Hybrid pumped storage power station intra-plant unit scheduling strategy fast solving method, device and medium based on approximate dynamic programming

Abstract

This invention relates to the field of pumped-storage unit operation optimization technology, and discloses a method, equipment, and medium for rapidly solving the scheduling strategy of hybrid pumped-storage power stations based on approximate dynamic programming. The method first establishes an in-plant scheduling model including multiple types of units; then, it reconstructs the model based on a Markov decision process, defining the virtual energy storage state of charge generated by a linear mapping of the upper and lower reservoir capacities as the core state variable; subsequently, it constructs a piecewise linear convex function with this state as the independent variable to approximate the function; the slope parameter of this function is trained offline during the day-ahead phase; during the intraday real-time phase, the trained function is invoked according to the real-time system state to rapidly solve the Bellman equation and output the optimal scheduling strategy. This invention solves the problems of high dimensionality and slow solution of in-plant unit scheduling models through virtual energy storage modeling and an "offline training-online application" mechanism, achieving real-time rapid decision-making while ensuring accuracy.

Description

Technical Field

[0001] This invention relates to the field of pumped-storage unit operation optimization technology, and in particular to a method, equipment and medium for rapidly solving the unit scheduling strategy of a hybrid pumped-storage power station based on approximate dynamic programming. Background Technology

[0002] Pumped storage, as a flexible resource, plays an irreplaceable role in the power system. It effectively mitigates the impact of fluctuations in renewable energy output (such as wind and solar power) and random load changes, thereby significantly improving the reliability and economy of the power system. Specifically, pumped storage power stations play a crucial role in various complex operating scenarios, such as absorbing surplus power during peak wind and solar power generation periods, flood control and peak shaving during the flood season, system emergency backup, and frequency and voltage regulation (frequency and phase regulation), serving as the core support for maintaining the dynamic stability of the power system.

[0003] However, the different operating scenarios described above have different characteristics, which requires pumped storage power stations to be able to flexibly switch and execute optimal operating strategies according to real-time changes in system demand. Traditional centralized optimization methods rely on obtaining complete information for the entire scheduling cycle to perform global solutions and make scheduling decisions. This post-optimization characteristic makes it difficult to meet the requirements of high-timeliness real-time scheduling.

[0004] A commonly used online optimization method based on Model Predictive Control (MPC) involves system operators continuously optimizing the system's operation based on constantly updated ultra-short-term forecast information. However, the effectiveness of this method is affected by the accuracy of the intraday ultra-short-term forecast information. If the forecast information has a large deviation, the accuracy of the MPC optimization results will be greatly reduced, ultimately making it difficult to guarantee the accuracy requirements of scheduling decisions.

[0005] In summary, existing solution methods generally suffer from limitations in addressing real-time scheduling problems of pumped storage units, such as insufficient real-time performance or susceptibility to prediction errors in accuracy. Summary of the Invention

[0006] The purpose of this invention is to propose a method, equipment, and medium for rapidly solving the scheduling strategy of pumped-storage power plants based on approximate dynamic programming, thereby solving the technical problem that existing solution methods generally suffer from insufficient real-time performance or are easily affected by prediction errors when dealing with real-time scheduling problems of pumped-storage units.

[0007] Specifically, this invention provides a rapid solution method for the scheduling strategy of hybrid pumped-storage power station units based on approximate dynamic programming. The core idea of ​​the method is to train corresponding approximate value functions offline for different typical operating scenarios. These value functions are efficiently expressed by piecewise linear functions. In the real-time scheduling stage, the pre-trained approximate value function that best matches the current scenario is directly called for rapid decision calculation. Based on the Markov Decision Process (MDP), by designing the state variables, decision variables, external information process, transition function, and objective function of the pumped-storage unit optimization problem, a mathematical model for the coordinated optimization operation of multiple types of units under the theoretical framework of approximate dynamic programming is established. Then, an online optimization strategy for pumped-storage units based on the approximate dynamic programming algorithm of piecewise linear functions is proposed. This method effectively avoids the dependence of traditional methods on global information or high-precision ultra-short-term prediction, aiming to achieve a balance between high accuracy and high timeliness.

[0008] The method mentioned in this invention includes the following steps: S1. Construct an in-plant coordinated operation and scheduling model that includes constant-speed pumped-storage units, variable-speed pumped-storage units, hydropower units and wind power output, and establish power balance constraints and reservoir capacity change constraints. S2. The plant-wide coordinated operation scheduling model is reconstructed based on the Markov decision process, defining system state variables, decision variables, and external information factors, including the virtual energy storage charge state; the virtual energy storage charge state is obtained by the reservoir capacity of the upper and lower reservoirs through a linear mapping relationship. S3. Using piecewise linear functions, construct a value function approximation model with the virtual energy storage state of charge as the core variable; S4. In the current phase, based on historical and predicted data, the slope parameter of the piecewise linear function is trained and updated offline iteratively. S5. During the intraday real-time operation phase, the current virtual energy storage state of charge is calculated based on the real-time system status. The trained value function approximation model is then invoked to quickly solve for the optimal decision and output the unit scheduling strategy.

[0009] A storage medium storing instructions and data for implementing a method for rapidly solving a hybrid pumped-storage power station unit scheduling strategy based on approximate dynamic programming.

[0010] A device for rapidly solving the scheduling strategy of hybrid pumped-storage power plants based on approximate dynamic programming includes: a processor and a storage medium; the processor loads and executes the instructions and data in the storage medium to implement a method for rapidly solving the scheduling strategy of hybrid pumped-storage power plants based on approximate dynamic programming.

[0011] The beneficial effects of this invention are as follows: Addressing the problems of high dimensionality and slow solution in in-plant coordinated operation scheduling models, this invention designs a fast solution method based on ADP (Aspect-Driven Programming) for solving in-plant unit scheduling strategies. By establishing a virtual energy storage SOC based on upper and lower reservoirs and introducing a piecewise linear function approximation function, this invention can train and update the slope of the piecewise linear function offline before the day, and apply the trained piecewise linear function to efficiently solve intraday in-plant unit scheduling strategies with the goal of minimizing costs. Attached Figure Description

[0012] Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 This is a schematic diagram of the virtual energy storage SOC model for the upper and lower reservoirs; Figure 3 This is a schematic diagram of the hardware device operation according to an embodiment of the present invention. Detailed Implementation

[0013] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be further described below with reference to the accompanying drawings.

[0014] Before formally describing the present invention, a general description of the solution of the present invention will be given first to facilitate understanding.

[0015] Example 1 Please refer to Figure 1 This invention provides a method for rapidly solving the scheduling strategy of hybrid pumped-storage power plants based on approximate dynamic programming, comprising the following steps: S1. Construct an in-plant coordinated operation and scheduling model that includes constant-speed pumped-storage units, variable-speed pumped-storage units, hydropower units and wind power output, and establish power balance constraints and reservoir capacity change constraints. It should be noted that step S1 specifically includes: S11. With the goal of minimizing the total system operating cost, construct the objective function of the plant coordinated operation scheduling model. The total operating cost includes the unit power generation cost, pumping cost, start-up and shutdown cost, and wind curtailment cost. Specifically, a coordinated operation scheduling model for the units within the plant was first established. Its objective function is shown in Equation (1), which is used to solve the operation strategy that takes into account both the minimization of unit operating costs and the minimization of unit operating water consumption.

[0016] (1) (2) (3) in, Represents a set of scenes; Represents a set of time periods; , , , , and The costs for starting and stopping a single constant-speed pumped-storage unit, a variable-speed pumped-storage unit, and a hydroelectric unit are respectively. The probability of each scenario occurring; The power generation cost of constant-speed pumped-storage units, variable-speed pumped-storage units, and hydropower units; The pumping cost of constant-speed pumped-storage units and variable-speed pumped-storage units; , and Indicates in the scene s Lower constant speed pumped storage unit i、 Variable speed pumped storage unit j、 Hydropower units k At any moment t The power generation capacity; , Indicates in the scene s Lower constant speed pumped storage unit i and variable speed pumped storage units j At any moment t Pumping power; Cost of wind curtailment; For the scene s Wind power output forecast; In the scene s Power output in actual wind power scenarios; , and They represent in s Constant speed pumped storage unit in the scenario i Variable speed pumped storage unit j and hydroelectric generator units k exist t The unit flow rate of electricity generated at any given moment; , They represent in s Constant speed pumped storage unit in the scenario i and variable speed pumped storage units j exist t The unit pumping flow rate at any given time.

[0017] S12. Construct operating constraints for constant-speed pumped-storage units, including pumping / power generation constraints, operating condition mutual exclusion constraints, minimum continuous operating time constraints, and vibration zone constraints. Specifically, constraints are established for constant-speed pumped-storage units, variable-speed pumped-storage units, and hydroelectric units to ensure their safe and stable operation. The specific constraints are as follows: Constraints of constant-speed pumped storage units: (4) (5) (6) (7) (8) (9) (10) (11) Equation (4) represents the pumping constraint of the constant-speed pumped-storage unit, Equation (5) represents the power generation constraint of the constant-speed pumped-storage unit, Equation (6) represents the mutual exclusion constraint between pumping and power generation of the constant-speed pumped-storage unit, Equation (7) represents the correlation constraint between start-up and shutdown states of the constant-speed pumped-storage unit, Equation (8) represents the mutual exclusion constraint between start-up and shutdown of the constant-speed pumped-storage unit, Equation (9) represents the minimum continuous start-up and shutdown time constraint of the constant-speed pumped-storage unit, Equation (10) represents the vibration zone constraint of the constant-speed pumped-storage unit, and Equation (11) represents the power characteristic constraint of the constant-speed pumped-storage unit.

[0018] In the formula: , Indicates a constant-speed pumped-storage unit i At any moment t Is it in pumping or power generation mode? If the unit is in pumping mode, then =1, otherwise =0, if the unit is in generating mode, then =1, otherwise =0; Indicates in the scene s Lower constant speed pumped storage unit i The rated installed capacity; Indicates a constant-speed pumped-storage unit i Minimum and maximum power generation; Indicates a constant-speed pumped-storage unit i At any moment t Startup and shutdown variables, when This indicates that the generator unit is started; otherwise, it equals 0. Similarly. For a moment t Water head.

[0019] S13. Construct the operating constraints of the variable speed pumped storage unit, including pumping / power generation constraints, operating condition mutual exclusion constraints, minimum continuous operating time constraints, and vibration zone constraints. Specifically, the constraints of variable speed pumped storage units are as follows: (12) (13) (14) (15) (16) (17) (18) (19) Equation (12) represents the pumping constraint of the variable speed pumped storage unit, Equation (13) represents the power generation constraint of the variable speed pumped storage unit, Equation (14) represents the mutual exclusion constraint between pumping and power generation of the variable speed pumped storage unit, Equation (15) represents the correlation constraint between start-up and shutdown states of the variable speed pumped storage unit, Equation (16) represents the mutual exclusion constraint between start-up and shutdown of the variable speed pumped storage unit, Equation (17) represents the minimum continuous start-up and shutdown time constraint of the variable speed pumped storage unit, Equation (18) represents the vibration zone constraint of the variable speed pumped storage unit; and Equation (19) represents the power characteristic constraint of the variable speed pumped storage unit.

[0020] In the formula: This indicates a variable-speed pumped-storage unit in scenario s. i At any moment t Pumping power; Indicates in the scene s Lower speed pumped storage unit i At any moment t The power generation capacity; Indicates variable speed pumped storage unit i At any moment t Is it in pumping or power generation mode? If the unit is in pumping mode, then =1, otherwise =0, if the unit is in generating mode, then ,on the contrary ; Indicates variable speed pumped storage unit i Minimum and maximum pumping power; Indicates variable speed pumped storage unit i Minimum and maximum power generation; Indicates variable speed pumped storage unit i The start and stop variables at time t, when This indicates that the generator unit is started; otherwise, it equals 0. Similarly.

[0021] S14. Construct operating constraints for hydropower units, including power generation constraints, minimum start-up and shutdown time constraints, and vibration zone constraints; Specifically, the constraints of the hydropower units are as follows: (20) (twenty one) (twenty two) (twenty three) (twenty four) (25) Equation (20) represents the power generation constraint of the hydropower unit, Equation (21) represents the start-up and shutdown state association constraint of the hydropower unit, Equation (22) represents the start-up and shutdown mutual exclusion constraint of the hydropower unit, Equation (23) represents the minimum start-up and shutdown time constraint of the hydropower unit, Equation (24) represents the vibration zone constraint of the hydropower unit; and Equation (25) represents the power characteristic constraint of the hydropower unit.

[0022] In the formula: Indicates in the scene s Submersible generator set i At any moment t The power generation capacity; Indicates hydroelectric power unit i At any moment t Is it in power generation mode? If the unit is in generating mode, then ,on the contrary ; Indicates hydroelectric power unit i Minimum and maximum power generation; Indicates hydroelectric power unit i At any moment t Startup and shutdown variables, when This indicates that the generator unit is started; otherwise, it equals 0. Similarly; Indicates hydroelectric power unit i In the m The minimum and maximum power generation for each feasible interval.

[0023] S15. Construct system power balance constraints to ensure that the total power generation, including wind power, is always in balance with the load power. Specifically, the power supply and demand of new energy power generation systems need to be balanced, that is, the amount of electricity generated equals the amount of electricity consumed, as shown below: (26) Equation (26) represents the power balance constraint. express t The workload of the moment.

[0024] (27) Equation (27) is the wind power constraint.

[0025] S16. Construct constraints on the changes in the upper and lower reservoir capacity, and describe the dynamic relationship between the reservoir's water storage and the unit's power generation flow, pumping flow, and natural inflow.

[0026] Specifically, the constraints on reservoir capacity changes are constructed using the following formula: (28) (29) (30) (31) Equations (28) and (29) represent the capacity range constraints of the upper and lower reservoirs, respectively; Equations (30) and (31) represent the capacity variation constraints of the upper and lower reservoirs.

[0027] In the formula, , Indicates in the scene s The upper and lower reservoirs are in t Storage capacity for a given period of time; Indicates in the scene s Xiashang Reservoir t Water volume during a given time period.

[0028] S2. The plant-wide coordinated operation scheduling model is reconstructed based on the Markov decision process, defining system state variables, decision variables, and external information factors, including the virtual energy storage charge state; the virtual energy storage charge state is obtained by the reservoir capacity of the upper and lower reservoirs through a linear mapping relationship. It should be noted that in step S2, the virtual energy storage charge state is defined as follows: based on the real-time capacity of the upper reservoir and the real-time capacity of the lower reservoir, a one-dimensional scalar is calculated by linear weighted summation, which is used as the virtual energy storage charge state; when the upper reservoir is full and the lower reservoir is empty, the state is 100%; when the upper reservoir is empty and the lower reservoir is full, the state is 0%.

[0029] Step S2 is as follows: The plant-wide coordinated operation scheduling model is reconstructed based on MDP, as follows: (32) Decision variables based on MDP reconstruction As shown in the following formula, it represents the system. t Scheduling decisions made based on the current state at all times.

[0030] (33) Within the MDP modeling framework, coordinated unit operation requires decisions on unit output and pumped power generation flow based on system status information, as well as external information factors. The system is represented by the following formula.t Random factors at any given time.

[0031] (34) The relationships between state variables, decision variables, and external information factors are given by the state transition equation. The representation is shown in the following formula.

[0032] (35) (36) (37) (38) System operating costs per time period This includes the cost of wind curtailment. Hydropower unit power generation cost and start-up and shutdown cost Pumping cost, power generation cost, and start-up and shutdown cost of variable speed pumped storage units In addition to the pumping cost, power generation cost, and start-up and shutdown cost of constant-speed pumped-storage units. .

[0033] (39) The constraints include equations (4)-(31).

[0034] S3. Using piecewise linear functions, construct a value function approximation model with the virtual energy storage state of charge as the core variable; Step S3 specifically involves: approximating the value function as a piecewise linear convex function with respect to the virtual energy storage state of charge, where each segment corresponds to a slope parameter to be trained. The segmentation points of the piecewise linear function are set based on the historical operating data distribution or empirical values ​​of the virtual energy storage state of charge.

[0035] Specifically, the approximation function for the piecewise linear function is as follows: (40) (41) In the formula, the value function For multiple types of units, the status is as follows: From time t to the end of the scheduling period The optimal total operating cost at time 10:00. The optimal total operating cost at time t+1 can be obtained by solving equation (40).

[0036] For constructing a virtual energy storage SOC model based on the upper and lower reservoirs, please refer to [reference needed]. Figure 2 ; (42) In the formula, a and b This represents the coefficient corresponding to the storage capacity of the upper and lower reservoirs. When the upper reservoir is full and the lower reservoir has the least amount of water, the virtual reservoir stores 100% of its capacity; when the upper reservoir has the least amount of water and the lower reservoir is full, the virtual reservoir stores 0% of its capacity.

[0037] Combining equations (28)-(31) and (42), the relationship between the reconstructed virtual reservoir capacity changes is shown in the following equation: (43) (44) In the formula, This represents the capacity of the virtual reservoir at time t.

[0038] S4. In the current phase, based on historical and predicted data, the slope parameter of the piecewise linear function is trained and updated offline iteratively. It should be noted that step S4 specifically includes: S41. Initialize the slope values ​​of the piecewise linear function in each segment interval; S42. Traverse multiple training scenarios and time-series states, and obtain the sampled estimate of the slope of the value function by solving the Bellman optimal equation; S43. Adjust the slope of adjacent intervals to ensure that the piecewise linear function as a whole remains convex; S44. Based on the sampled estimated values, update the slope values ​​of the corresponding segmented intervals using a difference calculation method; S45. Repeat S42 to S44 until the slope sequence of the piecewise linear function converges or the preset number of training iterations is reached.

[0039] Specifically, it is as follows: (45) (46) (47) (48) In the formula, The slope of the piecewise linear function of the virtual reservoir's capacity consumption is sampled and estimated. for The piecewise linear function it is located on, To update the step size, .

[0040] For slope updates, if ,like After each update, the slope sampling estimate calculated using the difference method according to formula (48) needs to be applied. Piecewise linear functions The slope of the segment.

[0041] S5. During the intraday real-time operation phase, the current virtual energy storage state of charge is calculated based on the real-time system status. The trained value function approximation model is then invoked to quickly solve for the optimal decision and output the unit scheduling strategy.

[0042] It should be noted that step S5 specifically includes: S51. Collect the precise operating status of each unit, the actual reservoir capacity, and ultra-short-term load and wind power forecast information at the current moment. S52. Determine the current state of the system based on the collected information, and calculate the current virtual energy storage state of charge; S53. Using the current virtual energy storage state of charge as an index, call the piecewise linear function that has been converged after offline training a day before to obtain the corresponding future cost marginal value information. S54. Substitute the future cost marginal value information into the Bellman optimal equation that is solved in real time to calculate the optimal power generation and pumping power commands of each unit that minimize the global cost at the current moment.

[0043] Specifically, during the daytime offline training process, the slope of the piecewise linear function is first initialized and the total number of training iterations is determined. Then, at each scheduling time, the current state of the system is determined and the Bellman equation is solved sequentially with the aid of the piecewise linear function. During the daytime real-time optimization process, accurate real-time operating information is first collected and the true state of the system is determined. Then, with the aid of the well-performing piecewise linear function slope obtained through training, the Bellman equation is solved to obtain real-time scheduling decisions.

[0044] Example 2: Please see Figure 3 , Figure 3 This is a schematic diagram of the hardware device in operation according to an embodiment of the present invention. The hardware device specifically includes: a device 401 for rapidly solving the scheduling strategy of the unit in the hybrid pumped storage power station based on approximate dynamic programming, a processor 402, and a storage medium 403.

[0045] A device 401 for rapidly solving the scheduling strategy of hybrid pumped storage power plants based on approximate dynamic programming: The device 401 implements the method for rapidly solving the scheduling strategy of hybrid pumped storage power plants based on approximate dynamic programming.

[0046] Processor 402: The processor 402 loads and executes the instructions and data in the storage medium 403 to implement the fast solution method for the scheduling strategy of the hybrid pumped storage power station units based on approximate dynamic programming.

[0047] Storage medium 403: The storage medium 403 stores instructions and data; the storage medium 403 is used to implement the method for quickly solving the scheduling strategy of hybrid pumped storage power station units based on approximate dynamic programming.

[0048] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A fast solution method for the scheduling strategy of hybrid pumped-storage power plants based on approximate dynamic programming, characterized in that: Includes the following steps: S1. Construct an in-plant coordinated operation and scheduling model that includes constant-speed pumped-storage units, variable-speed pumped-storage units, hydropower units and wind power output, and establish power balance constraints and reservoir capacity change constraints. S2. The plant-wide coordinated operation scheduling model is reconstructed based on the Markov decision process, defining system state variables, decision variables, and external information factors, including the virtual energy storage charge state; the virtual energy storage charge state is obtained by the reservoir capacity of the upper and lower reservoirs through a linear mapping relationship. S3. Using piecewise linear functions, construct a value function approximation model with the virtual energy storage state of charge as the core variable; S4. In the current phase, based on historical and predicted data, the slope parameter of the piecewise linear function is trained and updated offline iteratively. S5. During the intraday real-time operation phase, the current virtual energy storage state of charge is calculated based on the real-time system status. The trained value function approximation model is then invoked to quickly solve for the optimal decision and output the unit scheduling strategy.

2. The method for rapidly solving the scheduling strategy of hybrid pumped-storage power plants based on approximate dynamic programming as described in claim 1, characterized in that: Step S1 specifically includes: S11. With the goal of minimizing the total system operating cost and unit water consumption, construct the objective function of the plant coordinated operation scheduling model. The total operating cost includes the unit power generation cost, pumping cost, start-up and shutdown cost, and wind curtailment cost. S12. Construct operating constraints for constant-speed pumped-storage units, including pumping / power generation constraints, mutually exclusive operating conditions constraints, and vibration zone constraints. S13. Construct operating constraints for variable speed pumped storage units, including pumping / power generation constraints, operating condition mutual exclusion constraints, and vibration zone constraints. S14. Construct operating constraints for hydropower units, including power generation constraints, minimum start-up and shutdown time constraints, and vibration zone constraints; S15. Construct system power balance constraints to ensure that the total power generation, including wind power, is always in balance with the load power. S16. Construct constraints on the changes in the upper and lower reservoir capacity, and describe the dynamic relationship between the reservoir's water storage and the unit's power generation flow, pumping flow, and natural inflow.

3. The method for rapidly solving the scheduling strategy of hybrid pumped-storage power station units based on approximate dynamic programming as described in claim 1, characterized in that: In step S2, the virtual energy storage charge state is defined as follows: based on the real-time capacity of the upper reservoir and the real-time capacity of the lower reservoir, a one-dimensional scalar is calculated by linear weighted summation and used as the virtual energy storage charge state; when the upper reservoir is full and the lower reservoir is empty, the state is 100%; when the upper reservoir is empty and the lower reservoir is full, the state is 0%.

4. The method for rapidly solving the scheduling strategy of hybrid pumped-storage power station units based on approximate dynamic programming as described in claim 1, characterized in that: Step S3 specifically involves: approximating the value function as a piecewise linear convex function with respect to the virtual energy storage state of charge, with each segment corresponding to a slope parameter to be trained.

5. The method for rapidly solving the scheduling strategy of hybrid pumped-storage power station units based on approximate dynamic programming as described in claim 1, characterized in that: Step S4 specifically includes: S41. Initialize the slope values ​​of the piecewise linear function in each segment interval; S42. Traverse multiple training scenarios and time-series states, and obtain the sampled estimate of the slope of the value function by solving the Bellman optimal equation; S43. Adjust the slope of adjacent intervals to ensure that the piecewise linear function as a whole remains convex; S44. Based on the sampled estimated values, update the slope values ​​of the corresponding segmented intervals using a difference calculation method; S45. Repeat S42 to S44 until the slope sequence of the piecewise linear function converges or the preset number of training iterations is reached.

6. The method for rapidly solving the unit scheduling strategy of a hybrid pumped-storage power station based on approximate dynamic programming as described in claim 1, characterized in that, Step S5 specifically includes: S51. Collect the precise operating status of each unit, the actual reservoir capacity, and ultra-short-term load and wind power forecast information at the current moment. S52. Determine the current state of the system based on the collected information, and calculate the current virtual energy storage state of charge; S53. Using the current virtual energy storage state of charge as an index, call the piecewise linear function that has been converged after offline training a day before to obtain the corresponding future cost marginal value information. S54. Substitute the future cost marginal value information into the Bellman optimal equation that is solved in real time to calculate the optimal power generation and pumping power commands of each unit that minimize the global cost at the current moment.

7. A storage medium, characterized in that: The storage medium stores instructions and data to implement the method for quickly solving the scheduling strategy of hybrid pumped storage power station units based on approximate dynamic programming, as described in any one of claims 1 to 6.

8. A device for rapidly solving the scheduling strategy of hybrid pumped-storage power station units based on approximate dynamic programming, characterized in that: include: A processor and a storage medium; the processor loads and executes instructions and data in the storage medium to implement the fast solution method for the scheduling strategy of a hybrid pumped-storage power station based on approximate dynamic programming as described in any one of claims 1 to 6.

Images