Reservoir scheduling diagram optimization method, device, storage medium and equipment

Abstract

The present invention relates to the technical field of reservoir scheduling, and specifically to a method, device, storage medium and equipment for optimizing a reservoir scheduling diagram. In the present invention, the core research problem is the optimal control application mode of a reservoir based on a scheduling diagram (optimizing the scheduling diagram considering the flood and drought factors). On the basis of a conventional scheduling diagram, in pursuit of the maximum power generation efficiency, a differential evolution algorithm with round-line iteration is used to optimize the scheduling line in the scheduling diagram; at the same time, on the basis of the optimized scheduling diagram of a hydropower station, the runoff difference is further considered, the flood and drought factors of the incoming water are introduced, and the optimized scheduling lines for different years are determined. In addition, during scheduling, the influence of the forecast accuracy error can be corrected by the rolling forecast scheduling method, the difference between the calculation result of the optimized scheduling and the actual scheduling is reduced, and the reliability of the scheduling is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of reservoir scheduling, and in particular to a reservoir scheduling diagram optimization method, device, storage medium and equipment. Background Art

[0002] Reservoir operation involves the planned storage and release of water from the reservoir's natural inflow, based on the reservoir's established water conservancy tasks and operation principles, while ensuring dam safety. This is done to achieve the goal of promoting benefits and eliminating harm, and achieving comprehensive water resource utilization. Scientific reservoir operation can ensure reservoir safety and improve the efficiency of comprehensive water resource utilization with virtually no additional investment. Therefore, strengthening reservoir operation research is of paramount importance.

[0003] Reservoir operation can be categorized into conventional operation and optimized operation based on the scheduling method. Conventional operation refers to the operation method that guides actual reservoir operation based on a conventional operation chart. Conventional operation charts are drawn based on the reservoir's historical runoff data and the power station's position within the system. Due to their simplicity, ease of operation, and high reliability, they remain the most commonly used tool for guiding actual reservoir operation. Conventional operation charts also play an irreplaceable role in determining hydropower station parameters. Reservoir optimized operation uses an optimization algorithm to solve an optimized operation model based on the reservoir's water inflow process and the optimal reservoir operation criteria, generating the reservoir's optimal operation strategy. When reservoir water inflow is completely known, optimized operation can increase reservoir power generation compared to conventional power generation operation. However, due to the limitations of medium- and long-term runoff forecast accuracy, the predicted future inflow process of the reservoir cannot accurately reflect future water inflow conditions, resulting in significant discrepancies between the calculated results of optimized operation and actual operation. Furthermore, the complexity of reservoir optimized operation systems limits the scope of their application. Therefore, in-depth research on hydropower station operation charts remains of great practical significance for guiding reservoir operation.

[0004] The development of dispatch charts is a dynamic process of gradual improvement. Traditional dispatch charts are two-dimensional diagrams drawn based on historical runoff statistics and the position and role of hydropower stations in the system, placing a high emphasis on their system reliability. As the number of hydropower stations within the power system continues to expand, the system's reliability requirements for individual stations have decreased. However, for reservoirs with good regulation performance, traditional dispatch charts tend to be conservative in guiding actual reservoir operations, leading to a sharp conflict between water storage and water discharge at hydropower stations, resulting in low power generation efficiency. To fully leverage the regulatory role of reservoirs, increase the overall benefits of hydropower stations, and improve water resource utilization efficiency, it is particularly necessary to optimize traditional dispatch charts. Summary of the Invention

[0005] In view of this, the present invention provides a reservoir scheduling diagram optimization method, device, storage medium and equipment to solve the problem of how to achieve reservoir scheduling optimization.

[0006] In a first aspect, the present invention provides a method for optimizing a reservoir scheduling diagram, the method comprising: obtaining a long series of measured runoff data; based on the start time and end time of the water conservancy year, shifting the start time backward in sequence according to a preset period, and grouping the water conservancy years according to the shifted start time and end time; determining the corresponding year type according to the runoff data in each water conservancy year group, the year type is used to characterize the abundance and scarcity of runoff; based on a preset objective function, optimizing each scheduling line in a conventional scheduling diagram according to the year type using a differential evolution algorithm with round line iteration to obtain optimized scheduling lines of different year types, the preset objective function being constructed with the goal of maximizing the multi-year average power generation of the hydropower station; determining the year type according to the runoff data predicted according to the water conservancy year grouping for the water conservancy year to be scheduled, and performing scheduling according to the scheduling line of the corresponding year type.

[0007] This invention focuses on the optimal control of reservoirs based on dispatch charts (optimizing dispatch charts by considering both wet and dry seasons). Based on conventional dispatch charts, and striving for maximum power generation efficiency, a differential evolution algorithm with a round-robin iteration approach is employed to optimize the dispatch lines within the dispatch charts. Furthermore, based on the optimized dispatch charts for hydropower stations, this approach further considers runoff variability, incorporating wet and dry seasons to determine optimized dispatch lines for different years. Furthermore, a rolling forecast scheduling approach is used during dispatch to mitigate the impact of forecast accuracy errors, reduce the discrepancy between the calculated optimized dispatch results and the actual dispatch, and improve dispatch reliability.

[0008] In an optional embodiment, the corresponding year type is determined based on the runoff data in each water conservancy year group, including: extracting characteristic indicators of the runoff data; standardizing, principal component analysis and clustering the characteristic indicators to obtain clustering results; and determining the corresponding year type based on the clustering results.

[0009] In the present invention, since the impact of runoff on reservoir operation is not only reflected in the size of the runoff, but the runoff process also has a significant impact on reservoir operation, it is biased to use annual runoff as the only classification criterion when grouping runoff. Therefore, multiple indicators describing runoff characteristics are selected for wet and dry season identification, which improves the accuracy of the identification results. Principal component analysis of the extracted indicators can cover most of the information in the original indicators, while simplifying the analysis process and increasing the accuracy of the results. Finally, the annual type division is achieved through clustering.

[0010] In an optional embodiment, based on a preset objective function, a differential evolution algorithm with round-line iteration is used to optimize each scheduling line in the conventional scheduling diagram according to the year type to obtain optimized scheduling lines of different year types, including: obtaining multiple scheduling lines to be optimized in the conventional scheduling diagram, the multiple scheduling lines include limited output lines, anti-destruction lines and increased output lines; selecting the scheduling lines in sequence according to a preset order, and based on the preset objective function, using the differential evolution algorithm to optimize the limited output lines and the increased output lines; based on the runoff data of different years, using the differential evolution algorithm to optimize the anti-destruction lines according to the year type, and the optimized scheduling lines do not cross.

[0011] In this invention, to prevent large amounts of wasted water from being discarded and improve hydropower station efficiency, the plant should increase its output earlier in years with more abundant water. Therefore, the damage prevention line for years with abundant water should be lower than that for years with less abundant water. Based on this, when optimizing the damage line, the runoff factor is taken into account to determine the damage prevention line for different years.

[0012] In an optional embodiment, based on a preset objective function, a differential evolution algorithm is used to optimize the restricted output line and the increased output line, including: uniformly generating an initial scheduling line population based on the range values ​​formed by the upper and lower adjacent scheduling lines of the scheduling line to be optimized; performing mutation, crossover and selection operations on the initial scheduling line population to obtain the individual with the largest fitness value, and the fitness value is calculated based on the objective function; repeating the above mutation, crossover and selection operations until the preset requirements are met to obtain the optimized scheduling line.

[0013] In an optional embodiment, a differential evolution algorithm is used to optimize the anti-destruction line according to the year type, including: generating an initial anti-destruction line population based on the range values ​​formed by the upper and lower adjacent scheduling lines of the anti-destruction line of the year type to be optimized; performing mutation and crossover operations on the initial scheduling line population to obtain a new anti-destruction line population; extracting runoff data of the corresponding year type, using a preset objective function to calculate the fitness value corresponding to each anti-destruction line individual in the new anti-destruction line population, and selecting the individual with the largest fitness value; repeating the above mutation, crossover and selection operations until the preset requirements are met, and obtaining the optimized anti-destruction line of the corresponding year type.

[0014] In an optional embodiment, the preset objective function is constructed in the following manner: constructing constraint conditions, which include guarantee rate constraints, water balance constraints, scheduling line constraints and bending energy constraints, and the bending energy constraints are determined based on the slopes of each point on the scheduling line; based on the constraint conditions, the preset objective function is constructed with the goal of maximizing the average multi-year power generation of the hydropower station.

[0015] In the present invention, by ensuring rate constraints, water balance constraints, scheduling line constraints and other constraints, the optimized scheduling line is made more practical; at the same time, considering the bending energy, it is possible to avoid the optimized scheduling line having very obvious jagged edges.

[0016] In an optional embodiment, the runoff data grouped according to the water conservancy year to be scheduled is used to determine the year type to which it belongs, and scheduling is performed according to the scheduling line of the corresponding year type, including: predicting the runoff forecast data from the current time to the end of the water conservancy year based on the passage of time; determining the year type based on the distance between the runoff forecast data and the cluster center; and selecting the scheduling line of the corresponding year type for scheduling according to the year type.

[0017] In the second aspect, the present invention provides a reservoir scheduling diagram optimization device, the device including: a data acquisition module for acquiring a long series of measured runoff data; a grouping module for, based on the start time and end time of the water conservancy year, shifting the start time backward in sequence according to a preset period, and grouping the water conservancy year according to the shifted start time and end time; a year type division module for determining the corresponding year type based on the runoff data in each water conservancy year group, the year type being used to characterize the abundance and scarcity of runoff; an optimization module for optimizing each scheduling line in the conventional scheduling diagram according to the year type based on a preset objective function and adopting a differential evolution algorithm with wheel line iteration to obtain optimized scheduling lines of different year types, and the preset objective function is constructed with the goal of maximizing the multi-year average power generation of the hydropower station; a scheduling module for determining the year type based on the runoff data predicted according to the water conservancy year grouping for the water conservancy year to be scheduled, and performing scheduling according to the scheduling line of the corresponding year type.

[0018] In a third aspect, the present invention provides a computer device comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the reservoir scheduling diagram optimization method of the first aspect or any corresponding embodiment thereof by executing the computer instructions.

[0019] In a fourth aspect, the present invention provides a computer-readable storage medium having computer instructions stored thereon, the computer instructions being used to enable a computer to execute the reservoir scheduling diagram optimization method of the first aspect or any corresponding embodiment thereof.

[0020] In a fifth aspect, the present invention provides a computer program product comprising computer instructions for enabling a computer to execute the reservoir scheduling diagram optimization method of the first aspect or any corresponding embodiment thereof. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0022] Figure 1 is a flow chart of a reservoir scheduling diagram optimization method according to an embodiment of the present invention;

[0023] Figure 2 1 is a flow chart of a differential evolution algorithm based on round-line iteration according to an embodiment of the present invention;

[0024] Figure 3 1 is a flow chart of a differential evolution algorithm for wheel line iteration considering abundance and scarcity factors according to an embodiment of the present invention;

[0025] Figure 4 is a schematic diagram of an optimized scheduling diagram according to an embodiment of the present invention;

[0026] Figure 5 is a structural block diagram of a reservoir scheduling diagram optimization device according to an embodiment of the present invention;

[0027] Figure 6 Schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0028] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative efforts shall fall within the scope of protection of the present invention.

[0029] According to an embodiment of the present invention, an embodiment of a reservoir scheduling diagram optimization method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0030] In this embodiment, a reservoir scheduling diagram optimization method is provided, which can be used in electronic devices such as computers, mobile phones, tablet computers, etc. Figure 1 : is a flow chart of a reservoir scheduling diagram optimization method according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0031] Step S101: Acquire a long series of measured runoff data. Specifically, this data represents a series of runoff data obtained through long-term field observation, recording, or extrapolation, i.e., it includes multiple years of measured runoff data. Runoff data specifically includes flow rate data, which represents the amount of water passing through a specific section per unit time. It reflects the intensity of river flow at different times. Runoff data can be acquired through methods such as relevant hydrological surveys, and this embodiment does not specifically limit this acquisition method.

[0032] Step S102: Based on the start and end times of the water conservancy year, the start time is sequentially shifted backward according to a preset period, and the water conservancy years are grouped according to the shifted start and end times. Specifically, in water conservancy calculations, a water conservancy year is defined by the reservoir's storage and discharge cycle as the starting and ending points of the year. Typically, the year begins when the reservoir begins filling and ends when the reservoir is emptying. That is, the start and end times of a water conservancy year are typically determined based on local hydrological and meteorological characteristics and water resource patterns. For example, a water conservancy year may start in March and end in February of the following year; or it may start in October and end in September of the following year, and so on.

[0033] Among them, after determining the water conservancy year corresponding to the reservoir, this embodiment further groups the water conservancy year to more accurately determine the wet and dry year type of the water conservancy year. When grouping, the start time is shifted backward in sequence according to the preset period, and the water conservancy year is grouped according to the shifted start time and end time. For example, if the water conservancy year starts in March and ends in February of the following year, and the preset period is months, then March to February of the following year are first divided into a group, and then the start time is shifted by one month, and April to February of the following year are divided into a group; the start time is shifted by another month, and May and February of the following year are divided into one group, and so on, until February of the following year is divided into one group, thereby dividing the water conservancy year into 12 groups.

[0034] In practical applications, the preset period can be determined according to actual conditions. For example, the preset period can be ten days, a quarter, or other predefined time periods.

[0035] Step S103 determines the corresponding year type based on the runoff data within each water conservancy year grouping. The year type is used to characterize the abundance and scarcity of runoff. Specifically, after grouping the water conservancy years according to the above steps, the runoff data within each water conservancy year grouping is extracted using the acquired long series of measured runoff data, and the year type is divided based on the extracted runoff data. For example, if the acquired runoff data is from March 1950 to February 2020, the runoff data for each water conservancy year is first extracted, resulting in a total of 69 water conservancy years of runoff data. Then, according to the above grouping method, the runoff data of each water conservancy year are extracted. For example, for the water conservancy year from March 1950 to February 1951, the runoff data from March 1950 to February 1951 are extracted first, then the runoff data from April 1950 to February 1951, then the runoff data from May 1950 to February 1951, and so on, until the runoff data of February 1951 are extracted. Thus, 12 groups of runoff data are obtained in this water conservancy year.

[0036] For the runoff data within each water conservancy year group extracted, the year type corresponding to each group of runoff data can be determined by analyzing the runoff data, that is, the abundance and scarcity of water in the runoff in different periods can be determined. According to actual conditions, the year type can be divided into a flood season and a dry season, and a normal water season can be further added, or a more detailed division can be made. For example, in the present embodiment, the year type is divided into seven year types: an extremely abundant year group, a abundant water year group, a normal to slightly abundant year group, a normal water year group, a normal to slightly dry year group, a dry year group, and an extremely dry year group, and the water volume represented by the seven year types gradually decreases from abundant to scarce.

[0037] In step S104, based on the preset objective function, the differential evolution algorithm with round-line iteration is used to optimize each scheduling line in the conventional scheduling diagram according to the annual type to obtain the optimized scheduling lines of different annual types. The preset objective function is constructed with the goal of maximizing the multi-year average power generation of the hydropower station.

[0038] Specifically, the conventional dispatch diagram can be a dispatch diagram drawn using traditional methods. This diagram is based on flood control safety, takes the economic efficiency of the reservoir as the primary dispatching objective, and takes into account the reliability of the power supply of the power system. In other words, the conventional dispatch diagram can be a dispatch diagram for the reservoir to be optimized that has been obtained from related technologies. This embodiment performs optimization based on this conventional dispatch diagram.

[0039] The optimization process uses a differential evolution algorithm with round-robin iterations. This algorithm first uses round-robin iterations to sequentially optimize each dispatch line in the regular dispatch graph. Each dispatch line is optimized using a differential evolution algorithm. The differential evolution algorithm (DE) is a heuristic intelligent algorithm based on real number coding. It was proposed by Rainer Storn and Kenneth Price to solve Chebyshev polynomials. It is a randomized parallel search algorithm. The overall structure of DE is similar to that of a genetic algorithm, with a randomly generated initial population and new individuals generated through mutation, crossover, and selection. However, there are some differences between DE and genetic algorithms, primarily in the mutation operation. Genetic algorithms achieve mutation through chromosome mutation, while DE uses perturbations based on vector differences between individuals.

[0040] During the optimization process for each dispatch line, when a new individual is determined through a selection operation, this embodiment implements this process by calculating the individual's fitness value using a pre-constructed objective function. This preset objective function aims to maximize the average multi-year power generation efficiency of the hydropower station. Without considering factors such as losses, electricity sales costs, and time-of-use electricity prices, maximizing the power generation efficiency of the hydropower station is equivalent to maximizing the average multi-year power generation of the hydropower station. Therefore, maximizing the average multi-year power generation of the hydropower station can be understood as maximizing the power generation efficiency of the hydropower station. In addition, when optimizing the dispatch line, this embodiment further considers the year type to form dispatch lines corresponding to different year types, so that the dispatch line corresponding to the year type corresponding to the water conservancy year to be dispatched can be dispatched subsequently.

[0041] Step S105 determines the year type based on the runoff data forecasted by the water conservancy year grouping for the water conservancy year to be scheduled, and performs scheduling according to the scheduling line corresponding to the year type. Specifically, in this embodiment, to reduce the impact of forecast errors, corresponding runoff forecast data is obtained according to the water conservancy year grouping method described above. For example, if it is currently March, runoff forecast data from March to February of the following year is obtained. After determining the year type, the corresponding scheduling line is selected for scheduling. After April, runoff forecast data from April to February of the following year is re-obtained; after May, runoff forecast data from May to February of the following year is re-obtained. This process is repeated, with continuous rolling forecasting and scheduling to correct the impact of forecast errors.

[0042] The reservoir scheduling diagram optimization method provided by the embodiments of the present invention focuses on the optimal control application of reservoirs based on scheduling diagrams (optimizing scheduling diagrams by considering flood and drought factors). Based on conventional scheduling diagrams, and striving for maximum power generation efficiency, a differential evolution algorithm with round-robin iteration is used to optimize the scheduling lines within the scheduling diagram. Furthermore, based on the optimized hydropower station scheduling diagram, it further considers runoff variability, introduces flood and drought factors, and determines optimized scheduling lines for different years. Furthermore, during scheduling, a rolling forecast scheduling method is used to correct for the impact of forecast accuracy errors, reduce the discrepancy between the calculated optimized scheduling results and the actual scheduling, and improve scheduling reliability.

[0043] In this embodiment, a method for optimizing a reservoir scheduling diagram is provided, and the process includes the following steps:

[0044] Step S201: Obtain a long series of measured runoff data. Figure 1 Step S101 of the illustrated embodiment will not be described in detail here.

[0045] Step S202: Based on the start time and end time of the water conservancy year, the start time is shifted backwards in accordance with a preset period, and the water conservancy year is grouped according to the shifted start time and end time. Figure 1 Step S102 of the illustrated embodiment will not be described in detail here.

[0046] Step S203 , determining the corresponding year type according to the runoff data in each water conservancy year group, where the year type is used to characterize the abundance and scarcity of runoff.

[0047] Specifically, the above step S203 includes:

[0048] Step S2031, extracting characteristic indicators of runoff data; specifically, the impact of runoff on reservoir operation is not only reflected in the size of the runoff, but the process of runoff also has a great impact on reservoir operation. Therefore, when determining the abundance and scarcity of runoff, it is biased to use only the annual runoff as the only dividing standard. Based on this, when dividing the abundance and scarcity, the present embodiment extracts a plurality of indicators describing the runoff characteristics from the runoff data, and specifically, the corresponding indicators can be obtained by performing relevant calculations or processing on the runoff data. For example, specific indicators may include the annual runoff deviation percentage, the standardized runoff index, the flow duration curve, the change in the runoff coefficient, etc. In practical applications, the extracted indicators can be adjusted. The present embodiment does not limit the specific extracted indicators.

[0049] If the length of the long series of measured runoff data is m years, the m annual runoff samples are , considering the differences in the total amount of runoff and the runoff process in each year, n characteristic indicators are selected for each sample to fully reflect the characteristics of the runoff in each year, that is, m annual runoff samples, each sample has n characteristic indicators, forming a The data matrix , specifically expressed as:

[0050] .

[0051] Step S2032: standardize, perform principal component analysis, and perform clustering processing on the characteristic indicators to obtain clustering results.

[0052] Specifically, due to the incommensurability of the units and magnitudes of the various indicators, the results of the various indicators need to be standardized. This embodiment adopts the standard deviation standardization method, and the calculation formula is as follows:

[0053]

[0054] Where: , Indicators for each year expected value; Indicators for each year The mean square error of .

[0055] The standardized sample sequence constitutes a The data matrix , which can be expressed as:

[0056] .

[0057] In addition, when multiple characteristic indicators are selected, they will affect each other when there is a correlation between them. Principal component analysis can cover most of the information in the original indicators, while simplifying the analysis process and increasing the accuracy of the results. The main principles and steps of principal component analysis include:

[0058] (1) Calculate the covariance matrix.

[0059] The standardized data matrix is ​​calculated using the following formula: Building the covariance matrix ;

[0060]

[0061] (2) Calculate the eigenvalues ​​and eigenvectors of the covariance matrix.

[0062] Compute the eigenvalues ​​of the covariance matrix ,in ;

[0063] The corresponding eigenvector ,in .

[0064] (3) Calculate the contribution rate of each component and the cumulative contribution rate.

[0065] The contribution rate of each indicator is calculated using the following formula :

[0066]

[0067] The cumulative contribution rate is calculated using the following formula: :

[0068]

[0069] (4) Extract the principal components.

[0070] Set the cumulative contribution rate threshold , according to the cumulative contribution rate threshold Select the principal components and extract the corresponding values ​​of each principal component. , then the number of principal components is , extract the corresponding The values ​​of the principal components are generated Data Matrix , so far the principal component analysis has been completed.

[0071] In order to determine the corresponding year type using the characteristic indicators determined by principal component analysis, that is, to determine which year type category the characteristic indicators should be classified into, this embodiment uses clustering to divide them and determine the corresponding year type. Among clustering algorithms, K-means clustering has the characteristics of fast convergence speed and high accuracy. Therefore, this embodiment performs K-means clustering on the extracted principal components. The main process of K-means clustering is as follows:

[0072] (1) Input after principal component analysis Data Matrix , determine the number of groups .

[0073] (2) The selection methods of the initial center position include random selection, uniform selection based on the sample distribution range, and construction of an initial centroid position set. This embodiment uses random selection to determine Initial center position The number of k is determined by the year type categories that need to be divided. For example, if 7 year types need to be divided, then k=7.

[0074] (3) Calculate each sample and the corresponding cluster center This embodiment uses the Manhattan distance method to calculate distance. Manhattan distance is derived from the shortest driving path between cities planned as square building blocks, also known as city block distance. Manhattan distance is the sum of the distances projected onto the coordinate axes by the line segment formed by two sample points in a fixed rectangular coordinate system in Euclidean space. It depends on the rotation of the coordinate system and is not a simple translation or mapping of the system on the coordinate axes.

[0075] Each sample and the corresponding cluster center The distance is accumulated and used as an indicator to evaluate the clustering effect ,in:

[0076]

[0077] Where: express The set of all samples in a group.

[0078] (4) The minimum principle is to (re)assign each sample to the closest cluster.

[0079] (5) Update the average value of each cluster and use it as the new cluster center of each type.

[0080] (6) Continue iterating until Until no more changes occur.

[0081] Since the cluster centers are randomly selected, multiple calculations can be performed to reduce the randomness of the results, and the average of each result can be taken as the final classification result.

[0082] Step S2033 determines the corresponding year type based on the clustering results. Specifically, after the clustering results are determined, the corresponding year type is also determined. It should be noted that for runoff data within the water conservancy year grouping, the corresponding year type is determined using the process from steps S2031 to S2033 described above.

[0083] In step S204, based on the preset objective function, the differential evolution algorithm with round-line iteration is used to optimize each scheduling line in the conventional scheduling diagram according to the year type to obtain the optimized scheduling lines of different year types. The preset objective function is constructed with the goal of maximizing the multi-year average power generation of the hydropower station.

[0084] Among them, the preset objective function is constructed in the following way: constructing constraint conditions, which include guarantee rate constraint, water balance constraint, scheduling line constraint and bending energy constraint. The bending energy constraint is determined based on the slope of each point on the scheduling line; based on the constraint conditions, the preset objective function is constructed with the maximum average power generation of the hydropower station over many years as the goal.

[0085] Specifically, under the premise of flood control safety, with the goal of maximizing the power generation efficiency of the hydropower station, when factors such as losses, electricity sales costs, and time-of-use electricity prices are not considered, maximizing the power generation efficiency of the hydropower station is equivalent to maximizing the average power generation of the hydropower station over many years. Therefore, this embodiment establishes a mathematical model with the goal of maximizing the average power generation of the hydropower station over many years. The objective function is as follows:

[0086]

[0087] Where: The length of the measured inflow runoff data of the hydropower station, in years; is the total number of time periods within the year; For a long period of time; For the hydropower station Year The average output during the period, For the Year Reservoir water level during the period, For the Year The average inflow to the reservoir during a period, that is, the average output value of the hydropower station during a period, is related to the reservoir water level and the average inflow flow at that moment.

[0088] The constraints for the build are as follows:

[0089] (1) The guaranteed rate constraint is expressed by the following formula:

[0090]

[0091] Where, The power generation guarantee rate of the hydropower station calculated after the simulation and scheduling of the optimized scheduling diagram; It is the design guarantee rate of the hydropower station.

[0092] (2) The water balance constraint is expressed by the following formula:

[0093]

[0094] Where, Hydropower Station Year Reservoir water storage at the beginning and end of the time period; For the hydropower station Year Average inbound flow during the period; For the hydropower station Year Average outbound flow during the period; For the hydropower station Year Average loss flow during the period; Indicates a long period of time.

[0095] (3) The scheduling line constraint is expressed by the following formula:

[0096]

[0097] Where, These are the dead water level at the hydropower station at time t, the water level corresponding to the output limit line, the water level corresponding to the anti-destruction line, the water level corresponding to the increased output line, the water level corresponding to the anti-abandonment line, and the water level corresponding to the flood control dispatch line. During the dispatch diagram optimization process, the relative position of each dispatch line remains unchanged, and dispatch lines cannot intersect.

[0098] (4) Scheduling line bending energy constraint.

[0099] The dispatch line is a curve formed by multiple discrete points. From the perspective of aesthetics and practicality, the dispatch line needs to be as smooth as possible. If constraints in this regard are not set, the optimized dispatch line may have very obvious jagged edges. To this end, the following formula is used to determine the bending energy constraint:

[0100]

[0101]

[0102] Where m is the number of scheduling lines, n is the number of points on the scheduling line, is the slope of the i-th point of the j-th scheduling line; is the bending energy of the optimized scheduling graph, is the bending energy of the original scheduling diagram, The allowable increase margin.

[0103] Based on the above constraints and objective function, in order to pursue economic benefits while taking into account the guarantee rate of the hydropower station to the system, the requirement of the guarantee rate constraint is met by introducing a penalty function into the objective function. Therefore, the final form of the preset objective function in this embodiment is as follows:

[0104]

[0105] Where: 、 、 、 are penalty coefficients, which are determined according to the intensity of the penalty; the other variables have the same meanings as above.

[0106] Specifically, the above step S204 includes:

[0107] Step S2041: Multiple scheduling lines to be optimized in the conventional scheduling diagram are obtained. These multiple scheduling lines include a limited output line, an anti-destruction line, and an increased output line. Specifically, to ensure reservoir flood control safety, this embodiment optimizes the conventional scheduling diagram only for the reservoir's limited output line, anti-destruction line, and increased output line. Optimization of the flood control scheduling line is not considered. In actual applications, multiple scheduling lines in the conventional scheduling diagram can also be optimized.

[0108] In step S2042, the scheduling lines are selected in sequence according to the preset order, and based on the preset objective function, the differential evolution algorithm is used to optimize the restricted output line and the increased output line. Specifically, the optimization problem of the conventional scheduling diagram is a high-dimensional optimization problem, involving two levels: the optimization of each scheduling line in the conventional scheduling diagram and the time optimization of each time period of the specific scheduling line. The construction of the preset objective function represents the goal of optimizing the scheduling diagram, so the preset objective function can be used as a scheduling optimization model. The optimization process of the scheduling diagram is transformed into the solution process of the scheduling optimization model. In the solution process, based on the two optimization levels mentioned above, in order to reduce the difficulty of solving the scheduling optimization model, the differential evolution algorithm of the round-line iteration is used for solution.

[0109] Corresponding to the two levels of dispatch chart optimization, the differential evolution algorithm for line iteration also includes two layers. First, during the optimization of the reservoir dispatch chart, a line iteration method is used to reduce the dimension of the dispatch chart optimization while avoiding the intersection of dispatch lines. Second, when optimizing each dispatch line, a differential evolution algorithm is used to solve it. The output restriction line, the anti-destruction line, and the output increase line are optimized in sequence, and the optimized dispatch lines are finally obtained as the optimized dispatch chart. If the conventional dispatch chart includes I dispatch lines, these I dispatch lines can be sorted from bottom to top, and each dispatch line can be optimized in turn using the line iteration method.

[0110] The specific implementation steps of the wheel line iteration method are as follows:

[0111] (1) Based on the conventional scheduling diagram drawn by traditional methods, each initial scheduling line is represented as .

[0112] (2) Fix the remaining dispatch lines No change, optimize the first dispatch line (such as the output limit line) The specific optimization method uses the differential evolution algorithm to perform a long series of simulations and calculate the average power generation over many years. The scheduling line corresponding to the individual with the largest average power generation over many years in the population is taken as the first optimized scheduling line. .

[0113] (3) Let i be equal to , optimize the i-th scheduling line and fix the rest of the scheduling lines Recalculate the average power generation over the years and get the i-th optimal dispatch line .

[0114] In addition, for the multiple scheduling lines to be optimized in this embodiment, the differential evolution algorithm is directly used to schedule the output limiting line and the output increasing line, and the anti-destruction line can be optimized by further considering the abundance and scarcity factors.

[0115] In an optional implementation, the above step S2042 includes:

[0116] In step a1, an initial scheduling line population is uniformly generated based on the range values ​​formed by the upper and lower adjacent scheduling lines of the scheduling line to be optimized; specifically, when optimizing the i-th scheduling line, an initial population is uniformly generated based on the range values ​​between the scheduling line i-1 and the scheduling line i+1 to improve the convergence speed of the algorithm.

[0117] If the initial population size is assumed to be , the initial population can be expressed as:

[0118]

[0119] The dimension of the population is (correspond stage), then the initial population (i.e., generation 0) An individual can be represented as:

[0120]

[0121] In a certain dimension, the upper and lower limits of each individual should be the same. The curve formed by connecting the upper limit values ​​of each dimension, the curve formed by connecting the lower limit values ​​of each dimension, and the area between the two together constitute the search space of the population. The initial population should completely cover the search space, so the 0th generation individual No. Dimension elements can be randomly generated using the following formula:

[0122]

[0123] Where, Respectively represent The lower and upper bounds of the dimension element, A function that uniformly generates random numbers between [0,1].

[0124] Step a2: perform mutation, crossover, and selection operations on the initial scheduling line population to obtain the individual with the largest fitness value. The fitness value is calculated based on the objective function.

[0125] Specifically, the mutation operation is implemented using the following process: For any individual When performing mutation operation, the differential algorithm starts from the Randomly select two other different individuals from the population in , scale the vector difference of the two individuals and add it to the original individual On the above, we can get the mutant individual , the specific mutation method is implemented using the following formula:

[0126]

[0127] Where, is the scaling factor, which is used to control the influence of the difference vector, and is generally set between between.

[0128] The above mutation method is a common differential mutation strategy, called differential random strategy. Differential random strategy also has other forms of realization. It can also be achieved by randomly selecting four other different individual vectors, dividing them into two groups of two, and scaling the difference vectors of each group and comparing them with the individual vectors. In addition, the random strategy of differential variation can be improved by tending to the optimal individual strategy and the optimal individual strategy.

[0129] The crossover operation is implemented using the following process: Generation mutation intermediate Together they constitute the Generation of mutation vector population, mutation intermediate With the Generation population Perform cross-fusion operations between individuals to obtain the test vector population , No. The experimental individual can be expressed as This embodiment uses the following binomial crossover operator to perform the crossover operation:

[0130]

[0131] Where, is the crossover probability, is a constant, For uniform generation For the same individual, the dimensions are different but the random integers are the same. At least one element will be inherited to the experimental individual, so the crossover operation can increase the diversity of the population.

[0132] The selection operation is implemented using the following process: adopt the greedy selection strategy to select and update individuals, calculate each experimental individual With the Generation population individuals The fitness value of the two is compared, and the individual with better fitness value is selected to update the first Generation population, get the Generation population , the specific selection method is:

[0133]

[0134] Where: is the fitness calculation function (objective function value). In this embodiment, the objective function value is the multi-year average power generation. When calculating the objective function value, since there is no need to consider the annual type, the long series of measured runoff data is directly used for calculation.

[0135] Step a3, repeat the above mutation, crossover and selection operations until the preset requirements are met and the optimized scheduling line is obtained. Generation population The individual with the best fitness value is If the fitness value of the individual meets the error requirement or reaches the upper limit of the evolutionary generation, it is converged, the iteration is terminated, and the optimization is completed. If the relevant conditions are not met, the above mutation, crossover, and selection operations are repeated until the preset requirements are met. The individual with the best fitness value is used as the optimized scheduling line.

[0136] Step S2043: Based on the runoff data of different years, the anti-destruction line is optimized according to the year type using a differential evolution algorithm, and the optimized scheduling lines do not cross.

[0137] In an optional implementation, the above step S2043 includes:

[0138] Step b1, uniformly generating an initial anti-destruction line population based on the range values ​​formed by the upper and lower adjacent scheduling lines of the annual anti-destruction line to be optimized.

[0139] Step b2: perform mutation and crossover operations on the initial scheduling line population to obtain a new anti-destruction line population.

[0140] Step b3: extract the runoff data of the corresponding year, calculate the fitness value corresponding to each anti-tampering line individual in the new anti-tampering line population using a preset objective function, and select the individual with the largest fitness value.

[0141] Step b4, repeat the above mutation, crossover and selection operations until the preset requirements are met, and the optimized anti-destruction line of the corresponding year is obtained.

[0142] Specifically, to prevent hydropower stations from generating large amounts of ineffective water abandonment and improve their power generation efficiency, years with more abundant water inflow should increase their output earlier. Therefore, the anti-destruction line for years with more abundant water inflow should be lower than the anti-destruction line for years with less abundant water inflow. Therefore, this embodiment takes into account the runoff abundance and scarcity factors when optimizing the anti-destruction line. This results in a differential evolution algorithm that considers these factors. In this algorithm, the initial population generation, mutation, and crossover operations are identical to those of a differential evolution algorithm that does not consider these factors. During the selection process, this embodiment, based on the aforementioned year-type classification and the year type of the anti-destruction line to be optimized, extracts runoff data for the corresponding year type and calculates its fitness value. If the fitness value of the individual meets the error requirement or reaches the upper limit of the evolutionary generation, convergence is achieved, the iteration is terminated, and the optimization is complete. If the relevant conditions are not met, the mutation, crossover, and selection operations are repeated until the preset requirements are met. The individual with the best fitness value is then selected as the optimized anti-destruction scheduling line.

[0143] Among them, the optimized annual anti-destruction lines in this embodiment include seven types, specifically the optimized anti-destruction line of the extremely bumper year group, the optimized anti-destruction line of the flood year group, the optimized anti-destruction line of the normal to slightly bumper year group, the optimized anti-destruction line of the normal water year group, the optimized anti-destruction line of the normal to slightly dry year group, the optimized anti-destruction line of the dry year group, and the optimized anti-destruction line of the extremely dry year group.

[0144] Step S205 , determining the year type according to the runoff data grouped and predicted for the water conservancy year to be scheduled, and performing scheduling according to the scheduling line corresponding to the year type.

[0145] Specifically, the above step S205 includes:

[0146] Step S2051, forecast the runoff data from the current time to the end of the water conservancy year based on the passage of time. Specifically, in order to correct the impact of the forecast error, this embodiment forecasts the runoff data based on the passage of time. For example, the water conservancy year is from March to February of the following year, and the current time is March, then the runoff data from March to February of the following year are forecasted. After the preset period of time (such as the year as the preset period), when the time reaches April, the runoff data from April to February of the following year are re-forecasted; when the time reaches May, the runoff data from May to February of the following year are re-forecasted; and so on, until the forecast reaches February of the following year. In this way, by forecasting by time period, the impact of the forecast error is reduced compared to the method of directly forecasting the runoff data of the entire water conservancy year.

[0147] Step S2052: Determine the year type based on the distance between the forecast data and the cluster center. Specifically, the forecast runoff data is used to determine the year type through the method of step S203. The specific determination method can refer to the above steps and will not be repeated here.

[0148] Step S2053: Select a scheduling line corresponding to the year type for scheduling. Specifically, after the year type is determined, scheduling is performed based on the scheduling line corresponding to the year type.

[0149] As a specific application example of the embodiment of the present invention, the reservoir scheduling diagram optimization method is implemented using the following process:

[0150] The first step is to identify the flood and drought periods of a long series of runoff based on the clustering method.

[0151] The difference between high and low runoff is a relative concept, defined based on a specific criterion. The impact of runoff on reservoir operation is not only reflected in the magnitude of runoff, but also in the process of runoff itself. Therefore, using annual runoff as the sole criterion for runoff grouping is inappropriate. Therefore, it is crucial to select multiple indicators that describe runoff characteristics and employ appropriate methods to group runoff data.

[0152] 1.1, indicator selection.

[0153] Assume that the length of the long series of measured runoff data is Year, The annual runoff sample is Considering the differences in the total amount of runoff and runoff process in each year, each sample is selected characteristic indicators to fully reflect the characteristics of runoff in each year, namely . annual runoff samples, each with characteristic indicators, forming a The data matrix , specifically expressed as:

[0154] .

[0155] 1.2, indicator standardization.

[0156] Due to the incommensurability of the units and magnitudes of the indicators, the results of the indicators need to be standardized.

[0157] 1.3, principal component analysis.

[0158] When there is a correlation between multiple indicators, they will affect each other. Principal component analysis can cover most of the information in the original indicators, while simplifying the analysis process and increasing the accuracy of the results.

[0159] 1.4, K-means clustering method.

[0160] K-means clustering has the characteristics of fast convergence and high accuracy. This embodiment uses K-means clustering to cluster the extracted principal components. Because the cluster centers are randomly selected, multiple calculations can be performed to reduce the randomness of the results, and the average of each result is taken as the final classification result.

[0161] 1.5. According to the water conservancy year, long series runoff can be grouped by month or ten days.

[0162] For example, the hydropower year of a power station starts in March and ends in February of the following year. Assuming the scale is monthly, the grouping is as follows:

[0163] (1) First, for the runoff data from March to February of the following year, use 1 to 4 above to divide the annual type.

[0164] (2) Then, for the runoff data from April to February of the following year, the above 1-4 are used to divide the annual types in the same way.

[0165] (3) For runoff data until February of the following year, use 1-4 above to classify the annual type.

[0166] The same applies when using ten days as the scale.

[0167] This step provides support for building models and applying scheduling diagrams later.

[0168] In the second step, a hydropower station scheduling diagram optimization model considering bending energy is constructed.

[0169] Based on conventional dispatch diagrams drawn using traditional methods, with flood control safety as the prerequisite, the economic efficiency of the reservoir is the primary dispatching objective, and the reliability of the power supply to the power system is also taken into account to draw an optimized dispatch diagram for the hydropower station. To ensure the safety of reservoir flood control, the optimization of the conventional dispatch diagram only involves the reservoir's restricted output line, anti-destruction line, and increased output line, and does not consider the optimization of the flood control dispatch line. To achieve economic efficiency, this embodiment targets the maximum average multi-year power generation of the hydropower station. The consideration of the power system is reflected in the constraints. This chapter uses water balance constraints, guarantee rate constraints, and the non-intersection of the upper and lower lines of the dispatch diagram as constraints to establish a hydropower station dispatch diagram optimization model.

[0170] 2.1 Model construction

[0171] 2.1.1, Determine the objective function.

[0172] Under the premise of flood control safety, with the goal of maximizing the power generation efficiency of a hydropower station as the goal, when factors such as losses, electricity sales costs, and time-of-use electricity prices are not considered, maximizing the power generation efficiency of a hydropower station is equivalent to maximizing the average power generation of the hydropower station over many years. Therefore, this paper establishes a mathematical model with maximizing the average power generation of the hydropower station over many years as the goal.

[0173] 2.1.2, Constructing Constraints. The constraints constructed in this embodiment include guarantee rate constraints, water balance constraints, scheduling line constraints, and scheduling line bending energy constraints.

[0174] 2.1.3, Guaranteed rate and bending energy constraint processing.

[0175] The optimized dispatch chart is used to simulate the series dispatch of the hydropower station. Through statistical analysis of the calculation results, the guaranteed rate of the dispatch simulated by the optimized dispatch chart is obtained. In order to pursue economic benefits while taking into account the guaranteed rate of the hydropower station to the system, the guarantee rate constraint is met by introducing a penalty function into the objective function. Therefore, the final form of the objective function of the dispatch chart optimization model in this embodiment is as follows:

[0176]

[0177] 2.2, Model solving method.

[0178] Optimizing a regular schedule is a high-dimensional problem, involving two levels: optimizing each schedule line in the regular schedule and optimizing the schedules for each time period on a specific schedule line. To reduce the difficulty of solving the schedule optimization model, this embodiment uses a differential evolution algorithm with round-robin iterations.

[0179] 2.2.1, solution ideas.

[0180] Corresponding to the two levels of dispatch diagram optimization, the differential evolution algorithm for round-line iteration also includes two layers. First, during the reservoir dispatch diagram optimization process, a round-line iteration method is used to reduce the dimension of the dispatch diagram optimization while avoiding intersections of dispatch lines. Second, when optimizing each dispatch line, a differential evolution algorithm is used to solve the problem. The output restriction line, the anti-destruction line, and the output increase line are optimized in sequence, ultimately obtaining the optimized dispatch diagram for the hydropower station.

[0181] 2.2.2, the differential evolution algorithm with round-robin iteration is used to solve the problem. The specific process is as follows: Figure 2 shown.

[0182] Assuming that there are I scheduling lines in the hydropower station scheduling diagram, the I scheduling lines are sorted from bottom to top, and the round-robin iteration method is used to optimize each scheduling line in turn.

[0183] When optimizing the i-th scheduling line, first determine the parameters of the differential evolution algorithm. Among them, the parameters of the differential evolution algorithm mainly include the population size , dimension , scaling factor , crossover probability etc. Dimension Corresponding scheduling graph optimization stage, if the period is ten days, the dimension of the differential evolution population is 37, and the population size is and dimension The value of is related to the population size. Generally, the population size is 5 to 10 times the dimension. If the population size is too small, the population diversity is small and it is easy to fall into the local optimal solution. When the population size is too large, although the solution accuracy can be enhanced, the calculation time will be significantly increased. Therefore, the population size The determination of needs to be based on the nature of the calculation, the complexity, the size of the search space and the calculation time. Used to control the influence of the difference vector, scaling factor The larger the value, the greater the disturbance to the parent population. Too large will reduce the search efficiency, while too small will lead to insufficient disturbance, reduced diversity, and easy to fall into the local optimal solution. The general value is between The scaling factor is usually 0.5. The larger the value, the faster the convergence speed, but it is easy to lead to premature convergence and local optimal solution. The smaller the value, the more stable the algorithm. According to experience, 0.3 is a better choice.

[0184] An initial population is then generated uniformly between scheduling lines i-1 and i+1. Mutation, crossover, and selection are then applied to the initial population to generate new individuals. The newly generated individuals are then judged to see if they satisfy the constraints. If not, the individuals that violate the constraints are corrected. The parameter values ​​of the new individuals should satisfy the pre-set constraints. If not, they are adjusted to ensure they do. The specific correction process can employ boundary value correction methods, mapping methods, or penalty function methods.

[0185] When the newly generated individuals satisfy the constraints, the individual's fitness value is judged to meet the error requirement or reach the upper limit of the evolutionary generation number, which means convergence, the iteration is terminated, and the optimization is completed. If it does not meet the requirements, the mutation, crossover, and selection operations are repeated until the iterative convergence conditions are met. The optimization of the scheduling line is completed.

[0186] After that, let i=i+1, that is, optimize the next scheduling line. The specific optimization process refers to the above method and will not be repeated here. It should be noted that the scheduling lines optimized in this step include the increased output line and the limited output line.

[0187] The third step is to construct a hydropower station scheduling diagram optimization model considering the wet and dry factors.

[0188] To prevent large amounts of ineffective water abandonment and improve hydropower station efficiency, hydropower stations should increase their output earlier in years with more abundant water. Therefore, the anti-damage line for years with abundant water should be lower than that for years with less abundant water. Using the reservoir's initial water level and the hydrological year as decision-making indicators, the anti-damage lines are optimized separately for different runoff year groups, ensuring that the scheduling lines do not overlap.

[0189] Specifically, if Figure 3 As shown, the following process is used to optimize the anti-destruction line:

[0190] When optimizing the j-th anti-destruction line of the corresponding year, the parameters of the differential evolution algorithm are first determined.

[0191] An initial population is generated uniformly between dispatch lines j-1 and j+1. Mutation and crossover operations are performed on this initial population. Runoff data for the corresponding year is then extracted, and the fitness values ​​of each individual after the crossover operation are calculated using an objective function. The individual with the optimal fitness value is then selected through a selection operation. A determination is made as to whether the newly generated individuals satisfy the constraints. If not, the individuals violating the constraints are corrected. The parameter values ​​of the new individuals should satisfy the pre-set constraints. If not, they are adjusted to satisfy them. The specific correction process can employ boundary value correction, mapping, or penalty function methods.

[0192] When the newly generated individuals satisfy the constraints, the individual's fitness value is judged to meet the error requirement or reach the upper limit of the evolutionary generation, which means convergence, the iteration is terminated, and the optimization is completed. If not, the mutation, crossover, and selection operations are repeated until the iterative convergence conditions are met. The optimization of the anti-tampering line for that year is completed.

[0193] Next, let j = j + 1, optimizing the next year-type anti-destruction line. The specific optimization process is similar to the above method and will not be repeated here. In this embodiment, the optimized anti-destruction lines include the optimized anti-destruction lines for the exceptionally high-water year group, the optimized anti-destruction lines for the high-water year group, the optimized anti-destruction lines for the average-to-high-water year group, the optimized anti-destruction lines for the average-to-low-water year group, the optimized anti-destruction lines for the low-water year group, and the optimized anti-destruction lines for the exceptionally low-water year group.

[0194] It should be noted that during the first round of optimization, the release and destruction lines for different year types have not yet been generated. At this time, the optimized anti-destruction line for the exceptionally good year group is optimized first, and dispatching line j-1 and dispatching line j+1 are respectively used to increase output and limit output. Then, the optimized anti-destruction line for the high-water year group is optimized. At this time, dispatching line j-1 and dispatching line j+1 are respectively used to increase output and limit output, and so on. In the second round of optimization, all anti-destruction lines have been generated. When optimizing the optimized anti-destruction line for the exceptionally good year group, dispatching line j-1 and dispatching line j+1 are respectively used to increase output and limit output, and so on for the others.

[0195] By superimposing the optimized anti-destruction lines of each year group on the optimized dispatching diagram, the optimized dispatching diagram of the hydropower station considering the runoff factors can be drawn. Figure 4 As shown, line ① represents the optimized anti-damage line of the extremely good year group, line ② represents the optimized anti-damage line of the wet year group, line ③ represents the optimized anti-damage line of the normal to slightly good year group, line ④ represents the optimized anti-damage line of the normal water year group, line ⑤ represents the optimized anti-damage line of the normal to slightly dry year group, line ⑥ represents the optimized anti-damage line of the dry year group, and line ⑦ represents the optimized anti-damage line of the extremely dry year group.

[0196] The fourth step is to use the scheduling chart in conjunction with the hydrological rolling forecast.

[0197] 4.1. Determine the type of flood and drought.

[0198] 1. Obtain the runoff forecast results for each period for the remaining time of the scheduling year.

[0199] 2. Calculate the distance between it and each cluster center.

[0200] 3. Select the year type of the nearest cluster center as the year type for the scheduling period.

[0201] 4.2, Rolling Forecast Scheduling. The scheduling process is explained using the water conservancy year from March of the current year to February of the following year:

[0202] 1. For March, the hydrological forecast is for a wet or dry year from March of the current year to February of the following year. Select the corresponding dispatching line based on the forecast.

[0203] 2. In the next month, i.e. April, the hydrological forecast will predict the wet and dry year type from April to February of the following year, and then the corresponding dispatching line will be selected based on the forecast.

[0204] 3. Until the last scheduling period is forecast (February of the following year).

[0205] The advantage of this is that forecasts and scheduling can be carried out continuously and the impact of forecast errors can be continuously corrected.

[0206] In this invention, a scheduling diagram optimization model was established with the maximum average multi-year power generation as the objective function, and a differential evolution algorithm with round-robin iteration was used to solve the problem. This model overcomes the drawback of conventional scheduling diagrams, which place a high emphasis on the hydropower station's guarantee of the system (guaranteed output) and insufficient consideration of economic benefits. A long series of simulation scheduling processes comparing the conventional scheduling diagram with the optimized scheduling diagram yielded the following conclusions: Compared with the conventional scheduling diagram, the optimized scheduling diagram reduced the average water abandonment by 21.18% over the years and increased the average power generation by 1.05%. Among the 56 years of scheduling results, the years with increased number of ten-day disruptions to the optimized scheduling diagram all had less water inflow. The optimized scheduling diagram maintains head advantage and reduces power generation flow. Although this scheduling method reduces some of the duration guarantee rate and reduces the power generation in that year, it has positive significance for improving the long-term benefits of the reservoir.

[0207] In this invention, based on the optimization scheduling diagram, the factors of water abundance and drought are introduced, and a conceptual model of the optimization scheduling diagram considering the factors of abundance and drought is constructed. The optimized water level control method of the reservoir based on the scheduling diagram is explored, and the optimization scheduling diagram considering the factors of abundance and drought is drawn. Taking the Tankeng Reservoir as an example, 6 indicators are selected for principal component analysis, and K-means is used to identify the abundance and drought of 56 years of runoff data. The conceptual model of the optimization scheduling diagram considering the factors of abundance and drought is input, and the optimization scheduling diagram considering the factors of abundance and drought is drawn, and simulation scheduling is performed. Compared with the conventional scheduling diagram, the annual average water abandonment of the optimized scheduling diagram considering the factors of abundance and drought is reduced by 28.97%, and the power generation is increased by 1.289%. Compared with the optimized scheduling diagram, its multi-year average water abandonment is reduced by 8.96%, and the power generation is increased by 0.237%. This fully demonstrates that the optimization scheduling diagram considering the factors of abundance and drought can better coordinate the relationship between head and water volume, significantly reduce the ineffective water abandonment of hydropower stations, and improve the power generation efficiency of hydropower stations.

[0208] In this invention, rolling runoff forecasts are performed on a time-by-time basis to reduce scheduling deviations caused by forecast errors. The effectiveness of an optimized scheduling diagram that considers wet and dry conditions is also verified under both ideal forecast conditions and those with forecast errors. Case studies have shown that as the forecast error limit increases, the multi-year average power generation for both the conventional scheduling diagram and the optimized scheduling diagram that considers wet and dry conditions decreases, while the multi-year average water abandonment increases. Furthermore, when the forecast error ranges from 0% to 50%, the optimized scheduling diagram that considers wet and dry conditions consistently achieves greater power generation benefits and less ineffective water abandonment than the conventional scheduling diagram.

[0209] During the verification period, compared with the conventional scheduling diagram, the optimized scheduling diagram taking into account the factors of wet and dry seasons significantly reduced water abandonment by increasing the output in advance (April-September), in exchange for a greater water advantage. The output was appropriately reduced during the period of low water inflow, and the power generation head was re-accumulated to ensure that the power generation head would not be too low during the period of more water inflow (April-September). This allowed the reservoir to increase the power generation flow while still generating electricity at a higher efficiency, thereby improving power generation efficiency and significantly increasing the annual average power generation of the Tankeng Hydropower Station during the verification period.

[0210] In this embodiment, a reservoir scheduling diagram optimization device is also provided. The device is used to implement the above-mentioned embodiments and preferred embodiments. The details that have been described will not be repeated here. As used below, the term "module" can refer to a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation using hardware, or a combination of software and hardware, is also possible and contemplated.

[0211] This embodiment provides a reservoir scheduling diagram optimization device, such as Figure 5 As shown, including:

[0212] The data acquisition module 51 is used to obtain a long series of measured runoff data;

[0213] A grouping module 52 is configured to shift the start time backwards in accordance with a preset period based on the start time and end time of the water conservancy year, and to group the water conservancy years according to the shifted start time and end time;

[0214] The annual type classification module 53 is used to determine the corresponding annual type according to the runoff data in each water conservancy year group, and the annual type is used to characterize the abundance and scarcity of runoff;

[0215] An optimization module 54 is configured to optimize each dispatch line in the conventional dispatch diagram according to the year type using a differential evolution algorithm with round-line iteration based on a preset objective function, thereby obtaining optimized dispatch lines for different years. The preset objective function is constructed with the goal of maximizing the average multi-year power generation of the hydropower station;

[0216] The scheduling module 55 is used to determine the year type according to the runoff data predicted by the water conservancy year grouping according to the water conservancy year to be scheduled, and perform scheduling according to the scheduling line of the corresponding year type.

[0217] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0218] The embodiment of the present invention also provides a computer device having the above Figure 5 The reservoir operation diagram optimization device shown.

[0219] See also Figure 6 , Figure 6 is a structural diagram of a computer device provided by an optional embodiment of the present invention, such as Figure 6As shown, the computer device includes: one or more processors 10, memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components utilize different buses to communicate with each other and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in the memory or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Equally, multiple computer devices can be connected, and each device provides part of the necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 6 A processor 10 is taken as an example.

[0220] The processor 10 may be a central processing unit, a network processor, or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit, a programmable logic device, or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic, or any combination thereof.

[0221] The memory 20 stores instructions that can be executed by at least one processor 10, so as to enable at least one processor 10 to execute the method shown in the above embodiment.

[0222] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created based on the use of a computer device for displaying a small program landing page, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0223] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0224] The computer device further includes a communication interface 30 for the computer device to communicate with other devices or a communication network.

[0225] The embodiment of the present invention also provides a computer-readable storage medium. The above-mentioned method according to the embodiment of the present invention can be implemented in hardware, firmware, or implemented as a computer code that can be recorded in a storage medium, or implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and downloaded through a network and will be stored in a local storage medium, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state drive, etc.; further, the storage medium can also include a combination of the above-mentioned types of memory. It can be understood that a computer, a processor, a microprocessor controller or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor or hardware, the method shown in the above embodiment is implemented.

[0226] A portion of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the form in which the computer program instruction exists in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc. Accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium that can be accessed by the computer.

[0227] Although the embodiments of the present invention have been described with reference to the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention. Such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A reservoir dispatching diagram optimization method, characterized in that: The method comprises: Obtain a long series of measured runoff data; Based on the start time and end time of the water conservancy year, the start time is sequentially shifted backward according to a preset period, and the water conservancy year is grouped according to the shifted start time and end time; Determine the corresponding year type according to the runoff data in each water conservancy year group, wherein the year type is used to characterize the abundance and scarcity of runoff; Based on the preset objective function, the differential evolution algorithm with round-line iteration is used to optimize the dispatch lines in the conventional dispatch diagram according to the annual type, and the optimized dispatch lines for different years are obtained. The preset objective function is constructed with the goal of maximizing the average power generation of the hydropower station over many years. According to the runoff data predicted by the water conservancy year grouping for the water conservancy year to be dispatched, the year type to which it belongs is determined, and the dispatch is carried out according to the dispatch line corresponding to the year type; The preset objective function is constructed in the following way: Constructing constraints, including a guaranteed rate constraint, a water balance constraint, a scheduling line constraint, and a curved energy constraint, wherein the curved energy constraint is determined based on the slope of each point on the scheduling line; Based on the constraints, a preset objective function is constructed with the goal of maximizing the multi-year average power generation of the hydropower station.

2. The method according to claim 1, characterized in that The corresponding annual type is determined based on the runoff data within each water conservancy year group, including: Extract characteristic indicators of runoff data; Standardizing, principal component analysis, and clustering the characteristic indicators to obtain clustering results; Determine the year type based on the clustering results.

3. The method according to claim 1, characterized in that Based on the preset objective function, the differential evolution algorithm with round-line iteration is used to optimize the dispatch lines in the conventional dispatch diagram according to the year type. The optimized dispatch lines for different years are obtained, including: Acquire multiple scheduling lines to be optimized in a conventional scheduling diagram, the multiple scheduling lines including a power output limiting line, an anti-destruction line, and an increased power output line; The dispatch lines are selected in sequence according to the preset order, and the output limiting line and the output increasing line are optimized using the differential evolution algorithm based on the preset objective function; Based on the runoff data of different years, the differential evolution algorithm is used to optimize the anti-destruction line according to the year type, and the optimized scheduling lines do not cross.

4. The method according to claim 3, characterized in that Based on the preset objective function, the differential evolution algorithm is used to optimize the output limit line and the output increase line, including: The initial scheduling line population is uniformly generated based on the range values ​​formed by the upper and lower adjacent scheduling lines of the scheduling line to be optimized; Perform mutation, crossover, and selection operations on the initial scheduling line population to obtain the individual with the maximum fitness value, where the fitness value is calculated based on the objective function; Repeat the above mutation, crossover and selection operations until the preset requirements are met and the optimized scheduling line is obtained.

5. The method according to claim 3, characterized in that Differential evolution algorithm is used to optimize the anti-tampering line according to the year type, including: Generate an initial anti-destruction line population based on the range values ​​formed by the upper and lower adjacent scheduling lines of the annual anti-destruction line to be optimized; Perform mutation and crossover operations on the initial scheduling line population to obtain a new anti-destruction line population; Extract the runoff data of the corresponding year, use the preset objective function to calculate the fitness value of each anti-tampering line individual in the new anti-tampering line population, and select the individual with the largest fitness value; Repeat the above mutation, crossover and selection operations until the preset requirements are met, and the optimized anti-destruction line of the corresponding year is obtained.

6. The method according to claim 2, characterized in that Based on the runoff data forecasted by the water conservancy year group, the year type is determined and the dispatch is carried out according to the dispatch line of the corresponding year type, including: Runoff forecast data from the current time to the end of the water conservancy year is forecasted based on the passage of time; Determine the year type according to the distance between the runoff forecast data and the cluster center; According to the year type, the scheduling line corresponding to the year type is selected for scheduling.

7. A reservoir dispatching diagram optimization device, characterized in that: The device comprises: Data acquisition module, used to obtain long series of measured runoff data; A grouping module is used to shift the start time and end time of the water conservancy year in sequence according to a preset period, and group the water conservancy years according to the shifted start time and end time; The annual type classification module is used to determine the corresponding annual type according to the runoff data in each water conservancy year group, and the annual type is used to characterize the abundance and scarcity of runoff; The optimization module is used to optimize each dispatch line in the conventional dispatch diagram according to the year type based on the preset objective function and adopt the differential evolution algorithm with round-line iteration to obtain the optimized dispatch lines of different years. The preset objective function is constructed with the goal of maximizing the average power generation of the hydropower station over many years; The scheduling module is used to determine the year type according to the runoff data predicted by the water conservancy year group, and perform scheduling according to the scheduling line of the corresponding year type; The preset objective function is constructed in the following way: Constructing constraints, including a guaranteed rate constraint, a water balance constraint, a scheduling line constraint, and a curved energy constraint, wherein the curved energy constraint is determined based on the slope of each point on the scheduling line; Based on the constraints, a preset objective function is constructed with the goal of maximizing the multi-year average power generation of the hydropower station.

8. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the reservoir scheduling diagram optimization method according to any one of claims 1 to 6 by executing the computer instructions.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a computer to execute the reservoir scheduling diagram optimization method according to any one of claims 1 to 6.

Images