A method for peak-shifting dispatching based on dimension reduction optimization of multi-reservoir groups based on hotspot reservoir selection

Through the multi-reservoir group dimensionality reduction optimization peak staggered scheduling method selected based on hotspot reservoirs, the problems of 'dimensional disaster' and low efficiency in the joint flood control scheduling of reservoir groups are solved, efficient optimization scheduling is achieved, and calculation speed and methods are improved. practicality.

CN118780446BActive Publication Date: 2024-11-15DALIAN UNIV OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411262733.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-09-10
Publication Date
2024-11-15
Estimated Expiration
2044-09-10

AI Technical Summary

Technical Problem

The existing joint flood control scheduling method of reservoir groups has problems such as 'dimensional disaster', early maturity convergence, low solution efficiency, and inappropriate theory and practice, and has not been effectively applied to actual scheduling.

Method used

A multi-store reservoir group dimensionality reduction optimization peak scheduling method based on hotspot reservoir selection is proposed. By analyzing the flood control effect of each reservoir on the scene flood, hotspot reservoirs are selected and a multi-store joint peak scheduling optimization scheduling is generated using POA algorithm to meet flood control safety. plan.

Benefits of technology

It effectively overcomes the 'dimensional disaster', significantly improves the calculation speed and optimization capabilities of the optimization scheduling scheme, enhances the practicality of the method, and is suitable for joint flood control scheduling of multiple series and parallel forms of reservoirs.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118780446B_ABST
    Figure CN118780446B_ABST
Patent Text Reader

Abstract

A method for optimizing peak-shifting scheduling of a multi-reservoir group based on dimensionality reduction of hotspot reservoirs, which belongs to the field of reservoir group flood control optimization scheduling. It includes two major stages: selecting hotspot reservoirs and optimizing scheduling. In the first stage, the hotspot reservoirs and non-hotspot reservoirs of this flood are divided by analyzing the flood control effects of each reservoir in the basin on this flood; in the second stage, the initial solution is generated according to the regulation scheduling, and the multi-reservoir joint peak-shifting optimization scheduling scheme that meets flood control safety is quickly generated through the POA algorithm. The present invention considers various water inflow conditions in the upstream and interval to determine the objective function, and proposes a dimensionality reduction method based on hotspot reservoirs for indicators such as reservoir discharge, remaining flood control storage capacity, and comprehensive indicators of future flood control situation characteristics in view of various complex constraints. It can effectively overcome the problems of "dimensionality disaster" and low solution efficiency, improve the solution efficiency of the scheduling method, enhance the practicality of the method, and can be widely used in joint flood control scheduling of various series and parallel reservoirs.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the field of reservoir joint flood control optimization scheduling, and relates to a multi-reservoir reservoir group dimensionality reduction optimization peak-shifting scheduling method based on hotspot reservoir selection. Background Art

[0002] Joint dispatching of reservoir groups, giving full play to the runoff compensation effect between reservoir groups, and jointly defending against floods are currently important means of flood control dispatching of reservoir groups.

[0003] In 2011, Ji Changming et al. [Ji Changming, Yu Shan, Zhou Ting, et al. Application of ant colony algorithm in optimization of hydropower station dispatching function [J]. Automation of Electric Power Systems, 2011, 35(20): 103-107] used the ant colony algorithm (ACO) to solve the optimization model of cascade reservoir dispatching function in order to solve the problem of optimization space of dispatching function. The results showed that the optimized dispatching function has good operation efficiency. However, the traditional ACO is prone to fall into local optimum and is too dependent on the calibration of initial pheromone, which leads to low efficiency of the algorithm. In 2015, Jiang Zhiqiang et al. [Jiang Zhiqiang, Ji Changming, Sun Ping, et al. Comparative analysis of three parallel modes of multidimensional dynamic programming [J]. China Rural Water Conservancy and Hydropower, 2015(3): 168-173, 176] parallelized the multidimensional dynamic programming algorithm and constructed parallel algorithms between time periods, within time periods and a mixture of the two, which alleviated the problem of "curse of dimensionality" to a certain extent. In 2021, Chen et al. [Chen H Tao, Wang W Chuan, Chau K Wing, et al. Flood Control Operation of Reservoir Group Using Yin-Yang Firefly Algorithm[J]. Water Resources Management, 2021, 35(15): 5325-5345] established a flood control operation model for a reservoir group with the minimum flood storage capacity of each reservoir and the minimum peak flow at the downstream control point during the operation process. The actual example verified that the model can effectively reduce the flood control storage capacity and peak flow at the downstream control point of the basin.

[0004] From the above research, it can be seen that the research on joint flood control dispatching of reservoir groups has been extensive and has achieved a lot of results. Joint optimization dispatching models of various series-parallel reservoir groups have been established, and various ideas and methods have been introduced to solve and apply to actual basins. The problems existing in the research on joint flood control dispatching of reservoir groups have been solved from all aspects. However, there are still problems such as "dimensionality disaster", premature convergence, low solution efficiency, and the lack of relevance between theory and practice in joint flood control dispatching of reservoir groups. There is still no practical and mature dispatching method that can be truly applied to solve actual dispatching problems. Joint flood control optimization dispatching of reservoir groups is still in the research stage. Therefore, exploring reasonable and effective dimensionality reduction methods and applying relevant theoretical methods to the actual joint flood control optimization dispatching of reservoir groups is of great significance and practical value for theoretical guidance of practice.

[0005] Gao Mengping et al. [Gao Mengping, Cai Huaxiang, Li Xiufeng, et al. Reservoir operation method based on orthogonal successive approximation algorithm [P]. Yunnan Province: CN201310726041.4, 2016-08-31.] Orthogonal experimental design is used to alleviate the "curse of dimensionality" problem, but the efficiency and effect will be affected when dealing with large-scale reservoir group problems, which may not be enough to meet the needs of real-time operation. Continuous linear programming is suitable for dealing with linear constraint problems, but it may reduce the accuracy of the results when dealing with problems involving complex logical constraints or nonlinear constraints. Real-time operation is a dynamic problem, and this method is generally based on static model optimization, which may lead to insufficient timeliness and adaptability of the generated operation plan. [Lei Xiaohui, Li Xianxin, An Bo, et al. A reservoir optimization operation method based on continuous linear programming [P]. Beijing: CN201210142729.3, 2012-09-19.]. Summary of the invention

[0006] In view of the shortcomings of the existing technology, the present invention considers the number of reservoirs to be optimized for flood events, and proposes a dimensionality reduction optimization peak-shifting scheduling method for a multi-reservoir group based on the selection of hotspot reservoirs. The method can effectively overcome the problems of "dimensionality curse" and low solution efficiency, improve the solution efficiency of the multi-reservoir group joint flood control optimization peak-shifting scheduling method, enhance the practicality of the method, and can be widely used in joint flood control scheduling of various series and parallel reservoirs.

[0007] In order to solve the above technical problems, the technical solution of the present invention is:

[0008] A multi-reservoir group dimensionality reduction optimization peak-shifting scheduling method based on hotspot reservoir selection, the multi-reservoir group dimensionality reduction optimization peak-shifting scheduling method includes two stages: selecting hotspot reservoirs and optimizing scheduling. The first stage divides the hotspot reservoirs and non-hotspot reservoirs of this flood by analyzing the flood control effects of each reservoir in the basin on this flood; the second stage generates an initial solution according to the regulation scheduling, and quickly generates a multi-reservoir joint peak-shifting optimization scheduling plan that meets flood control safety through the POA algorithm. The output of the optimized scheduling plan is completed according to the following steps (1) to (11):

[0009] Step (1) All reservoirs are regularly dispatched for each flood event.

[0010] In step (1), "rule-based scheduling" refers to the corresponding rule-based scheduling methods already in place for each reservoir.

[0011] Step (2) Calculate the remaining flood control storage capacity of each reservoir according to the rule dispatch results of step (1) , Downflow , Comprehensive indicators of future flood control situation characteristics The subscript i represents the i-th reservoir. It represents the i-th time period under the rule scheduling, and its time period division is based on the general requirements of the field.

[0012] Step (3) comprehensively considers the sizes of the three indicators obtained in step (2), determines the flood control effect of each reservoir in this flood, and roughly screens out the hotspot reservoirs and non-hotspot reservoirs in this flood.

[0013] Furthermore, the principle for roughly screening hotspot reservoirs and non-hotspot reservoirs in step (3) is:

[0014] First, based on the existing reservoir data and flood information, the corresponding conditions of the eight indicators of each reservoir are determined, and the above three indicators are used as the main basis for judging candidate non-hotspot reservoirs, as follows:

[0015] 3.1) , is the flood control capacity of the reservoir, m 3 ; is a coefficient, ranging from 0.5 to 1.

[0016] 3.2) , is the safe discharge flow of the reservoir, m 3 / s; is a coefficient, and its value range is 0.5~1.

[0017] 3.3) The larger it is, the stronger the flood control function of the reservoir is.

[0018] Reservoirs that meet the requirements of indicators 3.1) and 3.2) at the same time and whose comprehensive ranking of indicator 3.3) is in the 80% are roughly screened hot reservoirs.

[0019] Step (4) combining the use constraints of the reservoir storage and discharge sequence, performing a secondary screening on the hotspot reservoirs roughly screened in step (3) to accurately determine the hotspot reservoirs;

[0020] Furthermore, in step (4), the principle of secondary screening is:

[0021] 4.1) Sort the indicators 3.1 corresponding to the hotspot reservoirs roughly screened out in step (3) from large to small, and select the top 80% of the roughly screened hotspot reservoirs as the final hotspot reservoirs.

[0022] 4.2) Combined with the information of the previous flood. If the previous flood has not subsided in the basin area controlled by a certain reservoir among the roughly screened hotspot reservoirs, its flood prevention effect on the current flood is relatively small and it will not be considered as a hotspot reservoir.

[0023] 4.3) Combined with short-term rainfall forecast information. If, among the roughly screened hotspot reservoirs, the next flood period of the basin area controlled by a certain reservoir is close to the current flood period, or the rainfall center is located in the area controlled by the reservoir, the storage capacity of the candidate hotspot reservoir is mainly reserved for preventing the next flood. It is considered that the flood prevention effect on the current flood is relatively small and it is not used as a hotspot reservoir.

[0024] Step (5) According to the rule dispatch result of step (1), it is used as the initial discharge trajectory of the hotspot reservoir POA optimization algorithm in step (4). , regular dispatching of non-hotspot reservoirs. Refers to the discharge process of the i-th reservoir under the single reservoir rule scheduling.

[0025] The rule-based dispatching result is: the dispatching result of each reservoir participating in the dispatching according to the single reservoir dispatching rule.

[0026] The initial discharge trajectory is determined by directly using the regular dispatching results of each reservoir involved in the dispatching.

[0027] The method of regular dispatching of non-hotspot reservoirs is: single reservoir dispatching rules for each non-hotspot reservoir.

[0028] Step (6) divides the total flood dispatching time T into a certain number of time periods according to the set optimization step size. The method of the present invention sets the total flood dispatching time of each reservoir to be the same. From this step, the POA algorithm iterative optimization is performed.

[0029] The algorithm involves parameters that need to be set in advance, including: optimization time step, number of optimization iterations, convergence accuracy, etc.

[0030] Algorithm-related settings include: the number of flood scheduling periods, the minimum water level allowed for each reservoir, the maximum water level allowed for each reservoir, the final water level of each reservoir, etc.

[0031] It refers to the objective function value of the i-th period during the j-th iteration when the POA algorithm performs optimal scheduling for hotspot reservoirs.

[0032] Refers to the objective function corresponding to the complete scheduling process after the nth optimization, that is, the algorithm optimizes each time period The objective function of the scheduling process generated under the optimal condition is calculated. Specifically refers to the objective function under the regular operation of all reservoirs without optimized operation.

[0033] 6.1) Divide the initial discharge trajectory of each hotspot reservoir into this time period, where the time period sequence number starts from 0;

[0034] 6.2) The regular dispatching process of non-hotspot reservoirs is divided into these time periods, and the time period numbers are sorted starting from 0.

[0035] Secure all reservoirs and , that is, the discharge flow in the 0th and 2nd periods of the initial discharge trajectory line.

[0036] Step (7) Discrete the discharge volume of each hotspot reservoir in the first period according to the algorithm , that is, the discharge between the 0th period and the 2nd period of each hotspot reservoir is calculated by 1 m 3 / s is the division value. The discharge flow of each non-hotspot reservoir in the first period is the discharge flow of the corresponding period of regular scheduling, and no adjustment is made; the discharge flow of each hotspot reservoir in the first period is adjusted.

[0037] Step (8) Determine the objective function based on the flood water characteristics and calculate the objective function . Minimize the objective function value in the 0th and 1st periods, Optimal means that the objective function value in the first optimization calculation of the algorithm is the smallest in the 0th and 1st time periods. Refers to the objective function value of the i-th period during the j-th iteration of the algorithm.

[0038] Furthermore, the flood water characteristics refer to the proportion of upstream reservoir water and interval water in the flood source composition. Define the ratio P of upstream reservoir water: interval water:

[0039] When P>P1, the objective function selects the maximum flood control safety guarantee criterion;

[0040] When P < P2, the objective function selects the maximum peak shaving criterion, where P1 > P2;

[0041] When P2 < P < P1, the objective function selects the combined weight criterion of the maximum peak shaving criterion and the maximum flood control safety guarantee criterion.

[0042] The above-mentioned P1 is 7:3, and P2 is 3:7.

[0043] In the above-mentioned step (8), the objective function is specifically:

[0044] 8.1) Taking the maximum flood control safety guarantee criterion as the flood control optimization criterion, that is, the objective function is to maximize or minimize the flood control storage capacity occupied by each reservoir: F = my ∑ i = 1 N [max( V i , 1 , V i , 2 , ⋯ , V i , t , ⋯ , V i , T )] ;

[0045] 8.2) Expressing the flood control optimization objective function with the maximum peak shaving criterion: = my ∑ t = 1 T ∑ i = 1 N [ Q seconds _ i , t + Q down _ i , t ] 2 ;

[0046] 8.3) The combined weight criterion of the maximum peak shaving criterion and the maximum flood control safety guarantee criterion, and its objective function can be expressed as: F = α 1 my ∑ t = 1 T ∑ i = 1 N [ Q seconds _ i , t + Q down _ i , t ] 2 + α 2 my ∑ i = 1 N [max( V i , 1 , V i , 2 , ⋯ , V i , t , ⋯ , V i , T )] 。

[0047] The so-called "the objective function no longer improves" here means that the difference between the two objective function values is less than 0.0001.

[0048] In step (9), the algorithm performs optimization for each time period one by one until the Tth time period. According to the principle of optimizing each time period in turn to make the overall objective function optimal, the discharge of each time period is readjusted until all time periods reach the optimal state, that is, the optimal objective function value is obtained. 。When adjusting the time period in step (7) a round of iterative calculation is completed, that is, a round of optimization for each time period by the algorithm ends;

[0049] In step (10), the objective function corresponding to the optimized complete scheduling process is compared with the objective function under the regular scheduling of all reservoirs to determine whether the optimized corresponding objective function meets the accuracy requirements or the objective function no longer improves. If neither of the above two determination conditions is satisfied, return to step (6) to start a new round of optimization, or when the set number of optimization iterations is reached, the iteration stops, and the optimal solution is obtained at this time. Among them refers to the objective function corresponding to the nth optimized complete scheduling process, specifically refers to the objective function under the regular scheduling of all reservoirs without optimization scheduling.

[0050] Furthermore, in the above-mentioned step (10), the accuracy requirement specifically refers to: the convergence accuracy is 0.0001.

[0051] The “objective function no longer improves” here means that the difference between two objective function values ​​is less than 0.0001.

[0052] The “return to step (6) to start a new round of optimization” here means: updating the complete dispatching process of each hotspot reservoir after this optimization to the initial discharge trajectory line, and the other parameter settings remain unchanged.

[0053] Step (11) outputs a flood control optimization peak-shifting scheduling plan based on the optimal solution obtained in step (10).

[0054] The beneficial effects of the present invention are:

[0055] While taking into account the flood control tasks of each reservoir, the present invention selects appropriate hotspot reservoirs and objective functions according to flood characteristics, carries out multi-reservoir joint dimensionality reduction flood control optimization peak shifting scheduling to achieve dimensionality reduction, effectively overcomes the "dimensionality disaster", and significantly improves the calculation speed of generating optimized scheduling schemes and improves optimization capabilities. The scheduling method provided by the present invention can enhance the practicality of the method and is widely used in joint flood control scheduling of reservoirs in various series and parallel forms. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 This is a flow chart of the peak-shifting dispatching of a multi-reservoir group based on the selection of hotspot reservoirs according to the present invention by reducing the dimension and optimizing the peak-shifting dispatching;

[0057] Figure 2 This is a generalized map of a group of reservoirs, where multiple reservoirs are used together to protect key flood control control points downstream. DETAILED DESCRIPTION

[0058] The present invention establishes a multi-reservoir group dimensionality reduction optimization peak shifting scheduling model. Under the premise of ensuring the flood control safety of the basin, it seeks a multi-reservoir joint flood control optimization scheduling method with high solution efficiency and good optimization effect. The specific process is detailed in Figure 1 .

[0059] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0060] The present invention consists of two parts: the selection of hotspot reservoirs and the solution of the optimal dispatching scheme. In the first stage, the hotspot reservoirs and non-hotspot reservoirs of this flood are divided by analyzing the flood control effects of each reservoir in the basin on this flood; in the second stage, the initial solution is generated according to the regulation dispatching, and the multi-reservoir joint peak-shifting optimization dispatching scheme that meets flood control safety is quickly generated through the POA algorithm.

[0061] 1. Selection of hotspot reservoirs;

[0062] Since the flood control effect of each reservoir varies in different flood events, we analyze the flood events from the perspective of the number of reservoirs optimized for each flood event, screen out reservoirs with strong flood control effects, and use them as hotspot reservoirs for optimized scheduling. Otherwise, use non-hotspot reservoirs for regular scheduling, thereby reducing decision variables and achieving dimensionality reduction.

[0063] The present invention provides the following indicators to judge hotspot reservoirs and non-hotspot reservoirs in a flood:

[0064] (1) Location of the center of the rainstorm: When the basin area is large, the flood inflow into the reservoir outside the center of the rainstorm is small and the flood control effect is weak, while the flood inflow into the reservoir in the center of the rainstorm is large and the flood control effect is strong.

[0065] (2) Current reservoir water storage status: that is, the corresponding reservoir water level when flood control is initiated. The closer the reservoir water level is to the flood limit water level, the more remaining flood control storage capacity there is, and the stronger the flood control effect is; conversely, the flood control effect is weaker.

[0066] (3) Flood control storage capacity: The flood control storage capacity of each reservoir in the basin is different, and the optimization space size of the downstream flood control section is different. Reservoirs with relatively large flood control storage capacity have stronger flood control effects; conversely, reservoirs with relatively large flood control capacity have weaker flood control effects.

[0067] (4) Remaining flood control storage capacity : Under regular dispatch, the difference between the peak time of the downstream flood control section and the time when the reservoir outflow evolves to the downstream flood control section is The remaining flood control storage capacity of the reservoir during the period. When the remaining flood control storage capacity is relatively large compared to the flood control storage capacity, it is considered that the flood control effect is stronger:

[0068] ;

[0069] Where: is the flood control capacity of the reservoir, m 3 ; is a coefficient, ranging from 0.5 to 1, and in this embodiment, the value is 0.7.

[0070] (5) Reservoir outflow : Under rule scheduling, The amount of water discharged from the reservoir during the period. When the discharge volume is relatively large compared to the downstream safe discharge volume, the reservoir can reduce the discharge volume more and the flood control effect is stronger:

[0071] ;

[0072] Where: is the safe discharge flow of the reservoir, m 3 / s; is a coefficient, and its value range is 0.5~1. In this embodiment, its value is 0.7.

[0073] (6) Ratio of reservoir discharge to peak flow at downstream flood control section : Under regular dispatch, the reservoir discharge flow in the period Peak flow rate of downstream flood control section When When the reservoir outflow is large, the peak flow of the downstream flood control section is large, and the flood control effect is strong:

[0074] ;

[0075] (7) Reservoir storage and discharge order: Based on the needs of flood protection objects and reservoir storage and discharge needs, the storage and discharge order of reservoir groups is considered in the order of "synchronous storage and discharge" and "asynchronous storage and discharge". Reservoirs that are given priority for interception and storage and have sufficient interception and storage capacity are considered hot reservoirs;

[0076] (8) Comprehensive index of future flood control situation characteristics G i : Analyze the quantitative indicators that can describe the strength of flood control and select qualitative principles, such as rainwater monitoring and forecasting capabilities, forecast rainfall information for a certain period after a flood, basin landforms, flood discharge capacity of rivers and estuaries, changes in the storage capacity of lakes and flood storage areas, etc. The weight of the characteristic quantity index is determined by the degree of influence on the flood control situation, and the final characteristic quantity comprehensive index is obtained through comprehensive analysis. The larger the characteristic quantity comprehensive index, the stronger the flood control effect.

[0077] To judge the future flood control situation, m characteristic indicators need to be considered, so the design frequency standard indicator value vector is established as:

[0078] G = [ G 1 , G 2 , ⋯ , G m ] ;

[0079] Where: G i is the comprehensive index of characteristic quantity corresponding to reservoir i.

[0080] The above indicators are used to jointly judge the hotspot reservoirs of the flood. The above indicators are ranked in order of importance: (4) = (5) > (1) > (3) = (2) > (8) > (7) > (6), and indicators (4), (5), and (8) are selected as the main judgment indicators. When the requirements of indicators (4) and (5) are met at the same time, and indicator (8) ranks in the top 60%, it is selected as a candidate hotspot reservoir. According to the degree of satisfaction of indicator (7), the flood control role of the candidate hotspot reservoir in this flood is judged, and the hotspot reservoir is accurately determined, otherwise it is a non-hotspot reservoir.

[0081] 2. Solve the optimization scheduling plan;

[0082] As the number of reservoirs increases and the calculation period lengthens, the scale of the problem will rise sharply, and the dynamic programming method will face a serious "curse of dimensionality" problem. The method of the present invention uses a stepwise optimization algorithm as a solution strategy based on the characteristics of the problem. Starting from an initial feasible solution, the method iteratively optimizes the size of the discharge flow in each period during the entire scheduling process until it converges to the global optimal solution. The solution process includes the following two main parts:

[0083] (1) Determine the initial feasible solution. Using the rule dispatch results, generate an initial feasible solution that is relatively stable in terms of discharge from each reservoir and satisfies the reservoir discharge constraints. The rule dispatch results refer to the dispatch operation results under the rule dispatch mode corresponding to each reservoir.

[0084] (2) Based on the initial feasible solution obtained in the first step, the discharge flow in each time period is readjusted according to the rule of optimizing the time periods in sequence to make the overall objective function optimal until all time periods reach the optimal state.

[0085] Different flood sources have different compositions. For upstream reservoirs with smaller inflows than interval water, the smaller the peak flow of the downstream flood control section, the better, provided that the maximum water level of the reservoir does not exceed the maximum allowable water level. This method uses the maximum peak reduction criterion to represent the flood control optimization objective function, as shown in the following formula:

[0086] = my ∑ t = 1 T ∑ i = 1 N [ Q seconds _ i , t + Q down _ i , t ] 2 ;

[0087] Where: F is the objective function value; T is the total time period; N is the total number of reservoirs; i is the reservoir number; t is the time period number; is the interval inflow of reservoir i in period t, m 3 / s; is the discharge flow of reservoir i at period t, evolving through Muskingum to the downstream flood control section, m 3 / s; , is the discharge flow of reservoir i in period t, m 3 / s; are the flow calculation coefficients calculated based on the data of the embodiment, which are the river section propagation time K, flow weight factor x, and calculation period under the condition of stable flow. The function corresponds to the discharge flow in period t+1 evolving from Muskingum to the downstream flood control section flow, the discharge flow in period t evolving from Muskingum to the downstream flood control section flow, and the reservoir discharge flow in period t.

[0088] When the water inflow to the upstream reservoir is larger than that to the interval, the flood control storage capacity of each reservoir should be used as little as possible, under the premise of meeting the safety of the downstream flood control section, in order to prepare for the subsequent possible major floods. This method uses the maximum flood control safety guarantee criterion as the flood control optimization criterion, that is, the maximum and minimum flood control storage capacity occupied by each reservoir is taken as the objective function:

[0089] F = my ∑ i = 1 N [max( V i , 1 , V i , 2 , ⋯ , V i , t , ⋯ , V i , T )] ;

[0090] Where: is the water storage capacity of reservoir i in period t, m 3 .

[0091] In the case of large inflows from upstream reservoirs and intervals, under the premise of ensuring the safety of the downstream flood control section and the maximum water level of the reservoir not exceeding the maximum allowable water level, it is necessary to minimize both the peak flow of the downstream flood control section and the utilization of the flood control capacity of each reservoir. The flood control optimization criterion of this method is a combined weighted criterion of the maximum peak reduction criterion and the maximum flood control safety guarantee criterion, and its objective function can be expressed as:

[0092] F = α 1 my ∑ t = 1 T ∑ i = 1 N [ Q seconds _ i , t + Q down _ i , t ] 2 + α 2 my ∑ i = 1 N [max( V i , 1 , V i , 2 , ⋯ , V i , t , ⋯ , V i , T )] ;

[0093] Where: is the weight coefficient of the maximum peak shaving criterion objective function; is the weight coefficient of the objective function of the maximum flood control safety guarantee criterion.

[0094] The constraint expressions that need to be satisfied to solve the flood control optimization scheduling problem of reservoir groups are as follows:

[0095] (1) Water balance constraints:

[0096] ;

[0097] Where: is the water storage capacity of reservoir i at the tth and t+1th periods, m 3 ; is the inflow to reservoir i in period t, m 3 / s; is the outflow of reservoir i in period t, m 3 / s; To calculate the long period, 3600s is taken.

[0098] (2) Water level constraint:

[0099] ;

[0100] Where: represents the water level of reservoir i in time period t, m; are the upper and lower limits of the water level allowed during operation of reservoir i, m.

[0101] (3) Final water level constraint:

[0102] ;

[0103] Where: is the water level of reservoir i in period T, m; is the final water level of reservoir i, m.

[0104] (4) Discharge capacity constraints:

[0105] ;

[0106] Where: is the discharge capacity of reservoir i at period t, m 3 / s.

[0107] (5) Reservoir discharge safety constraints:

[0108] ;

[0109] Where: is the safe discharge flow of reservoir i, m 3 / s.

[0110] (6) Discharge flow rate fluctuation constraints:

[0111] ;

[0112] In the formula, is the maximum outflow amplitude of reservoir i, m 3 / s.

[0113] (7) Downstream flood control safety constraints:

[0114] ;

[0115] Where: is the maximum flow allowed in the downstream flood control section, m 3 / s.

[0116] (8) Flood peak constraints for discharge into the reservoir:

[0117] ;

[0118] Where: is the flood peak in reservoir i, m 3 / s.

[0119] According to the above ideas, a complete optimization scheduling process is completed according to the following steps (1)-(11) to complete the specific steps of outputting the optimization scheduling plan:

[0120] (1) All reservoirs are regularly dispatched for flood events;

[0121] The “regular dispatching” mentioned above refers to the existing regular dispatching method corresponding to each reservoir, and the time periods are divided into 1 hour.

[0122] (2) Based on the regular dispatch results obtained in step (1), calculate the remaining flood control storage capacity of each reservoir in the flood event , Downflow , Comprehensive indicators of future flood control situation characteristics The subscript i represents the i-th reservoir. It represents the i-th time period under the regular scheduling, and its time period division is based on the conventional requirements of the field. In this embodiment, the time period is divided into 1 hour.

[0123] (3) Comprehensively consider the sizes of the three indicators obtained in step (2), judge the flood control role of each reservoir in this flood, and roughly screen out the hotspot reservoirs and non-hotspot reservoirs in this flood.

[0124] First, based on the existing reservoir data and flood information, the corresponding conditions of the eight indicators of each reservoir are determined, and the above three indicators are used as the main basis for judging candidate non-hotspot reservoirs, as follows:

[0125] 3.1) , is the flood control capacity of the reservoir, m 3 ; is a coefficient, ranging from 0.5 to 1. In this example, the value is 0.8.

[0126] 3.2) , is the safe discharge flow of the reservoir, m 3 / s; is a coefficient, ranging from 0.5 to 1. In this example, the value is 0.7.

[0127] 3.3) The larger it is, the stronger the flood control function of the reservoir is.

[0128] Reservoirs that meet the requirements of indicators 3.1 and 3.2 and whose comprehensive ranking of indicator 3.3 is in the 80% are roughly screened hot reservoirs.

[0129] (4) Combined with the usage constraints of the reservoir storage and discharge sequence, the hotspot reservoirs roughly screened out in step (3) are screened again to accurately determine the hotspot reservoirs.

[0130] In step (4) of this embodiment, the principle of secondary screening is:

[0131] The index 3.1 corresponding to the hotspot reservoirs roughly screened out in step (3) is sorted from large to small, and the hotspot reservoirs ranked in the top 80% are taken as the hotspot reservoirs finally determined.

[0132] In this embodiment, the Tiancun Reservoir has no gate control and no joint optimization scheduling is performed, so it is directly classified as a non-hotspot reservoir.

[0133] Table 1 Calculation results of hotspot reservoirs

[0134]

[0135] (5) According to the rule dispatch result of step (1), it is used as the initial discharge trajectory of the hotspot reservoir POA optimization algorithm in step (4) , non-hotspot reservoirs are regularly dispatched. Refers to the discharge process of the i-th reservoir under the single reservoir rule scheduling.

[0136] The rule-based dispatching result is: the dispatching result of each reservoir participating in the dispatching according to the single reservoir dispatching rule.

[0137] The initial discharge trajectory is determined by directly using the regular dispatching results of each reservoir involved in the dispatching.

[0138] The method of regular dispatching of non-hotspot reservoirs is: the single reservoir dispatching rules corresponding to each non-hotspot reservoir.

[0139] (6) The total flood dispatch time T is divided into a certain number of time periods according to the set optimization step size. From this step, the POA algorithm is iterated to find the optimal solution. In this embodiment, the relevant settings are as follows:

[0140] The algorithm parameters that need to be set in advance include: optimization time step of 1h, number of optimization iterations of 50 times, convergence accuracy of 0.0001, etc.

[0141] The algorithm-related settings include: the total time for flood dispatching of each reservoir is 137 hours, the minimum allowable water level of each reservoir is the corresponding flood limit water level of each reservoir, the maximum allowable water level of each reservoir is the corresponding flood control high water level of each reservoir, and the final water level of each reservoir is the corresponding flood limit water level of each reservoir.

[0142] When the POA algorithm optimizes the operation of hotspot reservoirs, the objective function value of the i-th period in the j-th iteration process is .

[0143] The complete scheduling process after the nth optimization corresponds to the objective function, that is, the algorithm optimizes each time period The objective function of the scheduling process calculated by the optimal situation is .

[0144] 6.1) Divide the initial discharge trajectory of each hotspot reservoir into this time period, where the time period sequence number starts from 0;

[0145] 6.2) The regular dispatching process of non-hotspot reservoirs is divided into these time periods, and the time period numbers are sorted starting from 0.

[0146] Secure all reservoirs and , that is, the discharge flow in the 0th and 2nd periods of the initial discharge trajectory line.

[0147] (7) The discharge volume of each discrete hotspot reservoir in the first period , that is, the discharge flow between the 0th period and the 2nd period is 1 m 3 / s is the division value for division.

[0148] In the specific example, the discharge of hotspot reservoir A in period 0 is 0 m 3 / s, the discharge flow in the second period is 20 m 3 / s, then the discrete leakage flow in the first period is 0, 1, 2, , 20 m 3 / s. The discharge flow of each non-hotspot reservoir in the first period is the discharge flow of the corresponding period of regular scheduling and will not be adjusted; the discharge flow of each hotspot reservoir in the first period will be adjusted.

[0149] (8) Determine the objective function based on the flood water characteristics and calculate the objective function . Minimize the objective function value in the 0th and 1st periods, that is Optimal means that the objective function value in the first optimization calculation of the algorithm is the smallest in the 0th and 1st time periods. Refers to the objective function value of the i-th period during the j-th iteration of the algorithm;

[0150] The flood water characteristics refer to the proportion of upstream reservoir water and interval water in the flood source composition. If the upstream reservoir water: interval water>7:3, the objective function selects the maximum flood control safety guarantee criterion; if the upstream reservoir water: interval water<3:7, the objective function selects the maximum peak reduction criterion; otherwise, the objective function selects the combined weight criterion of the maximum peak reduction criterion and the maximum flood control safety guarantee criterion.

[0151] The typical flood corresponding to this embodiment is: the water flow in the upstream reservoir is small and the water flow in the interval is large. Therefore, the flood control optimization criterion is the maximum peak reduction criterion. Under the premise that the maximum water level of the reservoir does not exceed the maximum allowable water level, the smaller the peak flow of the downstream flood control section, the better. Its objective function is to minimize the combined flow of the downstream flood control section: F = my ∑ t = 1 T ∑ i = 1 N [ Q down _ i , t + Q sec_i,t ] 2 .

[0152] (9) In this way, the optimal flow rate of each period is adjusted according to the principle of optimizing the period in sequence to make the overall objective function optimal, and the flow rate of each period is advanced to period T until all periods reach the optimal state, that is, the optimal objective function value is obtained. When the time period is adjusted in step (7) When , a round of iterative calculation is completed;

[0153] (10) The optimized complete scheduling process corresponds to the objective function The objective function under all reservoir regulation is Compare and judge the optimized objective function Whether the accuracy requirement is met or the objective function is no longer improved. If both of the above judgment conditions are not met, return to step (6) to start a new round of optimization, or reach the set number of optimization iterations, the iteration stops, and the optimal solution is obtained. Refers to the objective function corresponding to the complete scheduling process after the nth optimization. Specifically refers to the objective function under the regular operation of all reservoirs without optimized operation.

[0154] The “objective function no longer improves” here means that the difference between two objective function values ​​is less than 0.0001.

[0155] The accuracy requirement is: the convergence accuracy is 0.0001.

[0156] The “number of optimization iterations” mentioned above means that the number of optimization iterations set by the POA algorithm is 50 times.

[0157] The “return to step (6) to start a new round of optimization” here means: updating the complete dispatching process of each hotspot reservoir after this optimization to the initial discharge trajectory line, and the other parameter settings remain unchanged.

[0158] (11) Output flood control optimization peak-shifting scheduling plan.

[0159] This example takes the seven large and medium-sized reservoirs from Beiwang to Dawenkou in the Dawen River Basin as the research object, and aims at the flood control problem of the downstream Dawenkou Station. While taking into account the flood control tasks of each reservoir, according to the flood characteristics, appropriate hot reservoirs and objective functions are selected, and multi-reservoir joint dimensionality reduction flood control optimization peak-shifting scheduling is carried out to achieve dimensionality reduction, effectively overcome the "curse of dimensionality", and significantly improve the calculation speed of generating optimized scheduling plans, thereby enhancing the optimization capability.

[0160] The optimization scheduling scheme in which all reservoirs participate is compared with the optimization scheduling scheme of the method of the present invention. By comparison, it can be seen that when all reservoirs participate in the optimization scheduling, the overall optimization decision variables are as many as 822, the optimization calculation time is about 15 minutes, and there is a problem that the joint optimization scheduling results of 3 reservoirs are the same as the single reservoir rule scheduling results, and there is still a large flood control storage capacity that has not been utilized, that is, there is an insufficient optimization problem; the method of the present invention can select 4 reservoirs that need to participate in the optimization, the decision variables are reduced to 482, the optimization calculation time is shortened to about 8 minutes, and the remaining flood control storage capacity is more efficiently utilized and optimized more fully.

[0161] Table 2

[0162]

[0163] Table 3

[0164]

[0165] The above-described embodiments merely express the implementation methods of the present invention, but they cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that for those skilled in the art, several modifications and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. A method for peak-shifting scheduling based on dimensionality reduction optimization of multi-reservoir groups based on hotspot reservoir selection. It is characterized in that The multi-reservoir group dimensionality reduction optimization peak-shifting dispatching method includes two stages: selecting hotspot reservoirs and optimizing dispatching. In the first stage, the hotspot reservoirs and non-hotspot reservoirs of this flood are divided by analyzing the flood control effects of each reservoir in the basin on this flood. In the second stage, an initial solution is generated according to the scheduled regulation, and a multi-reservoir joint peak-shifting optimization scheduling scheme that meets flood control safety is quickly generated through the POA algorithm; the multi-reservoir group dimensionality reduction optimization peak-shifting scheduling method adopts steps (1) to (11) to complete the output of the optimization scheduling scheme: Step (1) All reservoirs are regularly dispatched for each flood event; Step (2) Calculate the remaining flood control storage capacity of each reservoir according to the rule dispatch results of step (1) , Downflow , Comprehensive indicators of future flood control situation characteristics ; The subscript i represents the i-th reservoir, It represents the i-th time period under the rule scheduling, and its time period division is based on the conventional requirements of the field; Step (3) comprehensively considers the sizes of the three indicators obtained in step (2), determines the flood control effect of each reservoir in this flood, and roughly selects the hotspot reservoirs and non-hotspot reservoirs in this flood; Step (4) combining the use constraints of the reservoir storage and discharge sequence, performing a secondary screening on the hotspot reservoirs roughly screened in step (3) to accurately determine the hotspot reservoirs; Step (5) According to the rule dispatch result of step (1), it is used as the initial discharge trajectory of the hotspot reservoir POA optimization algorithm in step (4). , regular dispatching of non-hotspot reservoirs; among them, Refers to the discharge process of the i-th reservoir under the single reservoir rule operation; Step (6) divide the total flood dispatching time T into a certain number of time periods according to the set optimization step size, and start the POA algorithm iteration optimization; set the total flood dispatching time of each reservoir to be the same; definition: Refers to the objective function value of the i-th period in the j-th iteration process when the POA algorithm performs optimal scheduling for hotspot reservoirs; Refers to the objective function corresponding to the complete scheduling process after the nth optimization, that is, the algorithm optimizes each time period The objective function of the scheduling process generated under the optimal condition is calculated; Specifically refers to the objective function under the regular operation of all reservoirs without optimized operation; Step (7) Discretize the discharge flow of each hotspot reservoir in the first period according to the POA algorithm , that is, the discharge between the 0th period and the 2nd period of each hotspot reservoir is calculated by 1 m 3 / s is the division value for division; the discharge flow of each non-hotspot reservoir in the first period is the discharge flow of the corresponding period of regular scheduling, and no adjustment is made; the discharge flow of each hotspot reservoir in the first period is adjusted; Step (8) Determine the objective function based on the flood water characteristics and calculate the objective function ; Minimize the objective function value in the 0th and 1st periods, Optimal means that the objective function value of the first optimization calculation of the algorithm in the 0th and 1st periods is the smallest; Refers to the objective function value of the i-th period during the j-th iteration of the algorithm; Step (9) The algorithm searches for the best value in each time period until the T period. According to the principle of optimizing the time periods in sequence to optimize the overall objective function, the discharge flow of each time period is readjusted until all time periods reach the optimal state, that is, the optimal objective function value is obtained. ; When the time period is adjusted in step (7) When , a round of iterative calculation is completed, that is, the algorithm ends one round of optimization in each time period; Step (10) corresponds the optimized complete scheduling process to the objective function The objective function under all reservoir regulation is Compare and determine the corresponding objective function after optimization Whether the accuracy requirement is met or the objective function is no longer improved; if both of the above judgment conditions are not met, return to step (6) to start a new round of optimization, or reach the set number of optimization iterations, the iteration stops, and the optimal solution is obtained; Refers to the objective function corresponding to the complete scheduling process after the nth optimization. Specifically refers to the objective function under the regular operation of all reservoirs without optimized operation; Step (11) outputs a flood control optimization peak-shifting scheduling plan based on the optimal solution obtained in step (10).

2. According to claim 1, a multi-reservoir group dimensionality reduction optimization peak shifting scheduling method based on hotspot reservoir selection, It is characterized in that The principle for roughly screening hotspot reservoirs and non-hotspot reservoirs in step (3) is: Based on the existing reservoir data and flood information, three indicators are used as the main basis for judging candidate non-hotspot reservoirs, as follows: 3.1) , is the flood control capacity of the reservoir, m 3 ; is the coefficient, ranging from 0.5 to 1; 3.2) , is the safe discharge flow of the reservoir, m 3 / s; is the coefficient, and its value range is 0.5~1; 3.3) The larger it is, the stronger the flood control function of the reservoir is. Reservoirs that meet the requirements of indicators 3.1) and 3.2) at the same time and whose comprehensive ranking of indicator 3.3) is in the 80% are roughly screened hot reservoirs.

3. According to claim 1, a multi-reservoir group dimensionality reduction optimization peak shifting scheduling method based on hotspot reservoir selection, It is characterized in that In step (4), the principle of secondary screening is: 4.1) Sort the indicators 3.1 corresponding to the hot reservoirs roughly screened out in step (3) from large to small, and select the top 80% of the roughly screened hot reservoirs as the final hot reservoirs; 4.2) Combined with the previous flood information; If, among the roughly screened hotspot reservoirs, the area of ​​the previous flood in the basin controlled by a certain reservoir has not subsided, then its flood control effect on the current flood is relatively small and it will not be considered as a hotspot reservoir; 4.3) Combined with short-term rainfall forecast information; if among the roughly screened hotspot reservoirs, the period of occurrence of the next flood in the basin area controlled by a certain reservoir is similar to that of the current flood, or the rainfall center is located in the area controlled by the reservoir, then the storage capacity of the alternative hotspot reservoir will be mainly reserved for preventing the next flood. It is considered that its flood prevention effect on the current flood is relatively small, and it will not be used as a hotspot reservoir.

4. According to claim 1, a method for peak-shifting scheduling based on dimensionality reduction optimization of a multi-reservoir group based on hotspot reservoir selection, It is characterized in that In the step (5): The rule-based scheduling result is: the scheduling results of each reservoir participating in the scheduling operating according to the single-reservoir scheduling rules; The determination method of the initial discharge trajectory line is: directly using the rule-based scheduling results of each reservoir participating in the scheduling; The method for performing rule-based scheduling on non-hotspot reservoirs is: the single-reservoir scheduling rules of each non-hotspot reservoir.

5. A multi-reservoir group dimensionality reduction optimization peak-shifting scheduling method based on hotspot reservoir selection according to claim 1, characterized in that in the step (6): The parameters preset for the iterative optimization of the POA algorithm include: optimization time step, optimization iteration times, convergence accuracy; the relevant settings of the POA algorithm include: the number of flood dispatch periods for each flood event, the allowable minimum water level of each reservoir, the allowable maximum water level of each reservoir, and the final water level of each reservoir; the specific process of the iterative optimization of the POA algorithm is as follows: 6.1) Divide the initial discharge trajectory lines of each hotspot reservoir according to this time period, where the time period serial numbers start from 0 and are sorted; 6.2) Divide the rule-based scheduling process of non-hotspot reservoirs according to this time period, and the time period serial numbers start from 0 and are sorted; Secure all reservoirs and , that is, the discharge flow in the 0th and 2nd periods of the initial discharge trajectory line.

6. A multi-reservoir group dimensionality reduction optimization peak-shifting scheduling method based on hotspot reservoir selection according to claim 1, characterized in that in the step (8): The inflow characteristics of each flood event refer to: in the composition of the flood source, the proportion of the inflow from upstream reservoirs to the inflow from the river reach between two reservoirs; define the ratio P of the inflow from upstream reservoirs to the inflow from the river reach between two reservoirs: When P > P1, the objective function selects the maximum flood control safety guarantee criterion; where P1 is 7 / 3 and P2 is 3 / 7; When P < P2, the objective function selects the maximum peak shaving criterion, and P1 > P2; When P2 < P < P1, the objective function selects the combined weight criterion of the maximum peak shaving criterion and the maximum flood control safety guarantee criterion.

7. A multi-reservoir group dimensionality reduction optimization peak-shifting scheduling method based on hotspot reservoir selection according to claim 1, characterized in that in the step (8), the objective function is specifically: 8.1) The maximum flood control safety guarantee criterion is used as the flood control optimization criterion, that is, the maximum and minimum flood control storage capacity occupied by each reservoir is taken as the objective function: ; 8.2) Express the flood control optimization objective function using the maximum peak reduction criterion: ; 8.3) The combined weight criterion of the maximum peak reduction criterion and the maximum flood control safety guarantee criterion, the objective function of which is expressed as: .

8. A multi-reservoir group dimensionality reduction optimization peak-shifting scheduling method based on hotspot reservoir selection according to claim 1, characterized in that in the step (10): The "accuracy requirement" refers to: the convergence accuracy is 0.0001; The "objective function no longer improves" refers to: the difference between two objective function values is less than 0.0001; The "return to step (6) to start a new round of optimization" means: update the complete scheduling process of each hotspot reservoir after this optimization as the initial discharge trajectory line, and keep the other parameter settings unchanged.