output hourly.

[0046] In a second aspect, the present invention provides a reservoir flood control dispatching method, comprising:

[0047] Use reservoir water forecasting models to forecast flood or runoff water;

[0048] Carry out flood control operations based on water balance and dynamic balance;

[0049] Real-time access to flood or runoff water forecast results, and real-time correction of forecast results;

[0050] Select the dispatching scheme or model according to the dispatching model configuration and water inflow forecast results, and calibrate the dispatching model at regular intervals;

[0051] Calculate the dispatch results in real time based on the real-time water inflow forecast results and the real-time dispatch model or plan;

[0052] The scheduling results generate scheduling instructions according to a fixed format.

[0053] In a third aspect, the present invention provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the reservoir flood control scheduling method is implemented when the processor executes the computer program.

[0054] In a fourth aspect, the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and the computer program implements the reservoir flood control scheduling method when executed by a processor.

[0055] In a fifth aspect, the present invention provides a computer program product, which includes computer instructions, and the computer instructions instruct a computer to execute the reservoir flood control scheduling method.

[0056] The beneficial effects of the present invention compared with the prior art are as follows:

[0057] The present invention proposes a reservoir flood control dispatching system based on a typical scenario of digital twins, aiming to construct a forecasting model for high-precision reservoir inflow forecasting, cooperate with digital twin technology and reservoir dispatching model, integrate and develop intelligent dispatching system, realize accurate advance forecasting of reservoir multi-core, multi-scheme, and multi-scale hydrological situation, effectively solve the problem of low digital level and insufficient calculation level of reservoir dispatching, improve the response efficiency of reservoir flood control dispatching, and enhance the flood control capacity of reservoir. Different from the existing forecasting model, the present invention combines distributed hydrological forecasting, high-resolution remote sensing, and machine learning technology to construct a reservoir inflow runoff forecasting model that considers the influence of human activities and the conditions of underlying surface inconsistency, significantly improving the accuracy and efficiency of reservoir inflow forecasting. The present invention integrates digital twin technology and traditional reservoir dispatching, develops a "simulation-analysis-dispatching" reservoir flood control integrated intelligent dispatching system, effectively solves the problem of low digital level and insufficient calculation level of reservoir dispatching, and improves the response efficiency of reservoir flood control dispatching. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 This is the flow chart of the Sanshuiyuan Xin'anjiang model;

[0059] Figure 2 Run the framework diagram for LSTM;

[0060] Figure 3 This is a diagram of the internal operation mechanism of LSTM;

[0061] Figure 4 This is the schematic diagram of the flood control dispatch model;

[0062] Figure 5 This is the flow chart of the water supply scheduling model;

[0063] Figure 6 This is the overall technical architecture diagram;

[0064] Figure 7 This is the interface diagram of the reservoir dispatching system;

[0065] Figure 8 Provide a technical roadmap for joint operation of a group of reservoirs;

[0066] Fig. 9 This is a diagram of the scheduling plan management interface. DETAILED DESCRIPTION

[0067] The embodiments of the present invention are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and are not to be construed as limitations of the present invention. For the step numbers in the following embodiments, they are only provided for the convenience of explanation, and the order between the steps is not limited in any way, and the execution order of each step in the embodiment can be adaptively adjusted according to the understanding of those skilled in the art.

[0068] Explanation of terms

[0069] VMD-APSO-LSTM is a joint algorithm of variational mode decomposition VMD, particle swarm optimization APSO, and long short-term memory network LSTM;

[0070] Variational Mode Decomposition (VMD) is a time-frequency analysis method for signal processing. It decomposes complex signals into a series of intrinsic mode functions (IMFs) with different center frequencies and frequency widths. These IMFs can be regarded as the intrinsic oscillation modes of the signal with different frequency characteristics and energy distribution. The goal of VMD is to automatically decompose the signal into a set of optimized IMFs to better represent the intrinsic structure of the signal.

[0071] The adaptive particle swarm optimization algorithm (APSO) is an optimization strategy that improves the performance of the algorithm by dynamically adjusting the inertia weight based on the standard particle swarm optimization algorithm (PSO).

[0072] Long Short-Term Memory (LSTM) is a special type of recurrent neural network (RNN) that can learn long-term dependencies in data. LSTM introduces three gates (input gate, forget gate, and output gate) to control the flow of information, thereby solving the gradient vanishing or exploding problem of traditional RNN when processing long sequence data. LSTM is particularly suitable for time series prediction tasks because it can effectively capture long-term and short-term patterns in time series data.

[0073] The present invention proposes a reservoir (group) flood control dispatching method based on a typical scenario of digital twins. The key technical points of the invention are:

[0074] Build a high-precision forecast model for reservoir inflow forecasting, cooperate with digital twin technology and reservoir scheduling model, integrate and develop intelligent scheduling system, realize accurate advance forecast of multi-core, multi-scheme and multi-scale hydrological conditions of reservoirs, effectively solve the problems of low digitalization level and insufficient computing level of reservoir scheduling, improve the response efficiency of reservoir flood control scheduling, and enhance the flood control capacity of reservoirs.

[0075] Distributed hydrological forecasting, high-resolution remote sensing, machine learning and other means are incorporated into the hydrological model to achieve accurate and advanced forecasts of multi-core, multi-scheme and multi-scale hydrological conditions in reservoirs. Information and digital technologies such as the Internet of Things, cloud computing, big data, BIM, artificial intelligence, and digital twins are combined with traditional reservoir scheduling to develop a "simulation-analysis-scheduling" integrated intelligent scheduling system for reservoir flood control. The calculation principle uses trial algorithms to replace the semi-graphic method of traditional hydrological calculations, thereby giving full play to the computing power of computers and effectively solving the problems of low digitalization and insufficient computing levels in reservoir scheduling.

[0076] The first object of the present invention is to provide a reservoir flood control dispatching system, comprising:

[0077] The water inflow forecast module uses the reservoir water inflow forecast model to forecast flood or runoff water inflow;

[0078] Flood control dispatch model, which conducts flood control dispatch based on water balance and dynamic balance;

[0079] Reservoir real-time dispatching module, which includes:

[0080] The water inflow forecast result correction unit is used to access the flood or runoff water inflow forecast results in real time and correct the forecast results in real time;

[0081] The dispatch model unit is used to select a dispatch scheme or model according to the dispatch model configuration and the water inflow forecast results, and to calibrate the dispatch model at regular intervals;

[0082] A dispatch result unit, used to calculate the dispatch result in real time according to the real-time water inflow forecast results and the real-time dispatch model or plan;

[0083] The scheduling instruction unit is used to generate scheduling instructions according to a fixed format based on the scheduling results.

[0084] The reservoir flood control dispatching system proposed in this invention integrates a high-precision water forecast model, a flood control dispatching model, and advanced digital and information technology to build an integrated "simulation-analysis-dispatching" intelligent system. This system not only significantly improves the accuracy and response efficiency of reservoir flood control dispatching, but also enhances the overall flood control capacity of the reservoir. The following is a detailed analysis of its principles and advantages:

[0085] The system principle of the present invention is analyzed as follows:

[0086] The model integrates a variety of technical means such as distributed hydrological forecasting, high-resolution remote sensing, and machine learning, and realizes multi-core, multi-scheme, and multi-scale accurate advance forecasts of floods or runoff. This method of multi-source data fusion and algorithm integration effectively improves the accuracy and timeliness of forecasts.

[0087] The flood control dispatch model based on water balance and dynamic balance can guide the flood control dispatch of reservoirs in a scientific and reasonable manner. The model calculates key parameters such as the reservoir's water storage capacity and outflow flow to ensure that the reservoir can effectively reduce downstream flood pressure and maintain the safety and stability of the reservoir itself during the flood control process.

[0088] Inflow forecast result correction unit: real-time access and correction of forecast results to ensure the accuracy and timeliness of the dispatch basis.

[0089] Scheduling model unit: According to the scheduling model configuration and real-time water inflow forecast results, the scheduling plan or model can be flexibly selected and corrected at regular intervals to adapt to the changing hydrological situation.

[0090] Scheduling result unit: Based on real-time data and scheduling models, it quickly calculates scheduling results and provides a scientific basis for decision-making.

[0091] Dispatch instruction unit: converts dispatch results into specific dispatch instructions to ensure the timeliness and accuracy of dispatch operations.

[0092] Through the application of technologies such as the Internet of Things, cloud computing, big data, BIM (Building Information Modeling), artificial intelligence, and digital twins, the comprehensive digitization and intelligentization of reservoir flood control dispatching have been achieved. These technologies not only improve the ability of data processing and analysis, but also make the dispatching process more intuitive, efficient, and controllable.

[0093] The fusion of multi-source data and the integration of multiple algorithms have significantly improved the forecast accuracy of flood or runoff water, providing a more reliable basis for flood control dispatching. Real-time access and correction of forecast results, as well as rapid calculation of dispatch results and generation of dispatch instructions, have greatly improved the response efficiency of reservoir flood control dispatching. Through scientific and reasonable flood control dispatching and real-time and accurate control, the flood pressure in downstream areas has been effectively reduced.

[0094] The use of advanced digital and information technology has significantly improved the digital level of reservoir flood control and dispatching, laying the foundation for intelligent management of reservoirs. Through the integrated application of high-precision forecasting models, advanced dispatching models, and digital and information technology, the reservoir flood control and dispatching has been made precise, intelligent, and efficient, providing strong support for improving the flood control capacity of reservoirs and protecting people's lives and property.

[0095] The present invention incorporates distributed hydrological forecasting, high-resolution remote sensing, machine learning and other means into the hydrological model, constructs a reservoir inflow forecast model that takes into account the impact of human activities and underlying surface inconsistency conditions, and realizes accurate advance forecasts of multi-core, multi-scheme and multi-scale hydrological conditions in reservoirs, significantly improving the accuracy and efficiency of reservoir inflow forecasts, and laying a solid foundation for reservoir flood control optimization scheduling; it utilizes information and digital technology means such as the Internet of Things, cloud computing, big data, BIM, artificial intelligence, digital twins, etc. in combination with traditional reservoir scheduling to develop an intelligent scheduling system. The calculation principle uses a trial algorithm to replace the semi-graphic method of traditional hydrological calculations, giving full play to the computing power advantages of computers, improving the digital level of reservoir scheduling, and enhancing the flood control capacity of reservoirs.

[0096] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0097] The present invention provides a reservoir (group) flood control dispatching method based on a typical digital twin scenario, including the construction and optimization of a reservoir inflow forecasting model, the construction and optimization of a reservoir flood control dispatching model, and the construction of an intelligent dispatching system.

[0098] 1. Construction and optimization of reservoir water inflow forecast model

[0099] This application uses the Sanshuiyuan Xin'anjiang model and API model to forecast the inflow flood of the reservoir, and uses the VMD-APSO-LSTM model for medium-term (monthly scale) runoff forecast.

[0100] Among them, VMD-APSO-LSTM is a hybrid prediction model that combines variational mode decomposition (VMD), adaptive adjusted inertialweight particle swarm optimization (APSO) and long short-term memory network (LSTM).

[0101] The VMD-APSO-LSTM model achieves high-precision prediction of complex time series data by combining the advantages of VMD, APSO and LSTM. VMD is used for data preprocessing to decompose the original signal into multiple stationary subsequences; APSO is used to optimize the parameter configuration of VMD and LSTM; and LSTM is used to predict each subsequence. This hybrid prediction model has broad application prospects in the fields of power load prediction and photovoltaic power prediction.

[0102] Reservoir group joint dispatching module, which includes:

[0103] The reservoir group optimization dispatching unit is used for flood control-oriented reservoir groups, which adopt compensation mode dispatching, with the upstream reservoir of the cascade reservoir or the parallel reservoir far from the flood protection zone being compensated first, so that the reservoir with a larger proportion of flood control, higher flood regulation capacity and closer to the downstream flood protection zone will control the discharge volume last;

[0104] The real-time dispatching unit of the reservoir group is used to realize joint flood control dispatching for the reservoir group and interval combined basin with dispatching capacity, taking into account the comprehensive peak-shaving capacity of the reservoir group, and ensuring the maximization of the comprehensive water storage capacity of the reservoir group under the premise of ensuring the safety of the reservoir group, interval and upstream and downstream.

[0105] The reservoir water inflow forecasting model of the present invention adopts the Sanshuiyuan Xin'anjiang model and the API model to forecast the reservoir inflow flood, and adopts the VMD-APSO-LSTM model to forecast the runoff;

[0106] The Sanshuiyuan Xin'anjiang model adopts the gradient descent method to determine the model parameter calibration;

[0107] The VMD-APSO-LSTM model is used to predict runoff, specifically including: determining the penalty factor α of runoff data at each scale by a parameter adjustment method, determining different decomposition levels K by a center frequency method, inputting the determined parameters into VMD to obtain multiple sub-modes; each sub-mode is input into the APSO-LSTM model as an independent sequence, setting the data before decomposition as input and the corresponding data of the next year as output for training, and using the model for forecasting after the training is completed; and accumulating the sub-sequences obtained by each forecast accordingly to obtain the final forecast result.

[0108] 1.1 Three water sources Xinanjiang model

[0109] (1) Sanshuiyuan Xin'anjiang model structure

[0110] The Sanshuiyuan Xin'anjiang model consists of four modules: evapotranspiration, runoff generation, water source division, and confluence. The evapotranspiration module uses a three-layer evapotranspiration model to calculate evaporation, the runoff generation module calculates runoff based on the full runoff generation mechanism, the water source division module uses a free water reservoir structure to divide water sources, and the confluence module uses a linear reservoir or hysteresis algorithm and Muskingum piecewise continuous algorithm to calculate confluence. The calculation process of the model is detailed in Figure 1 . Figure 1 This is the flow chart of the Sanshuiyuan Xin'anjiang model.

[0111] (2) Initial values ​​of the model parameters of Sanshuiyuan Xin'anjiang

[0112] The numerous parameters of the Three Water Sources Xin'anjiang model can be divided into four levels. The first level is the parameters for evapotranspiration calculation, including K, UM, LM, and C; the second level is the parameters for runoff calculation, including WM, B, and IM; the third level is the parameters for water source division, including SM, EX, KG, and KI; the fourth level is the parameters for confluence calculation, including CI, CG, CS, L, KE, and XE. The four levels of parameters are independent of each other, while the parameters in the same level have a certain degree of interdependence. The empirical value ranges of some parameters are shown in Table 1 below.

[0113] Table 1. Empirical range of parameters of the Sanshuiyuan Xin'anjiang model

[0114] parameter Parameter Description Parameter value range K Evapotranspiration calculation system 0.1-1.0 C Deep evapotranspiration coefficient 0.01-0.02 B Power of water storage capacity curve on permeable area 0.2-0.5 EX Power of free water storage flux curve 1.0-1.5 KG Outflow coefficient of free water reservoir replenishing groundwater runoff 0-0.7 KSS Outflow coefficient of free water reservoir supplementing subsoil flow 0-0.7 IM Ratio of impervious area to total basin area 0-0.05 KKG Subsurface runoff decay coefficient 0.3-0.9 KKSS Soil flow recession coefficient 0.3-0.9 SM Freewater Reservoir Storage Capacity 10-30 Cr Coefficient reflecting the characteristics of the river basin 0.02-0.03 UM Water content of upper soil layer 5-20 LM Water content of lower soil layer 60-90 DM Deep soil water content 0-60

[0115] (3) Model parameter calibration based on gradient descent method

[0116] The gradient decent algorithm (GDA) is one of the most basic and widely used algorithms in the field of deep learning. It is a commonly used method for solving unconstrained optimization problems. The algorithm can gradually iterate in the opposite direction of the gradient and finally find the minimum value of the loss function and its corresponding parameter value. The gradient descent method has the advantages of small storage, simple principle and structure, strong operability and practicality. In actual hydrological applications, if there are multiple reference variables, the parameters of the multivariate linear regression equation can be easily solved with the help of the gradient descent method. According to the different ways of processing the training set, the gradient descent method is divided into batch gradient descent method, stochastic gradient descent method and small batch gradient descent method. The small batch gradient descent method is selected for the parameter rate of this model.

[0117] The basic idea of ​​the mini-batch stochastic gradient descent method is: in each iteration, a small number of random samples are used to calculate the gradient, and then the parameter value is updated according to the gradient. The calculation steps are as follows:

[0118] The first step is to initialize the parameters w and b. Then, in each iteration, randomly select a small number of samples to calculate the gradient. If m samples are randomly selected from the training set, the gradient can be calculated using the following formula:

[0119]

[0120] Where L(w, b, x, y) is the loss function, represents the gradient of w and b.

[0121] In the second step, according to the value of the gradient, the parameter values ​​are updated using formulas (28) and (29).

[0122]

[0123] In the formula, α is the learning rate, which indicates the step size of each update.

[0124] The third step is to repeat the above process until the model converges.

[0125] 1.2 Medium-term (monthly) runoff forecast

[0126] The VMD-APSO-LSTM model is applied to medium-term (monthly) runoff forecast. The penalty factor α of runoff data at each scale is determined through a series of manual parameter adjustment methods. The center frequency method is used to determine the different decomposition levels K. The determined parameters are input into VMD to obtain multiple sub-modes. Each sub-mode is input into the APSO-LSTM model as an independent sequence. The data of the previous year (or month) obtained by decomposition is set as input and the corresponding data of the next year is set as output for training. After the model training is completed, it is used for forecasting. The sub-sequences obtained from each forecast are accumulated accordingly to obtain the final forecast result.

[0127] (1) VMD algorithm

[0128] VMD is an adaptive, completely non-recursive sequence decomposition method. This method can effectively reduce the "modal aliasing" phenomenon in empirical mode algorithms. VMD is more effective for non-stationary, nonlinear and complex signal sequences.

[0129] (2) LSTM model

[0130] Long short-term memory network (LSTM) is a recursive neural network that better captures the dependencies with large intervals in time series data and solves the possible decay and explosion in the gradient of hidden layer variables of recurrent neural network (RNN). The hidden state of LSTM includes hidden layer variables and memory cells. Its basic operating structure is as follows Figure 2 shown. Figure 2 Run the framework for LSTM; Figure 3 This is the internal operating mechanism of LSTM.

[0131] (3) APSO model

[0132] Particle swarm optimization (PSO) is a group intelligence optimization algorithm based on the study of bird predation behavior. Its basic idea is to find the optimal solution through collaboration and information sharing among individuals in the group. The main process is to find the optimal solution through iteration after generating a series of random particles (random solutions). In each iteration, particles update themselves by constantly approaching two extreme values. After obtaining the two extreme values, particles update their speed and position.

[0133] 2. Construction and optimization of reservoir flood control operation model

[0134] The flood control dispatch model selects a typical flood process from measured data, and then amplifies it according to the designed flood peak flow and designed flood volume to obtain the designed flood process line;

[0135] The model has built-in multiple flood processes. According to the relationship between the water level and storage capacity of the reservoir, the water level-storage capacity relationship curve is fitted, and each calculation point in the relationship line is obtained by linear interpolation. The calculation structure is constructed together with the reservoir water balance formula and the Saint-Venant equation group. The dynamic equation used is fitted and calculated using the storage-discharge relationship curve of the reservoir area, and the cyclic trial algorithm is used to calculate the discharge flow.

[0136] Furthermore, the real-time flood control dispatching process of the flood control dispatching model includes:

[0137] Extract and sort the peak flow of the design flood process line under each return period, and use it as the basis for classifying the flood level entering the reservoir; classify the flood into levels, read the peak flow data in the flood forecast data entering the reservoir, classify the predicted flood according to the data, and obtain the flood situation estimation level;

[0138] According to the forecast information of the forecast period, the operation strategy of the forecast period is obtained, and then the decision of the facing period is selected from the operation strategy for implementation;

[0139] Before implementing the optimization strategy for the next forecast period of the water diversion project, the water use forecast information for the next period is corrected using correlation analysis and other techniques based on the actual water use information of the previous periods, and the actual state of the project at the end of the previous period is used to replace the reservoir water storage calculated in the previous time, as the initial value of the actual state of the project for the next period. Then, flood dispatching for the next forecast period is continued, and the corresponding operation strategy is obtained, and the decision of the first period is selected for implementation. In this way, the forecast information and project status of the forecast period are continuously updated and corrected in sequence and time series, and real-time operation decisions are made for each period.

[0140] Furthermore, the modification process of the flood control dispatch model includes:

[0141] According to the conclusions drawn from the flood situation forecast, floods are divided into two categories: minor floods and super-standard floods. For minor floods, water storage is given priority while ensuring that the reservoir water level does not exceed the flood limit water level. The discharge flow is calculated based on the real-time forecast inflow data.

[0142] When the situation is predicted to be an extraordinary flood, the dispatching plan corresponding to the matching design flood process line is corrected in real time in combination with the real-time inflow forecast data, and the corresponding downstream flow data is output hourly.

[0143] 2.1 Flood control dispatch model

[0144] Flood control dispatch is based on water balance and dynamic balance (i.e., the continuity equation and motion equation of the Saint-Venant equations). The water balance is represented by the reservoir water balance equation, and the dynamic equation is represented by the reservoir storage and discharge equation (or storage and discharge curve). The calculation principle uses a trial algorithm to replace the semi-graphic method of traditional hydrological calculations, giving full play to the computing power of computers and solving these two equations continuously in each period. Figure 4 This is the schematic diagram of the flood control scheduling model.

[0145] Flood control dispatch is based on water balance and dynamic balance (i.e., the continuity equation and motion equation of the Saint-Venant equations). The water balance is represented by the reservoir water balance equation, and the dynamic equation is represented by the reservoir storage and discharge equation (or storage and discharge curve). The calculation principle uses a trial algorithm to replace the semi-graphic method of traditional hydrological calculations, giving full play to the computing power of computers and solving these two equations continuously in each period.

[0146] Reservoir water balance equation

[0147] (Q1+Q2) / 2×Δt-(q1+q2) / 2×Δt=V2-V1(4 1)

[0148] Where: Q1, Q2: inflow flow at the beginning and end of the period, m 3 / s; q1, q2: outbound flow at the beginning and end of the period, m 3 / s; V1, V2: reservoir water storage at the beginning and end of the period, m 3 ; Δt: calculation period, s. The length should be chosen based on the principle of accurately reflecting the shape of the flood process line. For steep rises and falls, Δt should be shorter.

[0149] The Saint-Venant equations are a set of partial differential equations that describe the motion laws of non-steady, gradually varying water flows in waterways and other shallow bodies of water with free surfaces. They consist of continuity equations that reflect the law of conservation of mass and motion equations that reflect the law of conservation of momentum. The typical form of the Saint-Venant equations is:

[0150] Continuity equation:

[0151] Equations of motion:

[0152]

[0153] Where t is time; s is the distance from a fixed section of the waterway along the flow path; h, v, and Z0 are the water depth, average flow velocity, and bottom elevation of the water section at s, respectively; H f is the energy drop caused by friction loss; g is the gravitational acceleration; t and s are independent variables; h and v are dependent variables; Z0, Hf It can be determined by s, h and v. Formula (2) is the continuity equation, which reflects the water balance in the waterway, that is, the rate of change of storage (the first term) should be equal to the rate of change of flow along the way (the second term). Formula (3) is the equation of motion. The first term reflects the local acceleration of a fixed point, and the second term reflects the convection acceleration caused by the uneven spatial distribution of flow velocity. The above two terms are called inertia terms. The third term reflects the gravity caused by the bottom slope, which is called the gravity term. The fourth term reflects the influence of water depth, which is called the pressure term. The third and fourth terms can be combined into one term, namely the water surface gradient. The fifth term is the friction loss inside and at the boundary of the water flow. This formula expresses that the combined effect of gravity and pressure enables the water flow to overcome the energy loss caused by inertia and friction and gain acceleration.

[0154] Reservoir storage and discharge equation or reservoir storage and discharge curve

[0155] When a reservoir discharges floodwater through a spillway, the amount of discharge, when the type and size of the spillway are constant, depends on the water head H at the top of the weir, that is, q = f(H). For a surface spillway without a gate or with the gate fully open, the discharge flow can be calculated according to the weir flow formula; the discharge flow of a deep-water spillway can be calculated according to the pressurized pipe flow formula. When the slope of the water surface in the reservoir is small and can be regarded as a static water surface, the discharge head H is only a function of the water storage V in the reservoir, that is, H = f(V). Therefore, the discharge flow q becomes a function of the water storage V, that is,

[0156] q=f(H) (4-4)

[0157] q=f(V) (4-5)

[0158] This formula is the reservoir discharge equation or q=f(V) curve when the reservoir water surface is assumed to be horizontal. This curve is a combination of the static reservoir capacity curve and the discharge calculation formula. For narrow river-type reservoirs, when flood flow passes through, the water surface often presents a significant slope due to the influence of backwater. In this case, flood control calculations based on the static reservoir capacity curve often lead to large errors. Therefore, in order to meet the requirements of the accuracy of the results, dynamic reservoir capacity must be used for flood control calculations.

[0159] 2.2 Flood control dispatch during typical flood processes

[0160] First, a typical flood process is selected from the measured data, and then it is enlarged according to the design flood peak flow and design flood volume to obtain the design flood process line. The selection of typical flood processes follows the following principles: ① Select the measured large flood process line with high data completeness and accuracy and flood peak flow and flood volume close to the design value; ② Select representative measured flood processes whose flood season, flood duration, peak-volume relationship, main peak location, peak type, etc. can represent the characteristics of large floods in the basin; ③ Select typical flood process lines that are unfavorable to engineering flood control, and try to select floods with high peaks and large volumes, and concentrated peak types with the main peak at the back.

[0161] The model has built-in 19 typical flood processes. According to the relationship between reservoir water level and storage capacity, the water level-storage capacity relationship curve is fitted, and each calculation point in the relationship line is obtained by linear interpolation. The calculation structure is constructed by the reservoir water balance formula and the Saint-Venant equations. The dynamic equation used is fitted with the storage-discharge relationship curve of the reservoir area, and the cyclic trial algorithm is used to replace the semi-graphic method of traditional hydrological calculation to calculate the discharge flow, giving full play to the computing power of computers.

[0162] 2.3 Real-time flood control dispatch

[0163] (1) Flood situation forecast

[0164] The peak flow of the design flood process line under each return period is extracted and sorted, which is used as the basis for classifying the flood level entering the reservoir. Floods are divided into small floods (less than once in five years), once in five years, once in ten years, etc. The peak flow data in the flood forecast data entering the reservoir is read, and the predicted floods are classified according to the data to obtain the flood situation estimation level.

[0165] (2) Real-time flood control

[0166] The process of real-time dispatch is a process of "typical flood-flood regulation-real-time correction" that is recursively pushed forward in a sequential and period-by-period manner. That is, first, based on the forecast information of the forecast period, the typical operation strategy of the project in the forecast period is obtained by using an appropriate typical dispatch model, and then the decision of the facing period is selected from this strategy for implementation. When this facing period ends, the actual water inflow and water consumption of this facing period and the engineering status at the end of this period are known. Therefore, before the optimization strategy of the next forecast period of the water diversion project is carried out, it is necessary to correct the water inflow forecast information of the next facing period based on the actual water inflow information of the previous few periods, and use correlation analysis and other technologies, and take the actual state of the project at the end of the previous facing period to replace the reservoir water storage calculated last time, etc., as the initial value of the actual state of the project in the next facing period (real-time correction), and then continue to carry out flood dispatch in the next forecast period, obtain the corresponding operation strategy, and select the decision of the first period (that is, the next facing period) to implement. In this way, the forecast information and engineering status of the forecast period are continuously updated and corrected in sequence and time, and real-time operation decisions are made period by period.

[0167] According to the conclusions drawn from the flood situation forecast, floods are divided into two categories: small floods and super-standard floods. For small floods, water storage is given priority under the condition that the reservoir water level does not exceed the flood limit water level, and the discharge flow is calculated based on the real-time forecast inflow flow data. When the situation is predicted to be a super-standard flood, the dispatching plan corresponding to the matching design flood process line is corrected in real time in combination with the real-time inflow flow forecast data, and the corresponding discharge flow data is output hourly.

[0168] The water supply scheduling module establishes a scientific and practical water scheduling model based on the actual needs of reservoir scheduling problems, assists in reservoir water scheduling decisions, and develops a reservoir water scheduling system to alleviate the contradiction between available water, required water and water delivery capacity, while ensuring that the formulated water supply plan can be implemented safely and efficiently. Figure 5 The figure shows the flow chart of water supply scheduling model.

[0169] Multi-objective optimization scheduling model, reservoir scheduling must take into account the contradictions between the benefits of the reservoir, which is usually a multi-objective decision-making process. Reservoir optimization scheduling is based on the inflow process of runoff. Under certain constraints, the optimization algorithm is used to obtain the optimal operating strategy when the objective function reaches the extreme value. The system analysis method is used to study the optimization problem of the system.

[0170] The construction of the water supply optimization scheduling model includes key elements such as objective function, decision variables, constraints and optimization algorithm, and then obtains the optimal operation strategy based on the input data. The reservoir optimization scheduling model is mainly a scheduling model with water supply as the goal (monthly scale). The following example illustrates:

[0171] Construction of water supply optimization dispatching model for reservoir 1

[0172] The water supply scheduling optimization model of reservoir 1 takes the minimum water shortage in the water demand process during the reservoir water supply process and the minimum water replenishment amount of the reservoir as the optimal criteria.

[0173] 1) Objective function

[0174] Ⅰ.

[0175] Among them, Fs1 is the water supply objective function value of reservoir 1, F refers to the objective function value, s refers to water supply, and "1" refers to reservoir 1; Dt is the water demand in period t; S t is the water supply in period t; T is the total number of periods.

[0176] Ⅱ. The goal of minimizing reservoir water replenishment is reflected by the minimum annual reservoir water replenishment designed and added during model calculation.

[0177] 2) Decision variables

[0178] The water level of the water replenishment judgment line, the water level of the water supply limit line and the water replenishment volume of the user in each period are taken as decision variables.

[0179] 3) Constraints

[0180] Ⅰ. Water balance constraint: V t+1 =V t +R t -S t -Q t (4-7)

[0181] Where V t+1 is the reservoir capacity at the end of period t, m 3 , V t is the initial reservoir capacity in period t, m 3 ; R t is the amount of water entering the reservoir during period t, m 3 ; S t is the water supply in period t, m3; Q t is the reservoir water loss in period t, m 3 .

[0182] II. Reservoir capacity constraints: V tmin ≤V t ≤V tmax (4-8)

[0183] Where V tmin 、V tmax are the minimum and maximum water storage allowed in the reservoir during period t, m 3 .

[0184] III. Water supply constraint: S tmin ≤S t ≤S tmin (4-9)

[0185] In the formula, S tmin , S tmin are the minimum and maximum water supply of the reservoir during period t, m 3 .

[0186] IV. Water demand constraint: 0≤S t ≤D t (4-10)

[0187] Among them, S t is the water supply in the period, m 3 , D t is the water demand for the period, m 3 .

[0188] (2) Construction of water supply optimization scheduling model for reservoir 2

[0189] The water supply scheduling optimization model of Reservoir 2 takes the minimum water shortage in the water demand process during the reservoir water supply process and the minimum water replenishment amount of the reservoir as the optimal criteria.

[0190] 1) Objective function

[0191] Ⅰ.

[0192] Among them, Fs2 is the water supply objective function value of reservoir 2, F refers to the objective function value, s refers to water supply, and "2" refers to reservoir 2; D t is the water demand during period t; S t is the water supply in period t; T is the total number of periods.

[0193] Ⅱ. The goal of minimizing reservoir water replenishment is reflected by the minimum annual reservoir water replenishment designed and added during model calculation.

[0194] 2) Decision variables

[0195] The water level of the water replenishment judgment line, the water level of the water supply limit line and the water replenishment volume of the user in each period are taken as decision variables.

[0196] 3) Constraints

[0197] Ⅰ. Water balance constraint: V t+1 =V t +R t -S t -Q t (4-12)

[0198] Where V t+1 is the reservoir capacity at the end of period t, m 3 , V t is the initial reservoir capacity in period t, m 3 ; R t is the amount of water entering the reservoir during period t, m 3 ; S t is the water supply in period t, m3; Q t is the reservoir water loss in period t, m 3 .

[0199] II. Reservoir capacity constraints: V tmin ≤V t ≤V tmax (4-13)

[0200] Where V tmin 、V tmax are the minimum and maximum water storage allowed in the reservoir during period t, m 3 .

[0201] III. Water supply constraint: S tmin ≤S t ≤S tmin(4-14)

[0202] In the formula, S tmin , S tmin are the minimum and maximum water supply of the reservoir during period t, m 3 .

[0203] IV. Water demand constraint: 0≤S t ≤D t (4-15)

[0204] Among them, S t is the water supply in the period, m 3 , D t is the water demand for the period, m 3 .

[0205] (3) Solving the optimization dispatching model based on a long series of historical runoff

[0206] Genetic algorithm is a theoretical result of bionics that applies biological principles to computer science. Due to its strong problem-solving ability, it has aroused the interest and participation of many scholars. It has been applied in 16 major fields and more than 250 minor fields, becoming one of the hot topics in academic interdisciplinary circles. In recent years, it has been widely used in the research of reservoir (group) optimization scheduling with good results. The water supply optimization scheduling model sets two objectives: water shortage rate and water replenishment. This study intends to use the NSGA-II algorithm for efficient optimization of dual-objective optimization problems to solve the model.

[0207] Based on a long series of historical monthly runoff data, according to the objective functions set in the aforementioned optimization models, based on the principle of genetic algorithm, the NSGA-II algorithm was improved and used, and the code was written to optimize and solve the decision variables to obtain the optimal scheduling strategy.

[0208] Typical annual water supply scheduling and annual water supply plan based on optimization

[0209] (1) Optimal water supply scheduling in a typical year

[0210] The typical year water supply optimization scheduling scheme is similar to the process of obtaining the reservoir water supply optimization scheduling. The construction of the typical year water supply model includes key elements such as objective function, decision variables, constraints and optimization algorithm, and then obtains the optimal operation strategy based on the input data. The reservoir optimization scheduling model is mainly a scheduling model with water supply as the goal. The objective function, decision variables and constraints of the typical year water supply optimization scheduling model are shown in Sections 1.1 and 1.2. The main difference from the aforementioned optimization scheduling model is the input data and the length of the calculation period.

[0211] This part first selects typical years based on long series runoff data through frequency calculation and P-Ⅲ curve matching. Typical years are selected based on five frequencies: 25%, 50%, 75%, 90%, and 95%.

[0212] Then, according to the runoff process of typical years with different frequencies, under certain constraints, the optimization algorithm is used to obtain the optimal operation strategy when the objective function reaches the extreme value, and the system analysis method is used to study the optimization problem of the system.

[0213] (2) Water supply plan for next year

[0214] The formulation of the water supply plan for the next year is mainly based on the long-term (annual) forecast runoff of the reservoir, and the typical year with the closest runoff is selected. This section provides the annual runoff corresponding to different frequencies of 25%, 50%, 75%, 90%, and 95%, and their corresponding monthly runoff processes. Users can select the typical year based on the forecast runoff for the next year.

[0215] According to the runoff of similar typical years and its annual distribution (i.e. the proportion of each month in the annual runoff), the long-term (annual) forecast runoff is proportionally distributed to 12 months as the monthly runoff process for the next year. The model reads the initial water level of the reservoir period, the water level-reservoir capacity curve, the special water level, and the water demand, and simulates and calculates the reservoir water supply process month by month according to the water supply scheduling diagram. Output the water level change process, water supply process, water replenishment process, water abandonment process for the next year, as well as the water supply guarantee rate and water shortage rate of the reservoir operation under the water supply plan.

[0216] Real-time water supply scheduling model: The real-time water supply scheduling model considers two modes: conventional water supply and optimized water supply. Conventional water supply provides a water balance plan. According to the water inflow and water demand, the amount of water replenishment required by the reservoir is calculated through the initial water level of the period and the water level at the end of the period that the reservoir needs to reach.

[0217] The optimized water supply is calculated based on the optimized water supply scheme obtained by the long series optimized water supply scheduling model. After selecting the water supply scheme according to the needs, the model directly calculates the reservoir water supply, water replenishment, water level at the end of the period, and water abandonment based on the initial water level of the period, the expected water inflow, and the water demand.

[0218] 3. Construction of intelligent scheduling system

[0219] 3.1 System Technical Architecture

[0220] The intelligent scheduling system is developed using B / S architecture, with Springboot as the backend, VUE as the frontend, and MySQL as the database. The overall architecture is as follows: Figure 6 As shown, Figure 6 For the overall technical architecture.

[0221] (1) Front-end technology selection

[0222] Table 2 Web front-end technology selection

[0223] type WEB front end Basic Framework VUE routing vue-router Permissions permission State Management vuex Code detection eslint Data Interaction axios UI Framework elementUI Data encryption Jsencrypt(RSA) Module Management webpack Visualization chart Echart GIS Applications NWHGIS / Openlayers / Cesium Message Push webSocket Packaging Tools webpack

[0224] (2) Backend technology selection

[0225] Table 3 Backend technology selection

[0226] type Springboot backend Basic Framework Springboot Safety Spring Security Oauth2.0 Interface Restful cache Redis Modular Management Maven Message Push MQTT, webSocket Token Verification JWT Database Operations Mybatis-plus Scheduled tasks Quartz / XXLJOB Data processing Kettle

[0227] (3) Database selection

[0228] Table 4 Database selection

[0229] type database Business Database Mysql Cache database Redis Temporary database Sqlite

[0230] 3.2 System Integration

[0231] (1) Conventional reservoir operation

[0232] The conventional operation of reservoirs adopts a semi-empirical and semi-theoretical method based on historical data. Using the basic theories of classical mathematics, hydrology, hydraulics and runoff regulation, the operation methods and rules are studied, operation diagrams are drawn, and operation regulations are compiled to guide reservoir operation. The conventional operation module of reservoirs integrates the reservoir operation plan and use conditions compiled by the reservoir management office.

[0233] (2) Reservoir Optimization Operation

[0234] Flood control calculations calculate the outflow process of the reservoir or the change in the reservoir water storage according to the reservoir dispatching rules. Common rules include controlled discharge dispatching, command dispatching and discharge curve dispatching. Controlled discharge dispatching is to manually set the future outflow process and conduct reservoir dispatching; command dispatching manually adjusts the gate opening height, automatically calculates the discharge flow at different gate opening heights, and performs reservoir flood control calculations based on the reservoir inflow; discharge curve dispatching is to perform dispatching calculations according to the reservoir discharge curve.

[0235] (3) Real-time reservoir dispatching

[0236] The real-time reservoir dispatching functions include: correction of inflow forecast results, dispatching model, dispatching results and dispatching instructions.

[0237] Correction of water inflow forecast results: Real-time access to flood or runoff water inflow forecast results. The forecast system corrects the forecast results in real time and can be corrected through manual intervention.

[0238] Scheduling model: Automatically select the scheduling plan or model based on the scheduling model configuration and water inflow forecast results, and calibrate the scheduling model at regular intervals. Manual intervention and correction are allowed.

[0239] Dispatching results: The dispatching results are calculated in real time based on the real-time water inflow forecast results and the real-time dispatching model or plan, and can be corrected by human intervention.

[0240] Scheduling instructions: The scheduling results generate scheduling instructions according to a fixed format. Figure 7 It is the interface of reservoir dispatching system.

[0241] (4) Joint Operation of Reservoirs

[0242] 1) Optimal Operation of Reservoir Groups

[0243] For reservoir groups used mainly for flood control, compensation is used for scheduling. Generally, the upstream reservoirs of the cascade reservoirs or the parallel reservoirs that are far away from the flood protection zone are compensated first, so that the reservoirs with a larger proportion of flood control, higher flood regulation capacity and closer to the downstream flood protection zone are the last to control the discharge.

[0244] The reservoir group optimization dispatching function includes reservoir group management, the construction and solution of reservoir group optimization dispatching model and the configuration of dispatching model.

[0245] 2) Real-time dispatching of reservoir groups

[0246] The reservoir group real-time dispatching unit comprises:

[0247] Establish a reservoir group joint dispatch model sub-unit, which is used to formulate a reservoir group joint dispatch mode and establish a reservoir group joint dispatch model based on the single reservoir dispatch, according to the geographical location of each reservoir, the topological relationship of the water system, the mutual influence of upstream and downstream, the proportion of flood control, the size of flood control storage capacity, the comprehensive utilization task, the interval combination and encounter of floods;

[0248] The joint dispatching scheme formulation subunit is used to carry out joint dispatching and optimization decision-making based on the water inflow forecast of each upstream and downstream reservoir and interval, and the joint dispatching model of the reservoir group, using the measures of early pre-discharge, compensation regulation, and peak shaving, to determine the dispatching indicators of the water storage index, flood discharge sequence, flood discharge time, and gate opening volume of each upstream and downstream reservoir, and formulate a joint dispatching scheme;

[0249] The sub-unit for issuing joint dispatching instructions is used to analyze whether the upstream and downstream reservoirs are safe, and whether the various sections and upstream and downstream rivers exceed the standards based on the joint dispatching plan; if unsafe or exceeding conditions occur, disaster warning information and disposal measures are generated in combination with the flood control plans of each reservoir and town, and joint dispatching instructions are issued. At the same time, changes in the water conditions of each reservoir gate and river channel are collected for comparison, and plans and warning decisions are revised in real time.

[0250] Figure 8It is a technical roadmap for joint dispatching of reservoir groups. For reservoir groups and interval combined basins with dispatching capabilities, the comprehensive peak-shaving capacity of the reservoir group is mainly considered to achieve joint flood control dispatching. Under the premise of ensuring the safety of the reservoir group, intervals, and upstream and downstream, the comprehensive water storage capacity of the reservoir group is maximized to achieve an organic combination of flood control and benefit development.

[0251] First, based on the single reservoir scheduling, a joint scheduling method for reservoir groups is formulated and a joint scheduling model for reservoir groups is established according to the geographical location of each reservoir, the topological relationship of the water system, the mutual influence between upstream and downstream, the proportion of flood control, the size of flood control storage capacity, the comprehensive utilization task, the interval combination and encounter conditions of floods.

[0252] Secondly, according to the water inflow forecasts of upstream and downstream reservoirs and sections, and based on the joint scheduling model of the reservoir group, we will make full use of measures such as early discharge, compensatory regulation, peak shaving and peak shifting to carry out joint scheduling and optimize decision-making, determine the water storage indicators, flood discharge sequence, flood discharge time, gate opening volume and other scheduling indicators of the upstream and downstream reservoirs, and formulate a joint scheduling plan.

[0253] Finally, the joint dispatching plan is used to analyze whether the upstream and downstream reservoirs are safe, and whether the various sections and upstream and downstream rivers exceed the standards. If unsafe or exceeding standards occur, disaster warning information and disposal measures are generated in combination with the flood control plans of each reservoir and town. The authorized person in charge reviews and issues joint dispatching instructions, and at the same time collects and compares the changes in the water conditions of each reservoir gate and river channel, and corrects the plan and warning decisions in real time.

[0254] The real-time dispatching functions of a reservoir group include: water inflow forecast results, dispatching models, dispatching results, and dispatching instructions.

[0255] (5) Scheduling plan management

[0256] The scheduling scheme management module includes:

[0257] A scheduling plan query unit, used to query the scheduling plan details by the plan name or the time when the scheduling plan was generated, and export the scheduling plan;

[0258] A scheduling scheme management unit, used to start implementing a scheduling scheme or to view and delete a scheduling scheme by scheme name or time when the scheduling scheme was generated;

[0259] The scheduling scheme comparison unit is used to compare and analyze different scheduling schemes for a group of reservoirs or a single reservoir in the basin, and compare and analyze their highest water levels and their occurrence time, maximum inflows and their occurrence time, maximum outflows and their occurrence time, and inflow process lines.

[0260] Fig. 9 This is the interface diagram of the scheduling plan management. The scheduling management module includes functions such as scheduling plan query, scheduling plan management, and scheduling plan comparison.

[0261] 1) Scheduling plan query: query the scheduling plan details by the plan name or the time when the scheduling plan was generated, and export the scheduling plan.

[0262] 2) Scheduling plan management: start implementing a scheduling plan or view and delete a scheduling plan by the plan name or the time the scheduling plan was generated.

[0263] 3) Comparison of dispatching schemes: Compare and analyze different dispatching schemes for reservoir groups or single reservoirs in the basin, including the highest water level and its occurrence time, maximum inflow and its occurrence time, maximum outflow and its occurrence time, inflow process line, etc.

[0264] Compared with the prior art, the present invention has the following beneficial effects:

[0265] (1) By adopting distributed hydrological forecasting, high-resolution remote sensing, machine learning and other technical means, a reservoir inflow forecast model was constructed that takes into account the impact of human activities and the non-uniformity of the underlying surface. This effectively solved the problems of inaccurate forecasting and poor timeliness of conventional means, thereby significantly improving the accuracy and efficiency of reservoir inflow forecasting.

[0266] (2) By combining information and digital technologies such as the Internet of Things, cloud computing, big data, BIM, artificial intelligence, and digital twins with traditional reservoir scheduling, an intelligent scheduling system has been developed. The calculation principle uses trial algorithms to replace the semi-graphic method of traditional hydrological calculations. Therefore, the computing power of computers is utilized to effectively solve the problems of low digitalization level and insufficient computing level in reservoir scheduling, and comprehensively promote the construction of computing data, algorithms, and computing power to safeguard reservoir flood control safety.

[0267] The second object of the present invention is to provide a reservoir flood control dispatching method, based on the above system, the method comprises the following steps:

[0268] Use reservoir water forecasting models to forecast flood or runoff water;

[0269] Carry out flood control operations based on water balance and dynamic balance;

[0270] Real-time access to flood or runoff water forecast results, and real-time correction of forecast results;

[0271] Select the dispatching scheme or model according to the dispatching model configuration and water inflow forecast results, and calibrate the dispatching model at regular intervals;

[0272] Calculate the dispatch results in real time based on the real-time water inflow forecast results and the real-time dispatch model or plan;

[0273] The scheduling results generate scheduling instructions according to a fixed format.

[0274] The third object of an embodiment of the present invention is to provide an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the reservoir flood control scheduling method when executing the computer program.

[0275] A fourth objective of an embodiment of the present invention is to provide a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the reservoir flood control scheduling method is implemented.

[0276] A fifth objective of an embodiment of the present invention is to provide a computer program product, wherein the computer program product comprises computer instructions, and the computer instructions instruct a computer to execute the reservoir flood control scheduling method.

[0277] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0278] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0279] The present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program codes.

Claims

1. A reservoir flood control dispatching system, characterized in that: include: The water inflow forecast module is used to forecast flood or runoff water using a reservoir water inflow forecast model; Flood control dispatch model, used to carry out flood control dispatch based on water balance and dynamic balance; Reservoir real-time dispatching module, which includes: The water inflow forecast result correction unit is used to access the flood or runoff water inflow forecast results in real time and correct the forecast results in real time; The dispatch model unit is used to select a dispatch scheme or model according to the dispatch model configuration and the water inflow forecast results, and to calibrate the dispatch model at regular intervals; A dispatch result unit, used to calculate the dispatch result in real time according to the real-time water inflow forecast results and the real-time dispatch model or plan; The scheduling instruction unit is used to generate scheduling instructions according to a fixed format based on the scheduling results.

2. A reservoir flood control dispatching system according to claim 1, characterized in that: Also includes: Reservoir group joint dispatching module, which includes: The reservoir group optimization dispatching unit is used for flood control-oriented reservoir groups, which adopt compensation mode dispatching, with the upstream reservoir of the cascade reservoir or the parallel reservoir far from the flood protection zone being compensated first, so that the reservoir with a larger proportion of flood control, higher flood regulation capacity and closer to the downstream flood protection zone will control the discharge volume last; The real-time dispatching unit of the reservoir group is used to realize joint flood control dispatching for the reservoir group and interval combined basin with dispatching capacity, taking into account the comprehensive peak-shaving capacity of the reservoir group, and ensuring the maximization of the comprehensive water storage capacity of the reservoir group under the premise of ensuring the safety of the reservoir group, interval and upstream and downstream.

3. A reservoir flood control dispatching system according to claim 2, characterized in that: The reservoir group real-time dispatching unit comprises: Establish a reservoir group joint dispatch model sub-unit, which is used to formulate a reservoir group joint dispatch mode and establish a reservoir group joint dispatch model based on the single reservoir dispatch, according to the geographical location of each reservoir, the topological relationship of the water system, the mutual influence of upstream and downstream, the proportion of flood control, the size of flood control storage capacity, the comprehensive utilization task, the interval combination and encounter of floods; The joint dispatching scheme formulation subunit is used to carry out joint dispatching and optimization decision-making based on the water inflow forecast of each upstream and downstream reservoir and interval, and the joint dispatching model of the reservoir group, using the measures of early pre-discharge, compensation regulation, and peak shaving, to determine the dispatching indicators of the water storage index, flood discharge sequence, flood discharge time, and gate opening volume of each upstream and downstream reservoir, and formulate a joint dispatching scheme; The sub-unit for issuing joint dispatching instructions is used to analyze whether the upstream and downstream reservoirs are safe, and whether the various sections and upstream and downstream rivers exceed the standards based on the joint dispatching plan; if unsafe or exceeding conditions occur, disaster warning information and disposal measures are generated in combination with the flood control plans of each reservoir and town, and joint dispatching instructions are issued. At the same time, changes in the water conditions of each reservoir gate and river channel are collected for comparison, and plans and warning decisions are revised in real time.

4. A reservoir flood control dispatching system according to claim 1, characterized in that: Also includes: Reservoir routine dispatch module, used to guide reservoir operation based on historical data; The reservoir optimization scheduling module is used to calculate the reservoir outflow process or reservoir water storage change according to the reservoir scheduling rules.

5. A reservoir flood control dispatching system according to claim 1, characterized in that: Also includes: The scheduling scheme management module includes: A scheduling plan query unit, used to query the scheduling plan details by the plan name or the time when the scheduling plan was generated, and export the scheduling plan; A scheduling scheme management unit, used to start implementing a scheduling scheme or to view and delete a scheduling scheme by scheme name or time when the scheduling scheme was generated; The scheduling scheme comparison unit is used to compare and analyze different scheduling schemes for a group of reservoirs or a single reservoir in the basin, and compare and analyze their highest water levels and their occurrence time, maximum inflows and their occurrence time, maximum outflows and their occurrence time, and inflow process lines.

6. A reservoir flood control dispatching system according to claim 1, characterized in that: The reservoir inflow forecast model adopts the Sanshuiyuan Xin'anjiang model and the API model to forecast the reservoir inflow flood, and adopts the VMD-APSO-LSTM model to forecast the medium-term runoff; The Sanshuiyuan Xin'anjiang model adopts the gradient descent method to determine the model parameter calibration; The VMD-APSO-LSTM model is used to predict runoff, specifically including: determining the penalty factor α of runoff data at each scale by a parameter adjustment method, determining different decomposition levels K by a center frequency method, inputting the determined parameters into VMD to obtain multiple sub-modes; each sub-mode is input into the APSO-LSTM model as an independent sequence, setting the data before decomposition as input and the corresponding data of the next year as output for training, and using the model for forecasting after the training is completed; and accumulating the sub-sequences obtained by each forecast accordingly to obtain the final forecast result.

7. A reservoir flood control dispatching system according to claim 1, characterized in that: The flood control dispatch model selects a typical flood process from measured data, and then amplifies it according to the designed flood peak flow and designed flood volume to obtain the designed flood process line; The model has built-in multiple flood processes. According to the relationship between the water level and storage capacity of the reservoir, the water level-storage capacity relationship curve is fitted, and each calculation point in the relationship line is obtained by linear interpolation. The calculation structure is constructed together with the reservoir water balance formula and the Saint-Venant equation group. The dynamic equation used is fitted and calculated using the storage-discharge relationship curve of the reservoir area, and the cyclic trial algorithm is used to calculate the discharge flow.

8. A reservoir flood control dispatching system according to claim 1, characterized in that: The real-time flood control dispatching process of the flood control dispatching model includes: Extract and sort the peak flow of the design flood process line under each return period, and use it as the basis for classifying the flood level entering the reservoir; classify the flood into levels, read the peak flow data in the flood forecast data entering the reservoir, classify the predicted flood according to the data, and obtain the flood situation estimation level; According to the forecast information of the forecast period, the operation strategy of the forecast period is obtained, and then the decision of the facing period is selected from the operation strategy for implementation; Before implementing the optimization strategy for the next forecast period of the water diversion project, the water use forecast information for the next period is corrected using correlation analysis and other techniques based on the actual water use information of the previous periods, and the actual state of the project at the end of the previous period is used to replace the reservoir water storage calculated in the previous time, as the initial value of the actual state of the project for the next period. Then, flood dispatching for the next forecast period is continued, and the corresponding operation strategy is obtained, and the decision of the first period is selected for implementation. In this way, the forecast information and project status of the forecast period are continuously updated and corrected in sequence and time series, and real-time operation decisions are made for each period.

9. A reservoir flood control dispatching system according to claim 1, characterized in that: The modification process of the flood control dispatch model includes: According to the conclusions drawn from the flood situation forecast, floods are divided into two categories: minor floods and super-standard floods. For minor floods, water storage is given priority while ensuring that the reservoir water level does not exceed the flood limit water level. The discharge flow is calculated based on the real-time forecast inflow data. When the situation is predicted to be an extraordinary flood, the dispatching plan corresponding to the matching design flood process line is corrected in real time in combination with the real-time inflow forecast data, and the corresponding downstream flow data is output hourly.

10. A reservoir flood control dispatching method, characterized in that: include: Use reservoir water forecasting models to forecast flood or runoff water; Carry out flood control operations based on water balance and dynamic balance; Real-time access to flood or runoff water forecast results, and real-time correction of forecast results; Select the dispatching scheme or model according to the dispatching model configuration and water inflow forecast results, and calibrate the dispatching model at regular intervals; Calculate the dispatch results in real time based on the real-time water inflow forecast results and the real-time dispatch model or plan; The scheduling results generate scheduling instructions according to a fixed format.

Images