Dynamic programming method and system for peak-shaving and valley-filling dispatching of pumped-storage power stations

Abstract

The present invention discloses a dynamic programming method and system for peak-shaving and valley-filling scheduling of a pumped-storage power station, the method comprising: extracting historical typical scenarios based on regional power grid load and output historical data, formulating typical scenarios of load and new energy output based on future development plans; taking minimizing the mean absolute deviation of the power grid's residual load as the optimization objective function, and the power generation and pumping power of the pumped-storage power station as decision variables, constructing a peak-shaving and valley-filling scheduling model for the pumped-storage power station; based on the typical prediction scenario of residual load, using a dynamic programming algorithm to optimize the power generation or pumping power process of the pumped-storage power station during the scheduling period, and compiling a scheduling plan that can smooth residual load fluctuations. The present invention can provide a scheduling plan that is adapted to the scheduling target demand of smoothing residual load fluctuations, reduce the impact of new energy output fluctuations on the power grid, improve the stability of thermal power unit operation by stabilizing the residual load process, and provide technical support for reducing regional power grid abandonment of water, wind, and light.

Description

Technical Field

[0001] The present invention belongs to the technical field of pumped storage power station dispatching, and in particular relates to a dynamic programming method and system for peak shaving and valley filling dispatching of a pumped storage power station. Background Art

[0002] Reasonable configuration of pumped storage power stations in the power system and formulation of reasonable operation plans through optimized scheduling can, on the one hand, alleviate the phenomenon of wind power and photovoltaic power generation due to the sharp increase in installed capacity. When the load is at a low level, the electricity generated by wind power and photovoltaic power generation can be stored through pumped storage power stations. When the load demand is large, the electricity stored in the pumped storage power station will be supplied to the system to fully absorb the output of new energy; on the other hand, it can effectively suppress the output fluctuation of new energy power generation, thereby stabilizing the residual load process, which is conducive to improving the stability of thermal power unit operation and equipment utilization, and ensuring the safe and stable operation of the new power system. At present, the solution of the mathematical model of pumped storage power station optimization scheduling mostly uses genetic algorithms, particle swarm algorithms and other intelligent optimization algorithms, but they have limitations in convergence speed, computational efficiency, local optimal problems, parameter tuning, etc. The dynamic programming algorithm can effectively make up for these limitations through its unique staged optimization capability, global optimal solution guarantee and parameter stability, and provides new ideas and methods for achieving stable operation of power systems and efficient utilization of new energy.

[0003] Compared with the application of intelligent optimization algorithms to solve the optimal dispatching model of pumped storage power stations, the technical difficulties and challenges of applying dynamic programming algorithms to the peak-shaving and valley-filling dispatching of pumped storage power stations include: ① The optimal dispatching of pumped storage power stations needs to consider a variety of complex constraints such as power grid power balance constraints and operating condition conversion constraints. The characteristic constraints such as operating condition conversion constraints make the decisions of adjacent time periods not completely independent, so that the problem has post-effects and does not meet the basic conditions for dynamic programming solutions. How to improve, modify and adjust the mathematical model and solution steps of the dynamic programming algorithm to make it suitable for solving the optimal dispatching problem of pumped storage power stations and ensure that the final decision combination selected in each stage meets the various constraints of the pumped storage power station. ② The optimization calculation with the mean absolute deviation as the objective function needs to be based on the known decision variables of the whole process. Due to the lack of post-effects, the traditional dynamic programming algorithm is difficult to directly optimize and calculate this type of objective function. How to process the indicator function, state transfer equation, recursive equation, etc. of the dynamic programming mathematical model so that the dynamic programming algorithm can be directly used for the optimization solution of this problem. Summary of the invention

[0004] In order to overcome the deficiencies of the above-mentioned prior art, the present invention provides a dynamic planning method and system for peak shaving and valley filling scheduling of a pumped-storage power station. According to the energy composition of the regional power grid and the operation requirements of the pumped-storage power station, typical prediction scenarios of load and renewable energy output are formulated, and a peak shaving and valley filling scheduling model for the pumped-storage power station is constructed. Based on the predicted scenarios and the constructed scheduling model, the power generation or pumping power process of the pumped-storage power station during the scheduling period is optimized to solve at least one of the above-mentioned technical problems.

[0005] According to one aspect of the present invention, a dynamic programming method for peak load shaving and valley filling scheduling of a pumped storage power station is provided, comprising:

[0006] According to the historical data of regional power grid load and output, historical typical scenarios are extracted, and based on the extracted historical typical scenarios and future development plans, typical scenarios for load and new energy output are formulated;

[0007] Taking minimizing the mean absolute deviation of the grid's residual load as the optimization objective function and the power generation and pumping power of the pumped-storage power station as the decision variables, a peak-shaving and valley-filling scheduling model for pumped-storage power stations is constructed.

[0008] Based on typical scenarios of load and renewable energy output prediction, combined with the peak shaving and valley filling scheduling model of pumped-storage power stations, a dynamic programming algorithm is used to optimize the power generation or pumping power process of the pumped-storage power station during the scheduling period, and a scheduling plan that can smooth out residual load fluctuations is formulated.

[0009] As a further technical solution, typical scenarios for load and renewable energy output prediction are proposed, including:

[0010] Analyze the energy composition of the regional power grid served by the pumped storage power station, collect long-term data of the power grid, and obtain historical data of load and output;

[0011] Based on the historical data of load and output, the fuzzy C-means clustering algorithm is used to extract the typical historical scenarios of load and renewable energy output in different months and output levels.

[0012] Based on the typical historical scenarios of load and renewable energy output and the future development plan of the regional power grid, the typical historical scenarios are adjusted to formulate predicted typical scenarios of future load and renewable energy output that are coordinated with the future development scenarios of the region.

[0013] As a further technical solution, the optimization objective function to minimize the mean absolute deviation of the residual load of the power grid is:

[0014]

[0015] Where F is the mean absolute deviation minimization objective function, T is the total number of time periods in the scheduling period; RL tis the residual load after deducting the output of various energy sources except thermal power from the original load in the tth period. The original load first deducts the output of new energy, which means that the power system gives priority to absorbing the electric energy generated by new energy into the power grid to meet the goal of giving priority to absorbing new energy; μ is the average value of the equivalent load in each period; L t is the total power load of the entire network system in period t; They represent the predicted output of wind power, photovoltaic power and hydropower in the t period respectively; is the output of the pumped storage power station at time period t, It can be positive or negative, with positive values ​​indicating power generation output and negative values ​​indicating pumping power consumption; is the regional renewable energy transmission power accepted by the power grid in period t; It is the power transmitted by renewable energy outside the region received by the power grid in period t.

[0016] As a further technical solution, when constructing a peak-shaving and valley-filling dispatching model for a pumped-storage power station, each reservoir / power station must meet the following constraints on power, water volume, and water-energy conversion:

[0017] Active power balance constraints of the entire network; power generation and pumping power restrictions of pumped-storage power stations; operating condition constraints of pumped-storage power stations; water energy conversion relationship constraints of pumped-storage power stations; water volume balance constraints of the upper and lower reservoirs of pumped-storage power stations; water storage constraints of the upper and lower reservoirs of pumped-storage power stations; and storage capacity constraints at the beginning and end of dispatching.

[0018] As a further technical solution, a dynamic programming algorithm is used to optimize the power generation or pumping power process of the pumped storage power station during the dispatch period, including: establishing a reverse recursive equation; discretizing state variables and judging the feasibility of decisions, determining the allowable decision set; and performing reverse recursive calculations.

[0019] According to one aspect of the present invention, a dynamic programming system for peak load shaving and valley filling scheduling of a pumped storage power station is provided, comprising:

[0020] The scenario extraction module is used to extract historical typical scenarios based on the historical data of regional power grid load and output, and formulate the predicted typical scenarios of load and new energy output based on the extracted historical typical scenarios and future development plans;

[0021] A model building module is used to build a peak-shaving and valley-filling dispatching model for a pumped-storage power station, taking minimizing the mean absolute deviation of the grid's residual load as the optimization objective function and the power generation and pumping power of the pumped-storage power station as the decision variables;

[0022] The optimization solution module is used to predict typical scenarios based on load and renewable energy output, combined with the peak shaving and valley filling scheduling model of pumped storage power stations, and uses dynamic programming algorithms to optimize the power generation or pumping power process of pumped storage power stations during the scheduling period, and compile a scheduling plan that can smooth out residual load fluctuations.

[0023] As a further technical solution, the scene extraction module is further used to execute the following instructions:

[0024] Analyze the energy composition of the regional power grid served by the pumped storage power station, collect long-term data of the power grid, and obtain historical data of load and output;

[0025] Based on the historical data of load and output, the fuzzy C-means clustering algorithm is used to extract the typical historical scenarios of load and renewable energy output in different months and output levels.

[0026] Based on the typical historical scenarios of load and renewable energy output and the future development plan of the regional power grid, the typical historical scenarios are adjusted to formulate predicted typical scenarios of future load and renewable energy output that are coordinated with the future development scenarios of the region.

[0027] As a further technical solution, the optimization solution module further includes:

[0028] The first submodule is used to establish a reverse recursive equation;

[0029] The second submodule is used to discretize state variables and determine the feasibility of decisions, and determine the set of permissible decisions;

[0030] The third submodule is used for reverse recursive calculation;

[0031] The fourth submodule is used to prepare a scheduling plan based on the reverse recursive calculation results.

[0032] According to one aspect of the present invention specification, there is provided an electronic device, which is configured in a pumped-storage power station, and includes a memory and a processor, wherein the memory is used to store a computer program, and when the processor runs the computer program stored in the memory, the processor executes the dynamic programming method for peak shaving and valley filling scheduling of the pumped-storage power station.

[0033] According to one aspect of the present specification, there is provided a non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions enable the computer to execute the dynamic programming method for peak shaving and valley filling scheduling of a pumped-storage power station.

[0034] Compared with the prior art, the present invention has the following beneficial effects:

[0035] 1. The present invention constructs a perfect dynamic programming mathematical model for peak-shaving and valley-filling scheduling of pumped-storage power stations that is adapted to the scheduling target of stabilizing the residual load of the regional power grid. By processing the objective function, the inconvenience of optimizing and solving the mean absolute deviation objective function caused by the lack of after-effect of the dynamic programming algorithm is solved. By limiting the allowable decision range, the decision process solved by the algorithm satisfies various constraints in each time period;

[0036] 2. The present invention constructs a peak-shaving and valley-filling dispatching model for a pumped-storage power station, and applies a dynamic programming algorithm to optimize and solve the dispatching model. The algorithm is highly flexible and does not require a complex parameter adjustment process. It not only reduces the difficulty of solving the problem, but also reflects the dynamic evolution of the smooth residual load curve of the pumped-storage power station. By searching for decision combinations in all stages through dynamic programming, it avoids falling into the situation of local optimal solutions, and improves the calculation efficiency and global optimization capability. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, a brief introduction is given below to the drawings used in the 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.

[0038] Figure 1 A schematic flow chart of a dynamic programming method for peak shaving and valley filling scheduling of a pumped storage power station provided in an embodiment of the present invention.

[0039] Figure 2 A schematic diagram of a dynamic programming algorithm solution model provided in an embodiment of the present invention.

[0040] Figure 3 A schematic diagram of typical scenarios of residual load in various seasons provided for an embodiment of the present invention.

[0041] Figure 4 An optimized dispatching scheme for a pumped-storage power station under different typical residual load scenarios in each season provided by an embodiment of the present invention, and a schematic diagram of the residual load process before and after optimization.

[0042] Figure 5 A schematic diagram of the structure of a dynamic programming system for peak shaving and valley filling scheduling of a pumped storage power station provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0043] The terms "including" and "having" and any variations thereof in the specification and claims of the present invention and the above-mentioned drawings are intended to cover non-exclusive inclusions, for example, a process, method, system, product or apparatus comprising a series of steps or units is not necessarily limited to the steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products or apparatuses.

[0044] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the 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 in the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. In addition, the technical features in the various embodiments or single embodiments provided by the present invention are arbitrarily combined with each other to form a new technical solution. This combination is not restricted by the sequence of steps and / or the structural composition mode, but must be based on the ability of ordinary technicians in this field to achieve. When the combination of technical solutions is contradictory or cannot be achieved, it should be considered that this combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0045] According to the historical data of regional power grid load and output, historical typical scenarios are extracted, and based on the extracted historical typical scenarios and future development plans, typical scenarios for load and new energy output are formulated;

[0046] Taking minimizing the mean absolute deviation of the grid's residual load as the optimization objective function and the power generation and pumping power of the pumped-storage power station as the decision variables, a peak-shaving and valley-filling scheduling model for pumped-storage power stations is constructed.

[0047] Based on typical scenarios of load and renewable energy output prediction, combined with the peak shaving and valley filling scheduling model of pumped-storage power stations, a dynamic programming algorithm is used to optimize the power generation or pumping power process of the pumped-storage power station during the scheduling period, and a scheduling plan that can smooth out residual load fluctuations is formulated.

[0048] like Figure 1 As shown, the present invention provides a dynamic programming method for peak shaving and valley filling scheduling of a pumped storage power station, comprising the following steps:

[0049] Step 1: Extract historical typical scenarios based on the historical data of regional power grid load and output, and formulate predicted typical scenarios for load and renewable energy output based on the extracted historical typical scenarios and future development plans.

[0050] First, the energy composition of the regional power grid served by the HMF pumped storage power station is analyzed, and the long series of data of the power grid are collected, including historical data such as actual total power load, actual hydropower output, actual wind power output, actual photovoltaic output, actual biomass output, internal network input new energy output and external network input new energy output, etc., to obtain a load and output data set with equal time intervals for k consecutive months (seasons), n days per month, and b times per day, such as a load and output data set with 96 equal time intervals for 12 consecutive months and every day, totaling 245280 data (7 factors * 365 days * 96 time periods). For each type of data, the data set for each month (season) is recorded as X = {X1, X2, ...X i , ..., X n}, where X i represents the output data of the i-th day of the month, X i =[x i，1 , x i，2 , ..., x i，m ] T , n is the total number of days in the month.

[0051] Based on the historical data of load and output, the fuzzy C-means clustering algorithm is used to extract typical historical scenarios of load and renewable energy output in different seasons and output levels. The data set X of a certain month is divided into c categories, where the set of cluster centers can be expressed as V = {V1, V2, ..., V c}.

[0052] Specifically, the data set X of each month is divided into three categories: high, medium, and low according to the load or output value, where the set of cluster centers can be expressed as V = {V1, V2, V3}. In fuzzy partitioning, a certain membership value is used to describe each data object as belonging to a certain category, without strictly dividing it into a certain category. The i-th data sample X in the data set X i The membership degree μ of the jth class ij It is expressed as the following numerical relationship:

[0053]

[0054] For the fuzzy C-means clustering algorithm, the essence of clustering solution is to solve the minimum value of the following objective function on the basis of satisfying the constraints in the above formula:

[0055]

[0056] Membership degree μ ij Calculated by the following formula:

[0057]

[0058] Cluster center Vj Calculated by the following formula:

[0059]

[0060] In the formula, F(X, V) is the objective function of iterative calculation, which represents the weighted sum of square distances from the data sample to the cluster center, and the weight is the data sample X i The membership degree μ of the jth class ij f to the power of f; f is the smoothing factor (fuzzy weighting parameter), f∈[1,+∞), in general calculations, it is taken as 2; d ij Represents the Euclidean distance from the i-th data sample to the j-th cluster center.

[0061] The specific process of clustering and dividing the monthly load or output data using the fuzzy C-means clustering algorithm is as follows:

[0062] (1) Set the number of clusters c = 3, the number of iterations k = 0, the maximum number of iterations K = 20, the termination error ε, and initialize the cluster center V 0 ;

[0063] (2) Use the membership calculation formula to calculate each membership value U k ;

[0064] (3) Use the cluster center calculation formula to update the next cluster center V k+1 ;

[0065] (4) Based on the updated cluster center, the membership value U is updated by the membership calculation formula k+1 ;

[0066] (5)If||U k+1 -U k ||<ε, end the calculation, otherwise, go to step (6);

[0067] (6) Let k = k + 1, and determine whether k > K. If so, end the calculation and output the clustering result; otherwise, return to step (2).

[0068] For the cluster calculation results of each season (month), after clustering, it is necessary to select the output curve of a certain day in each category as the typical scene of that category, which together constitute the typical scene set of load or output in the region. When selecting the typical scene in each type, we must first take the average of the output power of all historical days belonging to that category. Then, calculate the distance between the output power of each historical day and the average power, select the output curve of the closest historical day as the typical scene of this type, and obtain the typical scene set of the power grid load and new energy output in the region in turn. Finally, for each load and output factor, typical historical daily scenes of load and new energy output at three numerical levels of each season, high, medium and low are obtained, totaling 12 typical daily scenes.

[0069] Based on the typical historical scenarios of load and renewable energy output and the future development plan of the regional power grid, the typical historical scenarios are adjusted to formulate typical scenarios of future load and renewable energy output that are coordinated with the future development scenarios of the region. The residual load is obtained by subtracting the renewable energy output value from the load value of each period. The typical scenarios of residual load in each season are shown in Figure 3 ,Scenario 1, Scenario 2 and Scenario 3 shown in the figure represent typical residual load scenarios at low, medium and high numerical levels, respectively.

[0070] Step 2: Taking minimizing the average absolute deviation of the grid's residual load as the optimization objective function and the power generation and pumping power of the pumped-storage power station as the decision variables, a peak-shaving and valley-filling scheduling model for the pumped-storage power station is constructed.

[0071] The average absolute deviation is used to evaluate the fluctuation degree of load data, and the objective function of peak shaving and valley filling scheduling is selected to minimize the average absolute deviation of the remaining load of the power grid:

[0072]

[0073] Where: F is the mean absolute deviation minimization objective function, T is the total number of time periods in the scheduling period; RL t is the residual load after deducting the output of various energy sources except thermal power from the original load in the tth period. The original load first deducts the output of new energy, which means that the power system gives priority to absorbing the electric energy generated by new energy into the power grid to meet the goal of giving priority to absorbing new energy; μ is the average value of the equivalent load in each period; L t is the total power load of the entire network system in period t; They represent the predicted output of wind power, photovoltaic power and hydropower in the t period respectively; is the output of the pumped storage power station at time period t, It can be positive or negative, with positive values ​​indicating power generation output and negative values ​​indicating pumping power consumption; is the regional renewable energy transmission power accepted by the power grid in period t; It is the power transmitted by renewable energy outside the region received by the power grid in period t.

[0074] When establishing the dispatching and operation model of pumped storage power stations, the power generation and pumping power of pumped storage power stations are taken as the As decision variables, the peak shaving and valley filling and new energy consumption functions of the pumped-storage power station are considered, and the daily planning curve (daily power generation process) of the pumped-storage power station is derived.

[0075] Each reservoir / power station in step 2 must meet the following constraints on power, water volume, and water-energy conversion relationship:

[0076] (1) Active power balance constraints of the entire network:

[0077]

[0078] (2) Power generation and pumping capacity limits of pumped storage power stations:

[0079] ① When When the pumped storage power station operates in power generation mode, the power range is:

[0080]

[0081] In the formula, and They represent the minimum and maximum power values ​​of the pumped storage power station under the power generation condition in the tth period respectively;

[0082] ② When When , the pumped storage power station operates in pumping condition. The pumping power point of the pumping power station in pumping condition is discrete, and the mathematical expression is as follows:

[0083]

[0084] In the formula, and They represent the minimum and maximum power values ​​of the pumped-storage power station under the pumping condition in the tth period respectively;

[0085] (3) Pumped storage power station operating constraints:

[0086] Compared with conventional hydropower stations, pumped storage power stations have multiple operating conditions and there are constraints on the conversion between operating conditions. For the convenience of expression, a 0-1 variable is introduced, s g，t =1 means the power station is in power generation operation at the tth period, s g，t =0 means the power station is in a non-power generation state during this period, s p，t =1 means the power station is in pumping operation at time period t, s p，t =0 means that the power station is in non-pumping operation during this period;

[0087] From the overall perspective of the power station, the following constraints are generally considered:

[0088] ① The power station cannot be in power generation and pumping operation at the same time:

[0089] s g，t +s p，t ≤1

[0090] ② The switching between power generation operation and pumping operation of the power station shall be separated by at least one period:

[0091]

[0092] (4) Water energy conversion relationship of pumped storage power station:

[0093]

[0094] In the formula, K is the output coefficient, Q t represents the power station flow in period t, Q t >0 means water flows from the upper reservoir to the lower reservoir and the power station is in power generation operation. t Indicates the generating head; Q t <0 means water flows from the lower reservoir to the upper reservoir and the power station is in pumping operation. t Indicates the pumping head; η g and η p They represent the efficiency of the power station when generating electricity and when pumping water respectively;

[0095] The calculation formula for head and lift is as follows:

[0096]

[0097] In the formula, and Respectively represent the water levels of the upper reservoir and the lower reservoir at the beginning of the t period; Δh t Indicates the head loss in the pipe.

[0098] (5) Water balance between upper and lower reservoirs of pumped storage power stations:

[0099] Without considering other inflow supplements to the upper and lower reservoirs,

[0100]

[0101] In the formula, and They represent the water storage capacity (storage capacity) of the upper and lower reservoirs at the beginning of the tth period respectively; Δt is the length of the period, which is 15 minutes.

[0102] (6) Constraints on water storage capacity of upper and lower reservoirs of pumped storage power stations:

[0103]

[0104] In the formula, and Respectively represent the minimum and maximum water storage capacity of the upper reservoir (lower reservoir) at the beginning of the t period; V E Indicates the reserved emergency reserve storage capacity;

[0105] (7) Storage capacity constraints at the beginning and end of dispatch:

[0106] The storage capacity V1 at the beginning of the scheduling period is given in advance, and the storage capacity V at the end of the scheduling period is T+1 is also given. According to the scheduling arrangement, the constraint can be expressed as:

[0107]

[0108] Where δ is the allowable deviation rate.

[0109] Step three: Based on the typical scenarios of load and renewable energy output forecast, combined with the peak shaving and valley filling scheduling model of pumped-storage power stations, a dynamic programming algorithm is used to optimize the power generation or pumping power process of the pumped-storage power station during the scheduling period, and a scheduling plan that can smooth out residual load fluctuations is prepared.

[0110] Based on the predicted typical day scenario of load and renewable energy output extracted in step 1, and according to the peak shaving and valley filling dispatching model constructed in step 2, the dynamic programming algorithm is used to optimize the power generation or pumping power process of the pumped storage power station in each period during the dispatching period. The specific steps are as follows:

[0111] (1) Establishment of the reverse recursive equation

[0112] ① Stages and stage variables: The T time periods divided by the scheduling period are regarded as T stages, and t represents the stage variable (t = 1 to T). Then t is the facing stage, and t + 1 to T is the remaining period.

[0113] ②State variable: Select the initial reservoir water storage capacity at each stage As the state variable of the tth stage. From the scheduling and final storage capacity constraints considered in step 2, it can be seen that the initial state of the first stage and the final state of the Tth stage have only one value, so:

[0114]

[0115] ③Decision variables: Select the power generation or pumping power of the pumped storage power station stage is the decision variable. The feasible range of changes constitutes the allowed decision set D t ,but The sub-strategy is recorded as: Then u T，1 Represents a strategy for the entire process.

[0116] ④State transfer equation: The reservoir water balance equation is the state transfer equation, as shown below:

[0117]

[0118] ⑤Indicator function and optimal value function:

[0119] The objective function of the peak-shaving and valley-filling dispatching model of pumped storage power stations is:

[0120]

[0121] The absence of aftereffect means that the optimality of a decision does not depend on the results of future decisions. However, the mean absolute deviation is a statistic of the whole process. Its calculation needs to consider the results of the entire decision sequence and is affected by subsequent decisions. Therefore, the mean absolute deviation as an objective function does not meet the absence of aftereffect, and the minimum mean absolute deviation of the remaining load cannot be directly used as the objective function of dynamic programming.

[0122] Since the storage capacity of the reservoir at the beginning and end of the dispatching period is given, during the dispatching period, the pumped storage power station completes a cycle of work, the water level of the upper reservoir rises from the initial water level to the normal water level, and then drops from the normal water level to the water level at the end of the dispatching period, and the water level of the lower reservoir drops from the initial water level to the minimum operating water level (dead water level), and then rises from the minimum operating water level to the water level at the end of the dispatching period. Ignoring the different head calculation methods under different working conditions, the nonlinear relationship between head loss and flow, and the influence of efficiency coefficient on power calculation in each period, assuming that the pumped storage power station only completes one cycle during the dispatching period, the sum of the power of each stage during the dispatching period of the pumped storage power station can be approximately regarded as a constant, calculated by the following formula:

[0123]

[0124] In the formula, and They represent the water storage capacity of the upper reservoir at the beginning and end of the dispatching period respectively; and Respectively represent the upper reservoir (lower reservoir) water level at the beginning and end of the dispatch period; and They represent the normal water level of the upper reservoir and its corresponding storage capacity respectively; is the dead water level of the lower reservoir; H g and H p They represent the power generation head and pumping head respectively; Δh is the head loss; η g and η p They represent the power station's efficiency when generating electricity and when pumping water, respectively.

[0125] In the pumped storage power station optimization scheduling model, the load demand and the output of other energy sources in each period are given values. After the sum of the output values ​​of the pumped storage power station in each period is obtained, the mean value of the residual load in the whole period can be regarded as a constant and calculated by the following formula:

[0126]

[0127] The objective function can finally be simplified as:

[0128]

[0129] In the formula, It represents the original residual load in stage t, that is, the residual load when no pumped storage power station is set up.

[0130] The index function is used to measure the benefits obtained by the pumped storage power station after making a certain decision at a certain stage. The index function of the tth stage is:

[0131]

[0132] In the formula, Indicates that the initial state of stage t is The decision variables are , the index function value of the pumped storage power station in the tth stage.

[0133] The corresponding optimal value function can be obtained from the indicator function, that is, the optimal value of the sum of the indicator function values ​​in the T~t stage:

[0134]

[0135] In the formula, Indicates that the initial state of stage t is By seeking the optimal sub-strategy u T，t , so that the pumped storage power station can obtain the optimal objective function value in the T~t period.

[0136] ⑥ Recursive equation: According to the multi-stage decision-making principle, index function and optimal value function, the following reverse recursive equation can be listed:

[0137]

[0138] In the formula, In The initial state of stage t is When making decisions After that, the water storage at the end of stage t (beginning of stage t+1).

[0139] ⑦ Constraints: The constraints that need to be considered in the dynamic programming recursive calculation have been given in step 2.

[0140] (2) Discretization of state variables and decision feasibility judgment

[0141] According to the water storage capacity constraint of the upper reservoir of the pumped storage power station, the state of any stage can be Discrete into M values, that is,

[0142] The operating condition conversion constraints of pumped storage units make the optimization scheduling problem of pumped storage have aftereffects, that is, the decision at the current moment is affected by the decisions at the previous moment, and the optimal decision at the current moment depends on the past state and decisions. In order to eliminate this aftereffect, the decision space is restricted to only include the feasible decisions of the current state, so as to ensure that the optimal solution of each sub-problem depends only on the current state and is not affected by the past state and decisions.

[0143] Before the calculation of the tth stage (t=2~T-1), the initial states are obtained through the optimization calculation of the t+1th stage. The optimal value function under And record the optimal decision (output) made in each initial state at this stage

[0144] For the optimization calculation of the tth stage, when looking for a certain initial state In the process of the optimal value function and optimal strategy under With each final state Combine and calculate the decision value under each combination Considering the operating condition conversion constraints of the pumped storage power station, the final state The corresponding optimal decision in stage t+1 Will limit the range of decisions allowed at stage t D t :

[0145] ① When , the power station operates in the power generation state in the t+1 stage, so it can only continue to generate electricity or shut down in the t stage. The allowed decision range is:

[0146]

[0147] ② When , the power station operates in pumping mode in stage t+1, so it can only continue pumping or shut down in stage t. The allowed decision range is:

[0148]

[0149] ③ When , the power station is shut down in stage t+1, so it can generate electricity, pump water or shut down in stage t, and the allowed decision range is:

[0150]

[0151] Therefore, in the initial state With the final state When combining, you need to judge the decision under the combination Is it in the final state? The corresponding allowable decision range D t When D t When is the interval range, If it falls within the interval, it means that the decision is feasible; t When is a discrete value set, calculate The error between the decision value and a discrete value closest to the decision value is within the allowable error range, and the decision value is considered to be within the allowable decision range. Not in D t If the value is within the range, it means that the decision is not feasible and the initial and final state combination is discarded.

[0152] Considering that the pumping power of the power station can only take discrete values ​​under pumping conditions, D t may contain a discrete set of values, in which case the calculation The error with a discrete value closest to the decision value, if the error value is within the allowable error range, then the decision value is considered to be within the allowable decision range D t Inside. like Not in D t If the decision is within the allowed decision range D, it means that the decision is not feasible and the initial and final state combination is discarded. In order to avoid selecting infeasible decision combinations during optimization calculations at each stage, the penalty function idea is introduced. t If the value is within the range of , then the stage benefit of this stage is set to a value that is significantly different from the objective function.

[0153] (3) Reverse recursive calculation

[0154] ① T state calculation

[0155] There is only one state at the end of this stage, namely The initial state of the stage is M discrete values From the recursive equation, we can get the optimal state value at the end of each initial state period is The optimal function values ​​are

[0156] ② Calculation of the T-1~2 stage

[0157] Before the calculation of the tth stage (t=2~T-1), the initial states are obtained through the optimization calculation of the t+1th stage. The optimal value function under

[0158] For the optimization calculation of the tth stage, when looking for a certain initial state In the process of the optimal value function and optimal strategy under With each final state Combine them and calculate the stage decision value under each combination by the following formula:

[0159]

[0160] In the formula, The initial state is The final state is The decision value at stage t.

[0161] Calculate the decision value of each combination And M indicator functions are calculated by the decision value For any one Using the recursion equation By searching for the optimal function, we can get the corresponding optimal function value. Optimal final state value and optimal decision

[0162] ③ Phase 1 calculation

[0163] The initial state of stage 1 has only one value Using the recursion equation By searching for the optimal function value, we can get Optimal final state value and optimal decision

[0164] That is, the optimal value of the objective function within the dispatch period T can be obtained under the premise of satisfying various constraints. Based on the above recursive calculation results, the optimal state process of the whole process composed of the states of each stage (the change process of the water storage capacity of the upper reservoir) and the optimal strategy of the whole process (the power of the pumped storage power station in each period) can be obtained. That is, the daily dispatch plan of the pumped-storage power station.

[0165] The number of discrete states of the dynamic programming algorithm is set to 1000, and the maximum error between the pumping power decision value and the specified discrete value is set to 3% of the specified value. The optimal dispatching scheme of the pumped storage power station under different typical residual load scenarios in each season is obtained by calculation. The residual load process before and after optimization is shown in Figure 4 .

[0166] from Figure 4 It can be seen that through the optimized dispatch of pumped storage power stations, the optimized residual load process curves of each residual load scenario in each season are smoother and have smaller fluctuations, indicating that the volatility of the residual load has been effectively reduced. The pumped storage power station has fully played the role of peak shaving and valley filling. At the peak of the daily residual load curve, the pumped storage power station generates electricity and releases the previously stored water energy to reduce the peak load; at the trough of the load curve, the pumped storage power station uses excess electricity to pump water to the high-level reservoir to store energy, effectively increasing the residual load of the power grid during the trough period.

[0167] The implementation basis of each embodiment of the present invention is to implement programmed processing through a device with a processor function. Therefore, in engineering practice, the technical solutions and functions of each embodiment of the present invention are encapsulated into various modules. Based on this reality, on the basis of the above embodiments, an embodiment of the present invention provides a dynamic programming system for peak shaving and valley filling scheduling of a pumped storage power station, and the device is used to execute the dynamic programming method for peak shaving and valley filling scheduling of a pumped storage power station in the above method embodiment.

[0168] See also Figure 5 , the system comprises:

[0169] The dynamic planning system for peak-shaving and valley-filling scheduling of a pumped-storage power station provided by the embodiment of the present invention adopts Figure 5 Several modules in the system are used to construct a peak-shaving and valley-filling dispatching model for a pumped-storage power station, and a dynamic programming algorithm is applied to optimize and solve the dispatching model. The algorithm is highly flexible and does not require a complex parameter adjustment process. It not only reduces the difficulty of solving the problem, but also reflects the dynamic evolution of the smooth residual load curve of the pumped-storage power station. The embodiment of the present invention also avoids falling into the local optimal solution by searching for the decision combination of all stages through dynamic programming, thereby improving the computing efficiency and global optimization capability.

[0170] It should be noted that the system embodiment provided by the present invention is used to implement the methods in the above method embodiment as well as the methods in other method embodiments provided by the present invention. The only difference is that the corresponding functional modules are set. The principle is basically the same as the principle of the above system embodiment provided by the present invention. As long as the technical personnel in this field refer to the specific technical solutions in other method embodiments on the basis of the above system embodiment, obtain the corresponding technical means and the technical solutions composed of these technical means by combining the technical features, and on the premise of ensuring the practicality of the technical solutions, improve the modules in the above system embodiment to obtain the corresponding system class embodiments, which are used to implement the methods in other method class embodiments. For example:

[0171] Based on the content of the above system embodiment, as a preferred embodiment, a dynamic planning system for peak shaving and valley filling scheduling of a pumped storage power station provided in an embodiment of the present invention, the scenario extraction module is further used to execute the following instructions:

[0172] Analyze the energy composition of the regional power grid served by the pumped storage power station, collect long-term data of the power grid, and obtain historical data of load and output;

[0173] Based on the historical data of load and output, the fuzzy C-means clustering algorithm is used to extract the typical historical scenarios of load and renewable energy output in different months and output levels.

[0174] Based on the typical historical scenarios of load and renewable energy output and the future development plan of the regional power grid, the typical historical scenarios are adjusted to formulate predicted typical scenarios of future load and renewable energy output that are coordinated with the future development scenarios of the region.

[0175] Based on the content of the above system embodiment, as a preferred embodiment, a dynamic programming system for peak shaving and valley filling scheduling of a pumped storage power station provided in an embodiment of the present invention, the optimization solution module also includes:

[0176] The first submodule is used to establish a reverse recursive equation;

[0177] The second submodule is used to discretize state variables and determine the feasibility of decisions, and determine the set of permissible decisions;

[0178] The third submodule is used for reverse recursive calculation;

[0179] The fourth submodule is used to prepare a scheduling plan based on the reverse recursive calculation results.

[0180] An embodiment of the present invention also provides an electronic device, which is configured in a pumped-storage power station, including a memory and a processor, wherein the memory is used to store a computer program, and when the processor runs the computer program stored in the memory, the processor executes the dynamic programming method for peak shaving and valley filling scheduling of the pumped-storage power station.

[0181] When the electronic device based on the embodiment of the present invention performs dynamic planning of peak shaving and valley filling scheduling of a pumped-storage power station, typical scenarios of load and renewable energy output are extracted and formulated based on the energy composition of the regional power grid and the operation requirements of the pumped-storage power station, the power grid load, historical output data, and future development plans; a peak shaving and valley filling scheduling model for a pumped-storage power station is constructed by optimizing the objective function of minimizing the mean absolute deviation of the residual load of the power grid, and the power generation and pumping power of the pumped-storage power station as decision variables; based on typical historical scenarios and predicted scenarios of the residual load, a dynamic programming algorithm is used to optimize the power generation or pumping power process of the pumped-storage power station during the scheduling period, and a scheduling plan that can smooth the fluctuations of the residual load of the power grid is compiled.

[0182] The embodiment of the present invention further provides a non-transitory computer-readable storage medium, wherein the non-transitory computer-readable storage medium stores computer instructions, wherein the computer instructions enable the computer to execute the dynamic programming method for peak shaving and valley filling scheduling of a pumped storage power station, including:

[0183] According to the historical data of regional power grid load and output, historical typical scenarios are extracted, and based on the extracted historical typical scenarios and future development plans, typical scenarios for load and new energy output are formulated;

[0184] Taking minimizing the mean absolute deviation of the grid's residual load as the optimization objective function and the power generation and pumping power of the pumped-storage power station as the decision variables, a peak-shaving and valley-filling scheduling model for pumped-storage power stations is constructed.

[0185] Based on typical scenarios of load and renewable energy output prediction, combined with the peak shaving and valley filling scheduling model of pumped-storage power stations, a dynamic programming algorithm is used to optimize the power generation or pumping power process of the pumped-storage power station during the scheduling period, and a scheduling plan that can smooth out residual load fluctuations is formulated.

[0186] In summary, the present invention can provide a scheduling scheme that is adapted to the scheduling target demand of smoothing residual load fluctuations, reduce the impact of new energy output fluctuations on the power grid, improve the stability of thermal power unit operation by stabilizing the residual load process, and provide technical support for reducing regional power grid abandonment of water, wind and solar power.

[0187] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the technical solutions of the embodiments of the present invention.

Claims

1. A dynamic programming method for peak load shaving and valley filling scheduling of a pumped storage power station, characterized in that: include: According to the historical data of regional power grid load and output, historical typical scenarios are extracted, and based on the extracted historical typical scenarios and future development plans, typical scenarios for load and new energy output are formulated; Taking minimizing the mean absolute deviation of the grid's residual load as the optimization objective function and the power generation and pumping power of the pumped-storage power station as the decision variables, a peak-shaving and valley-filling scheduling model for pumped-storage power stations is constructed. Based on typical scenarios of load and renewable energy output prediction, combined with the peak shaving and valley filling scheduling model of pumped-storage power stations, a dynamic programming algorithm is used to optimize the power generation or pumping power process of the pumped-storage power station during the scheduling period, and a scheduling plan that can smooth the fluctuation of residual load is compiled; among them, the dynamic programming algorithm is used to optimize the power generation or pumping power process of the pumped-storage power station during the scheduling period, including: establishing a reverse recursive equation; discretizing state variables and judging the feasibility of decisions, determining the allowable decision set; and performing reverse recursive calculations.

2. The dynamic programming method for peak-shaving and valley-filling scheduling of a pumped-storage power station according to claim 1 is characterized in that: Propose typical scenarios for forecasting load and renewable energy output, including: Analyze the energy composition of the regional power grid served by the pumped storage power station, collect long-term data of the power grid, and obtain historical data of load and output; Based on the historical data of load and output, the fuzzy C-means clustering algorithm is used to extract typical historical scenarios of load and renewable energy output in different months and output levels. Based on the typical historical scenarios of load and renewable energy output and the future development plan of the regional power grid, the typical historical scenarios are adjusted to formulate predicted typical scenarios of future load and renewable energy output that are coordinated with the future development scenarios of the region.

3. The dynamic programming method for peak-shaving and valley-filling scheduling of a pumped-storage power station according to claim 1 is characterized in that: The optimization objective function to minimize the mean absolute deviation of the residual load of the power grid is: Where F is the mean absolute deviation minimization objective function, T is the total number of time periods in the scheduling period; RL t is the residual load after deducting the output of various energy sources except thermal power from the original load in the tth period. The original load first deducts the output of new energy, which means that the power system gives priority to absorbing the electric energy generated by new energy into the power grid to meet the goal of giving priority to absorbing new energy; μ is the average value of the equivalent load in each period; L t is the total power load of the entire network system in period t; They represent the predicted output of wind power, photovoltaic power and hydropower in the t period respectively; is the output of the pumped storage power station at time period t, It can be positive or negative, with positive values ​​indicating power generation output and negative values ​​indicating pumping power consumption; is the regional renewable energy transmission power accepted by the power grid in period t; It is the power transmitted by renewable energy outside the region received by the power grid in period t.

4. The dynamic programming method for peak-shaving and valley-filling scheduling of a pumped-storage power station according to claim 3 is characterized in that: When constructing a peak-shaving and valley-filling dispatching model for a pumped-storage power station, each reservoir / power station must meet the following constraints on power, water volume, and water-energy conversion: Active power balance constraints of the entire network; power generation and pumping power restrictions of pumped-storage power stations; operating condition constraints of pumped-storage power stations; water energy conversion relationship constraints of pumped-storage power stations; water volume balance constraints of the upper and lower reservoirs of pumped-storage power stations; water storage constraints of the upper and lower reservoirs of pumped-storage power stations; and storage capacity constraints at the beginning and end of dispatching.

5. A dynamic programming system for peak load shaving and valley filling scheduling of pumped storage power stations, characterized in that: include: The scenario extraction module is used to extract historical typical scenarios based on the historical data of regional power grid load and output, and formulate the predicted typical scenarios of load and new energy output based on the extracted historical typical scenarios and future development plans; A model building module is used to build a peak-shaving and valley-filling dispatching model for a pumped-storage power station, taking minimizing the mean absolute deviation of the grid's residual load as the optimization objective function and the power generation and pumping power of the pumped-storage power station as the decision variables; The optimization solution module is used to optimize the power generation or pumping power process of the pumped storage power station during the scheduling period based on typical scenarios of load and renewable energy output prediction, combined with the peak-shaving and valley-filling scheduling model of the pumped storage power station, and prepare a scheduling plan that can smooth the fluctuation of the residual load; The optimization solution module also includes: a first submodule, which is used to establish a reverse recursive equation; a second submodule, which is used to discretize state variables and judge the feasibility of decisions, and determine the allowable decision set; a third submodule, which is used to perform reverse recursive calculations; and a fourth submodule, which is used to prepare a scheduling plan based on the reverse recursive calculation results.

6. The dynamic programming system for peak-shaving and valley-filling scheduling of a pumped-storage power station according to claim 5 is characterized in that: The scene extraction module is also used to execute the following instructions: Analyze the energy composition of the regional power grid served by the pumped storage power station, collect long-term data of the power grid, and obtain historical data of load and output; Based on the historical data of load and output, the fuzzy C-means clustering algorithm is used to extract typical historical scenarios of load and renewable energy output in different months and output levels. Based on the typical historical scenarios of load and renewable energy output and the future development plan of the regional power grid, the typical historical scenarios are adjusted to formulate predicted typical scenarios of future load and renewable energy output that are coordinated with the future development scenarios of the region.

7. An electronic device, arranged in a pumped storage power station, characterized in that: It comprises a memory and a processor, the memory is used to store a computer program, and when the processor runs the computer program stored in the memory, the processor executes the dynamic programming method for peak shaving and valley filling scheduling of a pumped storage power station as described in any one of claims 1 to 4.

8. A non-transitory computer readable storage medium, characterized in that: The non-transitory computer-readable storage medium stores computer instructions, and the computer instructions enable the computer to execute the dynamic programming method for peak shaving and valley filling scheduling of a pumped-storage power station as described in any one of claims 1 to 4.

Images