Day-before-day multi-time scale optimization scheduling method for wind-light-storage alliance system

Abstract

The invention relates to the technical field of power system scheduling optimization, and discloses a wind-light-storage alliance system day-ahead-day multi-time scale optimization scheduling method, which comprises the following steps of: constructing a wind-light output scene generation and reduction model considering uncertainty, generating an initial scene set of wind and light output based on a Latin hypercube sampling method, and generating a wind-light output scene set; reducing the initial scene set by a scene reduction method based on a Kantorovich distance to obtain a typical wind and light output scene; constructing a day-ahead optimization scheduling model, establishing a multi-target stochastic programming model based on a typical wind and light output scene, and solving to obtain an optimization output plan of each unit in each day-ahead time period; and constructing an intra-day rolling optimization scheduling model, constructing an intra-day rolling optimization model based on a multi-target stochastic programming model, and performing real-time feedback and correction on the optimization output plan to obtain a final scheduling instruction of each time period in the day, wherein the constraint condition of the intra-day rolling optimization model is the same as that of the day-ahead optimization model. The method has the advantages of being capable of ensuring that the total revenue of the system is maximized and the grid-connected power is stable in the face of uncontrollable prediction errors.

Description

Technical Field

[0001] This invention relates to the field of power system optimization scheduling technology, and in particular to a day-ahead-intraday multi-timescale optimization scheduling method for wind-solar-storage alliance systems. Background Technology

[0002] Currently, the penetration rate of new energy sources such as wind power and photovoltaics in my country's power grid continues to rise. However, the output of new energy sources is characterized by randomness, volatility, and anti-peak-shaving features, posing significant challenges to the safe and stable operation of the power system. Large-scale absorption of these sources has become a bottleneck restricting the green transformation of the energy structure. Against this backdrop, there is an urgent need to introduce flexible regulation resources to mitigate wind and solar power fluctuations. Pumped storage hydroelectric power stations, due to their peak-shaving and valley-filling capabilities and large-scale energy storage advantages, have become one of the key technologies for solving this problem. How to fully tap the flexible regulation potential of pumped storage hydroelectric power stations and achieve synergistic and optimized operation with wind and solar power stations, thereby improving system absorption capacity, ensuring grid stability, and simultaneously enhancing the economic benefits of all stakeholders, has become a crucial issue in promoting the construction of new power systems and the sustainable development of the energy industry. Against this backdrop, Reference [1] proposed a capacity optimization allocation method based on multi-timescale decision-making in the power market to address the market positioning and investment return issues of pumped storage power stations. This method involves the participation of the wind-solar-spot alliance in the medium- and long-term market. The reference focuses on the long-term capacity planning problem. Its lower-level market model essentially provides signals for price and clearing volume for planning and analyzes the system's operating strategy on the day-to-day short-term timescale after the planned capacity is determined. Reference [2] quantifies the supply-demand imbalance of flexibility through convolutional probability weighting, establishes a hierarchical optimization scheduling model for pumped storage power stations, and improves the overall flexibility of the system and the economic benefits of pumping itself. However, it does not involve the collaborative operation of wind, solar, and pumped storage as a community of interests. Reference [3] constructs a power system scheduling framework that integrates the complementary characteristics of water, solar, and pumped storage with DC transmission characteristics. It collaboratively models the controllability of cascade hydropower, pumped storage units, and DC transmission systems to cope with the uncertainty of photovoltaic output. It also uses the sub-Bruker optimization method to seek the comprehensive optimal system operating cost and curtailment risk under the worst probability distribution. However, it focuses more on characterizing the uncertainty of wind and solar joint output through a large number of typical scenarios. Reference [4] constructed a comprehensive evaluation framework that considers direct economic benefits, carbon emission reduction benefits and grid-connected power stability, and carried out multi-objective optimization research on wind-solar-storage hybrid systems with pumped storage. However, it is essentially a single-stage optimization and fails to fully consider the random fluctuation characteristics of wind and solar power output in the day-ahead and intraday stages. In summary, existing research has certain limitations: most models are deterministic optimizations or rely on traditional dispatchable units as a stable basis. In addition, their optimization scales are mostly concentrated on a single time scale and lack a multi-stage collaborative decision-making framework that runs through day-ahead planning and intraday rolling adjustments.

[0003] Existing literature:

[0004] [1] Zhong Hao, Wu Fan, Zhang Lei, et al. Optimal allocation of pumped storage power station capacity for multi-timescale decision-making in the power market [J]. High Voltage Engineering, 2023, 49(10):4109-4119.

[0005] [2] Lü Wanyu, Zhao Hongsheng, Han Yingsheng, et al. Research on hierarchical optimization scheduling method to improve the flexibility of pumped storage system under supply and demand uncertainty [J / OL]. Journal of Electrical Engineering, 1-16 [2025-09-27].

[0006] [3] Tan Jing, He Chuan, Chen Baorui, et al. A distributed bar optimal scheduling method for power systems considering hydro-solar-storage complementarity and DC transmission [J]. Proceedings of the CSEE, 2024, 44(15): 5947-5960.

[0007] [4]Wang R, Yang W, Li X, et al. Day-ahead multi-objective optimal operation of Wind–PV–Pumped Storage hybrid system considering carbonemissions[J]. Energy Reports, 2022, 8: 1270-1279. Summary of the Invention

[0008] The purpose of this invention is to overcome the shortcomings of the prior art and provide a day-to-day multi-timescale optimization scheduling method for wind-solar-storage alliance systems.

[0009] The objective of this invention is achieved through the following technical solution: a day-ahead-intraday multi-timescale optimized scheduling method for a wind-solar-storage alliance system, the method comprising,

[0010] S1. Construct a wind and solar power output scene generation and reduction model that considers uncertainty. Generate an initial scene set of wind and solar power output based on the Latin hypercube sampling method, and reduce the initial scene set based on the scene reduction method of Kantorovich distance to obtain typical wind and solar power output scenes.

[0011] S2. Construct a day-ahead optimization scheduling model. Based on typical wind and solar power output scenarios, establish a multi-objective stochastic programming model with the goal of maximizing the joint economic benefits of the wind-solar-storage alliance system and minimizing the fluctuation of the grid-connected remaining load. Solve the model to obtain the optimized power output plan of each unit for each day-ahead period.

[0012] S3. Construct an intraday rolling optimization scheduling model. Based on a multi-objective stochastic programming model, with the objectives of maximizing the expected total revenue and minimizing the fluctuation of the remaining load within the rolling optimization window, construct an intraday rolling optimization model, and provide real-time feedback and correction to the optimized output plan to obtain the final scheduling instructions for each time period within the day.

[0013] Specifically, the generation of the initial scene set includes,

[0014] The initial sample value matrix, i.e., the initial scene set, is obtained by calculation:

[0015] ;

[0016] ;

[0017] In the formula, , Let be the distribution function loaded by the j-th variable and the j-th variable of the i-th sample, respectively. The sampled values ​​are obtained by hypercubic sampling; the initial sampled value matrix S contains... Let be the element in the i-th row and j-th column; N is the number of sample groups; each column of matrix P consists of random integers from 1 to N; each element in matrix R is a random number uniformly distributed in the interval [0,1).

[0018] Reorder the initial matrix: Reorder the elements of matrix P After processing, an intermediate matrix Y is obtained that follows a standard normal distribution, and its elements... As shown in the following formula:

[0019] ;

[0020] In the formula, It is the inverse function of the cumulative probability distribution function of the standard normal distribution;

[0021] Perform Cholesky decomposition on the covariance matrix COV(Y) of matrix Y:

[0022] ;

[0023] ;

[0024] In the formula, The lower triangular matrix obtained from Cholesky decomposition; This is the lower triangular matrix. The transpose of the matrix; It is a lower triangular matrix Its transpose The product of these variables is mathematically equal to the covariance matrix COV(Y) of the random variables.

[0025] Arrange matrix P according to matrix The order of the elements in the matrix is ​​rearranged to obtain a new matrix. Then based on the new matrix The initial sample value matrix is ​​calculated to obtain the effective control sample matrix.

[0026] Specifically, the scene reduction method based on Kantorovich distance involves the following steps for reducing the initial scene set:

[0027] S110. Assign equal initial probabilities to each scene in the initial scene set, and set the current scene set as the original scene set;

[0028] S120, Calculate each pair of scenes in the scene set. Kantorovich distance between :

[0029] ;

[0030] In the formula, Let i be the vector of the landscape scene. Let j be the vector of the landscape scene.

[0031] S130. For each target scene, find the scene closest to it, and calculate the product of that distance and the probability of the nearest scene:

[0032] ;

[0033] In the formula, For scene probabilities;

[0034] S140. Find the scene pair that minimizes the product, delete one of the scenes from the current scene set, and add the probability of the deleted scene to the other matching scene, and update the current scene set and the number of scenes.

[0035] S150. Repeat steps S120 to S140 until the number of scenes reaches the preset target number.

[0036] Specifically, the objective function of the multi-objective stochastic programming model is:

[0037] ;

[0038] ;

[0039] ;

[0040] ;

[0041] ;

[0042] In the formula, For the joint benefits of the wind-solar-storage alliance; , , These are the on-grid electricity price for time period t, the hydropower price, and the penalty price for curtailment of wind and solar power. , These are the system's on-grid electricity consumption and the amount of renewable energy abandoned; , These represent the on-grid electricity generated by wind power and photovoltaic power plants during time period t, respectively. , These represent the power generation and pumping power of the pumped storage power station during time period t, respectively. , These represent the amount of wind and solar power curtailed by wind power and solar power plants respectively during time period t. The cost of the pumped storage unit at time t; The start-up and shutdown costs during power generation; The start-up and shutdown costs during pumping; This refers to the number of pumped storage units; This represents the number of pumped-storage units started and stopped during time period t. , These represent the number of pumped-storage units connected to the grid during time period t and time period t-1, respectively. Fluctuations in the remaining load; Let t be the magnitude of the remaining load at time t; This represents the average value of the remaining load at time t. Let t be the load magnitude at time t.

[0043] Specifically, the constraints of the multi-objective stochastic programming model include:

[0044] Pumped storage power station reservoir capacity constraints:

[0045] ;

[0046] ;

[0047] ;

[0048] In the formula, Let t be the reservoir capacity of the pumped storage power station. The reservoir capacity of the pumped storage power station at time t-1; This is the conversion factor between electricity and storage capacity; To improve the power generation efficiency of pumped storage power stations; , These are the initial and final reservoir capacities for the optimized cycle of the pumped storage unit; This is the lower limit of the storage capacity. This is the maximum storage capacity.

[0049] Pumped storage power station operating condition transition constraints:

[0050] ;

[0051] ;

[0052] ;

[0053] In the formula, and These represent the pumped-storage unit being in power generation or pumping mode at time t, respectively. and These represent the number of times the pumped storage tank is started and stopped during power generation and pumping operations, respectively. , The pumped-storage unit is in power generation and pumping state respectively at time t-1;

[0054] Maximum start-stop frequency constraint for pumped storage units:

[0055] ;

[0056] ;

[0057] In the formula, and These represent the maximum number of start-ups and shutdowns of the pumped storage tank under power generation and pumping conditions, respectively.

[0058] Unit output and grid-connected power constraints:

[0059] ;

[0060] ;

[0061] ;

[0062] ;

[0063] ;

[0064] In the formula, , These represent the minimum and maximum power output of the pumped-storage unit during time period t, respectively. , These represent the minimum and maximum output of the pumped storage unit during time period t, respectively. , These represent the minimum and maximum power output of wind power during time period t, respectively. , These represent the upper and lower limits of the photovoltaic power station's output during time period t; , These represent the upper and lower limits of the power received by the power grid during time period t, respectively.

[0065] Constraints on wind and solar power curtailment:

[0066] ;

[0067] ;

[0068] In the formula, , These represent the predicted power output values ​​for wind power and solar power during time period t, respectively.

[0069] Specifically, the objective function of the intraday rolling optimization model is:

[0070] ;

[0071] ;

[0072] ;

[0073] ;

[0074] ;

[0075] ;

[0076] ;

[0077] ;

[0078] ;

[0079] ;

[0080] In the formula, For the past and present time periods, Forecast period in the future; To optimize the total predicted revenue of the time-complementary system for intraday time period t; , These are the real-time on-grid electricity price and pumping electricity price for the t1 time period, respectively. , These are the on-grid electricity price and pumping electricity price for the t2 period, respectively. , , , , , Optimized for the day at time t The power generation of wind and solar power, the amount of wind curtailment, the amount of solar curtailment, pumped storage power generation, and pumped hydro power are all subject to real-time dispatch. , , , , , They are respectively Real-time forecasts of wind and solar power generation, curtailed wind and solar power, pumped storage power generation, and pumped hydro power generation that are involved in the dispatching process; , The actual wind and solar power outputs at time t1 within the day are respectively. , The predicted wind and solar power outputs at time t2 are respectively. , They are respectively , Start-up and shutdown costs of pumping water during specific time periods; Costs associated with pumped storage start-up and shutdown; This represents the total number of pumped storage units. This is an indicator of the start-up / shutdown status change of the kth pumped storage unit during time period t within the day. Let $t$ be the startup cost of the k-th generating unit under power generation conditions during time period $t$. Let $t$ be the startup cost of the k-th unit in pumping operation during time period $t$. The online status of the k-th generating unit during time period t; This refers to the online status of the k-th unit during time period t-1. This refers to fluctuations in the remaining load during the intraday optimization phase. This represents the remaining load at time t during the intraday optimization phase. This represents the average value of the remaining load during the daily optimization phase. This represents the load size at time t during the intraday optimization phase.

[0081] The present invention has the following advantages:

[0082] This invention constructs a two-stage stochastic optimization model (day-ahead and intraday phases) to effectively mitigate wind and solar power output fluctuations, ensure grid stability, and significantly improve the overall economic benefits of the consortium and the individual returns of participating entities. Employing scenario generation and reduction techniques based on Latin hypercube sampling and Kantorovich distance, it effectively characterizes and reduces the uncertainty of wind and solar power output. Compared to deterministic optimization or simple scenario analysis methods, it significantly reduces decision-making errors caused by prediction bias. The established two-stage collaborative optimization framework (day-ahead and intraday phases) achieves a balance between planning and flexibility. In the day-ahead phase, an optimal power output plan is formulated based on typical scenarios, aiming to maximize expected returns; in the intraday phase, the plan is rolled over based on ultra-short-term forecasts, and the pumped-storage unit status is adjusted in real time to cope with wind and solar power fluctuations. This two-stage optimization framework ensures that the total system revenue is maximized and grid-connected power remains stable even when facing uncontrollable prediction errors. Attached Figure Description

[0083] Figure 1 This is a schematic diagram of the wind-solar-storage alliance system architecture of the present invention;

[0084] Figure 2 This is a schematic diagram of the wind power scenario reduction result of the present invention;

[0085] Figure 3 This is a schematic diagram of the photovoltaic scenario reduction result of the present invention;

[0086] Figure 4 This is a schematic diagram of the intraday rolling optimization scheduling process of the present invention;

[0087] Figure 5 This is a schematic diagram of the new energy prediction and actual output curves of the present invention;

[0088] Figure 6 This is a schematic diagram of the day-to-day multi-timescale optimization scheduling results of the present invention. Detailed Implementation

[0089] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only for explaining the invention and are not intended to limit the invention; that is, the described embodiments are merely some embodiments of the invention, and not all embodiments. The components of the embodiments of the invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0090] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0091] It should be noted that relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0092] The present invention will be further described below with reference to the accompanying drawings, but the scope of protection of the present invention is not limited to the following description.

[0093] like Figures 1 to 6 As shown, the wind-solar-storage alliance system employs a multi-timescale optimization scheduling method from day to day. This method includes:

[0094] S1. Construct a wind and solar power output scene generation and reduction model that considers uncertainties. Generate an initial scene set of wind and solar power output based on the Latin hypercube sampling method, and reduce the initial scene set based on the scene reduction method of Kantorovich distance to obtain typical wind and solar power output scenes. The Latin hypercube sampling method is a stratified sampling technique. By stratifying the probability distribution of input variables, it can efficiently and accurately simulate the original distribution function with fewer sampling times.

[0095] The generation of the initial scene set includes,

[0096] Assuming the wind and solar power output model contains K random variables, N sets of samples need to be generated. First, construct a K×N matrix P, where each column consists of randomly permuted integers from 1 to N. Simultaneously, construct another K×N matrix R, where each element is a uniformly distributed random number within the interval [0,1). Calculate the initial sampling value matrix, i.e., the initial scene set:

[0097] ;

[0098] ;

[0099] In the formula, , Let be the distribution function loaded by the j-th variable and the j-th variable of the i-th sample, respectively. The sampled values ​​are obtained by hypercubic sampling; the initial sampled value matrix S contains... Let be the element in the i-th row and j-th column; N is the number of sample groups; each column of matrix P consists of random integers from 1 to N; each element in matrix R is a random number uniformly distributed in the interval [0,1).

[0100] The initial sample values ​​generated by the above method have random correlations between variables. To reduce the correlation between the sample sequences of random variables and improve sampling accuracy, this invention uses the Cholesky decomposition method to reorder the initial matrix; firstly, the elements in matrix P are... After processing, an intermediate matrix Y is obtained that follows a standard normal distribution, and its elements... As shown in the following formula:

[0101] ;

[0102] In the formula, It is the inverse function of the cumulative probability distribution function of the standard normal distribution;

[0103] Perform Cholesky decomposition on the covariance matrix COV(Y) of matrix Y:

[0104] ;

[0105] ;

[0106] In the formula, The lower triangular matrix obtained from Cholesky decomposition; This is the lower triangular matrix. The transpose of the matrix; It is a lower triangular matrix Its transpose The product of these variables is mathematically equal to the covariance matrix COV(Y) of the random variables.

[0107] Using the matrix L obtained from the decomposition, matrix Y is transformed to obtain a new matrix with reduced correlation. ; Rearrange matrix P according to matrix The order of the elements in the matrix is ​​rearranged to obtain a new matrix. Then based on the new matrix The initial sample value matrix is ​​calculated to obtain the effectively controlled sample matrix S, which represents the total number of time dimensions of the scene. After generating a large number of initial scenes through Latin hypercube sampling, this invention employs a scene reduction method based on Kantorovich distance to merge similar scenes and reduce them into a representative few typical scenes. The specific steps are as follows:

[0108] S110. Initialization: Assign equal initial probabilities to each scene in the initial scene set. Set the current scene set as the original scene set, and the number of scenes. ;

[0109] S120, Distance calculation, calculates distance for each pair of scenes in the scene set. Kantorovich distance between The distance is defined as the distance between two landscape scene vectors i and j. , Euclidean distance of the difference in output at different times:

[0110] ;

[0111] In the formula, Let i be the vector of the landscape scene. Let j be the vector of the landscape scene.

[0112] S130. Find the minimum distance pair for each target scene. Find the scene closest to it. (Scenario probability is) ), and calculate the distance to the nearest scene. Product of probabilities :

[0113] ;

[0114] In the formula, For scene probabilities;

[0115] S140, Scene reduction, finding the product that makes the product... Minimal scene pair Remove one of the scenes from the current scene set (i.e., or Remove the element that minimizes the product from the list of elements in the list, and then remove the deleted element. probability Accumulate to another matching scene Above, that is Update the current scene set and the number of scenes. ;

[0116] S150, Iterate through the loop, repeating steps S120 to S140 until the number of scenes is reached. The preset target quantity has been achieved.

[0117] S2. Construct a day-ahead optimization scheduling model. Based on typical wind and solar power output scenarios, establish a multi-objective stochastic programming model with the goal of maximizing the joint economic benefits of the wind-solar-storage alliance system and minimizing the fluctuation of the grid-connected remaining load. Solve the model to obtain the optimized power output plan of each unit for each day-ahead period.

[0118] Maximizing the joint economic benefits of the wind-solar-storage alliance system aims to maximize the overall joint benefits of the wind-solar-storage complementary system in the day-ahead electricity market. Its main revenue components include the grid-connected electricity sales revenue from wind and solar power, as well as the electricity sales revenue from pumped storage power stations during power generation. Its main cost components include the electricity purchase costs incurred by pumped storage power stations during pumping, the cost of wind and solar curtailment penalties due to limited system regulation capacity, and the start-up and shutdown costs incurred by pumped storage units when switching between power generation and pumping modes. By optimizing the output plans of each unit in each time period, the total net revenue of the system is maximized.

[0119] Minimizing grid-connected residual load fluctuations aims to minimize the residual load fluctuations at the grid connection points of complementary systems, thereby improving the safety and stability of grid operation. Residual load is defined as the difference between the total system load and the total output of the wind-solar-storage combined system. By minimizing the variance or standard deviation of the residual load, the total output curve of the combined system can closely track changes in load demand, thus effectively mitigating the impact of wind and solar randomness on the grid.

[0120] The objective function of the multi-objective stochastic programming model is:

[0121] ;

[0122] ;

[0123] ;

[0124] ;

[0125] ;

[0126] In the formula, For the joint benefits of the wind-solar-storage alliance; , , These are the on-grid electricity price for time period t, the hydropower price, and the penalty price for curtailment of wind and solar power. , These are the system's on-grid electricity consumption and the amount of renewable energy abandoned; , These represent the on-grid electricity generated by wind power and photovoltaic power plants during time period t, respectively. , These represent the power generation and pumping power of the pumped storage power station during time period t, respectively. , These represent the amount of wind and solar power curtailed by wind power and solar power plants respectively during time period t. The cost of the pumped storage unit at time t; The start-up and shutdown costs during power generation; The start-up and shutdown costs during pumping; This refers to the number of pumped storage units; This represents the number of pumped-storage units started and stopped during time period t. , These represent the number of pumped-storage units connected to the grid during time period t and time period t-1, respectively. Fluctuations in the remaining load; Let t be the magnitude of the remaining load at time t; This represents the average value of the remaining load at time t. Let t be the load magnitude at time t.

[0127] The constraints of a multi-objective stochastic programming model include:

[0128] Pumped storage power station reservoir capacity constraints require that the reservoir capacity be equal at the beginning and end of the dispatch cycle to ensure the sustainability of the cyclical operation, and stipulate that the reservoir capacity must be within the allowable upper and lower limits at all times.

[0129] ;

[0130] ;

[0131] ;

[0132] In the formula, Let t be the reservoir capacity of the pumped storage power station. The reservoir capacity of the pumped storage power station at time t-1; This is the conversion factor between electricity and storage capacity; To improve the power generation efficiency of pumped storage power stations; , These are the initial and final reservoir capacities for the optimized cycle of the pumped storage unit; This is the lower limit of the storage capacity. This is the maximum storage capacity.

[0133] Pumped storage power station operating condition transition constraints: Pumped storage units can only be in one of three states at any given time: generating, pumping, or shut down. Simultaneous generation and pumping are prohibited.

[0134] ;

[0135] ;

[0136] ;

[0137] In the formula, and These represent the pumped-storage unit being in power generation or pumping mode at time t, respectively. and These represent the number of times the pumped storage tank is started and stopped during power generation and pumping operations, respectively. , The pumped-storage unit is in power generation and pumping state respectively at time t-1;

[0138] The maximum number of start-stop cycles for pumped storage units is constrained to prevent excessive equipment wear. This constraint limits the maximum number of start-ups and shutdowns allowed within a complete scheduling cycle.

[0139] ;

[0140] ;

[0141] In the formula, and These represent the maximum number of start-ups and shutdowns of the pumped storage tank under power generation and pumping conditions, respectively.

[0142] Unit output and grid-connected power constraints stipulate that the active power output of all power generation units (wind power, photovoltaic, pumped storage) must be between the minimum and maximum values ​​of their rated output. Simultaneously, to ensure grid security, the total grid-connected power transmitted from the combined system to the point of common coupling must also meet the grid's transmission limits.

[0143] ;

[0144] ;

[0145] ;

[0146] ;

[0147] ;

[0148] In the formula, , These represent the minimum and maximum power output of the pumped-storage unit during time period t, respectively. , These represent the minimum and maximum output of the pumped storage unit during time period t, respectively. , These represent the minimum and maximum power output of wind power during time period t, respectively. , These represent the upper and lower limits of the photovoltaic power station's output during time period t; , These represent the upper and lower limits of the power received by the power grid during time period t, respectively.

[0149] Constraints on wind and solar power curtailment:

[0150] ;

[0151] ;

[0152] In the formula, , These represent the predicted power output values ​​for wind power and solar power during time period t, respectively.

[0153] S3. Construct an intraday rolling optimization scheduling model. Based on a multi-objective stochastic programming model, with the objectives of maximizing the expected total revenue and minimizing the fluctuation of the remaining load within the rolling optimization window, an intraday rolling optimization model is constructed. The constraints of the intraday rolling optimization model are the same as those of the day-ahead optimization scheduling model. The optimized output plan is fed back and corrected in real time to obtain the final scheduling instructions for each time period within the day.

[0154] Maximizing the expected total revenue within the rolling optimization window aims to maximize the expected total revenue of the complementary system within the intraday rolling optimization window. It uses known actual wind and solar power output and real-time electricity prices for settlement. For current and future periods, it optimizes the operation strategy of pumped storage based on the latest ultra-short-term forecast data, while minimizing the potential penalty costs caused by wind and solar forecast deviations.

[0155] ;

[0156] In the formula, The combined total revenue of the complementary system during intraday optimization; For intraday returns during time period t;

[0157] When performing intraday rolling optimization, the actual electricity price and actual wind and solar power output will change over time. By performing rolling optimization at each moment, the optimization results at each moment can be obtained.

[0158] ;

[0159] ;

[0160] ;

[0161] (1x);

[0162] (2x);

[0163] ;

[0164] (3x);

[0165] (4x);

[0166] In the formula, For the past and present time periods, Forecast period in the future; To optimize the total predicted revenue of the time-complementary system for intraday time period t; , These are the real-time on-grid electricity price and pumping electricity price for the t1 time period, respectively. , These are the on-grid electricity price and pumping electricity price for the t2 period, respectively. , , , , , The optimized values ​​for time t are as follows: (At any given moment) The amount of wind and solar power fed into the grid, the amount of wind curtailment, the amount of solar curtailment, pumped storage power generation, and pumped hydro power that are all involved in the dispatching process at any given moment. , , , , , They are respectively Real-time forecasts of wind and solar power generation, curtailed wind and solar power, pumped storage power generation, and pumped hydro power generation that are involved in the dispatching process; , The actual wind and solar power outputs at time t1 within the day are respectively. , The predicted wind and solar power outputs at time t2 are respectively. , They are respectively , Start-up and shutdown costs of pumping water during specific time periods; Costs associated with pumped storage start-up and shutdown; This represents the total number of pumped storage units. This is the indicator of the start-up / shutdown status change of the kth pumped storage unit during time period t within the day (1 if start-up or shutdown occurs, 0 otherwise); Let $t$ be the startup cost of the k-th generating unit under power generation conditions during time period $t$. Let $t$ be the startup cost of the k-th unit in pumping operation during time period $t$. The online status of the kth unit during time period t (1 for running, 0 for shut down); The online status of the kth unit in time period t-1 (the previous time period); Equations 1x-2x and 3x-4x indicate that when optimizing time period t within the day, the wind and solar power output values ​​at time t1 before time t are taken as the actual known output values ​​as the initial values ​​for the current time period optimization, and the wind and solar power output values ​​at time t2 after time t are taken as the output values ​​that were predicted but not known before the day as the initial values ​​for the current time period optimization;

[0167] Minimizing residual load fluctuations aims to minimize the fluctuations in the residual load at the system's grid connection point during the rolling optimization cycle. By dynamically adjusting the output of each unit, the deviation between the total output curve of the combined system and the load is reduced, thereby further smoothing out wind and solar load fluctuations at the real-time level.

[0168] ;

[0169] (3-x);

[0170] In the formula, This refers to fluctuations in the remaining load during the intraday optimization phase. This represents the remaining load at time t during the intraday optimization phase. This represents the average value of the remaining load during the daily optimization phase. This represents the load size at time t during the intraday optimization phase.

[0171] The electricity consumption of each entity described in Formula 3-x represents the optimization result at time t during intraday rolling optimization. The optimization model at time t is as follows:

[0172] ;

[0173] ;

[0174] ;

[0175] In the formula, The fluctuation of the total remaining load during the optimization phase at time t within the day; This represents the remaining load at time t1 during the optimization phase at time t within the day. This represents the average of the total remaining load during the daily optimization phase. , , The on-grid power generation at time t1 is the optimization phase at time t within the day. The optimization is performed on a rolling basis at fixed time intervals. During each optimization, the actual wind and solar power output of the past period is used as a known fixed value. The future period is optimized based on the latest ultra-short-term forecast data. Only the output plan of pumped storage units in the future period is optimized, and the optimization results are applied to the next period.

[0176] Example system such as Figure 1The system comprises a pumped-storage power station, a wind farm, and a photovoltaic power station, forming a complementary alliance. Key technical parameters of each component are shown in Table 1.

[0177] Table 1

[0178] Parameter name numerical values Parameter name numerical values 、 / MWh 600 、 / ￥ 2000 、 / MWh 0 / ￥ / MW 5 / m³ 2000 0.5 / m³ 100 0.75 / m³ 600

[0179] Based on the output forecast curves of wind power and solar power, as well as the system load forecast curves, a multi-objective stochastic optimization scheduling model is established for the day-ahead phase, aiming to maximize the joint benefits of complementary systems and minimize the fluctuation rate of grid-connected residual load. The simulation scheduling cycle is set to 24 hours, with optimization performed in 1-hour increments. To demonstrate the adaptability of this method across multiple time scales, the intraday operation process is further simulated: considering that the actual output of wind power and solar power during the day may deviate from their day-ahead forecast values, and that real-time market electricity prices may also fluctuate. Therefore, the intraday rolling optimization strategy proposed in this invention is used to dynamically respond to and correct for the aforementioned uncertainties. Figure 2 The paper presents typical wind power output scenario curves generated based on Latin hypercube sampling and scenario reduction techniques to characterize uncertainties. Figure 3 It showcases typical photovoltaic output scenario curves reflecting different probabilities, generated based on Latin hypercube sampling and scenario reduction techniques; Figure 4 This demonstrates the rolling optimization principle in intraday scheduling, which utilizes updated ultra-short-term forecast data to dynamically advance the forecast domain and optimization period. Figure 5 It shows a comparison between the power curves of wind power and solar power during the day-ahead forecast phase and the actual operating power curves during the day, as well as the specific deviations. Figure 6 The study demonstrates the final operational results of the wind-solar-storage alliance system in achieving power balance and absorption through peak shaving and valley filling of pumped storage power stations under multi-timescale coordination.

[0180] The above description is merely a preferred embodiment of the present invention and does not constitute any limitation on the present invention. Any person skilled in the art can make many possible variations and modifications to the technical solution of the present invention, or modify it into equivalent embodiments, without departing from the scope of the present invention. Therefore, any modifications, equivalent changes, and alterations made to the above embodiments based on the technology of the present invention without departing from the scope of the present invention are within the protection scope of the present invention.

Claims

1. A multi-timescale optimization scheduling method for wind-solar-storage alliance systems, characterized by: The method includes, S1. Construct a wind and solar power output scene generation and reduction model that considers uncertainty. Generate an initial scene set of wind and solar power output based on the Latin hypercube sampling method, and reduce the initial scene set based on the scene reduction method of Kantorovich distance to obtain typical wind and solar power output scenes. S2. Construct a day-ahead optimization scheduling model. Based on typical wind and solar power output scenarios, establish a multi-objective stochastic programming model with the goal of maximizing the joint economic benefits of the wind-solar-storage alliance system and minimizing the fluctuation of the grid-connected remaining load. Solve the model to obtain the optimized power output plan of each unit for each day-ahead period. S3. Construct an intraday rolling optimization scheduling model. Based on a multi-objective stochastic programming model, with the objectives of maximizing the expected total revenue and minimizing the fluctuation of the remaining load within the rolling optimization window, construct an intraday rolling optimization model to provide real-time feedback and correction to the optimized output plan, and obtain the final scheduling instructions for each time period within the day.

2. The day-to-day multi-timescale optimization scheduling method for the wind-solar-storage alliance system according to claim 1, characterized in that: The generation of the initial scene set includes, The initial sample value matrix is ​​calculated as follows: ； ； In the formula, , Let be the distribution function loaded by the j-th variable and the j-th variable of the i-th sample, respectively. The sampled values ​​are obtained by hypercubic sampling; the initial sampled value matrix S contains... Let be the element in the i-th row and j-th column; N is the number of sample groups; each column of matrix P consists of random integers from 1 to N; each element in matrix R is a random number uniformly distributed in the interval [0,1). Reorder the initial matrix: Reorder the elements of matrix P After processing, an intermediate matrix Y is obtained that follows a standard normal distribution, and its elements... As shown in the following formula: ； In the formula, It is the inverse function of the cumulative probability distribution function of the standard normal distribution; Perform Cholesky decomposition on the covariance matrix COV(Y) of matrix Y: ； ； In the formula, The lower triangular matrix obtained from Cholesky decomposition. This is the lower triangular matrix. The transpose of the matrix, Represents a lower triangular matrix Its transpose The product of these variables is mathematically equal to the covariance matrix COV(Y) of the random variables. Arrange matrix P according to matrix The order of the elements in the matrix is ​​rearranged to obtain a new matrix. Then based on the new matrix The initial sample value matrix is ​​calculated to obtain the effective control sample matrix.

3. The day-to-day multi-timescale optimization scheduling method for the wind-solar-storage alliance system according to claim 2, characterized in that: The specific steps of the scene reduction method based on Kantorovich distance to reduce the initial scene set are as follows. S110. Assign equal initial probabilities to each scene in the initial scene set, and set the current scene set as the original scene set; S120, Calculate each pair of scenes in the scene set. Kantorovich distance between : ； In the formula, Let i be the vector of the landscape scene. Let j be the vector of the landscape scene. S130. For each target scene, find the scene closest to it, and calculate the product of that distance and the probability of the nearest scene: ； In the formula, For scene probabilities; S140. Find the scene pair that minimizes the product, delete one of the scenes from the current scene set, and add the probability of the deleted scene to the other matching scene, and update the current scene set and the number of scenes. S150. Repeat steps S120 to S140 until the number of scenes reaches the preset target number.

4. The day-to-day multi-timescale optimization scheduling method for the wind-solar-storage alliance system according to claim 1, characterized in that: The objective function of the multi-objective stochastic programming model is: ； ； ； ； ； In the formula, For the combined benefits of the wind-solar-storage alliance; , , These are the on-grid electricity price for time period t, the hydropower price, and the penalty price for curtailment of wind and solar power. , These are the system's on-grid electricity consumption and the amount of renewable energy abandoned; , These represent the on-grid electricity generated by wind power and photovoltaic power plants during time period t, respectively. , These represent the power generation and pumping power of the pumped storage power station during time period t, respectively. , These represent the amount of wind and solar power curtailment at wind power and solar power plants respectively during time period t; Let t be the cost of the pumped storage unit; The start-up and shutdown costs during power generation; The start-up and shutdown costs during pumping; This refers to the number of pumped storage units; This represents the number of pumped-storage units started and stopped during time period t. , These represent the number of pumped-storage units connected to the grid during time period t and time period t-1, respectively. Fluctuations in remaining load; Let t be the magnitude of the remaining load at time t; This represents the average value of the remaining load at time t. Let t be the load magnitude at time t.

5. The day-to-day multi-timescale optimization scheduling method for the wind-solar-storage alliance system according to claim 4, characterized in that: The constraints of the multi-objective stochastic programming model include: Pumped storage power station reservoir capacity constraints: ； ； ； In the formula, Let t be the reservoir capacity of the pumped storage power station. The reservoir capacity of the pumped storage power station at time t-1; This is the conversion factor between electricity and storage capacity; To improve the power generation efficiency of pumped storage power stations; , These are the initial and final reservoir capacities for the optimized cycle of the pumped storage unit; This is the lower limit of the storage capacity. This is the maximum storage capacity. Pumped storage power station operating condition transition constraints: ； ； ； In the formula, and These represent the pumped-storage unit being in power generation or pumping mode at time t, respectively. and These represent the number of times the pumped storage tank is started and stopped during power generation and pumping operations, respectively. , The pumped-storage unit is in power generation and pumping state respectively at time t-1; Maximum start-stop frequency constraint for pumped storage units: ； ； In the formula, and These represent the maximum number of start-ups and shutdowns of the pumped storage tank under power generation and pumping conditions, respectively. Unit output and grid-connected power constraints: ； ； ； ； ； In the formula, , These represent the minimum and maximum power output of the pumped-storage unit during time period t, respectively. , These represent the minimum and maximum output of the pumped storage unit during time period t, respectively. , These represent the minimum and maximum power output of wind power during time period t, respectively. , These represent the upper and lower limits of the photovoltaic power station's output during time period t; , These represent the upper and lower limits of the power received by the power grid during time period t, respectively. Constraints on wind and solar power curtailment: ； ； In the formula, , These represent the predicted power output values ​​for wind power and solar power during time period t, respectively.

6. The day-to-day multi-timescale optimization scheduling method for a wind-solar-storage alliance system according to claim 5, characterized in that: The objective function of the intraday rolling optimization model is: ； ； ； ； ； ； ； ； ； ； In the formula, For the past and present time periods, Forecast period in the future; To optimize the total predicted revenue of the time-complementary system for intraday time period t; , These are the real-time on-grid electricity price and pumping electricity price for the t1 time period, respectively. , These are the on-grid electricity price and pumping electricity price for the t2 period, respectively. , , , , , Optimized for the day at time t The power generation of wind and solar power, the amount of wind curtailment, the amount of solar curtailment, pumped storage power generation, and pumped hydro power are all subject to real-time dispatch. , , , , , They are respectively Real-time forecasts of wind and solar power generation, curtailed wind power, curtailed solar power, pumped storage power, and pumped hydro power generation participating in the dispatch; , The actual wind and solar power outputs at time t1 within the day are respectively. , The predicted wind and solar power outputs at time t2 are respectively. , They are respectively , Start-up and shutdown costs of pumping water during specific time periods; Costs associated with pumped storage start-up and shutdown; This represents the total number of pumped storage units. This is an indicator of the start-up / shutdown status change of the kth pumped storage unit during time period t within the day. Let $t$ be the startup cost of the k-th generating unit under power generation conditions during time period $t$. Let $t$ be the startup cost of the k-th unit in pumping mode during time period $t$. The online status of the k-th generating unit during time period t; This refers to the online status of the k-th unit during time period t-1. This refers to fluctuations in the remaining load during the intraday optimization phase. This represents the remaining load at time t during the intraday optimization phase. This represents the average value of the remaining load during the daily optimization phase. This represents the load size at time t during the intraday optimization phase.

Images